Point d'Inflection

THESE

présentée à

L’INSTITUT NATIONAL DES SCIENCES APPLIQUEES DE TOULOUSE

pour l’obtention du

DOCTORAT

SCIENCE DES PROCEDES

SPECIALITE : GENIE DES PROCEDES ET DE L’ENVIRONNEMENT

par

Srayut RACHU

Master of Environmental Engineering – Chulalongkorn University, Bangkok, Thaïlande

CONTRIBUTION A LA MISE AU POINT D’UN LOGICIEL DE CALCUL DE PROCEDES ET FILIERES DE TRAITEMENT D’EAUX

RESIDUAIRES HUILEUSES

Soutenance prévue le 16 Décembre 2005 devant la commission d’examen:

M. D. HADJIEV Professeur, IUT, Lorient Rapporteur

M. J. ROLS Professeur, UPS, Toulouse Rapporteur

M. Y. AURELLE Professeur, INSA, Toulouse Directeur de thèse

M. A. LINE Professeur, INSA, Toulouse Directeur de LIPE

M. R. BEN AIM Professeur Emérite, INSA, Toulouse

M. S. SAIPANICH PDG, Progress Technology Consultants Co.,Ltd., Thaïlande

M. H. ROQUE Professeur Emérite, INSA, Toulouse Invité

Executive summary

Sommaire

Sommaire

Le traitement des eaux résiduaires huileuses est l’un des sujets de recherche majeurs dans le laboratoire GPI. L’hydrocarbure ou l’huile est l’un des polluants de l’eau les plus importants. Une petite quantité de l’huile peut produire le film vastement couvrant la surface de l’eau, lequel affecte le transfert de l’oxygène et par conséquence ruine l’écosystème. Même s’il est biodégradable, il contribue à la demande biologique en oxygène (DBO) importante. En plus, étant donné ses propriétés, à la haute concentration, il cause l’effet nuisible dans le procédé de traitement biologique. Toutefois, l’hydrocarbure ou l’huile peut avoir la valeur ou être récupérée ou recyclée à condition où il peut être séparé de l’eau. En cet effet, les techniques de la séparation huile/eau sont parmi des recherches principales dans le laboratoire GPI. Il y a plusieurs études sur les techniques, les procédés et les innovations de la séparation d’huile initiée par le laboratoire GPI. Chaque étude peut être appropriée à certaine condition de l’opération ou certain caractéristique des eaux résiduaires.

Ainsi, les buts de cette thèse sont de réexaminer les recherches du laboratoire GPI, réalisées dans l’Equipe du Professeur AURELLE, sur le procédé de traitement des eaux résiduaires huileuses ; d’établir le design du procédure générale avec les précautions de tels procédés ; et, ensuite, valoriser et maximiser l’utilisation de ces connaissances établies sous forme du logiciel. Ainsi, les objectifs de la thèse ont été définis pour réaliser ces buts ci-dessus :

1. Réexaminer les technologies de traitement pour les eaux résiduaires huileuses ou les eaux résiduaires polluées par l’hydrocarbure, dans les recherches de doctorat réalisées dans l’Equipe du Professeur Yves AURELLE, du commencement jusqu’ au présent.

2. Généraliser et proposer le modèle de chaque procédé relatif.

3. Composer le textbook sur les eaux résiduaires hydrocarbure-polluées ou les procédés de traitement des eaux résiduaires huileuses, basés sur le résultat des 1er et 2ème objectifs.

4. Développer le prototype du logiciel pour la recommandation de sélection du procédé, le design des unités procédé et la simulation du plan de procédé de traitement des eaux résiduaires huileuses.

Sommaire des procédés de traitement des eaux résiduaires huileuses

Les trois premiers objectifs sont accomplis dans la Partie 1 à la Partie 3 de cette thèse. Tous les filières de procédé de récupération d’huile, étudiées dans le laboratoire GPI, avaient été réexaminées ; et ses modèles mathématiques correspondantes aussi bien que la limitation du design et les paramètres influents avaient été établis. Ces procédés qui sont réexaminés sous les buts et objectifs du travail de cette thèse sont ;

1. L’Écrémeur déshuileurs

Ce dispositif est développé pour récupérer sélectivement la couche d’huile de la surface de l’eau sans entraînant de l’eau avec lui. On a trouvé que le moyen pour réaliser la bonne sélection d’huile dépend de l’énergie de surface ou de la tension superficielle critique du matériel écrémeur. Le matériel avec la basse tension superficielle critique convient à être utilisé comme matériel écrémeur.

[1]

Sommaire

Les modèles mathématiques de deux types d’écrémeur : le tambour et le disque déshuileurs sont vérifiés ;

Pour le tambour déshuileur,

0.514g

L0.486oν

1.541N1.5413.035DP = m3/s {1.1}

Pour le disque déshuileur,

( ) ⎥⎦⎤

⎢⎣⎡ −−= 54.2)5.0(54.25.00.514g

0.486oν

1.5413.464NP IDD m3/s {1.2}

Où P = Productivité en huile du tambour ou du disque (relative à une surface) (m3/s)

D = Diamètre du tambour ou du disque (m) N = Vitesse de rotation (rev/s) νo = Viscosité cinématique de l’huile (m2/s) L = Largeur du tambour (m) I = Hauteur d’immersion totale du disque dans le liquide (m), I ≤

D/2. g = Accélération de la pesanteur (m/s2) γo = Tension superficielle tension de l’huile (kg/s2 or N/m = 1000

dyne/cm)

Tambour

Racleur

Vis d’Archimede

Nappe d’huile

Phase aqueuse

Disque

Racleur

Moteur

Goulotte de collecte

Fig. 1-1 Tambour déhuileur Fig. 1-2 Disque déhuileur (Par: Abanaki)

Les modèles sont valides dans ces hypothèses ; • La tension superficielle de l’huile est dans l’intervalle de 27 à 34 dynes/cm, ce qui

couvre pratiquement toute huile ordinaire. • Le nombre capillaire (Ca = μo V/γo) est de 0.04 à 3.6. • La densité de l’huile est d’environs 0.79 à 0.83 kg/m3. La viscosité dynamique d’huile

expérimentée est entre 1.35x10-3 et 291x10-3 (N.s)/m2 (1.35-291 cp).

[2]

Sommaire

• Pour le tambour déshuileur, la vitesse périphérique ne devrait pas être supérieure de 0.8 m/s. En fin d’éviter l’entraînement d’eau, la vitesse de 0.44 m/s ou moins est recommandée.

• Pour le disque déshuileur, la vitesse périphérique ne devrait pas être supérieure de 1.13 m/s. En fin d’éviter l’entraînement d’eau, la vitesse de 0.5 m/s ou moins est recommandée.

• Les modèles se prouvent valides pour les écrémeurs faites de SS, PP, PVC et PTFE.

2. Le Décanteur

Sa théorie importante, à savoir la loi de STOKE, est citée. Selon cette théorie, les gouttelettes peuvent être récupérées à condition qu’elles entrent dans le décanteur à la distance ascensionnelle nécessaire pour atteindre la surface de décantation, comme la surface de l’eau ou la surface de l’intercepteurs d’huile intérieurs, avant d’être entraînées hors du réservoir par l’eau. D’après ce concept, la relation typique de la taille de gouttelette et son efficacité de séparation par classe de goutte (ηd) est comme celles montrées dans la fig. 2-1.

VU

Q

d = dc

d < dc

d > dc

ENTREE SORTIEL

H, D

d

ηd

d c

d < d c

Zone 1 Zone 2

d = or > d c

API

PPI (n plaques)

L

H

H

H1

La cellule “Spiraloil”(section transversale)Hauteures sont differentes.

H2

Figure. 2-1 Schéma et courbe caractéristique de l’efficacitéd’un décanteur

Les modèles mathématiques générales pour les procédé de 3 variétés différentes, c’est-à-dire le décanteur classique (API), le décanteur lamellaires (PPI), et le décanteur compacte (Spiraloil) sont proposées et vérifiées :

Pour d ≥ diamètre coupure, dc %100=dη {2.1} Pour d ≤ dc %100⋅=

dc

dd U

Uη {2.2}

La vitesse ascensionnelle d’une gouttelette (Ud) c

ddgU

μρ18

2⋅⋅Δ= {2.3}

La diamètre de coupure, c’est-à-dire la plus petite taille de gouttelette dont l’efficacité de séparation est de 100%, est déterminée par les équations suivantes, Le décanteur simple, (A = aire ecoulement, Q = debit d’eau)

1/2

ΔρgLAc18HQμ

cd ⎟⎟⎠

⎞⎜⎜⎝

⎛=

{2.4}

[3]

Sommaire

Le PPI, 1/2

1)(NPΔρgSc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

+=

{2.5}

Pour un décanteur compact qui hauteur de décantation n’est pas bien connu.

1/2

dSgΔρc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

⋅⋅=

{2.6}

Les modèles sont valides à condition que la loi de STOKE soit valide (Re = 10-4 à 1). Pour le PPI, si les plaques sont inclinées, le paramètre « Sd » dans l’éq. 2.5 sera remplacé par « Sp cos α » (pour les plaques horizontales, α = 0o). Le Sd , pour le décanteur compacte, est l’espace totale de la surface de décantation qui peut intercepter de l’huile, sans regardant aux distances ascensionnelles des goulettes d’huile.

3. Le Coalesceur

Le coalesceur est le procédé de séparation accéléré, conçu pour favoriser la coalescence entre les gouttelettes d’huile en but de former les plus grosses gouttelettes d’huile. Selon la loi de STOKE, la vitesse ascensionnelle est proportionnelle au carré de diamètre de la gouttelette. Ainsi, en agrandissant la taille de la gouttelette, l’efficacité de séparation d’huile est augmentée considérablement.

A partir du coalesceur de lit granulaire, AURELLE a proposé que les mécanismes de fonctionnement du coalesceur puissent être divisés en 3 étapes ;

• 1ère étape : La Transportation, qui se constitue de 3 phénomènes de transport principaux : la sédimentation, l’interception directe et la diffusion. Le modèle mathématique de chaque phénomène est basé sur le modèle de filtration. Ces modèles fournissent les idées claires concernant les paramètres influents sur la performance du coalesceur. • 2ème étape : L’Adhésion-Coalescence. L’efficacité de cette étape dépend principalement de la mouillabilité du matériel coalesceur. • 3ème étape : La Relargage de la phase dispersée. Cette étape dépend de 4 paramètres, i.e. la mouillabilité du matériel coalesceur à la surface de relargage, la vitesse d’écoulement, la tension interfaciale, et la proportion de la phase dispersée et la phase continue dans l’émulsion à traitée.

Le diametre du collecteur =dp,Coefficient de vide = ε

V

H

ENTREE:Gouttelette

Dia. = d

SORTIE:Goutte grosse La surface de relargage

La grilles

V

Fig. 3-1 Schéma d’un coalesceur

[4]

Sommaire

U= Vitesse ascensionalleV= Vitesse d’ecoulment

V UVU

Gouttelette

Lignes de courant Collecteur

a) Transport par interception b) Sédimentation c) Diffusion

Fig. 3-2 Schéma des 3modes de transport de l’étape 1

Les mécanismes à 3-étapes peuvent être utilisés également pour décrire la performance d’autres types du coalesceur, par exemple le coalesceur de lit fibreux et celui de lit dynamique. Toutefois, pour la prédiction de l’efficacité par classe de goutte, les modèles empiriques sont recommandés car elles sont vérifiées par un ensemble de données plus large.

Les modèles empiriques, vérifiées par le biais d’une gamme de données relativement large, sont proposées pour les coalesceurs de 4 variétés : coalesceur de lit granulaire, coalesceur fibreux type brosse, coalesceur fibreux dynamique (brosse tournante), et coalesceur fibreux nonordonnée (laine d’acier). Veuillez noter que l’efficacité calculée supérieure de 100% sera comptée pour 100%.

Pour le coalesceur de lit granulaire,

%100)0.09)cρΔρ(0.09)

μcμd(0.08)

o/wγdpVcρ(0.12)

dpH(0.2)

dpd58(.0(dη ⋅−= {3.1}

Les modèles sont valides dans ces hypothèses ;

• La diemètre de collecteur (dp) = 0.36 - 0.94 mm. • Tension interfacielle (γow) = 11 - 42 dyne/cm. • Vitesse d’écoulement (V) = 0.09 - 0.54 cm/s. • La différence de mass spécifique entre deux phases (huile/eau) (Δρ) = 83 - 314 kg/m3. • Le lit est oleophile. • La concentration initiale de hydrocarbure < 1,000 mg/l.

Pure le coalesceur fibreux,

( ) %100)0.694)DH(0.35ε10.18)

DFd

(0.18)Dd(0.77)

cμVDcρ104.5((dη ⋅−−−=

{3.2}

[5]

Sommaire

Les modèles sont valides dans ces hypothèses ;

• 48 < Re < 1100. Re est le terme (ρcVD/μc) dans l’équation. 1 < H/D < 10. • La diamètre de coalesceur (D) est environ 5.0 cm. Pour le coalesceur de diamètre

important, il y a un risque de court-circuit par entrainant despassages préférentiels le long des parois du coalesceur.

• Le modèle est valide pour la diamètre de gouttelette (d) de 1 micron ou plus. • Vitesse d’écoulement (V) est entre 0.5 et 2.0 cm/s (1.8 - 120 m/h). • Le diamètre de fibre (dF) = 100 - 200 microns. • Coefficienct de vide (ε) est 0.845 - 0.96. • La concentration initiale de hydrocarbure < 1000 mg/l. • Les lits sont de type brosse et sont oleophiles.

Pour le coalesceur dynamique (ou brosse tournante),

%100)0.74V0.58

Fd0.03D

0.53N0.35H0.35ε)(10.580.67d(dη ⋅−

= {3.3}

Les modèles sont valides dans ces hypothèses,

• 52 < Re < 1164. Re est le terme (ρcVD/4c) dans l’équation. 1 < H/D < 2. • Vitesse de rotation (N) = 0.167 - 3.33 rps (10 to 200 rpm). • Vitesse d’écoulement (V) = 0.1 - 1.1 cm/s (3.6 to 39.6 m/h). • Diamètre de fibre (dF) = 100 - 300 microns. • Diameter de coalesceur (D) < 11.5 cm. • Le modèle est valide pour la diamètre de gouttelette (d) de 10 microns ou plus. • Les lits sont de type brosse et sont oleophiles.

Pout le coalesceur fibreux non-ordonnée (laine d’acier),

%100))()()()(35.3( 36.003..003.023.0 ⋅= −−

DH

Dd

DdVD F

c

cd μ

ρη

{3.4}

Les modèles sont valides dans ces hypothèses,

• Re est entre 840 - 2470. • La diamètre de fibre (dF) = 40-75 microns. • Coefficient de vide (ε) est environ 0.95. • La diamètre de coalesceur (D) = 5 cm. • La hauteur d’intercepteur tournant (H) = 0.07 - 0.21 m. • Vitesse (V) = 1 - 2.5 cm/s (36 to 90 m/h). • Le modèle est valide pour la diamètre de gouttelette (d) de 1 microns ou plus. • La concentration initiale de hydrocarbure est environ 1000 mg/l. • Le lit est oleophile.

[6]

Sommaire

Les modèles des pertes de charge sont aussi proposées ;

Pour le coalesceur granulaire, l’équation de Kozeny-Carman’s est proposée,

32

2)1(180ερεμ

⋅⋅⋅−

=dpgVH

P c m

ε = 0.13 à 0.23.

{3.5}

Pour les coalesceurs fibreux, l’équation de Hazen-William est utilisée,

167.185.1582.6 ⎟

⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛=

DL

CVPHW

m

CHW = 130. L = hauteur de l’intercepteur.

{3.6}

Finalement, plusieurs idées initiatives pour améliorer la performance du coalesceur, c’est-à-dire l’utilisation du guide pour augmenter la concentration d’huile entrante du coalesceur et l’utilisation du matériel de lit mélangée dans le coalesceur de lit granulaire pour traiter simultanément la mélange d’émulsion directe/inverse, sont réalisées.

4. La flottation à air dissous

La flottation à air dissous est le procédé de séparation accéléré, qui s’opère par l’augmentation de la différence masse volumique entre la phase continue et la phase dispersée. Ceci est accompli moyennant la création des bulles du gaz ou de l’air dans les eaux résiduaires en fin de favoriser la formation des agglomérats air-solides ou air-huileux.

Le modèle mathématique basé sur le concept de filtration, proposé par SIEM est réexaminé. Ce modèle est utile pour comprendre les effets des paramètres influents sur la performance de la DAF. Les équations de généralisation du modèle de SIEM, fondées sur la théorie de l’équilibre de population, sont établies dans cette thèse pour agrandir la gamme valide du modèle de SIEM et, donc couvrir la gamme des eaux résiduaires huileuses fréquemment trouvée. Les modèles sont comme montrés ci-dessous ;

V

Ub Ud

x

y

Bulle Gouttelette Concentration = C (en tenant compte de la dilution)

dH

Concentration afterintercepted by bubbles = C- dC

Vitesses d’ecoulement = V

Vitesse relative = Vr

Taille de bulle = db

Hauteur de la colonne= H

Section de la colonne = Ao

Fig. 4-1 Representation schématique de vitesses at vitesse relative et schéma de la colonne flottation

[7]

Sommaire

Ur= vitesse relative de decantationVr= vitesse relative d’ecoulement

Vr Ur

Vr Ur

Gouttelette

Lignes de courant

Bulle

a) Interception directe b) Sédimentation c) Diffusion Fig. 4-2 Schéma des 3modes de transport

SIEM’s model,

%100)1())(

23

(

,

exp

⋅−=Φ

−bd

HAV

refd eαη

η {4.1} 5919.0

exp )(009005.0)( theoηαη = {4.2}

diffIntsedtheo ηηηη ++= {4.3}

2)(23

bInt d

d=η {4.4}

rc

wateroilsed V

gdμ

ρη

18

2/Δ

= {4.5}

3/2)(9.0br

Diff ddVKT

μη = {4.6}

c

bwaterairbr

gdUV

μρ

18

2/Δ

== {4.7}

Où μc = Viscosité de phase contenue (eau) (L2/T) K = Constante de Boltzman (1.38*10-23) T = Température absolue (Kelvin) db = Taille de bulle (L) d = Diamètre de gouttelette (L) Vr = Vitesse relative enter des bulles et des gouttelettes (L/T)

Quant à l’équation de généralisation du modèle de SIEM, fondée sur le concept de l’équilibre de population, sa forme générale est comme montrée ci-dessous. Le κ2, ref est la constante, calculée par ηd,ref dans le modèle de SIEM. Mais, le point clé est le moyen d’adapter la valeur de Φ et τ pour s’approprier à la condition du design et contribue à la prédiction correcte de l’efficacité, c’est-à-dire, utiliser le Φ de SIEM avec le design τ, ou vice-versa, dans la calcul. La logique est assez compliquée pour résumer en un paragraphe court, et donc n’est pas citée ici. Pour de plus ample information, veuillez consulter le rapport principal de la thèse (Chapitre 6, Partie 3). La logique est d’ailleurs disponible dans le logiciel.

[8]

Sommaire

%100)1( )( ,2 ⋅−= Φ− τκη refed {4.8}

Les équations pour le design du pressurisateur ou du système de l’eau pressurisé sont aussi proposées et vérifiées dans cette thèse ;

La concentration du gaz dissous,

')().( HgasMWPygasConc ⋅⋅⋅= mg/l {4.9}

Le débit du gas ou d’ air dégazé,

))/(10082.0()(' 33 KmolmTPPQRHy atm ⋅×⋅⋅−⋅⋅⋅⋅=Φ − m3/s {4.10}

où H’ = constante de Henry (molair/(m3water-atm)) et y = fraction molaire du gaz

dans l’air. Pour l’air, y = 1.

La consommation énergétique de la compresseur,

pump

gaugepw

pump

atmpw PQPPQPower

ηη)()( ⋅

=−⋅

= Watt {4.11}

La consommation énergétique du pressurisateur au débit Qpw,

pwatm

air

comp

QgasMWgasConc

PPTRPower ⋅⋅

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡−⎥

⎦

⎤⎢⎣

⎡⋅⋅=

−

)()(1

4.01

)4.1

14.1(

η Watt {4.12}

Où Rair = Constante universal du gaz. T = température absolue

5. L’Hydrocyclone

L’hydrocyclone est le procédé de séparation accéléré. Son concept de travail principal est de remplacer l’accélération gravitationnelle qui dirige la vitesse de décantation au moyen de l’accélération centrifuge supérieure. MA et AURELLE ont proposé une approche nouvelle de prévoir l’efficacité hydrocyclone, appelée le modèle d’analyse trajectoire pour l’hydrocyclone biphasique (liquide-liquide), fondée sur la loi de STOKE. En fait, ce modèle peut prévoir, par l’équation théorique, l’efficacité de séparation par classe des gouttelettes d’huile de toute taille dans les eaux résiduaires ; à la différence des autres modèles qui sont basés sur les concepts du modèle empirique et de la similarité. Ainsi, il est très pratique de comprendre l’effet des paramètres reliés à la performance d’hydrocyclone. Le modèle est constitué des équations qui dirigent 3 éléments de vitesse des gouttelettes d’huile à l’intérieur du cyclone, comme montré ci-dessous.

∫=∫L

0 WdZdR

zvvR UdR {5.1}

RV 2

18μ

2ΔρdU = {5.2}

0.65)RnD

)(2iπD

Q(V = {5.3}

319.1

263.81233.3 ⎟⎟

⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zRR

zRR

zRR

zWW {5.4}

[9]

Sommaire

2))2/tan(Znπ(0.5D

QzW

β⋅−= {5.5}

)2

tan(2

β⋅−= Z

DR n

z {5.6}

)2/5.1tan()(25.0

onD

L = {5.7}

Pour le hydrocyclone de type THEW, quand Z = L:

)2/(186.0 nVZZ DR = {5.8}

d = dc

d > dcd < dc

Z

R

L

Rd

Parois

LZVV

0.5Dn

0.186Dn

d

ηd

d c

d < d c

Zone 1 Zone 2d = or > d c

100%

Fig. 5-1 Trajectories of oil droplets and typical efficiency curve from trajectory analysis model

L’efficacité par classes peut être calculée par les relations:

Pour d ≥ diamètre de coupure,dc %100=dη {5.9} Pour d < dc

%100)

2186.0()

2(

)2

186.0(

22

22

⋅−

−=

nn

nd

d DD

DRη

{5.10}

Les modèles sont valides dans ces conditions,

• Veuillez noter que les équations sont fondées sur la géométrie d’hydrocyclone liquide-liquide initiée par Professeur THEW, en Angleterre. Pour les utiliser avec d’autres types d’hydrocyclone, certaines constantes numériques (par exemple 0.63, 3.33, etc.) dans les éq. 5.3, 5.4 et 5.8 doivent être réexaminées pour s’approprier aux géométries nouvelles.

• Le modèle est valide pour la taille des gouttelettes (d) de 20 microns et plus. • Les équations sont valides pour l’hydroclone ayant seulement 2 entrées.

[10]

Sommaire

• Le débit de purge huile à surverse (Qoveflow) est généralement petite, pas plus de 10%. Son effet sur les profils de vitesse et d’efficacité est petit, et donc, négligeable. La Qoveflow recommandée est de 1.8 à 2 fois du débit entrant de l’huile.

• Le modèle tend à prévoir l’efficacité plus basse quand d < d80%, et celle plus haute quand d > d80%. L’erreur dans la prédiction de la diamètre de coupure est environs 10-20%, c’est-à-dire, si la diamètre de coupure prévue est 50 microns, la diamètre de coupure constatée devrait être environs 40-45 microns. Pour de plus ample information, veuillez consulter la référence.

MA et AURELLE a également inventé une nouvelle type d’hydrocyclone pour la séparation simultanée de l’huile, des matières en suspension (MES) et de l’eau, appelée l’hydrocyclone triphasique. Le modèle pour cet hydrocyclone est établi nouvellement dans cette thèse et également basé sur le concept d’analyse trajectoire.

Solid-liquid Liquid-liquid (de type Thew)

DoDDs

DiDu

Dp

L5 L3L1L3

L4

Note: Di/D=0.25 pour 1- entrée et 0.175 for 2- entrée, Do/D=0.43,Ds/D=0.28, Du/D=0.19, Dp/D=0.034, L1/D=0.4,L2/D=5, L3/D=15, L4/D=0.3, Solid-liquid :angle =12o, liquid-liquid angle =1.5o

Fig. 5-2 Hydrocyclone triphasique

Entree

Entree

Sortie eau

Purge MES

Purge huile

Fig. 5.-3 Trajectoires de l’huile (sphere) and MES (cube) dans le hydrocyclone triphasique

Solide-liquide(RIETEMA)

Liquide-liquide(THEW)

Fig. 5-4 Profils des vitesses axiales dans le hydrocyclone triphasique

[11]

Sommaire

Généralement, les équations pour l’hydrocyclone triphasique sont similaires à celles pour l’hydrocyclone biphasique avec un peu de modifications, c’est-à-dire ;

• Remplacer l’éq. 5.3 par l’éq. 5.11.

0.65)RnD

(2iD

4π(Q/2)0.676V

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

= {5.11}

• Remplacer, dans l’éq. 5.1, « Dn » par « Do » et « L » par « L5 ».

Pour l’efficacité solide-liquide, elle peut être calculée par le model de RIETEMA ou d’autres modèles compatibles dont la partie solide-liquide de l’hydrocyclone est conforme à la géométrie de l’hydrocyclone de RIETEMA.

Finalement, les équations de la fuite de pression ou la perte de charge pour l’hydrocyclone biphasique ou triphasique sont établies dans cette thèse :

Pour l’hydrocyclone biphasique, la perte de charge dans l’entrée/surverse (Po) et l’entrée/sousverse (Pu) peuvent être calculée par les équations suivantes,

1611.0

4

3.2

)1(6.216 ⎟

⎟⎠

⎞⎜⎜⎝

⎛

−⋅=Δ

fno RD

QP bar (V : m/s et D : meter) {5.12}

4

2.26.4

nu

D

QP =Δ bar {5.13}

Pour l’hydrocyclone triphasique, la perte de charge entre l’entrée et la sortie d’huile (Poil), entre la sortie d’eau (Pw) et la sortie de MES (PSS) peuvent être calculée par les équations suivantes,

4D

2.12Q49.8waterΔP = bar {5.14}

4D

2.34Q21ssΔP = bar {5.15}

4D

2.03Q55oilΔP = bar {5.16}

6. Les procédés membranaires Le principe de travail

Le procédé membranaire est le procédé de séparation qui est basé, principalement mais non entièrement, sur le concept de filtration. Physiquement, la membrane est le matériel perméable ou semi-perméable qui restreindre la motion de certaines espèces. Théoriquement, on peut toujours séparer certains composants d’émulsion à condition que la membrane choisie convient à la différence dimensionnelle.

[12]

Sommaire

Fig. 6-1 Classification des techniques membranaires de filtration selon la taille des pores

(Par Osmonics)

Les applications des procédés membranaires, c’est-à-dire, la microfiltration (MF), l’ultrafiltration (UF), le nanofiltration (NF) et l’osmose inverse (OI), au traitement d’eaux résiduaires huileuses dans le laboratoire GPI ont été réexaminées. Plusieurs techniques utiles émanant de ces recherches sont réalisés et les techniques de calcul sont établies dans cette thèse :

• L’augmentation du flux pour l’UF dans le traitement d’émulsion de l’huile de coupe par la déstabilisation partielle. L’addition du sel, de moindre quantité que celle pour la déstabilisation totale, dans l’alimentation contribue à l’augmentation du flux de perméat, et donc économiser la consommation d’énergie ainsi que la taille de la membrane. L’inconvénient de ce technique, inclus dans cette thèse, est que l’huile déstabilisée peut colmater la membrane si elle n’est pas correctement partitionnée ou séparée du courant d’alimentation recirculé.

• La régénération de la membrane UF après le traitement de l’huile de coupe par la microémulsion. Il peut efficacement laver de l’huile accumulée ou les foulants restant sur la membrane. Il peut être réutilisé pour plusieurs fois jusqu’à ce qu’il est saturé par l’huile.

• Le technique pour généraliser les données de la performance UF dans une seule condition pour couvrir des autres conditions, basé sur la combinaison du modèle de la couche de gel et modèle des résistances en série. Ce technique est proposé et vérifié dans cette thèse. Il est utile quand les données limitées sur des eaux résiduaires examinées sont disponibles. Il nous permet d’estimer l’évolution du débit ainsi que celle du volume de perméat en fonction du temps, lequel est utile pour le design du procédé.

[13]

Sommaire

• Le technique pour prévoir le débit du mélange de 2 émulsions différentes. Ce technique est également proposé et vérifié dan cette thèse. Il est utile quand l’émulsion mélangée peut être prévue et sa proportion tend à varier.

Permeat

Retentat

Membrane

pompe

Reservoir d’alimentation

Alimentation

Po

Pi

Pp

Echangeur de chaleur

Fig. 6-2 Schéma de l’unité ultrafiltration tangentielle en “batch”

Quant au modèle mathématique des procédés membranaires, on met l’accent sur le modèle de l’UF dans l’opération batch car il joue un rôle majeur dans le traitement d’eaux résiduaires huileuses. On a adopté deux modèles largement utilisés : le modèle de la couche de gel et le modèle des résistances en série. Les équations des deux modèles sont comme montrées ci-dessous,

Le modèle de la couche de gel

)ln(C

CkVJ gβ= l/(m2-h) {6.1}

Le modèle de résistance en série,

tm

t

PVRP

J⋅⋅+

= αφ' l/(m2-h) {6.2}

Les valeurs de k, α ,β, φ, R’m, et Cg varient selon les caractéristiques des eaux résiduaires et les types de la membrane.

7. Le procédé thermique

L’intérêt principal du procédé de ce type est à la distillation hétéroazéotropique (DH). L’application de la DH au traitement des slops de raffinerie ainsi que des retentats issus de l’UF de l’huile de coupe est réalisée par LUCENA et AURELLE. Le procédé est accompli par l’addition du produit chimique qui favorise la formation azéotropique (appelée extractant), généralement les hydrocarbures, dans l’eau résiduaire. Ceci réduira la température d’ébullition du système, et donc économisera l’énergie. Ces applications apportent la possibilité de revaloriser ces matériels résiduaires potentiels. Son application inverse, à savoir le stripping (pour récupérer la substance volatile de l’eau au moyen de l’addition du vapeur), est également réalisée.

L’hétéroazéotrope est l’un des phénomènes liés au équilibre vapeur-liquide-liquide (VLLE). Le diagramme isobare du mélange immiscible hétéroazéotropique est démontré dans la figure 7-1. Il est constaté que la courbe de bulle de ce diagramme est représentée par une ligne horizontale à la température constante, appelée « la température azéotropique ». A cette température, les deux liquides dans le mélange évaporent, quelle que soit la composition du mélange. La composition de la vapeur est toujours yH jusqu’à ce que il reste seulement une espèce de liquide dans la phase liquide.

[14]

Sommaire

A

xw = 1yw = 1

T P = const.

Huile pur Eau pur

xw = 0yw = 0

Huile +V

xw , yw

Point d’ebullitionde B

Vapeur (V)

Eau + V

Huile+eau

xw = xH yw = yH

Point d’ebullitiond’eau

THB C

D

xw,1

xw,2

xw,3 xw,4=1

xw,5=1xw,6=1

yw,1’ to yw,4 = yH

yw,5

yw,6

xw,1’

Azeotrope (H)

Fig. 7-1Diagramme d’équilibre isobar: température-concentration de l’eau residuaires

huileuses

Dans le cas des eaux résiduaires huileuses, spécialement celles concentrées comme le slop ou le retentat de l’UF dérivant de l’UF de l’huile de coupe, si l’extractant est choisi correctement, elles formeront une condition azéotropique et, pendant son évaporation, extraira le teneur en eau des eaux résiduaires. L’eau sera extraite jusqu’à ce que le résidu devienne l’hydrocarbure sans eau. Le vapeur condensera pour former le distillat contenant deux couches séparées du extractant (la couche supérieure) et l’eau (la couche inférieure) de la composition xH (= yH). Plus la valeur de yH est haute, mieux la capacité de l’extraction d’eau est. Les données théoriques de yH, basées sur la loi de Raoult et la loi de Dalton, sont également proposées comme montrées dans la table 7-1.

Table 7-1 Heteroazeotropic temperature and composition des certaines hydrocarbures

Extractant Molecular

weight (g/mol)

TH (deg. C)

yH (by molar)

yH (by volume)

y H observed (by volume)

C6H14 56 61.6 0.209 0.0351 C7H16 100 79.2 0.452 0.0922 C8H18 114 89.5 0.616 0.188 C9H20 128 94.8 0.827 0.3255 C10H22 142 97.6 0.914 0.495 0.468 C11H24 156 98.9 0.959 0.6663 C12H26 170 99.5 0.98 0.7953 0.767 C13H28 184 99.8 0.991 0.890 C14H30 198 99.95 0.996 0.9542 C15H32 212 99.999 0.998 0.9702 C16H34 226 ≈ 100 0.999 0.9840

Les équations utilisées dans le calcul de l’entraîneur ou extractant, dans le cas des eaux résiduaires huileuses et la vapeur, dans le cas du stripping, sont comme montrées ci-dessous,

[15]

Sommaire

H

Hwaterentrainer y

yVolumeVolume )1( −= {7.1}

)1(tanH

Htspollusteam y

yVolumeVolume−

= {7.2}

8. Le procédé chimique

Au cas où les eaux résiduaires huileuses contiennent l’émulsion stabilisée, c’est-à-dire la macro/microémulsion, les tailles des gouttelettes sont si petites et stables qu’elles ne peuvent pas être séparées par les procédés basés sur la loi de STOKES, par exemple le décanteur, le coalesceur ou l’hydrocyclone. A moins que les procédés membranaires ou thermiques soient appliqués, les gouttelettes doivent suivre un procédé chimique qui détruit sa stabilité et favorise la coalescence de ces gouttelettes. Ainsi, le procédé chimique dans ce cas s’appuie sur la déstabilisation chimique et la coagulation/floculation chimique. Selon ces recherches, on peut conclut que la stabilité de l’émulsion est fondée sur les facteurs suivants,

• La barrière électrique, laquelle peut être décrite par la théorie DLVO et la théorie de la couche double, comme montrée dans les figures 8-1 et 8-2.

• La barrière stérique, causée par la localisation des molécules tensioactives à la surface des gouttelettes d’huile. Elle s’agira comme un film qui empêche la coalescence des gouttelettes.

En fin de déstabiliser l’émulsion, les deux barrières doivent être éliminées. Selon les recherches du GPI, le mécanisme de déstabilisation (ou cassage) de l’émulsion par l’addition de plusieurs produits chimiques, c’est-à-dire du sel monovalent, des sels bivalents, les polyélectrolytes, l’acide et les agents adsorbants, sont réexaminées. Les mécanismes de déstabilisation des ces produits chimiques, fondés sur l’émulsion stabilisée par les agents tensioactifs anioniques, sont résumés comme ci-dessous,

• Les produits chimiques monovalents : Le procédé principal de déstabilisation est fondé sur la réduction de la couche diffuse au moyen de l’augmentation des ions chargés opposés. L’inconvénient principal des produits chimiques de ce groupe est la formation du saline ou des polluants acides au terme de la déstabilisation. Les exemples de ces produits chimiques sont NaCl, H2SO4.

• Les produits chimiques bivalents : Le mécanisme principal de déstabilisation est basé sur la précipitation des tensioactifs ; l’équilibre des tensioactifs est donc modifié. Les tensioactifs focalisants à la surface des gouttelettes d’huile seront retransmis dans la phase d’eau. Ainsi, la gouttelette est déstabilisée. La performance est dirigée par le produit de solubilité des tensioactifs. Les exemples des produits chimiques sont CaCl2, MgSO4 et MgCl2. Même si la dose requise pour ce cas est inférieure à celle des produits chimiques monovalents, elle cause encore les polluants résiduaires de saline. Des sels organiques, comme le formiate de calcium (Ca(COOH)2) seraient plus intéressants car les polluants résiduaires sont (COOH)2 qui est organique et biodégradable.

[16]

Sommaire

Huile + tensioactif

(TA) ++

+++ +

+

+

++

- -

--

-

--

-

- ----

- -- --

--

++

+

+

++

+

+

+

+

++

++

+

+

++

+

+

-+

-+

+-

+

+

+

-

+

++

+

++

- -

Couche Stern

Couch de diffusion Ion de charge positif avec

la solvation d’eau

Potentiel de SternPotentiel zeta

---

---

-

--

+

+

+

++

++

++

+

+

+

+

+

+

+

+

+

+

+

-

+

Couch Stern

Cou

ch d

e di

ffus

ion

Plan

de

cisa

illem

ent

-

-

+

Hui

le a

vec

agen

ts T

A/C

OTA

Distance

Potentiel

Fig. 8-1 Diagramme de potentiel électrique de gouttelette

Gouttelette+

+

+++ +

+

+

++

- -

--

-

--

-

- ----

- -- --

--

++

+

+

++

+

+

+

+

++

Gouttelette+

+

+++ +

+

+

++

- -

--

-

--

-

- ----

- -- --

--

++

+

+

++

+

+

+

+

++Force attractive de

Van Der Waal(court)

Repulsiveelectrostatique

(Long)

Force attactive

Distance

Force attractive prevaloir

Force repulsiveRepulsive electrostatique

Force total

Force deVal Der Waal

Fig. 8-2 Interaction et bilan de forces entre gouttelette

• Les électrolytes multivalents : Les mécanismes principaux de déstabilisation sont une combinaison de la précipitation des agents tensioactifs et la coagulation de type “sweep”. La concentration réelle est donc moins que celle calculée uniquement sur le produit de solubilité et est généralement le moins des trois premiers électrolytes. Les exemples de produit chimique de ce type sont le chlorure ferrique (FeCl3) et le sulfate d’aluminium.

• Les agents tensioactifs de la charge opposée : Le mécanisme principal de déstabilisation est l’adsorption et la neutralisation de la charge. Le surdosage doit être évité pour prévenir la restabilisation de l’émulsion par l’inversion des charges électriques. Les exemples des agents tensioactifs cationiques qui peuvent être utilisés dans la déstabilisation de l’émulsion sont le chlorure N-cetylpyridinium et des sels de l’hydroxyde d’ammonium quaternaire.

Dans le cas des émulsions stabilisées par des tensioactifs nonioniques, le

mécanisme principal de déstabilisation doit se fonder sur la précipitation des tensioactifs pour former des composés insolubles. ZHU a proposé la méthode classique de la déstabilisation chimique de l’émulsion stabilisée par les tensioactifs nonioniques de l’acide gras, basée sur l’addition des tensioactifs anioniques (comme l’akylsulfate des alcools gras) et les électrolytes trivalents cationiques (comme le sulfate d’aluminium, le chlorure ferrique).

[17]

Sommaire

Toutefois, il faut noter que il n’existe pas les produits chimiques et les doses universels qui sont valides pour toute émulsion. Les types des produits chimiques efficaces, de la concentration optimal et du niveau des polluants résiduaires doivent être évalués d’abord d’échelle de laboratoire avant le design du procédé chimique complet.

En tout cas, à l’égard des design du réacteur, ils sont identiques, quel que soit le produit chimique appliqué. Ils sont pratiquement identiques au réacteur utilisé pour la coagulation/floculation dans le traitement d’eau. Les équations utilisées pour l’évaluation du réacteur et de l’agitateur, c’est-à-dire 8.1 à 8.3, sont incluses dans le logiciel.

5.0

⎟⎟⎠

⎞⎜⎜⎝

⎛=

VPG

μ {8.1}

Où G = Gradient de vitisse (t-1, normalement en sec –1) μ = Viscosité d’eau, P = Consommation énergétique d’agitateur(ML2s-3,e.g. watt) V = Volume de réacteur (L3)

Pour les turbines,

ρ53 DnNP p= {8.2}

Pour les pales tournantes ,

μρ

VCnAv

G d

2

3

= {8.3}

Où Np = Nombre de puissance ρ = Mass volumique de l’eau usée (= celle de l’eau) n = Vitesse de rotation (rev/s) (e.q. 8.2) n = Nombre de pales (e.q. 8.3) D = Impeller diameter (m) A = Surface d’une pale (m2) v = vitesse périphérique (m2) Cd = Coefficient de traînée des pales (normalement = 0.6)

M

M

Separateur

Bassin de floculationBassin de coagulation

Produit chemique

Emulsion

G = 50 s-1 G = 30 s-1 G = 20 s-1

G = 100-300 s-1

Fig. 8-3 L’example de design de coagulation-floculation

[18]

Sommaire

9. Les procédés de finition

Bien que l’huile soit récupérée de l’eau par les procédés précédents, il y a normalement de l’huile laissée dans l’eau traitée. En plus, les polluants d’autres formes généralement présents dans les eaux résiduaires huileuses, surtout les agents tensioactifs/co-tensioactifs, sont encore présents dans l’eau traitée, et contribue au haut niveau de la DOT. Ainsi l’eau traitée est normalement transmise au procédé de finition avant de se décharger au corps d’eau recevant. Deux procédés de finition largement utilisés, c’est-à-dire le traitement biologique et l’adsorption sur le charbon actif, sont réalisés.

Comme le procédé biologique est lui-même la science majeure, le calcul détaillé n’est pas inclus dans les buts et objectifs du travail de cette thèse. Seulement les données utiles sur le procédé biologique, concernant le traitement des eaux résiduaires huileuses, sont réexaminées et incluses dans le rapport principal de cette thèse (Partie 3, Chapitre 11).

Les équations du design du filtre CAG, ainsi que la capacité d’adsorption (q) de certains co-tensioactifs, sont réexaminées et incluses, c’est-à-dire,

Le temps d’opération totale avant le remplacement de lit (tT):

Quand l’isotherme (q. VS.C relation) et les données du front d’adsorption (Qa et Ha comme les fonctions de C) sont disponible ;

)( eo

abT CCQ

QqHAt−

−⋅⋅=

ρ {9.1}

Où tT = Temps d’opération totale avant le remplacement de lit H = Hauteur de lit A = Section de la colonne ρb = Mass volumique de charbon actif q = Capacité d’adsorption Qa = Capacité d’adsorption disponible dans la zone de

l’adsorption Q = Débit Co,Ce = Concentration initiale et concentration de la solution en

soluté en équilibre avec l’adsorbant

Quand les données mentionnées ci-dessus ne sont pas disponibles ;

)( eo

bT CCQ

qHAEt−

⋅⋅⋅=

ρ {9.2}

E est la saturation efficace de lit. La valeur recommandée est environs 50-95%.

[19]

Sommaire

Ha

CCoCe

H

t = 0Sa

ture

d zo

neZo

ne

d’ad

sorp

tion

CCoCe

H

t = tT

Ha

Regeneration de lit est necessaire.

Lit fixe de

charbon actif

HT

ENTREE

SORTIE

Vitesse =V

Fig. 9-1 Evolution du front d’adsorption

Temps de séjour (τ):

)/( AQH

=τ {9.3}

Perte de charge (P):

La parte de charge peut être calculée par l’équation Kozeny-Carman.

32

2)1(180ερεμ

⋅⋅⋅−

=dpgVH

P c m {9.4}

10. La méthodologie générale pour la sélection des procédés

En fin de achever le premier objectif, la méthodologie sur la sélection du procédé de traitement des eaux résiduaires huileuses est proposé, comme montré dans la table 10-1. Les filières de procédé recommandés pour le traitement d’émulsion de l’huile de coupe et l’émulsion secondaire, fondés sur les recherches GPI, sont également proposés, comme montrés dans les figures 10-1 et 10-2.

[20]

Sommaire

Tabl

e 10

-1 L

a m

étho

dolo

gie

géné

rale

pou

r la

séle

ctio

n de

s pro

cédé

s

[21]

Sommaire

Distillation heteroazeotropic

M

X

Recyclage

SORTIE

boue

ClarificateurBassin d’aerationDistillat

TI

Chauffage

Permeate

Retentat

Membrane

Pompe

Buc d’alimentation

Alimentation

Po

Pi

Pp

Echangeurde chaleur Pe

rmea

t

Retentat

Membrane

Pompe

Po

Pi

PpSORTIE(Si possible)

ENTREE

Extractant

Traitement biologique

Ultrafiltration Osmose inverseRet

enta

t

Ret

enta

t

Echangeurde chaleur

Echangeurde chaleur

Fig. 10-1 Schéma d’une filière de traitement des émulsions d’huile de coupe

ENTREE M

X

Recyclage

SORTIE

ClarificateurBassin d’aeration

Traitement biologiqueAPI

huile

boue

Charbon actif(se necessaire)

huile

Coalesceur/hydrocyclone/

coalesceur

Fig. 10-2 Schéma d’une filière de traitement des émulsions non-stabilisées

[22]

Sommaire

Développement du logiciel

L’objectif final de cette thèse est de développer un logiciel de design et calcul d’une filière de traitement adoptée à l’épuration des eaux résiduaires huileuses. Ceci vise à valoriser et appliquer le savoir-faire ainsi que les résultats importants de la recherche, en les présentant de façon conviviale. Pour réussir à ces objectifs, le logiciel, à savoir le logiciel GPI, est divisé en 4 modes majeurs, c’est-à-dire,

• Le mode de documentation électronique : fournit les connaissances de fond ainsi que une base de données utile sur la pollution d’huile et les procédés de traitement. En fait, le textbook dans la Partie 3 est transformé en directoire « e-book » du logiciel.

• Le mode de recommandation du procédé : fournit les recommandations pour restreindre la gamme des procédés faisables pour tout influents débits. Le critère de sélection est comme proposé dans les méthodologies dans la Partie 3, Chapitre 12.

• Le mode de design (calcul) : est utilisé pour évaluer l’unité procédé. Les modèles utilisés dans le calcul sont comme résumés dans la Partie 3.

• Le mode d’analyse (simulation) : permet l’utilisateurs d’intégrer tout procédé de séparation qui est inclus dans la base de données du logiciel, en but d’établir leur propre filières de procédé de traitement. Et le logiciel va simuler la filière de procédé pour prévoir l’efficacité de chaque unité.

Le logiciel est développé pour être « upgradable ». Son architecture compose de la base de données sous forme de fichier texte ordinaire et les « sub-programmes ». Pour « upgrader » le logiciel, il peut se faire facilement en ajoutant les données, comme le nom de nouveau procédé et son paramètre de lien nécessaire pour le calcul, à la base de données. Le logiciel liera automatiquement le nouveau procédé à l’interface graphique d’utilisateur. Quant au calcul du « sub-programme » du nouveau procédé, il peut être développé séparément en utilisant la langue de programmation Visual Basic. Le moyen le plus facile est de copier le code de source d’un procédé existant et modifier l’équation pour convenir au nouveau procédé. Après la compilation dans un fichier exécutable, il peut être copié pour remplacer l’ancien fichier de logiciel GPI sans réinstaller le logiciel. Ainsi le logiciel pourrait se développé davantage jusqu’à ce qu’il peut couvrir plus de recherches et procédés dans l’avenir. Les interfaces du logiciel pour chaque mode sont comme montrées dans les figures suivantes.

Cette thèse est accomplie par les supports considérables de Professeur AURELLE, mon directeur de thèse, et M. Surapol SAIPANICH, PD-G de la société Progress Technology Consultants Co.,Ltd. (Thailand). Je suis très reconnaissant pour leur conseil et encouragement.

[23]

Sommaire

Tool barMain menu

Main program

Project window

Fig 1 L’interface graphique d’utilisateur du programme “GPI”

Fig. 2 L’interface graphique d’utilisateur du mode de documentation électronique

[24]

Sommaire

Fig. 3 L’interface graphique d’utilisateur du mode de recommendation

Fig. 4 Resultat du mode de recommendation

[25]

Sommaire

étape 1 “Wastewater data input” étape 2 “Process selection”

étape 3 “Process data input” étape 4 “Result”

Fig. 5 L’interface graphique d’utilisateur du mode de design

Graphic editing area

Basic drawing tool barCalculation button

Input and result screen(displayed when double clicking at the icon)

Category selection

Process selection

Fig. 6 L’interface graphique d’utilisateur du mode d’analyse

[26]

Résumé

Nom : RACHU Prénom : Srayut

CONTRIBUTION A LA MISE AU POINT D’UN LOGICIEL DE CALCUL DE PROCEDEDS ET FILIERES DE TRAITEMENT D’EAUX RESIDUAIRES HUILEUSE 562 pages Thèse de Doctorat : Science des procédés Spécialité :Génie des Procédés et de l’Environnement I.N.S.A. Toulouse, 2005, n°

Résumé : L’objectif de cette thèse est valorisation des diverses recherches développées au sein du Laboratoire d’Ingénierie des Procédés de l’Environnement de l’INSA dans le domaine spécifique du traitment des eaux résiduaires huileuse. Cette thèse résume les diverse recherchers rélisées et permet d’établir une procédure générale de calcul associée à un “Know -how” pour chaque procédé étudié qui permet l’utilisation de ces connaissances dans la mise au point d’un logiciel spécifique de design et de calcul de fillières de traitemant des eaux huileuse. La première partie du manuscript résume les principaux résultats obtenue dans chaque thèse relative au traitement des eaux huileuses réalisées dans l’Equipe du Professeur Y.AURELLE. Le deuxième partie est relative à la généralisation des modèles et procédures de calcul de chaque type de procédés. Dans cette partie, les modèles de calucul proposés dans les diverse thèses sont comparés aux autres données bibliographiques et généralisaés. Ainsi de nouveaux modèles de calcul sont proposés dans les cas où aucun modèle n’exist où dans les cas spécifiques où les modèles existants ont besoin d’être modifiés. La troisième partie permet d’établir un “textbook” qui intègre tous les rèsultats significatifs de chaque procédé ainsi que leur procédure généralisée de dimensionnement et leurs limitations. Ce textbook prend en compte les divers procédés de deshuilage des eaux susceptibles d’être intégrés dans une filière de traitement à savoir : les deshuileurs de surface ou écrémeurs de type tambours et disques déhuileurs, les décanteurs, les coalesceurs, les flottation à air dissous, les hydrocyclones, les procédés membranaires, les procédés thermique type distillation hétéroazéotropique, les procédés biologiques ainsi que l’adsorption sur charbon actif. Enfin ce textbook propose une méthodologie générale permettant la sélection des procédés selon le type d’eaux huileuses à traiter. La dernière partie de la thèse propose le développement d’un logiciel spécifique et original de design et calcul d’une filière de traitement adaptée à l’èpuration des eaux résiduaires huileuse notamment de l’industrile pétrolière s’appuyant sur le textbook précédent.

Mots clés : Déshuiluer, Décanteur, Coalesceur, Flottation, Hydrocyclone, Membrane, Distillation, Déstabilisation, Traitement biologique, Absorption, Logiciel, Eaux huileuses Jury et date de soutenance : 16 Décembre 2005 à l'I.N.S.A. de Toulouse M. D. HADJIEV Professeur, IUT, Lorient Rapporteur M. J. ROLS Professeur, UPS, Toulouse Rapporteur M. Y. AURELLE Professeur, INSA, Toulouse Directeur de thèse M. A. LINE Professeur, INSA, Toulouse Directeur du LIPE M. R. BEN AIM Professeur Emérite, INSA Toulouse M. S. SAIPANICH PDG, Progress Technology Consultants Co., Ltd., Thaïlande M. H. ROQUES Professeur Emérite, INSA Toulouse Invité Dépôt à la Bibliothèque Universitaire en 4 exemplaires.

Nom : RACHU Prénom : Srayut

COMPUTER PROGRAM DEVELOPMENT FOR OILY WASTEWATER TREATMENT PROCESS SELECTION, DESIGN AND SIMULATION 562 pages

Thèse de Doctorat : Science des procédés

Spécialité :Génie des Procédés et de l’Environnement

I.N.S.A. Toulouse, 2005, n°

Résumé :

The aims of this thesis is to summarize the researches of GPI lab on oily wastewater treatment processes and establish general design procedure and consideration for such processes, and then, value and maximize the use of these established knowledge in the form of computer program.

The first part of the thesis contributes to reviewing the related researches in GPI lab, directed by Prof. Y. AURELLE. Significant finding of each thesis is realized.

The second part of the thesis is generalization of models. In this part, models proposed in the researches are cross-verified with other researches and generalised. New models are also proposed when there is no existing model or the existing models need to be revised.

The third part of thesis contributes to composure of a textbook that includes all significant finding from every research as well as the generalized models and their limitations of every process found in the second part. The textbook includes these processes, i.e. skimmer, decanter, coalescer, dissolved air flotation, hydrocyclone, membrane processes, thermal process, chemical process, biological treatment and carbon adsorption, as well as guideline for process selection.

The final part of the thesis is program development. The program developed in this thesis consists of 4 main features, i.e. process recommendation for the wastewater being considered, design of unit process, simulation of process train and provision of knowledge on process design in the form of e-book, based on the text book in the third part.

The textbook, the program and its source code may be available upon request. For more information, please contact Pr. Y. AURELLE or [email protected]

Mots clés : Skimmer, Decanter, Coalescer, Dissolved air flotation, Hydrocyclone, Membrane,

Distillation, Destabilization, Biological treatment, Adsorption, Program, Oily wastewater

PRODUCTION SCIENTIFIQUE

Publications dans des actes de congrès avec comité de lecture, sur texte complet

S. Rachu, Y. Aurelle, S. Saipanich .

Simulation program on hydrocarbon polluted wastewater treatment processes

16th International Congress of Chemical and Process Engineering (CHISA 2004),

Prague, Czech Republic, 22-26 August 2004

A mes Parents

et à toute ma famille

Remerciements

Je remercie sincèrement Monsieur le professeur Yves AURELLE pour

m'avoir accueilli dans son équipe, pour m'avoir fait partager ses

connaissances et ses idées innovantes et pour son assistance tout au long

de ce travail.

I would like to thank Dr. Surapol SAIPANICH for his overwhelming

support and encouragement. If not because of him, this work would never happen.

J'adresse également tous mes remerciements à :

Monsieur Dimitre HADJIEV , professeur à l'USB de Lorient, , et

Monsieur Jean Luc ROLS, professeur à l'UPS de Toulouse pour avoir accepté

de juger ce travail en faisant partie du jury,

Monsieur BEN AIM, professeur émérite à l'INSA et Monsieur Alain

LINE, professeur à l'INSA, pour avoir accepté de participer au jury de

cette thèse,

Monsieur Henry ROQUES professeur émérite à l'INSA pour sa présence

au jury de cette thèse,

I would like to thank my colleagues at Progress Technology

Consultants Co., Ltd (Thailand) for their support and their kindness. Many

thanks to Krisana KHWANPAE for his great support on programming. My

gratitude to Bowornsak WANICHKUL, Chaiyaporn PUPRASERT and Pisut

PAINMANAKUL for their priceless advice. I owe you guys a boon.

Thank to the French Embassy in Thailand for their financial support

during the first two years of my studies. Thanks to the embassy personnels,

particularly, Khun Hataiporn, Khun Wanpen for their assistance, Khun

Chanida for her superb translation.

Mes remerciements s'adressent également à toutes les personnes qui

m'ont beaucoup aidé tant à Monsieur Louis LOPEZ, Monsieur Gilles HEBRARD,

Madame Eugénie BADORC, Madame Danièle CORRADI et Madame AURELLE. Je

remercie également mes amis du labo, particulièrement, Oliver LORAIN et

Eduardo LUCENA.

Enfin, je remercie mes amis Thaïlandais de Toulouse (P’Lek, P’Tou,

P’Aote, A, Pond, Por, Pew, Golf, Mon, Aey-Alexandre, 1, Fon, Chin, Chat,

Pat, Vee, Choke, Sith, Ju, Bomb, etc.) qui m'ont assuré par leur humour,

une ambiance sympathique et permis d’effectuer de grandes fêtes pendant mon

séjour en France.

Table des matières

Contents

- i -

Contents

Page

Part I Introduction and bibliography Chapter 1 Introduction 2

Chapter 2 Objectives 4

Chapter 3 Bibliography 3.1 Categories of hydrocarbon-polluted wastewater and treatment 6

processes 3.2 STOKE’s law 7 3.3 Decanting 7 3.4 Coalescer 10

3.4.1 Thesis of AURELLE [3] 10 3.4.2 Thesis of SANCHEZ MARTINEZ [6] 11 3.4.3 Thesis of DARME [7] 12 3.4.4 Thesis of TAPANEEYANGKUL [8] 14 3.4.5 Thesis of DAMAK [9] 15 3.4.6 Thesis of MA [16] 17 3.4.7 Thesis of SRIJAROONRAT [10] 17 3.4.8 Thesis of WANICHKUL [11] 18

3.5 Flotation 18 3.5.1 Thesis of SIEM [12] 18 3.5.1 Thesis of AOUDJEHANE [13] 19 3.5.1 Thesis of DUPRE [14] 20 3.5.1 Thesis of PONASSE [15] 21

3.6 Hydrocyclone 22 3.6.1 Thesis of MA [16] 22 3.6.2 Thesis of CAZAL [17] 24 3.6.3 Thesis of SRIJAROONRAT [10] 25 3.6.4 Thesis of WANICHKUL [11] 25 3.6.5 Thesis of PUPRASERT [25 ] 26

3.7 Ultrafiltration and other membrane processes 26 3.7.1 Thesis of BELKACEM [18] 27 3.7.2 Thesis of TOULGOAT [19] 29 3.7.3 Thesis of MATAMOROS [20] 30 3.7.4 Thesis of SRIJAROONRAT [10] 32 3.7.5 Thesis of WANICHKUL [11] 33

3.8 Thermal treatment 33 3.8.1 Thesis of LUCENA[24] 33 3.8.2 Thesis of LORRAIN[23] 34 3.8.3 Thesis of WANICHKUL [11] 34

3.9 Chemical treatment 34 3.9.1 Thesis of ZHU[21] 35 3.9.2 Thesis of YANG[22] 37

3.10 Biological treatment 38 3.9.10 Thesis of WANICHKUL [11] 38

Contents

- ii -

Contents (Con’t)

Page 3.11 Skimmer 38 3.12 Application researches 39

3.12.1 Thesis of SRIJAROONRAT [10] 40 3.12.2 Thesis of WANICHKUL [11] 41

Chapter 4 Conclusion 44

Contents

- iii -

Contents Page

Part II Generalization of models for oil-water separation process design Chapter 1 Decanting

1.1 Simple decanter or API tank 48 1.2 Lamella decanter or Parallel Plate Interceptor (PPI) 49 1.3 Model verification 50 1.4 Conclusion and generalized model of decanter 51

Chapter 2 Skimmer 2.1 Drum skimmer 53 2.2 Disk skimmer 54

Chapter 3 Coalescer 3.1 Granular bed coalescer 55

3.1.1 Filtration-based model 55 3.1.2 Dimensional analysis-based model 56 3.1.3 Model verification 57 3.1.4 Conclusion and generalized model of granular bed 58

coalescer 3.1.5 Generalized model for guide coalescer 59 3.1.6 Generalized model for mixed bed coalescer 60 3.1.7 Generalized model for pressure drop of granular bed 60

coalescer and guided coalescer 3.2 Fibrous bed coalescer 62

3.2.1 Dynamic fibrous bed coalescer model 62 3.2.2 Simple fibrous bed coalescer model 62 3.2.3 Model verification 62 3.2.4 Conclusion and generalized model of fibrous bed coalescer 65 3.2.5 Generalized model of random or disorderly fibrous bed 67

coalescer 3.2.6 Generalized model for pressure drop of fibrous bed 68

coalescer

Chapter 4 Dissolved air flotation 4.1 Dissolved Air Flotation (DAF) model for oily wastewater 69

treatment 4.2 Model verification 70

4.2.1 Modification of filtration-based model 70 4.2.2 Population balance method 72

4.3 Generalized model for DAF 74 4.4 Generalized equations for pressurized water system calculation 79

Chapter 5 Hydrocyclone

5.1 Two-phase hydrocyclone 81 5.1.1 Trajectory analysis-based model 81 5.1.2 Other models 82 5.1.3 Model verification 83

Contents

- iv -

Contents (Con’t) Page

5.1.4 Conclusion and generalized model of two-phase 84 hydrocyclone

5.1.5 Generalized model for pressure drop of two-phase 85 hydrocyclone

5.2 Three-phase hydrocyclone 87 5.2.1 Model development and verification for liquid-liquid 87

section 5.2.2 Model development and verification for solid-liquid section89 5.2.3 Generalized Model for pressure drop of three-phase 90

hydrocyclone

Chapter 6 Membrane process 6.1 Ultrafiltration 93

6.1.1 Resistance model 94 6.1.2 Film theory based model 96 6.1.3 Model verification 97 6.1.4 Flux prediction for mixture of cutting oil microemulsion 102

and macroemulsion 6.1.5 Theoretical flux prediction for batch cross-flow UF process 104 6.1.6 UF efficiency 107 6.1.7 Minimum and maximum transmembrane pressure and 108

power required 6.1.8 Conclusion and generalized model of UF 110

6.2 Nanofiltration and Reverse osmosis 110

Chapter 7 Heteroazeotropic Distillation 7.1 Theoretical model 113 7.2 Model verification 116 7.3 Conclusion and generalized model of heteroazeotropic distillation 116

Contents

- v -

Content

Page

Part III Summary of researches: Oily wastewater treatment Chapter 1 Oily or hydrocarbon-polluted wastewater

1.1 Introduction 119 1.2 Hydrocarbons and oils 119

1.2.1 Hydrocarbons 119 1.2.2 Fats and oils 124 1.2.3 Petroleum and petroleum products 124 1.2.4 Oils in term of oily wastewater 126

1.3 Other compositions of oily wastewater 126 1.3.1 Surfactants 126 1.3.2 Soaps 127 1.3.3 Co-surfactants 128 1.3.4 Suspended solids 1281 1.3.5 Other components 128

1.4 Categories of oily wastewater 128 1.4.1 Classification by the nature of the continuous phase 128 1.4.2 Classification by the stability of oily wastewater 128 1.4.3 Classification by the degree of dispersion 129

1.5 Characteristics of certain oily wastewaters 132 1.6 Standards, Laws, and Regulations 133

Chapter 2 Overview for oily wastewater treatment process design 2.1 Decantation velocity and STOKE’s law 138 2.2 Application of surface chemistry for oily wastewater treatment 139

2.2.1 Liquid-gas and liquid-liquid interfaces 139 2.2.2 Liquid-solid and liquid-liquid-solid interfaces 142 2.2.3 Capillary pressure and LAPLACE’s law 146

2.3 Important parameters in oily wastewater treatment and 147 their method of analysis 2.3.1 Oil concentration 147 2.3.2 Size distribution , spectrum or granulometry 149 2.3.3 Other parameters 155

2.4 Overview of oily wastewater treatment processes 155 2.4.1 Decanter 156 2.4.2 Coalescer 156 2.4.3 Hydrocyclone 156 2.4.4 Dissolved air flotation (DAF) 157 2.4.5 Skimmer 157 2.4.6 Membrane processes 157 2.4.7 Thermal processes 157 2.4.8 Chemical process 158 2.4.9 Finishing processes 158

Contents

- vi -

Content (Con’t)

Page

2.5 Determination of degree of treatment 158 2.5.1 Overall degree of treatment 158 2.5.2 Degree of treatment of each process 158

Chapter 3 Oil skimmer 3.1 General 162 3.2 Oil drum skimmer 164

3.2.1 Working principles 164 3.2.2 Design calculation and design consideration 169

3.3 Oil disc skimmer 171 3.3.1 Working principles 171 3.3.2 Design calculation and design consideration 173

3.4 Productivity comparison between drum and disc skimmer 173 3.5 Advantage and disadvantage of drum and disc skimmer 174

Chapter 4 Decanting

4.1 General 176 4.2 Simple Decanter or API tank 177

4.2.1 Working principles 177 4.2.2 Design calculation 179 4.2.3 Design considerations 182 4.2.4 Construction of simple decanters 183

4.3 Compact decanter 186 4.3.1 Working principles 186 4.3.2 Design calculation 190 4.3.3 Design considerations 192 4.3.4 Variations, advantage and disadvantage of compact 193

decanters

Chapter 5 Coalescer 5.1 General 195 5.2 Granular bed coalescer 195

5.2.1 Working principles 195 5.2.2 Design calculation 209 5.2.3 Design consideration 211 5.2.4 Variations, advantage and disadvantage of granular 212

bed coalescer 5.3 Guide coalescer 213

5.3.1 Working principles 213 5.3.2 Design calculation 215 5.3.3 Design consideration 215

5.4 Fibrous Bed coalescer 216 5.4.1 Working principles 216 5.4.2 Design calculation 226 5.4.3 Design consideration 227 5.4.4 Variations, advantage and disadvantage of fibrous bed 229

coalescer

Contents

- vii -

Content (Con’t)

Page

Chapter 6 Dissolved air flotation 6.1 General 232 6.2 Working principles 233

6.2.1 Filter based model 233 6.2.2 Population balance model 238 6.2.3 Generalized model of DAF from combination of 240

filtration based model and population balance model 6.2.4 Influent parameters 241

6.3 Design calculation 244 6.4 Design consideration and construction of DAF reactor 254 6.5 Pressurized water system or saturator 262

6.5.1 Working principle and design calculation 262 6.5.2 Type of saturator and injection valve 267

6.6 Variations, advantage and disadvantage of DAF 270

Chapter 7 Hydrocyclone 7.1 General 272 7.2 Two-phase hydrocyclone 273

7.2.1 Working principles 273 7.2.2 Design calculation 289 7.2.3 Design considerations 292 7.2.4 Variations, advantage and disadvantage of 295

hydrocy clone 7.3 Three-phase hydrocyclone 295

7.3.1 Working principles 295 7.3.2 Design calculation and design consideration 299 7.3.3 Advantage and disadvantage of three-phase 299

hydrocyclone

Chapter 8 Membrane process 8.1 General 300

8.1.1 Classification of membrane processes 300 8.1.2 Mode of operation of membrane processes 302 8.1.3 Membrane structure 302 8.1.4 Membrane material 303 8.1.5 Membrane module type 306

8.2 Ultrafiltration (UF) 311 8.2.1 Basic knowledge and working principles 311 8.2.2 UF process design for oily wastewater treatment 323 8.2.3 Design consideration and significant findings from 338

GPI’s researches 8.3 Microfiltration (MF) 349

8.3.1 Basic knowledge and working principles 349 8.3.2 Significant findings on MF for oily wastewater 350

treatment from GPI researches

Contents

- viii -

Content (Con’t)

Page

8.4 Reverse osmosis (RO) 352 8.4.1 Basic knowledge and working principles 352 8.4.2 Significant findings on RO for oily wastewater 353

treatment from GPI’s researches 8.5 Nanofiltration (NF) 356

8.5.1 Basic knowledge and working principles 356 8.5.2 Significant findings on NF for oily wastewater 356

treatment from GPI’s researches 8.6 Comparison of membrane processes on emulsion treatment 359

Chapter 9 Thermal processes 9.1 General 362 9.2 Basic knowledge on distillation 362

9.2.1 Basic knowledge on vapor/liquid equilibrium of 362 mixtures

9.2.2 Equilibrium of various mixtures 365 9.3 Heteroazeotropic distillation of oily wastewater 367

9.3.1 Working principles 367 9.3.2 Raoult’s law and Dalton’s law 368 9.3.3 Calculation of azeotropic temperature and composition, 369

dew curve and bubble curve. 9.3.4 Application of heteroazeotropic distillation on 371

treatment of inverse emulsion or concentrated oily wastewater

9.3.5 Application of heteroazeotropic distillation on 374 treatment of the wastes polluted by trace hydrocarbons: Steam stripping

9.3.6 Design calculation and design considerations 374 9.4 Classical or conventional distillation of oily wastewater 376

9.4.1 Working principles 376 9.4.2 Significant findings on classical distillation for oily 376

wastewater treatment from GPI’s researches Chapter 10 Chemical treatment processes

10.1 General 380 10.2 Basic knowledge 381

10.2.1 Stability of the emulsion 381 10.2.2 Surface-active agents 381 10.2.3 Important properties to obtain stable emulsion 383 10.2.4 Destabilization of emulsion 385

10.3 Process design 391 10.3.1 Rapid mixing 392 10.3.2 Flocculator 393 10.4 Design consideration 395

Contents

- ix -

Chapter 11 Finishing processes 11.1 General 397 11.2 Biological treatment 397

11.2.1 Basic knowledge 397 11.2.2 Design consideration and significant finding on 404

biological treatment for oily wastewater from GPI’s researches

11.3 Adsorption 405 11.3.1 Activated carbon (AC) 406 11.3.2 Basic knowledge 407 11.3.3 Design calculation 412

Chapter 12 Guideline for treatment process selection and examples of treatment processes for certain oily wastewaters 12.1 Guideline for treatment process selection 414

12.1.1 Oil film 414 12.1.2 Primary emulsion 416 12.1.3 Secondary emulsion 417 12.1.4 Macroemulsion and microemulsion 418 12.1.5 Concentrated oily wastewater or refinery slops 418

12.2 Examples of treatment processes for certain oily wastewaters 419 12.2.1 Treatment of cutting oil emulsion 419 12.2.2 Treatment of non-stabilized secondary emulsion 420

Contents

- x -

Contents

Page

Part IV Computer program development Chapter 1 Program overview

1.1 Introduction 423 1.2 Conceptual design of the program 423

1.2.1 E-book mode 424 1.2.2 Recommendation mode 426 1.2.3 Design mode 427 1.2.4 Analysis mode 428

1.3 Development tools 432 1.3.1 Main development software package 432 1.3.2 Special graphic user interface (GUI) component 433 1.3.3 The third party software 433

1.4 Program architecture 434 1.4.1 Forms 434 1.4.2 Modules 437 1.4.3 Modules 437 1.4.4 Class modules 437 1.4.5 Add-in project 437

1.5 Program development 438

Chapter 2 Program reference and user manual 2.1 Overview of the program 439

2.1.1 Main program 439 2.1.2 Project window 441 2.1.3 E-books worksheet 441 2.1.4 Recommend worksheet 442 2.1.5 Design worksheet 444 2.1.6 Analysis mode 445 2.1.7 Warning dialog box 447

2.2 Program capability 447 2.3 Program limitation 447 2.4 System requirement 449 2.5 User instruction 449

2.5.1 Program Installation 449 2.5.2 Starting the program 450 2.5.3 Using E-book mode 450 2.5.4 Using Recommend mode 450 2.5.5 Using Design mode 452 2.5.6 Using Analysis mode 454 2.5.7 Printing and file operation 457

2.6 Upgrading procedure and recommendation for further 458 development 2.6.1 Upgrading procedure 458 2.6.2 Recommendation for further development 460

Contents

- xi -

Contents

Page

Chapter 3 Process references 1) Drum skimmer 463 2) Disk skimmer 465 3) Simple decanter 467 4) Compact decanter 470 5) Customized decanter 473 6) Granular bed coalescer 476 7) Brush type bed coalescer 480 8) Dynamic fibrous bed coalescer 484 9) Metal wool bed coalescer 488 10) Dissolved air flotation 492 11) Two-phase hydrocyclone 498 12) Three-phase hydrocyclone 502 13) Ultrafiltration 506 14) Reverse osmosis 510 15) Heteroazeotropic distillation 513 16) Stripping 515 17) Chemical destabilization, coagulation-flocculation 517 18) Biological treatment 520 19) GAC filter 522 20) Customized concentrator 526 21) Customized oil separator 528 22) Customized inline concentrator 530 23) Inlet 532 24) Outlet 533 25) Flow merge 534 26) Flow split 536

Contents

- xii -

Contents

Page

General conclusion 537

Reference 540

Annexe 546

Nomenclatures

Nomenclature

xiii

Nomenclature

a Constant for population balance equation A Flow area (Cross sectional) area of decanter L2

A Cross section area of flotation column L2

A Flow area in membrane module (= HW) L2

C Considered or required or design oil concentration ML-3

C’g The 1st (or lower or pseudo) gel concentration in film model, used at the lower range of concentration before inflection point in flux vs. Log (Concentration) curve (in mass/ volume or volume/volume)

ML-3

Ca Capillary number = μo V/γo Cg Gel concentration in film model (in mass/ volume or

volume/volume) ML-3

Co Initial oil concentration of feed or influent wastewater ML-3

Cod Inlet concentration of the droplet diameter “d” M/L3

Cod Inlet concentration of the droplet diameter “d”, dilution effect from addition of pressurized water is not included.

M/L3

Conc(Air) Concentration of dissolved air in pressurized water M/L3

Conc(O2) Concentration of dissolved oxygen in pressurized water M/L3

Cy50 Cyclone number of d50% d Diameter of dispersed phase, in our case, oil L D Water depth or L Diameter of the skimmer or L Diameter of coalescer bed, such as diameter of brush or L Nominal diameter of 3-phases hydrocyclone (the largest

diameter of the cyclone) or L

Hydraulic diameter of flow channel in membrane module (Channel between membrane surface and membrane module wall)

L

db Average diameter of air bubble L dc Cut size of the decanter or API tank L dF Diameter of fiber in fibrous-bed coalescer L Di Diameter of inlet port of hydrocyclone L Dn Nominal diameter of hydrocyclone. ( = diameter of inlet of

lower conical part for Thew type hydrocyclone) L

dp Diameter of collector or coalescer bed material, such as resin L dxx% , dxx Diameter of droplet corresponding to removal efficiency of

“xx”%, such as d75%, etc. d100% or dc stands for cut size M

e Surface roughness of flow channel L f Friction factor of Darcy-Weisbach’s equation g Gravitational acceleration L/T2

G Turbulent intensity H Travelling or rising distance of oil drop, depending on

configuration of the decanter. Exp. For PPI, H = distance between plates or

L

Height of bed, or bed depth of coalescer or L Height of flow channel in membrane module or L Height of contact (or effective) zone of flotation column. If

both pressurized water and wastewater are fed at the bottom of the column, the contact zone is equal to column height.

L

Href H at reference operating condition of DAF reactor L Hreq H at design or required operating condition of DAF reactor L I Immersion depth of the disk in liquid or L

Nomenclature

xiv

The number of bubbles attached to the agglomerate J Permeate flux of membrane L/T K Boltzman constant (1.38*10-23) k Constant in film model L Length of inserted plates or interceptor surface or L Length of lower conical part for Thew type hydrocyclone or L Length of flow channel in membrane module L L3 Total length of RIETEMA part in 3-phases hydrocyclone,

includes conical and straight section. L

L4 Length of lower straight section of RIETEMA part in 3-phases hydrocyclone, which locate oil and solid outlet ports.

L

MW Molecular weight g/mol N Numbers of inserted plates or Rotational speed or T-1, such as rev/s

(not rad/s) Rotational speed of fibrous bed or T-1, such as rev/s

(not rad/s) Total number of bubbles n0, n1 The number of oil droplet attached by 0, 1, … air bubles P Oil productivity of the skimmer or L3/T Absolute pressure of the pressurized water system or MT-2 L-1

Pressure drop or LT-2M-1

Pressure LT-2M-1

p Partial pressure LT-2M-1

Pcap Capillary pressure LT-2M-1

Psat Vapor pressure (some references use “Πθb”) LT-2M-1

Pt Transmembrane pressure LT-2M-1

Q Wastewater flowrate or L3/T Inlet flow or L3/T Recirculation flowrate of membrane L3T-1

Qpw Pressurized water flowrate L3/T Qt total water flowrate (= Qpw + Qwastewater) L3/T R Universal gas constant (8.314 Pa. m3 / (mol.K)) r Radius of membrane pore L R Distance in radial axis of hydrocyclone L R’m Modified membrane resistance (= Rm+Rf) MT-1 L-2

Rd Distant from center line of hydrocyclone in radial axis to particle “d” considered

L

rd Radius of oil droplet L Re Reynolds number (ρVD/μ) Rf Split ratio (= Qoverflow/Q) Rf Fouling resistance MT-1 L-2

Rg Gel (or polarization) resistance MT-1 L-2

Rm Membrane resistance MT-1 L-2

Rz Distant from center line of hydrocyclone in radial axis to the wall of hydrocyclone at the axial distance “Z”

L

S Bottom projection area of the tank L2

Sp Inserted plate area L2

t Time T T Absolute temperature U Radial velocity of particle or oil drop in hydrocyclone L/T Ub Rising velocity of bubble L/T Ud Rising velocity of the droplet diameter “d” L/T Ud Rising velocity of oil droplet L/T V Tangential velocity or tip speed of the skimmer L/T

Nomenclature

xv

V Empty bed velocity or L/T Empty bed velocity of DAF (based on sum of pressurized water

and wastewater flow) or L/T

Tangential velocity of particle or oil drop in hydrocyclone or L/T Recirculation velocity in cross-flow membrane process L/T Vol Volume L3

Vr Relative velocity between bubble and oil drop L/T W Axial velocity of particle or oil drop in hydrocyclone or L/T Width of flow channel in membrane module L x Molar fraction in liquid (water) Mol/mol y Molar fraction in vapor Mol/mol Z Distance in axial axis of hydrocyclone L ΔP Pressure drop, (in m of water, for Kozeny-Carman’s equation) Or pressure drop in bar, for hydrocyclones ΔPo Pressure drop across inlet and overflow port Bar ΔPoil Pressure drop across inlet and oil outlet port Bar ΔPSS Pressure drop across inlet and suspended solids outlet port Bar ΔPu Pressure drop across inlet and underflow port Bar ΔPwater Pressure drop across inlet and water outlet port Bar

Greek Letter

Φ Air flowrate in flotation column L3/T Π Vapor pressure (some references use “Psat”.) LT-2M-1

α Probability of collision or Exponent of recirculation velocity in gel resistance equation α.ηexp Corrected experimental removal efficiency factor of the tank

for the droplet diameter “d”

α, α3φ, αThew Correction factor for inlet velocity of hydrocyclone β Conical angle of lower part of hydrocyclone or Exponent of recirculation velocity in film model of

membrane

β0, β1, … , βi Adhesion efficiency between bubbles and oil drop/ bubble agglomerate for population balance equation

ε Porosity or void ratio φ Constant in gel resistance equation γo Superficial tension of oil M/T2

γo/w Interfacial tension between oil and water M/T2

γo/w Interfacial tension between oil and water MT-2

ηoverall Overall efficiency of pump κ Collision rate constant for population balance equation T-1

κ2 Modified collision rate constant for population balance equation (κ = κ2Φ)

T-2

κ2,ref Modified collision rate constant at reference condition T-2

κ2,req Modified collision rate constant at design or required condition

T-2

μ Dynamic viscosity ML-1T-1

μC Dynamic viscosity of continuous phase, in our case, water L2/T μd Dynamic viscosity of dispersed phase, in our case, oil L2/T μo Dynamic viscosity of oil M/(L.T) νo Kinematic viscosity of oil L2/T θo/w Contact angle between oil and water surface (180o means the

oil drop in perfectly sphere.)

Nomenclature

xvi

ρ Density ML-3

ρair Density of air at required operating condition M/L3

ρc Density of continuous phases M/L3

ρm Density of emulsion M/L3

Δρ Difference between density of dispersed and continuous phases

M/L3

τ Retention time T ηd Removal efficiency of the tank for the droplet diameter “d” % ηd,ref Removal efficiency of DAF process for the droplet diameter

“d” at the reference retention time (25 minutes) %

ηDiff Efficiency factor from diffusion ηInt Efficiency factor from direct interception ηSed Efficiency factor from sedimentation ηSed Efficiency factor from sedimentation ηt Total Removal efficiency % ηtheo Theoretical removal efficiency factor of the tank for the

droplet diameter “d”

?A, ?BB Subscript indicating component A and B respectively ?H Subscript indicating heteroazeotropic point ?mac Subscript indicating macroemulsion ?mic Subscript indicating microemulsion ?mix Subscript indicating mixture ?o Subscript indicating initial condition ?ref Subscript indicating reference condition ?θb Superscript indicating boiling temperature

Part I Introduction and bibliography


I-i

Contents

Page

Part I Introduction and bibliography Chapter 1 Introduction I-2

Chapter 2 Objectives I-4

Chapter 3 Bibliography 3.1 Categories of hydrocarbon-polluted wastewater and treatment I-6

processes 3.2 STOKES law I-7 3.3 Decanting I-7 3.4 Coalescer I-10

3.4.1 Thesis of AURELLE [3] I-10 3.4.2 Thesis of SANCHEZ MARTINEZ [6] I-11 3.4.3 Thesis of DARME [7] I-12 3.4.4 Thesis of TAPANEEYANGKUL [8] I-14 3.4.5 Thesis of DAMAK [9] I-15 3.4.6 Thesis of MA [16] I-17 3.4.7 Thesis of SRIJAROONRAT [10] I-17 3.4.8 Thesis of WANICHKUL [11] I-18

3.5 Flotation I-18 3.5.1 Thesis of SIEM [12] I-18 3.5.1 Thesis of AOUDJEHANE [13] I-19 3.5.1 Thesis of DUPRE [14] I-20 3.5.1 Thesis of PONASSE [15] I-21

3.6 Hydrocyclone I-22 3.6.1 Thesis of MA [16] I-22 3.6.2 Thesis of CAZAL [17] I-24 3.6.3 Thesis of SRIJAROONRAT [10] I-25 3.6.4 Thesis of WANICHKUL [11] I-25 3.6.5 Thesis of PUPRASERT [25 ] I-26

3.7 Ultrafiltration and other membrane processes I-26 3.7.1 Thesis of BELKACEM [18] I-27 3.7.2 Thesis of TOULGOAT [19] I-29 3.7.3 Thesis of MATAMOROS [20] I-30 3.7.4 Thesis of SRIJAROONRAT [10] I-32 3.7.5 Thesis of WANICHKUL [11] I-33

3.8 Thermal treatment I-33 3.8.1 Thesis of LUCENA[24] I-33 3.8.2 Thesis of LORRAIN[23] I-34 3.8.3 Thesis of WANICHKUL [11] I-34

3.9 Chemical treatment I-34 3.9.1 Thesis of ZHU[21] I-35 3.9.2 Thesis of YANG[22] I-37

3.10 Biological treatment I-38 3.9.10 Thesis of WANICHKUL [11] I-38


I-ii

Contents (Con’t)

Page 3.11 Skimmer I-38 3.12 Application researches I-39

3.12.1 Thesis of SRIJAROONRAT [10] I-40 3.12.2 Thesis of WANICHKUL [11] I-41

Chapter 4 Conclusion I-44


I-iii

Table

Page Table 3.1 Summary of characteristics of wastewaters and sludges for “on site” I-24

experiment Table 3.2 Summary of characteristics of synthetic wastewaters I-24 Table 3.3 Membranes test by MATAMOROS I-31

Figure

Page Fig. 1-1 Summary of researches of Prof. AURELLE on hydrocarbon-polluted I-3

wastewater treatment Fig. 3.1 Schematic of decante I-8 Fig. 3.2 Schematic of Phase inversion coalescer I-15 Fig. 3.3 Three- phase hydrocyclone I-23 Fig. 3.4 Hydrocyclone tested by WANICHKUL I-26 Fig. 3.5 Ultrafiltration models used by BELKACEM I-29 Fig. 3.6 Treatment processes for macro- and microemulsion, recommended I-32

by MATAMORS


1 I-1


Chapter 1 Introduction

2 I-2

Chapter 1 Introduction

Water pollution is one of the most important environmental problems. Wastewater from agriculture and industrial processes, as well as domestic wastewater, is the main pollutant source that causes water pollution problem. There are many substances that can deteriorate water quality, thus classified as water pollutants, such as, organic matter from domestic wastewater, chemicals from industrial wastewater. Some valuable substances, such as sugar, flour, oil, will become major pollutants when discharged into water bodies.

Among various kinds of pollutants, hydrocarbon, or simply called oil, is one of the most severe pollutants because of its intrinsic properties. Small amount of hydrocarbon can spread over wide area of water surface and affect the oxygen transfer, so cause adverse effect to marine or water ecology. Furthermore, the hydrocarbon contributes to very high biochemical oxygen demand and is relatively difficult for biodegradation, which is the major natural self-purification process. So it can last relatively long in the water and causes long -term effect.

Our laboratory has researched for various treatment processes that cover various types of wastewater polluted by hydrocarbons. In the few decades of researches, many theses had been accomplished as shown in Fig. 1.1. Among these, many innovations had been created and some had been patented and commercialized. Some researches are the key steps to understand or improve the treatment efficiency and process design. However, because there are many kinds of oily wastewater, as well as, there are many kinds of treatment processes. Moreover, treatment efficiency of each treatment process will vary with characteristic of wastewater. Then, it may cause some difficulties in selecting or designing appropriate process train that can deal with the wastewater considered as well as predicting effluent quality accurately.

According to the difficulty stated above, this thesis had been initiated to provide the solution and tool about how to select and optimize the process or processes train to treat the specified wastewater to meet required effluent standard, as well as providing details about hydrocarbon polluted wastewater and each treatment process. It can be, also, applied for designing the process to recover valuable hydrocarbons from hydrocarbon/water mixtures in some industries, such as perfume or pharmaceutical industries.


3 I-3

Fi

g. 1

-1 S

umm

ary

of re

sear

ches

of P

rof.

AURE

LLE

on h

ydro

carb

on-p

ollu

ted

was

tew

ater

trea

tmen

t

Hyd

roca

rbon

-po

llute

d or

oily

was

tew

ater

treat

men

t

Dec

ante

r“S

pira

loil”

dec

ante

rC

HER

ID [4

], 19

86

Skim

mer

“Dru

m sk

imm

er”

&“d

isk

skim

mer

”TH

AN

GTO

NG

TAW

I [5]

, 198

8

Coa

lesc

er“G

ranu

lar b

ed c

oale

scer

”A

UR

ELLE

[3],

1980

“Gra

nula

r bed

coa

lesc

er fo

r s

tabi

lized

em

ulsi

on”

DA

RM

E [6

], 19

83

“Dyn

amic

fibr

ous b

ed c

oale

scer

”TA

PAN

EEY

AN

GK

UL

[8],

1989

“Pul

sed

gran

ular

bed

coa

lesc

er”

&“P

hase

inve

rsio

n co

ales

cer”

DA

MA

K [9

], 19

92

Flot

atio

n“F

lota

tion

for o

il-w

ater

sep

arat

ion”

SIEM

[12]

, 198

3

“App

licat

ion

of fl

otat

ion

on o

ily w

aste

wat

er tr

eatm

ent”

AO

UJE

HA

NE

[13]

, 198

6

“Mat

hem

atic

s mod

el o

f DA

F”D

UPR

E [1

4], 1

995

“Dee

p sh

aft D

AF”

PON

ASS

E [1

5], 1

997

Hyd

rocy

clon

e“2

& 3

-pha

se h

ydro

cycl

one”

MA

[16]

, 199

3

“2-p

hase

hyd

rocy

clon

e fo

r sto

rmw

ater

trea

tmen

t”C

AZA

L [1

7], 1

996

“Hyd

rocy

clon

e w

ith g

rit p

ot”

&“C

ombi

natio

n C

oagu

latio

n-D

AF-

Hyd

rocy

clon

e” se

para

tor

PUPR

ASE

RT

[25]

, 200

4

Ultr

afilt

ratio

n &

mem

bran

e pr

oces

ses

“Ultr

afilt

ratio

n fo

r cut

ting

oil

em

ulsi

on tr

eatm

ent”

BEL

KA

CEM

[18]

, 199

5

“Ultr

afilt

ratio

n fo

r the

rmal

em

ulsi

on tr

eatm

ent”

TOU

LGO

AT

[19]

, 199

6

“Mem

bran

e pr

oces

ses f

or c

uttin

g o

il em

ulsi

on tr

eatm

ent”

MA

TAM

OR

OS

[20]

, 199

7

Ther

mal

tre

atm

ent

“Cry

stal

lizat

ion

for o

il-w

ater

sepa

ratio

n”LO

RR

AIN

[23]

, 200

2

“Het

eroa

zeot

ropi

c di

still

atio

n” L

UC

ENA

[24]

, 200

4

Che

mic

al

treat

men

t“D

esta

bilis

atio

n of

em

ulsi

on”

ZHU

[21]

, 199

0“L

ow-p

ollu

tion

emul

sion

”Y

AN

G [2

2], 1

993

“Gui

ded

& M

ixed

bed

coa

lesc

er”

SAN

CH

EZ M

AR

TIN

EZ [5

], 19

82

App

licat

ion

“Tre

atm

ent o

f s

tabi

lized

em

ulsi

on”

WA

NIC

HK

UL

[

11]

, 20

00

“Tre

atm

ent o

f non

- s

tabi

lized

em

ulsi

on”

SRIJ

AR

OO

NR

AT

[10]

, 1

998

Chapter 2 Objectives

4 I-4

Chapter 2 Objectives

The objectives of this study are as described below;

1. To review the treatment technologies for oily wastewater or wastewater polluted by hydrocarbon from doctorate researches, directed by Professor Yves AURELLE in LIPE-GPI, from the beginning to the present.

To design and select the suitable treatment process or processes, firstly, one has to know of all applicable processes and understand their working mechanisms. So it becomes the first objective of this study to review these data. However, as previously described in Chapter 1, there are many processes for hydrocarbon-polluted wastewater treatment. For example, decanter can be subdivided into many types, such as simple API decanter, decanter with plate settler, etc. So there are many researches all over the world that had been conducted to study the mechanism or working principle of these processes and, then, to predict their efficiencies.

Among these researchers, Professor Yves Aurelle, with his team of researchers at GPI, INSA Toulouse, has dedicated his studies to oily and hydrocarbon-polluted wastewater treatment for a few decades. Through many years of study, the researches, conducted and led by him, cover relatively the whole processes of hydrocarbon-polluted wastewater treatment. So, to value and make ultimate use of these researches, the processes taken into account in this thesis are exclusive based on the researches of Professor Yves Aurelle with additional supports in some parts from related literatures for completeness of this study.

2. To generalize or propose the model of each related process.

Each treatment process has its own variations, adapted to improve working efficiency or to suit some certain circumstances. Thus results or models from the researches, which intended to study in detail of these variations can not be used or extrapolated beyond their experimental operating conditions. In order to provide the solution to the widest range, if not the entire range, of hydrocarbon-polluted wastewater, we set the 2nd objective to generalize or, if case arises, propose the model or models that allow the prediction of efficiency of each treatment process over the determined range of hydrocarbon-polluted wastewater.

3. To compose the textbook on hydrocarbon-polluted wastewater or oily wastewater treatment process, based on the result from the 1st and 2nd objective.

As previously described that there are a lot of researches dedicated to specific aspects of various treatment processes, it is very interesting to combine these researches together to get the whole picture. After we had reviewed and generalized all related researches as described in the 1st and 2nd objective, we set out to work on our 3rd objective of this thesis to compose the textbook. This textbook will feature the following topics;

• Background knowledge on oily and hydrocarbon-polluted wastewater

• Treatment technology overview

• Chapters devoted to details of each treatment process

• Guidance for selection and combination of processes

If we could compose it well as we intended. This textbook would provide readers sufficient practical data to understand properties oily wastewater, working mechanisms of


5 I-5

treatment processes and key parameters affected their designs and operations, that can lead to proper design and selection of treatment processes or creating their own variation of process that suit their own situation.

4. To develop the prototype of program for the design, comparison and simulation of hydrocarbon-polluted wastewater treatment processes.

At the present, computer comes to play major role in every field and become standard equipment in almost every household, office and academic institute. Because of its powerful logical and mathematical calculation, as well as its presentation and interaction capability, it is very interesting to use computer in the field of hydrocarbon-polluted wastewater treatment. Up till now, there are many commercial softwares on industrial and wastewater treatment process calculation. Anyway, those programs are not specifically designed to deal with hydrocarbon-polluted wastewater treatment. Furthermore, the commercial softwares are normally developed for expert users, so they do not provide much basic data, thus, render it difficult for non-expert users to use efficiently. Besides, they normally do not provide any data for decision supporting, for example they can not compare the efficiency between various processes or recommend the feasible processes for considered wastewater.

So for our 4th objective, we intend to develop the prototype of program for design, comparison and simulation of hydrocarbon-polluted wastewater treatment processes. The program will feature;

• E-book: provides background knowledge and useful database about the oil pollution and the treatment processes,

• Process recommendation part: provides recommendation or narrows the range of feasible processes for any input influent,

• Design (or calculation) part: used for sizing the process unit,

• Simulation part: allows users to integrate any separation processes, included in the program database, to build their own treatment process train. And the program will simulate the process train to forecast the efficiency of each unit.

The treatment processes, which can be calculated or simulated by the program built-in database, will then be based mainly upon the researches reviewed in the 1st to 3rd objective. However, the source code of the program will be available upon request to allow upgrading to include more processes in the future.


6 I-6

Chapter 3 Bibliography

3.1 Categories of hydrocarbon-polluted wastewater and treatment processes

As hydrocarbon or oil requires a great amount of oxygen or oxidizing agent to oxidize, moreover, the hydrocarbon is relatively difficult to biodegrade, thus, it becomes clear that the use of the biological treatment with high-concentration oily wastewater is not the economical alternative. Besides there are possibilities to reuse or recover the hydrocarbons in the wastewater. Then almost all of treatment processes that have been studied in our laboratory are based on separation, both physical and physico-chemical, techniques in order to separate oil from water.

Thus, it is more suitable to categorize the oily wastewater by its physical properties. Among these properties, the degree of dispersion of oil phase in water or the size of oil droplet is the key parameter that plays an important role in separation process selection. So we will categorize the oily wastewater into 4 groups, in accordance with its droplet size, i.e.,

• Hydrocarbon in form of film or layer on the surface of wastewater, or hydrocarbon in form of big oil drops in the wastewater

• Emulsion without surfactants, by the word “emulsion”, it can be easily described as the water with very fine dispersed oil drops

• Emulsion with surfactants • Dissolved hydrocarbon

For treatment processes, each process has its own characteristic or limitation so it can be used to separate some certain ranges of oil droplet. So each group of the oily wastewater may require certain process or train of processes to separate the oil from the water to the accepted degree. The treatment processes, studied by Professor AURELLE’s researchers, covered the entire range of the oily wastewater stated above and can be summarized as follow;

• Decanting • Skimmer • Coaleser • Flotation • Hydrocyclone • Ultrafiltration • Distillation • Biological treatment

Moreover, there are researches on chemical treatment, which relates to “breaking” the emulsion to allow the micro droplet to coalesce and make it possible to separate by the processes stated above. There, also, are some researches contributed to formulation of environmental-friendly oil product, which can be easily treated and still have the same essential working properties as the existing product’s.

Furthermore, some researches can be extensively used to solve the problems of some special types of oily waste, such as slop (viscous mixture between crude oil and water) or inverse emulsion (fine drops of water dispersed in oil)


7 I-7

Details of wastewater characteristics and treatment processes are thoroughly described in Part 3. So, this part emphasizes on reviewing of related researches, categorized by process, about their scope of work and significant finding, as described in next sections.

3.2 STOKES law

It is important to mention about STOKES law (eq. 3.1), because almost all of separation processes considered here are based upon modification of parameters in this equation. The STOKES equation is the relation between rising (or settling) velocity of spherical object (in our case, droplet of dispersed phase) with very small Reynolds number (10-4 to 1) and properties of dispersed phase and continuous phase.

c

EdgV

μρ18

2⋅⋅Δ= {3.1}

Where V = rising or settling velocity (based on density of the 2 phases) Δρ = Difference between density of dispersed phase and continuous

phase dE = Diameter of dispersed phase μC = Dynamic viscosity of continuous phase

In case of oily wastewater, the dispersed phase is hydrocarbon or oil and the continuous phase is water. Because the density of hydrocarbons in our wastewater is normally lower than water’s. It is prone to rise to surface of the water. Even though the STOKES equation is valid only for certain flow regimes, the equation covers the range of flow regime normally encountered in wastewater problem. It can be used to explain important phenomena or applied to many types of processes, even a bit beyond its valid regime, with satisfactory result. There are few modifications of STOKES law brought about by applying some correction factors into basic STOKES law. But the core equation usually remains the same as shown in eq. 3.1.

Form eq. 3.1, one can increase the rising velocity of the oil drop by modifying 4 variables properly. The separation processes, which are based on the results of STOKES law or modification of the variables in STOKES law, are decanting, coalescer, flotation process and hydrocyclone or various types of centrifugal process. The researches on each process can be summarized as follows.

3.3 Decanting

Decanting (or sedimentation) is the simplest separation process. It makes use of gravity force and density difference between oil and water to separate them. Rising velocity of oil or hydrocarbon drop in wastewater depends on its size, density, viscosity of water and gravity constant, as described by STOKES equation. As shown in fig. 3.1, when the homogeneous oil/water mixture flows uniformly pass through a control volume, some big oil drops will rise to the water surface and can be retained, then, separated from the water by means of proper equipment, such as overflow weir or skimmer. The small oil drop that can not reach the water surface will be entrained with the water and exit the control volume without separation.


8 I-8

VU

Q

d = cut size

d < cut size

d > cut sizeInfluent Effluent

Fig. 3.1 Schematic of decanter

When the size of the decanter (or settling tank) and the distribution of oil droplets in the wastewater are known, we can calculate the separation efficiency of the tank by comparing the time required for each size of oil droplet to reach the surface of the tank within hydraulic retention time of the tank. The time required for oil droplet to reach the surface can be calculated from STOKES equation and vertical travelling distance of the droplet. If the droplet can reach the surface before the wastewater will flow off the tank, we can say that the droplet can be separated by that decanter. American Petroleum Institute had recommended the geometry of the decanter, generally known as API tank. This type of decanter has been widely used.

Because all variables in STOKES equation are practically unchanged during the separation process by decanter. Then, the efficiency of decanter can be enhanced only by reducing the vertical travelling distance of oil droplet to the decanting surface. This fact leads to the modification of simple decanter by inserting submerged plates into the tank. These plates will act as oil interceptor. Instead of rising up to the water surface, oil droplets that reach the surfaces of these plates are intercepted, collected, and then separated from the tank. It can be said that these insertions reduce the vertical travelling time of oil drop without the reduction of hydraulic retention time.

Theoretical efficiency of plate-inserted decanter, known as lamella decanter, parallel plate interceptor (PPI), etc., can be calculated using the same equation as for API tank with a little modification on decanting area. However, the selection of the shape and installation of these plates are the state-of-art processes. The spacing between the plates should be as small as possible to minimize the size of decanter. However, the flow of water and the collected oil between the plates shall be taken into account to minimize shear force, thus minimize the effect of surface distortion or “snap-off”.

Thesis of CHERID [4] is contributed to study on the compact lamella decanter, known as “SPIRALOIL”, that shows far greater efficiency compared to the simple decanter of the same size.

The research consisted of,

• Study of interaction between oil drop and surface of lamella plate, and influence of wettability of lamella material, inclination of lamella plate and characteristic of wastewater to decanter operation

• Model development for lamella decanter


9 I-9

Experimental procedure

The experiment of phenomena in the lamella decanter was conducted in transparent model. Interaction between oil and surface of the lamella plate, as well as the influence of wettability of plate and inclination of plate, could be observed clearly in this model. The experiment to test the operation and efficiency of lamella decanter was conducted with 2 models of “SPIRALOIL”, the first one with normal spiral hydrophobic plate insertion, the other with the combination of corrugated hydrophilic plate and smooth hydrophobic plate insertion, also in spiral form. Kerosene/ water mixture was used for the research, with the addition of surfactant in some tests to vary the interfacial tension.

Result

The result of this study provides models for sizing and calculating the efficiency of two lamella decanters “SPIRALOIL”. The microscopic observation clearly shows the interaction between oil and surface of lamella plate and influence of wettability and inclination of the plate. The result leads to optimum configuration of the SPIRALOIL.

Significant findings

1. In this thesis, mathematical models for sizing and calculating efficiency of the decanters are proposed. The models are based upon theoretical model of classical decanters.

2. The microscopic study shows that the operation of the decanters consists of 2 steps, i.e., decanting and coalescing.

3. For hydrophobic plate, decanted oil will adhere to the surface of the plate, then coalesce to form a film on the surface. However, the following decanted oil drop will adhere and coalesce with the oil film with more difficulty than adhering to the plate itself.

4. For hydrophilic lamella plate, decanted oil will not form a film at the surface of the plate, but will accumulate in the form of big drop. This oil drop will be entrained with the water when it reaches sufficient size and, then, separated from water at the ends of the plate.

5. The inclination of the plate will affect the coalescing step. For hydrophobic plate, there is a possibility that the decanted oil drop will not coalesce with the film, but roll over the film until reaching the end of the plate. However, at the end of plate will appear the big oil drop, because the point will play the role of drip point or salting-out point. This big drop can intercept those non coalesced drops and becomes one larger oil drop until it reaches sufficient size to be snapped off by the water flow.

6. For hydrophilic inclined plate, the decanted oil drop will move along, rather than adhere to, the inclination of the plate. Anyway it can coalesce with other decanted oil drop along its way to be a larger drop in the same manner as a “snowball”.

7. Presence of surfactant will decrease the oil droplet size, thus hinder good decanting step. Furthermore, it will lower the interfacial tension, thus hinder good coalescing step and cause decreasing in the efficiency of decanter.


10 I-10

8. From the study, the optimum SPIRALOIL configuration is the one with combination of corrugated hydrophilic and smooth hydrophobic plate insertion, installed in horizontal position. This configuration will combine the advantage of enlarging of the oil drop both at the end of the plate (drip point enlargement) and within the plate (snowball enlargement). Furthermore, the horizontal installation (inclination = 0) will favor coalescing step.

3.4 Coalescer

From STOKES law (eq. 3.1), rising velocity of the oil droplet is proportional with square of droplet diameter. So increasing the droplet diameter will make rising velocity increasing at greater rate than other parameters in the equation. To increase the diameter, we need a process that can integrate small droplets into the big one. Coalescer is very effective process that enables coalescing or integrating of a number of small oil droplets in the wastewater into relatively big drops, which can be easily separated by ordinary decanter. Because coalescer operation depends upon several parameters or mechanisms and each of them can be optimized to obtain better efficiency or to fit some specific working conditions, then there are several theses related to coalescer, in order to cover every aspect of the process. Some researches are dedicated to detailed study of mechanism of the process. Some are contributed to various type of coalescer bed or modes of operation. While some are devoted to application on some specific wastewater or working condition. All of these related theses on coalescer can be outlined as follow.

3.4.1 Thesis of AURELLE [3]

This research contributed to the study on fundamental mechanism of granular bed coalescer, which was used as a basis and guideline for following researches.


• Influence of essential parameters, i.e., geometry of bed, and operation parameters, such as feed rate to efficiency of coalescer

• Fundamental mechanisms or phenomena, take place with in coalescer during its operation

• Model development for granular bed coalescer


The experiment was carefully planned to cover every aspect of coalescer operation, i.e.,

• Wastewater characteristic : Hydrocarbons, used in the research, included gasoline, kerosene at various concentrations, in form of both direct and inverse emulsion.

• Collector material : Alternatives of collector or bed material included; • Type of material : glass beads, resin, sand, chamotte (porous dried clay) • Wettability : Oleophilic or Hydrophilic • Granulometry or size and size distribution of the collectors • Geometry of colaescer • Diameter or size of the coalescer tank • Bed height • Empty bed flow velocity


11 I-11

Result

The result of this study indicate the important parameters which influent the efficiency of coalescer. It also shows complex phenomena in coalescer operation by means of visualization or photographic technique. Model for sizing the coalescer was proposed. The study also provided significant criteria to select the material of bed and guideline for optimization and for further development of coalescer


1. The main parameters, which effect the efficiency of the coalescer, consist of :

• Wettability of bed material, • Bed height, • Empty bed velocity, • Granulometry of bed, • Ratio of hydrocarbon in the wastewater

2. The phenomena or mechanism occur in the coalescer can be divided into 3 fundamental steps as follow.

• Step 1: Interception, which consists of 3 major transport phenomena, i.e., sedimentation, direct interception and diffusion. This step is normally the efficiency-determining step of coalescer. The research makes it possible to develop the model, based on model of filtration process, which governs all phenomena in this step.

• Step 2: Adhesion-Coalescence. The efficiency of this step depends mainly on wettability of bed material. So, this step can be optimized by using oleophilic material as coalescer bed.

• Step 3: Salting out or enlargement of coalesced liquid. This step depends on 4 parameters, i.e. wettability of bed material at the discharge surface, empty bed flow velocity, interfacial tension and ratio of dispersed phase and continuous phase in emulsion treated. So, for any given wastewater and bed material, one can optimize this step only by varying feed flowrate. However, the research shows that installation of guide, such as woven fibrous metal, attached to discharge surface of granular bed, can eliminate influences of the 4 parameters described above and allow the coalescer to operate at much higher velocity. This guide suppresses the limiting step 3 by channeling coalesced oil directly into decanted oil layer at the decanter surface.

This thesis provides very important concepts and mechanisms of coalescer that leads to further studies on various types of coalescer.

3.4.2 Thesis of SANCHEZ MARTINEZ [6]

This research contributed to extensive study on granular bed coalescer, first researched by AURELLE [3]. The research was emphasized on granular bed coalescer with guide, and mixed bed coalescer.


12 I-12


• Influence of essential parameters, i.e., geometry of guide, and operation parameters, such as feed rate and ratio of oil in wastewater to efficiency of classical coalescer and guided coalescer

• Study on mixed bed coalescer to treat direct and inverse emulsion simultaneously • Application on mixed bed coalescer with guide in phenol extraction


The experiment was conducted with transparent glass coalescer models. Coalescer bed materials used during the experiment were sand and glass bead. Both materials were specially coated to acquire oleophilic or hydrophilic property. 2 guides of different sizes of woven fiber and different porosity were tested. The author intended to study the operation of coalescer for liquid/liquid extraction. So he chose phenol extraction in this study. Phenol in wastewater can be extracted by dissolving into appropriate hydrocarbon solvent. Solvent and wastewater will be intensely agitated to maximize contact, thus, mass transfer. So, they normally become emulsion, both direct and inverse. Hydrocarbons used as solvent in the experiment were L.C.O, medium-cut petroleum, and gasoline.


1. Installation of guide, porous material preferably wetted by dispersed phase, helps optimizing the 3rd mechanism “salting out or enlarging of dispersed phase”, thus, allow the granular bed coalescer to work at the flowrate of 1.5 - 6 times higher than classical coalescer without guide. It helps preventing formation of mousse, jet at the bed outlet. It also prevents formation of zone of non-coalesced drops between decanted solvent and clarified water zone. The guided coalescer can be operated at higher ratio of dispersed phase/ continuous phase than the classical coalescer. The research confirms the advantage of guided coalescer proposed by Prof. AURELLE.

2. When both direct and inverse emulsion are simultaneously present in the wastewater, two-stages process of hydrophilic bed and hydrophobic bed coalescers, connected in series, may encounter a problem of re-dispersion or re-fragmentation, thus, lower the total efficiency.

3. Mixed bed coalescer, using combination of both hydrophilic and hydrophobic bed material, is proven to be the effective way to treat direct and inverse emulsion simultaneously.

4. Application of mixed bed coalescer with guide shows satisfactory result and is proven to be a good alternative in the field of liquid-liquid extraction process.

3.4.3 Thesis of DARME [7]

This research contributed to the application of coalescer on treatment of stabilized emulsion. This work was one of successions of the research of AURELLE [3] in order to understand profoundly about working mechanism, limitation and application of granular bed coalescer.


13 I-13


• Influence of surfactant on coalescer operation • Trials on optimization of each basic mechanisms of coalescer • Study on mixed bed electrocoalescer


The experiment was conducted with transparent glass coalescer models. Coalescer bed materials used during the experiment were oleophilic resin and glass bead. To study about influence of surfactant on wettability, glass, resin and steel were tested. For the study on mixed bed electrocoalescer, mixed material of aluminium and resin was tested. Emulsion of kerosene and water was used throughout the experiment, with various dosages of cationic, anionic and non-ionic surfactants.


1. Surfactants have important effects on emulsion and operation of coalescer as follow;

• Decreasing average size of droplet in emulsion, then, results in limiting the efficiency of the 1st mechanism “Interception step” of coalescer.

• Decreasing interfacial tension, increasing electric charge at the surface of the droplets and increasing viscosity of surface of the droplets, then, result in poor adhesion between droplets and bed, as well as, ineffective collision (collision without coalescence) between droplets. These leads to limiting the efficiency of the 2nd mechanism “Adhesion-coalescence step” of coalescer.

• Decreasing interfacial tension and adsorption of surfactant on enlargement surface (or grill) results in limiting the efficiency of the 3rd mechanism “Enlargement step”

2. Trials to optimize the 1st and 3rd mechanism were conducted by;

• Optimization of the 1st mechanism: increasing bed depth • Optimization of the 3rd mechanism: installation of “guide”

The efficiency of the coalescer after these 2 modifications is not improved. Hence, it shows that the 2nd mechanism “Adhesion-Coalescence” is the limiting step in this case.

3. The 2nd mechanism can be optimized by destabilizing the emulsion, chemically or electrically. To apply only the coalescer to treat stabilized emulsion, without additional chemical process to destabilize or “break” the emulsion, the author propose to destabilize the emulsion by electrostatic effect. To achieve the electrostatic effect required, the author proposes to use the combination of oleophilic resin and metal (such as aluminium) as coalescer bed. With carefully selected bed materials, the electrostatic effect, caused by electrical potential of the 2 materials, will be sufficient to subdue the surface charge of droplets, caused by surfactants. Then, it will improve adhesion and coalescence within the coalescer.

4. However, this mixed bed is effective only when the emulsion is stabilized by ionic surfactant. In case of non-ionic surfactant, coalescence is hindered by


14 I-14

mechanical, not electrical, barrier, caused by adsorption of the surfactant at the surface of droplets. This barrier cannot be subdued by electrostatic effect.

3.4.4 Thesis of TAPANEEYANGKUL [8]

This research was intended to study on a variation of coalescer, i.e., “Dynamic fibrous bed coalescer”. This type of coalescer has advantage in its anti-clogging property which allows it to be used with oily wastewater with high suspended solids concentration.


• Influence of essential parameters, i.e., geometry of fibrous bed, and operation parameters, such as feed rate, rotating speed of fibrous bed to efficiency of coalescer

• Model development for fibrous bed coalescer


The experiment was conducted with transparent glass coalescer model. Coalescer bed materials used during the experiment were of brush type, made of nylon (polyamine) and polypropylene fiber with various fiber diameters, brush sizes, and void ratios. The brush was mounted to a variable speed motor, so the rotation of the brush can be adjusted. Emulsion used in the experiment was kerosene emulsion.


1. The main parameters, which effect the efficiency of the coalescer, consist of;

• Granulometry of emulsion (sizes of droplets) • Bed height • Bed diameter • Void ratio or porosity of the bed • Diameter of fiber of the fibrous bed • Empty bed velocity • Rotating speed of bed

2. In this thesis, mathematical model for sizing and calculating efficiency of fibrous bed coalescer was proposed, i.e.;

%100)0.74PV0.58

Fd0.03D

0.53N0.35H0.35ε)(10.58E0.67d

( ⋅−

=dη

Where ηd = Efficiency of coalescer dE = Diameter of dispersed phase ε = Void ratio or porosity of bed H = Bed height N = Rotating speed D = Diameter of bed dF = Diameter of fiber Vp = Empty bed velocity


15 I-15

This model is valid when 52<Re<1164. However, scaling-up of the model will be limited by bed construction itself because the fiber elements of the bed tend to compact by their own weight. Moreover, if the bed is too large, void ratio at the tips of fibers will be very different from center’s. This may cause error in calculation. So, it is recommended to use a number of small coalescers instead of a single large one.

3. This coalescer has several advantages as follow;

• Anti-clogging. No regeneration or backwashing required • High void ratio, thus, head loss is very low • Very small size of fiber, compared to granular bed material, ensures good

interception of oil droplets • Treatment efficiency is adjustable by mean of adjusting rotating speed of bed

3.4.5 Thesis of DAMAK [9]

This research was intended to study another variation of coalescer, i.e., “Pulsed granular bed coalescer”. This type of coalescer had been initiated to enable the granular bed coalescer to treat oily wastewater with high suspended solids concentration without regeneration or backwashing process. This coalescer, in fact, is a classical up-flow coalescer, except it bottom end is equipped with rubber diaphragm, driven by pneumatic piston. The diaphragm will be periodically driven up, then released to go back down. This action will cause brief fluidization of bed and the trapped solids will be released from the bed.

The author also studied a new type of coalescer, called “Phase inversion coalescer”. For its operation principle, the wasted emulsion will be forced downward through small tubes, equipped at the top of coalscer column, to produce emulsion drops of required size. These drops will flow through thick layer of hydrocarbon, which is of the same type as dispersed phase in the emulsion. The drops play the role of micro decanter. Theoretically, hydrocarbon droplets in the emulsion drops will float to the top of the drops, then, coalesce with surrounding hydrocarbon layer. With appropriate depth of hydrocarbon layer and size of emulsion drop, the dispersed phase in emulsion drop will be totally separated and become only the drop of water phase when it flow out off the hydrocarbon layer into water phase underneath. The schematic of phase inversion coalescer will be as shown in Fig 3.2.

Hyd

roc a

rbon

la

y er

Orifice plate

Water layer

Decanting of droplets inemulsion drop

Influent emulsion

Top of hydrocarbon layer

Bottom of hydrocarbon layer

Emulsion jet from orifice

Emulsion drop

Effluent

Hydrocarbondroplets in emulsion drop

Fig. 3.2 Schematic of Phase inversion coalescer


16 I-16


• Influence of essential parameters, i.e.,diameters of droplets, height of hydrocarbon, etc. to efficiency of Phase inversion coalescer

• Study on up-flow pulsed granular bed coalescer • Model development for granular bed coalescer, based on dimensional analysis


For the phase inversion coalescer, the experiment was conducted with transparent glass coalescer model. Furthermore, to achieve a better visual study, a special tube equipped with micro-syringe was used to study internal phenomena in droplet. The tube was fixed at the tip of the syringe, to be within hydrocarbon stream. Emulsions, used in the experimental, were kerosene/water emulsion, hexane/water emulsion and T.I.O.A./water emulsion. T.I.O.A/water emulsion is the solution of kerosene, triisooctylamine (T.I.OA) and tributylphosphate. This solution is normally used as extracting solvent in hydro-metallurgical industries.

For up-flow pulsed granular bed coalescer, the experiment was conducted with glass coalescer with rubber diaphragm bottom. The diaphragm was connected to pneumatic piston that can drive the diaphragm up and down at preset interval. Coalescer materials were hydrophilic glass bead and stainless steel. Emulsion tested were kerosene/water emulsion and T.I.O.A./water emulsion with an addition of fly ash as suspended solids.

For model development of granular bed coalescer, transparent glass coalescer was used. Hydrophobic and hydrophilic glass beads were used as coalescer material. Hydrocarbons used in the experiment were kerosene, T.I.O.A., heptane, anisole and toluene.


1. The efficiency of phase inversion coalescer will increase with the modification of these parameters, i.e.;

• Increasing in height of hydrocarbon layer (until certain limiting height) • Decreasing in size or diameter of emulsion drop that will be forced through

the hydrocarbon layer (again, until lower limit) • Incresing in flowrate of treated emulsion • Increasing in height of emulsion-drop-accumulated or dense layer

The efficiency will also depend on characteristics of emulsion e.g. viscosity, interfacial tension, granulometry of emulsion etc.

2. While the emulsion drop travels pass through hydrocarbon layer, internal turbulence or circulation flow will be induced within the emulsion drop, which then disturbs good decanting of oil droplets, thus causes the decrease in the efficiency.

3. To avoid this turbulence, the flowrate should be increased to produce more emulsion drops. The rate of production has to be greater than the rate of coalescence of the drops, at the bottom of hydrocarbon layers. So, this results in formation of emulsion-drop dense bed or layer at the bottom of hydrocarbon layer. This layer will dampen the downward velocity of the drop, which results in decreasing of internal turbulence. The efficiency, then, will be improved.


17 I-17

4. This coalescer is suitable for treatment of primary emulsion (dE ≥50 μm). And the study shows that the efficiency is better than that of the classic decanter.

5. For pulsed coalescer, the study shows that brief pulsation, which causes the bed to fluidize, can regenerate the bed and clean up the accumulated matters.

6. In classical coalescer, coalescer bed material is usually lightweight resin, which requires the top grill to keep the bed in place without carrying over with the wastewater. However, in this study, it shows the possibility of using of relatively high density coalescer bed, e.g., stainless steel, without the top grill can replace the use of lightweight material. When the grill is not required, it allows us to use pulsating motion to fluidize, thus, regenerate the bed.

7. In this thesis, a mathematical model for sizing and calculating efficiency of granular bed coalescer was proposed. Unlike the model of AURELLE, which derived from theoretical mechanisms of coalescence, this model is based on dimensional analysis. So physical properties of wastewater, which are not shown in AURELLE’s model, are taken into account. Hence, it covers wider range of wastewater.

3.4.6 Thesis of MA [16]

This thesis was the main research on hydrocyclone for hydrocarbon/water separation. So it will be described in detail in section 3.6. However, the author had tested, for the first time, the efficiency of the combination process of hydrocyclone and coalescer, which worth describing here.


1. This study shows, for the first time, the possibility to use the combination process of hydrocyclone/coalescer and coalescer/hydrocyclone to improve the total efficiency of oil/water separation.

3.4.7 Thesis of SRIJAROONRAT [10]

This thesis is an application research on treatment of non-stabilized oil/water emulsion. This type of thesis provides important data that can be applied in real life situation. So we will devote one section to review these theses. However, as one part of her thesis, SRIJAROONRAT had studied on comparison between fibrous bed coalescer of brush type and random, disorderly type (like a steel wool). She also studied on combination of coalescer/hydrocyclone to treat non-stabilized emulsion. The idea is to use coalescer to increase droplet size, permitting a good separation by the following hydroclclone. Her study on coalescer will be briefly mentioned here to complete the entire coalescer studies.


1. At high empty bed velocity, the coalescer with random bed will coalesce and enlarge the droplets into relatively large drop, while the coalescer with brush type bed tends to produce stream of jet, containing small oil drop.

2. However, the disorderly fibrous (steel wool) bed coalescer tends to be clogged by suspended solids, usually presence in the oily wastewater. So the author proposed new configuration of fibrous bed in form of combination of 2 brushes. The internal one is of ordinary brush. The external one will look like coil spring


18 I-18

with its fiber elements protruding inward and toward the center. This type of bed is believed to provide good interception, as same as the disorderly bed, yet remain its anti-clogging properties, like brush-type bed.

3. Combination between coalescer/ hydrocyclone is proven to provide good efficiency on the wider range of feed flowrate and size of oil droplet.

3.4.8 Thesis of WANICHKUL [11]

This thesis is an application research on treatment of stabilized oil/water emulsion. Details of this thesis will be described in the section devoted to application thesis. However, as one part of his thesis, WANICHKUL had studied on the combination of coalescer/hydrocyclone and hydrocyclone/coalescer to treat non-stabilized emulsion, as well as the efficiency of multi stage fibrous bed coalescer.


1. Testing of the 2 types of the combination in industrial pilot plants shows satisfactory result. These combinations provide promising alternatives for industrial wastewater treatment process. Furthermore advanced combination of coalescer-hydrocyclone-coalescer is tested. It promisingly shows that 2 coalescers can be used, one before the hydrocyclone to increase droplet size entering the cyclone, and another at downstream of the cyclone to coalesce the concentrated oil/water mixture from the outlet of the hydrocyclone.

2. Result from multi stage fibrous bed coalescer test shows that this “modified bottlebrush” bed provides better efficiency than classic bottlebrush bed. This is because the gaps between each stage cause turbulance, then promote coalecence between droplets that, somehow, fail to coalesce in the previous stage of bed.

3.5 Flotation

Flotation is the separation process that makes use of increasing the density difference to increase the rising velocity of droplets of dispersed phase. The density difference can be increased by mean of integrating air or gas bubble with oil droplets. The bubble/droplet agglomerate will have lower density than droplet alone, thus, result in increasing of the rising velocity. Because flotation mechanisms are very complex, then there are several theses related to flotation, in order to cover every aspect of the process. All of related theses on flotation can be outlined as follow.

3.5.1 Thesis of SIEM [12]

This research contributed to the study on fundamental mechanism of Flotation for hydrocarbon/water separation, which was used as a basis and guideline for other followed researches.


• Study on the interaction between hydrocarbon droplets and air bubble, and overview of efficiency of dissolved air flotation and induced air flotation.

• Influence of essential parameters, coagulant dosage, pH, etc. to efficiency of flotation

• Model development for dissolved air flotation, based on filtration model.


19 I-19


Visual study of interaction between oil droplets and bubble had been conducted in special transparent model, equipped with a microscope and a VDO camera. For the study on flotation, transparent flotation model was used. Emulsion used during the experiment was gas-oil/water emulsion.


1. From visual study on interaction between oil droplet and air bubble, it shows that the bubble will agglomerate within the inside of oil droplet or the oil will form a thin skin, as a shell, around the air bubble.

2. In this thesis, mathematical model for sizing and calculating efficiency of dissolved air flotation is proposed. The model is based on filtration model by assuming that the air bubble is collector (sand or filter media in case of filter). Prediction results of the model fit well with experimental results. Anyway, the author did not account for the quantity of pressurized water that used to generate the bubbles. This amount of water can cause dilution effect, thus, contribute to reducing in wastewater concentration. So, the model proposed in this research should be reconsidered again to clarify the efficiency of interception of air bubble and the efficiency from dilution effect of pressurized water. The revised model will be described in Part 2 of this thesis.

3.5.2 Thesis of AOUDJEHANE [13]

This research contributed to application of flotation on treatment of hydrocarbon-polluted wastewater. This work was one of successions of the research of SIEM [12] in order to understand profoundly about working mechanism, limitation and application of flotation.


• Study on structure of bubble/droplet agglomerate and influence of additional of flotation reagent in form of the transfer compound( composé transferable)

• Influence of hardness and salinity of pressurized water on production of air bubble • Study on probability of collision and coalesce of air bubble and oil droplet


Visual study of the interaction between oil droplets and bubble had been conducted in special transparent models, equipped with a microscope and a VDO camera. The models are equipped with 2 syringes, located close to each other. These 2 syringes were used to supply air (or gas) and oil to form immobilized bubble and oil droplet respectively. Bubble and oil droplets were brought into contact to study the formation of the agglomerate. For study on flotation, transparent flotation model was used. Hydrocarbon used in this study was kerosene. For transfer compounds, ammonia gas, cationic surfactant and methanol were used as promoter of mass transfer between phases.


1. From visual study on interaction between oil droplet and air bubble, it confirms the result of SIEM [12] that the bubble will agglomerate within the inside of oil


20 I-20

droplet or the oil will form a thin skin, as a shell, around the air bubble. The agglomerate in from of bubble and droplet locating side by side of each other, which is exist in case of bubble/particle interaction, occurs only in unstable, transition forms. It also shows that coalescence time between bubble and oil droplet is less than that of the same species.

2. From the study with static transparent model, it shows that addition of transfer compound (as gas to bubble air or as transfer compound to water) affects, more or less, probability of coalescence between bubble/droplet, bubble/bubble and droplet/droplet. It can be described that mass transfer of these transfer compound from one phase to another (for example, bubble gas from bubble toward water, etc.) can cause disturbance in local surface tension and thinning of film, then, help increasing coalescence. This effect is well known in liquid-liquid extraction process as the Marangoni effect.

3. From the study with flotation lab-scale model, it shows that the effect of turbulence, caused by movement of air bubble, promotes the probability of collision both inter-species and same species of bubbles and oil droplets. However, this augmentation in collision increases flotation efficiency only slightly because the number of bubble, somehow, decreases from the effect of collision and coalescence, that leads to depletion of the number of bubble/droplet agglomerate. Anyway, the author notes that, because coagulation-flocculation process is not applied, the number of oil droplets is, then, always greater than the number of bubbles. Then, it may be interesting to study further on coalescence or coagulation of oil droplets.

3.5.3 Thesis of DUPRE [14]

This research contributed to the study on related mechanisms and mathematical model development of dissolved air flotation. Even though the author intended to study on solids/ liquid separation, some parts of the study on formation of air bubble and some general ideas can be applied in the field of hydrocarbon/water separation.


• Study on precipitation of dissolved gas to form bubbles • Study on formation of bubble/particle agglomerate • Model development for dissolved air flotation


To study the precipitation of dissolved gas, the author tested several pressure-reducing venturi tubes (convergence-divergence nozzles) of various divergent angles. Tube materials tested consisted of stainless steel (hydrophilic) and plastic (hydrophobic). Influence of divergent angle, wettability of tube, length of pipe after the pressure reducing device and addition of some chemicals were studied.

To study the formation of bubble/particle agglomerate and operation of flotation, the experiment was conducted using 3 apparatuses. The 1st one was a glass container with syringes and holders, for supplying bubbles and particles. Bubble can be fixed in place while particles were brought to contact with it or vice versa. The 2nd apparatus was glass container with sealed cover, connected to a vacuum pump. The container was filled with water and


21 I-21

various particles were added. Then, the air was pumped out to create vacuum. This apparatus was used to observe nucleation of bubble. The 3rd one was a transparent flotation model.


1. The study result, which can be applied to hydrocarbon/water separation, is the one about formation of bubble at pressure reducing valve. The study shows that efficiency of flotation can be improved if the design of pressure reducing valve is optimized to generate microbubble in majority.

2. The use of hydrophobic venturi or pipe can cause the increase in size of the largest bubble generated. These large bubbles, though small in the number, consume almost all of the gas volume. So the number of generated microbubbles, which plays an important role in flotation, is left but only in small proportion.

3. Addition of surfactant in pressurized water can cause augmentation in population of microbubbles. On the contrary, addition of polyelectrolyte will cause a decrease in electric charge, which is favorable for the coalescence of bubbles.

3.5.4 Thesis of PONASSE [15]

This research contributed to study and trial on improvement of flotation efficiency. The author had tested several ways to improve formation of microbubble, such as addition of chemicals, using ultrasound vibrator to create microbubble. She also performed a feasibility study on “Deep shaft” flotation unit.


• Study on the influence of pressure reducing valve on formation of bubble • Feasibility study on deep shaft flotation unit • Feasibility study on combination of pressure reducing valve and ultrasound

vibrator.


To study the influence of pressure reducing valve on formation of bubble, the author tested several pressure-reducing venturi tubes (convergence-divergence nozzles) of various divergent angles. Tube materials tested consisted of stainless steel (hydrophilic) and plastic (hydrophobic). Influence of divergent angle, wettability of tube, length of pipe after the pressure reducing device and addition of some chemicals were studied.

For study on deep well flotation unit, a 30-m depth deep well unit was tested. The unit consisted of 2 concentric pipes. The external pipe diameter was 0.15 m. On the top of the pipes placed settling tank of 1.2*1.2 m, water depth 0.7 m. The wastewater was charged with various suspended solids, both hydrophobic and hydrophilic. The air was introduced into the system by several methods, i.e.,

• Direct injection into the wastewater at static mixer • Via porous plate, located inside the wastewater pipe before enter the deep shaft

reactor • Via injection chamber, air will forced through small orifice into the chamber • Via membrane


22 I-22

• Via ejector • Injection in form of pressurized water via pressure reducing valve to form bubble,

then mix with inlet wastewater at the top of deep well reactor


1. We are interested only in the result that can be applied to hydrocarbon/water separation, i.e., formation of bubble. From the study, it shows that precipitation of dissolved gas appears firstly in form of big gas pocket within pressure reducing valve. Then the pocket is broken or fragmented by hydrodynamic force to form bubble.

2. Geometry of pressure reducing valve, which determines hydraulic condition, is the key parameter to obtain microbubbles.

3. To obtain good efficiency, pressure reducing device should be placed as close as possible to mixing zone between wastewater and pressurized water

3.6 Hydrocyclone

From STOKES law (eq. 3.1), rising velocity of the oil droplet is proportional to gravitational acceleration. So if one increases the acceleration, the rising velocity will be increased as well. To increase the acceleration, it can be achieved by replace gravitational acceleration with centrifugal acceleration. This can be done mechanically, such as the use of rotating machine like centrifuge, or by converting hydrodynamic force to centrifugal motion. Hydrocyclone is the process that uses the latter principle to increase the acceleration.

In our lab, we emphasize on hydrocyclone, rather than the centrifugal machine, because of its simplicity and economy. Furthermore, for most of oily wastewater we encounter, the centrifugal acceleration induced by hydrodynamic force alone is sufficient to provide satisfying separation. The hydrocyclone have no moving part. Its structure is relatively much more simple than centrifugal machine. And it uses the driving force only from its feed flowrate. So it is more economic than centrifuge both in investment cost and operating cost. Now, there are various types of hydrocyclone, commercialized by many manufacturers. Even so, basic concept of these products is identical. Understanding of the concept including related mechanisms and limitation of the process will lead to appropriate design, selection and operation, as well as improvement of the process. There are several theses related to hydrocyclone, in order to cover various aspects of the process. All of related theses on hydrocyclone can be outlined as follows.

3.6.1 Thesis of MA [16]

This thesis was the main research on hydrocyclone for hydrocarbon/water separation.


23 I-23


• Study on new approach for calculating two-phase hydrocyclone, used for liquid/liquid separation application

• Study on three-phase hydrocyclone, used for liquid/liquid/solid separation • Feasibility study on the combination process of three phase hydrocyclone and

coalescer


To study on a new approach for calculation on 2-phase hydrocyclone, the author based his experiment on “THEW” type hydrocyclone, initiated by Professor Thew, UK. For 3-phase hydrocyclone, he had initiated this hydrocyclone by integrating the liquid/liquid hydrocyclone of “THEW” type and the solid/liquid hydrocyclone of “RIETEMA” type to one unit. For coalescer tested, he used coalescers with various sizes of “brush” type beds. The emulsions used in the experiment were based on petroleum from the “Sud-Ouest” french oil rig, with various additions of very fine bentonite (3.7 μm) and calcium carbonate powder (6.2 or 16 μm). Average oil droplet in the emulsion tested was 15 to 50 μm.


1. In this thesis, a mathematical model for sizing and calculating efficiency of THEW type hydrocyclone was proposed. This model, based on calculation of decanter, can be used to calculate trajectories of oil droplets in the cyclone, as well as the separation efficiency for each size of oil droplet. Difference between model prediction and experimental result is around 10%.

2. It is recommended that the optimum angle of conical section of THEW hydrocyclone should be 8° to 12°. And ratio Dn/D (nominal diameter of cyclone/ diameter of cylindrical part of the hydrocyclone) should be around 0.5.

3. The author had designed the 3-phase hydrocyclone by integrating THEW type hydrocyclone with RIETEMA type hydrocyclone. To do so, the vortex finder of RIETEMA hydrocyclone is replaced by THEW hydrocyclone, as shown in fig. 3.3. Test results show that the hydrocyclone is capable to separate, efficiently and simultaneously, lightweight hydrocarbon and heavy suspended solids from water.

4. The study shows that separation efficiency of hydrocyclone will drop rapidly if the diameter of oil droplet is smaller than 20 μm.

Solid-liquid part (Rietema’s part) Liquid-liquid part (Thew’s part)

DoDDs

DiDu

Dp

L5 L3L1L3

L4

Fig. 3.3 Three- phase hydrocyclone

Note: Di/D=0.25, Do/D=0.43,Ds/D=0.28, Du/D=0.19, Dp/D=0.034, L1/D=0.4,L2/D=5, L3/D=15, L4/D=0.3


24 I-24

3.6.2 Thesis of CAZAL [17]

This research contributed to the feasibility study of solid/liquid separation hydrocyclone on treatment of storm water, domestic wastewater, and thickening of sludge. In spite of the fact that this research is mainly related to solid/liquid separation process, it provides useful details about model development and basic concepts of hydrocyclone. Furthermore, oily wastewater always contains some amount of suspended solids. So, some parts of this thesis can fulfil the study of hydrocyclone for oily wastewater treatment charged with suspended solids.


• Feasibility study on application of hydrocyclone for treatment of storm water, domestic wastewater, thickening of primary sludge and biological sludge

• Model development for solid/liquid separation hydrocyclone and study on influence of shape of suspended solids to separation efficiency

• Study and proposal on multi-hydrocyclone processes for treatment of storm water


For the feasibility study on hydrocyclone for wastewater treatment and sludge thickening, the author had performed the experiment “on site”, using a pilot plant. The pilot plant was equipped with storage tank, pumps and piping system that allow to test 2 hydrocyclones instantaneously, both in series and parallel. The hydrocyclones used in the experiment were product of NEYRTEC, equipped with replaceable outlet. The sizes of cyclones were tested, i.e., nominal diameter of 75 and 50 mm. Outlet ports of the cyclones can be changed to study the influence of their sizes on cyclone operation. The wastewater and sludges were provided by the wastewater treatment plant at Ginestous. Characteristics of wastewater were as summarized in table 3.1. For study in laboratory, The same pilot plant was used with synthetic wastewaters, summarized in table 3.2.

Table 3.1 Summary of characteristics of wastewaters and sludges for “on site” experiment

Domestic wastewater Storm water Primary sludge Biological

sludge

SS (mg/l) 150-490 41-700 22000-16000 4170-6660 Median diameter, d50 (μm)

24-37 10-28.8 160 36-96

Table 3.2 Summary of characteristics of synthetic wastewaters

Calcium carbonate Talc Talc

Ion exchange

resin 1

Ion exchange

resin 2

Name CaCO3 A60 Steamas 29 IRP 69 OG 4B Density (g/cm3) 2.7 2.7 2.7 1.57 1.57 Median diameter, (μm)

10 25 32 78 56


25 I-25


1. In this thesis, it shows that the efficiency of hydrocyclone for wastewater treatment and sludge thickening application are relatively low, compared to classic decanter or gravity thickener. Main reasons for poor efficiency consist of;

• Shear force from flow pattern in hydrocyclone, which causes deflocculation effect or decreasing in size of biological floc

• Nature of the wastewater, which normally contains high contents of fiber agglomerates. These fibers hinder separation phenomena in hydrocyclone

2. The author had tested the operation of hydrocyclones using the range of influent SS concentrations from 10 mg/l to 20 g/l and maximum purge ratio of 2%. The result shows that efficiency will increase if influent flowrate is increased.

3. Shape of suspended solid affects the efficiency of hydrocyclone. This influence is the function of particle’s size. The smaller the size is, the less the influence is.

4. In this study, mathematics model of hydrocyclone is proposed and form coefficient is introduced.

5. The author proposed the combination of 3 hydrocyclones in series as the recommended process for storm water treatment. She also proposed installation of grit pot (the short cylindical chamber), connected to discharge port of the hydrocyclone, to minimize the influence of particle shape.


This thesis is an application research on treatment of non-stabilized oil/water emulsion. SRIJAROONRAT, as one part of her thesis, had studied on the combination of coalescer/hydrocyclone to treat non-stabilized emulsion. The hydrocyclone used in this research is the product of Dorr Oliver, model DOXIE 5, nominal diameter of 1 cm, 10 cm long. Her study on hydrocyclone is briefly mentioned here to complete the entire hydrocyclone studies.




This thesis is an application research on treatment of stabilized oil/water emulsion. Details of this thesis will be described in the section devoted to application thesis. However, as one part of his thesis, WANICHKUL had studied on the combination of coalescer/hydrocyclone and hydrocyclone/coalescer to treat non-stabilized emulsion. It will be briefly mentioned here. WANICHKUL had been tested the combination of coalescer/hydrocyclone and hydrocyclone/coalescer. In his experiment, he used Plexiglas hydrocyclone as shown in fig. 3.4. The influent will be fed through 2 tangential inlet ports, located at the upper cylindrical section. The oil and treated water will be discharged at each corresponding outlet port. Because the oil and treated water outlet port locates at the same end of hydrocyclone, so this configuration of hydrocyclone is called “co-current”.


26 I-26

Fig. 3.4 Hydrocyclone tested by WANICHKUL


1. The results shows that key parameters that govern the efficiency of co-current hydrocyclone are influent flowrate, recovered oil draw-off rate and ratio between outlet velocity of oil and outlet velocity of treated water, calculated at each corresponding outlet port.

2. The author recommended that the velocity ratio of oil/ treated water mentioned above should be greater than 1.5, which can be precisely adjusted by using of oil draw-off pump.

3. Testing of the 2 types of the combination in industrial pilot plants showed satisfactory result. These combinations provide promising alternatives for industrial wastewater treatment process. Furthermore an advanced combination of coalescer-hydrocyclone-coalescer is tested. It promisingly shows that 2 coalescers can be used, one before the hydrocyclone to increase droplet size entering the cyclone, and another at downstream of the cyclone to coalesce the concentrated oil/water mixture from the outlet of the hydrocyclone. The oil in this case is separated, not just concentrated, from the wastewater.

3.6.5 Thesis of PUPRASERT [25 ]

This thesis is contributed to study on the efficiency of hydrocyclone equipped with “grit pot”, as recommended by CAZAL [17]. It is also related to the feasibility study of an innovation on solid/liquid separation, which is a hybrid of 3 processes, i.e., coagulation-flocculation, dissolved air flotation and centrifugal separation. However, it seems that this thesis is not related directly to oily wastewater treatment. So we will not mention about the result of this research here. Anyway, some information and ideas presented in PUPRASERT’s thesis will be integrated in the stage of textbook composing in Part 3 of this thesis.

3.7 Ultrafiltration and other membrane processes

Membrane process seems to be the future of separation technology. Main component of every membrane process is the membrane, which can be briefly described as a porous material that plays the role of selective barrier. Under appropriate transmembrane pressure, the membrane of appropriate characteristic will allow only certain components in wastewater


27 I-27

or emulsion to pass. The rest will be retained by the membrane. Working principle of membrane process can be approximately compared to that of filtration.

Membrane processes can be divided into several categories according to the pore size of membrane, i.e., microfiltration, ultrafiltration, nanofiltration and reverse osmosis. Each category has different permeability and can retain different size of components. So it is crucial to select the most appropriate membrane process to achieve the required effluent quality. Then several researches are contributed to membrane selection.

In membrane treatment system, wastewater will be circulated on one side of membrane. The water and some components that can be passed through membrane pore will flow to another side of membrane. This portion is called “filtrate” or “permeate”. The portion that can not pass the membrane will have higher oil concentration because it losses its water component. It is called “concentrate” or “retentate”. So, the membrane can be considered as a concentrating process because the pollutant, in our case, oil, is not exactly separated, just concentrated. Increasing in oil contents of concentrate will cause important phenomena called polarization concentration. This can be described as oil rich layer that can obstruct flow of permeate. Solid content or some components in wastewater can be trapped in membrane pore and cause clogging. Concentration polarization and clogging of membrane are the main problems of this process. So many researches are devoted to solve this problem. All of related theses on membrane process can be outlined as follows.

3.7.1 Thesis of BELKACEM [18]

This research is contributed to the application of ultrafiltration on treatment of cutting oil emulsion. This type of emulsion is one of important oily wastewater. In metal industries, especially metal forming plants and mechanical workshops, machine tools and products needs good lubrication and cooling to ensure good quality of products and prolonged working life of the tools. Water, by its property, is very good cooling liquid and a cost-effective choice. As well as oil is very good lubricant and anti-corrosion substance. So the combination of oil and water in form of emulsion provides users with both properties required. To fulfil this purpose, the emulsion is charged with surfactants and co-surfactants to increase its stabilization. From user point of view, it yields satisfying result.

However, from environmental point of view, presence of these surfactants or emulsifiers makes separation process very difficult. This is because the oil phase in the emulsion is dispersed in form of very fine droplets, which impossible to be decanted naturally. So it requires special treatment.


• Feasibility study on the application of ultrafiltration on treatment of cutting oil macroemulsion

• Study on treatment methodology of the cutting oil emulsion by ultrafiltration process

• Study on membrane washing solution • Feasibility study on the application of ultrafiltration on treatment of cutting oil

microemulsion


28 I-28


The experiment was conducted with a batch model and a pilot-scale model. For batch model, the author used AMICON membrane test module, (fig. 3.5). The batch model was transparent container with pressure-tight cover. The membrane was placed at the bottom of the container. Wastewater was added into the container, then the cover was locked in place and compressed air line was connected to the inlet port at the cover. The wastewater was forced through the membrane by mean of air pressure, then collected and examined. During ultrafiltration process, the wastewater was continuously stirred by mean of magnetic stirrer.

For pilot-scale model, the author used commercialized cross flow ultrafiltration test module, model PLEIADE UFP2 of Tech Sep co,.ltd. (fig. 3.5).Wastewater was circulated through narrow gap between membrane and transparent wall of the model. In this manner, wastewater flow was tangential, not perpendicular to, the membrane surface. Some components would pass through the membrane and become permeate. The rest would be returned to storage tank, then circulated pass membrane again until it reached some certain concentration. After being circulated for many times, the concentrate would gain in temperature, so heat exchanger was provided to cool down the flow before re-entered the UFP2 module.

He also tested the porous fiber membrane in the same way as the plain membrane, described before. For the membranes, he used plain and porous fiber membrane of various cut sizes, ranged from 40 to 150 Kdalton.

For the emulsions, he used commercialized cutting oil emulsion of both microemulsion and macroemulsion types. He also used cutting oil macroemulsion from mechanical workshops.


1. In this thesis, the author proposed a mathematical model for calculating ultrafiltration flux of macroemulsion, as function of non-dimensional numbers.

2. The study indicates interesting aspect in integrating ultrafiltration with partial destabilization, using salt. From this methodology, the membrane will play the roles of filter and coalescer, which help reducing the influence of polarization layer and concentration factor. This leads to increasing in ultrafiltration flux for treatment of macroemulsion.

3. To solve the problem about clogging, the author proposed to use new formulated microemulsion, which is not saturated by oil, to wash the membrane and slow down clogging process. With a careful selection of surfactants in the microemulsion, it can be easily treated by the combination process of polyelectrolyte breaking and ultrafiltration. Because the author used co-surfactant with poor water solubility, residual pollutant concentration after ultrafiltration, mainly in dissolved form, was relative low, compared to ordinary microemulsion.

4. From study on treatment of microemulsion by ultrafiltration, it shows that ultrafiltration provides good separation between oil and water. However, the filtrate is heavily polluted by dissolved pollutant, then additional treatment is inevitably required. The author proposed to use reverse osmosis process to treat


29 I-29

the ultrafiltrate. The result shows that reverse osmosis can treat this wastewater with very high efficiency (approx. 98%).

6

7

1

4

5

Air compressorPressure reducing valvePressure guageAMICON test moduleAgitatorBalanceComputer

21

345

23

67

Fig. 3.5 Ultrafiltration models used by BELKACEM

3.7.2 Thesis of TOULGOAT [19]

This research is contributed to the study on formation mechanisms of thermal emulsion and feasibility study on treatment of the emulsion by ultrafiltration. This type of milky emulsion is generally encountered in essential oil extraction process, such as in perfume industry. Not only this study provides information about one of the application of ultrafiltration. It also provides the general idea of formation of thermal emulsion, which leads to precaution to avoid it formation. This idea can be applied to heteroazeotropic distillation, which is one of treatment processes for oily wastewater.


• Influence of essential parameters, such as temperature, condensing surface, etc., on the formation and characteristic of thermal emulsion

• Preventive approach to avoid formation of the emulsion or, if case arise, limit its concentration

• Feasibility study on treatment of thermal emulsion by ultrafiltration


30 I-30


To study on the entraining of volatile compounds with steam, the author used glass distillation apparatus, equipped with microscope to observe the condensate. She had tested several hydrocarbon compounds, such as alcanes with the number of C atoms from 5 to 12 (pentane to dodecane), toluene, benzaldehyde, isobutanol, series of primary alcohols, etc.

To study mechanisms of thermal emulsion formation, hydrodistillation apparatus was used. It is, in fact, the same device as the first one, with an additional installation of a stirrer. For the plants used for the experiment, she used several plants, i.e., cinnamon, aniseed, celery, lavender, etc. Condensate was found in form of decanted essential oil, floating on the top, and milky thermal emulsion.

For ultrafiltration, she used tubular ceramic membrane of 10 nm. pore size. The experiment was conducted with membrane test module, model MEMBRALOX T1-70 of SCT co., ltd. For the emulsion treated, she used synthetic emulsion of kerosene/water as well as thermal emulsion from extraction of cinnamon, aniseed and celery with various doses of surfactants.


1. In this thesis, it shows that the most important parameter in thermal emulsion formation is the variation of solubility of essential oil with changes in temperature. She also proposed the minimum variation of solubility that can cause the formation of thermal emulsion.

2. Condensation time is the important parameter that governs the size distribution of the emulsion. If the condensation is relatively slow, average diameter of droplet in the emulsion will increase.

3. The author shows that if the following compounds are present in the essential oil, thermal emulsion will not be formed;

• Hydrocarbon of high vapor tension • Hydrophobic hydrocarbon that solubility is not sensitive to temperature

change.

4. To avoid formation of thermal emulsion, she proposed the process called “hydrodistillation under reduced pressure (sub-atmospheric pressure)”.

5. The study shows that “white water” or milky thermal emulsion, in fact, contains natural surfactants. So its properties are close to that of stabilized emulsion.

6. Under carefully selected operating condition, ultrafiltration can provide good separation efficiency between water and essential oil.

3.7.3 Thesis of MATAMOROS [20]

This research is contributed to the study on treatment of stabilized emulsion, especially cutting oil emulsion, by various membrane processes. This research provides many useful information and comparison data of several membrane processes.


31 I-31


• Feasibility study on the application of various membrane processes for treatment of cutting oil emulsion, of both macroemulsion and microemulsion types

• Feasibility study on the application of combination processes between one of membranes and chemical destabilizing process for treatment of cutting oil macroemulsion and microemulsion


The experiment in laboratory scale was conducted with two batch test modules, one was fabricated in GPI Lab, and another was AMICON test module. For pilot-scale experiment, the author used cross flow membrane test module, model PLEIADE UFP2 from Tech Sep co., ltd., for micro- and ultrafiltration process. For nanofiltration and reverse osmosis, he used test module from OSMONICS.

The membranes used in the research were as tubulated in table 3.3. For emulsions, he used various commercialized cutting oil of macro- and microemulsio typesn.

Table. 3.3 Membranes test by MATAMOROS

Membrane Cut size Initial permeate flux (m.s-1.pa-1) Material

Microfiltration 0.10 – 0.45 μm 655 - 125786 Cellulose compounds, Polyamide and PTFE

Ultrafiltration 40-50 Kda 750*10-12 Polyacrylonitrile, Acrylonitrile

Nanofiltration 150 – 2000 Da 1.72*10-12 - 20*10-12 Cellulose

Reverse osmosis 150 Da 2.2*10-12 Polyamide


1. The results from this research show that the combination between chemical destabilization process with micro- or ultrafiltration can increase working permeate fluxes of the membranes and still obtain good efficiency.

2. For the combination of chemical destabilization and microfiltration, the system can operate at relatively high flux with moderate transmembrane pressure. This leads to development of continuous treatment process of cutting oil because, in the past, working pressure is always the limiting factor of the process. However, it requires high amount of salt.

3. For the combination of chemical destabilization and ultrafiltration, the result conforms to BELKACEM’s [18] that the salt added plays important role, by double layer compression, and/or adsorption/partial neutralization, in the reduction of repulsive force and promotion of coagulation and coalescence of droplets to form a free oil layer. This oil layer can be entrained by recirculation stream, and then removed at the concentrate storage tank.


32 I-32

4. The permeates from the combination of two processes stated above contain high concentration of TOD, which is the result of dissolved pollutants, especially co-surfactants.

5. Combination between ultrafiltration and reverse osmosis shows good result both in oil separation and dissolved pollutant elimination in permeate. However, operating cost is relatively high.

6. The use of nanofiltration shows good efficiency result in good oil separation, as well as dissolved pollutant elimination, at lower energy consumption than reverse osmosis.

7. From experimental results, the author recommended 2 treatment process trains for micro- and macroemulsion as shown in fig. 3.6

Microfiltrtion+

CaCl2

Microfiltrtion+

CaCl2

Reverse osmosisor

Nanofiltration

Reverse osmosisor

Nanofiltration

Granular activated carbon or

Biological treatment


Biological treatmentEffluentMacroemulsion

DecanterDecanter Decanted oil

Surfactant +CaCl2

Recycle

NanofiltrationNanofiltrationGranular activated

carbon orBiological treatment


Biological treatmentEffluentMicroemulsion

Surfactant +Oil

Fig. 3.6 Treatment processes for macro- and microemulsion, recommended by MATAMOROS


This thesis is an application research on the treatment of non-stabilized oil/water emulsion. As one part of her thesis, SRIJAROONRAT had studied on the application of ultrafiltration to treat non-stabilized emulsion. In her thesis, SRIJAROONRAT had tested several types of ultrafiltration membranes, i.e., organic, inorganic and ceramic. Wastewater used in the experiment consisted of synthetic emulsion of kerosene/water, cutting oil emulsion, crude oil/water emulsion, and wastewater from textile plant. The result can be summarized as follows.


1. Ultrafiltration process can be used to treat non-stabilized emulsion with remarkable efficiency (approx. 100%).

2. Major problem of the ultrafiltration is the rapid decrease in permeate flux, caused by;

• Presence of suspended solids in wastewater, these solids will deposit on the membrane surface and/or in the pores and cause clogging.


33 I-33

• Presence of surfactant, the surfactant will be adsorbed or form micelles within the membrane pores, as well as membrane surface, thus cause changing in wettability of membrane.

3. Influence of surfactant on permeate flux also depends on material of membrane. Inorganic (mineral) based membrane is more sensitive to surfactants than other membranes.


This thesis is an application research on the treatment of stabilized oil/water emulsion. WANICHKUL had studied on the application of ultrafiltration on cutting oil emulsion treatment. He also studied on the treatment of ultrafiltrate by reverse osmosis.


1. The result on the treatment of cutting oil emulsion by ultrafiltration confirms the result of MATAMOROS that the process is feasible and efficiency of the process is satisfying. The author recommended that operating condition should be in permanent regime to avoid influence of concentration factor.

2. Using microemulsion with under-saturated oil concentration is an effective method to regenerate the membrane.

3. The author had studied the efficiency of reverse osmosis on ultrafiltrate treatment. The result shows that the process provides remarkable efficiency.

4. The result of the comparison on efficiency between ultrafiltration and distillation on cutting oil emulsion treatment can be concluded that ultrafiltration is suitable for macroemulsion treatment, while distillation is more efficient for microemulsion treatment. However, from economic point of view, energy consumption of ultrafiltration is always lower than distillation’s.

3.8 Thermal treatment

This treatment approach makes use of physical properties of hydrocarbon, water and their mixture. Because hydrocarbon is only slightly soluble in water, we can say that the hydrocarbon/water mixture is practically an immiscible binary system. From this fact, we can apply the basic of phase equilibrium to develop thermal processes for hydrocarbon/water separation. Such processes are distillation and crystallization. All related theses based on thermal treatment can be outlined as follows.

3.8.1 Thesis of LUCENA[24]

This research is contributed to the heteroazeotropic distillation for treatment of slop, the wasted viscous water-in-petroleum emulsion that is usually very difficult and costly to treat. This process is very interesting innovation that allows hydrocarbon recovery by separating from water from the slop. Certain type of hydrocarbons, considered as efficient water extractant or entrainer, is added to the slop, then the slop undergoes distillation to extract the water. After distillation, the distillate obtained is composed of 2 layers of clear liquid, the entrainer in the upper layer and the water at the lower. The water can be disposed or further treated and the entrainer can be reused for next cycle of distillation. The residue obtained is relatively fluid, contains no water, and can be recycled to refinery unit as crude oil.


34 I-34


The experiment in laboratory scale was conducted with a simple glass distillation apparatus. The author also used the pilot-scale model for on-site treatment tests.


1. The result on the treatment of slop shows that the water is totally separated from the slop. The residue consists of relatively water-free hydrocarbon. Distillate consists of 2 separate layers of entrainer and water.

2. Observed quantity of entrainer is slightly different (5-8%) from theoretical value. So the theoretical method for calculating entrainer quantity is accurate enough for practical use.

3. Kerosene and gasoline are proven to be good entrainer. Their water-extracting capacities are almost identical to that of decane and dodecane, respectively.

3.8.2 Thesis of LORRAIN[23]

This research is contributed to crystallization. However, this process has just been studied by GPI lab for the first time. So we will not include it in this research.


This thesis is an application research on the treatment of stabilized oil/water emulsion. WANICHKUL had studied on the application of distillation for the treatment of cutting oil emulsion and permeate from ultrafiltration. He also studied the possibility to treat concentrate from ultrafiltration of cutting oil emulsion by heteroazeotropic distillation.


1. Comparison on the efficiency between ultrafiltration and distillation on cutting oil emulsion treatment, the author concluded that ultrafiltration is suitable for macroemulsion treatment, while distillation is more efficient for microemulsion treatment. However, from economic point of view, energy consumption of ultrafiltration is always lower than distillation’s.

2. Heteroazeotropic distillation is proven to be an effective way to treat “mayonnaise-like” retentate from the ultrafiltration of cutting oil emulsion. The residue of distillation process is clear liquid, composed of oil without water. The distillate consists of 2 separate layers of the entrainer (in this case, decane) on the top and the water at the bottom. However, TOD of the distillate is still high (around 2,800 mg/l). So the additional treatment of the distillate (either by biological treatment or RO) is required.

3.9 Chemical treatment

When we talk about chemical treatment in oil/water separation process, we normally refer to chemical destabilization, coagulation and flocculation. The process, then, does not “destroy” oil. Its major role is to transform the oil into the form that facilitates oil/water separation. So, chemical treatment process usually follows by physical separation processes as we discussed in previous sections. Normally, chemical treatment will be required when oil or


35 I-35

hydrocarbon is present in the form of very stable emulsion, which will not be naturally coalesced. So it is impossible or very difficult to separate them by mean of physical process alone. Example of oily wastewater that requires chemical treatment process is stabilized emulsion, such as cutting oil emulsion.

All of related theses on chemical treatment can be outlined as follows.

3.9.1 Thesis of ZHU[21]

This research is contributed to the study on influence of cutting oil emulsion formula on its physico-chemical treatment.


• Study on chemical destabilization of various macroemulsions and microemulsions • Study on treatment of residual pollution in treated emulsions after physico-

chemical treatment • Feasibility study on formulation of cutting oil emulsion to minimize the residual

pollution


The author had performed the experiment using 2 commercial macroemulsions and 2 microemulsions. Some used emulsions, provided by mechanical workshops, were also tested. For chemical reagents for destabilizing (so-called “breaking”) the emulsions, he had tested several chemicals, i.e.,

• Inorganic electrolytes: Sulfuric acid and several salts, such as NaCl, CaCl2, Alum, FeCl3

• Organic electrolyte: Ca(HCOO)2 • Commercial adsorption reagents • Cationic polyelectrolytes • Anionic polyelectrolytes • Cationic surfactants

Each chemical reagent was added to the emulsions, stirred, and left to decant for 20 hours. Decanting results were carefully observed. If the oil in the emulsion was separated easily within 1 hour, the author concluded that the reagent used for destabilization was effective.

For the study on the elimination of residual pollution, the author proposed to use adsorption treatment by activated carbon. The experiment had been conducted with batch and continuous lab scale models.


1. From the experiment on macroemulsion, it shows that destabilization mechanism of electrolyte depends on the number of its valences, which can be described as follows;


36 I-36

• Mono-valence cationic electrolyte, main destabilization mechanisms are based on Zeta potential reduction and promoting micelle formation of surfactants.

• Bi-valence cationic electrolyte, main destabilization mechanism is based on precipitation of surfactant to insoluble form.

• Tri-valence cationic electrolyte, main destabilization mechanisms are based on the bridging of its cations and hydroxides and precipitation of surfactant to insoluble form.

2. The result shows that destabilization techniques of fresh macroemulsion can also be effective for used macroemulsion.

3. For breaking emulsion by acid, the study shows that destabilization mechanisms for micro- and macroemulsion are different, based on type of surfactant used in the emulsions.

• For microemulsion- surfactant used is soap. So it is sensitive to pH change. Then, addition of acid will disturb the equilibrium of ionization of fatty acid and make it lose its surfactant property. This type of destabilization requires relatively small amount of acid.

• For macroemulsion- surfactant used is sulfonate type. Main destabilization mechanisms are based on Zeta potential reduction and promoting micelle formation of surfactants. So it requires high dosage of the acid.

4. Ferric chloride is an effective destabilization chemical for breaking macroemulsion. However, it cannot be used to breaking microemulsion because it will react with surfactants and form complex, which is difficult to separate.

5. High residual pollution which presents after breaking the emulsion is caused by co-surfactant, which is fatty acid that dissolves very well in water.

6. To destabilize the emulsion stabilized by non ionic surfactant, the author recommended to use a mixture of anionic surfactant and tri-valence electrolyte (Alum, FeCL3) as destabilization reagent. Main breaking mechanism is based on selective reaction between non-ionic surfactant, anionic surfactant and electrolyte, which form insoluble complexes.

7. Breaking mechanisms of commercial adsorption reagents are based on their components. So, their efficiencies will be known only by testing.

8. Activated carbon adsorption is proven to be an effective treatment for residual pollution which is left after breaking the emulsion. However, investment and operating costs of this system, esp. cost for regeneration or replacement, are relatively high.

9. The author had proposed, for the first time, to replace fatty acid co-surfactant with other surfactant with less water solubility to solve the problem about residual pollutant after breaking the emulsion. In his case, he used decanol, which is slightly soluble in water, as a co-surfactant. The new emulsion has satisfying property. But vigorous mixing by ultrasound device is required to disperse the oil into droplets to create the emulsion.

10. He suggested that new formula of this low pollution emulsion should be further studied to make it soluble instantly in water, which is the way the users prefer.


37 I-37

3.9.2 Thesis of YANG[22]

This research is a succession of ZHU’s thesis and contributed to development of low pollution emulsion. This thesis was a new approach in pollution control because it was one of the first researches on pollution reduction from the source, as it is presently called “cleaner product, CP” research, which is the branch of pollution control that is well-known and widely acceptable now.


• Study on the formulation of low pollution cutting oil emulsion • Study on the treatment process of that new emulsion


The author had tried to formulate low pollution cutting oil emulsion, using carefully selected base components, i.e.,

• Base oil: commercial naphthene (cyclic aliphatic compound) based oils and paraffin based oils,

• Surfactants: various succinic and sulfonate surfactants, • Co-surfactants: alcohol-based, ester-based, and other co-surfactants with low water

solubility, • Corrosion inhibitors: fatty acid alcanolamide, oleylsarcosinic acid, and fatty acid

polydiethanolamide, • Anti-mousse reagents: polysiloxane and a commercial reagent, • Bactericides: 4-chloro-3methylphenol, isothiazolinone, and parahydroxybenzoic

ester with phenoxyethylic alcohol,

Each emulsion formula was subjected to these 5 tests, i.e.,

• Droplet size distribution measurement: using particle analyzer (granulometer) • Emulsification test: new products were diluted by water to observe its solubility

and stability. • Anti-rust test: new emulsions of various concentrations were used to coat on steel

chip and rust forming was observed. • Anti-mousse test: new emulsions were manually and intensely agitated and

characteristic and stability of mousse were observed. • Determination of residual pollutant after chemical destabilization process or

ultrafiltration.


1. The author proposed new formulae of low-residual-pollutant microemulsions and macroemulsions.

2. These new emulsions can be treated efficiently by breaking with calcium chloride or ultrafiltration. Residual pollution after treatment is 2-57 times lower than present commercial products.

3. Concentration of emulsion and hardness of dilution water are important parameters, which affect stability of emulsion.


38 I-38

3.10 Biological treatment

Biological treatment is the most common wastewater treatment process. Because of its versatile, efficiency and economic, biological treatment plants are widely used for both domestic and industrial wastewater treatments. Its working principles depend mainly on degradation and/or assimilation of pollutants by microorganisms, so called “biomass”, present in the biological reactor. Key to obtain satisfying treatment efficiency is to optimize the environment in the reactor to promote function of biomass. Even though oil or hydrocarbons are biodegradable, presence of some hydrocarbons at some concentrations can cause adverse effect on the biological treatment mechanisms by;

• Obstructing oxygen transfer: Oxygen content is an important factor in aerobic biodegradation. Oil layer that covers water surface will obstruct transfer of oxygen between air and water. Oil may surround biomass floc and impede oxygen transfer of the floc in biological reactor. This leads to the decrease in treatment efficiency.

• Its high oxygen demand: Oil requires high amount of oxygen for biodegradation. It will cause problem on insufficient aeration capability in some treatment plants that did not account for this type of pollutant in the design procedure. It also increases investment cost and operating cost for aeration system.

• Its toxicity: In petroleum product or industrial oil-based product, they, sometime, contain toxic substance, such as sulfide, heavy metals, which may be added to the oil product in form of inhibitor or additive to provide desired product properties. When these products are discharged to water and become wastewater, these substance will become toxicity to living water organisms, includes biomass.

There are many studies, includes some from our lab, on the effect of oil contents in wastewater to biological treatment process and possibility or optimum condition to use biological treatment to treat oily wastewater. However, none of doctoral thesis, directed by M.Aurelle, is fully devoted to biological treatment of oil wastewater. So we will work on this process by reviewing the available data from our lab researches as well as outside data. Anyway, only our lab researches will be summarized here. Synthesized result from reviewing will be presented as a section of textbook in Part 3 of this thesis.


This all-purpose thesis is also contributed to biological treatment. The author had performed biodegradability test of permeate from ultrafiltration of cutting oil emulsion.


1. The study shows that ultrafiltrate of cutting oil emulsion can be treated by aerobic biological process. The result from biodegradability test shows very high TOC reduction efficiency (82 to 90% at the retention time of 2 to 5 hours).

3.11 Skimmer

When hydrocarbon is in the form of layer on water surface. It can be simply removed from the surface by mean of overflow weir, overflow pipe or be scooped out manually. There is one thesis on skimmer, which is the thesis of THANGTONGTAWI [5]. This research is


39 I-39

contributed to study on oil drum skimmer and disk skimmer, which are very effective to recover film or layer of oil or hydrocarbon on the surface of water.


• Influence of skimming materials, geometry of devices and characteristic of wastewater to skimmer operation

• Model development for oil drum and disc skimmer • Behavior of skimmers under interesting working conditions, which may affect its

efficiency


The experiment was conducted with various sizes of disc and drum skimmers to verify effect of skimmer geometry. Various types of skimming material were tested, i.e., stainless steel, PVC, polypropylene and fluorocarbon coated material. Kerosene and 2 types of lubricant oil, as well as real wastewater from 4 refineries, were used in the experiment. Effect on interfacial tension was also considered by varying concentration of a commercial surfactant.

Result

The result of this study provides models for sizing and calculating oil quantity, recovered by the skimmers. The study, also, provides significant criteria for skimmer material selection to obtain selective property, which allow the devices to recover only oil, not water. Design consideration, such as limitation of model, working condition to be avoided, etc. is included in this study.


1. In this thesis, mathematical models for sizing and calculating oil productivity of the skimmer are proposed.

2. Influence of skimming material is studied and can be summarized as follow,

• Material of high surface energy, such as stainless steel, which is conditioned by submerging in oil, will effectively recover or separate the oil from water. However, when oil film on the surface of the tank is broken and the skimmer is exposed to water, it will start recover the water immediately. Even after the oil layer is present again, it will not resume its function of oil recovering.

• Material of low surface energy, such as PVC and PP, can effectively recover oil. But it will start recovering water after the oil film is broken and the skimmer is allowed to expose to the water surface for some times. Anyway, it will resume its function after the presence of oil layer.

• From the study, fluorocarbon coated material, which is patented by ELF, is the best oil-water selectivity. It always recovers only oil.

3.12 Application researches

To apply treatment processes previously described in real life situation, our lab had performed some researches that devoted to compare the efficiency of various stand-alone processes or combination of processes on treatment of certain type of oily wastewater. This


40 I-40

kind of research provides very useful information for process selection or process design. All of related theses on membrane process can be outlined as follows.


This thesis is an application research on the treatment of non-stabilized oil/water emulsion, which is one of the most encountered wastewaters.


• Study on treatment of non-stabilized emulsion by ultrafiltration • Influence of operating condition, presence of surfactant and type of membrane on

ultrafiltration process • Study on treatment of non-stabilized emulsion combination of coalescer/

hydrocyclone


The experiment on ultrafiltration was conducted with 3 models, i.e. commercial batch test module (effective area 37 cm2), lab scale cross flow model (effective area 100 cm2) and pilot scale cross flow model (effective area 1 m2). The non-stabilized emulsions tested consisted of synthetic emulsion kerosene/water and crude oil/water emulsion. However, the author also tested the emulsion with presence of surfactant to study the effect of surfactant on ultrafiltration. These stabilized emulsions were cutting oil emulsion and wastewater from textile factory, which contained oil, water, dye and some surfactant from cleansing chemical in form of emulsion.

For ultrafiltration membrane, she had tested various membrane materials, i.e., inorganic, organic and ceramic. She also used 2 forms of membrane, i.e. plain and tubular.

For combination of coalescer/hydrocyclone, two types of coalescer bed were used, bottle brush type and disorderly fibrous bed (like metal wool). For hydrocyclone, a commercial hydrocyclone, model DOXIE 5 of Dorr-Oliver, was used.


Results of this thesis are described in previous sections, corresponding to each process that had been studied in this thesis. However, to get the whole picture of this thesis, we will summarize overall significant findings of this thesis again as follow;

1. Ultrafiltration process can be use to treat non stabilized emulsion with remarkable efficiency (approx. 100%).

2. Major problem of the ultrafiltration is a rapid decrease in permeate flux, caused by;

• Presence of suspended solids in wastewater: these solids will lodge tightly in membrane pores and cause clogging.

• Presence of surfactant: the surfactant will be adsorbed or form micelles within the membrane pores, as well as membrane surface, then cause changing in wettability of membrane.


41 I-41

3. Influence of surfactant on permeate flux, also, depends on material of membrane. Inorganic (mineral) based membrane is more sensitive to surfactants than other membranes.

4. Backflushing the membrane with its permeate is proven to be an effective way to increase the limiting flux. The result shows that duration of ultrafiltrating and backflushing cycle should be optimized to obtain best efficiency. Normally, the 2 cycles is preferred to be short.

5. At high empty bed velocity, the coalescer with random bed will coalesce and enlarge the droplets into relatively large drop, while the coalescer with brush type bed tends to produce stream of jet, containing small oil drop.

6. However, the disorderly fibrous (steel wool) bed coalescer tends to be clogged by suspended solids, usually presence in the oily wastewater. So the author proposed new configuration of fibrous bed in form of the combination of 2 brushes. The internal one is of ordinary brush. The external one will look like coil spring with its fiber elements protruding inward and toward the center. This type of bed is believed to provide good interception, similar to the disorderly bed, yet remain its anti-clogging properties, like brush-type bed.



This thesis is an application research on the treatment of stabilized oil/water emulsion, one of the most encountered oily wastewaters. The author also extended the research of SRIJAROONRAT [10] on the combination of coalescer and hydrocyclone by studying the influence of order of the 2 processes. So, WANICHKUL tested both combinations of coalescer/hydrocyclone and hydrocyclone/coalescer as well as the combination on coalescer/hydrocyclone/coalescer.


• Study on treatment of stabilized emulsion by ultrafiltration and distillation • Study on treatment of permeate from ultrafiltration by reverse osmosis, biological

treatment and distillation • Study on treatment of retentate from ultrafiltration by heteroazotropic distillation • Study on treatment of non-stabilized emulsion combination of coalescer/

hydrocyclone and hydrocyclone/coalescer • Study on multi stage fibrous bed coalescer


The experiment on ultrafiltration was conducted with 3 models, i.e. commercial batch test module (effective area 37 cm2), lab scale cross flow model (effective area 100 cm2) and pilot scale cross flow model (effective area 1 m2). For reverse osmosis, he used commercial test module from OSMONIC co., ltd.

The stabilized emulsions tested consisted of 2 commercial cutting oil emulsions, 1 macroemulsion and 1 microemulsion. For membrane regeneration, he used the microemulsion


42 I-42

used in the experiment as a cleansing reagent. He also used water to rinse the membrane after each test.

For ultrafiltration membrane, he used the asymmetric type organic membrane from ORELIS S.A., pore size 50 Kdaltons. For reverse osmosis, he used hydrophilic organic membrane, cut size 150-200 Daltons.

For distillation, lab-scale glass distillation apparatus was used. For biodegradation, the apparatus, custom-made by our lab, called “Respirometer BIOS-R” was used. This apparatus can track reduction of organic material and convert to biodegradability.

For hydrocyclone/coalescer study, co-current transparent hydrocyclone and “bottle brush” fibrous bed coalescer were used. He also tested a new type of fibrous bed, multi stage fibrous bed”. In fact, the new bed is modified version of bottlebrush type with the rings of fibers not placed close to each other, but placed separately.


Results of this thesis are described in previous sections, corresponding to each process that had been studied in this thesis. However, to perceive the whole picture of this thesis, we have summarized overall significant findings of this thesis again as follows;

1. The result on the treatment of cutting oil emulsion by ultrafiltration confirms the result of MATAMOROS that the process is feasible and the efficiency of the process is satisfying. The author recommended that operating condition should be in permanent regime to avoid influence of factor of concentration.

2. Using microemulsion with under-saturated oil concentration is an effective method to regenerate the membrane.

3. The author had studied the efficiency of reverse osmosis on ultrafiltrate treatment. The result shows that the process provides remarkable efficiency.

4. Comparison on efficiency between ultrafiltration and distillation on cutting oil emulsion treatment, the author concluded that ultrafiltration is suitable for macroemulsion treatment, while distillation is more efficient for microemulsion treatment. However, from economic point of view, energy consumption of ultrafiltration is always lower than distillation’s.

5. Heteroazeotropic distillation is proven to be an effective way to treat “mayonnaise-like” retentate from ultrafiltration of cutting oil emulsion. The residue of distillation process is clear liquid, composed of oil without water.

6. The research results shows that key parameters that governs the efficiency of co-current hydrocyclone are;

• Influent flowrate • Recovered oil draw-off rate • Ratio between outlet velocity of oil and outlet velocity of treated water,

calculated at each corresponding outlet port

7. The author recommended that the velocity ratio mentioned above should be greater than 1.5.

8. Testing of the 2 types of the combination in industrial pilot plants shows satisfying result. These combinations provide promising alternatives for


43 I-43

industrial wastewater treatment process. Furthermore, advanced combination of coalescer-hydrocyclone-coalescer is tested. It promisingly shows that 2 coalescers can be used, one before the hydrocyclone to increase droplet size entering the cyclone, and another at downstream of the cyclone to coalesce the concentrated oil/water mixture from the outlet of the hydrocyclone

9. Result from Multi stage fibrous bed coalescer test shows that this “modified bottlebrush” bed provides better efficiency than the classic bottlebrush bed. This is because the gaps between each stage can cause turbulence, then promote coalescence between oil droplets that, somehow, fail to coalesce in the previous stage of bed.


44 I-44

Chapter 4 Conclusion

In this part, we have reviewed all of the researches in our lab, directed by Professor AURELLE. We can see the attempts to study many aspects of treatment processes to cope with various type of oily wastewater. The results of each thesis make us understand working principles and limitation of the processes in many specific cases or frameworks. In the next part, we will analyze these specific data from these theses. And then try to integrate and generalize them to formulate design criteria or mathematical models that can be used with the entire (or as wide as possible) range of oily wastewater.

Part II Generalization of models for oil-water separation process design


II-i

Contents Page

Part II Generalization of models for oil-water separation process design Chapter 1 Decanting

1.1 Simple decanter or API tank II-2 1.2 Lamella decanter or Parallel Plate Interceptor (PPI) II-3 1.3 Model verification II-4 1.4 Conclusion and generalized model of decanter II-7

Chapter 2 Skimmer 2.1 Drum skimmer II-9 2.2 Disk skimmer II-10

Chapter 3 Coalescer 3.1 Granular bed coalescer II-11

3.1.1 Filtration-based model II-11 3.1.2 Dimensional analysis-based model II-12 3.1.3 Model verification II-13 3.1.4 Conclusion and generalized model of granular bed II-14

coalescer 3.1.5 Generalized model for guide coalescer II-15 3.1.6 Generalized model for mixed bed coalescer II-16 3.1.7 Generalized model for pressure drop of granular bed II-16

coalescer and guided coalescer 3.2 Fibrous bed coalescer II-18

3.2.1 Dynamic fibrous bed coalescer model II-18 3.2.2 Simple fibrous bed coalescer model II-18 3.2.3 Model verification II-18 3.2.4 Conclusion and generalized model of fibrous bed coalescer II-21 3.2.5 Generalized model of random or disorderly fibrous bed II-23

coalescer 3.2.6 Generalized model for pressure drop of fibrous bed II-24

coalescer

Chapter 4 Dissolved air flotation 4.1 Dissolved Air Flotation (DAF) model for oily wastewater II-25

treatment 4.2 Model verification II-26

4.2.1 Modification of filtration-based model II-26 4.2.2 Population balance method II-28

4.3 Generalized model for DAF II-30 4.4 Generalized equations for pressurized water system calculation II-35


5.1 Two-phase hydrocyclone II-37 5.1.1 Trajectory analysis-based model II-37 5.1.2 Other models II-38 5.1.3 Model verification II-39

Contents

II-ii

Contents (Con’t) Page

5.1.4 Conclusion and generalized model of two-phase II-40 hydrocyclone

5.1.5 Generalized model for pressure drop of two-phase II-41 hydrocyclone

5.2 Three-phase hydrocyclone II-43 5.2.1 Model development and verification for liquid-liquid II-43

section 5.2.2 Model development and verification for solid-liquid section II-45 5.2.3 Generalized Model for pressure drop of three-phase II-46

hydrocyclone

Chapter 6 Membrane process 6.1 Ultrafiltration II-49

6.1.1 Resistance model II-50 6.1.2 Film theory based model II-52 6.1.3 Model verification II-53 6.1.4 Flux prediction for mixture of cutting oil microemulsion II-58

and macroemulsion 6.1.5 Theoretical flux prediction for batch cross-flow UF process II-60 6.1.6 UF efficiency II-63 6.1.7 Minimum and maximum transmembrane pressure and II-64

power required 6.1.8 Conclusion and generalized model of UF II-66

6.2 Nanofiltration and Reverse osmosis II-66

Chapter 7 Heteroazeotropic Distillation 7.1 Theoretical model II-69 7.2 Model verification II-72 7.3 Conclusion and generalized model of heteroazeotropic distillation II-72


II-iii

Table Page

Table 1.1 Summary of tested decanters and operating conditions II-6 Table 6.1 Summary of parameters of resistance model from UF researches on II-51

oilywastewater treatment (reference temperature = 20O C) Table 6.2 Summary of parameters of film model from UF researches on oily II-54

wastewater treatment (reference temperature = 20O C) Table 6.3a Summary of RO data on oily wastewater treatment II-67 Table 6.3b Summary of NF data on oily wastewater treatment II-68 Table 7.1 Heterotropic temperature and composition from various extractants II-72

Contents

II-iv

Figure Page

Fig. 1.1 Schematic and typical removal efficiency curve of simple decanter II-3 Fig. 1.2 Schematic of PPI decanter II-4 Fig. 1.3 “Spiraloil” decanter a) Simple spiral b) Mixed spiral II-5 Fig. 1.4a Comparison between observed efficiency (1',2',3') and predicted II-6

efficiency (1,2,3) for Simple Spiral "Spiraloil" decanter Fig. 1.4b Comparison between observed efficiency (1',2') and predicted II-6

efficiency (1,2) for Mixed Spiral "Spiraloil" decanter

Fig. 2.1 Drum and disk skimmer II-10 Fig. 3.1(a) Schematic diagram of granular bed coalescer, (b) photo of bed material II-11

with coalesced oil on their surface and (c) coalesced oil drops at the discharge surface of bed

Fig. 3.2a Relation between droplet diameter VS. model's error II-14 Fig. 3.2b Comparison between observed efficiency and predicted efficiency from II-14

DAMAK's model Fig. 3.3 Relation between observed pressure drop of granular bed coalescer and II-17

predicted upper & lower limits from Kozeny-Carman's porosity = 0.13 and 0.23 Fig. 3.4 Comparison between observed efficiency and predicted efficiency from II-19

SRIJAROONRAT's model, Verified by MA's and WANICHKUL's data

Fig. 3.5 Comparison between observed efficiency and predicted efficiency by II-

20TAPANEEYANGKUL's model for simple fibrous bed (Assume rotating speed = 450

rpm)

Fig. 3.6 Comparison between observed efficiency and predicted efficiency from II-21

modified SRIJAROONRAT's model (eq. 3.8b) Fig. 3.7 Relation between observed efficiency and predicted efficiency from random II-23(or

disorderly) fibrous bed coalescer model and simple fibrous bed model

Fig. 4.1 Schematic diagram of DAF II-25 Fig. 4.2 Relation between theoretical efficiency factor and observed efficiency II-27

factor Fig. 4.3 Relation between obseved efficiency and predicted efficiency of DAF II-27

from modified SIEM's model Fig. 4.4 Relation between absolute pressure in pressure tank and dissolved quantity II-36

and released air volume

Fig. 5.1 Schematic diagram and trajectories of droplets in two-phase hydrocyclone II-37 Fig. 5.2 Comparison between observed efficiency and predicted efficiency from II-39

Ma's and Thew-Colman's models

Fig. 5.3a Relation between observed pressure drop (inlet/overflow) of Thew cyclone II-42

and predicted pressure drop


II-v

Fig. 5.3b Relation between observed pressure drop (inlet/underflow) of Thew cyclone II-42

and predicted pressure drop

Fig. 5.4 Three-phase hydrocyclone II-43 Figure (Con’t)

Page

Fig. 5.5 Relation between observed efficiency and predicted efficiency of II-44

liquid-liquid (Thew) part of three-phase hydrocyclone

Fig. 5.6a Relation between observed pressure drop (bar) across inlet and water outlet II-47 and predicted value from 2 approaches

Fig. 5.6b Relation between observed pressure drop (bar) across inlet and SS outlet II-48

and predicted value from 2 approaches Fig. 5.6c Relation between observed pressure drop (bar) across inlet and oil outlet II-48

and predicted value from 2 approaches

Fig. 6.1 Typical schematic diagram of Cross-flow membrane process II-49 Fig. 6.2 Typical relation between flux and various parameters II-52 Fig. 6.3a Relation between flux and concentration at any recirculation velocity [38] II-52 Fig. 6.3b Typical characteristic curve of concentration VS. Flux II-

53(Log-Normal scale) [38] Fig. 6.4 Relation between UF permeate flux and Transmembrane pressure at II-56

reference concentration (C) of = 4%, V = 1.4 m/s and Predicted relations at C = 2 and 8% Fig. 6.5 Relation between observed and predicted flux by resistance model for II-57

ultrafiltration of macroemulsion 4% conc. and extend to cover other conc. by film model Fig. 6.6 Comparison between predicted flux and observed flux for UF of micro/ II-59

macroemulsion mixture (Conc. shown as % by volume of concentrate) Fig. 6.7 Relation between observed and predicted flux of micro/macroemulsion II-60

mixture from weighted average method between flux of whole micro and macroemulsion at the same total oil

concentration

Fig. 6.8a Relation between Flux VS. theoretical and observed permeate volume II-62 Fig. 6.8b Relation between time VS. theoretical and observed permeate volume II-62 Fig. 6.8c Relation between theoretical flux VS. concentration of oil in retentate II-63

Fig. 6.9 Relation between oil concentration VS. viscosity of emulsion II-65 Fig. 7.1 Isobar equilibrium diagram : Temperature-Concentration characteristic of II-70

immiscible binary mixture Fig. 7.2 Graphical method to find heteroazeotropic temperature II-70 Fig. 7.3 Graphical method to find dew curves II-71

Part II Generalization of model for oil-water separation process design

II-1

Part 2 Generalization of model for oil-water separation process design

In this part, the models or researches collected and summarized in the part I are analyzed and then generalized to obtain the model or models that governs the entire range (or as wide as possible) of oily wastewater that we are interested in. Basic concept of each model is briefly described here providing that it will be fully described in the next part. Then the models and their limitations will be presented. After that, models will be tested using various sets of data to cover the considered range of wastewater. Finally the generalized model will then be proposed.

45

Chapter 1 Decanting

II-2

Chapter 1 Decanting

Decanting or sedimentation is a non-accelerated process, which every input parameters in STOKES law are not modified. According to the researches reviewed in Part I, there are several types of decanters studied in GPI laboratory, i.e.,

1.1 Simple Decanter or API tank

Simple decanter, which is made well known and standardized by American Petroleum Institute (API), is the simplest oil-water separation process based on classical STOKES law. Concept of operation of the process is to provide sufficient time for droplets to float to the surface, where it will accumulate into oil layer, before it flows out with the water at the water outlet.

The model that governs the operation of the process is derived from comparing the time required for the droplet to reach the surface with retention time of the tank. Fig. 1.1 shows again the diagram of decanting process. From the figure, the longest path to reach the surface is the path starts at the bottom of the tank. The smallest droplet size that can reach the surface is called the cut size. The droplet of cut size or bigger is always separated from wastewater stream with 100% removal efficiency.

The smaller droplet can be also separated providing that it enters the tank near the water surface. When uniformly distributed influent flow is valid, which is true for almost all of properly designed tank, the removal efficiency of the droplets smaller than cut size is proportional to its corresponding rising velocity. From these concepts, the models of decanting process are as shown in eq. 1.1 to 1.4.

⎟⎠⎞

⎜⎝⎛=

SQ

dcU {1.1}

From STOKES law

c18μ

2dgΔρdU ⋅⋅= {1.2}

Then 1/2

SgΔρc18Qμ

cd ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⋅

= {1.3}

For oil droplet size, d ≥ cut size,dc

100%dη = {1.4a}

For oil droplet size, d ≤ cut size,dc

100%dcUdU

dη ⋅= {1.4b}

( ) 100%maxd

mind odCdη1

tη ⋅∑ ⋅=oCoutQ

Q {1.4c}

∑−=max

min

d

dodd

dout CQQQ η

ρ {1.4d}

46


II-3

UV

Q

d = cut size

d < cut size


d

ηd

d c

d < d c

Zone 1 Zone 2

d = or > d c

Fig. 1.1 Schematic and typical removal efficiency curve of simple decanter

Typical characteristic of the removal efficiency of decanter is as shown in fig. 1.1. The equations are valid while these conditions are satisfied i.e.,

1. Reynolds number, Re, of droplet is between 10-4 to 1, which is the range that STOKES law is valid.

2. The oil droplets are uniformly distributed across the cross section area of the tank.

3. The oil droplet is spherical, which is normally true.

1.2 Lamella Decanter or Parallel Plate Interceptor (PPI)

This type of decanter is the basic modification of the simple decanter. The concept is to decrease the rising distance of droplet to intercepting surface without decreasing the retention time. This can be achieved by inserting plates into the simple decanter to act as the interceptors for the rising oil droplets. Rising or travelling distance (H) is the distance between the plates, not the depth of water (D) as shown in fig. 1.2. Some times, the plates are inclined with angle (α) as related to horizontal axis. In this case, rising distance will become H / cos α.

The model that governs the operation of the process is modified from the model of simple decanter, as shown in eq. 1.5. From the equation and figure, it can be implied that the simple tank is divided into (N+1) small decanters.

47

Chapter 1 Decanting

II-4

Q

Separated oil floats to surface

Influent Effluent

L

D

H

Inserted plates (No. of plates = N)

Fig. 1.2 Schematic of PPI decanter

1/2

1)(NPΔρgSc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

+= {1.5}

Where Sp = Inserted plate area N = Numbers of inserted plates L = Length of inserted plates H = Rising distance of oil drop, in this case, distance between inserted

plates α = Inclination angle, related to horizontal axis

When incline plates are used, replace Sp with Sp cos α.

Removal efficiency can be calculated using eq. 1.4. Typical characteristic of the removal efficiency of decanter is identical to simple decanter’s, as shown in fig. 1.1.But the cut size of PPI tank will be smaller than the simple decanter’s, providing that they are the same size. The equations are valid while these conditions are satisfied i.e.,

1. The tank is operated under laminar flow regime. Reynolds number, Re, is between 10-4 to 1, which is the range that STOKES law is valid.

2. The oil droplets are uniformly distributed across the cross section area of the tank.

3. The oil droplet is spherical, which is normally true.

4. The plates are identical in size and are inserted evenly and in parallel.

1.3 Model verification

According to the concept of decanting model that derives from comparing detention time with rising time of droplet to intercepting surface, we can formulate the general model to calculate dc for any decanters as shown in eq. 1.9

General model for calculating the cut size, when H, L and A can be clearly defined. 1/2

ΔρgLAc18HQμ

cd ⎟⎟⎠

⎞⎜⎜⎝

⎛= {1.9a}

48


II-5

Where Q = Wastewater flowrate H = Rising distance of oil drop, depends on configuration of the decanter L = Length of interceptor surface A = Flow area (Cross sectional) area of decanter

Removal efficiency can be calculated using eq. 1.4. From the equations, we can see that the models of simple decanters and PPI are modified forms of eq. 1.9 by simple relations of flow velocity and tank geometry. However the models described before are derived from basic rectangular tank and flat insertion plates.

But, in real life situation, decanters are designed or produced in various forms, such as corrugated plate inserted tank, concentric annular insertion decanter, etc. So, sometimes, it is very difficult to define H. Then, we propose to simplify the general model by neglecting complicate analyzing to define H, and using concept of decanting area (Sd) instead. Sd is calculated from the sum of every surface area within the decanter that can intercept oil without considering whether the values H of these areas are identical or not. The other form of general model is shown in eq. 1.9b.

1/2

dSgΔρc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

⋅⋅= {1.9b}

To verify the theoretical model, we select the research of CHERID [4] on 2 sets of the “SPIRALOIL” decanter, i.e.,

• “Simple Spiraloil”; fabricated from concentric annular plates as shown in fig. 1.3a • “Mixed-spiral Spiraloil”; ”; fabricated from concentric annular plates with

corrugated plates spacer as shown in fig. 1.3b

We will use these decanters to compare the removal efficiency calculated from model to experimental result from the research. Geometry of the decanter and operating parameters used in the experiment are summarized in Table 1.1 and Annex A1.

Solid core, radius = r

R

H

e

Annular plates

a) b)

Fig. 1.3 “Spiraloil” decanter a) Simple spiral b) Mixed spiral

49

Chapter 1 Decanting

II-6

Table 1.1 Summary of tested decanters and operating conditions

Description Simple spiral “Spiraloil” Mixed spiral “Spiraloil”

Geometry Nominal radius R (mm) 50 50 Core radius r (mm) 10 10 Length of decanter L (mm) 300 300 Total length of inserted plates (m) 3.88 N/A Spacing between concentric annular plate (mm) 2 3 Plate thickness (mm) 0.3 0.2 for annular plates

0.4 for corrugated plates Interceptor (decanting) area (m2) 1.66 2.1 Inclination of decanter Horizontal Horizontal Operating conditions Flow velocity (cm/s) 0.4-1.6 0.5-1.5 (m/h) 14.4-57.6 18-54 Wastewater used Kerosene-water mixture Kerosene-water mixture

0.00%

10.00%

20.00%30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

0 10 20 30 40 50 6

Droplet diameter (micron)

Rem

oval

eff

icie

ncy

(%)

0

1'1 3'

2 2'

3

V = 1.6 cm/s (3, 3')

V = 0.8 cm/s (2, 2')

V = 0.4 cm/s (1, 1')

Fig. 1.4a Comparison between observed efficiency (1',2',3') and predicted efficiency (1,2,3) for Simple Spiral "Spiraloil" decanter

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

0 10 20 30 40 50 6


Rem

oval

eff

icie

ncy

(%)

0

1 (V = 0.5 cm/s)1'

2' 2 (V = 1.5 cm/s)

Fig. 1.4b Comparison between observed efficiency (1',2') and predicted efficiency (1,2) for Mixed Spiral "Spiraloil" decanter

50


II-7

Comparison graphs between observed efficiency and predicted efficiency from model for both Spiraloil decanters are shown in fig. 1.4. These graphs show that;

1. Predicted values for cut size are relatively accurate.

2. Predicted efficiencies of the droplets smaller than cut size are always lower than observed value because, in our simplified model, coalescing between rising oil drops is not accounted. Furthermore, in case that the plates are placed very close to each other like in “Spiraloil”, oil film will accumulated at the surface of plate, then helps reducing rising height oil droplet, thus increasing the efficiency.

3. Correction factor for efficiency prediction of these small droplets may not be established accurately. However, the predicted cut size can be used to design the tank with relative high accuracy. So it is reasonable to select the cut size to cover the majority of droplet size distribution. The predicted efficiency of smaller droplets, which is the minority part, will cause no harm but slightly underestimation on the total efficiency.

1.4 Conclusion and Generalized Model of decanter

From model verification result, we can conclude and propose the generalized model, as well as, its limitation as follows,

1. To solve oil or hydrocarbons removal efficiency of decanter, the cut size of the decanter will be determined first. Then, graded efficiency (efficiency of each size of droplet) and then total removal efficiency can be determined.

2. The cut size of the decanter can be determined from eq. 1.9. When configuration of decanter is not complicate and rising distance of oil drop to interceptor can be clearly determined, Using eq. 1.9a will give very accurate prediction. However, when configuration of the decanter is so complicate to determine the rising distance accurately, eq.1.9b provides relatively accurate result for the cut size. For PPI tank, the cut size can be calculated from eq. 1.5, which is the modified form of eq. 1.9a.

1/2

ΔρgLAc18HQμ

cd ⎟⎟⎠

⎞⎜⎜⎝

⎛= {1.9a}

1/2

dSgΔρc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

⋅⋅= {1.9b}

1/2

1)(NPΔρgSc18Qμ

cd⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

+= {1.5}

3. To determine graded and total removal efficiency, use eq. 1.4.

For oil droplet size, d ≥ cut size, dc

100%dη = {1.4a}

51

Chapter 1 Decanting

II-8


100%dcUdU

dη ⋅= {1.4b}

For total removal efficiency

( ) 100%maxd

mind odCdη1

tη ⋅∑ ⋅⋅⋅=oCoutQ

Q {1.4c}

∑−=max

min

d

dodd

dout CQQQ η

ρ {1.4d}

4. To use the models described above, the following conditions need to be satisfied and the assumptions and limitations would be noted.

1) Reynolds number, Re, of oil droplet is between 10-4 to 1, which is the range that STOKES law is valid.

2) The oil droplets are uniformly distributed across the cross section area of the tank, which can be achieved by proper design of inlet chamber. And the oil droplet is spherical, which is normally true.

3) For PPI or others forms of plate inserted decanter, the plates are identical in size and are inserted evenly and horizontally. For PPI tank with incline plates, Sp in eq. 1.5 will be replaced by Sp cos α. α is the inclination angle, related to horizontal axis.

4) If the decanter or the inserted plates are inclined, the rising distance will be the spacing between plates, but will be measured in vertical direction. So the shortest distance is obtained from the same spacing between plates, when the plates are located horizontally.

5) Prediction of cut size from eq.1.9b, even with its simplification, is relatively accurate for the cut size larger than 20 microns.

6) For droplets smaller than 20 microns, they are subject to Brownian motion and cause error in the prediction of the efficiency. So it is recommended to avoid using the decanter for the wastewater with majority part of oil droplets smaller than 20 microns. However, if these small droplets are the minority part of pollutants, the models can be used to predict the efficiency without any harm because its prediction is usually lower than observed value, thus make the prediction result on the safe side.

52


II-9

Chapter 2 Skimmer

Skimmer is the equipment designed to remove oil film or layer from water surface. It is superior to basic hydraulic elements such as bell-mouth pipe or weir for its oil-water selectivity. It will remove relatively water-free oil. According to the researches reviewed in Part I, there are 2 types of skimmers studied in GPI laboratory, i.e., drum skimmer and disk skimmer.

2.1 Drum skimmer

Drum skimmer is rotating drum or cylinder with specially selected surface material, which can adhere to oil, not water. The oil will adhere to the drum as a film and, then, lift up from water surface by The skimmer’s rotating movement. The skimmer will be equipped with scrapper blade to scrape the oil film from drum surface into receiving trough or gutter. Fig. 2.1 shows the schematic of oil drum skimmer.

The generalized model of drum skimmer had been proposed by THANGTONGTAWI [5]. His proposed model was developed by similarity analysis, using Buckingham Pi theory. He had conducted his research thoroughly and covered every necessary aspect of drum skimmer. So his model and its limitation will be quoted here again, as shown in eq.2.1, without the need for further verification.

0.514g

L0.486oν

1.541N1.5413.035DP = {2.1}

100%tη = {2.2}

The model will be valid when these conditions are satisfied, i.e.;

• Superficial tension of oil is in the range of 27 – 34 dynes/cm, which practically covers all common oil.

• Capillary number (Ca = μo V/γo) is in the range of 0.2 – 1.0.

• Oil density is around 0.8 g/cm3.

• Peripheral or tip velocity should not be greater than 0.8 m/s. To avoid water entraining, velocity of 0.44 m/s or less is recommended.

• Recommended immersion depth is 1.0-2.0 cm.

• Presence of surfactant has only little effect on skimmer’s operation.

• Total removal efficiency of the equipment is always 100%, providing that above conditions are satisfied, and then recovered oil is water-free.

53

Chapter 2 Skimmer

II-10

ScrapperD

Oil layerWater

I

Scrapper

Oil layerWater

a) b)

Fig. 2.1 Drum and disk skimmer

2.2 Disk skimmer

Disk skimmer’s operating principle is the same as drum skimmer’s. Major difference is the replacement of the drum with thin circular disk. fig. 2.1 shows the schematic of oil disk skimmer.

The generalized model of disk skimmer had been proposed, also, by THANGTONGTAWI [5]. His proposed model, as shown in eq. 2.3, was developed by similarity analysis, using Buckingham Pi theory.

0.332g

1.17I0.452oν

1.212N1.2581.328DP = {2.3}

100%tη = {2.2}




• Oil density is around 0.8 g/cm3.


• Presence of surfactant has only little effect on skimmer operation.

• Total removal efficiency of the equipment is always 100%, providing that above conditions are satisfied, and then recovered oil is water-free.

54


II-11

Chapter 3 Coalescer

Coalescer is an accelerated separation process that is designed to promote coalescing of oil droplets into bigger oil drop, which can be separated easily by relatively small decanter. There are several types and modification of coalescer studied in GPI laboratory, i.e.,

3.1 Granular bed coalescer

This type of coalescer uses granular material as a bed to promote coalescing between oil droplets, as shown in fig. 3.1. AURELLE [3] is the pioneer on coalescer study. He established the 3-step working mechanisms of the coalescer, based on filtration model [35]. Then there are several follow up researches on coalescer, based on AURELLE’s research. However, for mathematics model of coalescer, there are only 2 major mathematical models for coalescer. The first one is based on filtration concept and another is based on dimensional analysis.

Collector size =dp,

Void ratio = ε

V

H

IN:Micro drop

Dia. = d

OUT:Large drop Discharge

screen

Inlet screen/support

V

a) b) c)

Fig. 3.1 (a) Schematic diagram of granular bed coalescer, (b) photo of bed material with coalesced oil on their surface and (c) coalesced oil drops at the discharge surface of bed

3.1.1 Filtration-based model

The method is proposed by AURELLE [3]. In his study, AURELLE divided the mechanisms taking place with in coalescer bed into 3 steps, i.e.;

• Step 1: Interception, which consists of 3 major transport phenomena, i.e., sedimentation, direct interception and diffusion.

• Step 2: Adhesion-Coalescence. The efficiency of this step depends mainly on wettability of bed material.

• Step 3: Salting out or enlargement of coalesced liquid. This step depends on 4 parameters, i.e., wettability of bed material at the discharge surface (or screen), empty bed flow velocity, interfacial tension and ratio of dispersed phase and continuous phase in emulsion treated.

AURELLE suggested that step 2 and step 3 can be optimized by using oleophilic and hydrophilic material as coalescer bed and discharge screen, respectively, as well as keeping the feed flowrate within optimum range. Then step 1 will become rate determining step of the

55

Chapter 3 Coalescer

II-12

coalescer. To calculate removal efficiency of step 1, he proposed the model, based on filtration model, as shown in eq. 3.1 to 3.3.

%100)expη(α

dpHε)(1

23

e1dη ⋅

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛ ⋅−−−= {3.1}

0.5143)theo0.5484(ηexpηα =⋅ {3.2}

2/3

dpVdcμKT0.92)

dpd(

23

Vc18μ

2Δρgdtheoη ⎟⎟

⎠

⎞⎜⎜⎝

⎛⋅

++= {3.3}

( %maxd

mi)

nd odCdη1

tη ∑ ⋅⋅⋅=oCoutQ

Q {3.4a}

∑−=max

min

d

dodd

dout CQQQ η

ρ {3.4b}


• The model is valid when the shape of the collector is relatively spherical.

• The collector shall be wetted by dispersed phase. In case of direct emulsion (oil in water emulsion), the collector, then, shall be oleophilic. In this case, oleophilic resin is recommended.

• Range of empty bed velocity shall not be greater than 0.35 cm/s (12.6 m/h)

• Density difference between dispersed phase and continuous phase shall be approximately 200 kg/m3.

• The model is developed for inlet oil concentration between 100-200 mg/l.

• At velocity < 0.35 cm/s, efficiency of the coalescer is independent of velocity. Beyond this range, The efficiency will decrease when the velocity increases. The rate of the decreasing of efficiency varies with size and wettability and surface roughness of collector.

• The key assumption of this model is that mechanisms in steps 2 and 3 of the coalescer are optimized.

This research has been studied thoroughly and covered every important parameter. But the last assumption, related to mechanisms in steps 2 and 3, is difficult to verify. The evident is clearly shown in eq.3.2 where the relation between ηexp and ηtheo is not linear. This means there are other factors, besides the 3 interception steps, which should be included in the ηtheo. However this filtration-based model is good for understanding the effect of various parameters on coalescer efficiency.

3.1.2 Dimensional analysis-based model

The method is proposed by DAMAK [9]. The model is based on classic dimensional analysis, which is the efficient tool when exact theory of the processes can not be established, as described in the section 3.1.1. DAMAK’s model is shown in eq.3.5.

56


II-13

%1000.09)cρ

Δρ(0.09)μcμd(0.08)

o/wγ

dpVcρ(0.12)dpH(0.2)

dpd0.58(dη ⋅⎟

⎟

⎠

⎞

⎜⎜

⎝

⎛ −= {3.5}

( %maxd

mi)

nd odCdη1


Q {3.4a}

∑−=max

min

d

dodd

dout CQQQ η

ρ {3.4b}

The model is verified under these conditions i.e.;

• The model is tested at the range of dp from 0.36 to 0.94 mm. and interfacial tension of dispersed phase (oil) 11 to 42 dyne/cm. (T.I.O.A, heptane, anisole, toluene and kerosene)

• The velocity tested is in the range of 0.09 to 0.54 cm/s.

• Different density between dispersed phase (oil) and continuous phase (water) is between 83 to 314 kg/m3.

• The bed material used is spherical glass bead with silicon coated.

• The inlet concentration of hydrocarbon tested is around 1,000 mg/l.

3.1.3 Model verification

Inspite of the fact that AURELLE’s model is theoretical based and understandable, from experimental conditions of the two models stated above, DAMAK’s model covers wider range of oil and accounts for every parameter that affects coalescer operation. So, from application and design point of view, it is more reasonable here to use the second model as the generalized model of coalescer.

However, DAMAK’s model is tested at greater concentration of inlet oil (1,000 mg/l) than AURELLE’s (100-200 mg/l). In theory, the greater the concentration, the higher the probability of coalescing between droplets. So it is interesting to verify DAMAK’s model using AURELLE’s test data. The data used for model verification is tabulated in Annex A2.1.

However, firstly, difference between predicted and observed efficiency of both models have been checked to eliminate inaccurate or error data. Fig. 3.2a shows that, at droplet size smaller than 10 microns, the differences between predicted and observed efficiency of both models are very high. This may be caused by the re-fragmentation of coalesced oil by shear force, which both models can not predict. The error may also come from inaccurate measurement of these tiny droplets. Then, to be on the safe side, the droplet smaller than 10 microns will not be considered and assumed that their removal efficiencies are zero.

Comparison between AURELLE’s observed data and predicted data from DAMAK’s model (after error data removal) is as shown in fig. 3.2. From the graph, it shows that the error in prediction of DAMAK’s model for the whole range of wastewater tested by AURELLE and DAMAK is not greater than 10%. So DAMAK’s model can be used as the generalized model for granular bed coalescer.

57

Chapter 3 Coalescer

II-14

-30%-20%-10%

0%10%20%30%40%50%60%70%80%90%

100%

0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05 2.5E-05 3.0E-05 3.5E-05 4.0E-05 4.5E-05Droplet diameter (m)

Erro

r (%

)

DAMAK's model AURELLE's model

Fig. 3.2a Relation between droplet diameter VS. model's error

0%10%20%30%40%50%60%70%80%90%

100%110%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%Observed efficiency %

Pred

icte

d ef

ficie

ncy

%

ηd = 58 (dE/dp) 0.2 (H/dp)0.12 (ρc dp V2/γo/w )-0.08 (μd/μc)0.09 (Δρ/ρc)0.09

-10%

+10%

Fig. 3.2b Comparison between observed efficiency and predicted efficiency from DAMAK's model

3.1.4 Conclusion and generalized model of granular bed coalescer

From model verification result, we can conclude and propose the generalized model as well as its limitation as follow,

1. To predict removal efficiency of coalescer, graded efficiency (ηd) can be calculated by eq. 3.5. If the result from eq. 3.5 is greater than 100%, then it will be rounded up to 100%. Total removal efficiency can be calculated by eq. 3.4.

%1000.09)cρΔρ(0.09)

μcμd(0.08)

o/wγ

dpVcρ(0.12)dpH(0.2)

dpd58(dη ⋅

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛−= {3.5}

58


II-15

( %maxd

mi)

nd odCdη1


Q {3.4a}

∑−=max

min

d

dodd

dout CQQQ η

ρ {3.4b}

2. To use the models described above, the following conditions need to be satisfied and the assumptions and limitations would be noted;

1) Coalescer bed shall be oleophilic and relatively spherical in shape.Tested range of bed height is between 1 to 10 cm. However, bed height � 1 cm is not recommended. The greater the bed height, the safer coalescer operation. However, it also results in higher pressure drop.

2) The model is valid for inlet concentration between 100 - 1,000 mg/l.

3) Tested size of bed media is between 0.35 – 0.9 mm. The larger the media size, the lower the efficiency.

4) The density difference between dispersed phase (oil) and continuous phase (water) is between 80 to 315 kg/m3 (approx).

5) The velocity shall be in the range of 0.09 to 0.54 cm/s.(3.2 to 19.4 m/h) for coalescer without guide.

6) It is recommended to use the model only for the droplet size of 10 microns or greater. For smaller droplet, the model can also be applied, but for comparison purpose only.

3.1.5 Generalized model for guide coalescer

For guide coalescer, high porosity oleophilic material, such as steel wool, will be placed next to downstream end of the granular and extended up to water surface (actually, up to oil/water interface of the decanter). Coalesced oil drop will channel along this material until it combines with oil layer at the water surface. This material is called “guide”.

Installation of guide helps preventing formation of oil mousse or jet, which normally occurs in classical coalescer at high velocity or high concentration of oil. Thus, the maximum velocity before formation of mousse or jet will occur (so called critical velocity) of the guide coalescer is, at least, 1.5 times greater than usual [6].

Unfortunately, we do not have sufficient recorded data to develop the model for guide coalescer. However, it is proven that, at velocity range below critical value, the efficiency is approximately velocity-independent [3], [6]. It, also, can be confirmed by eq. 3.5, which shows that the exponent of V is very low (–0.08). Then, at recommended range of velocity, V-0.08 is approximately constant.

So, from this fact, we can estimate the efficiency of guide coalescer by using classical model of coalescer (eq. 3.5) with additional precautions and assumptions as follow,

• When V < 0.54 cm/s: Use eq. 3.4 and 3.5 directly.

• When 0.54 < V < 0.8 cm/s (1.5 times of 0.54): Use eq. 3.4 and 3.5 by using V = 0.54 cm/s for calculation. Using the real velocity instead of 0.54 cm/s will result in

59

Chapter 3 Coalescer

II-16

higher efficiency. However, because of the lack of verifying data, we recommend using low prediction value for safety.

• Velocity 0.8 to 1.9 cm/s may still be applicable at low concentration of inlet oil (phase ratio around 1-2).

• For oil inlet concentration, guide coalescer has been tested in liquid-liquid extraction application [6] with maximum phase ratio (oil/water) of 6, the result shows that it can operate effectively. So, for oily wastewater treatment process, which the concentration is much lower, installation of guide can assure that the coalescer should operate efficiently.

3.1.6 Generalized model for mixed bed coalescer

Mixed bed coalescer is another modified form of granular bed coalescer, used for mixed direct/inverse emulsion separation, which is usually found in liquid-liquid extraction process. In mixed bed coalescer, the bed will consist of separate layers of oleophilic and hydrophilic materials, placed in series in the same column. From research [6], ratio of oleophilic and hydrophilic material and order or configuration of column (upper hydrophilic layer/lower oleophilic layer or vice versa) depends on wastewater characteristic. So it is difficult to determine the efficiency of the coalescer by fixed equation. In this case, it is recommended to perform pilot test to find optimum design criteria.

3.1.7 Generalized model for pressure drop of granular bed coalescer and guided coalescer

Many GPI researches provide data of pressure drop of granular bed coalescers. But there is no model proposed and, unfortunately, there is insufficient data to develop new pressure drop model by dimensional analysis. So we have to develop the model based on available theory. According to the structure of coalescer, its components are similar to that of a deep bed filter. Then, we will use pressure drop models of deep bed filter as basis to develop coalescer’s pressure drop model.

Two pressure drop models are considered, i.e.,

• Kozeny Carman’s model: pressure drop is calculated from bed porosity.

• Darcy’s model: pressure drop is calculated from bed permeability.

From normal practice, porosity may be used more often to describe the bed. So we will use Kozeny-Carman’s equation (eq. 3.6) to predict coalescer pressure drop (ΔP), in m of water.

3ε2dpgmρ

2ε)V(1c180HμΔP

⋅⋅⋅

−= {3.6}

All variables except porosity (ε) will be determined by designer. For the porosity of coalescer bed, from our literature review, there is no data provided in any research. Then we will use the pressure drop data from researches to calculate back to find corresponding porosity. From many researches [3], [26], [27], it shows that bed porosity varies with bed depth and can be divided into 2 zones, i.e.,

60


II-17

• Lower zone or critical zone: This zone represents effective zone of coalescer bed. The maximum height of this zone is called “critical height (Hc)”. When bed height is greater than critical height, the efficiency will increase only slowly (From eq. 3.5: η ∝ H 0.12). In this zone, the bed will be soaked with oil so the porosity will be low.

• Upper zone: If the bed is higher than Hc, practically, all of oil will be trapped in critical zone. Then in higher zone, there will be enough oil in lower zone to flow continuously through the bed in form of “flow channeling”. So the porosity in this zone will be lower than critical zone.

We use data from various researches [3], [26], [27] to verify the value of bed porosity. The verification result is shown in Annex A2.2 and fig. 3.3. We can conclude that pressure drop of granular bed can be calculated by Kozeny-Carman’s equation (eq. 3.6), using the following recommendations, i.e.,

• When Hc is known (from literatures, etc.), pressure drop in the lower and upper part of bed can be calculated separately, using eq. 3.6. Recommended porosity (ε) for the lower (critical) part of bed (H<Hc) is between 0.14 to 0.19. Recommended porosity for the upper part of bed (H>Hc) is between 0.23 to 0.30.

• If it is certain that H design < Hc, use single step calculation with ε = 0.14 - 0.19.

• However, Hc is usually unknown, then it is recommended to use single step calculation with ε = 0.13 and 0.23 to estimate minimum and maximum pressure drop respectively (as shown in fig. 3.3).

• Because of the fact that “guide” of guided coalescer has relative high porosity (0.9 approx.), then, The pressure drop is very low, compared to granular bed, and can be negligible. So eq. 3.6 can also be used for guided coalescer.

0

20

40

60

80

100

120

140

160

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

Run number

Pres

sure

dro

p (m

)

Observed dataUpper limit (porosity = 0.13)Lower limit (porosity = 0.23)

Fig. 3.3 Relation between observed pressure drop of granular bed coalescer and predicted upper & lower limits from Kozeny-Carman's porosity = 0.13 and 0.23

61

Chapter 3 Coalescer

II-18

3.2 Fibrous bed coalescer

This type of coalescer uses relatively porous fibrous material as a bed to promote coalescing between oil droplets, as shown in fig. 3.4. Due to its high porosity, this type of bed is hardly clogged and can handle wastewater containing suspended solids efficiently. It also causes much less pressure drop than granular bed coalescer. However, fibrous element, which is normally very small, can be deflected, especially in large-scale unit, and causes unpredicted channeling, then decreasing in efficiency. Three basic steps for granular bed coalescer, proposed by AURELLE [3], can also be used to describe phenomena taking place within the coalescer. However, mathematical models, derived from dimensional analysis, are proven to be more accurate.

There are 2 main categories of fibrous bed coalescers, i.e., simple fibrous bed coalescer and dynamic (or rotating) fibrous bed coalescer. The latter is the modified form of the former, by the installation of driving unit to drive the bed.

3.2.1 Dynamic fibrous bed coalescer model

The method is proposed by TAPANEEYANGKUL [8]. The model is based on the classic dimensional analysis, which is an efficient tool when exact theory of the processes can not be established. The author had conducted his research thoroughly and covered every necessary aspect of the equipment. So his model and its limitation will be quoted here again, as shown in eq.3.7, without the need for further verification.

%1000.74V0.58

Fd0.03D

0.53N0.35H0.35ε)(10.580.67ddη ⋅

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛−

= {3.7}

The model is verified under these conditions i.e.;

• 52 < Re < 1164.

• Rotating speed of the bed is between 10 to 200 rpm.

• Empty bed velocity is between 0.1 to 1.1 cm/s (3.6 to 39.6 m/h).

• Diameter of coalescer bed is around 11.5 cm.

• Diameter of fiber is around 100 to 300 microns.

• The beds, used in the experiment, are “bottle brush” types, made of polyamide or polypropylene with stainless steel shaft.

3.2.2 Simple fibrous bed coalescer model

This coalescer is a predecessor of dynamic bed coalescer, described in the previous section. However, the researches on this type of coalescer are based mainly on its application and design consideration, rather than model development. So there is no model proposed by GPI researchers for this equipment.


After reviewing existing experimental data, it shows that each research presents only some parameters related to its own objectives. However we have tried to develop a model,

62


II-19

using dimensional analysis method, from SRIJAROONRAT’s data [10], which provided more details than others. And then we verified the model with the data from other researches [11], [16]. The proposed model is shown in eq. 3.8a. In our research, we will call it “SRIJAROONRAT’s simple fibrous bed model”.

%1000.694)DH(0.18)

DFd

(0.18)Dd(0.77)

cμ

VDcρ45.005(dη ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ −−= {3.8a}

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%

Observed efficiency (%)

Pred

icte

d ef

ficie

ncy

(%)

SRIJAROONRAT's dataMA's dataWANICHKUL's data

-20 %

+20 %

Fig. 3.4 Comparison between observed efficiency and predicted efficiency from SRIJAROONRAT's model, Verified by MA's and WANICHKUL's data

Comparison between observed and predicted efficiency is shown in fig. 3.4. From the graph, it shows that;

• The proposed model is valid for empty bed velocity up to 2 cm/s (72 m/h) and concentration of inlet oil up to 1000 mg/l.

• For higher empty bed velocity, there is not any detailed data on graded efficiency to verify the model. However, SRIJAROONRAT’s research shows that the total (or average) efficiency, more or less, still conforms to eq. 3.8 at the velocity as high as 5 cm/s and that the equation tends to underestimate the efficiency.

• For high inlet oil concentration, for example in case of WANICHKUL’s research, The model, again, tends to underestimate the efficiency.

• Because of the fact that there is not enough data to find the effect of porosity, porosity is omitted in the proposed model.

From the observations above, SRIJAROONRAT’s simple fibrous bed model does not include effect of porosity. However, from TAPANEEYANGKUL [8] research, he shows that the efficiency of dynamic coalescer, operating at low speed, is almost identical to the simple coalescer’s and his model readily includes the effect of porosity. So it may be a matter of interest to verify if the model of dynamic coalescer is possible to apply to simple coalescer by assuming a low rotational speed into the model.

63

Chapter 3 Coalescer

II-20

For this, data from several researches [10], [11], [16] have been used to verify TAPANNEYANGKUL’s model. The data, used to verify the model, has been limited to the droplet size of 10 microns and greater. For smaller droplet, it is difficult to measure the concentration of these droplets accurately. So observed data is not complete and seems erroneous. Comparison between observed and predicted efficiency is shown in fig. 3.5.

From the graph, it shows that TAPANEEYANGKUL’s model can be adapted to predict the efficiency of simple fibrous bed coalescer with acceptable degree of accuracy (± 20% error) when using the rotating speed = 450 rpm. This speed is too high to be reasonable. This may be due to the fact that TAPANEEYANGKUL’s model is verified from relatively short bed (H/D < 2) while other researches operate at H/D up to 10. So the exponent of (H/D) from both models is quite different. In this case, we can conclude that TAPANEEYANGKUL may not be applied to simple fiber bed coalescer.

However, it is still interesting to assume that the effect of porosity in SRIJAROONRAT’s model should be the same as TAPANEEYANGKUL’s. So we add the term “(1-ε)0.35” into eq. 3.8a and solve for a new constant to replace 45.005. The modified model is shown in eq. 3.8b. Comparison between observed and predicted efficiency is shown in fig. 3.6. The graph shows that MA’s data are better predicted, using eq. 3.8b. The model still cannot cover WANICHKUL’s data for it is tested at very high oil inlet concentration (7950 mg/l). However, the predicted efficiency is still on the safe side to use as a guideline.

( ) %1000.694)DH(0.35ε10.18)

DFd

(0.18)Dd(0.77)

cμ

VDcρ104.5(dη ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−−−= {3.8b}

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

120.0%

0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 120.0%


Pred

icte

d ef

ficie

ncy

(%)

-20%

+20%

Fig 3.5 Comparison between observed efficiency and predicted efficiency by TAPANEEYANGKUL's model for simple fibrous bed (Assume rotating speed = 450 rpm)

64


II-21

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%


Pred

icte

d ef

ficie

ncy

(%)


-20 %

+20 %

Fig. 3.6 Comparison between observed efficiency and predicted efficiency from modified SRIJAROONRAT's model (eq. 3.8b)

3.2.4 Conclusion and generalized model of fibrous bed coalescer

From model verification result, we can conclude and propose the generalized model as well as its limitation as follows,

Dynamic fibrous bed coalescer

1. To predict removal efficiency of coalescer, graded efficiency can be calculated by eq. 3.7. If the result from eq. 3.7 is greater than 100%, then it will be rounded up to 100%. Total removal efficiency can be calculated by eq. 3.4.

%1000.74V0.58

Fd0.03D

0.53N0.35H0.35ε)(10.580.67ddη ⋅

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛−

= {3.7}

( %maxd

mi)

nd odCdη1


Q {3.4a}

∑−=max

min

d

dodd

dout CQQQ η

ρ {3.4b}

2. The model is verified under these conditions i.e.;

• 54 < Re < 1164.

• 1 < H/D < 2. However, the maximum H/D shows in TAPANEEYANGKUL’s research is 6. Using H/D > 2 in the model can be also applied for comparison purpose only.

• Rotating speed of the bed is between 10 to 200 rpm. However, recommended minimum rotating speed is 75 rpm. Using lower speed may not provide any additional benefit over simple fibrous bed coalescer.

• Empty bed velocity is between 0.1 to 1.1 cm/s (3.6 to 39.6 m/h).

65

Chapter 3 Coalescer

II-22

• Diameter of fiber is around 100 to 300 microns and diameter of coalescer bed is around 11.5 cm. Using bigger coalescer diameter may cause deflection at the end of fibers from longer overhung length, which may cause error in calculation.

• The beds, used in the experiment, are “bottle brush” types, made of polyamide or polypropylene with stainless steel shaft.

• It is recommended to use the model only for the droplet size of 10 microns or greater. For smaller droplet, the model can also be applied, but for comparison purpose only.

3. Internal diameter of coalescer casing, which contains the bed, should be as close to the diameter of the bed as possible to avoid channeling problem.

Simple fibrous bed coalescer

1. To predict removal efficiency of coalescer, graded efficiency can be calculated by eq. 3.8b. If the result from eq. 3.8b is greater than 100%, then it will be rounded up to 100%. Total removal efficiency can be calculated by eq. 3.4.

( ) %1000.694)DH(0.35ε10.18)

DFd

(0.18)Dd(0.77)

cμ

VDcρ104.5(dη ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−−−= {3.8b}

( %maxd

mi)

nd odCdη1


Q {3.4a}

∑−=max

min

d

dodd

dout CQQQ η

ρ {3.4b}

2. The model is verified under these conditions i.e.;

• 48 < Re < 1100.

• 1 < H/D < 10.

• Unlike other granular bed model, eq. 3.8b is valid for droplet size of 1 microns and greater. The new technology on granulometer make it possible to measure these very tiny droplets with high accuracy.

• Empty bed velocity used in the researches is between 0.5 to 5.0 cm/s (1.8 to 180 m/h). However detailed data used to verify the model is between 0.5 to 2.0 cm/s. Using velocity > 2.0 cm/s may cause unpredictable error on calculated efficiency.

• Fiber size used in the researches is between 40 to 200 microns. However detailed data used to verify the model is between 100 to 200 microns. Using fiber size < 100 microns may cause unpredictable error on calculated efficiency.

• The model is verified at inlet oil concentration up to 1000 mg/l. Using the concentration > 1000 mg/l, again, will cause underestimation of predicted efficiency.

66


II-23

• Diameter of coalescer bed tested is around 5.0 cm. Using larger coalescer diameter may cause deflection at the end of fibers from longer overhung length, which may cause error in calculation.

• The beds used in these researches vary from “bottle brush” type, simple spiral type and combination of internal bed of “simple spiral” and concentric “coil spring” –like external bed with the tip of the fibers pointed to center line. However, they are all oleophilic. There is some difference in efficiency between each type, but there is too few data to make a conclusion. However, because of its rigidity, the “simple spiral in coil spring- like” bed tends to operate more stable without the decrease in efficiency with time, while others tend to be deflected by weight of accumulated oil drops. In fact, this type of bed is invented to take advantage of spiral bed for its non-clogging and disorderly bed for its rigidity.


3.2.5 Generalized model of random or disorderly fibrous bed coalescer

There is another special case of simple fibrous bed coalescer that uses random or disorderly fibrous material (such as metal wool, etc.) as coalescer bed. SRIJAROONRAT also studies this type of coalescer. In her research, she shows that the removal efficiency of this coalescer is higher than that of coalescer that uses brush type bed. For this, it can be concluded that tortuosity of bed also affects the removal efficiency. We use her data to develop the model, based on dimensional analysis, as shown in eq. 3.9. So we will call it “SRIJAROONRAT’s random fibrous bed model”. However, this model is developed from rather small set of data. So we have tried to apply the generalized model in eq. 3.8b to this type of coalescer. Comparison between eq. 3.9, 3.8b and observed value is shown in fig. 3.7.

%1000.36)DH(0.03)

DFd

(0.03)Dd(0.23)

cμ

VDcρ3.35(dη ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ −−= {3.9}

0.0%10.0%20.0%30.0%40.0%50.0%60.0%70.0%80.0%90.0%

100.0%110.0%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%Observed efficiency (%)

Pred

icte

d ef

ficie

ncy

(%)

SRIJAROONRAT's disorderlybed model

SRIJAROONRAT's simple bedmodel

+10%

-10%

Fig. 3.7 Relation between observed efficiency and predicted efficiency from random (or disorderly) fibrous bed coalescer model and simple fibrous bed model

67

Chapter 3 Coalescer

II-24

From the graph, it shows that SRIJAROONRAT’s random fibrous bed model (eq. 3.9) can accurately predict the efficiency of the coalescer. SRIJAROONRAT’s simple fibrous bed, (eq. 3.8b) is used to calculate the efficiency of the coalescer for comparison. However, from the graph, it shows that the result from eq. 3.8b tends to underestimate the efficiency from 2 to 6 times. From this, we recommend the following procedure to calculate the efficiency of metal wool bed coalescer.

1. Apply SRIJAROONRAT’s random fibrous bed model (eq. 3.9) if these conditions are satisfied, i.e.,

• Velocity is between 1 to 2.5 cm/s or 36 to 90 m/h. • Inlet concentration is around 1000 mg/l. • The model is developed from metal wool bed coalescer, diameter = 5 cm.,

fiber diameter = 40 microns, and porosity = 0.95. • The wool is coated with silicone, so it becomes oleophilic.

2. Even though eq. 3.9 is developed from small set of data and tortuosity can not be established in the form of numerical factor. But, from graph 3.6, it can be estimated that the disorderly bed coalescer is 2 to 6 times more efficient that simple bed. However, it will be clogged easily if suspended solids are present in the wastewater.


3.2.6 Generalized model for pressure drop of fibrous bed coalescer

Many researcher [8], [10], [11], [16] observed pressure drop of fibrous bed coalescers and reported that these coalescer causes very low pressure drop due to very high porosity of their beds. There is no proposed model on pressure drop.

In order to calculate the pressure drop, we recommend to use any general piping loss equations, such as Darcy’s, Colebrook-White’s or Hazen-William’s equation with the safety factor of 2 to 5, multiplied to the actual length of the bed. However, the pressure drop of fibrous bed coalescer is normally low (< 104 N/m2), compared to piping system pressure drop.

68


II-25

Chapter 4 Dissolved air flotation

Flotation is an accelerated separation process by increasing density difference between continuous phase and dispersed phase. This is accomplished by mean of adding gas or air into the wastewater to promote formation of air-solids or air-oil agglomerates. There are several researches on flotation, its modification and applications, studied in GPI laboratory, such as mechanical flotation, diffused air flotation, dissolved air flotation, etc. Here, we will consider only the application of dissolved air flotation (DAF) on oily wastewater treatment.

4.1 Dissolved Air Flotation (DAF) model for oily wastewater treatment

For Dissolved Air Flotation, air bubbles are generated from pressurized (or air –saturated) water. At GPI laboratory, the model of DAF for oily wastewater treatment is proposed by SIEM [12]. In his study, SIEM applies filtration-based model for granular bed coalescer, proposed by AURELLE [3], by assuming the air bubble as collector (or filter media), as shown in eq. 4.1. However, in this case, the media is also moving. Schematic diagram of DAF is shown in fig. 4.1.

SeparatedDroplet/bubbleagglomerate

Oil droplet

Pressurizedwater system

Oilywastewater

Airbubble

Clarifiedwater

Fig. 4.1 Schematic diagram of DAF

Removal efficiency (or probability) factor of DAF is calculated from summation of theoretical efficiency factors of 3 transport phenomena, i.e. sedimentation, direct interception, and diffusion, as shown in eq. 4.2, and adjusted to fit with observed data, as shown in eq. 4.3.

%100

)bd

H)exp(αAVΦ

23(

e1dη ⋅

⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛ −

−=

η {4.1}

diffηIntηsedηtheoη ++= {4.2a}

r18μ8

2Δρgdsedη = {4.2b}

69


II-26

2)bdd(

23

Intη = {4.2c}

2/3)bdrμdV

KT0.9(Diffη = {4.2d}

Vr = |V-Ub| {4.2e} 0.507)theo0.0074()exp(α ηη = {4.3a}


4.2.1 Modification of filtration-based model

From the model in section 4.1, air or gas flowrate (Φ) and the term “AV” can be written in form of Qt as shown in eq. 4.4.

airρConc(air)

tQpwQ

AVΦ

⋅= {4.4a}

)xtQ

pwQ(= {4.4b}

Normally, under certain design condition, Qpw/Qt is constant. Solubility of air in water (Conc(air)) and air density are intrinsic (internal) property, depends on pressure, and temperature of pressurized water, which are constant for any given pressurized water system. Then, from eq. 4.4, it shows that the term “Φ/(AV)” is flow-independent.

From, eq. 4.2, it shows that the effciency factors vary with flowrate via relative velocity between air bubble and oil droplet. Rising velocity of bubble (Ub) is calculated by STOKE’s law, so it does not depend on wastewater flowrate. However, if we consider eq. 4.2d, it can be implied that if we lower the flowrate until V = Ub, Vr is, therefore, equal to 0. Or when V >> U, it will seem to some one that happen to be on an oil droplet, which will be carried along with the flow, that he run pass very slow bubble, or bubble will stay in the reactor longer than oil drop. This cannot be true because the bubble will be carried along with the flow as well. Furthermore, from its lower density and its bigger size, the bubble will usually rise up faster than oil drop at the same diameter.

In fact, relative velocity (Vr) is equal to difference between absolute velocity of bubble (V+Ub) and flow velocity of water (V) (eq. 4.2f). If we replace Vr with Ub, it will cause some changes in eq. 4.3a. We have already recalculated the eq. 4.3a (see fig. 4.1) and found that it can be rewritten as shown in eq. 4.3b (modified SIEM’s model). Fig. 4.2 shows comparison between observed efficiency and predicted efficiency calculated from eq. 4.1, 4.2a to 4.2d, 4.2f and 4.3b. From the graph, prediction error is about 20%.

Vr = V+Ub-V = Ub {4.2f} 0.5919)theo0.009005(η)(α exp =η {4.3b}

70


II-27

η

y = 0.5919x - 4.71R2 = 0.9821

-9.0E+00

-8.0E+00

-7.0E+00

-6.0E+00

-5.0E+00

-4.0E+00

-3.0E+00

-2.0E+00

-1.0E+00

0.0E+00-7.0E+00-6.0E+00-5.0E+00-4.0E+00-3.0E+00-2.0E+00-1.0E+000.0E+00

ln(Theoritical efficiency)

ln (O

bser

ved

effic

ienc

y)

η

Fig. 4.2 Relation between theoretical efficiency factor and observed efficiency factor

0.0%10.0%20.0%30.0%40.0%50.0%60.0%70.0%80.0%90.0%

100.0%110.0%120.0%

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%Observed efficiency

Pred

icte

d ef

ficie

ncy

+ 20%

- 20%

Fig. 4.3 Relation between obseved efficiency and predicted efficiency of DAF from modified SIEM's model

However, since we replace Vr with Ub, it is interesting to note that the proposed model (eq. 4.1 to 4.2a to d) is mathematically flow-independent. In fact, the effect of wastewater flowrate, sometimes described in form of velocity or retention time, is studied by many researchers. It is widely accepted that DAF efficiency varies with retention time. So, considering SIEM’s research, it can be interpreted that the effect of retention time is already included in eq. 4.3. Since eq. 4.3 is evaluated from one set of operating condition and retention time (approx. 25 minutes), it may be valid only for that condition.

To expand valid range of model, we need to verify or adjust the model by other sets of data. However, from literature review, we cannot find enough data to do so. Then we have to find theoretical criteria or equation that allow us to estimate the effect of retention time on removal efficiency.

4.2.2 Population balance method

71


II-28

Population balance method is one of popular concepts, used by many researchers, to develop DAF model. In our lab, DUPRE [14] uses this method in her research on application of DAF for liquid-solid separation. Concept of this method is that rate of change in the number of oil droplets which are free or attached by 1, 2 or more air bubbles is the function of the number of air bubble and oil droplet, as shown in eq. 4.5. Oil droplet that is attached by at least 1 bubble, so-called oil-bubble agglomerate, will be separated. So the rate of change in number of oil droplets represents removal efficiency.

Main assumption of population balance method is that the number of bubble is assumed to be constant. From many researches [14], [34], It is proven that, for liquid-solid separation, the number of bubble (N) can be safely assumed as constant without serious error, because there are a lot more bubbles than pollutant if normal range of (Qpw/Qt) is applied. However, in case of oily wastewater, many researches [12], [13] show that bubble-oil agglomerates are in form of oil shell with air inside and these agglomerates are still able to intercept more oil droplets. Those researches also show that coalescence of oil and bubble is more effective than that of the same species. So it should be safely assumed that the number of bubble in this case is more or less constant and the population balance method is, then, applicable.

N0n0kβdt

0dn−= {4.5a}

NinikβN1in1ikβdt

idn−−−−= {4.5b}

i = 0 to imax

3bd

6πΦN = {4.5c}

no and ni represent the number of oil droplet which is free and attached by i bubbles respectively. κ represents collision rate constant, from Saffman and Turner’s (1956) coagulation theory, and β represents adhesion efficiency or probability that the collision between oil drop and bubble will be successful. Removal efficiency can be written as (ni/n0), which can be achieved by integrating eq. 4.5.

To integrate eq. 4.5, complex numerical method is required. MATSUI and LEPPINEN [34], [37] solve the equation by Laplace transforms and suggest the solution as shown in eq. 4.6.

i

12b/d2d

κτ

eκτ)(ei:xNin

⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜

⎝

⎛

−−⋅=⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛

{4.6a}

i = 0 to imax –1

)1)...(2)(1i(i1))(i1)...(xx(xi:x

−−−−

= {4.6b}

2b/d2dx = {4.6c}

72


II-29

When i = 0, the above function will be equal to 1.

0Φβ3))b(d/daG(1π6κ += {4.6d}

G is a measure of turbulence intensity, which is normally proportional to V but we do not know the exact relation. For i = 0, ni will become n0, which represents the number of free oil drop remaining from contacting with air bubble. So 1-(n0/N) represents removal efficiency of the process. Using i > 0 will result in greater value of efficiency. So, to be on the safe side, we will use i = 0. Then eq. 4.6, when i = 0, can be rewritten as shown in eq. 4.7. Removal efficiency of DAF is, then, can be written in form of eq. 4.8.

κτ)(eN0n −= {4.7}

κτ)(e1N0n

1η −−=−= {4.8}

It should be noted that eq. 4.8 is about the same form as eq. 4.1, but the value of κ, represented by eq. 4.6d, clearly represents effect of every interesting parameter, including retention time. The collision rate κ can be compared to the exponent term in eq. 4.1, as shown in eq. 4.9a and 4.9b.

bdH)exp(α

AVΦ

23τ0Φβ3))b(d/daG(1

π6κτ η=+= {4.9a}

At constant d and db, eq. 4.9a can be rewritten as;

Φτ2κGΦConstκτ =⋅= τ {4.9b} Φτ)2κ(

e1κτ)(e1η−

−=−−= {4.9c}

If “x” represents the variable Φ or τ that we want to vary from SIEM’s while another variable still conforms to SIEM’s condition, eq. 4.9c can be rewritten as show below.

xAe1η ⋅−−= {4.9d}

Where A= constant

If x = B.x ref {4.10a}

Where B = Constant

Xref = x at SIEM’s condition

Then

)refAx(e1

Ax)(e1

refηη

−−

−−= {4.10b}

)refAx(e1

)refABx(e1

refηη

−−

−−

= {4.10c}

73


II-30

B)refη(11η −−= {4.10d}

However, changing of τ or Φ will also cause some parameter in eq. 4.6d change. For example, increasing of retention time from SIEM’s condition will make V decrease. Then G will decrease. So the constant “A” at design condition is not equal to “A” at SIEM’s condition. In this case, eq. 4.10 is not valid. And it is not possible to know the value of “A” at design condition. So, the best estimation in this case is to use the value of A at some condition that we are sure that will underestimate efficiency at design condition. Design reactor in this case will be larger than it should be. Then it can be considered as safety factor.

Criteria to predict the efficiency from combination of SIEM’s model and population balance model will be described again in details in section 4.3.

In case of inlet oil concentration, SIEM’s model is verified at inlet oil concentration of 800 mg/l before adding pressurized water or around 400 mg/l after adding pressurized water, which is moderate value. If inlet oil concentration increases, the collision rate κ would increase since probability of bubble collision (a) will be increased. However, DUPRE’s study [14] shows that inlet concentration less than 600 mg/l does not have significant effect on the efficiency. So, to be on the safe side, we can imply that modified SIEM’s model is still valid when inlet concentration is not greater that 600 mg/l (dilution included) or 1,200 mg/l (dilution not included). Using the model for higher oil concentration will result in underestimating of efficiency.

4.3 Generalized Model for DAF

From 2 models stated above, we can conclude and propose the generalized model as well as its limitation as follows,

1. To predict removal efficiency of DAF, reference graded efficiency (based on Qt) can be calculated by eq. 4.1, 4.2 and 4.3. If the result from the equations is greater than 100%, then it will be rounded up to 100%. The values of (Φ/AV) have to be the same as SIEM’s condition (see 2.) or its staightforward scale up.

%100

)bd

H)exp(αAVΦ

23(

e1refd,η ⋅

⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛ −

−=

η

{4.1a}

or

%100

)bd

H)exp(αairρ

Conc(air)

tQpwQ

23(

e1refd,η ⋅

⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜

⎝

⎛⋅⋅−

−=

η {4.1b}

diffηIntηsedηtheoη ++= {4.2a}

74


II-31

r18μ8

2gdoil/waterΔρ

sedη = {4.2b}

2)bdd(

23

Intη = {4.2c}

2/3)bdrμdV

KT0.9(Diffη = {4.2d}

0.5919)theo0.009005(η)exp(α =η {4.3b}

c18μ

2bgdair/waterΔρ

bUrV == {4.2f}

2. To use the models described above, the following conditions need to be satisfied;

1) Inlet oil concentration should not be greater than 1,200 mg/l (before dilution) or 435 mg/l (after dilution). Using the model with higher oil concentration will result in underestimating the efficiency.

2) The model is tested at the following operating conditions;

• Φ/AV = 0.0516. Only this value must be used in the equations. As long as this value is fixed, SIEM’s operating condition still holds and the model is still valid.

• Retention time (τ), based on Qt, is around 25 minutes. • Droplet diameter (d) tested is between 2 to 40 microns. • Diameter of air bubbles (db) varies from 15 to 130 microns. Tested

average diameter is 70 microns, which is used to verify the model, and standard deviation of bubble diameters is 34.5 microns. The range of bubble sizes is common for commercial pressurized water system or saturator. The pressure of the test system is 4 atm (absolute).

• Tested air flowrate (Φ) is 0.42 cm3/s (4.2e-7 m3/s). • Tested wastewater flowrate (Q) is 3.9 cm3/s (3.9e-6 m3/s) • Tested effective water depth (H) is 0.70 m. The value of H can be

freely changed as long as (Φ/AV) is fixed. However, H between 1.8 to 2.7 is recommended by API [45].

• Diameter of flotation column is 0.15 m Cross section area of column (A) is 0.01767 m2.

• Ratio of pressurized water to wastewater (Qpw/Q) is 1.76. • Air to pollutants ratio used is around 0.12 kg. air/ kg. oil. • Ratio of number of bubble/ oil droplet tested is around 1.4 oil droplet/ 1

air bubble. • Hydraulic loading rate or flow velocity (V), based on Qt, is 1.6 m/h

3. Because of the limitation of the pilot model, tested ratio of pressurized water to wastewater is quite high (around 92%), compared to that of general DAF for solid/liquid separation (less than 50%) [13]. However, API [45] recommended air/wastewater ratio of 0.35 std. ft3/ 100 gal of total flow for full-flow DAF process. This value is equivalent to 84% of 4-atm (abs) pressurized water/

75


II-32

wastewater. Anyway, it is interesting to adapt the model to calculate the efficiency at lower ratio of pressurized water.

4. Tested hydraulic loading rate or overflow rate (based on Qt) is 1.6 m/h, which is relatively low, compared to normal rate of 3-15 m/h for domestic wastewater treatment. The value recommended by the American Petroleum Institute (API) [45] is between 4.8-6.1 m/h. So it is also interesting to adapt the model to calculate the efficiency at higher overflow rate.

5. To calculate graded removal efficiency at other operating condition, esp. at higher overflow rate or lower ratio of pressurized water/wastewater, the following procedure is recommended.

1) Calculate the reference efficiency of the model (ηd,ref) at our required height (Hreq), average bubble diameter (db) and droplet sizes (d) using eq. 4.1 to eq.4.3. Use Φ/AV = 0.0516 in order that the operating condition of SIEM still holds.

2) Scale up the area from 0.01767 m2 to required area (Areq), which may be calculated from general design criteria. Other operating condition from 1) still holds. So efficiency from 1) remains the same.

3) Find Φref, corresponding to the area Areq, by the following equation,

0.01767reqA7104.2

modelΦmodelA

reqArefΦ

−×=⋅= m3/s {4.11}

4) Find τref, corresponding to the area Areq, by the following equation,

36001

0.00046req25H

modelτmodelH

reqHref ⋅=⋅=τ hour {4.12}

5) Find κ2,ref, corresponding to Areq, Hreq, τref and Φref from the reference efficiency (from 1)) by the following equations. Please note that, at this point, SIEM’s constraint still holds.

refτrefΦ

)refd,ηln(1ref2,κ

⋅

−−= {4.13}

Or )refτrefΦref2,κ(

e1refd,η−

−= {4.14}

6) To change Φ or τ from SIEM’s condition, the following procedure is recommended and precaution should be noted.

• To decrease Φ (Φreq < Φref) and increase τ (τreq > τref):

This will cause decreasing of V, so G will decrease. Then κ2 (see eq. 4.9) will be lower. However we do not know how much exactly. So, to be on the safe side, we will assume that only Φ decrease but τ = τref. In this case, κ2 will remain the same and be equal to κ2,ref. The efficiency can be calculated by eq. 4.15a.

76


II-33

)refτreqΦref2,κ(e1dη−

−= {4.15a}

Because we use τref, instead of τreq, the calculated efficiency will be lower than the real value.

• To decrease Φ (Φreq < Φref), as well as, τ (τreq < τref):

This will cause increasing of V, so G will increase. Then κ2 (see eq. 4.9) will be higher. Again, we do not know how much exactly. So, to be on the safe side, we will assume that κ2 = κ2,ref. The efficiency can be calculated by eq. 4.15b. And again, the calculated efficiency will be lower than the real value.

)reqτreqΦref2,κ(e1dη−

−= {4.15b}

• To increase Φ (Φreq > Φref) and decrease τ(τreq < τref):

This can be done by increasing pressurized water flowrate. However, the ratio of pressurized water/ wastewater is already high (176%). Thus, this case is unlikely to heppen. In this case, κ2(see eq. 4.9) will be higher. Like the former case, the efficiency can be calculated by eq. 4.15b.

• To increase Φ (Φreq > Φref), as well as, τ(τreq > τref):

This case does not exist because it means that we have to decrease wastewater flowrate. As stated above, the ratio of pressurized water/ wastewater is already high (92%). If we decrease wastewater flow, quantity of pressurized water flow will exceed that of wastewater, which is not feasible because we have to recycle effluent at 100% plus additional makeup water to feed the pressurized water system.

There is no obvious limit for the 4 adaptations, shown above. However we recommend using the values within general range, shown in item 2. to 4.

7) Outlet concentration can be calculated from eq. 4.16a. If DAF effluent is recycled to pressurized system, Cod,dil will be calculated as shown in item 8. If pressurized water comes from additional clean water, Outlet concentration can be calculated by eq. 4.16b. In this case, effluent quantity is equal to Qt, not Q. The subscript “dil” represents the condition after dilution with pressurized water.

dilod,)Cdη(1dC −⋅=outQQ {4.16a}

)tQ

Q(od)Cdη(1dC −⋅=outQQ {4.16b}

∑⋅−=max

min

d

dodd

dout CQQQ η

ρ {4.16c}

DAF efficiency (ηDAF), which is the efficiency based on flotation effect alone, can be calculated by the following equation.

77


II-34

100%dilo,C

maxd

mind dilod,C)dη-(1

DAFη ⋅

∑ ⎟⎠⎞⎜

⎝⎛ ⋅

= {4.16d}

In case that, additional clean water is used as pressurized water, eq. 4.16d can be rewritten as follow.

( )100%

oC

maxd

mind odC)dη-(1

)tQ

Q(oC

maxd

mind)

tQQ(odC)dη-(1

DAFη ⋅

∑ ⋅

=

∑ ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

=

{4.16e}

Total removal efficiency, which is the efficiency calculated from ratio mass between oil removed and initial mass of oil, can be calculated by eq. 4.16e.

100%oC

CoCtη ⋅

−= {4.16f}

In case that, additional clean water is used as pressurized water, eq. 4.16e and be rewritten as follow.

100%oC

)Q

tQC(oC

QoCtQCQoC

tη ⋅−

=⋅−

= {4.16g}

8) If the effluent from flotation is recycled to use in pressurized water system, some oil left in the effluent will be returned to the system. In this case, mass balance of oil has to take into account. Thus, Cod will be modified by adding this returned oil repeatedly, as shown in eq. 4.17. Theoretically, rmax in eq.4.17 should be infinity. The value of C’od will eventually convert to an asymptote. Anyway, for ηd greater than 50%, using rmax between 5 to 10 will practically give the result sufficient accuracy, esp. when ηd > 20%.

tQ

QodCmaxr

1r

1r

)dη(1tQ

pwQ

dilod,C ⋅⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

∑=

−

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛−=

{4.17a}

Where Qpw = RQ, Qt = Q + RQ. Then, eq. 4.17a can be simplified as follows,

tQ

QodC

1-)η(1R1

R

1)1maxr

))η(1R1

R((

dilod,Cd

d

⋅−

+

−−

−+= {4.17b}

9) It must be noted that there are many other parameters that can affect efficiency of DAF, such as reactor hydraulic design, contact zone configuration, design of pressurized water system (or saturator) and pressure regulating device, etc. These parameters can cause some discrepancies in efficiency prediction. However, the model will be a useful tool for preliminary evaluation of the efficiency of DAF reactor.

78


II-35

Batch experiment, such as Flota-test, will provide valuable result, if not exact, for design propose and can be used to compare with model result.

4.4 Generalized equations for pressurized water system calculation

An important system that plays very significant role in DAF process is pressurized water system, since, as shown in previous section, it is the source of air bubble and ratio of air/oil droplet is one of the key parameters in DAF process. It is still the component that consumes almost all of the energy required in DAF process. The other required energy is pressure drop of the process caused by hydraulic elements of the reactor, such as pipeline, outlet weir, etc., which can be calculated by familiar pressure drop equations. This pressure drop is substantially lower than required energy of the pressurized water system.

The power for pressurized water pump can be calculated by normal pump equation (Power =ρgQpwH/ηpump). Head of pump (H) can be assumed to equal the absolute pressure of the pressurized water system. For the power required for an air compressor as well as the quantity of air released by the pressurized water will be as show below.

To calculate quantity of pressurized water that can supplied the required amount of air bubbles, our researches show that theoretical equations based on Henry’s and Dalton’s law give relatively accurate result. The equations can be summarized as follows,

1. To predict molar fraction of dissolved gas in the water (x), Henry’s law (eq. 4.18) will be applied. Normally, we use the molar fraction of air (yair), which equals to 1 , to calculate. Anyway, if we need to know the quantity of some certain gases, i.e., oxygen or nitrogen, etc. It can also be calculated by Henry’s law, using molar fraction of oxygen gas and nitrogen gas in air (yO2 and yN) of 0.11 and 0.89 respectively (in permanent regime). For the absolute pressure of the saturator or pressurized water system (P), it is recommended to use pressure within the standard of commercial equipment range (around 4 atm(abs)). The higher the pressure, the more the amount of bubble generated and the greater the energy required. Henry’s constant (H) of air, oxygen, and nitrogen are 4.04 x104, 8.04 x104, 6.64 x104 atm/mole respectively

HyPx = {4.18}

2. When we know the molar fraction of dissolved air or gas (x) in water, we can convert it to mg/l of dissolved air or gas or volume of air or gas per unit volume of water by general conversion factor. The equation for calculating mg/l of dissolved oxygen or air in pressurized water is shown in eq. 4.19a and 4.19b respectively (using P in atm). Ratio of air flow to pressurized water flow at 20 o

C, assuming %saturation of air or gas in water is equal to 95%, are as shown in eq. 4.20 (using P in atm).

4.5893P)2Conc.(O = {4.19a}

22.965PConc.(air) = {4.19b}

.01910.0191P)pwQΦ( −= {4.19c}

79


II-36

y = 0.0191x - 0.0191R2 = 1

y = 22.965xR2 = 1

y = 4.5893xR2 = 1

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0Absolute pressure (atm)

mg

of g

as /l

of w

ater

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

Rel

ease

d ai

r vol

ume

per

volu

me

of p

ress

uriz

ed w

ater

(l ga

s/ l

wat

er)

Dissolved oxygen Dissolved air Released air per volume of pressurized water

Fig. 4.4 Relation between absolute pressure in pressure tank and dissolved quantity and released air volume

3. To estimate power required for the air compressor, SIEM shows that theoretical equation (eq. 4.20a), based on adiabatic process (PV1.4 = const.), can be applied with acceptable accuracy. Please note that the equation does not include the overall efficiency (ηoverall) of the system, which can be safely assumed to be around 60-70%. So the overall power required can be calculated by eq. 4.20b.

pwQMW(air)Conc(air)1

)1.411.4(

)atmP

2P(

0.4RT

theoPower ⋅⋅⎥⎥

⎦

⎤

⎢⎢

⎣

⎡−

−= {4.20a}

overallηtheoPower

Power = {4.20b}

Using R = 8.314 Pa.m3/(mol. K), Conc(air) in g/l and Qpw in m3/s will result in Power theo in Watt. Molecular weight (MW) of air is 28.95 g/mol. For power consumption of pressurized water pump, it can be calculated by the following equation (P as atm).

pump

atmc

pump

atmpw PPgQRPPQPower

ηρ

η ⋅⋅−⋅⋅⋅⋅

=⋅

−⋅=

360010)(

3600)( watt {4.20c}

80


II-37


Hydrocyclone is an accelerated separation process by replacing gravitational acceleration with higher centrifugal acceleration. Hydrocyclones are widely used in many processes, i.e., classification and separation between solid-liquid and liquid-liquid. In our scope of work, we will consider mainly on the mathematical model of liquid-liquid hydrocyclone. There are 2 main types of hydrocyclones studied in GPI laboratory, i.e., two-phase hydrocyclone and three-phase hydrocyclone.

5.1 Two-phase hydrocyclone

5.1.1 Trajectory analysis-based model

Every commercial hydrocyclone has its own shape and ratio between each component. So it is difficult to develop the model to cover every type of them. In our lab, main research on hydrocyclone is based upon MA’s study [16] on two-phase hydrocyclone for oil/water separation. In his research, he bases his experiment on “THEW” type hydrocyclone, which is initiated by Prof. THEW, UK. So, in our research, we will emphasize only on this type of hydrocyclone.

MA develops hydrocyclone model, based upon trajectory analysis, which is normally used in decanter calculation. The concept of the model is that oil droplet in the hydrocyclone is subjected to 3 velocity components, i.e., radial velocity (U), tangential velocity (V) and axial or vertical velocity (W), as shown in fig. 5.1. For the tangential velocity, he assumed that the oil droplet has the same tangential velocity as the liquid, which follows the free vortex pattern (VRn = const.). In the case of THEW’s hydrocyclone, n is equal to 0.65. The equation of V for THEW’s hydrocyclone is a sshown in eq. 5.1c. He also assumed that vertical velocity (W) of the droplet is similar to that of the liquid. Vertical velocity profile can be devided into 2 regions. The external region, near to the wall of the hydrocyclone, has downward flow, while the internal region has upward flow. The equation of W is in 3rd order polynomial form as shown in eq. 5.1d and 5.1e. For the radial velocity of the droplet, he assumed that it is governed by STOKE’s law. His assumption is that if the droplet can reach certain radial distance where the vertical velocity starts changing from downward to upward direction (R = RZVV, “ZVV” means zero vertical velocity), it will be carried upward to the overflow port, then separated from the main stream, which will flow out at the bottom (underflow) outlet.

Dn

W

VU

Di

Do

R

Z

β

D

Ds

Z

d = dc d > dcd < dc

R

Z

R

Z

RRd

L

%100=η%100=η a) Schematic diagram b) Trajectory of each size of oil droplets

Note: Dn/D=0.5, Ds/D=0.25, Do/D<0.05, Di/D=0.25, β=1.5 deg

Fig. 5.1 Schematic diagram and trajectories of droplets in two-phase hydrocyclone

81


II-38

MA’s model is presented in eq. 5.1 to 5.2. Steps of calculation of the model will be identical to decanter’s, beginning with cut size calculation. After finding the cut size, graded and total removal efficiency, then, can be calculated.

∫=∫L

0 WdZdR

zvvR UdR {5.1a}

RV 2

18μ

2ΔρdU ⋅= {5.1b}

0.65)RnD

(2iD

4π(Q/2)0.5V

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

= {5.1c}

3

zRR1.19

2

zRR8.63

zRR123.33

zWW

⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−= {5.1d}

2/2))tan(Znπ(0.5D

QzW

β⋅−= {5.1e}

Cut size (dc) is the smallest oil droplet diameter that moves from Rd = Dn/2 (eq.5.1a) and can reach the central core of cyclone at Z = L. When d ≥ dc, removal efficiency will be 100%. For THEW’s hydrocyclone, RZVV = 0.186 (Dn/2) at Z = L.

100%dη = {5.2a}

For d < dc, the oil drops that enter the hydrocyclone at relatively short distance from the center line can also be separated, while those that enters at further distance will be carried over with wastewater through the underflow outlet. The efficiency in this case can be calculated by ratio of the area corresponding to the entering distance “Rd” to the whole cross section area, as shown in eq. 5.2b. We can assume ηd (η50%, η75%, etc.) and calculate the corresponding Rd, then use eq. 5.1 to calculate corresponding d (d50%, d75%, etc.) from the Rd.

%100)2)

2nD

(0.1862)2nD

((

)2)2nD

(0.1862d(R

dη ⋅−

−= {5.2b}

Advantage of the concept of trajectory is that the concept can be applied to other shape of hydrocyclone, if the equation of tangential velocity V and W are known, since the equations of U is assumed to conform to STOKE’s law. For the equation of V, the general form is normally as shown in eq. 5.1c except the exponent (for THEW type, = 0.65) that will vary with the shape of the hydrocyclone.

5.1.2 Other models

There are several researches that suggests the model to predict the efficiency of hydrocyclone, both theoretical and empirical based, such as Bradley’s, Rietema’s, Dahlstrom’s, Chebelin’s, Plitt’s, Lynch’s, etc. [16],[28]- [34]. However, most of models are developed from solid-liquid hydrocyclone. Some models are developed for specific

82


II-39

commercial liquid-liquid hydrocyclone, such as Vertoil. So it should be applied only with that specific hydrocyclone. Extrapolation of model is normally not guaranteed.

For THEW hydrocyclone, used by MA in his research, Prof. THEW, himself, and his colleague, COLMAN, have proposed the model for the hydrocyclone (eq. 5.3a). However, it is an empirical model, which seems to be obtained from curve fitting. CHEBELIN [29] quoted THEW-COLMAN’s correlation for solving d75% in his research, as shown in eq. 5.3b.

%100)e1(dη0.19))

dd

1.8((75% ⋅−=

−− {5.3a}

0.5

QΔρ

3)n(0.001D0.00001μ61075%d⎥⎥⎦

⎤

⎢⎢⎣

⎡

⋅⋅

= {5.3b}


To verify the models, we will compare predicted efficiency, calculated from MA’s model, with observed data. Moreover, since MA’s model is developed from THEW hydrocyclone. So it will be interesting to compare the MA’s model with THEW’s model. We use the data from MA’s study, based on THEW hydrocyclone, nominal diameter 2 cm. Operating condition used is tabulated in table 5.1. For COLMAN’s model, we used observed value of d75% in eq. 5.3.Comparison between result from the 2 models and observed data is shown in fig. 5.2.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

110%

0.00 10.00 20.00 30.00 40.00 50.00 60.00Droplet diameter (micron)

Effic

ienc

y (%

)

Colman's modelObserved dataMa's trajectory analysis

Fig. 5.2 Comparison between observed efficiency and predicted efficiency from Ma's and Thew-Colman's models

From the graph, it shows that, at droplet size > 20 microns, MA’s and COLMAN’s models give relatively accurate result (± 10% error). However, at d > d 80%, COLMAN’s model seems to cause higher degree of error and predict too high value of cut size. This may because the researchers used different assumptions or operating condition to develop their models. In effect, it is very difficult to point out that which model is more accurate.

83


II-40

However, Ma’s model is developed from theoretical assumption and does include many parameters, such as viscosity, feed flowrate, and size of the hydrocyclone, etc., which can be used to explain or verify the effect of these parameters to the efficiency. So, in this research, we will base our model on MA’s.

5.1.4 Conclusion and Generalized model of two-phase hydrocyclone

From model verification result, we can conclude and propose the generalized model as well as its limitation as follows,

1. To predict removal efficiency of the hydrocyclone, graded efficiency can be calculated by eq. 5.1 and 5.2. If Rd in eq. 5.1a is equal to Dn/2, the corresponding “d” will be equal to dc.

∫=∫L

0 WdZdR

zvvR UdR {5.1a}

RV 2

18μ

2ΔρdU ⋅= {5.1b}

0.65)RnD

)(2iπD

Q(V = {5.1c}

319.1

263.81233.3 ⎟⎟

⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zRR

zRR

zRR

zWW {5.1d}

2/2))tan(Znπ(0.5D

QzW

β⋅−= {5.1e}

For THEW’s type hydrocyclone, when Z = L:

)2/(186.0 nVZZ DR = {5.1f}

If Rd in eq. 5.1a is equal to Dn/2, the corresponding “d” will be equal to dc.

For d ≥ dc

%100=η {5.2a}

For d < dc,

%1002)

2nD

(0.1862)2nD

((

2)2nD

(0.1862d(R

dη ⋅−

−= {5.2b}

The equations are, somehow, very complex and require complicate numerical method, such as Range-Kutta, to solve. However, our computer program, developed in scope of work of this thesis, will be able to calculate these equations.

84


II-41

2. To design the hydrocyclone by MA’s model, it is recommended to select the cut size that covers majority of oil droplet in the wastewater and provide a safety factor around 10% to 20%, because the model is apt to predict too small cut size. For example, if the desired cut size is 50 microns, it is recommended to select 50(1-0.20) = 40 microns for eq. 5.1 and 5.2.


1) The model is valid for THEW hydrocyclone or other cyclones with relative identical geometry.

2) It is recommended to use the model only for droplet diameter of 20 microns or greater. For smaller droplet, it can also be applied, but for comparison only.

3) Eq. 5.2c is valid for the hydrocyclone with 2 inlet ports only. If the hydrocyclone has only 1 inlet port, Q in eq. 5.1c will be modified as shown in eq. 5.1c’. However, using 2 inlet ports is recommended for its hydraulics stability. Please note that the size of 2 inlet ports will be smaller than a single inlet port to keep the inlet area constant.

0.65)RnD

)(2iπD

Q2(V = {5.1c’}

4) Overflow quantity is usually small, not greater than 10%. Its effect on velocity profiles and efficiency is small, thus, negligible.

5.1.5 Generalized Model for Pressure drop of two-phase hydrocyclone

Many literatures [16], [28] – [34] have studied pressure drop of hydrocyclone. Some researchers also proposed models of pressure drop, obtained from their experimental data, i.e., Bradley’s, Hotta’s, Rietema’s. From these models, we find that, even derived from different assumptions, the general form of pressure drop equation is as shown in eq. 5.4

)4n/D2.xxf(QΔP = {5.4}

We use the data from MA [16] and THEW [28] to find the constant in eq. 5.2. The model, developed from these data, is as shown in eq. 5.5 (Q in m3/s and Dn in meter). Comparison between observed data and predicted data is shown in fig. 5.3.

For pressure drop (bar) across inlet and overflow port (oil outlet);

4nD

2.3Q16oΔP = {5.5a}

For pressure drop (bar) across inlet and underflow port (water oultet);

4nD

2.2Q4.6uΔP = {5.5b}

85


II-42

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00Observed pressure drop (bar)

Pred

icte

d pr

essu

re d

rop

(bar

)

+10%

-10%

Δpo (bar) = 16 Q 2.3 /Dn 4

Fig. 5.3a Relation between observed pressure drop (inlet/overflow) of Thew cyclone and predicted pressure drop

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0Observed pressure drop (bar)

Pred

icte

d pr

essu

re d

rop

(bar

)

+10%

-10%

Δpu (bar) = 4.6 Q 2.2 /Dn 4

Fig 5.3b Relation between observed pressure drop (inlet/underflow) of Thew cyclone and predicted pressure drop

From THEW’s research [28], Split ratio (Rf) or ratio between overflow and inlet flow has some effect on pressure drop. However the effect of this parameter on under flow pressure drop is very small, thus, negligible. There is more effect on overflow pressure drop. From THEW’s and MA’s data, we can find the empirical correlation between split ratio and then can transform eq. 5.5a to account for the split ratio. The modified equation is as follow;

0.1611

)fR(12.6

4nD

2.3Q16oΔp⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

−⋅= {5.5c}

However, the correlation is developed from relatively small set of data. Furthermore, range of split ratio, generally used, is around 1 to 10%. Within this range, eq. 5.5a alone can predict the efficiency with an error of only 10-20%. So we recommend using eq. 5.5a and

86


II-43

5.5b to predict the pressure drop of the hydrocyclone. Anyway, to allow for prediction error, safety factor of 1.2 should be applied.

5.2 Three-phase hydrocyclone

Prof. AURELLE and MA develop the three-phase hydrocyclone in order to create a new type of hydrocyclone that can separate both oil and suspended solids simultaneously in the same unit. From this concept, MA proposed the new hydrocyclone, which is the combination between THEW liquid-liquid hydrocyclone and RIETEMA solid-liquid hydrocyclone, as shown in fig. 5.4. Geometry of this hydrocyclone is nearly identical to the 2 originals with slight adaptation.

Solid-liquid part Liquid-liquid part (Thew’s part)

DoDDs

DiDu

Dp

L5 L3L1L2

L4

Note: Di/D=0.25 for 1- inlet and 0.175 for 2- inlet, Do/D=0.43,Ds/D=0.28, Du/D=0.19, Dp/D=0.034, L1/D=0.4,L2/D=5, L3/D=15, L4/D=0.3, Solid-liquid part cone angle=12o, for liquid-liquid part=1.5o

Fig. 5.4 Three-phase hydrocyclone

MA testd the performance of the prototype of this hydrocyclone. It showed very good efficiency, which relatively conforms to the efficiency obtaining from separate solid-liquid and liquid-liquid hydrocyclone. He also studied the influence of important parameters to efficiency of this new hydrocyclone. However he did not propose the model. So we have to develop new model for three-phase hydrocyclone. Model development detail will be described in section 5.2.1.

5.2.1 Model development and verification for liquid-liquid section

From the fact that three-phase cyclone is the combination between THEW and RIETEMA hydrocyclone. New model for liquid-liquid and solid-liquid part should conform to that of each separate hydrocyclone.

For liquid-liquid hydrocyclone, we will apply the model of MA’s, as stated in section 5.1, to three-phase hydrocyclone. The problem is how to adapt MA’s model to this new hydrocyclone. From MA’s research, he observed that the phenomena in three-phase hydrocyclone, such as oil central core formation, etc., are relatively identical to THEW hydrocyclone. From this, we assume that the driving force of oil part in the hydrocyclone should be identical to normal THEW hydrocyclone at the same flowrate and nominal diameter (Dn in fig.5.1 = Do in fig. 5.4 = ND.). The driving force in hydrocyclone is generated by energy of feed flowrate. From geometry in fig. 5.1 and 5.4, we get the following equations.

( ) i(Thew)VThewα3iV3α ⋅=⋅ φφ

87


II-44

( ) ⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛⋅=

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛⋅ 2)i(Thew)π(D

4Q0.52)3iπ(D

4Q3α

φφ

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⋅=⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⋅2))

0.5NDπ(0.25(

4Q0.52))

0.43NDπ(0.25(

4Q3α φ

676.02

0.430.50.53α =⎟

⎠⎞

⎜⎝⎛⋅=⋅φ {5.6}

We use α3φ in eq. 5.6 with the model in eq. 5.1 and 5.2 to predict the efficiency of the oil part of three-phase hydrocyclone and compare the result with observed value from MA’s data [16]. Comparison result in fig. 5.5 shows that the error from prediction is ± 20%, which is acceptable. We have tried to select the value of α3φ arbitrarily and found that;

• If the value of α3φ is lower, the model will predict too low efficiency.

• For higher value of α3φ, it may provide better curve fitting but we do not have any data to support the use of it. The value of 0.676 seems more appropriate.

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

120.0%

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%Observed efficiency (%)

Pred

icte

d ef

ficie

ncy

(%)

+20%

-20%

Fig. 5.5 Relation between observed efficiency and predicted efficiency of liquid-liquid (Thew) part of three-phase hydrocyclone

From model verification result, we can conclude and propose the model of liquid-liquid part of three-phase hydrocyclone as well as its limitation as follows,

1. To predict removal efficiency of three-phase hydrocyclone, graded efficiency can be calculated by eq. 5.7 and 5.8.Dn in this case is equal to Dc. And L in this case is equal to L5.

∫=∫L

0 WdZdR

vzzR UdR {5.7a}

88


II-45

RV 2

18μ

2ΔρdU ⋅= {5.7b}

0.65)RnD

(2iD

4π(Q/2)0.676V

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

= {5.7c}

3

zRR1.19

2

zRR8.63

zRR123.33

zWW

⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−= {5.7d}

2))2/tan(Znπ(0.5D

QzW

β⋅−= {5.7e}

For d ≥ dc,

100%dη = {5.8a}

For d < dc,

%100)2)

2nD

(0.1862)2nD

((

)2)2nD

(0.1862d(R

dη ⋅−

−= {5.8b}

2. To design the hydrocyclone by the model, it is recommended to select the cut size that covers majority of oil droplet in the wastewater and provide a safety factor around 10% to 25%, because the model is apt to predict too small cut size. For example, if the desired cut size is 50 microns, it is recommended to select 50(1-0.25) = 37.5 microns for eq. 5.7 and 5.8.


1) The model is valid for three phase hydrocyclone with geometry of the oil part conforms to that of THEW.

2) It is recommended to use the model only for droplet diameter of 20 microns or greater. For smaller droplet, the model can also be applied, but for comparison purpose only.

3) Eq. 5.7c is valid for the hydrocyclone with 2 inlet ports only. If the hydrocyclone has only 1 inlet port, replace Q/2 with Q. However, using 2 inlet ports is recommended for its hydraulics stability. Please note that the size of 2 inlet ports will be smaller than single inlet port to keep the inlet area constant.

5.2.2 Model development and verification for solid-liquid section

From the same reason as oil-part model development, we will base our model for solid-liquid separation on RIETEMA hydrocyclone’s model. In MA’s research, he used only 2 sizes of suspended solids. So the data is not sufficient to develop the model. However, geometry of this part of three-phase hydrocyclone is identical to RIETEMA’s. So RITEMA’s model should be applied without any modification.

89


II-46

RIETEMA [30] defined the performance of the hydrocyclone in form of d50% and the dimensionless number Cy50, as shown in eq. 5.9. To find graded efficiency besides d50%, correlation of YOSHIOKA and HOTTA [30], for 2% < ηd <98%, may be applied (eq. 5.10).

50CyQcρcη

Δp)4L2)(Lcρss(ρ250d

=⋅−−

{5.9}

%100)

30.115)50%d

d(e1(dη ⋅

−−

−= {5.10}

The value of Cy50 of MA’s hydrocyclone can not be calculated due to lacking of data. However, RIETEMA recommended the value of Cy50 around 3.5 for RIETEMA type hydrocyclone, which MA used in his hydrocyclone. So we recommend to use Cy50 =3.5 to calculate the efficiency of solid part of three-phase hydrocyclone.

5.2.3 Generalized Model for pressure drop of three-phase hydrocyclone

MA proposed the model for pressure drop calculation for his prototype hydrocyclone. For the prototype, he used Do = 14 mm and D = 32 mm. Pressure drops across various ports of the prototype (in bar, m3/s) can be calculated from the following equations,

2.111.364QwaterΔP = {5.11a}

2.340.951QssΔP = {5.11b} 03.2140.3 QPoil =Δ {5.11c}

The equations are valid only for 14/32-mm. three-phase hydrocyclone. Thus, to extend the valid range of the equations or to develop generalized model. We consider 2 approaches, i.e.,

• Similarity approach

From TRAWINSKY [33], he suggested that hydrocyclone, like other centrifugal machines, is subject to concept of similarity or affinity law.

Δρ250%d1DΔP −−∝ {5.12}

So combination of eq. 5.11 and 5.12 can be used to predict the pressure drops of any size of three-phase hydrocyclone by calculating the pressure drops of 14/32-mm. hydrocyclone at given flowrate by eq. 5.11 first, then use eq. 5.12 to find the flowrate at given hydrocyclone diameter and given characteristic of wastewater.

• Empirical approach

Similarity approach is theoretical based. In practice, many factors may cause some discrepancies from theoretical value. To account for these factors, empirical approach is introduced. We base our model on eq. 5.4 that we successfully applied for two-phase hydrocyclone.

90


II-47

To develop and verify empirical model, we use the data from MA’s and THEW’s research [16], [18]. The empirical models are as shown in eq. 5.13 (ΔP in Bar, Q in m3/s, D in m.).

4D

2.12Q49.8waterΔP = {5.13a}

4D

2.34Q21ssΔP = {5.13b}

4D

2.03Q55oilΔP = {5.13c}

Comparisons between the predicted pressure drops of the 2 approaches are shown in fig. 5.6. The graphs show that the two approaches give very accurate results. However, using eq. 5.11 and 5.12 may cause some difficulty because it requires calculation for d50% first. So using eq. 5.13 may be more convenient.

0.000

0.500

1.000

1.500

2.000

2.500

3.000

3.500

4.000

4.500

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50Observed pressure drop (bar)

Pred

icte

d pr

essu

re d

rop

(bar

)

Method 2- Similarlity approachMethod 1- Empirical approach

(Empirical approach: Δp D4/Q2.12 = 49.8 )

Fig. 5.6a Relation between observed pressure drop (bar) across inlet and water outlet and predicted value from 2 approaches

91


II-48

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

Observed pressure drop (bar)

Pred

icte

d pr

essu

re d

rop

(bar

)

Method 2- Similarity approachMethod 1- Empirical approach

(Empirical approach: Δp D4/Q2.34= 21.0 )

Fig. 5.6b Relation between observed pressure drop (bar) across inlet and SS outlet and predicted value from 2 approaches

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00


Pred

icte

d pr

essu

re d

rop

(bar

)

Method 2- Similarity approachMethod 1- Empirical approach

(Empirical approach: Δp D4/Q2.03= 55 )

Fig. 5.6c Relation between observed pressure drop (bar) across inlet and oil outlet and predicted value from 2 approaches

92


II-49

Chapter 6 Membrane process

Membrane process is a separation process based, mainly but not entirely, on filtration concept. It can be said that the membrane is a very fine screen or filter. Theoretically, we can always separate one or more components from fluid stream providing that the filter chosen is suitable for size difference. Membranes can be categorized by their separation characteristics, i.e., microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). MF and UF have relatively large pore sizes, so they work rather like screens or sieve filters. While separation by NF and RO, which have very tiny pore sizes, is not simply by size alone but involves more complex factors, such as osmotic pressure. So we can group the membranes processes into 2 categories, i.e., 1) MF and UF, 2) NF and RO.

Membrane process design depends mainly on wastewater characteristics. So, the best way to design the membrane process is to perform feasible study in lab scale or pilot scale. However, our lab has several researches on UF. From these researches, we have gained some understanding of phenomena taking place in the process. Then we can use our result to set a guideline for conceptual design or preliminary evaluation of membrane process.

6.1 Ultrafiltration

Because of their pore sizes, MF and UF are suitable for finely dispersed emulsion, such as, secondary emulsion, macroemulsion and microemulsion. However, there is only one research on MF of oily wastewater treatment in Prof. AURELLE’s team. Then, despite of its feasibility on cutting oil emulsion treatment, which is one of the important applications of membrane processes on oily wastewater treatment, we will not include MF in our thesis for there is insufficient data. So we will emphasize on UF.

There are 2 main types of UF, as well as other, membrane processes, based on flow pattern, i.e., dead-end and cross-flow. In dead end reactor, wastewater will be fed in perpendicular direction to membrane surface. So this mode of operation is rather like cake filtration. Dead-end process is normally used in bench scale experiment. For cross-flow, wastewater will be fed in tangential direction, parallel to membrane surface. It is this mode of operation that is widely used in real life situation. So our researches are related to this process.

For wastewater treatment, our aim is to reduce the quantity of the wastewater as much as possible while the effluent quality still meets effluent standard. So wastewater will be recycled repeatedly until its volume reaches required limit. In this case, UF system is normally designed as batch processes, as shown in fig. 6.1.

Permeate

Retentate

Membrane

Feed pump

Storage tank

Fig. 6.1 Typical schematic diagram of Cross-flow membrane process

93


II-50

From fig. 6.1, we have to apply enough pressure to force components that can pass through the membrane pores to the other side. The components of influent that can pass through the membrane is called permeate. The rest that can not pass is called retentate. The pressure difference between retentate side and permeate is called transmembrane pressure (Pt). The retained components will accumulate at the surface of membrane and its concentration will be increased. This accumalation and high concentratio, called concentration polarization or gel polarization, will hinder the permeate flow. Some components may lodge in the pores. These phenomena will make the membrane clog and rate of permeate passing through the membrane or permeate flux will decrease. In cross-flow process, we can alter the superficial (or recirculation) velocity at the membrane surface (V) by changing recirculation flowrate to reduce the effect of polarization. From this brief operating principle, we can see that there are several parameters, relates to membrane process design. So models of membrane will be developed to describe the relation between these parameters.

There are several researches in our lab [10], [11], [18]-[22] on membrane processes. However, they were based mainly on application. Only some of them provided models. Furthermore those models are limited by their scopes of the experiment, then they are not in general forms to apply to general case. In this thesis, we will try to develop the generalized models, based on well-accepted theoretical models, i.e. resistance model and film model.

6.1.1 Resistance model

General form of resistance model [38], [54], [18], [11] is similar to the equation for electrical calculation as shown in eq. 6.1.

gRfRmRtP

J++

= {6.1a}

Membrane resistance (Rm) and fouling resistance (Rf) are property of membrane and relatively unaffected by operating condition [38]. So it is normally be summed together and called intrinsic membrane resistance (R’m). Eq. 6.1a , then, will become,

gRm'RtP

J+

= {6.1b}

Gel resistance (Rg) is the function of Pt and V as shown in eq. 6.2. α and φ are numerical constants.

tPαVgR ⋅= φ {6.2}

This model can give the accurate evolution of flux with Pt, starting from pressure controlled region, which the flux varies with Pt, to mass transfer controlled region, which the flux is relatively constant, as shown in fig. 6.2.

From researches in our lab, we can summarize the value of R’m, φ and α for many operating conditions, as tabulated in table 6.1. φ and α are dependent of wastewater inlet concentration so it can be applied only to their corresponding concentration only.

94


II-51

Tabl

e 6.

1 Su

mm

ary

of p

aram

eter

s of r

esis

tanc

e m

odel

from

UF

rese

arch

es o

n oi

ly w

aste

wat

er tr

eatm

ent (

refe

renc

e te

mpe

ratu

re =

20O

C)

95


II-52

Pressure controlled region

Mass transfer controlled region

Wat

er fl

ux

Higher recirculation flowHiger temperatureLower concentration

Transmembrane pressure (Pt)

Flux

Fig. 6.2 Typical relation between flux and various parameters

6.1.2 Film theory based model

This model is developed under the assumption that the process is in the mass transfer controlled region. So flux is assumed to be controlled by mass transfer phenomena and independent of Pt. Operating UF in this region will maximize the flux, thus minimize the size of membrane. The general form of film theory based model [38], [54] is shown in eq. 6.3.

)oCgC

ln(βkVJ = {6.3}

From the equation, k is mass transfer coefficient. Cg is gel concentration, which depends on type of wastewater. V is recirculation velocity. Two typical characteristic curves of flux VS. retentate concentration are shown in fig. 6.3b. (Other curves also exist, but they are very rare cases)

For the first type, the curve is flat without inflection point. This type of curve will cross the horizontal axis at C . This Cg g remains constant for the whole range of retentate concentration.

For the second type, The curve presents an inflection point. In this case, we can say that there are two Cg. The first one is obtained from extending the steeper part of the graph to cross the X-axis. But it is not the real Cg and used only for flux calculation at lower range of retentate concentration. The real Cg is obtained from the graph after inflection point. UF of cutting oil emulsion will be in this category. In this case, the real Cg normally crosses the X-axis at approximately 100% concentration. This can imply that, theoretically, we can use UF to filter the oily wastewater until the retentate become water-free oil. However, the flux will become very low and the operation may become unacceptable from economic point of view.

Log (Concentration)

V1

Cg

Flux V2>V1

Fig. 6.3a Relation between flux and concentration at any recirculation velocity [38]

96


II-53

Log (Concentration)

1 2

Cg case 1

Cg case 2

C’g case 2

Flux

Fig. 6.3b Typical characteristic curve of concentration VS. Flux (Log-Normal scale) [38]

From researches in our lab, we can summarize the value of k, Cg and β for various kinds of membranes for many types and inlet concentration of wastewater, as tabulated in table 6.2.


From section 6.1.1, it shows that the resistance model can predict flux at any pressure. However, the constants in eq. 6.2 are valid for only their own specific operating conditions, esp. the influent oil concentration. On the other hand, eq. 6.3, even though it is supposed to be valid only in mass transfer region, can be used to predict flux at any influent oil concentration, providing that the Cg is known. So, it is interesting to combine eq. 6.2 and 6.3 to see if they can be used to predict flux at any influent concentration and pressure or not. The procedures used to combine film theory and resistance theory, which are divided into 2 cases, are as shown below.

Case 1: Know k, β and Cg

1. Find J C,Vref

If we know the value of k and β at one known velocity (called reference velocity, Vref), we can calculate limiting flux at any required concentration (called C) by the film theory.

)C

gCln(β

refkVVrefC,J = {6.4a}

However, even the film theory is supposed to be pressure-independent, it should be noted that k and β are obtained from experimental data, which are conducted at certain value of pressure. At that reference pressure (Pref), the flux may not yet fully reach the mass transfer controlled region. In fact, we frequently find that flux/concentration curve never the reach really flat part within the recommended operating range of pressure (generally 0 to 4 bar), especially when velocity is high.

In this case, we have tried to verify if the film theory is still valid. We studied the flux/concentration relation of cutting oil macroemulsion (Elf SeraftA) at P = 2 and 3.5 bar, V from 0.7 to 2.8 m/s from Belkacem’s data [18]. We found that eq. 6.3 is valid at P = 3.5 bar, where is in the mass transfer controlled region. And we also found that the form of eq.6.3 still holds at P = 2, where flux does not yet reach fully mass transfer controlled region at low C and/or high V. But there are some little differences in the values of k, β and Cg.

97


II-54

O C

) Ta

ble

6.2

Sum

mar

y of

par

amet

ers o

f film

mod

el fr

om U

F re

sear

ches

on

oily

was

tew

ater

trea

tmen

t (re

fere

nce

tem

pera

ture

= 2

0

98


II-55

So we conclude that eq. 6.3 is valid even when the mass transfer controlled region is not fully reached in some condition at the reference pressure. However, it is recommended to select the reference pressure as high as possible, at the very least, more than 50% of recommended operating range of the membrane. It should be noted that, in this case, JC,Vref from eq. 6.4a may not exactly be limiting flux but it represents the flux at required concentration at the reference velocity and pressure. The equation may be rewritten as

)C

gCln(β

refkVPrefVref,C,J = {6.4b}

2. Find J C,V

Again from the film theory, we can find flux at required concentration and velocity and at reference pressure by eq. 6.5.

)C

gCln(βkVPrefV,C,J = {6.5}

3. Find αC and φC

From the resistance theory, if we know at least 2 fluxs at the same concentration but at different velocity, we can find α and φ at that concentration (α and φC C) by the following equations.

)VrefV

ln(

refPmRVC,JJKrefPmRVC,J

ln

Cα⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

−

−

= {6.6a}

β)refVV(

PrefVref,C,JPrefV,C,J

JK == {6.6b}

So eq. 6.6a can be rewritten as,

)VrefV

ln(

refPmRVC,J

β)refV

V(refPmRVC,Jln

Cα⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

−

−

= {6.6c}

PrefV,C,JrefPαcV

mRPrefV,C,JrefPC

−=φ {6.6d}

Membrane resistance (Rm) (or modified resistance R’m) depends on type of membrane and emulsion treated. Then it is usually constant for any given membrane and emulsion. In case of cutting oil emulsion, we found that there is, more or less, no fouling occurred. So we can safely replace R’m by Rm, as shown in eq. 6.6.

99


II-56

Besides using the reference velocity and required velocity, other dummy velocities can also be used to cross-check the value of αc and φc. Calculated values of the two parameters may not be exactly the same as those calculated from Vref and V, since there is normally some calculation error. Using average values of these αc and φc is recommended. From αc and φc, we can calculate flux at any transmembrane pressure and at the required concentration by the resistance theory (eq. 6.7).

tPαcVCmRtP

PtV,C,Jφ+

= {6.7}

To verify the procedure from eq. 6.4 to eq. 6.6, we use data from many researches [10], [11], [18], [20], which φ, Cg, and α of corresponding cases are available. Example of verification data is provided in Annex A5. In this case, we use φ, Cg, and α obtained from UF test of cutting oil macroemulsion (Elf SeraftA) at influent concentration (C ) of 4% V/V, Pref ref = 2 bar, Vref = 1.4 m/s as a reference condition to predict flux/ pressure relation at C = 2 and 8 % V/V. Predicted relations and comparison between observed flux and predicted flux are as in fig. 6.4 and 6.5. From the graphs, it shows that eq. 6.4 to 6.6 can be effectively used to extend the range of eq. 6.1, 6.2 and 6.3 to cover any influent concentration.

0

20

40

60

80

100

120

140

160

180

0 0.5 1 1.5 2 2.5 3

Transmembrane Pressure (Bar)

Pred

icte

d Fl

ux (l

/ (h.

m2 ))

Observed, C = 2% Predicted, C = 2% Observed, C = 8% Predicted, C = 8% Reference, C = 4%

Fig. 6.4 Relation between UF permeate flux and Transmembrane pressure at reference concentration (C) of = 4%, V = 1.4 m/s and Predicted relations at C = 2 and 8%

100


II-57

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

180.00

0 20 40 60 80 100 120 140 160 180

Observed flux (l/ (h.m2))

Pred

icte

d Fl

ux (l

/ (h.

m2 ))

-10%

+10%

Fig. 6.5 Relation between observed and predicted flux by resistance model for ultrafiltration of macroemulsion 4% conc. and extend to cover other conc. by film model

Case 2: Know α, φ and Cg

1. Find J Cref,V

From resistance theory, we can find J , when φ and αCref,V Cref Cref are known, by the following equation.

refPαCrefVCrefm'RrefP

VCref,J⋅+

=φ

{6.8}

Again, we recommend selecting the reference pressure as high as possible to make sure that the calculated flux is in the mass transfer controlled region.

2. Find JC,V

From film theory, we can write that;

)refCgC

ln(

)C

gCln(

VCref,JVC,J = {6.9}

3. Find J C,Vref

In the same way as 1 and 2, we can find J , when φ and αC,Vref Cref Cref are known, by the following equation.

101


II-58

⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜

⎝

⎛

⋅⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛

⋅+=

)refCgC

ln(

)C

gCln(

refPαCrefrefVCrefm'R

refPVrefC,J

φ {6.10}

4. Find αC and φC

αc and φc can be calculated by the same method as shown in case 1, no. 3.

6.1.4 Flux prediction for mixture of cutting oil microemulsion and macroemulsion

Wasted cutting oil emulsion is one of the most difficult-to-treat oily wastewater, mainly because its stability and its very tiny oil droplet. From the researches stated above, we could see that treatment of this wastewater is the main application of UF on oily wastewater treatment. From table 6.1 and 6.2, we have some information on UF of microemulsion and macroemulsion, summarized from many researches. However, in real life situation, some mechanical workshops or factories may use more than 1 type of cutting oil, to meet individual process requirement. So it will be interesting to predict the flux of the mixture of them.

To do this, we start with components of the cutting oil emulsion. Two main components of cutting oil concentrate are base oil and surfactants/co-surfactants. Other additives are also present but their quantities are usually very low. For macroemulsion, the ratio of oil in cutting oil concentrate is around 80%. For microemulsion, the ratio of oil is around 30 to 40%. The surfactants and co-surfactants are the components that give the emulsion its characteristic.

When 2 types of emulsion are mixed. The oil components will be summed up. Concentration of oil in the mixture is, therefore, the summation of oil concentration in each emulsion. Excess surfactants/co-surfactants in microemulsion part will reduce oil droplet size of macroemulsion, as they do in microemulsion. However, their quantities will not be sufficient to convert all of macroemulsion into microemulsion. The resultant flux, then, will be neither the same as the flux of pure microemulsion, nor that of pure macroemulsion. But it will surely fall between these 2 extreme cases. Accurate prediction may require the knowledge of the chemistry of surfactants/co-surfactants.

However, since the excess amount of surfactants/co-surfactants in the mixture is directly proportional to the ratio of the microemulsion presenting in the mixture, we may estimate that flux of the mixture is the weighted average of the two emulsions.

To formulate the equations, consider the mixture of Cmac % by V of macroemulsion concentrate and Cmic % by V of microemulsion concentrate. Ratios of oil in the macroemulsion and microemulsion concentrate are R and Rmac mic %, respectively. Total oil concentration in mixture can be written as follows.

{6.10a} micRmicCmacRmacCmixoil,C +=

However, if the concentrations are presented in %V of oil (not concentrate), total oil in the mixture is a simple summation of the two concentrations (eq. 6.10b).

102


II-59

{6.10b} micoil,Cmacoil,Cmixoil,C +=

Flux of macroemulsion and microemulsion at the total oil concentration (Jmac:Coil,mix and Jmic:Coil, mix) can be calculated by the procedure described in the previous section. Therefore, the flux of the mixture can be calculated by the following equation.

mixoil,CmixCoil,mic,Jmicoil,CmixCoil,:macJmacoil,C

mixJ+

= {6.11}

To verify this idea, we used reference data of flux of pure macroemulsion and microemulsion, as used in section 6.1.3, to predict flux of various ratios of micro/macroemulsion mixture. Relations between predicted values and observed data are presented in fig. 6.6 and 6.7. Even though, the error from prediction is around 20%, this procedure will be a useful tool for the preliminary estimation of mixture flux, especially when UF test data on the wastewater is not available

(Module: UFP2, Membrane: IRIS 3042, T = 20o C., data from [11]. The lines show predicted value.)

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

180.00

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

Pressure (bar)

Flux

(l/(s

q.m

.h))

4% Macro1% MicroV = 2.8 m/s



Fig 6.6 Comparison between predicted flux and observed flux for UF of micro/ macroemulsion mixture (Conc. shown as % by volume of concentrate)

103


II-60

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

180.00

0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00

Observed flux (l/(sq.m.h)

Pred

icte

d flu

x (l/

(sq.

m.h

)

Macro 4%, Micro 1% Macro 4%, Micro 2% Macro4%, Micro 4%

+20%

Fig 6.7 Relation between observed and predicted flux of micro/macroemulsion mixture from weighted average method between flux of whole micro and macroemulsion at the same total oil concentration

6.1.5 Theoretical flux prediction for batch cross-flowUF process

Models in section 6.1.1 and 6.1.2 are developed under the condition that permeate is recycled back to feed storage tank, so the inlet concentration is kept constant. For real situation, the process may be designed as batch process, continuous process, single stage process or multi stage process. In these cases, mass balance will be taken into account.

However, for wastewater treatment process, UF is normally designed as batch system, as shown in fig. 6.1. In batch process, the concentration of retentate will be increased up to required limit or as much as possible. From film theory, flux will decrease when the concentration increases. And from our procedure to predict flux in section 6.1.3, it will be interesting to predict evolution of retentate volume, permeate volume, and flux with time. These data is important to design UF process to meet the required operating time. However, it must be noted that these evolutions are based on the assumption of fresh (not used) emulsion thus there is no fouling from any other foreign materials, which, in practice, hardly exists. Anyway, these data will provide the idea about the approximate size of membrane required.

To find permeate volume, we consider that a small volume of permeate (dVol), passing through a UF membrane of area A at a small time (dt), will be defined by the following equation.

{6.12} AdtJ(c)dVol ⋅=

If the system is operated in the mass transfer controlled region, the film theory can be applied. Eq. 6.12 can be rewritten as follows,

AdtCgC

lnβkVdVol ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅= {6.13}

If we start with initial wastewater volume Vol and concentration Co o and the rejection of oil is totally completed, which is true from our every test, the concentration C will be the function of the permeate volume at that moment V.

104


II-61

Vol)o(VoloVoloC

C−

= {6.14}

Eq. 6.9 can be rewritten as,

AdtoVoloC

gVol)Co(VollnβkVdVol ⎟

⎟

⎠

⎞

⎜⎜

⎝

⎛ −⋅= {6.15a}

∫=∫

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ −

t

0AdtβkV

lfinalVo

oVol

oVoloCgVol)Co(Vol

ln

dVol {6.15b}

However, the system may not be operated in the mass transfer controlled region for the whole time. In this case, the function J(c) in eq. 6.12 can be calculated by eq 6.7. Eq. 6.12 and 6.15b are in the form of integration of [1/ln (x)], so they can not be written in general form since the equation will be infinity at x =1. However, a definite integration is possible, using numerical method that can be calculated by computer.

In this thesis, we have compared the theoretical result, using reference data stated in section 6.1.3, with observed results from UF test on used cutting oil macroemulsion (3% V of concentrate or 2.4%V of oil, from Willamette SAS factory) from WANICHKUL’s research [11], as shown in fig. 6.8a. The x-marked circles indicate the observed flux of fresh emulsion. From the graph, it shows that the model can accurately predict flux of fresh (unused) emulsion as the circles are closed to the theoretical flux curve.

Comparing with observed flux of the used emulsion, the graph shows that, at low concentration, theoretical flux is greater than observed value. This simply because additional fouling from foreign material in the emulsion. However, at high concentration, theoretical flux is, somehow, lower than observed value. This can be explained by partial degradation of the used emulsion. During its working lifetime, cutting oil emulsion will subject to many foreign material, such as coated oil on specimen surface, small scraps of specimen, leaked lubricant, and heat. So its quality, as well as its stability, will gradually deteriorate. This is proven by milky appearance of used emulsion, compared with the translucent or transparent characteristic of fresh emulsion. When partial oil is destabilized to be free oil, this means the concentration of oil in emulsion form may be lower than its initial value. At lower concentration, the effect of fouling overwhelms the effect of reduced concentration, so the theoretical flux is higher than the observed value. However, at higher concentration where the concentration effect is stronger, the theoretical flux shows lower value.

Fig. 6.8b shows evolution of permeate volume with time, the result from integration of eq. 6.12. From the observed data, it confirms that the macroemulsion can be ultrafitrated until the retentate is relatively pure oil. In this case, initial volume of 1643 l of 2.4%V of oil is ultrafiltrated to the final volume of 40 l. The theoretical time required to do so is 45 hours, compared to observed value of 34 hours. However, the theoretical volume of permeate is higher until almost at the end of the operation. Fig.6.8c shows evolution of theoretical flux with concentration. It must be noted that eq 6.12 to 6.15 are based on the assumption that no additional emulsion is added to the storage tank. If the emulsion is added, eq. 6.12 must be modified to account for it.

105


II-62

(Used cutting oil macroemulsion :initial volume 1643 l, final volume 40 l,initial concentration 2.4% by volume of oil (not concentrate): Module UFP10: membrane IRIS 3042)

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 200 400 600 800 1000 1200 1400 1600 1800

Permeate volume (l)

flux

(l/(s

q.m

.h))

Theoretical data at P = 1.0 Bar, v = 1.17 m/s) Observed data Theoretical data at P = 1.5 Bar, v = 1.40 m/s)

= Observed data from UF test ofnew cutting oil at the same condition

P = 1.0 bar, v = 1.17 m/s P = 1.5 bar, v = 1.40 m/s

Fig. 6.8a Relation between Flux VS. theoretical and observed permeate volume

(Used cutting oil macroemulsion :initial volume 1643 l, final volume 40 l,

initial concentration 2.4% by volume as oil (not as concentrate): Module UFP10: membrane IRIS 3042)

0

200

400

600

800

1000

1200

1400

1600

1800

0.000 5.000 10.000 15.000 20.000 25.000 30.000 35.000 40.000 45.000 50.000

Time (h)

Perm

eate

vol

ume

(l)

Theoretical data at P = 1.0 Bar, v = 1.17 m/s) Theoretical data at P = 1.5 Bar, v = 1.4 m/s) Observed data

P = 1.0 bar, v = 1.17 m/s P = 1.5 bar, v = 1.40 m/s

Fig 6.8b Relation between time VS. theoretical and observed permeate volume

106


II-63

(Elf SeralfABS cutiing oil macroemulsion: Membrane IRIS 3042, 20oC)

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

1 10 100

Retentate concentration (% V of oil)

Perm

eate

flux

(l/m

2 .h)

P = 1 bar, V = 1.17 m/s P = 1.5 bar, V = 1.4 m/s

Fig. 6.8c Relation between theoretical flux VS. concentration of oil in retentate

6.1.6 UF efficiency

From many researches, with appropriate UF membrane pore size, it clearly shows that oil, even in form of very tiny droplets in macroemulsion and microemulsion, cannot pass the membrane. We can say that the removal efficiency of UF is 100%. Selection of membrane involves many parameters. However, AURELLE [quoted by [18]] have grouped these parameters and proposed 3 brief criteria, i.e.,

• Pore size of membrane: To prevent oil droplets to pass through the membrane pore, the size of the pore, firstly, must be smaller than the droplets. From researches in our lab, we recommend that minimum pore size should be 1/4 to 1/3 of average droplet size.

• Characteristic of membrane: for oil/water separation, membrane should be hydrophilic. Membrane material should not react with the wastewater, which can cause pour clogging. Hydrophilic material, such as polyacrylic, cellulose acetate, zirconium oxide, etc., is recommended. Naturally without special treatment or coating, polysulfonate tends to be fouled by oil, resulting in low flux and frequent washing.

• Operating condition: Operating pressure should be less than capillary pressure required to force the oil droplets through the membrane pores. Capillary pressure increases with the hydrophilicity of membrane and decreasing of pore size. However, if pore size and hydrophilicity are carefully selected, the capillary pressure is normally higher than recommended maximum pressure of the membrane.

107


II-64

However, other soluble pollutants, especially surfactants and co-surfactants, can pass through the membrane. So, after UF, these chemicals are still present in permeate, resulting in high TOD, and require further treatment.

6.1.7 Minimum and maximum transmembrane pressure and power required

Transmembrane pressure and recirculation velocity are important factors in UF process. In case of the pressure, many researches [11], [18] recommended that using moderate pressure would reduce fouling, thus increasing operating time before cleaning. Increasing in velocity will reduce polarization effect and lead to increasing of flux. However, using high recirculation velocity inevitably cause high pressure drop, then the transmembrane pressure will be increased accordingly.

Thus, the minimum transmembrane pressure required for any recirculation velocity will be equal to pressure drop caused by that velocity. The pressure drop can be calculated, as a friction loss of retentate flowing through narrow channel between membrane surface and UF module wall, by general pressure drop formula, such as Darcy-Weisbach’s equation. It must be noted that the friction factor (f) depends on Reynolds number, which, in turn, depends on viscosity of liquid. For oily wastewater, the viscosity varies with concentration of oil. Fig. 6.9 shows relation of viscosity and oil concentration (as % V of concentrate) of the cutting oil macroemulsion (Elf Seraft ABS). So the pressure drop will be calculated using the highest viscosity in the selected operating range of concentration.

2

2VDLfminP = {6.16a}

For laminar flow,

Re64f = {6.16b}

For turbulent flow (Colebrook’s equation),

)0.5fRe

2.513.7e/Dlog(20.5f

1

⋅+⋅−= {6.16c}

e/D is ratio of roughness to diameter. For plate membrane, the flow channel is usually rectangular. D of this channel can be calculated as hydraulic diameter, as shown in eq. 6.16d. H and W are height and width of the rectangular channel respectively.

W)2(H4HWD+

= {6.16d}

However, since L of the UF module is small, even though the velocity is as high as 3 m/s and D is as low as 1.0 mm., P is still well below 0.5 bar. min

The maximum transmembrane pressure will be controlled by the lower pressure among 1) capillary pressure of oil droplet , 2) the maximum operating pressure, recommended by membrane manufacturer. Capillary pressure of oil droplet of diameter rd and membrane pore of diameter r can be calculated by the following equation [10].

108


II-65

⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

−

⎟⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜⎜

⎝

⎛

+−+⎟⎟⎠

⎞⎜⎜⎝

⎛

−−= 1

1/3

)o/wθ3sino/wsinθ(2o/wcosθ3

rdr4

2o/wθ3coso/w3cosθ

ro/wcosθ

o/w2γcapP {6.17}

Normally, the value of the capillary pressure is higher than the maximum operating pressure. For example, for UF of macroemulsion, θo/w= 135o, γo/w = 0.033 N/m, rd = 150 nm and r = 100 nm, the capillary pressure is 11.45 bar while the maximum pressure recommended by the manufacturer is only 4 bar.

(Elf SeralfABS cutiing oil macroemulsion, 20oC, 100% = pure cutting oil concentrate = 80% V(approx) oil )

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

0 10 20 30 40 50 60 70 80 90 10

Oil concentration (% V of cutting oil concentrate in water)

Vis

cosi

ty (c

p)

0

Fig. 6.9 Relation between oil concentration VS. viscosity of emulsion

For the power requirement of UF system, main power consumption is the power required to maintain transmembrane pressure at require flowrate (or recirculation velocity). This power can be calculated straightforwardly by the basic equation

overallηAVP

overallηQPPower ⋅⋅

=⋅

= {6.19}

Q in this case means recirculation flowrate, not the permeate flowrate. V is the recirculation velocity and A is the flow area of liquid in UF module (the channel between the membrane surface and the UF module wall). P in the equation is pump discharge pressure, which is the summation of transmembrane pressure and other headloss from pipe and values system. It should be noted that transmembrane pressure is average value of the pressure at the inlet and outlet of UF module. Overall efficiency of pump depends on pump type. For progressive cavity pump or progressive screw pump, the efficiency should be around 50-70%.

109


II-66

6.1.8 Conclusion and generalized model of UF

Both resistance and film theory based models can give accurate result if the numerical constants from table 6.1 and 6.2 are chosen correctly. So both models can be used as generalized models for UF process. It should be noted before using any parameters from any UF researches that every parameter varies with type of emulsion and membrane. Experimental procedure (such as recycling of permeate, preparation of emulsion by tap water or demineralised water, reporting concentration of oil concentration as %V of concentrate or % V of oil, etc.), as well as the difference from lot-to-lot of membrane manufacturing, also affects the result. So, if possible, we strongly recommend performing a feasibility test in bench scale or pilot scale, using real wastewater.

In this thesis, the combination of resistance and film models (section 6.1.3) is proposed and verified. Advantage of this combine model is that, if we know k, β and Cg and the flux at one influent concentration (Co), we can find flux at any concentration. This fact is very useful to extend the resistance model (section 6.1.1) at specific concentration to any concentration. The constant in table 6.1 and 6.2 can be used to preliminarily calculate UF process of cutting oil emulsion.

Estimation of flux of mixed micro/macroemulsion is verified in section 6.1.4. Furthermore, prediction of permeate flux of batch UF process is proven to be useful tool for process design for it can provide general idea of evolution of permeate flux with time (section 6.1.5). The oil removal efficiency of UF is considered to be 100% providing that the pore size is suitably chosen. Minimum transmembrane pressure is controlled by required pressure drop across UF module at the design recirculation velocity, which is usually lower than 0.5 bar. Maximum theoretical pressure is controlled by the lower value among capillary pressure and manufacturer recommended maximum pressure. Normally the former is much higher than the latter.

6.2 Nanofiltration and Reverse osmosis

NF and RO are the membrane processes that use very fine pore membranes. So it can be retain almost all of components in the fluid stream and let only the water pass through. Despite of their separation capability, they are normally costly and consume very high energy. Thus, they are not economical choices for wastewater treatment. However, it is proven by our lab [11], [18], [20] that it is useful for the post treatment of permeate from UF and MF for it can retain dissolved pollutants such as surfactant and co-surfactant.

The phenomena, taking place in NF and RO, are more complex than those of MF and UF. They involve osmotic pressure. So the model of RO and NF will be relatively complicate. However, because we use them only as post treatment, which the characteristic of influent is quite typical, so the process can be safely designed using simple criteria, such as aerial loading rate. From researches of our lab, we can summarize such criteria, as tabulated in table 6.3.

110


II-67

Ta

ble

6.3

a S

umm

ary

of R

O d

ata

on o

ily w

aste

wat

er tr

eatm

ent

111


II-68

Tabl

e 6

.3b

Sum

mar

y of

NF

data

on

oily

was

tew

ater

trea

tmen

t

112


II-69

Chapter 7 Heteroazeotropic Distillation

Heteroazeotropic distillation is a thermal treatment process that makes use of vapor/liquid/liquid equilibrium (VLLE) to separate oil from water. Concepts of VLLE and heteroazeotropic distillation are classical concepts for process engineers. However, the application on oily wastewater treatment is not realized, or if any, very few.

Heteroazeoptropic distillation phenomena normally take place in a distillation column. But, after condensation, evaporated water and hydrocarbons are condensed together and form distillate mixture of hydrocarbons and water again.

Our lab studies find that it is feasible to apply heteroazeotropic distillation process on treatment of oil/water emulsion by adding proper entrainer or extractant, which is relatively volatile hydrocarbons, into the mixture. The distillate in this case will consist of 2 separate layers, one of extractant on the top, and another of water at the bottom. The residue from distillation is water-free hydrocarbons (efficiency = 100%), which can be reused or recycled. Anyway, due to its relatively high-energy consumption, its application should be considered, also, from economic point of view, especially for relatively dilute wastewater.

Our lab emphasizes on its application for highly viscous emulsion treatment, like wasted slop from refineries and retentate left from emulsion treatment by membrane processes. These applications are feasible because it provides a chance to valorize these suppose-to-be-wasted materials.

Another application of heteroazeotropic distillation is the steam stripping, in which steam is supplied into mixture of relatively non-volatile and volatile components to strip the volatile components from the mixture. The steam is this case will act as the entrainer to extract the volatile components. This process can be viewed as the inverse process of heteroazeotropic distillation. If the mixture is a hydrocarbon/water mixture, the water will act as an entrainer. So it is called a self-entraining mixture.

Researches in our lab are based mainly on the study on types of entrainers and influent parameters, rather than developing distillation column. Main researches on heteroazeotropic distillation in our lab are thesis of LUCENA [24] on slop treatment and thesis of WANICHKUL [11] on retentate treatment. However, the concept of calculation, based on classical theory, can be applied to other oil/water mixtures.

7.1 Theoretical model

Models or procedure for heteroazeotropic distillation, in our case, are based on non-miscible mixture, like oil and water. For non-miscible binary mixture of water and hydrocarbon, typical isobar diagram of VLLE is as shown in fig. 7.1.

113


II-70

2 ph.vapor

Bubble curve

Pure H2O

Temperature

Azeotrope (H)2 ph. liquid

1 ph.vapor +1 ph. liquid

Pure hydrocarbon x,y yH

Dew curvesBoiling point of hydrocarbon

Boiling point of water

TH

Fig. 7.1 Isobar equilibrium diagram : Temperature-Concentration characteristic of immiscible binary mixture

From the figure, we can set the steps of calculation to create isobar diagram as follows,

1) Find bubble curve

For binary mixture satBPsat

AP +=+=+= θbBΠθb

AΠθbBAΠP {7.1}

Relations between vapor pressures (Πθb or Psat) and boiling temperature of A and B, in our case, water and hydrocarbons, can be found in any standard property tables, such as PERRY’s [2]. Then, we obtain the relation between vapor pressure and boiling temperature of the mixture from the summation of the vapor pressure of A and B, as shown in eq. 7.1. This can be done easily by a graphical method, as shown in fig. 7.2. When we select our design pressure (normally 1 atm), we can obtain heteroazeotropic temperature (TH) from the graph.

TH Temperature

Pressure

Pure A

Pure B

A+B

Pdesign

Fig. 7.2 Graphical method to find heteroazeotropic temperature

114


II-71

2) Find dew curves

From DALTON’s law

PAyAp ⋅= {7.2a} And

PByBp ⋅= {7.2b}

On the dew curves, partial pressure is equal to vapor pressure, then For condensation of A

P

θdAΠ

AyθdAΠPAyAp =→=⋅= {7.3a}

For condensation of B

P

θdBΠ

ByθdBΠPByBp =→=⋅= {7.3b}

At any temperature between boiling points of pure substances and heteroazotropic temperature (TH), we can find their corresponding vapor pressures (Π) from fig. 7.3a. Then, from eq. 7.3, we can obtain yA and yB. This can be done easily, again, by the graphical method as shown in fig. 7.3

T1 Temperature

Pressure

Pure A

Pure B

A+B

ΠA 1ΠB 1

T2

ΠA 2ΠB 2

Pure H2O

Temperature


TH

T1

T2

yB 1 yB 2 yA 2

yA 1

Calculate yA and yB from ΠA,ΠBby eq. 7.3, then, plot T,yA and T,yB to obtain dew curves

Fig. 7.3 Graphical method to find dew curves

3) Find heteroazeotropic composition

Heteroazeotropic composition (yH) can be calculated, based on eq. 7.1 and 7.3, as shown in eq. 7.4,

P

θbAΠ

AyHy == {7.3a}

From eq. 7.1,

θbBΠθb

AΠ

θbAΠ

Hy+

= {7.4}

115


II-72

Vapor pressures of A and B (water and hydrocarbons) can be obtained from any standard property tables [2]. yH indicates capability of extractant to extract water from wastewater. For example, if yH = 0.5 mole/mole, it means 1 mole of extractant can extract 1 mole of water. The higher the yH , the better the extractant.


From section 7.1, we can find yH of pure hydrocarbon that can be used as an extractant, i.e., alkane with the number of carbon from 6 atoms (n-Hexane) to 16 (n-Hexadecane). Theoretical values of yH of each alkane are tabulated in table 7.1. To compare the theoretical value to experimental result, we use data from LUCENA’s thesis on slop treatment, using decane and dodecane. Comparison result in table 7.1 shows that the theoretical values are slightly higher (3.5-5.5%) than observed values.

Table 7.1 Heterotropic temperature and composition from various extractants

Extractant Molecular

weight (g/mol)

TH (deg. C)

yH (by molar)

yH (by volume)

y H observed (by volume)

[24]


7.3 Conclusion and generalized model of heteroazeotropic distillation

From data verification result, we can conclude that the theoretical model, stated in section 7.1, provides very good prediction of heteroazotropic composition (yH). However, the safety factor (S.F) of 1.05 or 1.1 is recommended. When we know the water volume in our treated wastewater and determine the type of extractant, we can find quantity of extractant required for extracting the water from the following equation. yH in eq. 7.5 will be by volume basis, determined by the model in section 7.1. Heteroazeotropic distillation is suitable to treat or recover valuable residue, such as slop or retentate from UF of cutting oil emulsion, which is contaminated by relatively small amount of water.

Hy

)Hy(1waterVolumeS.FextractantVolume

−⋅= {7.5}

116


II-73

On the other hand, for stripping, we can find the theoretical quantity of steam or water required for extracting the design volume of volatile hydrocarbons from the following equations. Please note that the calculated volume of steam is the quantity required for heteroazeotropic distillation only. Additional heat may be required to raise the temperature up to the design point. Stripping is suitable to treat the waste polluted by small amount of volatile substance. The concept is also applied to essential-oil extraction from herbs or flowers in perfume or chemical industries.

)Hy(1Hy

nhydrocarboVolumeS.FsteamVolume−

⋅= {7.6a}

Or

)Hy(1Hy

wastewaterVolumenhydrocarboionConcentratS.FsteamVolume−

⋅= {7.6b}

117

Part III Summary of researches : Oily wastewater treatment

Part III Summary of researches: Oily wastewater treatment

III-i

Content

Page

Part III Summary of researches: Oily wastewater treatment Chapter 1 Oily or hydrocarbon-polluted wastewater

1.1 Introduction III-2 1.2 Hydrocarbons and oils III-2

1.2.1 Hydrocarbons III-2 1.2.2 Fats and oils III-7 1.2.3 Petroleum and petroleum products III-7 1.2.4 Oils in term of oily wastewater III-9

1.3 Other compositions of oily wastewater III-9 1.3.1 Surfactants III-9 1.3.2 Soaps III-10 1.3.3 Co-surfactants III-11 1.3.4 Suspended solids III-11 1.3.5 Other components III-11

1.4 Categories of oily wastewater III-11 1.4.1 Classification by the nature of the continuous phase III-11 1.4.2 Classification by the stability of oily wastewater III-11 1.4.3 Classification by the degree of dispersion III-12

1.5 Characteristics of certain oily wastewaters III-15 1.6 Standards, Laws, and Regulations III-16

Chapter 2 Overview for oily wastewater treatment process design 2.1 Decantation velocity and STOKE’s law III-21 2.2 Application of surface chemistry for oily wastewater treatment III-22

2.2.1 Liquid-gas and liquid-liquid interfaces III-22 2.2.2 Liquid-solid and liquid-liquid-solid interfaces III-25 2.2.3 Capillary pressure and LAPLACE’s law III-29

2.3 Important parameters in oily wastewater treatment and III-30 their method of analysis 2.3.1 Oil concentration III-30 2.3.2 Size distribution , spectrum or granulometry III-32 2.3.3 Other parameters III-38

2.4 Overview of oily wastewater treatment processes III-38 2.4.1 Decanter III-39 2.4.2 Coalescer III-39 2.4.3 Hydrocyclone III-39 2.4.4 Dissolved air flotation (DAF) III-40 2.4.5 Skimmer III-40 2.4.6 Membrane processes III-40 2.4.7 Thermal processes III-40 2.4.8 Chemical process III-41 2.4.9 Finishing processes III-41

Chapter 1 Oily or hydrocarbon-polluted wastewater

III-ii

Content (Con’t)

Page

2.5 Determination of degree of treatment III-41 2.5.1 Overall degree of treatment III-41 2.5.2 Degree of treatment of each process III-41

Chapter 3 Oil skimmer 3.1 General III-45 3.1 Oil drum skimmer III-47

3.1.1 Working principles III-47 3.1.2 Design calculation and design consideration III-52

3.2 Oil disc skimmer III-54 3.2.1 Working principles III-54 3.2.2 Design calculation and design consideration III-56

3.3 Productivity comparison between drum and disc skimmer III-56 3.4 Advantage and disadvantage of drum and disc skimmer III-57

Chapter 4 Decanting 4.1 General III-59 4.2 Simple Decanter or API tank III-60

4.2.1 Working principles III-60 4.2.2 Design calculation III-62 4.2.3 Design considerations III-65 4.2.4 Construction of simple decanters III-66

4.3 Compact decanter III-69 4.3.1 Working principles III-69 4.3.2 Design calculation III-73 4.3.3 Design considerations III-75 4.3.4 Variations, advantage and disadvantage of compact III-76

decanters

Chapter 5 Coalescer

5.1 General III-78 5.2 Granular bed coalescer III-78

5.2.1 Working principles III-78 5.2.2 Design calculation III-92 5.2.3 Design consideration III-94 5.2.4 Variations, advantage and disadvantage of granular III-95

bed coalescer 5.3 Guide coalescer III-96

5.3.1 Working principles III-96 5.3.2 Design calculation III-98 5.3.3 Design consideration III-98


III-iii

Content (Con’t)

Page

5.4 Fibrous Bed coalescer III-99 5.4.1 Working principles III-99 5.4.2 Design calculation III-109 5.4.3 Design consideration III-110 5.4.4 Variations, advantage and disadvantage of fibrous bed III-112

coalescer


6.1 General III-115 6.2 Working principles III-116

6.2.1 Filter based model III-116 6.2.2 Population balance model III-121 6.2.3 Generalized model of DAF from combination of III-123

filtration based model and population balance model 6.2.4 Influent parameters III-124

6.3 Design calculation III-127 6.4 Design consideration and construction of DAF reactor III-137 6.5 Pressurized water system or saturator III-145

6.5.1 Working principle and design calculation III-145 6.5.2 Type of saturator and injection valve III-150

6.6 Variations, advantage and disadvantage of DAF III-153

Chapter 7 Hydrocyclone 7.1 General III-155 7.2 Two-phase hydrocyclone III-156

7.2.1 Working principles III-156 7.2.2 Design calculation III-172 7.2.3 Design considerations III-175 7.2.4 Variations, advantage and disadvantage of III-178

hydrocy clone 7.3 Three-phase hydrocyclone III-178

7.3.1 Working principles III-178 7.3.2 Design calculation and design consideration III-182 7.3.3 Advantage and disadvantage of three-phase III-182

hydrocyclone

Chapter 8 Membrane process 8.1 General III-183

8.1.1 Classification of membrane processes III-183 8.1.2 Mode of operation of membrane processes III-185 8.1.3 Membrane structure III-185 8.1.4 Membrane material III-186 8.1.5 Membrane module type III-189


III-iv

Content (Con’t)

Page

8.2 Ultrafiltration (UF) III-194 8.2.1 Basic knowledge and working principles III-194 8.2.2 UF process design for oily wastewater treatment III-206 8.2.3 Design consideration and significant findings from III-221

GPI’s researches 8.3 Microfiltration (MF) III-232

8.3.1 Basic knowledge and working principles III-232 8.3.2 Significant findings on MF for oily wastewater III-233

treatment from GPI researches 8.4 Reverse osmosis (RO) III-235

8.4.1 Basic knowledge and working principles III-235 8.4.2 Significant findings on RO for oily wastewater III-236

treatment from GPI’s researches 8.5 Nanofiltration (NF) III-239

8.5.1 Basic knowledge and working principles III-239 8.5.2 Significant findings on NF for oily wastewater III-239

treatment from GPI’s researches 8.6 Comparison of membrane processes on emulsion treatment III-242

Chapter 9 Thermal processes 9.1 General III-245 9.2 Basic knowledge on distillation III-245

9.2.1 Basic knowledge on vapor/liquid equilibrium of III-245 mixtures

9.2.2 Equilibrium of various mixtures III-248 9.3 Heteroazeotropic distillation of oily wastewater III-250

9.3.1 Working principles III-250 9.3.2 Raoult’s law and Dalton’s law III-251 9.3.3 Calculation of azeotropic temperature and composition, III-252

dew curve and bubble curve. 9.3.4 Application of heteroazeotropic distillation on III-254

treatment of inverse emulsion or concentrated oily wastewater

9.3.5 Application of heteroazeotropic distillation on III-257 treatment of the wastes polluted by trace hydrocarbons: Steam stripping

9.3.6 Design calculation and design considerations III-257 9.4 Classical or conventional distillation of oily wastewater III-259

9.4.1 Working principles III-259 9.4.2 Significant findings on classical distillation for oily III-259

wastewater treatment from GPI’s researches


III-v

Content (Con’t)

Page

Chapter 10 Chemical treatment processes 10.1 General III-263 10.2 Basic knowledge III-264

10.2.1 Stability of the emulsion III-264 10.2.2 Surface-active agents III-264 10.2.3 Important properties to obtain stable emulsion III-266 10.2.4 Destabilization of emulsion III-268

10.3 Process design III-274 10.3.1 Rapid mixing III-275 10.3.2 Flocculator III-276

10.4 Design consideration III-278

Chapter 11 Finishing processes

11.1 General III-280 11.2 Biological treatment III-280

11.2.1 Basic knowledge III-280 11.2.2 Design consideration and significant finding on III-287

biological treatment for oily wastewater from GPI’s researches

11.3 Adsorption III-288 11.3.1 Activated carbon (AC) III-289 11.3.2 Basic knowledge III-290 11.3.3 Design calculation III-295

Chapter 12 Guideline for treatment process selection and examples of treatment processes for certain oily wastewaters 12.1 Guideline for treatment process selection III-297

12.1.1 Oil film III-297 12.1.2 Primary emulsion III-299 12.1.3 Secondary emulsion III-300 12.1.4 Macroemulsion and microemulsion III-301 12.1.5 Concentrated oily wastewater or refinery slops III-301

12.2 Examples of treatment processes for certain oily wastewaters III-302 12.2.1 Treatment of cutting oil emulsion III-302 12.2.2 Treatment of non-stabilized secondary emulsion III-303


III-vi

Table

Page

Table 1.2.1-1 Properties of some normal alkanes [1],[2] III-3 Table 1.2.1-2 Properties of some unsaturated aliphatic hydrocarbons [1],[2] III-5 Table 1.2.1-3 Properties of some benzene-series hydrocarbons [1],[2] III-6 Table 1.4.3-2 Summary of oily wastewater classification III-14 Table 1.5-1a Data of pollution from certain oils, oil products and serfactants. III-15 Table 1.6-1a Industrial effluent standard of Thailand III-17 Table 1.6-1b Industrial effluent standard of France (1998) III-18 Table 1.6-1c Industrial effluent standards of USA, categorized by pollutant sources III-19 Table 2.1-1 Summary of equations for rising velocity calculation [41] III-22 Table 2.2.1-1a Data of surface and interfacial tension (N/m) for some liquids at 20°C III-24 Table 2.2.2-1 Relation between contact angle, work adhesion, interfacial tension III-28

and spread coefficient of oil/water/solid system [42] Table 2.3.2-1a Example of the decanting test of oil/water emulsio III-35 Table 2.3.2-1b Example of the decanting test of oil/water emulsion: sorting of the III-36

result from table 2.3.2-1a Table 2.3.2-2 Criteria for visual observation of size distribution [22] III-37 Table 2.3.2-3 Criteria for visual observation of surface oil film [45] III-38 Table 3.2.1-1 Work adhesion and contact angles of oil, water and various materials III-48 Table 3.2.2-1 Critical surface tensions of certain materials [42] III-53 Table 5.4.1-1 Coalesced kerosene droplet size at various velocities and bed heights III-108

from oleophilic “bottle brush” coalescer (dF = 100 microns, D = 0.05 m,Co = 1 g/l, 120oC) [10]

Table 5.4.1-2 Comparison between the individual efficiency of oleophilic bottle III-108 brush coalescer, hydrocyclone, theoretical and observed efficiency of the coupling of coalescer/hydrocyclone [10]

Table 6.3-1 General design criteria of DAF from various literatures and III-130 manufactures (hydraulic loading rate us based on total flowrate)

Table 6.4-1 Data on efficiency and coagulant concentration of various oily III-141 wastewater treatments by DAF [51]

Table 6.5-1 Constant for calculation of concentration and quantity of gas for III-148 saturator design (Operating pressure of DAF = Patm, T = 20oC)

Table 6.5-2 Air characteristic and solubility at Patm [51] III-148 Table 8.1.4-1 Advantages and disadvantages of inorganic membrane [38] III-188 Table 8.1.4-2 Summary on membrane polymeric materials [38], [54], [57] III-189 Table 8.1.5-1 General characteristic of various membrane module [38], [54], [57] III-193 Table 8.2.2-1 General design criteria of UF process from various literatures III-208 Table 8.2.2-2 Summary of parameters of film model from UF researches on oily III-209

wastewater treatment (reference temperature = 20oC) Table 8.2.2-3 Summary of parameters of resistance model from UF researches on oily III-210

wastewater treatment (ref. temperature = 20oC)


III-vii

Table (Con’t)

Page

Table 8.2.2-4 Procedure to predict flux/pressure relation for case 1: Know k, III-211 β, Cg and R’m

Table 8.2.2-5 Procedure to predict flux/pressure relation for case 2: Know φ, α, III-213 Cg and R’m

Table 8.2.3-1 Composition of the special cleaning microemulsion concentrate [18] III-230 Table 8.4.2-1 Summary of RO data on oily wastewater treatment III-240 Table 8.5.2-1 Summary of NF data on oily wastewater treatment III-243 Table 8.6-1 Comparison of membrane processes on cutting macroemulsion III-244

treatments (based on Elf Seraft ABS at 4% by V of oil) Table 9.3.3-1 Heterotropic temperature and composition from various III-254

hydrocarbons [24] Table 9.3.4-1 Water extracting performance of various commercial hydrocarbons [24] III-257 Table 10.2.4-1 Results from ZHU’s research on destabilization of various emulsions III-272 Table 10.3.2-1 Recommended value of gradient and detention time III-277 Table 11-2.1-1 Rate coefficient for selected wastewaters [51] III-282 Table 11.2.1-2 Biodegradability and biotoxicity data [51] III-284 Table 11.2.1-3 Concentration of certain metals affecting biological systems [45] III-287 Table 11.2.2-1 Case studies on biological treatment of oily wastewater [66] III-287 Table 11.3.1-1 Examples of PAC and GAC properties III-289 Table 11.3.1-2 Adsorptive capacity of AC for some hydrocarbons [65] III-291 Table 11.3.3-1 Adsorption isotherm and MTZ data of some co-surfactants [21] III-295 Table 12.1-1 Guideline for oily wastewater process selection (based on III-298

GPI’s researches)


III-viii

Figure

Page

Fig. 1.2.1-1 Examples of normal alkanes and derivatives with equal number of III-4 carbon atoms

Fig. 1.2.1-2 Examples of benzene series III-6 Fig. 1.4.3-1 Relation between droplet sizes and rising velocity of primary and III-13

secondary emulsion [11] Fig. 1.4.3-2 Classification of oily wastewater by degree of dispersion III-14 Fig. 2.1-1 Free body diagram of oil drop in water and relation between III-21

Cd and Re [55] Fig. 2.2.1-1 Force diagram of oil drop in water III-23 Fig. 2.2.1-2 Model for visualization of γ [40][42] III-24 Fig. 2.2.1-3 Model for visualization of adhesion work III-25 Fig. 2.2.2-1 Diagram of Oil drop on solid surface in water III-25 Fig. 2.2.2-2 Model for visualization of cohesion work III-26 Fig. 2.2.2-3 Effect of surface roughness on contact angle III-27 Fig. 2.2.3-1 Section of air bubble in water III-29 Fig. 2.2.3-2 Diagram for visualizing the capillary pressure III-30 Fig. 2.3.2-1 Example of the size distribution of oil droplet in cutting oil III-32

macroemulsion (Elf Seraft 4% V of concentrate), measured by Coultronics nanosizer NDM4 [11]

Fig. 2.3.2-2 Granulometer (Source: above - Ankermid Techcross / below - CILAS) III-33 Fig. 2.3.2-3 Decanting test column III-33 Fig. 2.3.2-4 Relation between accumulated C/Co (% of C/Co when the droplet size III-36 is

equal or smaller than the given dE) and oil droplet size

Fig. 2.3.2-5 Example of estimated size distribution of oil droplets from III-37

decanting test

Fig. 2.5.2-1 Cut size determination III-43 Fig. 2.2.5-2 Economics of processes (least cost criteria) III-44 Fig. 3.1-1 Examples of oil skimming devices III-46 Fig. 3.2.1-1 Lab-scale drum skimmer: Major components are shown. III-47

(Source: GPI lab) Fig. 3.2.1-2 Surface energy or superficial tension of materials, oil and water III-48 Fig. 3.2.1-3 Influent parameters on drum skimmer performance III-52 Fig. 3.2.2-1 Occurrence of eddy currents from drum operation and no oil zone III-54

or non-productive zone that affects the productivity of the skimmer (Source: Oil Spill Cleanup)

Fig. 3.5-1 Application of the skimmers III-58


III-ix

Fig. 4.1-1 Examples of decanters III-59 Fig. 4.2.1-1 Schematic and typical removal efficiency curve of simple decanter III-61 Fig. 4.2.2-1 Relation between rising velocity, hydraulic loading rate of simple III-64

decanter and kerosene droplet size

Figure (Con’t)

Page

Fig. 4.2.4-1 Variations of simple decanters III-67 Fig. 4.2.4-2 Important components of simple decanter III-67 Fig. 4.3.1-1 Schematic of PPI decanter III-69 Fig. 4.3.1-2 “Spiraloil” decanter a) Simple spiral b) Mixed spiral III-70 Fig. 4.3.1-3a Comparison between observed efficiency (1',2',3') and predicted III-71

efficiency (1,2,3) for Simple Spiral "Spiraloil" decanter Fig. 4.3.1-3 Comparison between observed efficiency (1',2') and predicted III-71

efficiency (1,2) for Mixed Spiral "Spiraloil" decanter

Fig. 4.3.1-4 Effect of wettability of inserted plates III-72 Fig. 4.3.1-5 Schematic of inclined and horizontal decanter III-73 Fig. 4.3.4-1 Examples of compact decanters III-77 Fig. 5.1-1 Examples of coalescer III-79 Fig. 5.2.1.1 Schematic diagram of granular bed coalescer Mechanisms taking place III-79

inside the coalescer bed Fig. 5.2.1-2 Schematic diagram of the 3 transport phenomena III-80 Fig. 5.2.1-3 Relation between oil droplet diameter and efficiency factors of each III-82

transportation phenomena Fig. 5.2.1-4 Schematic of a single collector and the entire bed of coalescer III-83 Fig. 5.2.1-5 Phenomena in step 2: Adhesion-Coalescence III-85 Fig. 5.2.1-6 Phenomenon in step 3: “Salting out” or enlargemant of coalesced oil III-86 Fig. 5.2.1-7 Relation between experimental (or observed) efficiency factor and III-88

theoretical efficiency factor Fig. 5.2.1-8 Typical relation between efficiency of granular bed coalescer and III-90

various parameters Fig 5.2.2-1 Relation between cut size and coalescer dimension III-93 Fig. 5.3.1-1 Guide coalescer III-97 Fig. 5.4.1-1 Fibrous bed coalescer (Source: GPI lab) III-100 Fig. 5.4.1-2 Comparison between observed efficiency and predicted efficiency from III-102

SRIJAROONRAT's model, Verified by MA's and WANICHKUL's data. Fig. 5.4.1-3 Relation between observed efficiency and predicted efficiency from III-104

random fibrous bed coalescer model and simple fibrous bed model Fig. 5.4.1-4 Random or disorderly fibrous bed, used in the research of III-105

SRIJAROONRAT Fig 5.4.1-5 Typical relation between efficiency of fibrous bed coalescer and various III-106

parameters


III-x

Fig. 5.4.3-1 Examples of fibrous bed coalescer casing III-111 Fig. 5.4.3-2 Various types of bed tested by WANICHKUL [11]: From left, a) III-112

Simple spiral, b) Double spiral, c) Solid plastic spiral, d) A module of multi-stage bed [No. of stage can be added by increase the number of module into the same center shaft]

Figure (Con’t)

Page

Fig. 5.4.3-3 Characteristics of oil and water drops on silicon coated (oleophilic) III-112 steel fibers and non-coated (hydrophilic) steel fibers

Fig. 6.2-1 Example of schematic diagram of DAF (Source: Aquatec Maxcon) III-115 Fig. 6.2.1-1 Diagram for considering relative velocity of bubble and oil in flotation III-117

column Fig. 6.2.1-2 Schematic diagram of the 3 transport phenomena III-117 Fig. 6.2.1-3 Schematic of single bubble and entire height of flotation column III-118 Fig. 6.2.1-4 Relation between theoritical efficiency factor and observed efficiency III-120

factor Fig. 6.2.4-1 Typical relation between efficiency of DAF and various parameters III-124 Fig. 6.2.4-2 Solid-bubble agglomerate and formation of oil-bubble agglomerate [14] III-126 Fig. 6.3-1 Pilot-scale DAF test and Flota-test III-135 Fig. 6.4-1 Example of necessary equipment and component details of DAF system III-

140(Source: Environ Treatment System) Fig. 6.4-2 Necessary equipment and reactor components of DAF system III-141 Fig. 6.4-3 Examples of characteristics of scum from DAF processes III-145 Fig. 6.5-1 Example of good bubble formation from pressurized water III-146

(Source: Cornell DAF pump) Fig. 6.5-2 Relation between power required for pump, compressor and absolute III-149

pressure of saturator for pressurized water flowrate of 10 m3/h (assume %air saturation = 95%)

Fig. 6.5-3 Schematic diagrams of saturator systems (Source: Edur pump) III-151 Fig. 6.5-4 Examples of saturator system III-152 Fig. 6.5-5 Examples of injection valve III-153 Fig. 7.1-1 Basic flow pattern and examples of hydrocyclones III-155 Fig. 7.2.1-1 General flow pattern and features of hydrocyclones III-158 Fig. 7.2.1-1 General flow pattern and features of hydrocyclones III-159 Fig. 7.2.1-2 Velocity components in hydrocyclone III-159 Fig. 7.2.1-3 Tangential velocity profile in hydrocyclone and various typed of vortex III-160 Fig. 7.2.1-4 Examples of tangential velocity profile III-161 Fig. 7.2.1-5 Example of axial or vertical velocity profile III-162 Fig. 7.2.1-6 Example of radial velocity profile [30] III-163 Fig. 7.2.1-7 Forces on oil droplets or particles in hydrocyclone III-163 Fig. 7.2.1-8 Components of velocity of oil droplets or particles in hydrocyclone III-163


III-xi

Fig. 7.2.1-9 Trajectories of oil droplets and typical efficiency curvefrom trajectory III-166 analysis model

Fig. 7.2.1-10 Comparison between observed efficiency and predicted efficiency III-168 form MA’s model and THEW-COLMAN’s model

Fig. 7.2.1-1 Corresponding flowrate and velocities of various sizes of hydrocyclones III-173 at overflow pressure drop (Po) = 3 and 5 bars

Figure (Con’t)

Page

Fig. 7.2.3-1 An example of the coupling between solid-liquid and liquid III-176 hydrocyclone for the treatment of oily wastewater containing suspended solids (Source: Ultraspin)

Fig. 7.2.4-1 Relation between oil and water outlet velocity in co-current III-177 hydrocyclone

Fig. 7.3.1-1 Three-phase hydrocyclone III-178 Fig. 7.3.1-2a Oil drop (sphere) and particle (cube) trajectories in three-phase III-179

hydrocyclone Fig. 7.3.1-2b Typical vertical velocity profiles in three-phase hydrocyclone III-179 Fig. 7.3.1-3 Comparison between observed pressure drop and predicted valued from III-182

the 2 approaches (similarity and semi-empirical model) Fig. 8.1.1-1 Material sizes and corresponding membrane processes III-

184(Source: Osmonics) Fig. 8.1.1-2 Separation characteristic of membrane processes (Source: III-184

Koch membrane system) Fig. 8.1.2-1 Mode of operation of membrane process [11] III-185 Fig. 8.1.3-1 Membrane structures (Source: SCT, Millipore) III-186 Fig. 8.1.4-1 Membrane materials (Structures depend on manufacturing processes.) III-187

(Source: Millipore, Orelis, WWW.Scienceinafica.co.za) Fig. 8.1.5-1 Membrane modules III-191 Fig. 8.2.1-1 Relation between pore sizes and MWCO [59] III-194 Fig. 8.2.1-2 Typical characteristic curves of UF membrane III-195 Fig. 8.2.1-3 Oil drop at membrane pore III-196 Fig. 8.2.1-4 Typical schematic of cross-flow UF for wastwater treatment III-196 Fig. 8.2.1-5a Examples of pure water flux and solution flux characteristics III-197 Fig. 8.2.1-5b Typical characteristic curves of water and solution flux VS. III-197

transmembrane pressure when feed conc. is constant Fig. 8.2.1-6 Characteristic curves of permeate flux and feed concentration III-198 Fig. 8.2.1-7 Typical characteristics of permeate flux VS. time for constant feed conc III-198

system Fig. 8.2.1-8 An example of characteristic of flux VS. time for non-constant feed III-199

conc. system


III-xii

Fig. 8.2.1-9 Diagram of polarization layer and effect of velocity on flux in film III-201 theory

Fig. 8.2.1-10 Various resistances in UF processes III-203 Fig. 8.2.2-1 Examples of flux VS. oil concentration in UF of macroemulsion III-211

Fig. 8.2.2-2 Case 1: find flux/pressure relation when k, β, Cg and R’m are known III-211 Fig. 8.2.2-3 Relation between UF permeate flux and Transmembrane pressure III-212 at

Cref = 4% by volume of oil , V = 1.4 m/s and Predicted relations at C = 2 and 8% (Oil: Elf SeraftA

cutting oil macroemulsion, Membrane: IRIS 3042 PAN ) Figure (Con’t)

Page

Fig. 8.2.2-4 Relation between Flux VS. theoretical and observed permeate volume III-216 Fig 8.2.2-5 Relation between time VS. theoretical and observed permeate volume III-217

Fig. 8.2.2-6 Calculation of required storage volume of equalization tank by III-218 graphical method

Fig. 8.2.2-7 Relation between oil concentration VS. viscosity of emulsion III-220 Fig. 8.2.3-1a Flux of non-stabilized emulsion III-222 Fig. 8.2.3-1b Photographs of non-stabilized emulsion influent, retentate and permeate III-

222(from left) [10] Fig. 8.2.3-2 Typical relation between CaCl2 concentration and flux at low Pt III-223 Fig. 8.2.3-3a Partially destabilization by salt and coalesce of oil droplets III-224 Fig. 8.2.3-3b Magnified images (x100) of oil droplets from original (left) and III-224

partially destabilized macroemulsion (right) [11] Fig. 8.2.3-3c Magnified images of new membrane surface (left) and the surface after III-224

UF of macroemulsion at 3 bars without salt (middle) and with salt addition (600 mg/l CaCl2) [11]

Fig. 8.2.3-4 Relation between fluxes VS. pressure and calculated VS. observed oil III-225 concentration in retentate for partially destabilized emulsion

Fig. 8.2.3-5 Examples of feed emulsions and their corresponding UF permeates III-226 Fig. 8.2.3-6 Examples of flux enhance techniques [38] III-228 Fig. 8.2.3-7 Example of evolution of flux of macroemulsion UF with periodical III-229

cleaning with macroemulsion (membrane IRIS 3042, P = 1 bar, V=1.5 m/s. 25oC) and schematic of interaction between membrane/ surfactants [18]

Fig. 8.2.3-8 Examples of UF test modules III-231 Fig. 8.3.1-1 Examples of MF membranes III-232

Fig. 8.3.2-1 Evolution of flux from MF (with salt addition) of macroemulsion [20] III-234 Fig. 8.4-1 Examples of RO membranes III-235 Fig. 8.4.1-1 Working principles of reverse osmosis III-235 Fig. 8.4.2-1a Typical relation between flux and pressure of RO III-237 Fig. 8.4.2-1b Typical relation between flux and log of concentration III-237


III-xiii

Fig. 8.4.2-2 Examples of flux evolution from RO of the UF permeates of various III-237 emulsions(the RO permeate are not recycled) [11]

Fig. 8.4.2-3a Relation between rejection, TOD of permeate and concentration factor III-238 Fig. 8.4.2-3b Relation between rejection and transmembrane pressure III-238 Fig. 8.4.2-2 Examples of permeate TOD evolution from RO of the UF permeates III-238

of various emulsions (the RO permeate are not recycled) [11] Fig. 8.5.2-1 Relation of flux and transmembrane pressure of NF for macroemulsion III-

241(Elf Seraft ABS) and microemulsion treatment (Elf Emulself G3 EAB), using desal5 membrane at 20oC [20]

Figure (Con’t)

Page

Fig. 8.5.2-2 Relation of flux and theoretical feed concentration of NF for macro- III-242 and microemulsion treatment [20]

Fig. 9.2.1-1 Diagram of a binary mixture system III-246 Fig. 9.2.1-2 Evolution of temperature of a binary mixture at a constant P III-246 Fig. 9.2.1-3 Examples of P-T diagram III-246 Fig 9.2.1-4 Examples of T-x-y and P-x-y diagrams III-247 Fig. 9.2.1-5 An example of P-T-x-y diagram III-247 Fig. 9.2.1-6 An example of T-x-y diagram III-247 Fig. 9.2.2-2 Examples of evolution of VLE and VLLE with pressure III-248 Fig. 9.2.2-1 Examples of various types of vapor/liquid equilibrium III-249 Fig. 9.3.1-1 Typical isobar diagram oily wastewater III-250 Fig. 9.3.3-1 Graphical method to find heteroazeotropic temperature III-252 Fig. 9.3.3-2 Graphical method to find dew curve III-253 Fig. 9.3.4-1 Examples of lab-scale apparatus for heteroazeotropic III-255

distillation [11], [24] Fig. 9.3.4-2 Example of evolution of temperature with time from the treatment of UF III-255

permeate of cutting oil emulsion, using decane as entrainer [11] Fig. 9.3.4-3 Pictures of feed, residue and retentate of slop and UF retentate of used III-256

macroemulsion (30% by volume of oil) [24], [11] Fig. 9.4.2-1 Evolution of temperature, volume of distillate, TOD of mixed distillate III-260

and TOD of water part of the distillate (Based on Elf Seraft ABS, 4% by volume of concentrate) [11]

Fig. 9.4.2-2 The feed, residue and distillate from distillation of the III-260 macroemulsion [11]

Fig. 9.4.2-3 Evolution of temperature, volume of distillate , TOD of mixed distillate III-261 and TOD of water part of the distillate (Based on Elf G3 EAB, 4% by volume of concentrate) [11]

Fig. 9.4.2-4 The feed, residue and distillate from distillation of the III-261 microemulsion [11]

Fig. 9.4.2-5 Evolution of temperature, volume of distillate, and TOD of distillate III-262(Based on UF of used macroemulsion, TOD of feed =4400 mg/l) [11]


III-xiv

Fig. 10.2.2-1 Symbols of surface active agent and its localization at oil/water III-264 interfaces

Fig. 0.2.2-2 Diagram of the electrical double layer III-266 Fig. 10.2.3-1 Interfacial of oil and water with the presence of surfactants III-266 Fig. 10.2.3-2 Coalescence and redistribution of droplets in thermodynamiclly III-266

stabilized emulsion Fig. 10.2.3-3 Diagram of the electrical double layer III-267 Fig. 10.2.4-1 Force diagram of oil droplets and relation of repulsive, attractive and III-268

resulting force with the distance between oil droplets Figure (Con’t)

Page

Fig. 10.3-1 Schematic diagram of chemical process for emulsion destabilization III-274 Fig. 10.3.1-1 Static mixer (leftmost) and impeller types (from left): flat turbine, III-275

pitched blade turbine, propeller type (Source: Memko, Sharp mixers) Fig. 10.3.1-2 Relation between Np and Re [63] III-276 Fig. 10.3.2.1 Paddle type mixers (Source: Norfolk WTP, Aqua Pak) III-277 Fig. 10.4-1 Jar test equipment (Source: ECE engineering) III-278 Fig. 10.4-2 Example of stabilized emulsion before (right) and after chemical III-279

treatment (left) Fig. 11.2.1-1 General schematic diagram of activated sludge III-281 Fig. 11.3-1 Schematic diagram of adsorption III-289 Fig. 11.3.2-1 Examples of adsorption isotherm diagrams III-291 Fig. 11.3.2-2 Evolution of pollutant concentration along the bed depth III-292 Fig. 11.3.2-3 C/V curve and C/H curve III-292 Fig. 11.3.3-1 Examples of AC and GAC filter III-296 Fig. 12.2.1-1 Schematic diagram of cutting oil emulsion treatment system III-303 Fig. 12.2.2-1 Schematic diagram of conventional oily wastewater treatment system III-304 Fig. 12.2.2-2 Schematic diagram of compact oily wastewater treatment system for III-304

Non-stabilized secondary emulsion


III-1

Part 3 Summary of researches: Oily or hydrocarbon-polluted wastewater treatment

In this part, we will use the data reviewed and verified in Part 1 and Part 2 to summarize and compose the textbook that covers every research of Prof. AURELLE. The text will consists of;

• Types and characteristics of oily wastewater

• Related theories on oily wastewater treatment

• Oily wastewater treatment processes, based mainly on researches of Prof. AURELLE with additional data from well-proven literatures

• Special treatment process trains for some specific oily wastewater, based on Professor AURELLE’ s researches

118


III-2


1.1 Introduction

Water pollution is one of the most important environmental problems. Wastewater from agriculture and industrial processes, as well as domestic wastewater, is main pollutant source that causes water pollution problem. There are many substances that can deteriorate water quality and, then, can be counted as water pollutants, such as, organic matter from domestic wastewater, chemicals from industrial wastewater. Some valuable substances, such as sugar, flour, oil, will become major pollutants when discharged into water bodies.

Among various kinds of pollutants, hydrocarbon or oil is one of the most severe pollutants because of its intrinsic properties. Examples of adverse effects of hydrocarbons or oil when it is discharged into water bodies or environment can be summarized as follow;

• Even the small amount of oil can cause unpleasant odor and taste, so the water can not be used in potable water production system. Some hydrocarbons, such as benzene series, are noted for their carcinogen property.

• Presence of oil or hydrocarbons in visible form on the water surface is objectionable from aesthetic and recreation point of view.

• For ecosystem, floating oil layer is dangerous for it can directly harm aquatic animals such as fishes and waterfowls. It may coat and destroy algae thereby destroying food sources of aquatic animals.

• Small amount of hydrocarbon can spread over wide area of water surface. API reports that only 40 liters of oil can cover 1 km2 of water surface in form of visible film. It can affect photosynthesis and oxygen transfer, so causes adverse effect to marine or water ecology.

• The hydrocarbon contributes to very high biochemical oxygen demand (BOD) and is relatively difficult for biodegradation, which is main natural self-purification process. So it can last relatively long in the water and cause long term effect.

Therefor, it is obligatory to separate or remove oil from wastewater before disposal. In European countries, standard for hydrocarbons in effluent is normally 5 mg/l (see section 1.6).

The first step to proper oily wastewater treatment design is to have basic knowledge on types and characteristics of oily wastewater, which are described in the following sections

1.2 Hydrocarbons and oils

This section will provide background knowledge on definitions and basic chemistry of various from of hydrocarbons and oils, normally present in oily wastewater.

1.2.1 Hydrocarbons

Chemically, the hydrocarbons are compounds of carbon and hydrogen. They are also in the family of organic compounds, thus can be divided into 3 types, according to formation of characteristic groups i.e., aliphatic hydrocarbons, aromatic hydrocarbons and heterocyclic hydrocarbons.

119


III-3

1.2.1.1 Aliphatic Hydrocarbons

The aliphatic hydrocarbons are the hydrocarbon compounds those in which the carbon atoms are linked to a straight line. There can be subdivided into 2 groups i.e., saturated and unsaturated aliphatic hydrocarbons.

1. Saturated aliphatic hydrocarbons

The saturated aliphatic hydrocarbons are the compounds in form of CnH2n+2.The compounds are also called Methane series, Alkanes or Paraffins. Principle source of these compounds is petroleum.

Properties: These saturated hydrocarbons are normally colorless, relatively odorless, and relatively insoluble, particularly those with five or more carbon atoms. They are present in form of gas, liquid and solid. Alkanes with 1 to 5 carbon atoms are gases at ambient, those from C6 to C17 are liquids, and those above C17 are solids. Saturated aliphatic hydrocarbons are quite inert toward most chemical reagents. They are readily soluble in many solvents. However, their water solubility is relatively low. Solubility decreases with increasing size or molecular weight, while boiling and melting point increase with increasing molecular weight. Like common chemicals, properties of these hydrocarbons are subject to change with pressure and temperature.

Nomenclature: The hydrocarbons that form to a straight line are the basic form, called normal alkanes or n-alkanes, such as n-butane. The names and properties of some normal alkanes are shown in table 1.2.1-1. There are also another formations of hydrocarbons with the same number of carbon atoms as shown in fig. 1.2.1-1. These hydrocarbons are named differently from n-alkanes. To name such hydrocarbons, the IUPAC system is commonly used, somehow, will not be presented here.

Table. 1.2.1-1 Properties of some normal alkanes [1],[2]

Name Formula Melting

point(°C/ 1 atm)

Boiling point (°C/1 atm)

Specific gravity (20/4°)

Solubility (at 25

°C)(mg/l)

Methane CH4 -182.4 -161.5 0.423 -162C Ethane C2H6 -182.8 -88.6 0.545 -89C Propane C3H8 -187.6 -42.1 0.493 25C Butane C4H10 -138.2 -0.5 0.573 25C n-Pentane C5H12 -129.7 36 0.626 40 n-Hexane C6H14 -95.3 68.7 0.655 10 Cyclohexane C6H14 55 n-Heptane C7H16 -90.6 98.5 0.684 3 n-Octane C8H18 -56.8 125.6 0.699 25C 0.66 Isooctane C8H18 2 Nonane C9H20 -53.5 150.8 0.718 Decane C10H22 -29.7 174.1 0.730

120


III-4

C C C C

H

H

H

H

H

H

H

H

H H

C C C H

H

H

CHC

H

H

H

H

C

a) n-Butane b) Isobutane

Fig. 1.2.1-1 Examples of normal alkanes and derivatives with equal number of carbon atoms

Basic chemical reactions: The hydrocarbons in this category are quite inert. In ambient, they do not react with strong bases, acids or aqueous solutions of oxidizing agents. However, at elevated temperature, strong oxidizing agents, such as concentrated sulfuric acid, oxidize the compounds to CO2 and water. Other important reactions of the compounds are as follow;

• Oxidation with oxygen or air and heat: Ex: CH4 + 2O2 CO2 + 2H2O

• Substitution of hydrogen by halogens (scarcely occurs): Ex: CH4 +Cl2 HCl + CH3Cl

• Pyrolysis or cracking: High-mol.-wt. Alkanes + Heat + Pressure Low-mol.-wt. Alkanes + alkenes +

hydrogen + naphthenes +carbon

• Biological oxidation: 1st step: Alcohol forming (relatively slow) Ex: 2CH3CH2CH3 + O2 +bact. 2CH3CH2CH2OH 2nd step: Ex: 2CH3CH2CH2OH + 5O2 3CO2 + 4H2O

2. Unsaturated aliphatic hydrocarbons

The unsaturated aliphatic hydrocarbons are can be divided into 4 groups, catagorized by the number of double or triple bond in between carbon atoms, i.e., Alkenes, Diolefins, Alkadienes, and Alkynes. Examples of these compounds and their properties are shown in table 1.2.1-2. Solubility of these hydrocarbons is very low, normally less < 1.0 g/100g water.

Alkenes: The hydrocarbons in this category contain one double bond between two adjacent carbon atoms. Their general formula is CnH2n, such as Ethylene (C2H4), Butene (C4H8). They are also called “Olefins”. The compounds are produced in large number during the cracking of petroleum.

Diolefins: The hydrocarbons in this category contain two double bonds, such as 1,3-butadiene (CH2=CHCH=CH2), which has been used to make polymers.

121


III-5

Table. 1.2.1-2 Properties of some unsaturated aliphatic hydrocarbons [1],[2]

Name Formula Melting point (°C/ 1 atm)

Boiling point (°C/1 atm)


Solubility (at 25

°C)(mg/l)

Ethene CH2=CH2 -169 -103.7 0.568-104C

Propene CH2=CHCH3 -185.2 -47.6 0.50525C

1-Butene CH2=CHCH2CH3 -185.3 -6.2 0.58825C

1-Pentene CH2=CH(CH2)2CH3 -165.2 29.9 0.640

1-Hexene CH2=CH(CH2)3CH3 -139.7 63.4 0.673 50

1-Heptene CH2=CH(CH2)4CH3 -119.7 93.6 0.697

1-Octene CH2=CH(CH2)5CH3 -101.7 121.2 0.715 2.7

1-Nonene CH2=CH(CH2)6CH3 -81.3 149.9 0.72525C

1-Decene CH2=CH(CH2)7CH3 -66.3 170.5 0.741

Alkadienes: The hydrocarbons in this category contain more than two doubles, such as the red coloring matter of tomatoes, lycopene. Example of presence of these compounds in industrial wastes is wastewater from vegetable canneries. They require extremely high demand of oxidant.

Alkynes: The hydrocarbons in this category contain one triple bond, for example, H-C≡C-H. They are found to some extent in industrial wastes, such as synthetic rubber industries. Solubility

Basic chemical reactions: These compounds have many properties in common, regardless of the type of compound in which they exist. Their important reactions are as follow;

• Oxidation: Oxidizing agents, such as potassium permanganate, easily oxidizes the compounds, in an aqueous solution, to glycol.

• Reduction: Hydrogen may be added to double or triple bonds under proper temperature and pressure.

• Addition: Halohen acids, hypochlorous acids and halogens can be added at double or triples bonds.

• Polymerization: The compounds prone to combine with each other, by their unsaturated link, to form polymers.

• Biological oxidation: These compounds are easier to be biodegraded by bacteria than saturated compounds.

3. Cyclic aliphatic compounds (Naphthenes)

These compounds are the special cases of aliphatic hydrocarbons. They are also called “Naphthenes” and can be count as saturated hydrocarbons. The compounds can be easily described as a ring of CH2 molecules. They are found in petroleum product. Examples

122


III-6

of these compounds are cyclopropane, cyclopentane, cyclohexane, etc. At the same number of carbon atoms, naphthenes have higher density and lower melting point than paraffins.

1.2.1.2 Aromatic hydrocarbons

The aromatic hydrocarbons are the hydrocarbon compounds that contain ring structure. The difference between these compounds and the cyclic aliphatic compounds is that a carbon atom has one single bond and one double bond to the adjacent carbon atoms (see fig. 1.2.1-2). For the cyclic aliphatic compounds, they do not have double bond. There are 2 major series of the aromatic hydrocarbons, i.e., Benzene and Polyring series.

1. Benzene series

Benzene (C6H6) ring is the simplest aromatic ring, containing 6 carbon atoms. Benzene ring is knows as the parent compound of the aromatic series. The formula of benzene is shown in fig. 1.2.1-2. The compounds in benzene series are made up of alkyl (CnH2n+1) substitution product of benzene. Examples of the compounds in benzene series are as shown in table 1.2.1-3 and fig .1.2.1-2.

These compounds are widely used as solvents and in chemical synthesis. They are commonly found in petroleum products, such as gasoline. The compounds are hazardous for human health. They are known to be carcinogenic and cause leukemia. Eventhough they are relatively insoluble in water, there are limits of these compounds, such as xylene, toluene, set in drinking water standards.

Table. 1.2.1-3 Properties of some benzene-series hydrocarbons [1],[2]

Name Formula Melting point(°C)

Boiling point(°C/1

atm)


Solubility (at 20 °C)

Benzene C6H6 5.5 80.0 0.877 1,600 Toluene C6H5CH3 -94.9 110.6 0.867 500 o-Xylene C6H4(CH3)2 -25.2 144.5 0.88 170 m-Xylene -47.8 139.1 0.864 p-Xylene 13.2 138.3 0.861

Ethybenzene C6H5C2H5 -94.9 136.1 0.867 130

HC

HC

CH

HC

CH

CH

CH3

CH3

CH3

a) Benzene ring b) simplified formula of benzene

c) Toluene d) o-Xylene

Fig. 1.2.1-2 Examples of benzene series

123


III-7

2. Polyring series

The compounds contain more than 1 benzene ring. Some compounds are in from of solids at ambient, such as Naphthalene (mothball). Other examples of the compounds in this group are Anthracene and Phenanthrene (C14H10). They are known as carcinogen, thus also hazardous to human health. Furthermore, the larger compounds (5 or more rings) are very difficult to be biodegraded.

3. Chlorinated aromatic hydrocarbons

The compounds are derivatives of aromatic hydrocarbons that are of interest because of their industrial importance, as well as, their toxicity. Well-known example of these compounds is the polychlorinated biphenyls (PCBs), which are widely used as coolants in transformer, plasticizer, solvents and hydraulic fluid for they are very stable. The compounds are known to be very hazardous to human health.

1.2.1.3 Heterocyclic hydrocarbons

The heterocyclic hydrocarbons are the hydrocarbon compounds that have one other element in the ring in addition to carbon. An important hydrocarbon in this group is epoxides. Epoxides are three-membered rings where oxygen is bonded with two carbons. They are very reactive. Example of epoxides is ethylene oxide, which is in synthetic surfactant and pesticide.

1.2.2 Fats and oils

Chemically, fats and oils are esters (product formed by reaction between acid and alcohol) of the trihydroxy alcohol, glycerol. They are also described as glycerides of fatty acids. These glycerides that are liquid at room temperature are called oils and those that are solids are called fats. The oils, here, are mainly vegetable or animal oil and normally edible.

Basic chemical reactions: Their important reactions are as follow;

• Oxidation: For some oils, such as linseed oil, oxygen may be added to the double bonds and forms resin-like material. These drying oils are major component in all oil-based paints.

• Hydrolysis: since they are esters, oils can be easily hydrolyzed into glycerol plus fatty acids. If hydrolysis is accomplished by using of NaOH, it is called saponification. The product from saponification is soap.

• Addition: Oils containing unsaturated acids can add chlorine at the double bonds. This process is normally slow. However, it may represent a significant part of chlorine demand in some wastes. Oils that contains large amount of oleic and linoleic acids may be converted to fats by the process of hydrogenation. Thus, low-priced oils can be converted into magarine.

1.2.3 Petroleum and petroleum products

1.2.3.1 Petroleum or crude oils

Petroleum and petroleum products are one of the major sources of oil in oily wastewater. Petroleum is the mixture of various hydrocarbons, as well as other chemicals. If contains (by weight) about 80% of carbon, 10-15% of hydrogen, 1.55 of oxygen, 0.1-t% of

124


III-8

sulfur, 0.65 of nitrogen and also trace amount of iron, calcium, sodium, potassium, and other elements [39]. The density of crude oils is around 650 to 1,200 kg/m3.

Crude oils normally consist of paraffins, olefins, naphthenes and aromatic hydrocarbons. Oxygen, sulfur and nitrogen are present in the forms of compounds. Normally, paraffins of 5 to 16 atoms of carbon are present in the form of liquid. They are important components in gasoline and kerosene, as well as, desire components of diesel and lubricant oil. Paraffins of 17 carbon atoms or more are solids.

Olefins are found relatively rarely and in significant amount. Because of their highly reactive and easily polymerizing, they are relatively undesirable components in fuels and lubricant oils.

Naphthenes are normally found in forms of cyclo-pentane and cyclo-hexane. They are also important components of fuels and lubricant oils for their temperature resistance property.

Aromatic hydrocarbons of 1 to 4 rings are also found in crude oils. These compounds have the greatest density, compared to other hydrocarbons. They provide good viscosity-temperature property to petroleum product. They are also good solvents

Solubility of crude oils is low. The solubility of petroleum fractions decreases in the following order: aromatic compounds – naphthenes – paraffins. It also decreases with increasing in molecular weight.

1.2.3.2 Petroleum products

Petroleum products are the product derived from crude oils through processes such as catalytic cracking and fractional distillation. These products have physical and chemical characteristics that differ according to the type of crude oils and subsequent refining processes. Several examples of refined petroleum products and their physical properties are as follow:

Gasoline: a lightweight product that flows easily, spreads quickly, and may evaporate completely in a few hours under temperate conditions. It poses a risk of fire and explosion because of its high volatility and flammability, and is more toxic than crude oil. Gasoline is amenable to biodegradation, but the use of dispersants is not appropriate unless the vapors pose a significant human health or safety hazard.

Kerosene: a lightweight product that flows easily, spreads rapidly, and evaporates quickly. Kerosene is easily dispersed, but is also relatively persistent in the environment.

No. 2 Fuel Oil: a lightweight product that flows easily, spreads quickly, and is easily dispersed. This fuel oil is neither volatile nor likely to form emulsions, and is relatively non-persistent in the environment.

No. 4 Fuel Oil: a medium weight product that flows easily, and is easily dispersed if treated promptly. This fuel oil has a low volatility and moderate flash point, and is fairly persistent in the environment.

125


III-9

No. 5 Fuel Oil (Bunker B): a mediumweight to heavyweight product with a low volatility and moderate flash point. Preheating may be necessary in cold climates, and this fuel oil is difficult, if not impossible, to disperse.

No. 6 Fuel Oil (Bunker C): a heavyweight product that is difficult to pump and requires preheating for use. This fuel oil may be heavier than water, is not likely to dissolve, is difficult or impossible to disperse, and is likely to form tar balls, lumps, and emulsions. It has a low volatility and moderate flash point.

Lubricating Oil, a medium weight product that flows easily and is easily dispersed if treated promptly. This oil has a low volatility and moderate flash point, but is fairly persistent in the environment

Because of the difference in their properties, as stated above, it is helpful to know which type of oils is present in the oily wastewater considering.

1.2.4 Oils in term of oily wastewater

From section 1.2.1 to 1.2.3, the definitions and properties of hydrocarbons and other substances have been described. Those substances can be accounted as “oil” in oily wastewater. Even though they are some discrepancies in their chemical formula, their source of origin or some or their properties, they are some physical properties in common. Since the treatment processes studied in GPI lab are based on separation process, which depends mainly on these physical properties, it can be concluded that substances that have identical properties can be accounted as “oil” by the meaning of this book. These properties include:

• Liquidity: oils, in our case, must be liquids only. So some solids hydrocarbons and fats will not be accounted as “oil” in our case.

• Solubility: oils, in our case, must have very low solubility in water. So it can be treated as non-miscible binary mixture, which is main assumption or condition that governs operating principles of most separation processes. However, dissolved pollutants in oily wastewater are considered in some treatment processes, even though their major components come from other additives in oil, rather than the dissolved portion of oil its self.

• Specific gravity or density: oils, in our case, must have less density or specific gravity than water.

So, after this section, the word “oils” found herein this book will normally refer to the substances that conform to the above criteria.

1.3 Other compositions of oily wastewater

Apart from “oil” and water, oily wastewater can contain other substances that may affect its properties, as well as selection and design of its treatment processes. Important components of oily wastewater include:

1.3.1 Surfactants

In many applications, such as in cleansing processes, it is necessary to change the level of free surface energy. By the surface energy (or interfacial tension, in case of liquid/liquid

126


III-10

interfaces), it means the work required to form new surfaces. The substances that can decrease the level of surface energy are called “surface-active substances” or “surfactants”. They are also called by other names, such as, detergents (for solid/liquid interface, e.g. in washing or cleaning process), emulsifier, or dispersant, etc. With proper amount of surfactants, surface energy can be substantially decreased until almost to zero. This makes it possible to create a lot of new surface.

In our interest, surfactants can help making a lot of new surfaces of oil in the water. This means the oils can be divided into very small droplets, which contributes to many new surfaces, in the water, called emulsion. Theory of surface energy will be described in chapter 2. Effect and some further details of surfactants and co-surfactants will be described again in chapter 10 “Chemical treatment processes”.

All surfactants have rather large functional groups. One end of the molecule, which normally is organic group, is particularly soluble in oil. The other (polar group) is soluble in waters. There are 4 main types of surfactants i.e., cationic, anionic, non-ionic and ampholytic surfactants

1.3.1.1 Cationic surfactants

These surfactants are generally salts of quaternary ammonium hydroxide. The surface-active properties are contained in the cations. Besides their surface-active properties, the compounds are known for their disinfecting properties.

1.3.1.2 Anionic surfactants

These surfactants are generally sodium or potassium salts. The common ones are sulfates and sulfonates. The surface-active properties are contained in the anions.

1.3.1.3 Non-ionic surfactants

These surfactants do not ionize and have to depend upon groups in the molecule to render them soluble. All depend on polymers of ethylene oxide (C2H4O) to give them this property.

1.3.1.4 Ampholytic (or amphoteric) surfactants

For these surfactants, they contain both cationic and anionic functional groups.

1.3.2 Soaps

Ordinary soaps are derived from fats and oils by saponification with sodium hydroxide. Saponification is the special case of hydrolysis in which an alkaline agent is present to neutralize the fatty acids as they are formed. The fats and oils are split into glycerol and sodium soaps. The nature of the soap depends on the type of fat and oil used. Sodium and potassium soaps are normally soluble. If the water contains hardness, calcium, magnesium and other hardness-causing ions will precipitate soap to form metallic soap. Soap will precipitate all hardness ions first, after that it becomes a surfactant.

127


III-11

1.3.3 Co-surfactants

In emulsions, especially microemulsion, they contain co-surfactants. These compounds are normally alcohol, such as butyldiglycol, or benzylic alcohol. These compounds are readily soluble in water thus provide good solubility to certain products, such as cutting oil emulsion. They also increase the stability to the emulsion by promoting both electrical stability and mechanical stability.

1.3.4 Suspended solids

Suspended solids are commonly found in oily wastewater, especially when combined collection systems are used. Types and concentration of suspended solids, present in oily wastewater, depends on the sources of the wastewater. Some oil separation processes, esp. the ones that are based on filtration concept such as granular bed coalescer and membrames, are sensitive to the suspended solids. Suspended solids can cause adverse effect on oil separation processes, unless it has already accounted for and mitigation measure is provided.

1.3.5 Other components

Others components found in oily wastewater depend on their source. Normally they are the additives of the oil product, such as anti-foaming agents, anti-mousse agents, bactericides, dyes, or anti-corrosion agents, etc. Some can present adverse effect on treatment processes. Some may be present as residual pollutants in the effluent after oil separation processes. So it will be very helpful if we know in advance the components of the oily wastewater considering.

1.4 Categories of oily wastewater

Oil/water mixture systems can be categorized by many criteria. However, as stated before that the treatment processes summarized in this book are based mainly on separation processes, the criteria, used to categorize these oil/water mixtures, are based upon physical properties. There are 3 main criteria, i.e., the nature of continuous phase, the stability of oily wastewater, and the degree of dispersion.

1.4.1 Classification by the nature of the continuous phase

In binary mixture systems of oil and water, their components can be divided into 2 main phases, i.e., continuous phase, which is the majority of the two, and dispersed phase.

Direct emulsion: If continuous phase is the water, the mixture will be called “direct emulsion”, and can be written as “o/w emulsion”, which stands for emulsion of oil in water.

Inverse emulsion: On the other hand, if the continuous phase is oil, the mixture will be called “inverse emulsion” or “w/o emulsion”.

The treatment processes described herein this book will emphasize on oily wastewater, which is direct emulsion. However, some processes can also be applied to inverse emulsion.

1.4.2 Classification by the stability of oily wastewater

By these criteria, oily wastewater can be divided into 2 groups:

128


III-12

1. Non-stabilized emulsion

This type of wastewater mainly consists of 2 components, i.e., oil and water, without presence of surfactants. So its stability is not enhanced by surfactants and depends mainly on degree of dispersion (the size of oil droplets) in the wastewater. If the droplet size is small, the oil drop will take long time to rise to the surface, so it stays relatively long within the water phase. As long as there are oil dispersed in the water, the mixture is still called emulsion. The longer the time that droplets stay in the water, the greater the stability of the emulsion. Without surfactants, the degree of dispersion depends on energy added to the system to disperse the oil phase. This energy is normally supplied by mechanical means, such as agitation, pump, or hydraulic mixing, such as venturi.

2. Stabilized emulsion

This type of wastewater contains oil, water and surface-active agent, which may consists of only one surfactant or both surfactant and co-surfactant. Presence of these surface-active substances cause decrease in interfacial tension between oil and water, thus make the oil disperse into very fine droplets, down to micron size. Furthermore, localization or orientation of the surface-active agents causes electric barrier as well as mechanical barrier that prevent collision and coalescence between droplets. So the droplets remain small (See chapter 10). Their rising velocity is then very small and can be negligible, compared to Brownian motion. So this type of emulsion is stable, as it is called, and not prone to decant naturally.

1.4.3 Classification by the degree of dispersion

This classification is based on rising velocity of oil droplet, which is controlled by properties of the oil/water system and its size or its degree of dispersion. For common treatment processes, this velocity is governed by STOKES law. Theory of rising velocity will be described in chapter 2. By these criteria, oily wastewater can be divided into 5 groups:

1. Film or layer of oil on water surface

This type of wastewater is relatively easy to treat for the oil and the water are readily separated. However, extraction or removal of oil from the water surface should be carefully performed to prevent entraining of water with oil.

2. Primary emulsion

Oily wastewater will be categorized into this type if the oil droplets are greater than 100 microns.

3. Secondary emulsion

Oily wastewater will be categorized into this type if the oil droplets are smaller than 20 microns.

Classification of primary and secondary emulsion can be summarized as shown in fig. 1.4.3-1.

129


III-13

10-3

1 10 100 1000Droplet diameters (microns)

Ris

ing

velo

city

(m/s

)

10-4

10-5

10-6

10-7

10-8

10-9

10-10

0.60 0.70 0.750.80 0.98 0.90

Oildensity

Secondary emulsion

Primaryemulsion

Secondaryemulsion

Primaryemulsion

Fig. 1.4.3-1 Relation between droplet sizes and rising velocity of primary and secondary emulsion [11]

4. Macroemulsion

This type of emulsion contains droplet size between 0.06 to 1.0 microns. It usually contains surfactant (and co-surfactant). This type of emulsion has a milky appearance. A common example of this emulsion is cutting oil macroemulsion.

5. Microemulsion

This type of emulsion contains droplet size between 10 to 60 nm. It contains extensive amount of surfactants and cosurfactants. Since the droplets is very small, this type of emulsion is usually transparent (sometimes, translucent if dyes are present or amount of o-surfactants is low). A common example of this emulsion is cutting oil microemulsion.

Details of microemulsion amd macroemulsion will be described in details in the chapter 10, related to “emulsion destabilization”.

From the three criteria stated above, some classes of oily wastewater from different criteria might be overlapped. So, to summarize all of three criteria, oily wastewater can be categorized as shown in fig. 1.4.3-1 and table 1.4.3-2.

130


III-14

Solubility Forms of oil Continuous phase Stability Dispersion Type of emulsion

SolubleSoluble

NonsolubleNon

soluble FilmFilm

EmulsionEmulsion InverseInverse

DirectDirect Non stabilized

Non stabilized

StabilizedStabilized

Primary emulsionPrimary emulsion

Secondaryemulsion

Secondaryemulsion

Macro-emulsionMacro-

emulsion

Micro-emulsionMicro-

emulsion

OilOilNon

stabilizedNon

stabilized

Stabilized

< 1 μ

< 60 nm

> 100 μ

< 20 μ

Stabilized

Fig. 1.4.3-2 Classification of oily wastewater by degree of dispersion

Table. 1.4.3-2 Summary of oily wastewater classification

Degree of dispersion Continuous

phase Stability Primary emulsion

Secondary emulsion Macroemulsion Microemulsion

Non-stabilized emulsion

Exist Exist Exist Not exist Direct emulsion

(o/w) Stabilized emulsion

Exist Exist Exist Exist

1.5 Characteristics of certain oily wastewaters

From the previous sections, it can be concluded that the characteristics of oily wastewater can greatly very from wide varieties of oils and compositions of oily wastewater, as well as degree of dispersion. So, the best solution to characterize the oily wastewater is to have it analyzed by standard methods to get important data necessary for treatment process consideration. Important parameters and their method of analysis will be present in chapter 2. However, in this section, data of oily wastewater from many literatures, include GPI’s researches, will be presented as shown in table 1.5-1. These data can be used as guideline for treatment process consideration, when the real data is not available.

131


III-15

Ta

ble

1.5-

1a D

ata

of p

ollu

tion

from

cer

tain

oils

, oil

prod

ucts

and

serf

acta

nts.

132


III-16

1.6 Standards, Laws, and Regulations

To study or design wastewater treatment process, it is necessary to determine the quality of the effluent after the treatment. In case that the effluent will be recycled or reused, the treatment process will be designed to meet the requirement for reuse or recycle. However, in general case, the effluent will be discharged to public collection system or receiving water bodies. To do so, The effluent quality will have to meet the requirement of that receiving water bodies, set by local laws, standards, or related regulations. For oily wastewater treatment process, apart from oil concentration, other parameters that usually relate to oil or come with oil product, e.g. BOD and COD (from surfactants in cutting oil), phenol (from petroleum product), should be considered.

Table 1.6-1 shows effluent standards data from various sources. It must be noted that,

• Applicable effluent standards or laws for individual site usually depend on type of pollutant source and receiving water body as well as location of the source. The same type of industries locating at different towns may subject to different effluent standards.

• Summary of standards that should be considered for each site generally consists of,

• National standards (e.g. “Normes françaises pour les eaux useé” in France, “Notification of the Ministry of Science, Technology and Environment” in Thailand, or “Code of Federal Regulations on Environment” in the USA)

• Regional and specific industrial standard (e.g. standards of Toulouse, Bangkok metropolitan administration standard, standards for petroleum industries, etc.)

• Receiving water body standards (e.g. standards for coastal aquaculture, etc.)

• Generally, there are many conditions and exceptions in the laws. In some countries, such as the USA, the standards are meticulous categorized based on treatment technology, e.g. the best available technology (BAT) and the best practicable technology (BPT), which makes it more complicate to select the right applicable standard. Thus, conditions of the laws shall be studied carefully.

133


III-17

Table 1.6-1a Industrial effluent standard of Thailand

Parameters Standard Values Method for Examination1. pH value 5.5-9.0 pH Meter

2. Total Dissolved Solids (TDS)

not more than 3,000 mg/l depending on receiving water or type of industry under consideration of PCC but not exceed 5,000 mg/l not more than 5,000 mg/l exceed TDS of receiving water having salinity of more than 2,000 mg/l or TDS of sea if discharge to sea

Dry Evaporation 103-105 °C, 1 hour

3. Suspended solids (SS)

not more than 50 mg/l depending on receiving water or type of industry or wastewater treatment system under consideration of PCC but not exceed 150 mg/l

Glass Fiber Filter Disc

4. Temperature not more than 40°C Thermometer on-site 5. Color and Odor not objectionable Not specified 6. Sulphide as H2S not more than 1.0 mg/l Titrate

7. Cyanide as HCN not more than 0.2 mg/l Distillation and Pyridine Barbituric Acid Method

8. Fat, Oil & Grease (FOG)

not more than 5.0 mg/l depending of receiving water or type of industry under consideration of PCC but not exceed 15.0 mg/l

Solvent Extraction by Weight

9. Formaldehyde not more than 1.0 mg/l Spectrophotometry

10.Phenols not more than 1.0 mg/l Distillation and 4-Aminoantipyrine Method

11.Free Chlorine not more than 1.0 mg/l lodometric Method 12.Pesticides not detectable Gas-Chromatography 13.Biochemical

Oxygen Demand (BOD)

not more than 20 mg/l depending on receiving water or type of industry under consideration of PCC but not exceed 60 mg/l

-Azide Modification at 20 °C , 5 days

14.Total Kjedahl Nitrogen (TKN)

not more than 100 mg/l depending on receiving water or type of industry under consideration of PCC but not exceed 200 mg/l

Kjeldahl

15.Chemical Oxygen Demand (COD)

not more than 120 mg/l depending on receiving water of type of industry under consideration of PCC but not exceed 400 mg/l

Potassium Dichromate Digestion

16.Heavy metals 1. Zinc (Zn) not more than 5.0 mg/l

2. Chromium (6) not more than 0.25 mg/l 3. Chromium (3) not more than 0.75 mg/l 4. Copper (Cu) not more than 2.0 mg/l 5. Cadmium (Cd) not more than 0.03 mg/l 6. Barium (Ba) not more than 1.0 mg/l 7. Lead (Pb) not more than 0.2 mg/l 8. Nickel (Ni) not more than 1.0 mg/l 9. Manganese (Mn) not more than 5.0 mg/l

Atomic Absorption Spectro Photometry; Direct Aspiration or Plasma Emission Spectroscopy ; Inductively Coupled Plama : ICP

10. Arsenic (As) not more than 0.25 mg/l

11. Selenium (Se) not more than 0.02 mg/l

Atomic Absorption (AA)Spectrophotometry; Hydride Generation, or Plasma Emission Spectroscopy; Inductively Coupled Plasma : ICP

12. Mercury (Hg) not more than 0.005 mg/l AA Cold Vapour

134


III-18

Table 1.6-1b Industrial effluent standard of France (1998)

Parameters Standard Values Method for Examination1. pH value 5.5-9.5, 9.5 if there is alkaline neutralization pH Meter 2. Total Dissolved

Solids (TDS) Depend on receiving water or type of industry Dry Evaporation 103-105 °C, 1 hour

3. Total suspended solids (TSS)

100 mg/l for maximum daily rate not more than 15 kg/d, 35 mg/l for those that exceed, 150 mg/l for lagoon treatment system

Glass Fiber Filter Disc

4. Temperature Not more than 30°C Thermometer on-site 5. Color and Odor Not specified Not specified 6. Sulphide as H2S Not specified Not specified 7. Cyanide as HCN not more than 0.1 mg/l Not specified 8. Fat, Oil & Grease

(FOG) Depend of receiving water or type of industry Not specified

9. Formaldehyde Not specified Not specified 10.Phenols not more than 0.3 mg/l Not specified 11.Organic

halogenated compounds

not more than 1.0 mg/l if discharge > 30 g/d As AOX or EOX

12.Pesticides Not specified Not specified 13.Biochemical

Oxygen Demand (BOD)

Depend on receiving water or type of industry, Level A to D: 30-100 mg/l, Level E: not more than 30 mg/l, Level F: not more than 15 mg/l

At 20 °C , 5 days

14.Nitrogen (N)

Depend on receiving water or type of industry, TKN and ammonia N not more than 10-40 mg/l for level NK1 to NK3, TKN+AmmoniaN+Nitrite+Nitrate not more than 20 mg/l for level NGL1 and NGL2.

Kjeldahl

15.Chemical Oxygen Demand (COD)

Depend on receiving water or type of industry, Level A to D: 125-150 mg/l, Level E: not more than 50 mg/l, Level F: not more than 90 mg/l

Potassium permanganate Digestion

16.Heavy metals 1. Zinc (Zn) not more than 2.0 mg/l if discharge > 20 g/d

2. Chromium (6} not more than 0.1 mg/l if discharge > 1 g/d

3. Chromium (total) not more than 0.5 mg/l if discharge > 5 g/d 4. Copper (Cu) not more than 0.5 mg/l if discharge > 5 g/d 5. Cadmium (Cd) Not specified 6. Barium (Ba) Not specified 7. Lead (Pb) not more than 0.5 mg/l if discharge > 5 g/d 8. Nickel (Ni) not more than 0.5 mg/l if discharge > 5 g/d 9. Manganese (Mn) not more than 1.0 mg/l if discharge > 5 g/d 10. Tin not more than 2.0 mg/l if discharge > 20 g/d 11. Iron, aluminium not more than 5.0 mg/l if discharge > 20 g/d 12. Mercury (Hg) Not specified

Not specified

Note: 1 Extracted form Les normes de rejet des eaux usées et d'effluents (normes guides) (Source: Web site of Grenoble eau pure)

2. The standards may be amended and some special remarks for each type of receiving water and type of industry, which refers to other notification or decree, are not completely included.

135


III-19

Table 1.6-1c Industrial effluent standards of USA, categorized by pollutant sources

(1) CFR 40 part 419—Petroleum refining point source category

Kg / 1000 m3 of feed stock

Effluent characteristics Maximum for any 1 day Average of daily values for 30

consecutive days shall not exceed

BOD5 34.6 18.4 COD 210.0 109.0 TSS 23.4 14.8 Oil and grease 11.1 5.9 Phenolic compound 0.25 0.120 Ammonia as N 23.4 10.6 Sulfide 0.22 0.099 Total chromium 0.52 0.30 Chromium (hexavalent) 0.046 0.020 pH 6.0-9.0 6.0-9.0

Note: 1 The limits shown in the table are to be multiplied by the process factors and size factor to calculate the maximum for any one day and maximum average of daily values for thirty consecutive days

(2) CFR 40 part 437—The centralized waste treatment point source category

Effluent characteristics Maximum daily (mg/l) Maximum monthly average (mg/l)

Oil and grease 127.0 38.0 pH 6.0-9.0 6.0-9.0 TSS 74.1 30.6 Metals Arsenic 2.95 1.33 Cadmium 0.0172 0.0102 Chromium 0.746 0.323 Cobalt 56.4 18.8 Copper 0.500 0.242 Lead 0.350 0.160 Mercury 0.0172 0.00647 Tin 0.335 0.165 Zinc 8.26 4.50 Organic parameters Bis(2-ethylhexyl) phthalate 0.215 0.101 Butylbenzyl phthalate 0.188 0.0887 Carbazole 0.598 0.276 n-Decane 0.948 0.437 Fluoranthene 0.0537 0.0268 n-Octadecane 0.589 0.302

Note: 1 The limits shown in the table arebased on best practicable control technology currently available (BPT)

General note: The standards may be amended and some special remarks for each type of receiving water and type of industry, which refers to other notification or decree, are not completely included in this chapter.

136


III-20

Table 1.6-1c Industrial effluent standards of USA, categorized by pollutant sources (Cont.)

(3) CFR 40 part 417—Soap and detergent manufacturing point source category

Kg / 1000 kg of anhydrous product

Effluent characteristics Maximum for any 1 day Average of daily values for 30

consecutive days shall not exceed

BOD5 0.90 0.30

COD 4.05 1.35

TSS 0.09 0.03

Surfactants 0.90 0.30

Oil and grease 0.15 0.05

pH 6.0-9.0 6.0-9.0

(4) CFR 40 part 438—Metal products and machinery point source category

Effluent characteristics Maximum daily (mg/l)

TSS 62.0

Oil and grease (as HEM) 46.0

General note: 1 The standards may be amended and some special remarks for each type of receiving water and type of industry, which refers to other notification or decree, are not completely included in this chapter.

137


III-21

Chapter 2 Overview for oily wastewater treatment process design

This chapter provides basic theories, which are important to understand the phenomena, taking place in oily wastewater, and to understand operating principle of oily wastewater treatment processes. The analysis methods of some important parameters and overview of oily wastewater treatment processes are also described.

2.1 Decantation velocity and STOKES law

Oily wastewater can be considered as an immiscible binary mixture system of oil and water, with the oil as a dispersed phase, because the solubility of oil in water is very low (< 1g/ 100 g of water). So the oil will be present in the water in form of oil droplets, which is the result of surface tension, which will be described in the next section. Because of its low specific gravity, these droplets will try to separate themselves from the water by “floating” or “rising” to the water surface. Consider a single oil drop, when the oil drop is rising, it will be subject to 2 forces as shown in fig. 2.1-1. The first one (F1) is the gravitational force, calculated from oil drop size and density difference of oil and water (eq. 2.1.1). The second force is called “drag force”, which is the resultant of the resistance inertia force and viscous force (eq. 2.1.2).

F2

F1

Water

Oil drop,dia. = d

AV

gdF c )(6

13

ρρπ−=

{2.1.1}

2

212 AVCF cd ρ= {2.1.2}

Fig. 2.1-1 Free body diagram of oil drop in water and relation between Cd and Re [55]

Cd is a numerical constant, called drag coefficient, which depends on the flow region, represented by Reynolds number (Re) as shown in fig. 2.1-1. A is the projection area of oil drop. In this case, the oil drop is spherical, then A is the cross sectional area of the sphere.

The oil drop will start to rise from a stationary position (V = 0 m/s, F2 = 0). After some time, it will reach the terminal velocity or rising velocity, which will remain constant along its course. At this velocity, F1 = F2. So, from this condition, the rising velocity (V) can be written as follows.

2/1

)1(34

⎟⎟⎠

⎞⎜⎜⎝

⎛−= g

cdV

cd ρρ {2.1.3}

138

Chapter 2 Basic theory for oily wastewater treatment process design

III-22

By substituting the relation between Cd and Re at various ranges of Re into eq. 2.1.3 [41], The rising velocity can be calculated by the equations, summarized in table 2.1-1. In most of oil separation processes described hereafter, they normally operate in laminar flow regime. So rising velocity are governed by STOKES law (Table 2.1-1, eq. 2.1.4).

From STOKES law, one can increase the rising velocity by modifying related parameters in eq. 2.1.4. To increase the rising velocity is to increase the oil separation efficiency. Several treatment processes in this book are based upon this idea of modification. And because some parameters in STOKES law have been modified from natural condition, these processes will be called “accelerated separation process”. Examples of these processes include coalescer, flotation and hydrocycloning.

Table. 2.1-1 Summary of equations for rising velocity calculation [41]

Equation Flow Regime Re Cd V

STOKES law 2.1.4}

Laminar 10-4 - 1 Re24

c

c gdμρρ

18)( 2− or

21)1(545.0 dc

−− γρρ

1 - 10 77.0Re

26 439.1625.0814.0)1(57.0 dc

−− γρρ

10 – 100 65.0Re

20 222.1481.0741.0)1(73.0 dc

−− γρρ

ALLEN’s law 2.1.5a,b,c}

Intermediate

100 – 1000 316.0Re

92.4 813.0209.0604.0)1(81.1 dc

−− γρρ

NEWTON’s law

{2.1.6}

Tubulent (Eddy flow)

103 - 2*105 0.44 5.0

5.0)1(40.5 dc

−ρρ

Turbulent > 2*105 Newton’s law is applicable

2.2 Application of surface chemistry for oily wastewater treatment

2.2.1 Liquid-gas and liquid-liquid interfaces

In liquid, there are short-length attraction forces (Van de waal and polar forces) reacting between adjacent liquid molecules. The molecules located inside the bulk of liquid are subject to equal force in all direction, as shown in Fig 2.2.1-1. However, the molecules located at the surface of air/liquid are subject to unbalanced force, resulting in net inward pull (R) (see fig. 2.2.1-1). From this force, the molecules will try to move from the surface into bulk liquid. The surface of liquid, thus, tends to contract spontaneously. That is the reason why the oil or air present in the water tends to form a sphere shape, as oil drop or bubble (Laplace’s law).

This phenomenon is identical for the surface between 2 liquids. To avoid confusion, when talking about the surface between 2 substances, not the surface between one substance and air, this surface will be called “interface”

139


III-23

R

Fig. 2.2.1-1 Force diagram of oil drop in water

Surface tension (between liquid-air) or interfacial tension (between liquid-liquid or liquid-solid) [40] is defined as the work required to increase, isothermally and reversibly, the surface or interface area by 1 unit. Sometimes, it is also called “surface free energy” or “surface energy” (esp. in case of solid surface). It is measured in unit of work per unit area (ex. Nm/m2) or unit of force per unit length (ex. N/m or dyne/cm).

From this definition, interfacial tension between oil and water (γow) and between oil and air can be written in form of the following equations. Interfacial tension of liquid normally decreases with increasing temperature in a nearly linear fashion [40]. Data of surface tension and interfacial tension for some liquids are presented in table 2.2.1-1

owow A

W=γ {2.2.1a}

oo A

W=γ {2.2.1b}

W represents effective work required to increase the interfacial area between oil and air or water by Ao or Aow (or Ao/w) m2 respectively. For emulsion, Aow is the total area of every oil droplet. This effective work will be used to calculate mechanical work of the mixer to create the emulsion providing that the efficiency of the mixer is known. This concept will be used to create the emulsion of required degree of dispersion.

140


III-24

Tip To visualize the surface tension [42], we will consider a tank that one side wall can be moved as shown in fig. 2.2.1-2. If we try to move the wall to extend the surface of liquid, we have to apply the force F. Then, the wall move by the distance dl. From this, the work done or effective work can be written as:

dlFW ⋅= {2.2.2a}

When the wall is moved, we increase the surface area of the liquid, which can be written as shown in eq. 2.2.2b.The increasing in energy of the system is as shown in eq. 2.2.2c.

L

dl

F

F

Section

Plan

dlLA ⋅= {2.2.2b} AG ⋅=Δ γ {2.2.2c}

We assume the process is isothermal and reversible, so there is no heat loss. Then, work done will be equal to energy (eq. 2.2.2d). From this, we will see that the surface tension can be derived as shown in eq. 2.2.2e and eq. 2.2.1

dlLdlF ⋅⋅=⋅ γ {2.2.2d}

LFor

AW

dlLdlF

==⋅⋅

=γ {2.2.2e}

Fig. 2.2.1-2 Model for Visualization of γ

Table. 2.2.1-1a Data of surface and interfacial tension (N/m) for some liquids at 20°C [40][42]

Name Surface tension

Interfacial tension (/water)

Name Surface tension

Interfacial tension (/water)

Water 72.8 Ethanol 22.3

Benzene 28.9 35 n-Octanol 27.5 8.5

Acetic acid 27.6 n-Hexane 18.4 51.1

Acetone 23.7 n-Octane 21.8 5038

CCl4 26.8 45.1 Mercury 485 375

Polar organic compounds

22-50 Aqueous detergent solutions

24-40

Hydrocarbon 18-30 Fluorocarbon 8-15

141


III-25

Adhesion work: Interfacial tension between oil and water (or any two liquids) can relate to surface tension of each liquid by DUPRE’s equation (eq. 2.2.3a). This equation can be also applied to solid-liquid interface, as shown in eq. 2.2.3b. γs represents surface energy of solid.

)(owadhwoow W−+= γγγ {2.2.3a}

)(soadhosso W−+= γγγ {2.2.3b}

Tip To visualize the concept of adhesion work [42], we will consider a tank that one side wall can be moved, filled with 2 liquids (oil and water). When we move the wall, from eq. 2.2.2, it can be described that:

γo

γw

γo/w F

OilWater

• We increase the surface of oil and water by γo and γw, respectively.

• Then we attach the oil and water surface together. For this, we recover an amount of energy, called “adhesion work”

Fig. 2.2.1-3 Model for visualization of adhesion work

2.2.2 Liquid-solid and liquid-liquid-solid interfaces

2.2.2.1 Wetting, Contact angles, Adhesion work, and YOUNG’s equation

Wetting is the displacement from a surface of one fluid by another [40]. Therefore, it involves three phases and, at least two phases must be fluid. To understand this, we will consider the oil drop on a solid surface in water, as shown in fig. 2.2.2-1.

θso

γow

γsoγsw

Water

Solid

Oil droplet

Fig. 2.2.2-1 Diagram of Oil drop on solid surface in water

Contact angle: The angle between tangential line of oil drop’s surface and the surface of solid is called “contact angle”. In case that the oil remains as a drop, it means the system is in equilibrium. We will have definite contact angle θso. Equilibrium of force on the droplet can be written as shown in eq. 2.2.4, known as YOUNG’s equation.

142


III-26

For oil/water/solid system

soowsosw θγγγ cos+= {2.2.4a}

For oil/air/solid system

soosos θγγγ cos+= {2.2.4b}

If we combine DUPRE’s equation (eq. 2.2.3) with YOUNG’s equation, we will get an equation, called YOUNG-DUPRE equation, as shown in eq. 2.2.5. This equation allows us to calculate the adhesion work of the system.

For oil/water/solid system

)cos1()( soowwsoadhW θγ += {2.2.5a}

For oil/air/solid system

)cos1()( soosoadhW θγ += {2.2.5b}

To define the degree of wetting or “wettablity”, we use the simple criteria as follow:

• If the contact angle is equal or less than 90°, it can be said that the solid is wetted by oil. So it has “oleophilic” property.

• If the contact angle is greater than 90°, it can be said that the solid is not wetted by oil. So it has “hydrophilic” property. The shape of oil drop is almost spherical in this case.

2.2.2.2 Cohesion work and Spreading

Fig. 2.2.2-2 Model for visualization of cohesion work

Cohesion work: If we have a tank of oil and we want to separate the oil to create air-oil surface of 1 unit area within the oil bulk, we can apply eq. 2.2.1 to find the work required. In this case, we will have 2 surface of oil. Each surface has the area of 1 unit (see fig. 2.2.2-2).

From eq. 2.2.1, the required work will be calculated as shown in eq. 2.2.6a, which we call “Cohesion work”.

γ2)( =ocW {2.2.6a}

When applied to oil/water system, cohesion work, in order to separate oil in presence of water, can be written as follows:

owwocW γ2)( = {2.2.6b}

143


III-27

Spreading: From YOUNG-DUPRE’s equation, if contact angle is zero, it means that the oil surface is parallel to solid surface. In this case, we call that the oil spreads over the solid surface. In this case, adhesion work can be calculated as follows:

owowwsoadhW γγ 2))0cos(1()( =+= {2.2.7}

From eq. 2.2.7, it shows that adhesion work, in this case, is equal to cohesion work. From this fact, HARKINS [quoted by AURELLE [42]] sets the criteria to consider if the liquid will spread over the solid or not, by a constant called “spreading coefficient”(φ).

wocwsoadh WW )()( −=φ {2.2.8}

• If φ ≥ 0, or Wadh(so)w ≥ Wc(o)w, then the oil will spread over the solid.

• If φ < 0, or Wadh(so)w < Wc(o)w, then the oil will not spread over the solid.

Concept of spreading is very useful for oil/water separation process. When we bring the emulsion to contact with the solid, if the solid is olephilic or the oil prefer to spread over the solid, the oil that collides with the solid will attach to solid surface, thus, be separated from the emulsion. From this concept, we can define the relation between contact angle, adhesion work, interfacial tension, and spread coefficient as shown in table 2.2.2-1.

Critical surface tension (or critical surface energy): Zisman [quoted by AURELLE [42]] studies that if the oil drop is placed on the solid surface, the contact angle will decrease linearly with decreasing surface tension of oil. When the surface tension of the oil is low enough, the contact angle will be zero, then that oil will spread over the surface. That surface tension will be called “Critical surface tension” of the solid. The value can be interpreted that any oil will spread over the solid surface when surface tension of that oil is equal or lower than critical surface tension of the solid, as shown in eq. 2.2.9. This concept is very useful for material selection in several separation processes, such as skimmer (chapter 3).

If γo ≤ γc , the oil can spread over the solid surface. {2.2.9}

Effect of roughness of solid surface: Surface roughness of solid can effect its wettability. If the surface is not smooth, the contact angle will increase as described by WENZEL’s equation (eq. 2.2.10, fig 2.2.2-3). r represents the roughness coefficient and is equal to ratio between actual surface area to geometrical surface area . The value of r is equal to 1, when the surface is smooth. Otherwise, r will always be greater than 1. The effect of roughness should be considered when design the oil/water separation process.

)cos()'cos( soso r θθ ⋅= {2.2.10}

θθ

Smooth Rough

θ θ

Smooth Rough

Fig. 2.2.2-3 Effect of surface roughness on contact angle

144


III-28

Table 2.2.2-1 Relation between contact angle, work adhesion, interfacial tension and spread coefficient of oil/water/solid system [42]

Contact angle θso

Wadh(so)w φ Relation beween interfacial tension Characteristic of oil drop

= 0° > 2γow > 0 γsw > γso + γow

γso

γow

γsw

= 0° = 2γow = 0 γsw = γso + γow

γso

γow

γsw

0°<θso<90o γow <W< 2γow < 0 γsw > γso θγso

γow

γsw

= 90o W = γow < 0 γsw = γso θ

γso

γow

γsw

90°<θso<180o 0<W < γow < 0 γsw < γso θγso

γow

γsw

=180o W = 0 < 0 γso = γsw+γow

γso γsw

γow

145


III-29

2.2.3 Capillary pressure and LAPLACE’s law

2.2.3.1 LAPLACE’s law

LAPLACE’s law is the relation of pressure difference across the surface of liquid of surface tension γ. If we consider the air bubble in water, at one fraction of the bubble with the radius of curvature R1, and R2 on 2 perpendicular plains (fig. 2.2.3-1), the pressure difference, by LAPLACE’s law, will be as shown in eq. 2.2.11:

R1 R2Section of surface of air bubble in water

P (Outside pressure)

P’(Inside pressure)

Fig. 2.2.3-1 Section of air bubble in water

)11('21 RR

PPP +=−=Δ γ {2.2.11}

From the equation, we can calculate the pressure difference of the following systems, i.e., • A soap bubble of radius R in air

RP γ4=Δ

Because there are 2 interfaces of air/soap. • An air bubble of radius R in water

RP γ2=Δ

• A cylindrical jet of water in air γP =ΔR

Because R2 is infinity.

2.2.3.2 Capillary pressure

Capillary effect is an important phenomenon for oil/water separation process when we consider flowing through porous media, such as ultrafiltration membrane. This phenomenon will be described by fig. 2.2.3-2. When the tube with very small diameter (capillary tube) is plunged into liquid surface. Because of surface tension of liquid, the liquid will rise into the tube to the height h. From equilibrium, we can calculate the capillary pressure (Pc) as follow:

catmb PghPP ==− ρ {2.2.12}

From LAPLACE’s law,

rRPP atmb

θγγ cos22==− {2.2.13}

From eq. 2.2.12 and 2.2.13,

rPc

θγ cos2= {2.2.14a}

And

grh

ρθγ cos2

= {2.2.14b}

146


III-30

From eq. 2.2.14, it shows that the value of Pc or h will vary with the size of capillary tube. In case that the contact angle is greater than 90o, the surface of the liquid in the tube will be lower than that outside of the tube. So if we want to force the liquid through the tube, the applied force must be greater than capillary pressure plus the resistance of tube. This concept is very useful to understand membrane process.

a) b)

Fig. 2.2.3-2 Diagram for visualizing the capillary pressure

2.3 Important parameters in oily wastewater treatment and their method of analysis

From chapter 1, it shows that characteristics of oily wastewater can greatly vary, depending on their sources, degree of dispersion and compositions. However, to design the oily wastewater treatment process, it is, somehow, necessary to know its characteristic. In case that characteristic data is not available, we have to have it analyzed by an acceptable method. This chapter will suggest the important parameters necessary to consider or design the treatment processes, esp. those that are included in this book.

For the methods of analysis, if the result will be used to report to authorities, such as effluent quality report, method of analysis used has to be those that are approved by authorities or related standards. However, if the result is used for preliminary consideration, or standard method is not available, this section will provide simple technique that can be performed by basic or commonly available apparatus.

2.3.1 Oil concentration

Oil concentration is one of important parameters for oily wastewater treatment consideration or design. Effluent standard is set by oil concentration. Inlet oil concentration and effluent standard are the parameters that determine the degree of treatment required. Furthermore, some separation processes are limited by oil concentration. Normally, the oil concentration is reported in the form of mg/l. However, for inlet oil concentration, it may be measured in the form of % by volume.

147


III-31

Method of analysis

1. Solvent extraction [1]

This method is the standard method of examination of oil and grease. They are at least 4 variations. However, they all involve an initial extraction of oil and grease by hexane or CFC-113. The 4 variations include,

• Partition-gravimetric method: Hexane will be used to extract the oil. Then it will be separated from water and then evaporated. The residue will be used as an inditator of oil and grease content.

• Partition-infrared method: CFC-113 will be used as extractant. Then it will be measured with infrared scanning. This method is faster than gravimetric method. The accuracy depends on selection of oil standards for calibration. The oil used for calibration should be the same type as oil in the wastewater.

• Soxhlet extraction: It involves the initial step of acidification and then hexane extraction. This method tends to retain more volatile hydrocarbons. However, it is time consuming.

• Hydrocarbon analysis: Silica gel is added to hexane to remove fatty materials. And then the hexane will be analyzed by the first method.

2. Turbidity measurement

Oil concentration can be measured by light loss (turbidity) or light scatter from oil droplets in the wastewater [43]. The main advantage of this method is its simplicity and fast response. However, it will be difficult to distinguish between oil droplets and other particles. Some manufacturer claim that this effect can be minimized by the selection of the angle of scatter and detector. In GPI lab, some researchers use this method to measure the oil concentration. It is suitable to use for relatively clear wastewater, such as condensate. However, in most situations, oily wastewaters are usually contaminated by color materials and suspended solids. This method, thus, may not provide accurate result on oil concentration.

3. TOD measurement

If total oxygen demand (TOD) measuring instrument is available, oil concentration can also be measured as TOD. Again, the main advantage of this method is its fast response. However, it will be difficult to distinguish between TOD of oil and other substances, especially surfactants and co-surfactants. So this fact should be taken into account. In GPI lab, some researchers use this method to measure the oil concentration. The data from the researches are presented in chapter 1. It must be noted that TOD includes oxygen demand from non-organic factors, such as ammonia, sulfide. Thus its value is usually higher than BOD.

4. Other methods

Total organic carbon (TOC) is also widely used to measure pollutant concentration. Its working principle bases on oxidizing of the sample, usually at high temperature, to convert carbons to carbondioxide, which will be measured by IR, UV probe. TOC machine can cover from ppm to ppt range. It is very useful for online control. But it does not account for nitrogen compounds, which have much higher oxygen demand than carbons.

148


III-32

There are also others techniques to measure the oil concentration. However, some techniques are developed for specific application. Some are not widely used or not suitable for wastewater analysis. So they will not be described in detail here. Such techniques include direct infrared absorption, ultraviolet absorption, optical fiber sensor, gas chromatography, etc.

2.3.2 Size distribution , spectrum or granulometry

The size distribution, spectrum or granulometry is the data on the number of each size of oil drop, air bubbles, particles or other dispersed material in continuos phase. Normally, it is reported in the form of percent of total number of dispersed material or percent by volume or percent by weight. Example of the size distribution of oil droplet in cutting oil macroemulsion is shown in fig. 2.3.2-1.

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

10.0 17.8 31.6 56.2 100.0 178.0 316.0 562.0 1,000.0 1,780.0

Droplet diameter (nm)

Perc

ent b

y w

eigh

t (%

)

Fig. 2.3.2-1 Example of the size distribution of oil droplet in cutting oil macroemulsion (Elf Seraft 4% V of

concentrate), measured by Coultronics nanosizer NDM4 [11]

Oil drop or particle’s diameter is one of the parameters in the STOKES equation. Since treatment efficiency of several separation processes, which will be described in the following chapters, are based on the rising velocity that is governed by STOKES law. So the diameter is one of the key parameters to calculate the efficiency of the process. Furthermore, each process can provide acceptable or competitive efficiency in only a certain range of droplet size. So the size distribution is an important data to select feasible treatment process.

Method of analysis

1. Particle size analyzer or Granulometer

“Particle size analyzer” or “granulometer” or “nanosizer” is the specific equipment designed for measuring size and quantity of particle in liquid. There are many generations of granulometer with different operating principles. However, present granulometers are based on laser detection technology. This equipment can be used to measure size distribution of oil droplet or particle down to the range of submicron size (depend on manufacturers). It can report the result in the form of table or graph of size

149


III-33

distribution, both by number and by volume or weight. An example in fig. 2.3.2-1 was also measured by the granulometer. Fig. 2.3.2-2 shows the picture of granulometer.

Fig. 2.3.2-2 Granulometer (Source: above - Ankermid Techcross / below - CILAS)

2. Decanting test

When the use of granulometer is not available, there is a simple method to measure the size distribution, which is called “decanting test”. This method requires a simple equipment as shown in fig. 2.3.2-3.

0 min 5 min 10 min …. t min

h1h2

h3h4

Fig. 2.3.2-3 Decanting test column

150


III-34

Analytical procedure

1) Pour the wastewater into the decanting test column and stir. Make sure it is homogeneous.

2) Leave the wastewater to decant, taking the water samples from various heights of the column at preset interval.

3) Analyze the samples to find the oil concentration or other parameters that are directly proportional to the oil concentration, such as TOD, turbidity, etc.

4) From the collected data, we can calculate the granulometry, using this procedure:

• At t = 0, the wastewater is homogeneous, then the oil concentration is Co throughout the column.

• At t = t1 and height (from the bottom of the column) = h1, the corresponding rising velocity Vh1t1 = h1/t1.

• So at this point, the oil drop that have rising velocity greater than Vh1t1 will rise past the height h1 then will not be found in the sample taken from h1. From Vh1t1, we can calculate the corresponding droplet diameter dh1t1 by STOKES law.

• If the oil concentration at this point is Ch1t1, and the calculated droplet diameter is dh1t1, we can conclude that: The fraction of d < dh1t1 = Ch1t1/Co

5) Repeat step 4), at other t and h. then, use the fraction of d < dhx tx to plot the size distribution curve.

6) Concentration of each droplet size can be roughly represented by the differences between value of adjacent % of accumulated C/Co. Please note that the size and %C/Co are estimated values only. But it can provide an idea of how the sizes of droplet are distributed.

An example of the analysis of decanting test data is shown in tab. 2.3.2-1 and fig. 2.3.2-4 and 2.3.2-5.

151


III-35

Table 2.3.2-1a Example of the decanting test of oil/water emulsion

Time At sampling height (from the bottom

of column)

Corresponding rising velocity

Oil concentration of the sample

% accumulated C/Co

when d < dEht

Corresponding droplet size (from STOKES

law)

h Vh,t Ch,t % Ch,t/Co dEh,t

sec m m/s mg/l % micron

0 0.5 86 100

900 0.5 5.56E-04 57 66.28 72.8

1800 0.5 2.78E-04 25 29.07 51.4

2700 0.5 1.85E-04 8 9.30 42.0

3600 0.5 1.39E-04 3 3.49 36.4

5400 0.5 9.26E-05 1 1.16 29.7

7200 0.5 6.94E-05 0 0.00 25.7

0 1.25 86 100

900 1.25 1.39E-03 83 96.51 115.0

1800 1.25 6.94E-04 63 73.26 81.3

2700 1.25 4.63E-04 49 56.98 66.4

3600 1.25 3.47E-04 37 43.02 57.5

5400 1.25 2.31E-04 16 18.60 47.0

7200 1.25 1.74E-04 6 6.98 40.7

7200 1.25 2.06E-03 80 93.02 140.0

Note: Assume the oil is kerosene so the density (20o C) = 790 kg/m3. Water dynamic viscosity = 1.08E-3 (N.S)/m2.

152


III-36

Table. 2.3.2-1b Example of the decanting test of oil/water emulsion: sorting of the result from table 2.3.2-1a

Corresponding droplet size Corresponding rising velocity % accumulated C/Co

when d < dEht

Estimated C/Co for each droplet size

dE Vh,t % Ch,t/Co %C/Co

micron m/s % %

25.72 6.94E-05 0

29.70 9.26E-05 1.16 1.16

36.38 1.39E-04 3.49 2.33

40.67 1.74E-04 6.98 3.49

42.00 1.85E-04 9.30 2.33

46.96 2.31E-04 18.60 9.30

51.44 2.78E-04 29.07 10.47

57.52 3.47E-04 43.02 13.95

66.41 4.63E-04 56.98 13.95

72.75 5.56E-04 66.28 9.30

81.34 6.94E-04 73.26 6.98

115.03 1.39E-03 96.51 23.26

0

10

20

30

40

50

60

70

80

90

100

0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00

Oil droplet size (dE) (micron)

% A

ccum

ulet

ed C

/Co

Accumulated conc. curve h = 0.5 m h = 1.25 m

Fig. 2.3.2-4 Relation between accumulated C/Co (% of C/Co when the droplet size is equal or smaller than the

153


III-37

given dE) and oil droplet size

0

5

10

15

20

25

26 30 36 41 42 47 51 58 66 73 81 115

Droplet size (micron)

% b

y w

eigh

t

Fig. 2.3.2-5 Example of estimated size distribution of oil droplets from decanting test

3. Visual observation

To roughly estimate the granulometry of oily wastewater, there are simple criteria for visual observation as shown in table 2.3.2-2. However, these criteria may be used only when the wastewater is not polluted by suspended solids or other substances that makes it opaque. In case of surface oil layer, API [45] recommended useful guide to evaluate oil thickness as shown in table 2.3.2-3.

Table 2.3.2-2 Criteria for visual observation of size distribution [22]

Visual aspect Average particle size Example

Particles visible by eye 500 micron Very fine sand

Not clearly visible by eye 100 micron Starch

Opaque milky 10 micron Milk

Whitish milky 1 micron Homogenized milk

Bluish milky 0.1 micron Macroemulsion

Semitransparent 50 nm Microemulsion

Transparent 10 nm Microemulsion

Transparent 2-6 nm Micelles

Transparent 1 nm Molecules

154


III-38

Table 2.3.2-3 Criteria for visual observation of surface oil film [45]

Film thickness (micron) Appearance Approximate volume of oil

for film 1 km2 in area (l)

38.1 Barely visible under most favorable light conditions

37

76.2 Visible as silvery sheen on surface of water

74

152.4 First trace of color may be observed

148

304.8 Bright bands of color are visible 296

1016 (1 mm) Colors begin to turn dull 985

2032 (2 mm) Colors are much darker 1970

2.3.3 Other parameters

Each treatment process may require some special data for its calculation. However, common data required by several processes include:

• Interfacial tension • Viscosity of oil

These data can be analyzed by specialized laboratories. However, if we know the source of the oily wastewater, we can estimate these parameters. Even when we can not specify the exact source of oil, we still can estimate them, using the general data of oil or from references, without too much error. These data can also be found in chapter 1 and this chapter.

2.4 Overview of oily wastewater treatment processes

As hydrocarbon or oil requires a great amount of oxygen or oxidizing agent to oxidize, moreover, the hydrocarbon is relatively difficult to biodegrade, thus, it becomes clear that the use of biological treatment with high-concentration oily wastewater may not be the economical alternative. Besides there are possibilities to reuse or recover the hydrocarbons in the wastewater. Then, if possible, it is reasonable to separate oil from the wastewater, rather than destroy or change them into other forms.

Thus almost all of treatment processes that have been studied in GPI laboratory are based on separation process, both physical and physico-chemical, techniques in order to separate oil from water. From chapter 1, it shows that oily wastewaters can be divided into several categories, mainly depending on their stability and degree of dispersion. And in the previous sections on this chapter, it shows that physical separation of oil from water depends on STOKES law.

Thus, to enhance separation between water and oil, it can be archived by modification of parameters in STOKES law in the manner that make the rising velocity of oil droplets increase. Actually, almost all of oil/water separation processes covered by Prof. AURELLE’s research team are STOKES law-based processes, i.e. decanter, coalescer, hydrocyclone, and

155


III-39

dissolved air flotation (DAF). Thus each process will has its own limitation, bound by assumptions of STOKES law as well as its unique working characteristic.

Apart from STOKES law-based processes, there are other oil/water separation processes that based on other physical properties between oil and water. In GPI lab, Prof. AURELLE’s teams also studied these processes, i.e. skimmer (which is based on interfacial tension concept), membrane process (which depend on mass transfer phenomena), thermal process (which depends on thermodynamic properties), and chemical process (which depends on chemical reaction).

Thus, to design an efficient oily wastewater treatment process, designer should understand working principle and limitation of each process. The following chapters will devote to each oily wastewater treatment processes studied and perfected by Prof. AURELLE’s team within 3 decades. However, in this section, description of each process will be described briefly as follows,

STOKES law-based processes

2.4.1 Decanter

Decanter is an oil separation process that depends purely on STOKES law without modification of their parameters. The oil droplets in wastewater will allow to decant (or rise) naturally to the surface of water .It is very simple and proven to be very good process for separation of free oil and primary emulsion. Rising velocity depends mainly on the droplet size. In case of very small droplets, it may take too long time for the droplets to reach the surface or it may need so large tank to become economical. However, performance of decanter can be enhanced by reducing of rising distance of droplets, such as by lamella-plate insertion. Actually, GPI lab had initiated a very compact decanter by this concept, called “Spiraloil”, which is commercialized by Elf Total Fina. Decanter is described in details in chapter 4.

2.4.2 Coalescer

Coalescer is a modified or accelerated STOKES law-based separation process by increasing the size of oil droplet. Since rising velocity is proportional to square of droplet size, theoretically, coalescer can enhance oil separation efficiency more rapidly than other process that work on other parameters. AURELLE is one of the pioneers on granular bed coalescer, which was patented and, later, commercialized by Elf Total Fina. He also initiated other type of coalescer, i.e, granular bed coalscer with guide to handle wastewater with high oil concentration, fibrouse bed coalescer that can handle wastewater with SS. Coalescer can handle secondary emulsion effectively. However, its performance with stabilized emulsion is not so good since the droplets are very table and unlikely to coalesce with each other. Coalescer is described in details in chapter 5.

2.4.3 Hydrocyclone

Hydrocyclone is a modified or accelerated STOKES law-based separation process by increasing the acceleration of the system. Hydrocyclone replaces gravity acceleration (g) with centrifugal acceleration, which can reach several hundreds times of g. Thus it is very compact separation process. AURELLE and MA had initiated three-phase hydrocyclone that can separate oil, solids and water simulteneously. Hydrocyclone use its inlet flow energy to

156


III-40

operate. So it is noted for its increase in efficiency with the increase of inlet flow, which is unique advantage of the system. It can handle secondary emulsion effectively. However, its operation has high shear characteristic. Furthermore, it driving force come only from its inlet flow so the centrifugal force is limited. So it can not separate very small droplets (<10 microns). It must also be noted that hydrocyclone is actually a concentrator. It can not separate water-free oil. Separated oil usually contains some water. Thus it need to be further treated by other process. Hydrocyclone is described in details in chapter 6.

2.4.4 Dissolved air flotation (DAF)

DAF is a modified or accelerated STOKES law-based separation process by decreasing density difference between oil and water. It can be obtained by addition of gas bubble or air bubble to wastewater to form agglomerates with oil droplets. Since air or gas has low density the agglomerates, then, has lower density than oil drops, resulting in higher rising velocity. DAF is noted for its versatility. It can separate both solids and oil. Its efficiency depends on formation of oil-bubble agglomeration. Thus its performance is low if droplets are very small, since collision between air bubble and oil drops is difficult. In practice, DAF will be used after coagulation-flocculation process, which can increase the droplet sizes chemically. DAF is described in details in chapter 7.

Other processes

2.4.5 Skimmer

Skimmer is the device designed to remove oil film from the water surface. GPI lab had studied and perfected skimmer’s oil performance so it can selectively remove only oil without drawing the water with it. Operating principle of the oleophilic oil skimmer is based on surface tension concepts. The skimmer studied in GPI lab was also commercialized. Details about skimmer are described in chapter 3.

2.4.6 Membrane processes

Membrane processes are the promising separation processes that gain popularity in these recent years. The key part of the processes is the membrane, which is porous material with various ranges of pore size to suit the sizes of material to be retained or separated. Its working principle can be compared, but not exactly identical, to that of filtration process. With properly selected pore size, membrane processes can separate target materials with relatively high efficiency. GPI lab had studied application of various membrane processes, i.e., microfiltration, ultrafiltration, nanofiltration and reverse osmosis on oily wastewater treatment. The lab also initiated studies on capacity (or flux) enhancement of membrane process for cutting oil emulsion treatment. Membrane processes are described in details in chapter 8.

2.4.7 Thermal processes

Thermal process is the separation process based on thermodynamic properties of oil/water mixture. GPI lab emphasizes on heteroazeotropic distillation, which can be carried out at the temperature that is lower than those of pure water and pure hydrocarbons. It is achieved by addition of proper hydrocarbons, called entrainer or extractant, into the wastewater to azeotrope point, where oil, water and vapor co-exist, during distillation. The process can be used to separate water from refinery slops and permeate from membrane

157


III-41

process of oily wastewater treatment. Thus it provides opportunity to recover these supposed-to-be-wasted oil and treat the most problematic waste. Thermal processes are described in details in chapter 9.

2.4.8 Chemical process

Chemical process for oily wastewater treatment is mainly destabilization process for destabilizing of stable emulsion. It consists of 2 main mechanisms, i.e. destabilization of oil droplets and coalescence or flocculation of destabilized oils. This process is necessary when the wastewater to be treated contains stable or stabilized emulsion, otherwise it can not be treated by STOKES law-based processes. Destabilization chemicals includes salts, acids, and polyelectrolytes. GPI lag had studied destabilization mechanisms of each chemical which provide understanding and knowledge on chemical selection. The detail is discussed in chapter 10.

2.4.9 Finishing processes

To meet the effluent standards, effluent from aforementioned processes may need further treatment before discharge to receiving water body. The most widely used processes for finishing propose are biological treatment and adsorption (by activated carbons). These processes are not the main interest in Prof. AURELLE’s team, which emphasize of physical processes. However, to fulfil the whole oily wastewater treatment processes, biological treatment and adsorption are briefly described in chapter 11.

The theories and details of each processes described in section 3 to 11 will, then, be used to develop the software for design and simulation of oily wastewater process train, which is the scope of work of this thesis.

2.5 Determination of degree of treatment

2.5.1 Overall degree of treatment

Before designing oily wastewater process, required overall degree of treatment or effluent quality of the treatment system must be set. In case that the effluent is discharged to public receiving water body, such as sewage system or natural stream, the final effluent quality must conform to every applicable law. Examples of effluent standards are shown in chapter 1. In case that the water is recycled or reused, required water quality for those purposes, such as standard of recycled water quality, standard of recycled cooling water, etc., will determine the overall degree of treatment.

2.5.2 Degree of treatment of each process

Since the performance of every treatment process depends mainly on wastewater characteristic, for existing facilities, wastewater must be thoroughly analyzed to obtain its detailed characteristic. For new facilities, its characteristic must be carefully estimated. In case of oily wastewater, performances of oil separation processes are based on degree of dispersion and stability of the wastewater. When these data are available and limitation as well as expected performance of each treatment process is thoroughly understand, it is possible to select feasible process train that can treat the wastewater to meet the required degree of treatment.

158


III-42

Example A: consider wastewater from in-land refinery contains 230 mg/l of non-stabilized secondary emulsion and 10 mg/l of emulsified oil. If applicable effluent standard in this case is 12.0 mg/l, in this case, the degree of treatment of the system is (10+230-12)/(10+230)*100 = 95%. We can separate the secondary emulsion by compact decanter, coalescer or DAF, which secondary emulsion removal efficiency is expected at 90%. So it requires further treatment and the degree of treatment required in the next stage is at least 100-((100-950)/(100-80)) = 50%. In this case, It can be archieved by sending to the water to mix with domestic wastewater and treat by biological treatment. The concentration of 12.0 mg/l may be met by dilution effect of domestic wastewater alone. Moreover the biological process can be reduced the oil concentration at least 50%. Thus it can be certain that the effluent concentration of 12 mg/l is guaranteed at all times.

2.5.2.1 Graded efficiency and total removal efficiency

As stated before that efficiency of most separation process depends on the size of oil droplets, most of researches in GPI lab had common aim to establish the relation between the efficiency and droplet size, which will be shown in the following chapters. If granulometry data (see fig. 2.3.2-5) is available, it will provide clear data on percentage of each size of oil droplets. So it is very useful for removal efficiency calculation of each droplet size (graded removal efficiency) of each process. Summation of graded efficiency of each process results in total removal efficiency of that process corresponding to that particular wastewater, which is more accurate than the estimated value from literatures alone. In case that the treatment system consists of several processes, overall efficiency can be determined from summation of greaded effeciecny of ecah process, as shown in eq. 2.5.1.

100))1)....(1)(1(1( ,2.1,, ⋅−−−−= idddoveralld ηηηη % {2.5.1a}

( )%100

,

,,

,

max

min ⋅⋅

=∑

io

d

diodid

it C

Cηη {2.5.1b}

( )%100

max

min

,

⋅⋅

=∑

o

d

dodoveralld

overall C

Cηη {2.5.1b}

Where ηoverall = Overall efficiency of the treatment process train ηd,overall = Graded removal efficiency of the treatment process train ηd,i = Graded removal efficiency of the process “i”, calculated by

equation of each process shown in the following chapters ηt,i = Total removal efficiency of the process “i” Co = Oil inlet concentration of the treatment process train Co,i = Oil inlet concentration of the process “i” Cod,i = Graded oil inlet concentration of the process “i” i = the number of processes in the process train, i ≥ 1

To achieve the required effluent standard, overall efficiency of the system must be equal of higher than the required degree of treatment, established in section 2.5.1.

159


III-43

2.5.2.2 Cut size determination

Since the bigger oil drops are realtively easier to remove than the smaller ones, it is recommended to select the treatment process to handle the range of oil drops from the biggest to the smallest size that contributes to the required degree of treatment (in case that the granulometry data is known). This smallest size in the range described above is called cut size.

For example, if the granulometry of the wastewater is as shown in fig. 2.3.2-4, if the required degree of treatment is 90%, from the curve, the cut size will be 90 microns.

The cut size is very useful in process design for it can be used as a representation of the whole wastewater to calculation the required size of separation process. It can save calculation time when many processes are to be compared. However, sizing of the process calculated from this theoretical cut size may not be exactly equal to the degree of treatment since the cut size is selected based on the assumption that the graded removal efficiency of droplet equal to larger than cut size is 100% (see fig. 2.5.2-1). Thus some adjustments should be made after detailed calculation of the efficiency by eq. 2.5.1.

Droplet size

Req

uire

d de

gree

of

trea

tmen

t

Theo. cut size

100%

Acc. %(by weight)

Droplet diameter

ηoverall

Real eff.Theo. eff

ηR = Required degree of treatment

Theo.cut size

Realcut size

100%

ηR

Fig. 2.5.2-1 Cut size determination

2.5.2.3 Economics of the processes

Performance of each separation process also depends on other parameters apart from the droplet size. Change in these parameters will also make the performance of the process change. Thus the process train can be optimized by changing influent parameters, usually the geometry, of each process in the process train to make the overall efficiency as close to the required degree of treatment as possible.

Treatment capability of processes may overlap in some range of droplet sizes, which will be summarized again in chapter 12. For example, droplet size around 20 microns can be treated by coalescer or compact decanter. This leads to several combinations of process trains. For example, the process train in example A may be divided into 2 alternatives, i.e. (1) Compact decanter+activated sludge (AS), (2) Coalescer+Activated sludge. In each alternative, the sizes of equipment and reactors can be varied within its own valid range. Thus, to select the best alternative and process sizes, economics of each alternative should be compared.

160


III-44

Since the expected degree of treatment is identical, least cost criteria are usually adopted. The most suitable process train is practically the system that can treat the water to meet the required degree of treatment at the least capital cost and operating expense (see fig. 2.5.2-2).

Size or cost

η

HydrocycloneCoalescer

Overall eff.

Required degree of treatment

Most suitable(economic) size

Technical acceptable range

Fig. 2.2.5-2 Economics of processes (least cost criteria)

However, to select such a system, many alternatives must be calculated. This is the reason that the computer program for oily wastewater treatment process calculation, according to the scope of work of this thesis, is developed. The program will help perform complex calculation and handle calculation of many alternatives within very short time. So the results will lead to the effective oily wastewater treatment system design. Details of the program is described in Part 4.

161

Chapter 3 Oil skimmer

III-45


3.1 General

The most basic form of oily wastewater is the presence of oil film or layer on the water surface. The most severe case of this form of oily wastewater is the large-scale oil spill, caused by tanker or offshore platform accident. The oil will spread over wide area and cause adverse effect on ecosystem. When the accident occurs, the oil confinement, such as booms, is arranged to limit the spreading area. Then, the spilled oil is, sometimes, dispersed by chemical dispersant to form tiny droplets, then left to biodegrade. Biological accelerating agents may be added to promote biological degradation. Sometimes, specific adsorbents are added to the oil to make it settleable or in more manageable form, such as floating agglomerate scum. These methods may still cause adverse effect on the environment. Thus, if possible, the oil layer is preferably removed in form of water-free oil for it can be reused.

In oily wastewater treatment process, the goal of almost all of the separation processes is to separate the oil droplets and form an oil layer on the water surface. After that, it is also crucial to remove this oil layer as water-free oil, otherwise it will become new oily wastewater. In a small container, the oil layer can be removed manually or by simple devices, such as weir, bell mouth pipe. Anyway, in case of oil spill at sea, or upscale tank, it is difficult to remove only oil without getting water with it even though the oil film is visibly stratified from the water. In these cases, specific devices that have good oil selectivity are required. There are many variations of these devices as shown in fig. 3.1-1. However, they can be generally divided into 2 concepts i.e.,

• Pumping or hydraulic devices: The oil will be directly pumped out or intercepted by controlled hydraulic devices, such as adjustable-weir. Important factor that governs the performance is the oil layer thickness. So, some mechanisms are provided to locally thicken the oil layer before removing it. Examples of the devices include pump skimmer and weir skimmer.

• The devices based on adsorption property: This group of devices makes use of difference between oil and water adsorption property of material to remove the oil. The oil selectivity of material is based on the concept of interfacial tension, as described in chapter 2. When design properly, the device can remove virtually water-free oil from the wastewater. Examples of the devices include drum skimmer, disc skimmer, belt skimmer, etc.

This chapter emphasizes on the drum and disk skimmer, which are thoroughly researched in GPI lab.

162


III-46

a) Disc skimmer (source: Highland tank) b) Belt skimmer (Source: Ultraspin)

c) Drum skimmer (source: ELF/GPI lab) d) Mop skimmer (source: Ultraspin)

e) Weir skimmer (Source: Skimoil) f) Pump skimmer (Source: Ro cleandesmi)

g) Dispersant spray (Source: Ro-cleandesmi) h) Absorbent (Source: Ro-cleandesmi)

Fig. 3.1-1 Examples of oil skimming devices

163


III-47

3.2 Oil drum skimmer

3.2.1 Working principles

Oil drum skimmer is the equipment that has a rotating drum which its surface acts as an oil-skimming surface (fig. 3.2.1-1). The oil will adhere to the skimming surface, lift up from the surface by the skimmer’s rotating movement, and then be scraped off by a scrapper blade into a receiving channel or container.

Drum skimmer

Scrapper

Oil receiving trough

Oil layerWater

Fig. 3.2.1-1 Lab-scale drum skimmer: Major components are shown. (Source: GPI lab)

The working principle of oil drum skimmer, as well as other absorption based devices, is based on surface energies of hydrocarbon, water and skimmer material. To obtain water-free oil, the material must have good oil selectivity. THANGTONGTAWI [5] had studied the effect of surface energies on oil selectivity and concluded that oil selectivity of the material depends on the difference between its critical surface tension (see chapter 2) and the superficial (surface) tensions of hydrocarbon and water.

Water normally has higher superficial tension (around 72 dyne/cm, depending on the temperature) than oil (25 – 40 dyne/cm). The diagram in fig. 3.2.1-2 shows surface energy of water, oil and various materials. Adhesion of oil or water at the material surface can be explained in term of the adhesion work (see chapter 2, eq. 5.2.5) as shown in table 3.2.1-1. If the adhesion work of oil on solid surface in presence of water (Wadh(so)w ) is greater than that of water on the surface in presence of oil (Wadh(sw)o), the oil film will adhere to the skimmer surface and not be replaced by water film when the skimmer is rotated until the oil film submerges in water. Thus, it still adheres to the surface when it is lifted from the liquid and then removed by the scraper.

From this concept, the oil selectivity of materials can be summarized as follows,

• Material of high surface energy (or critical surface tension), such as stainless steel (γ > 72 dyne/cm), tends to be adhered by water, rather than oil.

• Material that have the surface tension in the vicinity of oil’s, such as polyvinylchroride (PVC), polypropylene (PP), tends to adhered by oil, rather than water. It can be adhered by water if the oil film is not present. However, it will start recovering oil once the oil is present again.

164


III-48

• Material that has very low critical surface tension (lower than surface tension of oil), such as PTFE and fluorocarbon variations, is proven to have very good oil selectivity since its surface energy is so low that it is hardly adhered by water. When used as a skimmer material, it recovers only oil and hardly or does not recover water.

2010 30 40 50 60 70

PTFE(19)

PP(30)

PVC(39)

Water(72)

Stainless steel(> 72)

25 35

Material of low surface energy

Material of high surface energy

oil

Surface energy(dyne/cm)

Fig. 3.2.1-2 Surface energy or superficial tension of materials, oil and water

Table 3.2.1-1 Work adhesion and contact angles of oil, water and various materials

Material Contact angle of oil drop in water (θ)

Wadh(so)w = (γow(1+cos θ))

Wadh(sw)o = (γow(1+cos (180−θ))

Stainless steel (SS) 124o 0.01543 N/m 0.05457 N/m

PVC 26o 0.06650 N/m 0.0035 N/m

PP 17o 0.0685 N/m 0.0015 N/m

PTFE 47o 0.0589 N/m 0.0111 N/m

Note: The oil used is kerosene. γow (at 20o C) = 35 dyne/cm.

It is interesting to note that the stainless steel, which is the most oleophilic material (oil can spread over its surface), makes the worst oil selectivity. This, sometimes, causes confusion since the device is, sometimes, called oleophilic oil drum skimmer. So, It should be reminded that the device got its name from its oil selectivity, but the skimming material itself, even though can considered as oleophilic (contact angle < 90o), is selected from its low surface energy, rather than its oleophilicity.

3.2.1.1 Exposure history

Another important characteristic of material related to oil selectivity is the “exposure history” or the order and duration of contact between solid and liquids. THANGTONGTAWI study shows that,

Material of high critical surface tension

• Even the high critical surface tension material can remove oil if it is exposed to pure oil first. Performance from SS, PVC, PP and PTFE skimmers in this case are about identical. In case of SS, this does not seem to conform to the adhesion work

165


III-49

calculation result. This can be explained by the effect of “masking”. The oil film covers overall surface of the skimmer so the water cannot contact directly to the skimmer to form a water film.

• However, if the oil is entirely recovered, or oil film locally raptures, the SS will be immediately adhered by water and stop recovering oil. Then it will recover substantial amount of water.

• Even though the water film is removed by scrapper, there is always a thin film left due to very high adhesion work of the film that causes by chemi-sorption or polar bond to the solid surface. When the oil is present again, the oil cannot contact directly to the solid surface, then cannot be recovered.

Material of medium critical surface tension (PVC, PP, etc.)

• When oil film is present, PVC or PP skimmer will recover oil. After the oil is totally removed, the skimmer is still not immediately adhered by water film. This can be explained by the presence of residual oil film on the surface of the skimmer.

• The residual oil film will gradually disappear. So, after some times (about 24 hours), the skimmer will be wetted by water and start recover water.

• If the oil is present again, the skimmer will immediately resume oil recovery since, from concept of work adhesion, it is preferably adhered by oil.

Material of very low critical surface tension (PTFE, etc.)

• Because of its low critical surface tension, this type of material is not affected by exposure history for it is hardly adhered by water. PTFE skimmer always recover oil without water even when it is left to operate without the presence of oil layer for a long period of time.

3.2.1.2 Mathematical model of drum skimmer

Mathematical model of oil drum skimmer, proposed by THANGTONGTAWI, was derived from dimensional analysis approach, using Buckingham Pi theory under a wide range of material and wastewater properties as well as operating parameters. The model is as shown below.

0.514g

L0.486oν

1.541N1.5413.035DP = {3.2.1}

Where P = Oil productivity of the skimmer (m3/s) D = Diameter of skimmer (m) N = Rotational speed of the skimmer (rev/s) νo = Kinematics viscosity of oil (m2/s) L = Length of the skimmer (m) G = Gravitational acceleration (m/s2) γo = Superficial tension of oil (kg/s2 or N/m = 1000 dyne/cm)

166


III-50




• Oil density is around 0.79 – 0.83 kg/m3. Oil dynamic viscosity (μ) tested is between 1.35x10-3 to 291x10-3 (N.s)/m2 (1.35-291 cp).


• Recommended minimum immersion depth is 1.0-2.0 cm.

• Drum skimmer surface used for model development is polypropylene. But it is proven to be valid for SS, PVC, and PTFE [5]. As shown in fig. 3.2.1-2.

Oil removal efficiency of the skimmer is usually 100%, if the conditions stated above are satisfied.

100%tη = {3.2.2}

3.2.1.3 Influent parameters

Parameters that affects the performance of drum skimmer are as summarized below. Graphical presentation of effect of various parameters on the performance of the skimmer is shown in fig. 3.2.1-3.

1. Diameter of drum

The oil productivity is proportional to drum diameter because it directly relates to the skimming surface area. The larger the diameter, the bigger the area, thus the productivity. However, the study on the oil film thickness on the surface of the drum [5] shows that the diameter has only little effect, thus negligible, on the film thickness, which direct relates to the skimmer performance. So the diameter of the drum could be selected to suit the tank freeboard.

2. Length of drum

The oil productivity is proportional to drum length because, again, it directly relates to the skimming surface area. The longer the drum, the bigger the area, thus the productivity. It does not affect the oil film thickness on the surface of the drum. So there is no effect on the performance. However, the rotation of the drum will cause eddy current at the 2 ends of the drum. This turbulence will propel the oil film and prevent it to contact with the parts of drum surface. If the drum is long, the area of eddy zone, compared to the entire length of drum, will be small. So the length of drum will help improving the skimmer performance in this sense.

3. Oil layer thickness

Oil layer thickness on the water surface will affect the oil productivity of the skimmer when it is thinned out until it cannot supply the oil fast enough to the skimmer. Then there will be some part of skimmer that does not contact with the oil layer and has no

167


III-51

productivity. So the overall productivity will drop. The layer thickness of 0.4 – 0.5 cm is considered thick enough to ensure continuous supply of oil to the skimmer.

4. Immersion depth of drum

Immersion depth means the height of the part of the drum that is lower than water surface, measured from the bottom of oil layer. At immersion depth of 1-2 cm. THANGTONGTAWI reported that this parameters does not practically affect the performance of the skimmer. The author has tested at the immersion depth around 5 - 7 cm. The result still confirmed THANGTONGTAWI’s conclusion.

5. Immersion depth of drum

Immersion depth means the height of the part of the drum that is lower than water surface, measured from the bottom of oil layer. At immersion depth of 1-2 cm. THANGTONGTAWI reported that this parameters does not practically affect the performance of the skimmer. The author has tested at the immersion depth around 5 - 7 cm. The result still confirmed THANGTONGTAWI’s conclusion.

6. Roughness of drum surface

Surface roughness affects the contact angle, thus wettability, as described in chapter 2. However, in case oil skimmer, this change is overshadow its affect on difficulty to remove film from the drum surface. Concavity on the surface may cause some difficulty to scrape the oil film off the drum surface even though the oil thickness may be increased by the roughness. So it is recommended to use relatively smooth drum surface.

7. Velocity of drum

Rotating velocity directly affects the oil productivity as shown in eq. 3.2.2. The higher the velocity, the higher the productivity. However, the velocity also governs the entraining of water by the dynamic force. If the velocity is too high, the rotation of the drum will draw the water up too fast until it can reach the scrapper blade and entrain with the skimmed oil. It is recommended to use peripheral velocity (or tip speed) of 0.44 m/s or less to prevent water entraining.

8. Viscosity of oil

Oil productivity, as well as the oil film thickness on the surface of the drum, will increase if the viscosity of the oil increase, as clearly shown in eq. 3.2.1.

9. Superficial tension of water and interfacial tension of oil/water

Surfactant is usually present in oily wastewater. The presence of the surfactants lowers the superficial tension of water and interfacial tension of oil/water. Effect of surfactant on the skimmer operation can be divided into 2 cases, i.e.,

• When oil layer is present: The efficiency of the drum skimmer is proven to be practically independent of the presence of the surfactant [5]. So change in superficial tension of water and the interfacial tension of oil/water does not affect the performance of the skimmer.

168


III-52

• When oil layer is absent: After the oil is removed and the skimming surface is exposed to water. The surfactant will lower superficial tension of water until it is close to naturally interfacial tension of oil/water. So there is no difference between oil and the water. Thus the skimmer will lose its selectivity, even in the case of PTFE, and start recovering the water.

Oil

Prod

uctiv

ity (P

)

Tip speedof drum

Increase D, LIncrease oil viscosity

0.8 m/s(max of model)

0.44 m/s(recommended)

a) Oil productivity of various drum materials when the oil layer is not a limiting factor [5]

b) Effect of parameters on drum skimmer performance

Fig. 3.2.1-3 Influent parameters on drum skimmer performance

3.2.2 Design calculation and design consideration

1 Drum skimmer sizing

The size of the skimmer can be calculated by eq. 3.2.1. If oil concentration or quantity of inlet oil is known, it could be used as required oil productivity of the skimmer. Then, the oil productivity, geometry of tank (width, length), tank freeboard and available installation space of the skimmer and oil outlet pipe should be taken into account in order to select a suitable size of the skimmer.

Energy requirement of the skimmer is the energy for driving the skimmer. It can be calculated by simple product of torque and speed. The torque required depends on the structure, weight and size of the drum.

2 Design consideration

2.1 Limitations of the equation

The mathematical model of drum skimmer (eq 3.2.1) is valid only under its limitations shown in section 3.2.1. Application of the model beyond its limitation may cause unpredictable error.

2.2 Practical design consideration

Besides the limitations of models shown in section 3.2.1, there are some assumptions or operating condition that affect the performance of the skimmer but cannot be expressed in the form of equation. To design a skimmer, these assumption and precaution, as described below, should be taken into account to ensure good performance of the skimmer.

169


III-53

1) The model is developed under the assumption that the scrapper can totally remove or scrape the oil film from the drum surface. So in the real situation, scrapper should be designed properly to make sure that it will not be a limiting factor in the operation. Scrapper should be made of flexible material to ensure its close contact to the entire length of the drum. However it should have good abrasive resistance.

2) The oil productivity in eq. 3.2.1 is valid under the condition that the oil layer is not the limiting factor and the entire length of drum can contact to the oil film. To ensure these condition, the following precaution should be considered;

• The drum should be properly sized to accommodate the inlet oil quantity. So it can operate continuously without the problem about lack of oil layer.

• Automatic control should be provided to stop the skimmer when the oil layer in the tank is too thin to avoid the recovery of water, esp. when high or moderate critical surface tension is used as skimming material, or when there is a risk of the presence of surfactant in wastewater.

• If possible, the skimming surface should be of low critical surface tension material, such as variants of flouorocarbon, to guarantee good oil selectivity and avoid exposure history problem. Critical surface tensions of certain materials are listed in table 3.2.2-1.

Table 3.2.2-1 Critical surface tensions of certain materials [42]

Material γc (dyne/cm, 20 oC)

Poly(1,1-dihydroperfluoroctyl methacrylate) 10.6 Polyhexafluoropropylene 16.2 Polytetrafluoropropylene 18.5 Polytrifluoretylene (PTFE) 22 Poly (vinylidene fluoride) (PVDF) 25 Poly(vinyl fluoride) 28 Polyethylene ({PE) 31 Polytrifluorchloroethylene 31 Polystyrene 33 Poly (vinyl alcohol) 37 Starch 39 Poly (methylmethacrylate) 39 Poly (vinyl chloride) (PVC) 39 Poly (vinylidene chloride) 40 Poly (ethylene terephthalate) 43 Cellulose 45 Poly (hexamethylene adipiamide) 46

• Operation of the skimmer can cause eddy current around the ends of the drum that causes the oil layer in this area vanished quicker than other area. It results in non-productive zone of the skimmer (see fig. 3.2.2-1). To avoid this, the oil layer should be kept at

170


III-54

certain thickness to cope with this effect. The thickness of 1.0 cm is considered safe.

• If the length of the drum is smaller than the width of the tank to be skimmed, guide walls should be installed to help guiding the oil layer to the drum.

a) Eddy currents b) No oil zone or non-productive zone

Fig. 3.2.2-1 Occurrence of eddy currents from drum operation and no oil zone or non-productive zone that affects the productivity of the skimmer (Source: Oil Spill Cleanup)

No oil zone

Eddy

3.3 Oil disc skimmer


Oil disc skimmer is the equipment that has a rotating disc or discs, instead of drum, which surface act as oil-skimming surface (fig. 3.3.1-1). The oil will adhere to the skimming surface, lift up from the surface by the skimmer’s rotating movement, and then be scraped off by a scrapper blade into a receiving channel or container.

Working concept of the disc skimmer is identical to that of the drum skimmer except the skimming surface. The concept of surface tension and exposure history described in the previous is still valid. For the skimming surface area, in case of disc skimmer, it is circular surface of the disc.

Disc

Scrapper

Drive unit

Oil trough

Components of disc skimmer

(Source: Abanaki) Eddy current from rotating of the disc

(see section 3.3.2)

Fig. 3.3.1-1 Disc skimmer

Eddy

171


III-55

3.3.1.1 Mathematical model of disc skimmer

Mathematical model of oil disc skimmer, proposed by THANGTONGTAWI, was derived from dimensional analysis approach, using Buckingham Pi theory under a wide range of material and wastewater properties as well as operating parameters. The model is as shown in eq. 3.3.1

0.332g

1.17I0.452oν

1.212N1.2581.328DP = {3.3.1}

Where P = Oil productivity of two sides of the skimmer (m3/s) D = Diameter of skimmer (m) N = Rotational speed of the skimmer (rev/s) νo = Kinematics viscosity of oil (m2/s) I = Immersion depth of the skimmer (m), I ≤ D/2. g = Gravitational acceleration (m/s2) γo = Superficial tension of oil (kg/s2 or N/m = 1000 dyne/cm)




• Oil density is around 0.79 – 0.83 kg/m3. Oil dynamic viscosity (μ) tested is between 1.35x10-3 to 291x10-3 (N.s)/m2 (1.35-291 cp).


• Disc skimmer surface used for model development is PVC. But it is proven to be valid for SS, PP, and PTFE [5].

The major difference between drum and disc model is that the disc model is the function of immersion depth while this parameter hardly affects the drum skimmer performance.

Oil removal efficiency of the skimmer is usually 100%, if the conditions stated above are satisfied.

100%tη = {3.3.2}


Parameters that affect the performance of disc skimmer are identical to that of the drum skimmer, as described in section 3.2.1.2, except the immersion depth. Graphical presentation of effect of various parameters on the performance of the skimmer is shown in fig. 3.2.1-2.

Immersion depth of disc skimmer, in this case, represents the depth of a part of the disc that is under the liquid free surface. From eq. 3.3.1, it shows that oil productivity incerase

172


III-56

with increase in the immersion depth. However the immersion depth is limited by the diameter of the disc (I ≤ D/2)


1 Disc skimmer sizing

The size of the skimmer can be calculated by eq. 3.3.1. If oil concentration or quantity of inlet oil is known, it could be used as required oil productivity of the skimmer. Then, the oil productivity, geometry of tank (width, length), tank freeboard and available installation space of the skimmer and oil outlet pipe should be taken into account in order to select a suitable size of the skimmer.

Like the drum skimmer, energy requirement of the skimmer is the energy for driving the skimmer. It can be calculated by simple product of torque and speed. The torque required depends on the structure, weight and size of the drum.

2 Design consideration

2.1 Limitations of the equation

The mathematical model of disc skimmer (eq 3.3.1) is valid only under its limitations shown in section 3.3.1. Application of the model beyond its limitation may cause unpredictable error.

It must be noted that the oil productivity from eq. 3.3.1 is for two sides of the disc. If scrapper is installed at only one side of the disc, the productivity can be safely assumed to be 50 % of the value from eq. 3.3.1. Productivity of several discs are the product of the result from eq. 3.3.1 and the number of the disc “n”.

2.2 Practical design consideration

Like the case of drum skimmer, besides the limitations of models shown in section 3.3.1, there are some assumptions or operating condition that affect the performance of the skimmer but cannot be expressed in the form of equation. The precaution proposed for the drum skimmer can also be applied for the disc skimmer. Like the drum skimmer, operation of disc skimmer also results in non-productive zone of the skimmer, starting at the axis of the skimmer (see fig. 3.3.1-1), when the oil layer is almost totally recovered. To avoid this, the oil layer should be kept at certain thickness to cope with this effect. The thickness of 1.0 cm is considered safe [5].

3.4 Productivity comparison between drum and disc skimmer

To compare the productivity of drum disc skimmer, THANGTONGTAWI had proposed the alternative model of the disc skimmer (eq. 3.4.1) that has the same exponents as the drum skimmer except for that of geometric parameters (D, L, I).

( ) ⎥⎦⎤

⎢⎣⎡ −−= 54.2)5.0(54.25.00.514g

0.486oν

1.5413.464NP IDD {3.4.1}

173


III-57

From eq. 3.4.1 and 3.2.1, it the productivity of the two skimmers are assumed to be equal, we can write the relation between the length of the drum and immersion depth of the disc at the same diameter as shown in eq. 3.4.2. This equation is useful for comparison the dimensions of the two devices. The dimension can be used for preliminary cost estimation, which can help determining which type of skimmer is more suitable for the project.

( ) ⎥⎦⎤

⎢⎣⎡ −−

⋅

⋅= 54.2)5.0(54.25.0

541.1035.3

464.3L IDDD

n {3.4.2}

Where n = the number of the disc (n ≥ 1)

3.5 Advantage and disadvantage of drum and disc skimmer

Advantages: Main advantage of drum skimmer and disc skimmer, as well as other adsorption based skimming devices, is its oil selectivity. The devices require no additional chemicals and the oil recovered from these devices will relatively water-free, which is more suitable to recycle or reuse. Application on oil-spill recovery, developed by ELF-INSA collaboration, is one of the outstanding achievements of the skimmer (fig 3.1-1). The device helps solving the environmental problem by collect the oil out off water surface.

Disadvantages: An inconvenience of disk and drum skimmer may be due to its geometry. The scrapper cannot to be placed higher than the top of the drum or disc. This means the device cannot lift the oil higher than its diameter. So, in case that the tank freeboard is large or in case of the tank with variable water level, it may cause an inconvenience on installation of oil outlet pipe. However, this inconvenience can be solved if outlet pipe sleeve through the wall is designed and installed in advance. In case of a variable-water level tank, the skimmers can be installed on a pontoon and a small oil pump can be used to lift the oil to the desire elevation. Otherwise a high-lift device of the same type, such as belt skimmer, can be used. But, from its more complex structure, it is always more costly.

174


III-58

a) Drum skimmers and booms for oil spill recovery (ELF/ GPI lab)

b) Installation of skimmer in variable water level tank. An oil pump (below) is used to lift the

skimmed oil. (Source: SkimOil, Ro-clean Desmi

Fig. 3.5-1 Application of the skimmers

175

Chapter 4 Decanting

III-59

Chapter 4 Decanting

4.1 General

Decanting or sedimentation is the most basic separation process. Its working principle bases only upon properties of oil and water, which tend to separate from each other naturally. This process is widely used, even becomes legal standard for some industries, such as refineries in USA. Because of its working principle, which bases on natural unadapted properties, decanting is suitable for separating oil in form of big oil drop or primary emulsion. There are many variations of decanting process, however they can be categorized into 2 groups, i.e.,

• Simple decanter: This group of decanter is the most basic process. They may vary in geometry and details of some components. But they all use the same working principle of natural unadapted decanting. The well known example of this type of decanter is API tank, as shown in fig. 4.1-1.

• Compact decanter: This group of decanter is the modified version of the first group, intended to upgrade the capacity of the existing simple decanter. Or it can be newly designed equipment. The objective of compact decanter design is to enhance the efficiency of decanter, making it handle more capacity, using smaller footprint. This can be achieved by the modification of decanter geometry, such as the insertion of lamella plates. However, the working principle is still identical to that of the simple decanter as, theoretically, there is no modification of any natural properties. Examples of this type of decanter are shown in fig. 4.1-1. Among these, GPI lab has developed one of the most compact decanter, called “Spiraloil”. Its special features and technical design consideration will be presented hereinafter.

a) API tank (Source: Pan American

Environmental)

b) “Spiraloil” compact decanter c) Oil layer at the surface of API tank

Fig. 4.1-1 Examples of decanters

176


III-60

d) Example of up-scale API tank with tank enclosure (Source: USFilter)

e) Graphical image of lamella plate decanter (Source: USFilter)

f) Corrugated plate module of corrugated plate decanter (Source: Dewaterworks)

Fig. 4.1-1 Examples of decanters (Con’d)

4.2 Simple Decanter or API tank


Simple decanter, which is made well known and standardized by the American Petroleum Institute (API), is the simplest oil-water separation process. Its working principle bases on classical STOKES law (cf. Chapter 2). Concept of the operation of the process is to provide sufficient time for oil droplets to float to the water surface and accumulate into oil layer before they have a chance to flow out with the water at the water outlet. The equation that governs the operation of the process is derived from comparing the time required for the droplet to reach the surface with retention time of the tank (τ), as shown in eq. 4.2.1.

177

Chapter 4 Decanting

III-61

τ ≥ (Rising distance of droplets / Rising velocity of droplets) {4.2.1}

Ud

V

Q

d = cut size

d < cut size


L

H, D

Oil droplets

Separatedoil layer

d

ηd

d c

d < d c

Zone 1 Zone 2

d = or > d c

Figure. 4.2.1-1 Schematic and typical removal efficiency curve of simple decanter

Figure 4.2.1-1 shows the diagram of decanting process. From the figure, the longest path to reach the surface is the path starts at the bottom of the tank. The smallest droplet size that can reach the surface is called the cut size (dc). The droplets of cut size or bigger are always separated from wastewater stream with 100% removal efficiency.

The smaller droplets can be also separated providing that it enters the tank near the water surface. When uniformly distributed influent flow is valid, which is true for almost all of properly designed tanks, the removal efficiency of the droplets smaller than cut size are proportional to theirs corresponding rising velocity. From these concepts, the models of decanting process are as shown in Eq. 4.2.2 to 4.2.5.

For the simple decanter of the length “L”, the width “W” and the water depth “D”, from eq. 4.2.1, we have,

dcUH

WDQ

L==

)(τ {4.2.2a}

For the simple decanter, rising distance of the droplet (H) is equal to the water depth (D), then

⎟⎠⎞

⎜⎝⎛=⎟

⎠⎞

⎜⎝⎛=

SQ

LWQU dc {4.2.2b}

In laminar flow regime (Re < 1), STOKES law is applicable. So, rising velocity of droplet of diameter “d“ (Ud) can be written as shown in eq. 4.2.3.

cd

dgUμ

ρ18

2⋅⋅Δ= {4.2.3}

178


III-62

Then 2/118

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⋅Δ

=Sg

Qd c

c ρμ {4.2.4}


%100=dη {4.2.5a}


%100⋅=dc

dd U

Uη {4.2.5b}

Where dc = Cut size of the decanter D = Diameter of dispersed phase, in this case, oil S = Bottom projection area of the tank Q = Wastewater flowrate A = Flow (Cross section) area H = Travelling or rising distance of oil drop D = Water depth ρc = Density of continuous phase, which is water, for oily wastewater Δρ = Difference between density of dispersed phase and continuous

phase μC = Dynamic (or absolute) viscosity of continuous phase, which is

water, for oily wastewater Ud = Rising velocity of the droplet diameter “d” Udc = Rising velocity of the droplet at cut size “dc” Co = Total inlet concentration of oil ηd = Removal efficiency of the tank for the droplet diameter “d”, or

graded efficiency

Typical characteristic of the removal efficiency of decanter is as shown in Figure. 4.2.1-1. Zone 1 in the graph represents the droplet sizes smaller than dc. Zone 2 represents those, which are equal or bigger than dc.

4.2.2 Design calculation

Design procedure for simple decanter is based upon the equations, shown in the previous section. To design the simple decanter, the required cut size will be determined first. After that, the size of the tank can be calculated. Then, graded efficiency (efficiency of each size of droplet) and then total removal efficiency can be determined. Calculation in each step is described below.

1. Cut size determination

The cut size can be determined from the degree of treatment required as well as from the limitation of the decanting processes. Cut size determination from the degree of treatment is described in chapter 3. For the limitations of the decanting process, they come from 2 main parts, i.e.,

179

Chapter 4 Decanting

III-63

• Theoretical limitation: Since the decanter’s working principle is based on STOKES law, it can be applied only when the droplet behavior conforms to the law. Thus, it can not be used with very small droplet sizes for they are subjected to Brownian’s motion and their rising velocity are not governed by STOKES law. Commonly, the simple decanter is used for primary emulsion treatment (d ≥ 100 microns).

• Economic limitation: As stated before that the process is a non-accelerated process, its required footprint or installation area may be unacceptable if too small cut size is selected.

Thus, the suitable cut size will be considered by accounting for the 2 limitations above. API recommends the cut size of 150 microns for API tank.

2. Decanter sizing

The size of the decanter can be determined, based on the equation in section 4.2.1, as follow:

Bottom surface area (S): The decanter size is based mainly on its bottom surface area. In general case, it is identical to the tank surface area. Presence of some structures or components, such as effluent trough or gutter or tank cover (at the water surface), may affect the efficiency of the decanter. However, if these structures are present in the ways that make the rising distance of oil droplets decrease, they will help enhancing the efficiency. However, their effects are normally small, thus, negligible, unless they substantially reduce the rising distance of oil drops. In latter case, the tank will become a compact decanter, which will be described in the next section. Bottom surface area of the simple decanter can be calculated from the rising velocity of the cut size, as shown in eq. 4.2.6. The example of relation between kerosene droplet size and its corresponding rising velocity as shown in fig. 4.2.2-1.

⎟⎟⎠

⎞⎜⎜⎝

⎛=

dcUQS {4.2.6a}

c

cdc

dgU

μρ18

2⋅⋅Δ= {4.2.6b}

Flow velocity (V): Flow velocity means the velocity of total wastewater along the tank, determined from wastewater flowrate and cross sectional area of the tank. Using high flow velocity may cause turbulent or eddy current, especially when the wastewater collides to the far end of the tank. Turbulence in the tank may interfere the decanting of droplets, as well as suspended solids, which are usually present in the wastewater. This may cause carry-over of oil drop and suspended solids with the effluent of the tank.

API recommends that V should not be greater than 0.15 m/s or greater than 15Vdc, whichever is smaller.

Water depth (D) and width (W): Theoretically, the efficiency of the tank does not depend directly on these parameters. However, they have some effects on the tank operation since they are ones of the parameters that govern the flow regime of the tank. Furthermore, if they are not selected properly, they can cause adverse effect, such as eddy, or short circuit.

180


III-64

0

20

40

60

80

100

120

140

160

180

200

0 100 200 300 400 500 600 700 800

Cut size (micron)

Hyd

raul

ic lo

adin

g ra

te ((

m3 /h)/m

2 )

0

0.01

0.02

0.03

0.04

0.05

Ris

ing

velo

city

of d

ropl

et (m

/s)

Fig. 4.2.2-1 Relation between rising velocity, hydraulic loading rate of simple decanter and kerosene droplet size

API recommends that the ratio D/W should not be smaller than 0.3. The value of 0.5 is recommended. The depth (D) should be in the range of 0.9 to 2.5 m.

3. Removal efficiency

To determine the removal efficiency, the graded efficiency (ηd) will be calculated first by eq. 4.2.7a and b. Then, the total removal efficiency (ηt) can be determined from eq. 4.2.7e.


%100=dη {4.2.7a}


%100⋅=dc

dd U

Uη {4.2.7b}

It must be noted that the graded efficiency described above is not yet accounted for effect of flow splitting between water outlet flow and separated oil outlet flow. To calculate graded outlet oil concentration, the effect must be taken into account, as shown in eq. 4.2.7c.

oddout

d CQ

QC )1( η−= {4.2.7c}

Qout is outlet flow at treated water outlet port of the process. Qout is calculated from difference between inlet flow and separated oil (in relatively pure condition) as shown below.

d

d

dodd

out

CQQQ

ρ

η∑ ⋅−=

max

min

)( {4.2.7d}

181

Chapter 4 Decanting

III-65


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {4.2.7e}

ηt = Total Removal efficiency of the tank ρd = Density of dispersed phase, which is oil, for oily wastewater Cod = Inlet concentration of the droplet diameter “d”

4. Energy required

The simple decanter does not require extra energy to make it function. The energy is required only to feed the water into the tank, then the water will flow, naturally, through the tank by gravity. Pressure drop across the tank and piping system depend on tank and piping design. This pressure drop can be calculated by general hydraulic equations, such as Darcy-Weisbach’s, Manning’s, or Hazen-William’s equation, thus will not de described here.

4.2.3 Design considerations

1. Limitations of the equations

The equations described above are developed from the following assumptions. Thus, it is necessary to ensure that these assumptions are valid when design your decanter.

1) Reynolds number, Re, of droplet is between 10-4 to 1, which is the range that STOKES law is valid.

c

dc dUμ

ρ ⋅⋅=Re {4.2.8}

2) The oil droplets are uniformly distributed across the cross section area of the tank, which can be achieved by proper design of inlet chamber.

3) The oil droplet is spherical, which is normally true.

4) For droplets smaller than 20 microns, they are subjected to Brownian motion and cause error in the prediction of the efficiency. So it is recommended to avoid using the decanter for the wastewater with the majority part of oil droplets smaller than 20 microns. However, if these small droplets are the minority part of pollutants, the models can be used to predict the efficiency without any harm because its prediction is usually lower than observed value, thus make the prediction result on the safe side.

2. Safety factors

Since the design equations are simplified by many assumptions, it is recommended to provide some safety factors to the sizing, calculated from the equations, to cover some unexpected effects such as short circuit flow or local turbulent flow.

API recommends the safety factor for turbulent flow (Ft) as the function of (V/U), as shown in eq. 4.2.9. U represents hydraulic loading rate or overflow rate, based on total flow and bottom area. For effect of short circuit, API recommends the safety factor (Fs)

182


III-66

of 1.2. Then the total safety factor (F), by API recommendation, is the product of Ft and Fs (eq. 4.2.9c). Factor F will be used to multiplied the value of S, calculated from eq. 4.2.6.

9617.0)(0355.0)(0005.0 2 ++−=UV

UVFt {4.2.9a}

tst FFFF ⋅=⋅= 2.1 {4.2.9b}

4.2.4 Construction of simple decanters

There are many variations of simple decanters, as shown in fig. 4.2.4-1. The most famous one is API tank, which is the rectangular tank. However, they use the same working principles, can be calculated by the same equations and consists of relatively the same components.

To design the decanter, besides the sizing stated above, proper details of construction of decanter components are also important to guarantee good efficiency. Design considerations for important components of decanter are described hereby.

1. Inlet chamber

Inlet section, as shown in fig. 4.2.4-2, is an important of decanter for it helps assuring that uniformly distribution of oil droplet is achieved. For this purpose, the inlet chamber should be equipped with baffles or energy dissipation devices or structure. Widely used structures include vertical columns, perforated wall or vertical partition. Design of inlet section is a state of art process. So it is recommended to study the successful design from many references and adapt to fit with the condition considering.

2. Separation section

Sizing of this section is obtained from the equations. In this section, not only the oil will be separated, suspended solids will also settle. So sludge hopper or sludge draw-off pipe, or other provisions for sludge removal should be provided. Geometry of this section should be as recommended in the previous section to ensure good hydraulic condition. This section may be covered, if necessary, to prevent accidental ignition and to prevent the loss of volatile hydrocarbons by evaporation. Sludge hopper, sludge scrapper and surface skimmer, should be provided, as shown in fig. 4.2.4-2.

3. Effluent and oil removal devices

Effluent and oil will be removed at the downstream end of the tank. Normally oil retention underflow baffle is installed to prevent the oil to flow out with the effluent. API recommends that this baffle should be installed with a maximum submergence of 55% of the water depth and should be located as close as possible to the oil removal device. The baffle should be extended to the top of the tank or, at least, higher than water surface.

For oil removal devices, in small unit, the weir or slotted pipe is sufficient. For upscale tank, those simple devices may draw the water off along with the oil. So the separated oil still contains some water, and may not be suitable for downstream reuse or recycle process. In this case, the device with more selective property, such as rotating slotted pipe or oil skimmer, is required. There is a chapter in this book, devoted to oil skimming process.

183

Chapter 4 Decanting

III-67

For effluent, it is normally removed from the tank by a weir. In some small units, bell mouth pipe is acceptable. Design consideration for effluent removal device is that it should be of sufficient size to prevent undesired turbulence and provide good flow distribution in the tank. Normally, the weir across the width of the tank is sufficient.

a) Example of up-scale simple decanter with inlet diffuser wall, equipped with

mechanical skimmer (Source: Monroe)

b) Example of small to medium size simple decanter with simple elbow type inlet diffuser and weir for collecting oil (Source: Pan American)

Fig. 4.2.4-1 Variations of simple decanters

a) Graphic image shows important components of API tank (Source: US filter)

Fig. 4.2.4-2 Important components of simple decanter

184


III-68

Flow

15o

Plan Elevation

b) Example of inlet diffusion wall (Source: API [45])

c) Example of inlet diffusion nozzle

(Source: API [45]) d) Chain and flight skimmer (and

scrapper)(Source: Envirinmentalleverage.com)

e) Example of rotating skimmer pipe (Source: Environmentalleverage)

f) Chain and flight scrapper and oil drum skimmer (Source: US Filter)

Fig. 4.2.4-2 Important components of simple decanter (Con’d)

185

Chapter 4 Decanting

III-69

4.3 Compact decanter


This type of decanter is the modification of the simple decanter. The concept is to decrease the rising distance of droplet to intercepting surface without decreasing the retention time. This can be achieved by inserting plates into the simple decanter to act as the interceptors for the rising oil droplets. Rising or travelling distance (H) is the distance between the plates, not the depth of water (D) as shown in fig. 4.2.1-1. There are several variations of this type of decanter. Normally they are named after configurations of their inserted plates, such as:

Parallel plate interceptor (PPI), which its inserted plates are flat sheets, placed parallel to each other.

Tilted plate separator (TPS), which its inserted plates are tilted or inclined.

Corrugated plate interceptor (CPI), which used corrugated plates instead of flat plates.

The equation that governs the operation of the process is modified from the model of simple decanter, as shown in Eq. 4.3.1. From the equation and the figure, it is implied that, in a compact decanter, the simple tank is divided into (N+1) small decanters but the flow velocity through the tank remains the same.

Q

Separated oil floats to surface

Influent Effluent

L

D

H


Fig. 4.3.1-1 Schematic of PPI decanter

2/1

)1(18

⎟⎟⎠

⎞⎜⎜⎝

⎛+Δ

=NgS

Qd

P

cc ρ

μ {4.3.1}

Where Sp = Inserted plate area N = Numbers of inserted plates L = Length of inserted plates H = Rising distance of oil drop, in this case, distance between inserted

plates

Removal efficiency can be calculated using eq. 4.2.5. Typical characteristic of the removal efficiency of the compact decanter is identical to the simple decanter’s, as shown in

186


III-70

fig. 4.2.1-1.But the cut size of this tank will be smaller than the simple decanter’s, providing that they are of the same geometry.

4.3.1.1 General model of plate-inserted decanters

According to the concept of decanting model that derived from comparing detention time with rising time of droplet to intercepting surface, we can formulate the general model to calculate dc for any decanter as shown in eq. 4.3.2

General model for calculating the cut size, when H, L and A can be clearly defined, is as shown in eq 4.3.2.

2/118⎟⎟⎠

⎞⎜⎜⎝

⎛Δ

=gLA

HQd c

c ρμ {4.3.2}

Where Q = Wastewater flowrate H = Rising distance of oil drop, depends on the configuration of decanter L = Length of interceptor surface A = Flow area (or cross section area) of decanter

4.3.1.2 Plate-closely-inserted decanters

GPI lab has also developed a decanter, called “Spiraloil” (see fig. 4.3.1-2) This special decanter has its lamella plates placed very close to each other. Then the rising distance is reduced dramatically. For the spiraloil, the plates are rolled into cylindrical shape, so it is very compact in size. However, it is difficult to determine the rising distance (H). GPI lab has been studies the model that governs this type of decanter and found that the modified theoretical model, as shown in eq. 4.3.3, can provide good result. This model is developed from the general model (eq. 4.3.2) by neglecting the complicated analysis to define H, and using the concept of decanting area (Sd) instead. Sd is calculated from the sum of every surface area within the decanter that can intercept oil without considering if the values H of these areas are identical or not. This model or equation has been confirmed by CHERID [4] for its accuracy, as shown in fig. 4.3.2-3.

2/118⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⋅Δ

=d

cc Sg

Qd

ρμ {4.3.3}

Solid core, radius = r

R

H

e

Annular plates

a) b)

187

Chapter 4 Decanting

III-71

Figure. 4.3.1-2 “Spiraloil” decanter a) Simple spiral b) Mixed spiral

0.00%10.00%20.00%30.00%40.00%50.00%60.00%70.00%80.00%90.00%

100.00%

0 10 20 30 40 50 6


Rem

oval

eff

icie

ncy

(%)

0

1' 1 3'

2 2'

3

V = 1.6 cm/s (3, 3')

V = 0.8 cm/s (2, 2')V = 0.4 cm/s (1, 1')

Fig. 4.3.1-3a Comparison between observed efficiency (1',2',3') and predicted efficiency (1,2,3) for Simple Spiral "Spiraloil" decanter

0.00%10.00%20.00%30.00%40.00%50.00%60.00%70.00%80.00%90.00%

100.00%

0 10 20 30 40 50 6


Rem

oval

eff

icie

ncy

(%)

0

1 (V = 0.5 cm/s)1'

2' 2 (V = 1.5 cm/s)

Fig. 4.3.1-3 Comparison between observed efficiency (1',2') and predicted efficiency (1,2) for Mixed Spiral "Spiraloil"

decanter

From the graphs, it shows that;

1. Predicted values for cut size are relatively accurate.

2. Predicted efficiencies of the droplets, smaller than cut size, are always lower than observed value. This can be explained by the phenomena taking place within the decanter, which will be described hereby.

3. Correction factor for efficiency prediction of these small droplets may not be established accurately. However, the predicted cut size can be used to design the

188


III-72

tank with relatively high accuracy. So it is reasonable to select the cut size to cover the majority of droplet size distribution. The predicted efficiency of smaller droplets, which is the minority part, will cause no harm but slightly underestimation on the total efficiency.

Because of its very close spacing, This type of decanter provides the separation both from decanting phenomenon and coalescing phenomenon. So the oil drops from the decanter is relatively big (> 2 mm), and easy to separate at the downstream end of the decanter.


Phenomena taking place in the decanter have also been studied by CHERID. It shows that, besides the decanting phenomenon, coalescing of oil drop is also taking place. From the study, it shows that there are 3 parameters that affect the decanting and coalescing phenomenon in the plate inserted decanter, i.e.,

1. Wettability of inserted plates

• Oleophilic plates: The oil will adhere and then coalesce to form a film at the surface of the plates. However, when the film is totally coated the plate surface, it will be more difficult for the following oil drops to coalesce to the film than adhere to the surface. These following oil drops, sometimes, roll along the film, rather than coalesce into it, along with the wastewater flow. However, at the end of plates, the oil film, driven by the hydraulic force of the wastewater, will form big oil drops. It can be said that these points will play the roles of the drip points or salting-out points, the same as in the coalescer. These big oil drops can intercept those non-coalesced oil drops to form even bigger drops until they reach certain size that can be snapped off by the water flow.

• Hydrophilic plates: The oil will not form a film on the plate surface. But they will accumulate as an oil droplet and coalesce with other droplets to form big oil drops. Then they will roll along with the wastewater flow and coalesce with other drops along their ways like snowballs. These big oil drops will leave the plate for the surface when they reach sufficient size.

Some following oil drops do not coalesce

with film but roll along the film

Oleophilic plate

Oil film

Drip point

Hydrophilic plate

Drip point

Snowball effect

a) Oleophilic plates b) Hydrophilic plates

Fig. 4.3.1-4 Effect of wettability of inserted plates

2. Inclination of inserted plates

189

Chapter 4 Decanting

III-73

In many designs, the plates are tilted for the designers believe that it will help increasing the oil separation efficiency. CHERID has studied the effect of inclination. From theory and research result, the influent of the inclination to the decanter can be concluded and described below:

• Effect on decanting phenomena: An inclined decanting channel is compared with a horizontal channel of the same distance between 2 adjacent plates (fig. 4.3.1-5). From the figure, it shows that the oil drops in the inclined decanter will take more time to rise to the intercepting surface since rising distance is greater (H/cos θ, instead of H). So, theoretically, inclination will not help promoting decanting phenomena of oil drop. However, in large-scale tank, inclination helps draining the sludge, that is always present in wastewater, thus, preventing clogging of the decanter.

• Effect on coalescing phenomena: CHERID found that the inclination of plate affects the coalescing phenomena. The decanted oil drops in inclined decanter may not coalesce into oil film at the surface of the plates, even the olephilic ones, but roll along with the water flow. Moreover, the inclination will make the film snap off easier. Thus it will reduce the thickness of film accumulating at the surface of plates. For big tank, it may not affect the efficiency much. But in plate-closely-inserted decanter, like spiraloil, this film helps reducing the rising distance in considerable degree. So, from these two reasons, CHERID proposes that, for plate-closely-inserted decanter, inclination should be zero. Or the decanter should be installed horizontally.

=V( Flow velocity)

U (Rising velocity)H

VU

H

θ

Rising distance =(H/ cos θ) > Η

a) Horizontal decanter b) Inclined decanter

Fig. 4.3.1-5 Schematic of inclined and horizontal decanter

3. Presence of surfactants The presence of surfactant will decrease the sizes of decanted oil drops, thus

hinder good decanting phenomenon. Furthermore, it will prevent coalescing between oil drops. As a result, the total efficiency of decanter will be decreased. It has been reported that [4], when tested with kerosene-water emulsion at the concentration of 100 mg/l with the concentration of surfactants around 100 mg/l and the average droplet size of 40 microns, the efficiency of spiraloil will decrease about 15 – 20%. The interfacial tension between kerosene and water will drop from 42 N.m to 8.5 – 32.1 N.m, depending on the type of the surfactants. Non ionic surfactant is the most effective surfactant for lowering the interfacial tension. The second best is cationic surfactant, then anionic surfactant, respectively.


190


III-74

Design procedure for compact decanter is relatively similar to the simple decanter’s and based upon the equations shown in the previous section.


The cut size can be determined from the degree of treatment required as well as from the limitation of the decanting processes. Cut size determination from degree of treatment is described in chapter 3. For the limitations of the decanting process, they come from 2 main parts, i.e., theoretical and economic limitations, as stated in section 4.2.2.

2. Decanter sizing

The size of the decanter can be determined, based on the equation in section 4.3.1, as follows:

Decanting area (Sd), inserted plate area (Sp), footprint area (S): The decanter area of the compact decanter depends on the types or configuration of the decanter. However the decanting area (Sd) in this case is not equal to footprint area (S).

Parallel plate interceptor, Corrugated plate separator or the decanter of identical concept

⎟⎟⎠

⎞⎜⎜⎝

⎛=

dcd U

QS {4.3.4a}

c

cdc

dgU

μρ18

2⋅⋅Δ= {4.3.4b}

)1( += NSS pd {4.3.4c}

pSS = {4.3.4d}

N represents the number of plates. Please note that the plates must be inserted at an evenly interval throughout the water depth.

General compact decanter that rising distance (H) can not be exactly specified.

⎟⎟⎠

⎞⎜⎜⎝

⎛=

dcd U

QS {4.3.5a}

c

cdc

dgU

μρ18

2⋅⋅Δ= {4.3.5b}

In this case, the footprint area has to be specially calculated, based on the geometry of the decanter. Sd can be specifically calculated, depending on the shape of the decanter. Surface of material or free water surface that can intercept oil drops will be accounted for calculating the decanting area (Sd). In case of commercial decanter, required Sd calculated from eq. 4.3.5a will be used to select the model and/or the number of the product required for the design flowrate.

Flow velocity (V): Flow velocity means the velocity of total wastewater along the decanter, determined from wastewater flowrate and full cross sectional area of the decanter. Velocity is the main parameter that governs flow regime and hydraulic condition,

191

Chapter 4 Decanting

III-75

such as hydraulic force, head loss, etc., in the decanter. CHERID [4] recommends the value of 0.4-1.6 cm/s or 14.4-54 m/h., 2.6 to 10.6 times higher than simple decanter.


To determine the removal efficiency, the graded efficiency (ηd) will be calculated first by eq. 4.3.6a and b. For graded outlet concentration (Cd), flow splitting effect between treated water outlet port and separated oil outlet flow must be taken into account, as shown in eq. 4.2.7c and 4.2.7d. Then, the total removal efficiency (ηt) can be determined from eq. 4.3.6c.


%100=dη {4.3.6a}


%100⋅=dc

dd U

Uη {4.3.6b}


( ) %1001 max

min

⋅⋅⋅= ∑d

dodd

ot C

Cηη {4.3.6c}

ηt = Total Removal efficiency of the tank Cod = Inlet concentration of the droplet diameter “d” Qout = Treated water outlet flow (eq. 4.2.7d)

5. Energy required

The compact decanter does not require extra energy to make it function. The energy is required only to feed the water through the decanter. Pressure drop across the decanter and piping system depends on decanter and piping design. This pressure drop can be calculated by general hydraulic equations, such as Darcy-Weisbach’s, Manning’s, or Hazen-William’s equation, thus will not de described here. However, velocity through decanter is relatively low, compared to velocity in pipe (0.6-2.5 m/s). So the estimated value of pressure drop across the decanter around 0.5-1.0 m. is normally acceptable, regardless of its size and configuration.



The equations described above are developed from the following assumptions and limitations. Thus, it is necessary to ensure that these assumptions are valid when design your decanter.

1) The tank is operated under laminar flow regime. Reynolds number, Re, is between 10-4 to 1, which is the range that STOKES law is valid.

2) The oil droplets are uniformly distributed across the cross section area of the tank, which can be achieved by the proper design of inlet chamber.

192


III-76

3) The oil droplet is spherical, which is normally true.

4) For droplets smaller than 20 microns, they are subjected to Brownian motion and cause error in the prediction of the efficiency. So it is recommended to avoid using the decanter for the wastewater with the majority part of oil droplets smaller than 20 microns. However, if these small droplets are the minority part of pollutants, the models can be used to predict the efficiency without any harm because its prediction is usually lower than observed value, thus make the prediction result on the safe side.

2. Material selection

There is no evidence that provides strongly recommendation or rule to select the decanter material, esp. its inserted plates. CHERID has studied the effect of hydrophilic and oleophilic material but did not point out which one is suitable. Moreover, in his spiraloil decanter, he used the combination of corrugated oleophilic plates with smooth hydrophilic insertions. However, theoretically, olephilic material may provide more benefit for it promotes formation of oil film on its surface, which helps reducing the rising distance, as described in section 4.3.1.3. Oleophilic property can be achieved by selecting the material with very low surface energy, such as PTFE, Teflon. It can also be archived by coating the surface of any material by oleophilic coating, such as silicone. In the latter case, we can select base material that provides good machinability and cost-effectiveness, regardless of its wettability, and then have it coated later.

4.3.4 Variations, advantage and disadvantage of compact decanters

There are many variations of compact decanters. However, the components of the decanter can be designed by the same concept as that of the simple decanter. Nowadays, there are several suppliers that commercialize these products, such as ELF, Johnson lamella separator, etc. So designers can find further information from these suppliers. Some examples of these decanters are shown in fig. 4.3.4-1

Advantages: Main advantage of compact decanter is, of course, its compactness. It requires 2.6 –10.6 times (or more) of footprint area less than simple decanter at the same capacity. The concept of plate insertion can be used to upgrade the existing simple tank without constructing the new tank.

Disadvantages: Because of its relatively small flow passage, esp. for closely inserted decanter, it may have a chance of clogging by suspended solids. So this fact should been taken into account before design or select the decanter.

193

Chapter 4 Decanting

III-77

a) Spiraloil test (Source: GPI lab) b) Spiraloil installation on off-shore platform

(Source: ELF, GPI lab)

c) Graphical drawing of parallel plate

interceptor (Source: Anch Tank) d) Parallel plate interceptor aboveground

installation (Source: Anch Tank)

e) PPI in form of cylindrical tank

(Source: Knitmesh) f) Parallel plate module for cylindrical

tank(Source: Knitmesh)

g) Parallel plate module

(Source: Industrial heating.com) h) Corrogated plate module

(Source: Brentwood)

Fig. 4.3.4-1 Examples of compact decanters

194

Chapter 5 Coalescer

III-78

Chapter 5 Coalescer

5.1 General

From STOKES law, particle or oil droplet diameter is the most influent parameter because it is powered by 2. Thus, if we have a process that can increase the size of the oil droplet 2 times, the rising velocity will be 4 time higher. This can be achieved by making the oil drops combine or coalesce to each other. Coalescer is the process that is designed to promote coalescing of oil droplets into bigger oil drops, which can be separated easily by relatively small decanter. Because of its working principle involves the modification of parameters of STOKES law. So it is one of accelerated or adapted separation process, along with other processes such as hydrocyclone and flotation. There are several types and modification of coalescers. However, they can be categorized into 2 groups, based on the type of the media used, i.e.,

• Granular bed coalescer: This group of coalescer uses granular material, such as resin, sand or glass beads, as a coalescer bed.

• Fibrous bed coalescer: This group of coalescer uses fibrous material, such as metal wool or plastic brush, as a coalescer bed.

The pictures of various coalescers are shown in fig. 5.1-1.

5.2 Granular bed coalescer


This type of coalescer uses granular material as a bed to promote coalescing between oil droplets. The concept of the process is that the oily wastewater will flow through the coalescer bed, as shown in fig. 5.2.1-1. Oil droplets in the wastewater will undergo several steps to coalesce them into big oil drops. Then, these big oil drops will be separated by a small integrated decanter at the downstream of the bed.

Professor AURELLE [3] is the pioneer on coalescer research. He establishes the 3-step working mechanisms of the coalescer and the model for efficiency prediction, based on YAO’s filter model [35]. Then there are several follow-up researches on coalescer , based on AURELLE’s research

To understand the working principles of coalescer, the 3-step working mechanisms of the coalescer ise described hereby.

195


III-79

a) Graphical image of Granular bed coalescer (Source: Elf, GPI lab)

b) Graphical image of one type of fibrous bed coalescer (Source: Knit mesh)

c) Granular bed coalescer installation for

refinery’s condensate treatment (2x125m3/h)(Source: Donges refineries, Elf)

d) Photo of coalesced oil drop of dynamic (rotating) fibrous bed coalescer

(Source: GPI lab)

Fig. 5.1-1 Examples of coalescer

Collector size =dp,

Void ratio = ε

V

H

IN:Micro drop

Dia. = d


screen


V

Fig. 5.2.1.1 Schematic diagram of granular bed coalescer

196

Chapter 5 Coalescer

III-80

Mechanisms taking place inside the coalescer bed

AURELLE divided the mechanisms taking place with in coalescer bed into 3 steps, i.e.;

• Step 1: Interception, which consists of 3 major transport phenomena, i.e., sedimentation, direct interception and diffusion.

• Step 2: Adhesion-Coalescence. The efficiency of this step depends mainly on wettability of bed material.

• Step 3: Salting out or enlargement of coalesced liquid. This step depends on 4 parameters, such as wettability of bed material at the discharge surface, empty bed flow velocity, interfacial tension and ratio of dispersed phase and continuous phase in emulsion treated.

Details of each step is described hereby.

5.2.1.1 Step 1: Interception

This step is the step of transportation of the oil droplets through the bed of coalescer. AURELLE considered that this step is identical to the mechanism of in-depth filtration. So he applied the phenomena taking place in the filter to the colescer. This step will be divided into 3 transport phenomena, i.e., sedimentation, direct interception and diffusion. Other transport phenomena, such as that of electrical force, might take place, but it is proven that their effect is small, thus, negligible. To simplify the model, we will consider the phenomena, taking place at 1 collector or 1 piece of media. Schematic diagram of the 3 phenomena is shown in fig. 5.2.1-1.

U= Rising velocityV= Flow velocity

V UVU

Oil droplet

Stream line Media, Collector

a) Direct interception b) Sedimentation c) Diffusion

Fig. 5.2.1-2 Schematic diagram of the 3 transport phenomena

1. Direct interception

When oil drops of diameter “d” flow along with the streamline, the oil drops that pass within the distance less than d/2 from the media will be intercepted by the media, as shown in fig. 5.2.1-1a. The efficiency factor for this phenomenon can be calculated by eq. 5.2.1. dp is the diameter of collector or media particle.

197


III-81

2)(23

dpd

Int =η {5.2.1}

2. Sedimentation

Consider oil drops of diameter “d” flow along with the streamline. Because of its density, the droplets will be subjected to rising velocity (U), calculated by STOKES law. When they are far from the media, the rising velocity and flow velocity (V) will have the same direction. When they come close to the media, the flow velocity will deviate, as shown by the streamline, while the oil drops will be subjected to both flow velocity and their own rising velocity. So the resultant velocity will not totally conform to streamline. And in some cases, it will make the oil drops collide to, thus, sediment on the media. The efficiency factor for this phenomenon can be calculated by eq. 5.2.2.

Vgd

VU

csed μ

ρη

18

2Δ== {5.2.2}

3. Diffusion

For very small droplets (d < 5 microns), They will be subjected to Brownian’s motion. These random motions can cause the droplets to collide to the media. The efficiency factor for this phenomenon can be calculated by eq. 5.2.3. K , in this equation, represents the Bolzmann constant and T represents the absolute temperature.

3/29.0 ⎟⎟

⎠

⎞⎜⎜⎝

⎛

⋅=

dpVdKT

cDiff μ

η {5.2.3}

4. Combined efficiency of step1 : Interception

From steps 1 to 3, we can calculate the efficiency factors of each transport phenomena for single collector. Theoretical efficiency factor of interception step for single collector is the summation of the efficiency factors of those three transport phenomena, as shown in eq. 5.2.4. From the equation, it shows that, at the same operation condition, the theoretical efficiency factor of the single collector will vary with the droplet diameter, as shown in fig. 5.2.1-2.

DiffSedInttheo ηηηη ++=

Thus

3/22

29.0)(

23

18 ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

++Δ

=dpVd

KTdpd

Vgd

cctheo μμ

ρη {5.2.4}

198

Chapter 5 Coalescer

III-82

1.00E-05

1.00E-04

1.00E-03

1.00E-02

1.00E-01

0.01 0.1 1 10 100


Effc

iecn

cy fa

ctor

Diffusion efficiency factor Interception efficiency factorSedimentation efficiency factor Theoretical efficiency factor

Fig. 5.2.1-3 Relation between oil droplet diameter and efficiency factors of each transportation phenomena

From the graph, the range where the theoretical efficiency factors are minimal is between 0.25 to 5 microns. So the droplets in this range is theoretically the most difficult to separate.

To adapt the efficiency of a single collector to the entire coalescer bed, we will consider the simplified diagram of single spherical collector, placed in laminar flow regime, as shown in fig. 5.2.1-3a. V represents the flow velocity. The fraction of wastewater flowing pass the single collector will be the flow that passes through the projected area of the collector (q), as shown in eq. 5.2.5a. Then, some oil drops in this fraction of the wastewater will be intercepted by the collector. The quantity of intercepted oil drops of the single collector (c’) will be calculated from the theoretical efficiency factor, as shown in eq. 5.2.5b.

Vdq p ⋅= 24π {5.2.5a}

CVdc ptheo ⋅⋅⋅= 24

' πη {5.2.5b}

199


III-83

q = Vπdp2/4

V

dp

Collector size =dp,Void ratio = ε

V

H

IN:Micro drop

Dia. = d


screen


V

a) b)

Fig. 5.2.1-4 Schematic of a single collector and the entire bed of coalescer

C is the inlet concentration of oily wastewater. For the entire coalescer bed, we will consider a very small slice of bed of the height dH (see fig. 5.2.1-3b). The number of collector particles in this slice can be calculated from the cross sectional area of bed (Ao), the size of collector (dp) and the void ratio of the bed (ε), as shown in eq. 5.2.5c.

Then, the total concentration of intercepted oil for this slice of bed will be equal to the product of c’ and the number of collector particles. However, not all of the intercepted oil drops will adhere to the collector. So the probability coefficient (α) will be applied to adapt the quantity of intercepted oil drops to the quantity of adhered oil drops (c”), as shown in eq. 5.2.5d.

The number of collector particles in the slice dH 3

6

)1(

p

o

d

AdHπ

ε−= {5.2.5c}

32

6

)1(4

"p

optheo

d

AdHCVdc

πεπ

ηα−

⋅⋅⋅⋅⋅= , α < 1 {5.2.5d}

If dC represents the concentration of oil reduced after passing through the bed dH, then we have got eq. 5.2.5e and f;

"cdCAV o =⋅⋅− {5.2.5e}

32

6

)1(4

p

optheoo

d

AdHCVddCAV

πεπ

ηα−

⋅⋅⋅⋅⋅=⋅⋅− {5.2.5f}

Therefore,

dHdC

dCtheo

pαηε )1(

23

−−= {5.2.5g}

Integration of eq. 5.2.5g will give the value of the oil concentration reduced by the entire bed, as shown in eq. 5.2.6.

200

Chapter 5 Coalescer

III-84

ptheo

o dH

CC

αηε )1(23

)log( −−= {5.2.6}

Thus, the theoretical removal efficiency of the coalescer, based on step 1: “Interception”, can be written as shown in eq. 5.2.7.

%1001%1001)()1(

23

, ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−=⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛−=

⋅−− theodpH

otheod e

CC ηαε

η {5.2.7}

5.2.1.2 Step 2: Adhesion and coalescence

After being intercepted from step 1, the oil droplets will be separated from the wastewater stream if they can adhere to the surface of the collectors. After that, in good coalescers, the adhering oil drops will coalesce to each other and flow separately from the water stream, as oil stream within the bed. Phenomena taking place in this step is shown in fig. 5.2.1-4. So the mechanism in this step depends mainly on the wettability of the bed. AURELLE had tested the effect of wettability, using hydrophilic and olephilic material as coalescer bed for secondary o/w emulsion treatment. The result shows that:

• For Oleophilic material: The oil droplets will adhere to the collector surfaces then form the oil film around the collectors. The oil films will accumulate in the bed until it reaches saturate level at certain height of the bed, called “critical height” (Hc). After that the oil will start to flow as a separated oil stream, called “channeling”.

• For Hydrophilic material: The oil droplets will be trapped in the void between collectors. These trapped oil droplets will, then, be play the role of collectors by their own to intercept the following oil droplets. Then they will coalesce into bigger drops and flow through the void with the wastewater stream.

Even though both materials can cause coalescence, from test result, the oleophilic material yields good efficiency up to higher range of flow velocity. It means that oleophilic bed coalescer can be used at higher loading rates, thus makes it more compact in size. So it is recommended to use oleophilic material as the bed for direct emulsion treatment.

5.2.1.3 Step 3: “Salting out” or enlargement of coalesced oil

The coalesced oil that flows through the bed in the manner of channeling will finally reach the topmost of the bed. Then it will leave the bed for the water surface. In good coalescer, the coalesced oil will leave the bed in the form of big oil drops (diameter 2-3 mm or more). Characteristic or size of the oil drops that leave the bed depends on the phenomenon, taking place at the top surface of the bed. AURELLE has studied this phenomenon, comparing between oleophilic material and hydrophilic material. The result shows that:

201


III-85

a) Oil drops adhere to collectors b) Several oil drops coalesce to form oil film or stream, separating from water stream

Oleophilic bed

a) Oil drops are intercepted or lodge between interceptor. Note that contact angles > 90o.

b) Intercepted oil drops coalesce to form bigger drops in the void between collectors

Hydrophilic bed

Fig. 5.2.1-5 Phenomena in step 2: Adhesion-Coalescence

• For Oleophilic material: The coalesced oil will flow along its preferable “channels” out of the bed, called “drip points”. At the drip point, the oil will adhere and cover the top surface of the material as a film. To flow out of the bed, the water has to flow inevitably through the covering oil film, then forms the “oil mousse”, as shown in fig. 5.2.1-5. The separated oil will contain high content of water. The water flow will also cause “re-fragmentation” of the oil into small oil droplets again when the mousses rapture, esp. in case that the interfacial tension is low. Thus, the efficiency is decreased.

202

Chapter 5 Coalescer

III-86

a) “Oil mousse” forming. Note that contact angle < 90o.

b) Re-fragmentation of oil when the mousse raptures

Oleophilic “salting out” surface

c) Formation of big oil drops. Note that contact angle > 90o.

d) Oil jet and re-fragmentation of oil at high velocity or high oil concentration

Hydrophilic “salting out” surface

Fig. 5.2.1-6 Phenomenon in step 3: “Salting out” or enlargemant of coalesced oil

• For Hydrophilic material: The drip points will appear as well. But the oil will not form the film over the collector surface. But the film will locate between the collectors and will grow to big oil drops and, then, be snapped off as big oil drops, without forming oil mousse. However, at high flow velocity or high concentration of oil, hydraulic force from water flowing out of the bed will be high. The oil will be snapped off as a jet of oil and then, from the high hydraulic force, will be “re-fragmented” to small droplets.

203


III-87

From the result described above, it shows that efficiency of this step depends on 4 parameters, i.e., wettability of discharge surface, flow velocity or empty bed velocity, interfacial tension, and concentration or ratio of oil to water. To optimize the efficiency of this step, AURELLE suggestd the following methods:

• Design the coalescer at the proper velocity to avoid the re-fragmentation effect.

• Use hydrophilic material as the top layer of the bed or use top grill of hydrophilic material, to avoid mousse formation. At the same removal efficiency, it is proven that the bed with hydrophilic material on the top layer can be operated at higher velocity than that with the oleophilic top layer.

5.2.1.4 Comparison between observed efficiency and theoretical efficiency

The 3 steps mechanisms, proposed by AURELLE, can effectively describe the phenomena taking place within the coalescer bed. However, to apply it for efficiency prediction, we have to use many assumptions to simplify them. The most important assumption is that the efficiency mechanisms in steps 2 and 3 are optimized, so their efficiency is 100%. It means all of oil that has been intercepted and adhered to collectors in step 1 can be separated. So the removal efficiency of coalescer can be calculated by eq. 5.2.4 and 5.2.7.

AURELLE had studies the relation between observed efficiency (α.ηexp) and the theoretical efficiency and found that the observed values are quite different from the calculated value in complex fashion, as shown in fig. 5.2.1-6. Relation between observed efficiency and theoretical efficiency, suggested by AURELLE, is shown in eq. 5.2.8a. So the theoretical efficiency in eq.5.2.7 can be modified by replacing αηtheo with α.ηexp. The efficiency of granular bed coalescer, then, can be calculated by eq.5.2.8b.

5143.0exp )(5484.0 theoηηα =⋅ {5.2.8a}

%1001%1001)()1(

23

exp

⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−=⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛−=

⋅−− ηαεη dp

H

od e

CC

{5.2.8b}

AURELLE’s model is developed under these following assumptions and conditions:

• The key assumption of this model is that mechanisms in step 2 and step 3 of the coalescer are optimized.

• The shape of the collector is relatively spherical. The size of the collector tested (dp) is between 0.2 – 1 mm.

• The collector shall be wetted by dispersed phase. In case of direct (oil in water) emulsion, the collector, then, shall be oleophilic. For inverse emulsion, oleophilic resin is recommended.

• Range of empty bed velocity (V) shall be not greater than 0.35 cm/s (12.6 m/h)

204

Chapter 5 Coalescer

III-88

• Density difference between dispersed phase and continuous phase (Δρ) is approximately 200 kg/m3.

• The model is developed for inlet oil concentration between 100-200 mg/l.

• At velocity < 0.35 cm/s, the efficiency of the coalescer is independent of velocity. Beyond this range, The efficiency will decrease when the velocity increases. The rate of efficiency decreasing varies with size, wettability and surface roughness of collector.

• If the result from eq. 5.2.8b is greater than 100%, then it will be rounded up to 100%.

1.00E-04

1.00E-03

1.00E-02

1.00E-01

1.00E+00

1.00E-041.00E-031.00E-021.00E-011.00E+00

Theoretical efficiency factor

Exp

erim

enta

l eff

icie

ncy

fact

or

Fig. 5.2.1-7 Relation between experimental (or observed) efficiency factor and theoretical efficiency factor

5.2.1.5 Empirical model of granular bed coalescer

From fig. 5.2.1-6 and eq. 5.2.8, the relation is not in linear fashion, then it can be interpreted that some assumptions in theoretical model are violated or not exactly true. Many researchers had tried to fine-tune the assumptions. HAPPEL [44] had changed the flowrate corresponding to single collector (eq. 5.2.5b) by replacing the value “dp” with an imaginary diameter called “b” to account for the void area between collectors. The value of b is calculated from the cross sectional area of bed (Ao) and the number of collectors by eq. 5.2.9.

oAbN =24π

{5.2.9a}

24

)1(

p

o

d

AN

πε−⋅

= {5.2.9b}

205


III-89

DAMAK [9] used HAPPEL’s theory to adapt AURELLE’s theoretical efficiency equation, however, the accuracy is not improved. So it seems that the mechanism taking place in the coalescer bed is so complex that it cannot be simplified to develop an accurate theoretical model. Thus, when the exact theoretical based model can not be found. DAMAK proposed an empirical equation for coalescer efficiency prediction, based on the dimensional analysis, as shown in eq. 5.2.10. Typical characteristic curve between efficiency and droplet size is shown in fig.5.2.1-7.

%100)()()()()(58.0 09.009.008.0

/

12.02.0 ⋅⎟⎟⎠

⎞⎜⎜⎝

⎛ Δ= −

cc

d

wo

cd

dpVdpH

dpd

ρρ

μμ

γρ

η {5.2.10}

The model is developed under these conditions i.e.;

• The model is tested at the range of dp from 0.36 to 0.94 mm. and interfacial tension (γow) of 11 to 42 dyne/cm. (T.I.O.A, Heptane, Anisole, Toluene and Kerosene)

• The velocity (V) tested is in the range of 0.09 to 0.54 cm/s.

• The density difference between dispersed phase (oil) and continuous phase (water) (Δρ) is between 83 to 314 kg/m3.

• The bed used is spherical glass bead with silicon coated to achieve oleophilicity.

• The inlet concentration of hydrocarbon tested is around 1,000 mg/l.

• If the result from eq. 5.2.10 is greater than 100%, it will be rounded up to 100%.

Verification result, using AURELLE’s experimental data, confirms that DAMAK’s model can predict the removal efficiency with only ±10% error. It implies that the model is also valid at the operating condition tested by AURELLE. From these test conditions, it seems that this empirical model covers the range of the oily wastewater commonly found in real situation. Thus, it is recommended to use this model (eq.5.2.10) for granular bed coalescer calculation.


Main parameters that affect the efficiency of the granular bed coalescer include:

1. Bed height (H)

Typical relation between the bed height and the removal efficiency is as shown in fig. 5.2.1-7. From the graph, the efficiency of coalescer will increase with increasing height, then it will stay relatively constant. This height is called “critical height (Hc)”. The occurrence of critical height can be described by phenomena in step 2 : adhesion- coalescence, as described before in section 5.2.1.2. If the bed height is shorter than the critical height, it can be said that there is not enough accumulated oil film to trap the oil droplets and provide continuous coalesced oil stream to form perfect “channeling”. Hc depends on wettability, roughness and size of media. From eq. 5.2.10, it shows that the efficiency is proportional to H0.12

206

Chapter 5 Coalescer

III-90

Droplet size (d)

Efficiency (η)

100%

Lower limit ofthe model: 10 microns

Bed height (H)

Efficiency (η)

Increase V, dpMore hydrophilicLess roughness

Hc

Fig. 5.2.1-8 Typical relation between efficiency of granular bed coalescer and various parameters

2. Size of collector or media particle (dp)

The effect of the size of media is shown in fig. 5.2.1-7. Under the same operating condition, the smaller the collector size, the better the efficiency. From eq. 5.2.10, it shows that the efficiency is proportional to dp

-0.4

3. Flow velocity or empty bed velocity (V)

The effect of velocity is relatively small, compared to other parameter. From eq. 5.2.10, it shows that the efficiency is proportional to V-0.08

. However, this is true only within the valid range of the model. At higher velocity, mousse or jet formation will occur, thus the efficiency will drop rapidly and no longer conform to eq. 5.2.10. The velocity that the mousse or jet starts appearing is called “critical velocity”. It is recommended to use velocity not more than 0.54 cm/s to avoid mousse and jet formation.

4. Wettability of bed material

The effect of wettabiblity of media is shown in fig. 5.2.1-7. It is recommended to use oleophilic material as the coalescer bed with the thin top layer of hydrophilic material as drip point surface.

5. Ratio of oil to water

Even though this parameter is not included in eq. 5.2.10, but it is an important parameter that limit the working range of coalescer. In fact, the model in eq. 5.2.10 is valid for oil concentration not greater than 1,000 mg/l. Higher oil concentration will result in mousse or jet forming, which will greatly lower the efficiency of the coalescer. To expand the working range of granular bed coalescer, it is necessary to modify basic granular bed coalescer by additional installation of oil guide, which will be described in the following section.

207


III-91

6. Interfacial tension and presence of surfactants

Surfactant will lower the interfacial tension, then make the oil droplets more stabilized. This leads to poor adhesion between droplets and surface, ineffective collision (collision without coalescence), and re-fragmentation of oil drops. So the presence of surfactant directly limits the efficiency of step 2 “adhesion-coalescence”. To optimize the efficiency of step 2, it is recommended to destabilize (or breaking or cracking) the emulsion before sending to the coalescer.

7. Temperature

Properties of oil and water, such as viscosity, change with temperature. So temperature is inevitable an important parameter that affects coalescer efficiency. Then it is important to design the process by taking the possible range of temperature into account. Normally, the efficiency will, more or less, increase with the temperature. The equations in this section are developed under the temperature range between 15 to 25oC, which is the general practical operating range.

8. Surface roughness of bed material

Surface roughness affects adhesion of oil on the surface of the solid, as shown in chapter 2. The roughness makes the wettablity of the surface more eminent. So oleophilic material will show more oleophilicity if its surface is rough. Contact angle of oil on the material surface will be lower. In this case, it helps promoting direct emulsion separation.

5.2.1.7 Pressure drop

Pressure drop across coalescer bed (p), in metre, can be calculated by Kozeny-Carman’s equation (eq. 5.2.11) (use SI unit, e.g. m, kg, second).

32

2)1(180

ερ

εμ

⋅⋅⋅

−=

dpg

VHp c m {5.2.11}

All variables except porosity (ε) will be determined by designer. For the porosity of coalescer bed, from many researches [3], [26], [27], it shows that bed porosity varies with bed depth and can be divided into 2 zones, i.e.,

• Lower zone or critical zone: This zone represents an effective zone of coalescer bed. The maximum height of this zone is called “critical height (Hc)”. When bed height is greater than the critical height, the efficiency will increase only slowly (from eq. 5.2.10: η ∝ H 0.12). In this zone, the bed will be soaked with oil so the effective porosity will be low.

• Upper zone: If the bed is higher than Hc, practically, all of oil will be trapped in critical zone. Then in the higher zone, there will be enough oil in lower zone to flow continuously through the bed in form of “channeling”. So the effective porosity in this zone will be lower than critical zone.

208

Chapter 5 Coalescer

III-92

From data of various researches [3], [26], [27], it can be concluded that the pressure drop of granular bed can be calculated by Kozeny-Carman’s equation, using the following recommendations, i.e.,

• When Hc is known (from literatures, etc.), pressure drop lower and upper part of bed can be calculated separately, using eq. 5.2.11 and corresponding H of each zone. For example, if Hc is 0.1 m. and total bed height is 0.15 m, H for lower zone will be equal to Hc because Hc is lower the total height. And H for the upper zone will be 0.15 - 0.1 = 0.05 m.

• Recommended porosity for the lower (critical) part of bed (H<Hc) is between 0.14 to 0.19.

• Recommended porosity for the upper part of bed (H>Hc) is between 0.23 to 0.30. • If it is certain that H design < Hc, use single step calculation with ε = 0.14 - 0.19. • However, Hc is usually unknown, then it is recommended to use the single step

calculation with ε = 0.13 and 0.23 to estimate minimum and maximum pressure drop, respectively.

• The value of ε described above can be used for the range of dp from 0.20 to 1.0 mm.


Design procedure for granular bed coalescer is based upon the equations, shown in the previous section. To design the coalescer, the required cut size will be determined first. After that, the size of the coalescer can be calculated. Then, graded efficiency (efficiency of each size of droplet) and, hence, the total removal efficiency can be determined. Calculation procedure for each step is described below.


The cut size can be determined from the degree of treatment required as well as from the limitation of the coalescing processes. The cut size determination from degree of treatment is described in chapter 3. For the limitations of the process, it is mainly model limitation. Since the model used for calculation is based on empirical data, it can be applied only within its valid range. Extrapolation of model may cause unpredictable errors. The limitation of model will be described in section 5.2.3. It should be noted that the cut size must be greater than 10 microns since it is the lower limit of the model. For smaller droplets, the efficiency will be very low and unpredictable.

2. Coalescer sizing

The size of the granular bed coalescer can be determined, based on the equation in section 5.2.1. The main equation for coalescer designed will be based on the empirical model, as shown is eq. 5.2.10. Design cut size will be used to calculate the dimension of the coalescer by assuming that the graded efficiency at the cut size is 100%.

%100)()()()()(58.0 09.009.008.012.02.0 ⋅⎟⎟⎠

⎞⎜⎜⎝

⎛ Δ= −

cc

d

ow

cd

dpVdpH

dpd

ρρ

μμ

γρ

η {5.2.10}

209


III-93

Dimension of the coalescer can be arbitrarily selected under the limitations of the model, as shown in section. 5.2.3, to make eq. 5.2.10 consistent. However, it is recommended to select the lowest possible media diameter and highest velocity for the first trial because it will give the most compact diameter of the coalescer. If the result is acceptable, it can be fine-adjusted to get the most suitable dimension. If the result is unacceptable, normally too big, it may be necessary to increase the number of units.

Fig. 5.2.2-1 shows the result of calculation from eq. 5.2.10 at various cut sizes. The calculation is based on kerosene-water emulsion, which may be used as a guideline for coalescer size estimation. The collector diameter, used in the graph, is 0.35 mm.

00.10.20.30.40.50.60.70.80.9

11.11.21.31.4

0 10 20 30 40 50 60

Cut size (micron)

Bed

hei

ght (

m)

V = 1.94 m/hV = 3.2 4m/hV = 11.3 m/h

Fig 5.2.2-1 Relation between cut size and coalescer dimension


To determine the removal efficiency, the graded efficiency (ηd) will be calculated first by eq. 5.2.10. If the result from eq. 5.2.10 is greater than 100%, then it will be rounded up to 100%.

It must be noted that the graded efficiency from the equation is not yet accounted for effect of flow splitting between water outlet flow and separated oil outlet flow. To calculate graded outlet oil concentration, the effect must be taken into account, as shown in eq. 5.2.12a.

oddout

d CQ

QC )1( η−= {5.2.12a}


d

d

dodd

out

CQQQ

ρ

η∑ ⋅−=

max

min

)( {5.2.12b}

210

Chapter 5 Coalescer

III-94


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {5.2.12c}


4. Pressure drop

Pressure drop (in metre) can be calculated from Kozeny-Carman’s equation (eq. 5.2.11). Recommended value of bed porosity (ε) and other criteria are as shown in section 5.2.1.7. For diluted wastewater, the values of μ and ρ can be replaced by those of water (μc, ρc).

32

2)1(180ερεμ⋅⋅⋅−

=dpg

VHp m {5.2.11}

5.2.3 Design consideration


The equations described above are developed from the following assumptions and limitations. Thus, it is necessary to make sure that these assumptions are valid when design your coalescer.

1) Coalescer bed shall be oleophilic and relatively spherical in shape.

2) Tested size of bed media (dp) is between 0.20 – 1.0 mm. The larger the media size, the lower the efficiency.

3) Tested range of bed height (H) of the model is between 1 to 10 cm. However, bed height as low as 1 cm is not recommended. The greater the bed height, the safer the coalescer operation. However, it also results in higher pressure drop.

4) The velocity (V) should be in the range of 0.09 to 0.54 cm/s or 3.2 to 19.4 m/h.

5) Tested interfacial tension (γow) is between 11 to 42 dyne/cm or 0.011 to 0.042 N/m. (such as, T.I.O.A, Heptane, Anisole, Toluene and Kerosene)

6) Different density between dispersed phase (oil) and continuous phase (water) (Δρ) is between 80 to 315 kg/m3 (approx.).

7) The equation is valid for droplet size (d) of 10 microns or bigger. For smaller droplets, result from eq. 5.2.10 may not be accurate because it is beyond the data that has been used to verify the model.

8) The model is valid for inlet concentration between 100 to 1,000 mg/l. At higher concentration, mousse or jet formation may occur, resulting in unpredictable decreasing of the efficiency.

211


III-95

2. Construction and material selection

Granular bed coalescer consists of 2 major components, i.e., casing and bed.

1) The casing

The casing is the component that contains the bed and other important appurtenances, such as wastewater inlet, wastewater outlet, oil outlet, clean out, etc. Researches in GPI lab normally emphasize on working mechanisms and developing of a model so there is no direct study on characteristic of the casing. However, because of the fact that the coalescer is usually designed as a pressure vessel in the same manner as a pressure filter, then, casing or tank construction criteria of the pressure filter can be readily applied to the coalescer. However, some details, such as oil outlet pipe, water outlet pipe and internal baffle should be adapted to ensure good separation between oil and water. As outlined in chapter 4 “Decanting”, the internal baffle should be installed in the manner that it can promote oil drop decanting, prevent short circuit of oil drops to the water outlet pipe. The pipe and baffle arrangement shown in fig. 5.1-1 can be used as a guideline for casing design.

2) The bed

From many GPI researches, it is recommended that resin is the most suitable material for coalescer bed, because:

• It is widely used in many industries, and available in every country. • It is produced in wide range of size and normally spherical. • It can be coated to achieve required wettability. • Its cost is competitive. • Its physical properties, such as hardness, etc., are also good.

Other materials that have been used as coalescer bed include glass beads, small stainless steel balls. Material that is heavier than water, such as stainless steel, has an advantage for it can be used without installation of top grille to prevent it from carry-over. So it can be cleaned by backwash process or scouring process because the material is not retained by top grille, then can be expanded or floated freely.

For wettability, oleophilic bed with thin layer of hydrophilic material on the top or with hydrophilic top grille is recommended (see section 5.2.1.2). Guideline on oleophilicity of materials is described in chapter 2. Some hydrophilic material that has good mechanical properties can acquire oleophilic property by coating with proper substance, such as silicone. Oleophilicity of bed may change with time from deterioration of the coating or reaction with wastewater. It may need re-coating or replacement if decreasing in efficiency is unacceptable.

5.2.4 Variations, advantage and disadvantage of granular bed coalescer

Variations of granular bed coalescer: There are several modifications of granular bed colaescers, normally, on bed materials, such ad resin bed coalescer, glass bead bed coalescer. There are 3 major modified granular bed coalescer studied at GPI lab, i.e., mixed bed coalescer,

212

Chapter 5 Coalescer

III-96

down-flow coalescer and guide coalescer. The latter will be fully described in the next section.

Mixed bed coalescer is another modified form of granular bed coalescer, used for mixed direct/inverse emulsion separation, which is usually found in liquid-liquid extraction process. In mixed bed coalescer, the bed will consist of separate layers of oleophilic and hydrophilic materials, placed in series in the same column. From research [6], ratio of oleophilic and hydrophilic material and order or configuration of column (upper hydrophilic layer/lower oleophilic layer or vice versa) depends on wastewater characteristic. So it is difficult to determine the efficiency of the coalescer by fixed equation. In this case, it is recommenced to perform pilot test to find optimum design criteria.

Furthermore, there are also variations of coalescers, based on flow patterns i.e., up-flow coalescer and down-flow coalescer. The researchs conducted in GPI lab are generally based on up-flow coalescer. For down-flow coalescer, the flow pattern in this case will be identical to that of deep-bed filter. This type of coalescer is intentionally designed to use with the oily wastewater that contains suspended solids. The wastewater will be fed from the top of the bed. The oil will be coalesced in the same manner as the up-flow coalescer. However, the coalesced oil will flow up against the wastewater stream to the inlet water surface, then be skimmed out of the reactor. This flow patterns can eliminate the top grille that is used for preventing carry-over of bed media in up-flow coalescer. This allows us to clean the bed by air scouring, back washing or any proper mechanism, such as pneumatic pulsation that can make the bed move up or partially fluidize to unclog the trapped solids. Working principles of this coalescer is generally identical to that of up-flow coalescer. However, there is not enough research data to develop the math model. So it is recommended to use the equation of up-flow coalescer for roughly estimation of efficiency. Anyway, the exact efficiency would be obtained from pilot scale testing.

Advantage: Granular bed coalescer has a major advantage in its compactness. Tested loading rate of coalescer is 3.2 – 19.4 m/h and it can be used with the droplet size from 10 microns while, for simple decanter, the loading rate is about 0.04 m/h for 10-micron droplet separation.

Disadvantage: The bed of this type of coalescer has relatively low porosity (0.14-0.19). So, at high loading rate, the pressure drop may be very high. Moreover, it can make the bed clog relatively easily. So the granular bed coalescer is sensitive to the presence of suspended solids.

5.3 Guide coalescer


Guide coalescer is the modified form of simple granular bed coalescer. In guide coalescer, high-porosity oleophilic material, such as woven metal fiber or woven mesh, will be placed next to downstream end of the granular and extended up to water surface. Coalesced oil drop will be guided along this material until it combines with oil layer at the water surface. So, this material is called “guide”. Structure of the guide shall be self-sustained or installed in perforated structure, so the treated water can flow freely out of the guide structure, then be discharged from the coalescer. The pictures of guide coalescer are shown in fig. 5.3.1-1.

213


III-97

Working principles of guide coalescer is relatively identical to those of basic granular bed coalescer. However, iinstallation of guide helps preventing formation of oil mousse or jet, which normally occurs in classical coalescer at high velocity or high concentration of oil because it virtually eliminates the drip point (step 3 of the 3 mechanisms, see section 5.2.1). Thus, the maximum velocity before formation of mousse or jet will occur (so called critical velocity) of the guide coalescer is 1.5 times greater than usual [6]. Furthermore, it can be operated at ratio of oil to water (Phase ratio) as high as 6 [6].

a) Graphic image of “guide” coalescer b) Steel wool or steel mesh that can be used as a guide

Fig. 5.3.1-1 Guide coalescer

From literature review, there is no proposed model for guide coalcescer. However, it is proven that, at velocity range below critical value, the efficiency is approximately velocity-independent [3], [6]. It, also, can be confirmed by eq. 5.2.10, which shows that the exponent of V is very low (–0.08). Then, at recommended range of velocity, V-0.08 is approximately constant.

So, from this fact, the efficiency of guide coalescer could be calculated by the model of basic granular bed coalescer (eq. 5.2.10) with additional precautions or assumptions as follows,

• When V < 0.54 cm/s: Use eq. 5.2.10 directly.

• When 0.54 < V < 0.8 cm/s (1.5 times of 0.54): Use eq. 5.2.10 by using V = 0.54 cm/s for calculation. Using the real velocity instead of 0.54 cm/s will result in higher efficiency. However, because of lacking of support data, it is recommended using low prediction value for safety.

• Velocity 0.8 to 1.9 cm/s may still be applicable at low concentration of inlet oil (phase ratio around 1-2).

• For inlet oil concentration, guide coalescer is tested in liquid-liquid extraction application [6] with maximum phase ratio (oil/water) of 6, the result shows that it can operate effectively. So, for oily wastewater treatment process, which the concentration is much lower, installation of guide can assure that the coalescer should be operate

214

Chapter 5 Coalescer

III-98

efficiently. To apply to higher concentration or beyond the limit stated above, bench scale or pilot scale testing is highly recommended

This type of coalescer have been developed for 2 purposes, 1) to increase the hydraulic loading rate of the granular bed coalescer, 2) to make it possible to use the granular bed with the water containing very high concentration of oil, such as in liquid-liquid extraction process.


Steps of calculation, as well as cut size determination, for the guide coalescer is identical to those of basic granular bed coalescer. However, there are some difference in details of sizing and pressure drop calculation as described below.

1. Coalescer sizing

The size of guide coalescer can be calculated by eq. 5.2.10. For velocity less than 0.54 m/s, actual velocity will be used in the equations. However, if design velocity is greater than 0.54, the velocity of 0.54 m/s will be used, regardless of the actual velocity. The size from the calculation can be, theoretically, used at velocity from 0.54 to 0.8 cm/s.


The removal efficiency can be calculated from the selected dimension, using eq. 5.2.10 and 5.2.12.

3. Pressure drop

Because of the fact that the “guide” in guided coalescer has relative high porosity (0.9 approx.), then, The pressure drop is very low, compared to the granular part, and can be negligible. So Kozeny-Carman’s equation (eq. 5.2.11) and recommended value of porosity from section 5.2.2 can also be used to predict pressure drop of guided coalescer.



The limitations in this case are identical to those of granular bed coalescer (see section 5.2.3) except for the velocity and oil concentration, which will conform to section 5.3.1.

2. Material selection

For direct emulsion treatment, granular bed in guide coalescer shall be oleophilic material. However, instead of top layer or grill of hydrophilic media, guide will be placed on the top of the granular bed. The guide should be made of metal or rigid material and properly coated to obtain oleophilic property. It should have high porosity and self-sustained structure because it is not supported by the wall of the coalescer. If the selected guide is not self-sustained, perforated oleophilic material shall be provided to support the guide. Commercial woven wire-meshes or metal wool, such as MuitiKnit™, Knit meshTM, 3MTM, etc., can be efficiently used as a guide.

215


III-99

5.4 Fibrous Bed coalescer


This type of coalescer uses high porous fibrous material, such as fibrous bottle brush, as a bed to promote coalescing between oil droplets. Due to its high porosity, this type of bed is hardly clogged and can handle wastewater containing suspended solids efficiently. It also causes much less pressure drop than granular bed coalescer. However, fibrous element, which is normally very small, may deflect at its tip, especially in large-scale unit, and causes unpredictable channeling of untreated wastwater, then decreasing in efficiency.

From other point of view, this coalescer can be considered as a modified form of guide coalescer that granular bed is removed and the guide is additionally used to intercept oil, besides its role to provide coalesced oil flow channel.

Three basic steps for granular bed coalescer, proposed by AURELLE [3], can also be used to describe phenomena taking place within the coalescer. However, mathematics models, derived from dimensional analysis, are proven to be more accurate.

There are 2 main categories of fibrous bed coalescers, i.e., Simple fibrous bed coalescer and dynamic (or rotating) fibrous bed coalescer, as shown in fig. 5.4.1-1. The latter is the modified form of the former, by installation of driving unit to drive the bed.

5.4.1.1 Simple fibrous bed coalescer model

This type of coalesceruses uses fibrous material, normally, in the form of “bottle brush” as a coalescer bed. The brush has relatively high porosity, compared to granular material. So this coalescer causes much lower pressure drop and is hardly clogged. Furthermore, from the three-step phenomena of coalescer, described in section 5.2.1, it shows that the efficiency of coalescer will increase if the size of the bed media is small. For fibrous bed coalescer, the size of fiber is around 100- 200 microns, much smaller than granular media’s. By this way, the efficiency can be improved.

GPI lab has been studied the possibility to use this type of material as a bed for some times. However, the researches on this type of coalescer are based mainly on its application and design consideration, rather than model development. So there is no model proposed for this coalescer. Anyway, these researches, especially the study of SRIJAROONRAT [10], provide sufficient raw data to formulate an empirical model. This newly formulated model, which will be called “SRIJAROONRAT’s model”, is as shown in eq. 5.4.1. The model has been verified, using data from MA’s and WANICHKUL’s researches, as shown in fig. 5.4.1-2.

( ) %100)(1)()()(5.104 694.035.018.018.077.0 ⋅⎟⎟⎠

⎞⎜⎜⎝

⎛−= −−

DH

Dd

DdVD F

c

cd ε

μρ

η {5.4.1}

216

Chapter 5 Coalescer

III-100

a) Simple fibrous bed coalescer [11] b) Graphical image of dynamic fibrous bed coalescer

c) Examples of “brush” d) Coalesced oil drops from dynamic fibrous bed coalescer

Fig. 5.4.1-1 Fibrous bed coalescer (Source: GPI lab)

217


III-101

Where dF = Diameter of fiber d = Diameter of dispersed phase, in our case, oil D = Diameter of coalescer bed, such as diameter of brush H = Height of bed, or bed depth V = Empty bed velocity ε = Porosity or void ratio of the bed μC = Dynamic viscosity of continuous phase, in our case, water ρc = Density of continuous phases

The model is developed and verified under these conditions i.e.;

• 48 < Re < 1100. Re is the term (ρcVD/μc) in the equation.

• 1 < H/D < 10.

• Diameter of coalescer bed (D) tested is around 5.0 cm. Using bigger coalescer diameter may cause deflection at the tips of fibers because of longer overhung length, which may cause channeling of untreated wastewater and error in efficiency calculation.

• The model is valid for droplet size (d) of 1 microns and greater.

• Empty bed velocity (V) used in the researches [10], [11], [16] is between 0.5 to 5.0 cm/s (1.8 to 180 m/h). However available raw data used to verify the model is between 0.5 to 2.0 cm/s. Using velocity > 2.0 cm/s may cause unpredictable error on calculated efficiency.

• Fiber size (dF) used in the researches [10], [11], [16] is between 40 to 200 microns. However available raw data used to verify the model is between 100 to 200 microns. Using fiber size < 100 microns may cause unpredictable error on calculated efficiency.

• Void ratio of the bed (ε) is around 0.845 to 0.96.

• The model is valid for droplet size (d) of 1 microns and greater.

• The model is verified at inlet oil concentration up to 1000 mg/l. Applying the model to the concentration > 1000 mg/l will cause underestimation of predicted efficiency, as shown in fig. 5.4.1-2 for WANICHKUL’s data (C = 7950 mg/l) [11].

• The beds used in these researches vary from “bottle brush” type, simple spiral type and combination of internal bed of “simple spiral” and concentric “coil spring–like” external bed with the tip of the fibers pointed to the centerline. However, they are all oleophilic. There is some difference in efficiency between each type, but there is too few data to make a conclusion. However, because of its rigidity, the “simple spiral in coil spring- like” bed tends to operate more stable without decreasing in efficiency with time, while others tend to be deflected by weight of accumulated oil drops. In fact, this type of bed is invented to take advantage of spiral bed for its non-clogging and disorderly bed (section 5.4.1.3) for its rigidity and good interception efficiency.

218

Chapter 5 Coalescer

III-102

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%


Pred

icte

d ef

ficie

ncy

(%)


-20 %

+20 %

Fig. 5.4.1-2 Comparison between observed efficiency and predicted efficiency from SRIJAROONRAT's model, Verified by MA's and WANICHKUL's data.

5.4.1.2 Dynamic fibrous bed coalescer model

Dynamic brush coalescer is the modified form of simple fibrous brush coalescer by installing the prime mover to drive the brush. From TAPANEEYANGKUL [8], rotating motion of the brush provides 2 major benefits, i.e.,

• It can increase the probability of interception or collision between fiber elements and oil droplets. Thus, the efficiency will be improved. Furthermore, if the variable speed driving system is installed, the efficiency of the coalescer will be adjustable.

• Rotating motion also helps reducing possibility of clogging.

The equation for efficiency prediction of this coalescer is proposed by TAPANEEYANGKUL [8]. It is based on classic dimensional analysis, which is the efficient tool when exact theory of the processes can not be established. The equation will be as shown in eq. 5.4.2.

( ) %100)()(1)()()(76.1 53.035.035.058.058.021.0 ⋅⎟⎟⎠

⎞⎜⎜⎝

⎛ ⋅−= −−

VND

DH

Dd

DdVD F

c

cd ε

μρ

η {5.4.2a}

Or

%100)1(67.074.058.003.0

53.035.035.058.0

⋅⎟⎟⎠

⎞⎜⎜⎝

⎛ −=

VdDNHd

Fd

εη {5.4.2b}

219


III-103

The equation is developed and verified under these conditions i.e.;

• 52 < Re < 1164. Re is the term (ρcVD/μc) in the equation.

• 1 < H/D < 2. However, the maximum H/D shown in TAPANEEYANGKUL’s research [8] is 6. Using H/D > 2 in the model can be also applied for comparison purpose only.

• Rotating speed of the bed (N) is between 0.167 to 3.33 rps (10 to 200 rpm). Please note that N is in form of revolution per unit time, not radian per unit time). Recommended minimum rotating speed is 75 rpm. Using lower speed may not provide any additional benefit over simple fibrous bed coalescer because the effect of rotating on interception probability may be cancelled out by the shear effect, which causes fragmentation of oil drops.

• Empty bed velocity (V) is between 0.1 to 1.1 cm/s (3.6 to 39.6 m/h) .

• Diameter of fiber (dF) is around 100 to 300 microns

• Diameter of coalescer bed (D) is not greater than 11.5 cm. Using bigger diameter may cause deflection at the end of fibers from longer overhung lengths, which may cause channeling of untreated wastewater and error in calculation.

• It is recommended to use the model only for the droplet size (d) of 10 microns or greater. For smaller droplet, the model can also be applied, but for comparison purpose only.

• The beds, used in the experiment, are “bottle brush” types, made of oleophilic polyamide or polypropylene with stainless steel shaft, as shown in fig. 5.4.1-1c.

5.4.1.3 Random or disorderly fibrous bed coalescer

There is another special case of simple fibrous bed coalescer that uses random or disorderly fibrous material (such as metal wool, etc.) as coalescer bed. SRIJAROONRAT’s research shows that removal efficiency of this coalescer is higher than that of coalescer that uses brush type bed. For this, it can be concluded that tortuosity of bed also affects the removal efficiency. From SRIJAROONRAT’s raw data, an empirical model, based on dimensional analysis, can be derived as shown in eq. 5.4.3. However, this model is developed from rather small set of data. There is an effort to verify if the simple bed model (eq. 5.4.1b) is still valid for random fibrous bed coalescer. Comparison between efficiency from eq. 5.4.1b, eq. 5.4.3 and observed value is shown in fig. 5.4.1-3.

%100)()()()(35.3 36.003..003.023.0 ⋅⎟⎟⎠

⎞⎜⎜⎝

⎛= −−

DH

Dd

DdVD F

c

cd μ

ρη {5.4.3}

220

Chapter 5 Coalescer

III-104

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

110.00%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%


Pred

icte

d Ef

ficie

ncy

(%)

+10%

-10%

Fig. 5.4.1-3 Relation between observed efficiency and predicted efficiency fromrandom fibrous bed coalescer model and simple fibrous bed model

From the graph, it shows that SRIJAROONRAT’s random fibrous bed model (eq. 5.4.3) can accurately predict the efficiency of the coalescer while the result from the simple bed model (eq. 5.4.1b) tends to underestimate the efficiency from 2 to 6 times. This confirms that tortuosity of the bed plays very important role in efficiency of fibrous bed coalescer. However, tortousity can not be effectively established in form of numerical value so it can not be included as a parameter in mathematics model. Eq. 5.4.3, then, can be applied only when design condition is close to the test condition from SRIJAROONRAT’s research.

Eq. 5.4.3 is developed and verified under these conditions i.e.;

• The beds used in the experiment are highly disorderly bulk of stainless steel fiber, dF = 75 microns, and steel wool, dF = 40 microns (see fig. 5.4.1-4). However, only the latter case, which raw experimental data is available, is used to develop the model. The minimum size of oil droplet tested is 1 micron.

• Tested Reynolds number is between 840 to 2470.

• Porosity of the bed (ε) is around 0.95.

• Diameter of the coalescer (D)= 5 cm.

• Height of the coalescer bed (H) is between 0.07 to 0.21 m.

• Velocity (V) is between 1 to 2.5 cm/s or 36 to 90 m/h.

• Inlet concentration is around 1000 mg/l.

In case that the conditions stated above are not fully compliant, it is recommended to use eq. 5.4.1b to calculate the efficiency of the coalescer because it is proven to be valid within wider range. In this case, predicted result from eq. 5.4.1b tends to underestimate the efficiency of the coalescer.

221


III-105

a) Highly disorderly stainless steel fibers b) Steel Wool

c) Magnified picture of (a) d) Magnified picture of (b)

Fig. 5.4.1-4 Random or disorderly fibrous bed, used in the research of SRIJAROONRAT


Even though each type of fibrous bed coalescer is governed by different equation, all of the equations is in about the same mathematics form, as shown in eq. 5.4.4. Thus, main parameters that affect the efficiency of the fibrous bed coalescer can be commonly summarized as shown in fig. 5.4.1-5.

))1(,,,,1,1,1( εη −= NHddDV

fF

d {5.4.4}

222

Chapter 5 Coalescer

III-106

Droplet size (d)

Efficiency (η)

100%

Lower limit ofthe model: 1 micron

Droplet size (d)

Efficiency (η)

Increase V, D, dFDecrease H, (1-ε), NLess tortousity

100%

Fig 5.4.1-5 Typical relation between efficiency of fibrous bed coalescer and various parameters

However, besides the parameters shown in eq. 5.4.4, there are some important parameters that affect the efficiency of the coalescer, i.e.,

1. Tortousity

If the bed is more tortuous, the probability to intercept oil droplets will be increased. The efficiency, then, is increased, as clearly shown in the case of wool coalescer. However, it will tend to be clogged if suspended solids are present in wastewater.

2. Wettability of bed

From the 3 mechanisms of AURELLE [3] (section 5.4.1.1), wettability of the bed is the key parameter that governs the step 2 “adhesion-coalescence”. In case of direct (oil in water) emulsion treatment, he bed material shall be oleophilic to optimize the adhesion-coalescence phenomena.

3. Interfacial tension and presence of surfactants

Surfactant will lower the interfacial tension, then make the oil droplets more stabilized. This leads to poor adhesion between droplets and surface, ineffective collision (collision without coalescence), and re-fragmentation of oil drops. So presence of surfactant directly limits the efficiency of step 2 “adhesion-coalescence”. To optimize the efficiency of step 2, it is recommended to destabilize, (breaking or cracking) the emulsion before sending to the coalescer

4. Temperature

Properties of oil and water, such as viscosity, change with temperature. So temperature is inevitable an important parameter that affect coalescer efficiency. Then it is important to design the process by taking the possible range of temperature into account. Normally, the efficiency will, more or less, increase with the temperature. The equations in this

223


III-107

section are developed under the temperature range between 15 to 25oC, which is general practical operating range.

5. Size of bed shaft

Another precaution about the bed is the ratio of bed center shaft to diameter of bed. Since the effective part of the bed is its fibers, efficiency of coalescer of the same diameter will decrease it the shaft is relatively large. Equations in sections 15.4.1 already include the effect of center shaft. The sizes of the shafts used in those researches are around 1.0 to 1.5 cm., or about 20 to 30% of diameter of the bed. If the selected shaft does not conform to this criterion, diameter of shaft used in the equations will be adapted to the equivalent diameter that has the same effective area of fiber as the real bed with the shaft size conforms to the criterion.

5.4.1.5 Pressure drop of fibrous bed coalescer

Many researcher [8], [10], [11], [16] observe pressure drop of fibrous bed coalescers and report that these coalescer causes very low pressure drop due to very high porosity of their beds. However, there is not any proposed model on pressure drop.

In order to calculate the pressure drop, it is recommended to use any general piping loss equations, such as Darcy’s, Colebrook-White’s or Hazen-William’s equation with the safety factor of 2 to 5, multiplied to the actual length of the bed. However, the pressure drop of fibrous bed coalescer is normally low (< 104 N/m2), compared to piping system pressure drop.

5.4.1.6 Size of coalesced oil drop

Observed size of coalesced oil drop from fibrous bed coalescer with oleophilic bed is reported [3], [10] to be between 0.3 mm to 8.4 mm with the average value between 1.8 to 6.8 mm (table 5.4.1-1). The size of the oil drop depends on operating condition, esp. velocity, and material properties, such as interfacial tension, fiber wettability. The size will decrease if the velocity increases or interfacial tension decreases or the fiber is hydrophilic.

From STOKES law, rising velocity of the droplet of 1.8 mm diameter is around 0.33 m/s, which is greater than maximum tested velocity in the fibrous bed coalescer (around 5 cm/s). So the oil drop is positively separated. Furthermore, if oil and water discharge pipes of the coalescer is properly arranged, even the smallest drop of 0.3 mm can still be separated. So, it can be safely said that all of coalesced oil can be decanted by downstream end of the coalescer because it requires less decanting area than the size of the coalescer itself. However, using moderate velocity (around 2.0 cm/s) and sufficient height (H/D>2) will guarantee sufficient size of coalesced oil drop for good decanting.

Anyway, it is interesting to consider granulometry of remaining oil droplets that are not fully coalesced to form big oil drops. To help considering this, there are some researches on coupling or combination of coalescer and hydrocyclone [10], [11]. Example of the result from [10] is show in table 5.4.1-2. From the table, it shows that the observed total efficiency of the coupling is higher than the product of the two separate processes. So it may be implied that the size of the smallest droplets from coalescer is increased. Therefore, with these bigger oil droplets, the efficiency of the hydrocyclone is increased. However, there is not enough data for further

224

Chapter 5 Coalescer

III-108

analysis. Furthermore, if this partial coalescing exists, its effect should be already included in the efficiency prediction models (eq. 5.4.1 to 5.4.3). So, for calculation, it is safer to assume that there is no partial coalescing. Prediction of coupling of processes would be calculated from the product of the efficiency of each individual process.

Table 5.4.1-1 Coalesced kerosene droplet size at various velocities and bed heights from oleophilic “bottle brush” coalescer (dF = 100 microns, D = 0.05 m, Co = 1 g/l, 120oC) [10]

Bed height

Empty bed velcity Coalesced oil diameter

H V min max weighted average

cm cm/s mm mm mm

7 1.8 1.4 5 3.3

2.8 0.6 5 2.2

4.25 0.4 3.3 2

5.3 0.3 3.3 1.8

14 1.8 6 6 6

2.8 1.4 6 3.5

4.25 0.8 6 2.3

5.3 0.6 5 1.8

21 1.8 7.2 8.4 8

2.8 6.4 7.6 6.8

4.25 5 7.2 6.5

5.3 0.6 6.4 2.7

Table 5.4.1-2 Comparison between the individual efficiency of oleophilic bottle brush coalescer, hydrocyclone, theoretical and observed efficiency of the coupling of coalescer/hydrocyclone [10]

Initial droplet

diameter

Empty bed

velcity

Observed efficiency of

hydrocyclone (a)

Observed efficiency of

coalescer (b)

Theoretical efficiency of the coupling 1-(1-a)(1-b))

Observed efficiency of the coupling

micron m/s % % % %

11 0.02 35% 54.19% 70% 82%

11 0.03 50% 39.65% 70% 82%

11 0.04 61% 31.78% 73% 82%

11 0.05 69% 26.76% 77% 82%

225


III-109


Steps of calculation, as well as cut size determination, for fibrous bed coalescer is identical to those of basic granular bed coalescer. However, there are some difference in details of sizing and pressure drop calculation as described below.


The ut size can be determined from the degree of treatment required as well as from the limitation of the coalescing processes. Cut size determination from degree of treatment is described in chapter 3. For the limitations of the process, it is mainly model limitation. Since the model used for calculation is based on empirical data, it can be applied only within its valid range. Extrapolation of model may cause unpredictable error. The limitation of each model is described in section 5.4.1.

2. Coalescer sizing

The size of various types of fibrous bed coalescer can be calculated by eq. 5.4.1 to 5.4.3. Design cut size will be used to calculate the dimension of the coalescer by assuming that the graded efficiency at the cut size is 100%. Dimension of the coalescer can be arbitrarily selected under the limitations of each equation, as described in section 5.4.1.1 to 5.4.1.3. However, it is recommended to select the possible lowest fiber diameter and highest velocity for the first trial because it will give the most compact diameter of the coalescer. If the result is acceptable, we may try to fine-adjust the parameters to get the most suitable dimension. If the result is unacceptable, normally too big, we may have to increase the number of units.


After the dimension of the coalescer is determined, graded efficiency (ηd) can be calculated by eq. 5.4.1, 5.4.2 or 5.4.3. If the result from eq. 5.2.10 is greater than 100%, then it will be rounded up to 100%.

It must be noted that the graded efficiency from the equation is not yet accounted for effect of flow splitting between water outlet flow and separated oil outlet flow. To calculate graded outlet oil concentration, the effect must be taken into account, as shown in eq. 5.2.12a.

oddout

d CQ

QC )1( η−= {5.2.12a}


d

d

dodd

out

CQQQ

ρ

η∑ ⋅−=

max

min

)( {5.2.12b}

226

Chapter 5 Coalescer

III-110


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {5.2.12c}


4. Pressure drop

As stated in section 5.4.1.5, in order to calculate the pressure drop, it is recommended to use any general piping loss equations, such as Darcy’s, Colebrook-White’s or Hazen-William’s equation (eq. 5.4.5) with the safety factor of 2 to 5, multiplied to the actual height (H) of the bed. CHW is Hazen william’s constant, depending on surface roughness of coalescer. Recommended value is 110-130 for steel column.

167.185.1582.6 ⎟

⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛=

DH

CVPHW

m {5.4.5}



The equations described above are developed from the following assumptions and limitations. Thus, it is necessary to make sure that these assumptions are valid when design the coalescer. Limitation of each design equation (eq. 5.4.4 to 5.4.3) is described in section 5.4.1.

2. Construction and material selection

Fibrous bed coalescer consists of 2 major components, i.e., casing and bed.

1) The casing

The casing is the component that contains the bed and other important appurtenances, such as wastewater inlet, wastewater outlet, decanting section, oil outlet, clean out, etc. Researches in GPI lab normally emphasize on working mechanisms and developing of a model so there is no direct study on characteristic of the casing. However, the size of the fibrous bed coalescer is usually small to avoid deflection at the tip of fibers, it is, then, designed as inline unit or in the form of pipe (see fig. 5.4.1-1 and 5.4.3-1). Batteries or multi-coalescer unit is also available. Downstream decanter may be integrated or placed separately from the coalescer as shown in fig. 5.4.3-1. As described in section 5.4.1, diameter of the decanter is normally equal to that of coalescer or a bit bigger to facilitate placement and construction of internal baffle, water and oil outlet pipe.Casing of coalescer can be made of any material (plastic, steel, glass, etc.) that can withstand the operating condition. If possible, it should be oleophilic.

227


III-111

a) Lab scale coalescer with integrated decanter (The bed is placed in the steel

column just below the decanter)

b) Inline-style fibrous bed coalescer [11] c) Assembly of the coalsecer and separated decanter (Source: Elf, GPI lab)

Fig. 5.4.3-1 Examples of fibrous bed coalescer casing

2) The bed

The bed is the most important part in the coalescer. From section 5.4.1, there are 2 major types of fibrous beds, i.e., simple or brush type and random or disorderly type. Difference in operating efficiency between the two is already discussed. However, one property, which both beds have in common, is that, for direct emulsion treatment, the bed shall be oleophilic.

Material of the bed varies from steel, stainless steel and variety of plastic, scuh as polyamide, polypropylene as shown in fig. 5.4.1-1, 5.4.1-4 and 5.4.3-2. For simple brush type, the efficiency of the coalescer varies only slightly for various types of bed shown in the figure. The factor that governs the efficiency for this type of bed is its oleophilicity. Guideline on wettability of material is described in chapter 2. To preliminary test olephilic property of the bed, it can be done easily by dip or have the bed contact with oil or water. If the drop of oil or water is almost sphere, it can be approximately concluded that the material is hydrophilic or oleophilic respectively (see fig. 5.4.3-3)

228

Chapter 5 Coalescer

III-112

Fig. 5.4.3-2 Various types of bed tested by WANICHKUL [11]: From left, a) Simple spiral, b) Double spiral, c) Solid plastic spiral, d) A module of multi-stage bed [No. of stage can be added by increase the number of module into the same center shaft

a) water drop on

silicon coated fibers b) water drop on non-

coated fibers c) oil drop on silicon

coated fibers d) oil drop on non-

coated fibers

Fig. 5.4.3-3 Characteristics of oil and water drops on silicon coated (oleophilic) steel fibers and non-coated (hydrophilic) steel fibers

5.4.4 Variations, advantage and disadvantage of fibrous bed coalescer

Variations of fibrous bed coalescer: There are several modifications of fibrous bed colaescers, normally, on bed type and materials, such as simple fibrous bed, steel wool bed, rotating or dynamic bed, as described in section 5.4.1. Each type of bed can be sub-divided into sub-types, characterized by specific features of beds. The efficiencies of these “brush” beds are almost identical, as discussed before. However, there are some variations of these coalescers worth to mention here for their initiative idea.

Double spiral bed (fig. 5.4.3-2b) is one of the innovations of GPI lab [10], [11]. It is designed to mitigate the fiber tips deflection effect. The bed provides homogeneous void ratio at any radial distance. External “coil spring-like” part will help intercepting oil droplets without the problem about defection since the fibers just protrude a short distance from their coil spring-like support. This type of bed can also reduce undesirable channeling, taking place at high velocity at the tips of simple fiber bed. Anyway, it tends to clog easier if suspended solids are presence.

229


III-113

Multi-stage simple brush coalescer is another initiative idea to improve the efficiency of simple fibrous bed coalescer [11]. Eq. 5.4.1 to 5.4.3 show that efficiency varies with Ha, when 0< a <1. If a coalescer of bed height “H” and efficiency of one stage “η1-stage” is divided into n sets of bed height (H/n), theoretical total efficiency of the series of these coalescers will be as shown in eq. 5.4.5.

na

stagestagesn n

)1(1 1−− −−=

ηη {5.4.5}

From the equation, when η1-stage and “a” is lower than 1, the value of ηn-stages is always higher than η1-stage. From this fact, WANICHKUL [11] had been developed annular brush module (fig. 5.4.3-2d) that could be placed close to each other or with some spacing between them. When there is a sufficient distance between modules, it can be assumed that the coalescer becomes multi-stage. Theoretically, the efficiency will increase as shown in eq. 5.4.5. WANICHKUL use spacing of 1.0 cm. between each module. The result shows that the total efficiency of this multi-stage coalescer is actually higher than the single stage one. However, efficiency of the multi-stage coalescer is approximately equal to efficiency of the single stage coalescer plus 10%, not as high as calculated from eq. 5.4.5. This is simply because the efficiency is also a function of oil concentration. After each stage, the oil concentration for the next stage will decrease, so eq. 5.4.5 will be no longer valid. However, this research confirms the idea of efficiency improvement by simply adding some spacing between each module of bed.

Coupling coalescer and cyclone is another simple modification of 2 well known processes that can give an impressive efficiency. This coupling is designed on the basic knowledge that.

• Efficiency of hydrocyclone increases with increasing velocity.

• Hydrocyclone can not work efficiently at droplet size less than 20 microns.

• Efficiency of coalescer decreases with increasing flowrate.

• At high velocity, coalescer can intercept these tiny droplets and, somehow, form bigger droplets, even though the efficiency is low.

Combination of upstream coalescer with downstream hydrocyclone makes the system work at relatively high and constant efficiency, regardless of the velocity, as shown in table 5.4.1-2. This combination will be described again in chapter 12.

Advantage:

Major advantages of fibrous bed coalescer are summarized below;

• Fibrous bed has high porosity, so it is hardly clogged. Then regeneration or backwashing is normally not required.

• Hydraulic loading rate is very high (up to 72 m/h), compared to simple decanter. So the size is very compact.

• Pressure drop is very low for the bed has high void ratio.

230

Chapter 5 Coalescer

III-114

• Very small size of fiber ensures good interception of oil droplets.

• The coalescer in form of in-line unit can be designed as a preliminary treatment for other downstream oil/water separation process, such as hydrocyclone.

Disadvantage: Major disadvantage of fibrous bed coalescer is undesirable channeling of untreated wastewater from fiber deflection, either by hydraulic force form wastewater flow or accumulation of oil and solids on the fibers. From this constraint, size of the coalescer is limited at several centimeters, normally less than 10 cm. So, at high wastewater flowrate, it may require too many units to be economical. For disorderly bed, inspite of its good efficiency, it is sensitive to presence of suspended solids. Moreover, from its tortuosity, it is very difficult or impossible to unclog.

231


III-115


6.1 General

Flotation is an accelerated separation process, operated by increasing density difference between continuous phase and dispersed phase. This is accomplished by mean of adding gas or air into the wastewater to promote formation of air-solids or air-oil agglomerates. There are several researches on flotation, its modification and applications of various types of flotation such as mechanical flotation, diffused air flotation, dissolved air flotation (DAF), etc. Among these, DAF, with its finest air bubble size, is the most efficient process. So it became the main flotation process studied in GPI lab.

6.2 Working principles

From STOKES law (eq. 6.2.1), difference density between dispersed phase and continuos phase (Δρ) is one of the parameters that governs rising or decanting velocity of the dispersed phase. Flotation is the separation processes of which aims to increase the density difference to increase the decanting velocity.

c

gdUμ

ρ18

2Δ= {6.2.1}

For dissolved air flotation, pressurized water which is (or almost) saturated with air or gas will be fed to wastewater at lower, usually, atmospheric pressure. The air or gas will be released from the pressurized water in form of tiny bubbles. These bubbles, while rise up to the surface, will collide with dispersed phase, in case of oily wastewater oil droplets, in the wastewater. Some will attach to the droplets and form oil-bubble agglomerates. Since the bubble has much less density than oil and water. Density of these agglomerates will be lower than that of oil droplets alone, thus, make the rising velocity increase. General schematic diagram of DAF process is as shown in fig. 6.2-1.

Fig. 6.2-1 Example of schematic diagram of DAF (Source: Aquatec Maxcon)

232


III-116

There are many attempts to predict the efficiency of DAF or to formulate mathematics model of DAF. However, there is no general model so far that can predict the efficiency of the process accurately at any operating condition. Almost all of the models are on empirical. However, there are some researches in GPI lab to formulate the models of DAF that are semi-empirical or based on some theories. So they can be used as tools to understand the effect of some parameters on DAF operation and to adapt empirical data on DAF efficiency at one condition to estimate, only roughly, the efficiency at other condition. There are 2 approaches to formulate such a model, studied in GPI lab, i.e.,

• Filtration based model • Population balance model

The two approaches are described in section 6.2.1 and 6.2.2.

6.2.1 Filter based model

From the concept that air bubbles are used to intercept oil droplets and form oil-bubble agglomerate, AURELLE and SIEM [12] considered that this concept is relatively close to that of filtration with the air bubbles play the role of collector. So they formulated model of DAF based on the filtration model, proposed by YAO et al. [35]. Details of model formulation are as shown in the following sections.

6.2.1.1 Applying filtration model to DAF

From filtration model, interception of dispersed phase by a collector of filter media is based on 3 transport phenomena, i.e., sedimentation, direct interception and diffusion. Other transport phenomena, such as that of electrical force, might take place, but it is proven that their effect is small, thus, negligible. However, for DAF, the collectors, which are air bubbles, are not stationary but have their own rising velocity and also carried along with the water. So concept of relative velocity will be applied in stead of absolute velocity.

To calculate relative velocity of air bubble/ oil droplets in flotation column, consider a flotation column which wastewater and pressurized water are fed at the bottom. This causes flow velocity “V” in the column (see fig. 6.2.1-1). Air bubbles generated from pressurized water, as well as, oil droplets in the wastewater will be carried along with the liquid. However, from STOKES law, the air bubble and the oil droplets also have their own rising velocities, “Ub“and “Ud“ respectively. So, for the observer looking from outside of the column, the absolute velocity of bubble and oil droplets will be equal to “V+Ub“ and “V+Ud “ respectively. Then relative velocity between air bubble and water (Vr) will be as shown in eq. 6.2.2a while the relative velocity between bubble and oil droplets (Ur) is shown in eq. 6.2.2b.

bbr UVUVV =−+= )()( {6.2.2a}

dbdbr UUUVUVU −=+−+= )()( {6.2.2b}

From above equations, if an observer stands on the bubble, it will seem to him that the water is flowing past him at velocity “Ub”, while oil droplets move toward him at slightly lower velocity “Ub-Ud”. To apply filtration model to DAF, we will consider, firstly, the phenomena, taking place at 1 collector (or 1 bubble). Assume that frame of reference is on the bubble, schematic diagram of the 3 transport phenomena is shown in fig. 6.2.1-2. Efficiency of each phenomenon is described as follow.

233


III-117

V

Ub Ud

x

y

Bubble,Collector

Oil droplet

Fig. 6.2.1-1 Diagram for considering relative velocity of bubble and oil in flotation column

Ur= Relative decanting velocityVr= Relative flow velocity

Vr Ur

Vr Ur

Oil droplet

Stream lineBubble,Collector

a) Direct interception b) Sedimentation c) Diffusion

Fig. 6.2.1-2 Schematic diagram of the 3 transport phenomena

1. Direct interception

When oil drops of diameter “d” flow along with the streamline, the oil drops that pass within the distance less than d/2 from the bubble will be intercepted by the bubble, as shown in fig. 6.2.1-1a. The efficiency factor for this phenomenon can be calculated by eq. 6.2.3. db is the diameter of bubble.

2)(23

bInt d

d=η {6.2.3}

2. Sedimentation

Consider oil drops of diameter “d” flow along with the streamline. When they are far from the media, the relative decanting velocity (Vr) and relative flow velocity (Ub) will have the same direction. When they come close to the media, the flow velocity will deviate, as shown by the streamline, while the oil drops will be subjected to both flow velocity and their own decanting velocity. So the resultant velocity will not totally conform to streamline. And in some cases, it will make the oil drops collide to, thus, sediment on the bubble. The efficiency factor for this phenomenon can be calculated by eq. 6.2.4.

rc

wateroilsed V

gdμ

ρη

18

2/Δ

= {6.2.4}

234


III-118

3. Diffusion

For very small droplets (d < 5 microns), They will be subjected to Brownien’s motion. These random motions can cause the droplets to collide to the media. The efficiency factor for this phenomenon can be calculated by eq. 6.2.5. K, in this equation, represents the Bolzmann constant and T represents the absolute temperature.

3/2

9.0 ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

=rbc

Diff VddKT

μη {6.2.5}

4. Combined theoretical efficiency

From 1 to 3, we can calculate the efficiency factors of each transport phenomena for single collector (or bubble). Theoretical efficiency factor of single collector is the summation of the efficiency factors of those three transport phenomena, as shown in eq. 6.2.6.

DiffSedInttheo ηηηη ++=

Thus

3/22

2/ 9.0)(

23

18 ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

++Δ

=rbcbrc

wateroiltheo Vdd

KTdd

Vgd

μμρ

η {6.2.6}

Assume that frame of reference is on the bubble, it can be implied that the bubble is stationary within the flow of wastewater. Then, consider one bubble as shown in fig. 6.2.1-3, The fraction of wastewater flowing past the single collector will be the flow that passes through the projected area of the collector (q), as shown in eq. 6.2.7a. Then, some oil droplets in this fraction of the wastewater will be intercepted by the collector. The quantity of intercepted oil drops of the single collector (c’) will be calculated from the theoretical efficiency factor, as shown in eq. 6.2.7b.

rb Vdq ⋅= 2

4π

{6.2.7a}

CVdc rbtheo ⋅⋅⋅= 2

4' πη {6.2.7b}

q = Vrπdb2/4

Vr

db

EffectiveConcentration = C (include dilution)

dH

Concentration afterintercepted by bubbles = C- dC

Flow velocity = V

Relative velocity between bubble and water = Vr

Bubble size = db

Column height= H

Cross sectionarea = Ao

a) b)

Fig. 6.2.1-3 Schematic of single bubble and entire height of flotation column

235


III-119

C is inlet concentration of oily wastewater. For the entire flotation column, we will consider a very small slice of rising bubbles of the height dH (see fig. 6.2.1-3b). The number of bubble particles in this slice can be calculated from the cross section area of bed (A), the size of the bubble (db) and the ratio of volume of water to total volume (ε), as shown in eq. 6.2.7c.

Then, the total concentration of intercepted oil for this slice of rising bubbles will be equal to the product of c’ and the number of bubbles. However, not all of the intercepted oil drops will adhere to the bubbles. So the probability coefficient (α) will be applied to adapt the quantity of intercepted oil drops to the quantity of adhered oil drops (c”), as shown in eq. 6.2.7d.

The number of bubbles in the slice dH 3

6

)1(

bd

AdHπ

ε−= {6.2.7c}

3

2

6

)1(4

"b

btheo

d

AdHCVdcπ

επηα −⋅⋅⋅⋅⋅= , α < 1 {6.2.7d}

If dC represents the concentration of oil reduced after past through the slice of bubbles dH, then we have got eq. 6.2.7e and f;

"cdCAV =⋅⋅− {6.2.7}

3

2

6

)1(4

b

btheo

d

AdHCVddCAV

πεπηα

−⋅⋅⋅⋅⋅=⋅⋅− {6.2.7f}

Therefore,

dHdC

dCtheo

b

αηε )1(2

3−−= {6.2.7g}

Integration of eq. 6.2.7g will give the value of the oil concentration reduced by the entire column height, as shown in eq. 6.2.8a.

btheo

o dH

CC αηε )1(

23)log( −−= {6.2.8a}

However, void ratio (e) of DAF system can be calculated directly from air flowrate discharged from pressurized water and total water flow as shown in eq. 6.2.8b.

AVQt

Φ=

Φ=− ε1 {6.2.8b}

Thus, the theoretical removal efficiency DAF, based on filtration model, can be written as shown in eq. 6.2.8c.

%1001%1001)(

23

, ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−=⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛−=

⋅⋅Φ

⋅− theodpH

AV

otheod e

CC ηα

η {6.2.8c}

236


III-120


The filtration based model, proposed by AURELLE and SIEM, can be effectively described the phenomena taking place within DAF reactor and effect of parameters to DAF efficiency. However, formation of bubble-oil agglomerates, which is rather complex phenomenon into itself, is simplified to only one parameter “α”. The value of α would vary with many parameters such as characteristic of wastewater, operating condition, etc, which may not be included in the 3 transport phenomena. This causes some discrepancies between theoretical efficiency and observed efficiency. So the efficiency from eq. 6.2.8c should be corrected by a correction factor.

SIEM had studies the relation between observed efficiency (α.ηexp) and the theoretical efficiency and found that the observed values are quite different from the theoretical value from eq. 6.2.8c. Relation between observed efficiency and theoretical efficiency, suggested by SIEM, is shown in eq. 6.2.9a and fig. 6.2.1-4. So the theoretical efficiency in eq.6.2.8c can be modified by replacing αηtheo with α.ηexp. The efficiency of DAF, then, can be calculated by eq.6.2.9b.

5919.0exp )(009005.0 theoηηα =⋅ {6.2.9a}

%1001%1001)(

23

exp

⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−=⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛−=

⋅⋅Φ

⋅− ηαη dp

HAV

od e

CC {6.2.9b}

y = 0.5919x - 4.71R2 = 0.9821

-9.0E+00

-8.0E+00

-7.0E+00

-6.0E+00

-5.0E+00

-4.0E+00

-3.0E+00

-2.0E+00

-1.0E+00

0.0E+00-7.0E+00-6.0E+00-5.0E+00-4.0E+00-3.0E+00-2.0E+00-1.0E+000.0E+00

ln(Theoritical efficiency)

ln (O

bser

ved

effic

ienc

y)

(αηt)= 0.009005 (ηt theo)0.5919

Fig. 6.2.1-4 Relation between theoritical efficiency factor and observed efficiency factor

6.2.1.3 Limitation of filtration based model of DAF

From eq. 6.2.3 to 6.2.5, it shows that the three efficiency factors does not depend on flowrate because there is no parameter relating to the flowrate, such as velocity, etc. in the equations. Furthermore, from eq. 6.2.9b, the term Φ/AV can be rewritten in form of pressurized water (Qpw) and wastewater (Qt) flowrate, as shown in eq. 6.2.10.

237


III-121

airt

pw airConcQ

QAV ρ

)(⋅=

Φ {6.2.10a}

Or

XRQ

RQXQ

Q

t

pw ⋅+

=⋅=1

{6.2.10b}

Normally, under certain design condition, Qpw/Qt or Qpw/Q (= R)is constant. Solubility of air in water (X) and air density are intrinsic (internal) properties, which are constant at any given pressure and temperature. So, from eq. 6.2.10, it shows that Φ/AV is flow-independent.

From these, it seems that predicted DAF efficiency from the filtration-based model is mathematically flow-independent. However, In fact, effect of wastewater flowrate, sometimes, described in form of velocity or retention time, is studied by many researchers. It is widely accepted that DAF efficiency varies with retention time. So, considering SIEM’s research, it can be interpreted that the effect of retention time is already included in eq. 6.2.9a. Since eq. 6.2.9a is evaluated from one set of operating condition (Φ/AV = 0.0516) and retention time (25 minutes (approx.)), it may be valid only for that condition. To expand valid range of SIEM’s model, some criteria or model that contains flowrate-dependent parameter (such as retention time, etc.) is required.

6.2.2 Population balance model

Population balance model is one of popular concepts, used by many researchers, to develop DAF model. In GPI lab, DUPRE [14] uses this method in her research on application of DAF for liquid-solid separation. Concept of this method is that rate of change in the number of pollutants which is free or attached by 1, 2 or more air bubbles is the function of the number of air bubble and oil droplet, as shown in eq. 6.2.11. Oil droplet that is attached by at least 1 bubble, so-called oil-bubble agglomerate, will be separated. So rate of change in number of oil droplets represents removal efficiency.

Main assumption of population balance method is that the number of bubble is assumed to be constant. From many researches [14], [34], It is proven that, for liquid-solid separation, the number of bubble (N) can be safely assumed as constant without serious error, because there are a lot more bubbles than pollutant if normal range of (Qpw/Qt) is applied. However, in case of oily wastewater, many researches [12], [13] show that bubble-oil agglomerates are in form of oil shell with air inside and these agglomerates are still able to intercept more oil droplets. Those researches also show that coalescence of oil and bubble is more effective than that of the same species. So it should be safely assumed that the number of bubble in this case is more or less constant and the population balance method is, then, applicable.

Nnkdt

dn00

0 β−= {6.2.11a}

NnkNnkdtdn

iiiii ββ −−= −− 11 {6.2.11b}

i = 0 to imax

238


III-122

3

6 bdN π

Φ= {6.2.11c}

Where κ = Collision rate constant for population balance equation (T-1) βi = Adhesion efficiency between bubbles and oil drop/ bubble

agglomerate at the number of bubble of the oil/bubble agglomerate = “i”

N = the number of bubble

no and ni represent the number of oil droplet which is free and attached by i bubbles respectively. κ represents collision rate constant, from Saffman and Turner’s (1956) coagulation theory, and β represents adhesion efficiency or probability that the collision between oil drop and bubble will be successful. Removal efficiency can be written as (ni/n0), which can be achieved by integrating eq. 6.2.11.

To integrate eq. 6.2.11, complex numerical method is required. MATSUI and LEPPINEN [34], [37] solve the equation by Laplace transforms and suggest the solution as shown in eq. 6.2.11.

i

bddi eeixNn

⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜

⎝

⎛

−⋅=⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛

− 1:2/2

)(

κτ

κτ {6.2.11a}

i = 0 to imax –1

)1)(2)...(1())1()...(1(:

−−−−

=ii

ixxxix {6.2.11b}

22 / bddx = {6.2.11c}

When i = 0, the above function will be equal to 1.

03))/(1(6 β

πκ Φ+= bddaG {6.2.11d}

Where a = numerical constant

G is a measure of turbulence intensity, which is normally proportional to V but the exact relation depends on DAF design and operating condition. For i = 0, ni will become n0, which represents the number of free oil drop remaining from contacting with air bubble. So 1-(n0/N) represents removal efficiency of the process. Using i > 0 will result in greater value of efficiency. So, to be on the safe side, we will use i = 0. Then eq. 6.2.11, when i = 0, can be rewritten as shown in eq. 6.2.12. Removal efficiency of DAF is, then, can be written in form of eq. 6.2.13.

)(0 κτ−= eNn {6.2.12}

)(0 11 κτη −−=−= eNn {6.2.13}

239


III-123

It should be noted that eq. 6.2.13 is about the same form as eq. 6.2.9b, but the value of κ, represented by eq. 6.2.11d, clearly represents effect of every interesting parameter, including retention time. So population balance model is flow-dependent and, then, can be used to extend the valid range of SIEM’s model, as described in section 6.2.3.

6.2.3 Generalized model of DAF from combination of filtration based model and population balance model

From section 6.2.1, it shows that SIEM’s filtration based model can be used to predict the efficiency of DAF at any diameters of oil droplets and bubbles, as long as design retention time and F/AV still conforms to SIEM’s condition. And from section 6.2.2, it shows that population balance model gives the equation that is flow-dependent. So it is possible to use these two models to extend the valid range of SIEM’s model to another operating condition.

To do so, firstly, SIEM’s model will be used to calculate the efficiency at his operating condition. Then, at any d and db, eq. 6.2.11d and 6.2.13 can be rewritten in form of a function of a parameter that we want to vary from SIEM’s condition, i.e., retention time and gas flowrate. while other parameters still conform to SIEM’s (eq. 6.2.14). When the efficiency at SIEM’s condition is known, the efficiency at other condition can be predicted from eq. 6.2.14 and relation between design value and SIEM’s value of that parameters (eq. 6.2.15 and 6.2.16).

τκτκτ Φ=Φ= 2.Gconst {6.2.14a} )( 211 τκκτη Φ−− −=−= ee {6.2.14b}

If “x” represents the variable Φ or τ that we want to vary from SIEM’s while another variable still conforms to SIEM’s condition, eq. 6.2.14b can be rewritten as show below.

xAe ⋅−−=1η {6.2.14c}

Where A = constant

If x = B.x ref {6.2.15}

Where B = Constant

Xref = x at SIEM’s condition

Then

)(

)(

11

refAx

Ax

ref ee

−

−

−

−=

ηη {6.2.16a}

)(

)(

11

ref

ref

Ax

ABx

ref ee

−

−

−

−=

ηη {6.2.16b}

Bref )1(1 ηη −−= {6.2.16c}

However, changing of τ of Φ will also cause some parameter in eq. 6.2.11d change. For example, increasing of retention time from SIEM’s condition will make V decrease. Then G will decrease. So the constant “A” at design condition is not equal to “A” at SIEM’s condition. In this case, eq. 6.2.16 is not valid. And it is not possible to know the value of “A” at design condition. So, the best estimation in this case is to use the value of A at some

240


III-124

condition that we are sure that will underestimate efficiency at design condition. Design reactor in this case will be larger than it should be. Then it can be considered as safety factor.

Guideline to predict the efficiency from combination of SIEM’s model and population balance model will be described again in details in section 6.3.

6.2.4 Influent parameters

From models described in section 6.2.1 and 6.2.2, main parameters that affect the efficiency of DAF can be summarized as shown in fig. 6.2.4-1.

Droplet size (d)

Efficiency (η)

100%

Lower limit ofthe model: 2 micron

Droplet size (d)

Efficiency (η)

Increase dbDecrease τ, Φ, H

100%

Fig. 6.2.4-1 Typical relation between efficiency of DAF and various parameters

From the models, it can be summarized that, efficiency of DAF can be improved by:

• Increase the size of oil droplets. This can be done by coagulation-flocculation process.

• Decrease the size and quantities of bubbles. This can be achieved by using high performance injection valve.

• Increase quantity of air/pollutant ratio by increasing pressurized water flowrate or using higher saturator pressure. However, operating cost should be considered.

• Increase column height to increase retention time.

However, there are some more phenomena and parameters that also effect the efficiency of DAF, which will be described below,

1. Characteristic of bubble-pollutant agglomerate and transfer compound

Characteristic of bubble-pollutant agglomerate

Since DAF performance depends on formation of bubble-pollutant agglomerate, some researches in GPI lab [13], [14] had been conducted to study characteristics of the agglomerate. Fig. 6.2.4-2 shows characteristic of bubble-solid and bubble-oil agglomerate.

For bubble-solid agglomerate, certainly, bubbles will attach to the surface of solid in side-by-side manner. On the contrary, for bubble-oil agglomerate, side-by-side characteristic is only temporary or intermediate condition. Finally, oil and bubble will integrate in form of spherical thin film or shell of oil with air core inside. This can be

241


III-125

explained by the concept of spreading, described in chapter 2, section 2.2.2.2. Consider interfacial tension between kerosene, water and air as shown below,

Oil-water: γow = 35 dynes/cm Air-water: γw = 72 dynes/cm Air-oil: γo = 28 dynes/cm

From chapter 2, oil will spread over the air bubble when γw > γo+γow. From the values of interfacial tension, it clearly shows that this condition is confirmed (72 > 28 + 35). So the oil will spread over air bubble and form a thin film around the air.

The four ways to improve the efficiency of DAF described earlier in fig. 6.2.4-1 is actually the way to improve collision probability. However, one of the most encountered problems in DAF operation is collision between bubble and pollutant without attachment. According to the study in static transparent model [13], to form agglomerate, bubble and pollutant must be collide or come close to each other at some certain distance for sufficient period of time to drain water film between them and made bond to each other. To solve this problem, some chemically active agents are used to adapt properties of bubble and wastewater. The chemicals are called “transfer compound”.

Effect of transfer compound on bubble-pollutant agglomerate

Transfer compound is chemical added either as gas to bubble-forming air or as chemical to water. Major role of the compound is to help promoting bubble-pollutant agglomerate and strengthen the bond between them. It will cause mass transfer, where it got its name, from bubble to water or vice versa that change local interfacial or surface tension of pollutant or bubble. This change will facilitate agglomerate formation. This effect of local change in surface tension is known as the Marongoni effect. There are 2 major types of the compounds, i.e., transfer gas and transfer compound for water.

• Transfer compound for water is practically a surfactant added to pressurized water. The surfactant will reduce surface tension of water. In case of solid separation, the effect is relatively the same as that when using surfactant in a washing machine. Bubbles and solids will attach to each other easier and stronger. Efficiency will be improved. This idea is actually commercialized by many companies, such as SAFTM (suspended air flotation) from Enprotec.

• Transfer gas is typically a gas of highly soluble chemicals, such as ammonia, which is added to bubble-forming air. The concept is that the gas would transfer from bubble into water and make the water film rapture easier. This will help interaction time between bubble and pollutant and make them attach immediately after collision. In GPI lab, AOUDJEHANE [13] shows that small amount of ammonium gas added into the air actually helps reducing coalescence time between bubble and oil droplet. However, it also promotes coalescence between bubble and bubble. So, these effects tend to cancel out each other. The efficiency is relatively the same.

In case of oily wastewater, however, presence of surfactant in water causes the droplet smaller and more stable. So it make the efficiency decrease. Anyway, AOUDJEHANE shows that if the surfactant is firstly added to oil and mass transfer of

242


III-126

surfactant is in the direction from oil toward water. It will help rupturing the water film then improving efficiency of coalescence between bubble and oil. However, doing so in real situation may not be possible because we cannot add surfactant into oil before it becomes wastewater.

a) Solid-bubble agglomerate. Please notice side-by-side formation. [14]

b) Bubble and oil droplet come close to each other.

c) The bubble and droplet collide. Water film between them is draining.

d) Water film rapture after been draing for some times. Oil starts to enclose air bubble

in very brief time.

e) Then, oil-bubble agglomerate is form of thin oil film around air core is generated.

Fig. 6.2.4-2 Solid-bubble agglomerate and formation of oil-bubble agglomerate [14]

2. Turbulence in flotation column

As shown in eq. 6.2.11 and 6.2.13, the efficiency of DAF depends on gradient or turbulence within the reactor. If turbulence increases, probability of collision between bubble and pollutants will also increases. However, turbulence also causes adverse effect on fragmentation of oil droplets or flocs. AOUDJEHANE had studied the effect of turbulence in special DAF model equipped with mechanical mixer, reported that efficiency of DAF increases with increasing turbulence or mixing intensity until some certain limit. Then the efficiency will be stable or slightly decrease with increasing turbulence. It seems that turbulence has both advantage and disadvantage. Normally, DAF is designed in smooth laminar flow region. Increasing the efficiency of DAF by increasing turbulence is to be done very carefully. Pilot-scale observation or CFD analysis should be conducted.

243


III-127

3. Presence of surfactant

Effect of surfactant in wastewater is as discussed in “effect of transfer compound”. In presence of surfactant, oil droplets will become more stable and their sizes will be smaller. Efficiency of DAF, then, will decrease. So it is recommended to destabilize (or break or crack), coagulate/flocculate the wastewater before using DAF.

4. Formation of air bubbles

From, filtration-based model efficiency of DAF will increase if the size of bubbles decreases. So pressurized water system shall be designed to make majority of very tiny droplets and avoid generating big bubbles. DUPRE [14] had studied bubble generation in venturi-style injection vale and reported that using hydrophobic material for valve construction may result in bigger bubbles. She also reported that addition of surfactants in pressurized water causes augmentation of population of microbubbles, while addition of polyelectrolyte gives the opposite result. Details of pressurized water system or saturator will be described in section 6.5.

6.3 Design calculation

From previous sections, it shows that there are many parameters, such as wastewater characteristic, reactor configuration and operating condition, that effect efficiency of DAF reactor. So, design of DAF reactor is generally based on hydraulic loading rate and some proven reactor configuration. However, to valorize the researches conducted in GPI lab, design procedure for DAF, recommended in this section, will be based upon the models, shown in the previous sections as well as general design practices. Calculation in each step is described below.

1. Cut size determination and required efficiency

The cut size can be determined from the degree of treatment required as well as from the limitation of the DAF processes. Cut size determination from degree of treatment is described in chapter 3. However, in general practice, DAF is hardly used alone but will be combined with coagulation-floculation process (see chapter 10). After coagulation-flocculation process, granulometry of size distribution of dispersed phase in the wastewater will be changed from that of initial wastewater. The size distribution after coagulation-flocculation may be roughly estimated, as shown in chapter 10. However, it is clear that it is no longer possible to determine the cut size from the initial size distribution. Then the degree of treatment (such as effluent oil concentration, etc.) will be used to calculate the required efficiency. This required efficiency or the required degree of treatment would be used to compare to general recommended criteria of DAF to determine if it is feasible to use DAF for such wastewater.

In case that there is no general design criteria or recommended efficiency for the wastewater to be treated, it is strongly recommended to perform DAF test, such as Flota-test, to evaluate the feasibility and efficiency of DAF before designing the real unit.

2. DAF reactor sizing

The size of the DAF can be determined based on (1) mathematics models from section 6.2, (2) general design criteria and (3) pilot-scale test result. Designing an efficient

244


III-128

DAF reactor, as well as other processes, is the state-of-art and requires experience. However, result from the models and some useful criteria acquired from many researches, esp. GPI researches, will be provided here to be a guideline for DAF reactor sizing.

2.1. General design criteria

Recommended design criteria from various literatures and manufacturers are summarized in table 6.3-1. It should be noted that the values in the table are summarized from experiences and researches with various operating conditions and types of wastewater, then, should be used as a guideline. It is recommended to review related researches for the type of wastewater to be treated. If possible, DAF test, using lab-scale batch test, such as flota-test, or pilot-scale test with the real wastewater should be performed.

2.2. DAF design by combination of filtration based model and population balance model

From the 2 models in section 6.2, they can be used to size DAF reactor and estimate graded efficiency of DAF process at any operating condition. If direct scale-up of SIEM’s model ( the same V and Φ/AV) is used, the cut size at H =0.70 m. is around 35 microns.

If DAF is used alone without coagulation-flocculation process and cut size can be determined, the cut size can be used to calculate surface area of DAF reactor (A) by assuming that the efficiency at the cut size is 100%.

If cut size can not be determined, approximate size of DAF reactor can be calculated from general criteria, described in table 6.3-1. It is recommended to use the reactor size calculated from the general criteria as a guideline to calculate the efficiency. Then, it can be fine adjusted again to fit specific constraint of each project.

Calculation procedure and model limitation will be as described below.

2.2.1. Design DAF at SIEM’s condition (�/AV is at SIEM’s condition. H can be varies)

1. To predict removal efficiency of DAF, reference graded efficiency (based on Qt) can be calculated by SIEM’s model. If the result from the equations is greater than 100%, then it will be rounded up to 100%. The values of (Φ/AV) have to be the same as SIEM’s condition (see item 3).

%1001))(

23(

,

exp

⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−=

Φ−

bdH

AVrefd e

αη

η {6.2.9b}

5919.0exp )(009005.0)( theoηαη = {6.2.9a}

diffIntsedtheo ηηηη ++=

2)(23

bInt d

d=η {6.2.3}

245


III-129

rc

wateroilsed V

gdμ

ρη

18

2/Δ

= {6.2.4}

3/2)(9.0br

Diff ddVKT

μη = {6.2.5}

c

bwaterairbr

gdUV

μρ

18

2/Δ

== {6.2.2a}

Where μc = viscosity of continuous phase (water) (L2/T) K = Boltzman constant (1.38*10-23) T = Absolute temperature (Kelvin)

2. To use the equations described above, the following conditions will be satisfied;

1) Inlet oil concentration should not be greater than 1,200 mg/l (before dilution) or 435 mg/l (after dilution). Using the model with higher oil concentration will result in underestimating of efficiency. • Φ/AV = 0.0516. Only this value must be used in the

equations. As long as this value is fixed, SIEM’s operating condition still holds and the model is still valid.

• Retention time, based on total flowrate (Qt), is around 25 minutes.

• Droplet diameter (d) tested is between 2 to 40 microns. • Diameter of air bubbles (db) varies from 15 to 130 microns.

Tested average bubble diameter is 70 microns, which is used to verify the model, and standard deviation of bubble diameters is 34.5 microns. The range of bubble sizes is common for commercial pressurized water system or saturator. The pressure of the test system is 4 atm (absolute).

• Tested air flowrate (Φ) is 0.42 cm3/s (4.2e-7 m3/s). • Tested wastewater flowrate (Q) is 3.9 cm3/s (3.9e-6 m3/s) • Tested effective water depth (H) is 0.70 m. The value of H

can be freely changed as long as (Φ/AV) is fixed. However, H between 1.8 to 2.7 is recommended by API [45].

• Diameter of flotation column is 0.15 m Cross section area of column (A) is 0.01767 m2.

• Ratio of pressurized water to wastewater (Qpw/Q) is 1.76. • Air to pollutants ratio used is around 0.12 kg. air/ kg. oil. • Ratio of number of bubble/ oil droplet tested is around 1.4

oil droplet/ 1 air bubble. • Hydraulic loading rate or flow velocity (V), based on Qt, is

1.6 m/h

2) he model is tested at the following operating condition;

246


III-130

Tabl

e 6.

3-1

Gen

eral

des

ign

crite

ria

of D

AF fr

om v

ario

us li

tera

ture

s and

man

ufac

ture

rs (h

ydra

ulic

load

ing

rate

is b

ased

on

tota

l flo

wra

te)

247


III-131

Tabl

e 6.

3-1

Gen

eral

des

ign

crite

ria

of D

AF fr

om v

ario

us li

tera

ture

s and

man

ufac

ture

rs (h

ydra

ulic

load

ing

rate

is b

ased

on

tota

l flo

wra

te)

248


III-132

3. To apply SIEM’s condition to other wastewater flowrate, area (Aref) and gas flowrate (Φref) corresponding to that flowrate can be calculated from the following equations. The subscript of A and Φ in this case is “ref” to indicate that SIEM’s condition (or reference condition) still holds.

01767.0109.3 6mod

mod

⋅×

=⋅= −req

elel

reqref

QA

QQ

A m2 (Q as m3/s) {6.3.1}

01767.0102.4 7

modmod

reqel

el

reqref

AAA −×

=Φ⋅=Φ m3/s {6.3.2}

4. If H is changed from 0.70 m. to Hreq, τ will be changed from 25 min. to τref by the following equation. The subscript of τ in this case is “ref” to indicate that SIEM’s condition (or reference condition) still holds.

36001

00046.0modmod

⋅=⋅= reqel

el

reqref

HHH

ττ hour {6.3.3}

5. Because of limitation of the pilot model, tested ratio of pressurized water to wastewater is quite high (around 92%), compared to that of general DAF for solid/liquid separation (less than 50%) [13]. However, API [45] recommended air/wastewater ratio of 0.35 std. ft3/ 100 gal of total flow for full-flow DAF process. This value is equivalent to 84% of 4-atm (abs) pressurized water/ wastewater. Anyway, it is interesting to adapt the model to calculate the efficiency at lower ratio of pressurized water by the procedure in item 2.2.2.

6. Tested hydraulic loading rate or overflow rate (based on Qt) is 1.6 m/h, which is relatively low, compared to normal rate of 3-15 m/h for domestic wastewater treatment. The value recommended by the American Petroleum Institute (API) [45] is between 4.8-6.1 m/h. So it is also interesting to adapt the model to calculate the efficiency at higher overflow rate by the procedure in item 2.2.2.

2.2.2. Design DAF at other condition from SIEM’s condition

To calculate graded removal efficiency at other operating condition than SIEM’s model, esp. at higher overflow rate or lower ratio of pressurized water/wastewater, calculation procedure will start from direct scale-up of SIEM’s condition (the same as 2.2.1). After that, adaptation by population balance model will be applied.

Direct scale-up of SIEM’s condition (Φ/AV is at SIEM’s condition. H can be varied.)

1. Calculate the reference efficiency (ηd,ref) of the model at required height (Hreq), average bubble diameter (db) and droplet sizes (d) using eq. 6.2.2 to eq.6.2.5 and eq. 6.2.9. Use Φ/AV = 0.0516 in order that the operating condition of SIEM still holds.

249


III-133

2. Scale up the area from 0.01767 m2 to required area (Areq). Other operating condition from section 2.2.1 still holds. So efficiency from section 2.2.1 remains the same. This required area could be approximated from recommended hydraulic loading rate (Vreq) and ratio of pressurized water to wastewater ((Qpw/Q)req) as shown in table 6.3-1.

req

reqpw

req

req V

QQQ

A⎟⎠⎞

⎜⎝⎛ +

=)(1

{6.3.4}

3. Find Φref, corresponding to the area Areq, by following equation,

01767.0102.4 7

modmod

reqel

el

reqref

AAA −×

=Φ⋅=Φ m3/s {6.3.2}

4. Find τref, corresponding to the height Hreq, by following equation,

6025

70.0modmod

⋅=⋅= reqel

el

reqref

HHH

ττ Hour {6.3.3}

Change Φ and τ from SIEM’s condition by population balance model

5. From population balance method, calculate κ2,ref corresponding to Areq, Hreq, τref and Φref from the reference efficiency (from item 1) by the following equations. Please note that, at this point, SIEM’s condition still holds. κ2,req has to be calculated separately for each droplet diameter.

refref

refdref τ

ηκ

⋅Φ

−−=

)1ln( ,,2

{6.3.5a}

Or )(

,,21 refrefreferefd

τκη Φ−−= {6.3.5b}

6. Find Φreq from required ratio of pressurized water to wastewater (see table 6.3-1 for the recommended value) by following equations.

QRVolumeVolume

QVolumeVolume

pw

airpw

pw

airreq ⋅⋅=⋅=Φ {6.3.6a}

Then, from Henry’s law (section 6.5)

)10082.0()('3

3

KmolmTPPQRHy atmreq ⋅

⋅−⋅⋅⋅=Φ − {6.3.6b}

For air, y =1. H’ in this case in in the form of molair/(m3 water. atm). Henry’s constant of air at any temperature can be calculated from an empirical equation, shown in section 6.5.

7. Choose τreq from recommended criteria (see table 6.3-1).

8. To change Φ and τ from SIEM’s, the following procedure is recommended and precautions should be noted.

250


III-134


This will cause decreasing of V, so G will decrease. Then κ2 (see eq. 6.2.11, 6.2.14) will be lower. However we do not know how much exactly. So, to be on the safe side, we will assume that only Φ decrease but τ = τref. In this case, κ2 will remain the same and be equal to κ2,ref. The efficiency can be estimated by eq. 6.3.7a.

)( ,21 refreqrefedτκη Φ−−= {6.3.7a}

Because we use τref, instead of τreq, the calculated efficiency will be lower than the real value.


This will cause increasing of V, so G will increase. Then κ2 (see eq. 6.2.11, 6.2.14) will be higher. Again, we do not know how much exactly. So, to be on the safe side, we will assume that κ2 = κ2,ref. The efficiency can be calculated by eq. 6.3.7b. And again, the calculated efficiency will be lower than the real value.

)( ,21 reqreqrefedτκη Φ−−= {6.3.7b}


This can be done by increasing pressurized water flowrate. However, the ratio of pressurized water/ wastewater is already high (92%). So there is only a small gap to increase Φ (from Qpw/Q = 92% to 100%). In this case, κ2(see eq. 6.2.11, 6.2.14) will be higher. Like the former case, the efficiency can be calculated by eq. 6.3.7b.


This case is not feasible because it means that we have to decrease wastewater flowrate and increase pressurized water flowrate. As stated above, the ratio of pressurized water/ wastewater is already high (92%). If we decrease wastewater flow, quantity of pressurized water flow will exceed that of wastewater, which is not feasible because we have to recycle effluent at 100% plus additional makeup water to feed the pressurized water system.

There is no obvious limit for the 4 adaptations, shown above. However it is recommended to use the values of each parameter (d, db, C, etc.) within general range, shown in item 2.2.1 and table 6.3-1.

The procedure described in paragraph 2.2 will result in efficiency of DAF reactor corresponding to firstly approximate size. If the calculated efficiency dose not meet the required efficiency, the reactor size will be adjusted until it gives a satisfying efficiency.

251


III-135

2.3. DAF test

It must be noted that there are many other parameters that can affect efficiency of DAF, such as presence of coagulation-flocculation process, reactor hydraulic design, contact zone configuration, design of pressurized water system (or saturator) and injection valve, etc. These parameters can cause some discrepancies in efficiency prediction especially when coagulation-flocculation is used since the granulometry of influent will be totally changed. In this case, pilot-scale or small batch experiment, such as Flota-test, will provide valuable data for design propose and can be used to compare with model result to determine the final design criteria. In fact, it is strongly recommended to perform DAF test, if possible, every time before designing DAF system. Pictures of DAF test are as shown in fig. 6.3-1

a) Pilot-scale for DAF test b) Flota-test set (Source: GPI lab)

c) Flota-test of cutting oil emulsion with

coagulant addition. Notice the float at the surface of the reactors. (Source: GPI lab)

d) Magnified photos of non-coagulant float (left) and flocculated float from Flota-

test(Source: GPI lab)

Fig. 6.3-1 Pilot-scale DAF test and Flota-test

3. Removal efficiency and outlet concentration

3.1. Outlet concentration

Outlet concentration and graded concentration of oil after DAF (C and Cd) can be calculated by the following equation

diloddout

d CQ

QC ,)1( η−= {6.3.8a}

252


III-136

∑ −=max

min

,)1(d

ddilodd

out

CQ

QC η {6.3.8b}

∑−=max

min

,

d

ddilodd

dout CQQQ η

ρ {6.3.8c}

Co,dil and Cod,dil are the inlet concentration after dilution with pressurized water. Q/Qout represents the effect of flow splitting between treated water and oil outlet port. In lab scale, where pressurized water comes from oil-free water, outlet concentration can be written in form of Co and Cod, which are the initial concentration of oil in wastewater before dilution by pressurized water, as shown in eq 6.3.8d. However, in this case, effluent quantity will increase from the initial wastewater flow (Q) to total flow (Qt), which is summation of wastewater flow and pressurized water flow.

∑ −⋅=max

min

)()1(d

d todd

out QQC

QQC η {6.3.8d}

If pressurized water comes from treated effluent, remaining oil in the effluent will effect the inlet concentration after dilution. The value of Cod,dil and Co,dil in this case can be calculated with accounting for mass balance, as shown in item 3.4.

3.2. DAF efficiency

DAF efficiency (ηDAF), which is the efficiency based on flotation effect alone, can be calculated by the following equation.

( ) %100)1(1 max

min

, ⋅⋅−⋅= ∑d

ddilodd

oDAF C

Cηη {6.3.9a}

In lab scale, where pressurized water comes from oil-free water. DAF efficiency can be written in form of Co and Cod, which are the initial concentration of oil in wastewater before dilution with pressurized water, as shown in eq 6.3.9b.

( )( ) %100)1(1

)(

)()1()1(max

min

max

min

max

min

,

,

⋅⋅−=⎟⎟⎠

⎞⎜⎜⎝

⎛⋅−

=⋅−

= ∑∑∑ d

dodd

o

to

d

d todld

dilo

d

ddilodd

DAF CC

QQC

QQC

C

Cη

ηηη

{6.3.9b}

However, if treated water is recycled to the pressurized water system, the values of Co,dil and Cod,dil will be affected by remaining concentration in recycled stream and will be required more complex calculation as shown in item 3.4.

3.3. Total removal efficiency

When pressurized water comes from DAF effluent: Total removal efficiency (ηt), which is defined as ratio between oil mass removed from water and initial oil mass, can be calculated by the following equation.

%100⋅−

=o

ot C

CCη {6.3.10a}

When pressurized water comes from additional clean water: Total removal efficiency (ηt) can be calculated by the following equation. Please note that effluent quantity in this case is equal to Qt, not Q.

253


III-137

%100)(

⋅−

=⋅−

=o

to

o

tot C

QQ

CC

QCQCQC

η {6.3.10b}

3.4. The value of Cod,dil and Co,dil when DAF effluent is used in the pressurized water system

If the effluent from DAF is recycled to pressurized water system, some oil left in the effluent will be returned to the system. In this case, mass balance of oil has to be taken into account. Thus, Cod,dil will be modified by adding this return oil repeatedly, as shown in eq. 6.3.11. Theoretically, rmax in eq.6.3.11 should be infinity. The value of Cod,dil will eventually convert to an asymptote. Anyway, using rmax around 30 will practically give the result sufficient accuracy, esp. when ηd > 20%. However, if coagulation-flocculation is used before DAF process, the value of ηd will be so high that Cod,dil in this case is only slightly higher than Cod.Q/Qt.

t

odr

r

rd

t

pwdilod Q

QCQ

QC

x

⋅⎥⎥⎦

⎤

⎢⎢⎣

⎡−= ∑

=

−max

1

1, )1( η {6.3.11a}

Where Qpw = R.Q. Above summation can be simplified as shown in eq. 6.3.11b.

t

od

d

rd

dilod QQC

RR

RR

C ⋅−−

+

−−+=

−

1)1(1

)1))1(1

(( 1

,

max

η

η {6.3.11b}

( )∑=max

min

,,

d

ddiloddilo CC {6.3.11c}

4. Energy required

DAF reactor does not require extra energy to make it function. The energy is required only to feed the water and pressurized water into the tank, then the water will flow, naturally, through the tank by gravity. Pressure drop across the tank and piping system depend on tank and piping design. This pressure drop can be calculated by general hydraulic equations, such as Darcy-Weisbach’s, Manning’s, or Hazen-William’s equation, weir equation, orifice head loss equation, thus will not de described here.

However, it is the pressurized water system that requires most energy in DAF system. Design of pressurized water system or saturator will be described separately in section 6.5

6.4 Design consideration and Construction of DAF reactor

1. Reactor material

DAF reactor can be made of concrete, steel or any material that can withstand wastewater characteristic and operating condition. Special protective coating may be necessary in case of corrosive wastewater. Otherwise normal coating is sufficient.

254


III-138

2. Reactor geometry and internal baffle configuration

As shown in section 6.2.4 that turbulence in reactor effects efficiency of DAF for it effects probability of collision between oil droplets and air bubbles. Theoretically, efficiency increases with increasing turbulence or velocity gradient (G). This concept could be applied in DAF without coagulation. However, for DAF with coagulation-flocculation, high gradient may result in re-fragmentation of flocs, which makes efficiency decrease. Besides, more turbulence may cause some undesirable eddy, which can cause carry-over of droplet along with effluent. So geometry or configuration of internal baffle in the tank are usually designed to minimize undesirable turbulence. However, to maximize contact or collision between bubbles and oil droplets, as well as, to ensure uniform flow distribution in the tank, DAF reactor is normally equipped with inlet baffle as shown in fig. 6.4-1. The configuration of internal baffle is based mainly of experience. Nowadays, as computerized fluid dynamics (CFD) tool is easier affordable, it is wildly used as a tool to design or perfect the configuration of DAF reactor

There are some attempts to design of DAF reactor to imitate the condition within the flota-test, which is closest to quiescent condition. Examples of this design are KROFTA reactor and DAF Corp. reactor, as shown in fig. 6.4-2. In KROFTA tank, wastewater and pressurized water are fed by the rotating arm into the reactor in order that there is least lateral velocity.

For tank geometry, DAF reactor can be designed as rectangular or circular tank. Both shape of tank, when properly designed, can operate at about the same efficiency, even though some literatures [49] might claim that circular shape is superior for its better bubble and water distribution. So other factors, such as shape of land, convenience of construction, availability and O&M cost of necessary equipment, such as scraper, skimmer, etc. should be taken into account to determine the most suitable shape of tank.

3. DAF necessary equipment and component details

Like other treatment processes, DAF performance depends both on proper calculation and proper equipment and component design. Nowadays, there are a lot of packaged DAF systems and commercial products related to DAF process. Designer or owner can contact various suppliers for more details. However, in this chapter, some details of necessary equipment and component details will be provided as a guideline of what is necessary to consider in DAF process.

Fig. 6.4-1 is a good example to show necessary equipment and components of DAF system. Brief of each equipment and component will be as described as follow.

• Inlet port or inlet channel: Inlet port or inlet channel should be designed to guarantee well bubbles and water distribution. Details of inlet port might vary from simple branch pipes with throttling valves to big manifolds with flow control devices (fig. 6.4-2). For circular tank, water is normally fed at the center feed well of the tank. In this case, annular perforated pipe will be used. Combination of good inlet port and inlet baffle, described in previous item, will contribute to good performance of DAF system. Pressure reducing valve or injection value or injector of pressurized water is one of the most important component of DAF system for it plays important role in bubble

255


III-139

formation. The finer the bubbles, the better the efficiency. The pressure reducing valve will be described in section 6.5.

• Effluent outlet: Effluent outlet is normally designed as weir or bell mouth pipe. It plays important role to control water level in the reactor so it should be designed to be adjustable. Effluent from DAF will overflow off the weir of bell month into gutter or trough, which is normally equipped with a pipe to recycle some portion of effluent to saturator system. For rectangular channel, outlet port normally locates at the opposite end of inlet port. However, in some designs, outlet port will locate at the same side as inlet port to maximize agglomerate paths. Internal baffle or lateral draw-off pipe should be properly located to ensure that only clarified water will be withdrawn to the weir or bell mouth.

• Scum or float skimmer: air-pollutant agglomerate will float and form a layer at the surface of the tank. The float layer is normally removed by means of skimmer. Typical skimmer for rectangular tank is chain and flight type. For circular tank, skimmer is normally rotating skimmer blade, attached to the same central driving unit or rotating bridge as the bottom scraper. In some designs, skimmer for circular tank is designed as a rotating scoop (see fig. 6.4-2) to scoop off the scum and transport via its center trough that also serves as a shaft to scum hopper. Scum hopper or scum receiving structure will be placed a little higher than water outlet port in order to prevent water from overflow into the structure.

• Sludge draw-off: Settleable solids are normally present in the wastewater and will be settled within any tanks, include DAF reactor. So sludge draw-off system should be provided. Components of the system might vary from simple draw-off pipes to sludge hopper with sludge scrapper. For circular tank, scraper will be either peripheral driven or central driven type. For rectangular tank, there are several types of scraper, i.e., chain and flight, auger or screw conveyor.

4. Add-on process of DAF

Coagulation-flocculation: Efficiency of DAF depends on many factors included influent oil droplet sizes. Even though DAF can be used as stand alone process, its performance is normally erratic esp. when oil droplet size is very small. To make the performance rather steady, coagulation-flocculation process is usually included to increase the size of pollutant to be floated. Coagulation-flocculation process consists of 2 main components, i.e., rapid (or flash) mixing part for coagulant-water mixing, and flocculation (or slow mixing) part for floc formation. There are a lot of coagulants using for wastewater treatment, such as metal salts, polyelectrolyte. Type and dosage of coagulant vary with wastewater characteristic and can be archived by pilot testing, such as jar test or flota-test. Details of this add-on process will be described in chapter 10. Pictures of coagulation-flocculation from flota-test of cutting oil emulsion are shown in fig. 6.3-1. Examples of case studies on oily wastewater treatment are shown in table. 6.4-1.

256


III-140

Fig.

6.4

-1 E

xam

ple

of n

eces

sary

equ

ipm

ent a

nd c

ompo

nent

det

ails

of D

AF sy

stem

(Sou

rce:

Env

iron

Tre

atm

ent S

yste

m)

257


III-141

Table. 6.4-1 Data on efficiency and coagulant concentration of various oily wastewater treatments by DAF [51]

Oil concentration Wastewater Coagulant

(mg/l) Influent Effluent %Removal

Refinery 0 125-170 35-52 70-72 100 alum 100 10 90 130 alum 580 68 88 Oil tanker ballast water 100 alum + 1 polymer 133 15 89 Paint manufacture 150 alum + 1polymer 1900 0 100 Aircraft maintenance 30 alum + 10 activated silica 250-700 20-50 >90 Meat packing 3830 270 93 4360 170 96

Addition of transfer compound: Sometimes, efficiency of DAF is limited by probability of air-pollutant agglomerate. In case of oily wastewater, the most frequent encountered problem is ineffective collision between bubble and oil droplets (collision without agglomerate forming). There are some attempts to solve this problem by adding chemically active agent, called transfer compound, to change some properties of bubble or oil droplet to facilitate oil-bubble agglomerate forming. These chemicals can be added to either wastewater or as a gas to bubbles. Examples of these chemicals are ammonium gas, surfactants, etc. In GPI lab, DUPRE [14] had studied the effect of various transfer compound and concluded that addition of surfactant in pressurized water can cause augmentation in population of microbubbles. On the contrary, addition of polyelectrolyte will cause decreasing in electrical charge, then favor coalescence of bubbles.

Even though this method is used with some success for water-solid separation, which the chemical helps forming stronger bond between bubble and solid surface, there is no clearly evidence, at least in GPI lab, that it is feasible on oily wastewater treatment (see section 6.2.4). So it is recommended to perform DAF test with the real wastewater before design this add-on system. Otherwise, coagulation-flocculation may be more suitable.

a) KROFTA’s inlet manifold with flow control devices (Source: KROFTA)

b) Example of inlet branch pipes (Source: Leopold)

Fig. 6.4-2 Necessary equipment and reactor components of DAF system

258


III-142

c) Inclined inlet baffle at the center of the

tank (Source: WR DAF) d) Perforated pipe inlet

(Source: HUBER)

e) Bell mouth pipe outlet (Source: ETS) f) Example of weir outlet (Source: HUBER)

g) Lateral perforated pipes for effluent

draw-off (Source: Leopold) h) Chain and flight scum skimmer

Fig. 6.4-2 Necessary equipment and reactor components of DAF system (Con’t)

259


III-143

i) Rotating scoop skimmer (Source: DAF corp.)

j) Detail of rotating scoop skimmer (Source: DAF corp.)

k) Example of CFD analysis on bubble

density in DAF reactor (Source: InfilcoDegrémont)

l) Example of CFD analysis on velocity distribution in DAF reactor (Source: InfilcoDegrémont)

m) Typical flow pattern in DAF system (Source : InfilcoDegrémont)

n) Folded flow pattern in DAF reactor (Source: WR DAF)

o) Rectangular DAF reactor (Source: WR DAF)

p) Circular DAF reactor (Source: KROFTA)


260


III-144

q) Example of shallow DAF reactor (Source: WR DAF)

r) Graphical image of shallow DAF reactor (Source: KROFTA)

s) Feed manifold and rotating internal baffle in shallow DAF reactor (Source: WR DAF)

t) 4 effluent draw-off pipes in shallow DAF reactor (Source: WR DAF)

u) Flocculation tank (Source: Aqua- pak systems)

v) Flocculation pipe (the array of pipe beside DAF reactor) (Source: WR DAF)


5. Characteristic of float or scum

Characteristic of float or scum of DAF depends on characteristic of wastewater. Fig. 6.4-3 shows some examples of scum from DAF. However, for refinery wastewater, petroleum or oil-related industries, which oils are components of the wastewater, scum will be

261


III-145

rich in oil. Some literatures [50] report that oil concentration in DAF scum is as high as 50% by weight for DAF process with coagulation-flocculation. For oil-rich scum, it could be possible to use oleophilic oil skimmer, which is widely used for API tank.

However, solids and other pollutants are usually present in oily wastewater, the scum, then, contains these solids and pollutants and is in form of thick froth layer. DAF system can produce scum, which is normally more dense than settled sludge from settling tank. Average concentration of 1% (solid content) could be expected. Higher concentration, such as 3%, is reported by many sources [51].

a) Scum for petroleum industry (Source:

www.environmentalleverage.com) b) Scum from high rate DAF reactor

(Source: InfilcoDegrémont)

c) Scum from circular DAF reactor

(Source: KROFTA) d) Example of side window of DAF tank. Notice

float layer and bubble (Source: KROFTA)

Fig. 6.4-3 Examples of characteristics of scum from DAF processes

6.5 Pressurized water system or saturator

6.5.1 Working Principle and design calculation

Because air bubble is the heart of DAF operation, pressurized water system, then, is a very important component of DAF for it is the source of air bubbles. Basic principle of pressurized water system or saturator is to dissolve the air into the water at high pressure. After that the pressurized water will be injected into the wastewater at operating pressure, normally ambience. When the pressure decreases, solubility of air in water will decrease so the excess air will become gas and cause air bubble within the water.

262


III-146

There are 3 major modes of pressurization in DAF process, i.e.,

• Full-flow pressurization: In this mode, whole wastewater is pressurized. It might operate at lower pressure since the quantity of pressurized water is high, low dissolved concentration of air is sufficient. Reactor in this case is designed only to handle wastewater quantity. However, pollutants, esp. suspended solids, may cause some problems to the saturator system, esp. injection valve and packed saturation tank. It can not be used if coagulation-flocculation is employed for it may cause re-fragmentation of floc by the saturator packing.

• Partial flow pressurization: In this mode, only some portion of wastewater is pressurized and then sent to mix with the rest of wastewater to reactor. However, pollutants in wastewater might still cause some problem to the saturator system.

• Recycled flow pressurization: This mode of operation is the most popular. Some portion of treated water is recycled to use as pressurized water. Quantity of the water is relatively good for it has been treated. So it does not cause any problem to the packed saturation tank. However, the reactor has to handle wastewater plus recycled flowrate.

However, the pressurized water system for every mode of pressurization is identical. Good saturator system has to provide a number of very tiny bubbles. Characteristic of good pressurized water is that, when it is discharged into the reactor, it will have cloudy and milk appearance for relatively long period of time, as shown in fig. 6.5-1.

a) Example of propagation of bubbles generated by pressurized water when injected

into the top of test tank.

b) Cloudy, milky appearance of bubbles generated by pressurized water

Fig 6.5-1 Example of good bubble formation from pressurized water (Source: Cornell DAF pump)

263


III-147

From GPI research [12], theoretical equations, such as Henry’s law and Dalton’s law, can be used to calculate pressurized water system with relatively high accuracy. To size the pressurized water system, the following procedure is proposed.

1. Quantity of air or gas in pressurized water

Solubility or concentration of air in water in saturator is practically governed by Henry’s law (eq. 6.5.1).

HyPx = {6.5.1}

Where x = molar fraction of dissolved gas in water (mol/mol) Y = molar fraction of gas in air (mol/mol) For air: y = 1 For oxygen:

y(O2) = 0.21 (=0.11 in permanent regime) For nitrogen: y(N) = 0.79 (= 0.89 in permanent regime)

P = absolute pressure of saturator (atm, or conforms to unit of H used)

H = Henry’s constant, depending on type of gas (20oC), i.e., For air: H = 4.04x104 atm/mol For oxygen: H = 8.04x104 atm/mol For nitrogen: H = 6.64x104 atm/mol

Eq. 6.5.1 can be used to calculate the value of “x” or molar solubility of gas in saturator at any given pressure (P). Normally, saturator is operated at around 4 atm (abs). Concentration of air or gas discharged within DAF reactor at ambience condition can be calculated by eq. 6.5.2.

1000)(

)().( ⋅

⋅⋅=

waterMWgasMWx

gasConc waterρ mg/l {6.5.2a}

Where MW(gas) and MW(water) are molecular weight of gas and water respectively.

Substitute x from eq. 6.5.1.

PAHwaterMW

gasMWPygasConc water ⋅=⋅

⋅⋅⋅⋅

= 1000)()(

).(ρ {6.5.2b}

1000)(

)(⋅

⋅⋅⋅

=HwaterMW

gasMWyA waterρ {6.5.2c}

To facilitate calculation of dissolved air or gas, Henry’s constant can be expressed in the form of molair(or gas)/(m3 water.atm), called H’. Eq. 6.5.2d can be used to calculate the value of H’ at any temperature. Temperature is in Kelvin. From this concept, eq. 6.5.2b can be rewritten as shown in eq. 6.5.2e.

2472.1)273(02745.0)273(00039.0)273(000002.0' 23 +−−−+−−= TTTH {6.5.2d}

1000')().( HgasMWPygasConc ⋅⋅

= mg/l {6.5.2e}

For quantity of gas discharged in DAF reactor, it can be calculated from difference between concentration of dissolved gas in at saturator’s pressure and that at ambience pressure, 20 as shown in eq. 6.5.3.

264


III-148

))/(10082.0()(' 33 KmolmTPPQRHy atm ⋅×⋅⋅−⋅⋅⋅⋅=Φ − m3/s {6.5.3a}

at T = 20oC (293 K)

)(33

)( )/10026.24(' gaugegaugegas PBmolmPRHy ⋅=×⋅⋅⋅⋅=Φ − m3/s {6.5.3b}

Every constant in eq.6.5.2 and the value of “A” for eq. 6.5.2c, as well as, “B” for eq. 6.5.3b are tabulated in table 6.5-1. Table 6.5-2 shows the value of “A” at various temperature. The value of “A” and “B” in the table are theoretical value under an assumption that the water is saturated by gas. In fact, general saturator can dissolve gas into water at around 70-95% of saturated concentration. So value of Conc.(gas) and φgas from eq. 6.5.2c and 6.5.3b should be multiplied by the factor of 0.70 to 0.95 to yield the practical or observed value.

Table. 6.5-1 Constant for calculation of concentration and quantity of gas for saturator design (Operating pressure of DAF = Patm, T = 20oC)

Gas Y

(mol gas / mol water)

H (atm/mol)

MW(g/mol)

A ((mg/l)/atm)

B ((l gas/l water)/atm. gauge)

Air 1.0 4.04x104 28.95 24.17 0.020

Oxygen (O2)

0.21 8.04x104 32 9.222 0.0076

Table. 6.5-2 Air characteristic and solubility at Patm [51]

Temperature(oC)

Weight solubility: A((mg/l)/atm.)

Volume solubility: B ((ml gas/l water)/atm gauge))

Air density (g/l)

0 37.2 28.8 1.293

10 29.3 23.5 1.249

20 24.2 20.1 1.206

30 20.9 17.9 1.166

40 18.5 16.4 1.130

50 17.0 15.6 1.093

60 15.9 15.0 1.061

70 15.3 14.9 1.030

80 15.0 15.0 1.000

90 14.9 15.3 0.974

100 15.0 15.9 0.949

2. Energy required

Main energy consumption for saturator is the energy for 1) compressing the gas into water at required rate and 2) pumping the water at design pressure and flowrate. Normally, pressure drop of piping system is relatively low, compared to working pressure of the saturator, then can be negligible.

265


III-149

Energy for compressor can be practically calculated by adiabatic process equation (PV1.4 = Constant) with satisfactory accuracy (see eq. 6.5.4). ηcomp is overall efficiency of compressor, which can be safely assumed to be around 60-70%.

pwatmcomp

QgasMWgasConc

PPRTPower ⋅⋅

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡−⎥

⎦

⎤⎢⎣

⎡⋅=

−

)()(1

4.01

)4.1

14.1(

η {6.5.4}

Using R = 8.314 Pa.m3/(mol. K), Conc(air) in g/l and Qpw in m3/s will result in power as Watt. Molecular weight (MW) of air is 28.95 g/mol. Fig. 6.5-2 shows relation between calculated power required for compressor and absolute pressure of saturator (P) for 10 m3/h of pressurized water.

Energy for pressurized water pump can be practically calculated by general equation by assuming overall efficiency of the pump (ηpump) around 60-70%.

pump

gaugepw

pump

atmpw PQPPQPower

ηη)()( ⋅

=−⋅

= {6.5.5}

Using Qpw in m3/s and pressure (P) in Pa. will result in power as Watt. Fig. 6.5-2 shows relation between calculated power required for pump and absolute pressure of saturator (P) for 10 m3/h of pressurized water.

Energy for DAF pump The DAF pump is a special type of saturator that integrate the function of compressor and pressurized water pump into a single machine. Details of the pump will be described in the next section. Its energy required is theoretically equal to summation of energy for compressor and energy for pressurized water pump. However, single step estimation, using eq. 6.5.5 alone with ηpump around 50-60% is acceptable. The exact energy required can be obtained from manufacturer’s information.

0

20

40

60

80

100

120

140

160

180

200

0 1 2 3 4 5 6 7 8

Saturator pressure (atm. (absolute))

Pow

er re

quire

d fo

r com

pres

sor

(Wat

t / 1

0 m

3 /h o

f wat

er)

0

500

1000

1500

2000

2500Po

wer

requ

ired

for p

ump

(Wat

t / 1

0 m

3/h

of w

ater

)

Compressor(η = 100%)

Compressorη = 70%

Pump(η = 100%)

Pumpη = 70%

Fig. 6.5-2 Relation between power required for pump, compressor and absolute pressure of saturator for

pressurized water flowrate of 10 m3/h (assume %air saturation = 95%)

266


III-150

6.5.2 Type of saturator and injection valve

1. Saturator

Saturator can be divided into 2 main categories, i.e.,

• Combination of pump and saturation tank: Schematic diagram of this system is generally as shown in fig. 6.5-3. Working principle of the system is relatively the same as packaged booster pump or pump/pressure tank system in water distribution work. However, the saturation tank has to be specially designed to maximize air dissolution into water. For conventional system, the tank is designed as packed column, using ring packing or likewise, as shown in fig. 6.5-4. Saturation tank in this case is classified as pressure vessel, so, in some countries, its design must conform to applicable law or regulation of pressure vessel.

• In some designs, the pressure tank is replaced by other air-water mixing device, which is equivalent in performance but less complex and less expensive. An example of this device is mixing pipe as shown in fig. 6.5-4.

• Generally, air compressor for DAF is installed. However, in case that there is common compressed air supply, it can be readily used in DAF. In few design, the pressure tank with automatic air intake system, which is widely used in packaged pump-pressure tank system, is used.

• DAF pump: This device is combination of pump and air saturation system into one machine. Air will be either sucked from ambience by the pump itself or supplied by compressor into pump suction or pump casing. The air will mix with the water within the pump. It eliminates the use of saturation tank. Schematic diagram and examples of DAF pump are as shown in fig. 6.5-3 and 6.5-4.

2. Injection valve

Injection valve or pressure reducing valve or pressure release valve is a device that practically generates air bubbles at the working point. The valve will reduce the pressure of pressurized water to ambience, which cause excess dissolved gas to convert to gas bubbles. In GPI lab, there are few researches on the value [14], [15]. From the researches, it shows that the gas will be firstly generated in form of large gas pocket and then will be fragmented by hydrodynamic force to form microbubbles (see fig. 6.5-5). So geometry of the valve, which determines hydraulic condition, is the key parameter to obtain microbubbles. Many types of valves are shown in fig. 6.5-5. Selection of the valve is based mainly on experience. However, it is recommended to install the value as close to mixing zone between wastewater and pressurized water as possible.

267


III-151

a) Conventional pressurized water system or saturator system

b) Saturator system using DAF pump

Fig. 6.5-3 Schematic diagrams of saturator systems (Source: Edur pump)

268


III-152

a) Packed tower to air saturation (Source: Leopold)

b) Packed column for air saturation (Source: Enprotec)

c) Air mixing tube for air saturation (Source: DAF corp.)

d) Installetion of DAF pump (Source: WR DAF)

e) Example of DAF pump (Source: Edur pump)

f) Graphical image shows working principle of DAF pump (Source: Hellbender pump)

Fig. 6.5-4 Examples of saturator system

269


III-153

a) Examples of injection valves [14]

b) Magnified photos of air pocket formation and fragmentation in convergence- divergence nozzle type injection valve [14]

Fig. 6.5-5 Examples of injection valve

6.6 Variations, advantage and disadvantage of DAF

There are many variations of DAF system in many ways such as tank shape, flow pattern, equipment, loading rate, as described before in this chapter. So selection and design of DAF is state-of-art, like many other processes, and based mainly on experience. However, data from DAF test, such as Flota-test, and studies on successful projects for the same or related type of wastewater are useful to design DAF system.

Advantages:

1. It requires less footprint area than conventional decanter. 2. Since it is accelerated process, we have some control over many operating

parameters, such as coagulant dosage, quantity of pressurized water, etc. Then it has more flexibility to handle variation of wastewater characteristic.

3. DAF is less effected by temperature, compared to other processes.

270


III-154

Disadvantages:

1. Capital cost and operating is higher than conventional decanter.

2. DAF system consists of a lot of equipment, so the maintenance is higher than conventional decanter. Moreover, it may require more skill of the operator.

3. Its efficiency depends on many parameters as described before, so it may decrease if some parameters in not optimized.

4. Coagulation-flocculation process is normally required. This causes more expense on chemicals.

271


III-155


7.1 General

Hydroyclone is an accelerated separation process. Its main working concept is to replace the gravitational acceleration that governs decanting velocity by higher centrifugal acceleration. Hydrocyclones are widely used in many processes, i.e., classification and separation between solid-liquid and liquid-liquid. Unlike other centrifugal machines, driving force of hydrocyclone is generated solely from its inlet velocity. The higher the velocity (or flowrate), the higher the efficiency. Or it can be implied that, at the same flowrate, smaller will have higher performance that the smaller one. This is the most interesting advantage of hydrocyclone.

Commercial hydrocyclones come in various shapes (single conical, long double conical, etc.) and configurations (co-current or counter current, based on direction of water and oil outlet flow). However, because it is used to make the separation between 2 phases (such as, liquid-liquid, solid-liquid.), it will be called here as “two-phase hydrocyclone”. In GPI lab, almost all of the researches are based on this type of hydrocyclone. In this chapter, we will consider particularly on application of hydrocyclone on oil/water separation. For oil/water separation, the hydrocyclones usually come in the form of elongated conical hydrocyclone, as shown in fig 7.1-1a. Vortex or spiral movement in hydrocyclone will make oil move inward to the center where it is purged to an overflow port, as shown in fig. 7.1-1. Treated water will flow out at an underflow port. Studies on two-phase hydrocyclone for oil/water separation will be described in section 7.2.

However, there is another type of hydrocyclone that is initiated by AURELLE and MA [16] in GPI lab. This special hydrocyclone is an innovation designed for simultaneous separation of oil and suspended solids from water. Studies on this “three-phase hydrocyclone” will be summarized in section 7.3.

a) Graphical image shows vortex and oil

core in 2-phase counter current hydrocyclone (Source: Ultraspin)

b) Graphical image of three-phase hydrocyclone [16]

Fig. 7.1-1 Basic flow pattern and examples of hydrocyclones

272


III-156

120255

53

428

10

50

Entrée

Sortie eau

Sortie huile

Entrée

Sortie eau

Sortie huile

Tube d'aspiration

Surface intérieure

Surface extérieure

c) Industrial scale counter current 2-phase hydrocyclone (Source: Krebs)

d) Co-current lab scale 2-phase hydrocyclone (Source: GPI lab)

Fig. 7.1-1 Basic flow pattern and examples of hydrocyclones (Con’d)

7.2 Two-phase hydrocyclone


As mentioned earlier, working principle of hydrocyclone is based on modifying an acceleration in STOKES law (eq. 7.2.1). From the law, rising or decanting velocity (Ud) of oil droplet is proportional to acceleration (a). Generally, this acceleration means gravitational acceleration (g). However, in centrifugal machine, liquid is forced to spin or centrifuged around the axis of the machine. So the liquid is subjected to another acceleration, which is centrifugal acceleration. If the liquid spins fast enough, the centrifugal acceleration (ac) will overcome the gravity acceleration. The decanting velocity of oil droplets, as well as, removal efficiency, will increase. The higher the centrifugal acceleration is, the better the efficiency is.

cd

daUμ

ρ18

2⋅⋅Δ= {7.2.1}

Where d = Diameter of dispersed phase, in this case, oil droplets a = Corresponding acceleration of oil droplets Δρ = Difference between density of dispersed phase and continuous phase μC = Dynamic (or absolute) viscosity of continuous phase, which is water, for

oily wastewater Ud = Rising or decanting velocity of the droplet diameter “d”

So, efficiency of hydrocyclone can be described in term of ratio between centrifugal and gravity acceleration, called the factor of acceleration (ζ), as shown in eq. 7.2.2 .For hydrocyclone, the driving force that makes the liquid rotate or spin comes from energy of its own inlet flowrate. Theoretically, pressure head of feed flowrate can be converted to velocity by energy conservation law, as shown in eq. 7.2.3.

DVac

22= {7.2.2a}

gDV

gac

22==ς {7.2.2a}

gHV 2= {7.2.3}

273


III-157

From eq. 7.2.2a and 7.2.3, ζ can be rewritten as follow,

gDgH 2)2(2

=ς {7.2.4a}

DH4

=ς {7.2.4b}

Where ζ = Factor of acceleration V = Tangential velocity in hydrocyclone D = Diameter of rotation, in this case, hydrocyclone diameter H = Head or pressure drop across hydrocyclone

Eq. 7.2.4b signifies that,

• Efficiency of hydrocyclone is direct proportional to its pressure drop. • As stated earlier in section 7.1, efficiency of hydrocyclone will increase if the size

of hydrocyclone decreases. It can be implied that, at the same flowrate, smaller hydrocyclone is always more efficient than the bigger one.

7.2.1.1 General flow pattern and feature of hydrocyclone

General feature

Generally, hydrocyclone consists of cylindrical section, where inlet port(s) is tangentially placed, and conical section, which may be a smooth cone (fig. 7.1-1d and 7.2.1-1a) or has an inflection point (fig. 7.1.1a,c and 7.2.1-b).

There may be one or two inlet ports. And it may have circular or rectangular opening. For typical hydrocyclone, there will be an outlet port at the center of the cover of cylindrical section. This port is always called overflow port. The port locate at the pointed end of conical section is called underflow port. The wall of overflow port is, usually but not always, protruded into the hydrocyclone. If it exists, this part is called vortex finder.

However, many manufacturers had separately engineered the shapes of their hydrocyclones to fit their applications. For examples, there might be another cylindrical section at the end of the conical section. Or there might be 2 concentric underflow without overflow ports. Anyway, their working principal is always based on centrifugal acceleration.

Liquid-liquid hydrocyclone for oil/water seperation

Because of relatively low density difference between oil and water, compared to that of solid and water, shape of liquid-liquid hydrocyclone for oil/water seperation has to be adapted from typical solid-liquid hydrocyclone as shown in fig. 7.1-1 and 7.2.1-1f. The shape of the oil/water hydrocyclone shown in the figure was researched and proposed by Prof. THEW of University of Souththamton, UK. From the figure, another conical section with small cone angle is added, and then followed by another cylindrical section. Overflow oil outlet port is smaller than that of solid-liquid one. And there is no vortex finder since it is not essential to prevent short circuit of oil to oil outlet. Moreover, THEW also shows that presence of vortex finder resulted in lower efficiency. This hydrocyclone is widely accepted

274


III-158

as a typical design of oil/water separation. In GPI lab, MA [16] also used THEW’s type hydrocyclone for his model development.

General flow pattern

As mentioned earlier, general spiral flow pattern of typical hydrocyclone, which has 1 overflow and 1 underflow port, is as shown in fig. 7.2.1-1a. From the figure, there are both upward and downward flow. Upward and downward flow pattern might be difference in case of co-current hydrocyclone, which has 2 concentric underflow ports. Nevertheless, flow pattern will always be spiral, which causes centrifugal acceleration.

From centrifugal acceleration and density difference between oil and water, oil droplets will tend to move toward the axis of hydrocyclone and form an oil rich zone or oil core at the axis (fig. 7.2.1-1d) that can be purged to an oil outlet port. Water, on the other hand, will be centrifuged outward and flow out at water outlet port, as shown in fig. 7.1-1a and 7.2.1-1b. Oil core normally contains both oil and water (fig. 7.2.1-1e). So oil removed from hydrocyclone still contains some amount of water. Then hydrocyclone is rather oil concentrator than oil separator. To separate oil from this concentrated mixture, other separation process, such as coalescer, may be required.

Sometimes, there is an air core forming at the center of oil core (fig. 7.2.1-1c). It does not provide any separation efficiency and, sometimes, makes the operation unstable. Presence of air core can be avoided by providing back pressure, such as valve throttling, at the outlet port.

Oily wastewater

Concentrated oil

Treated water

Underflow port

Overflow portInlet port

a) General spiral flow

pattern in typical hydrocyclone [30]

b) Typical trajectory of oil droplets in liquid-liquid hydrocyclone

c) Example of presence of air core

[30]

Fig. 7.2.1-1 General flow pattern and features of hydrocyclones

275


III-159

d) Oil core formation (thin red strip at the center of the hydrocyclone) [11]

e) Magnified photo of oil core, shows tiny oil droplets suspending in water [16]

Di

D

Dn

Ds

L1

L3

Doθ

β

Dn/D=0.5, Ds/D=0.25, Do/D<0.05, L1/D=1, L3/D=15-20 , β=1.50o, θ=20o

Di/D=0.25 for 1 inlet and 0.175 for 2 inlet ports, total length/D =45 (approx.)

f) THEW’s type liquid-liquid hydrocyclone for oil/water separation

Fig. 7.2.1-1 General flow pattern and features of hydrocyclones

7.2.1.2 Velocity distribution in hydrocyclone

Flow pattern in hydorcyclone, as shown in the previous section, can be described in the form of velocity within the hydrocyclone. Velocity at any point in the hydrocyclone can be divided into 3 components, i.e., tangential, radial and axial velocity (fig. 7.2.1-2). Characteristic of each velocity components will be described as follow.

VzVt

Vr

Z

R

Axial axis

Hydrocyclone w

all

Vr = Radial velocity Vt = Tangential velocity Vz = Axial (or vertical) velocity

Fig. 7.2.1-2 Velocity components in hydrocyclone

276


III-160

1. Tangential velocity

Outer zone: When water is fed tangentially into the hydrocyclone, it will cause vortex flow pattern around axial axis on the hydrocyclone. Generally, in accordance with conservation of angular momentum, tangential velocity (Vt) will increase as radial distance from the axis of the hydrocyclone (R) decreases. Relation between tangential velocity and radial distance in this zone is as shown in eq. 7.2.5.

.constRV nt = {7.2.5}

If “n” is equal to 1, this means conservation of the momentum is complete. The vortex formed under this condition is called “free vortex”. However, for general hydrocyclone, there is some loss within the hydrocyclone so there is no complete conservation of the momentum. Normally, the value of “n” is between 0.5 –1.0, depending on configuration of hydrocyclone. This type of vortex, when “n” in not equal to 1.0, is called “semi-free” vortex.

Inner zone: When small values of R are reached, flow pattern will change in the manner that tangential velocity decreases with decrease in radial distance. The relation between tangential velocity and radius becomes that of solid body rotation, corresponding to constant angular velocity, as shown in eq. 7.2.6. This flow pattern is called “forced” vortex.

.1 constRVt =− {7.2.6}

Then, tangential velocity profile within general hydrocyclone will be as shown in fig. 7.2.1-3c, where tangential velocity profiles of various types of vortex are also shown.

R

VtRn = Const. n=1n<1

R

Vt/R = Const.

R

Ra

Vt/R = Const. VtRn = Const

a) Free vortex (n=1) and semi-free vortex (n<1)

b) Forced vortex c) Combined vortex (general case of hydrocyclone)

Fig. 7.2.1-3 Tangential velocity profile in hydrocyclone and various typed of vortex

Driving force of hydrocyclone come from inlet velocity. Theoretically, tangential velocity at the wall of hydrocyclone at the inlet port level (Vc) should be equal to inlet velocity (Vi). However, since there is some loss within the hydrocyclone, there is some reduction of the tangential velocity from the inlet velocity. The ratio of the tangential velocity to inlet velocity is called “α”.

i

c

VV

=α , α < 1 {7.2.7}

277


III-161

The value of α varies with the shape of hydrocyclone. From eq. 7.2.5 and 7.2.7, tangential velocity at any radial distance can be written as shown below.

ni

nc

nt

DVDVRV )2

()2

( α== {7.2.8a}

nit R

DVV )2

(α= {7.2.8b}

Where VI = Velocity in the inlet port of hydrocyclone D = Diameter of hydrocyclone (measured at the same level as the

inlet port) R = Radial distance

It should be noted that tangential velocity depends on “radial distance (R), and is independent of axial distance (Z). The velocity at the same “R” will be practically identical at any axial distance from inlet level of hydrocyclone (Z), as shown in fig. 7.2.1-4.

Vr1

Vr1

Vr1

Vr1

R= r1

a) Source: Krebs b) Data from KELSALL[30]

Fig. 7.2.1-4 Examples of tangential velocity profile

2. Axial velocity

As shown in fig. 7.1-1 and 7.2.1-1 that the outer and inner layers of liquid in hydrocyclone move in opposite direction, axial velocity (Vz) profile will consist of both upward and downward velocity, as shown in fig. 7.2.1-5. Axial velocity profile depends on shape of hydrocyclone, such as angle of cone, and operating condition, such as presence of air core. Even in the same hydrocyclone, characteristic of axial velocity profile at various zones, such as conical zone, inlet cylindrical zone or near apex zone. etc., are not similar. However, typical axial velocity profiles at the effective conical section of hydrocyclone can be simplified as shown in fig. 7.2.1-5b and 7.2.1-5c. Fig. 7.2.1-5b is typical axial velocity profile for solid-liquid separation, proposed by RIETEMA [quoted by [16]]. For fig 7.2.1-3c, it is the

278


III-162

axial velocity profile, proposed by THEW and COLMAN [quoted by [16]] for of liquid-liquid hydrocyclone, as well as, proposed by KASSELL [quoted by [30]] for solid-liquid separation.

There is no general relation or equation of axial velocity for hydrocyclone. However, equations of axial velocity profile for specific shapes of hydrocyclone are available, such as COLMAN’s equation for THEW’s type liquid-liquid hydrocyclone. The equation (shown in the next section) will be used for 2-phase hydrocyclone model development.

(Source: Krebs) (Source: [52]) (Source: Bradley [30])

a) Examples of axial velocity profiles from various sources

Locus of zero vertical velocity

Locus of zero vertical velocity

R

Z

Locu

s of z

ero

verti

cal v

eloc

ity

Hyd

rocy

clon

e w

all

Axial axis

b) Axial velocity profile with 2 downward and 1

upward flows

c) Axial velocity profile with 1 downward and 1

upward flows

d) Axial velocity profile predicted by COLMAN’s

equation

Fig. 7.2.1-5 Example of axial or vertical velocity profile

279


III-163

3. Radial velocity

Radial velocity (Vr) normally has a maximum value near to the wall of hydrocyclone in inward direction. Then it will decrease with decreasing of radial distance. It might or might not reach zero value and become outward radial velocity, depending on the shape and axial position considering. Example of radial velocity profile is shown in fig. 7.2.1-6. KELSALL [30] proposed the estimated value of radial velocity in the conical section as shown in eq. 7.2.9.

)2tan(θ⋅=WU {7.2.9}

Fig. 7.2.1-6 Example of radial velocity profile [30]

7.2.1.3 Forces on particle in hydrocyclone

From the velocity fields of liquid or continuous phase decsribed above, dispersed particles, in this case, oil droplets, suspended in hydrocyclone will subject to many components of forces. Consider an oil droplet (or particle) in hydrocyclone as shown in fig. 7.2.1-7. The oil droplet will subject to several clearly-defined forces, such as,

F4Fr

F1

Z

R

F2

F3

F5

WV

U

Z

R

Axial axis

Hydrocyclone w

all

Resulting V

Fig. 7.2.1-7 Forces on oil droplets or particles in hydrocyclone

Fig. 7.2.1-8 Components of velocity of oil droplets or particles in hydrocyclone

280


III-164

• Centrifugal force in radial axis (F1)

6

32

1d

RVF πρ ⋅⋅Δ= {7.2.5a}

• Drag force or hydraulic force in radial axis (F2)

421 2

22

dUCdF cπρ ⋅⋅= {7.2.5b}

• Gravity force (F3)

6

3

3dgF πρ ⋅Δ= {7.2.5c}

• Drag force or hydraulic force in axial axis (F4)

421 2

24

dWCdF cπρ ⋅⋅= {7.2.5d}

• Drag force or hydraulic force in tangential direction (F5)

421 2

25

dVCdF cπρ ⋅⋅= {7.2.5e}

Where U = Radial velocity (or decanting velocity) of oil droplets V = Tangential velocity of oil droplets W = Axial velocity of oil droplets

Please note that three components of velocity are of oil droplets (fig 7.2.1-8), which are not exactly the same as the three velocities (Vx, Vy, Vz) of liquid, described in the previous sections. There are also other forces, such as lift force, etc. These forces will make the oil droplets move in hydrocyclone and determine if those droplets can be totally, partially separated or can not be separated by the hydrocyclone. Normally, it is not possible to exactly calculate some components of forces or account for all forces to find the exact resultant force that governs movement of oil droplets.

Thus, to predict if oil droplets can be separated by hydrocyclone or, on the other hand, efficiency of hydrocyclone, there are 2 major approaches, i.e.,

• Theoretical based model, such as trajectory analysis, equilibrium orbit or retention time based model. Well proven laws, such as STOKES law, are applied to these models under some assumption or simplification. This type of model might or might not be very accurate, compared to empirical model derived from experimental data. The accuracy of the type of model depends on degree of simplification used by the researchers. But it is useful tool to understand the effect of parameters to the effciency of hydrocyclone.

• Empirical based model, such as the model and reducing efficiency curve, proposed by Yoshioka and Hotta [30], Dalstorm’s, and Plitt’s , as well as THEW-COLMAN’s model.

7.2.1.4 Trajectory analysis model

GPI’s trajectory analysis model for oil/water separation is the work of MA [16]. Concept of MA’s model is adapted from the concept of decanter. Major assumptions that underlines this model are that;

281


III-165

• The concept of decanter that oil droplets will be separated by hydrocyclone, if the oil droplets can reach the decanting surface before flowing out with water is applied (see fig. 7.2.1-9). Decanting surface in this case is assumed to be the locus of zero vertical velocity (LZVV). If the droplets can reach this line, next moment, they will start to travel upward to oil outlet port, then, are separated from the wastewater.

• There is equilibrium of forces in radial direction and oil droplet travels in the radial direction at terminal velocity of immersed object in fluid stream.

• Radial velocity is governed by STOKES law.

• For tangential and axial velocity of oil droplet, it is assumed that oil droplets are entrained by the water and have the same tangential and axial velocity as the water at that point. So V = Vt and W = Vz.

• THEW’s type hydrocyclone is used. So tangential and axial velocity profiles are calculated from the equations proposed by THEW, COLMAN [quoted by [16]] for this type of hydrocyclone.

Three velocity components, used for model development, are governed by eq. 7.2.6 to 7.2.8.

Radial velocity: derived from STOKES law;

R

VdUc

22

18⋅

Δ=

μρ {7.2.6}

Tangential velocity: derived from energy conservation law (see eq. 7.2.8) and verified by experimental data to find α and n. For THEW hydrocyclone, α = 0.50 and n = 0.65. Please note that eq. 7.2.7b is valid only for 2-inlet ports hydrocyclone only.

65.0)2

(50.0R

DVV i= {7.2.7a}

Or 65.0

2

4

)2(50.0 ⎟

⎠

⎞⎜⎝

⎛

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

=R

D

D

QV n

iπ

{7.2.7b}

Axial velocity: COLMAN [quoted by 16]] observed axial velocity profile in lower conical section of THEW hydrocyclone, which is the effective section, and developed axial velocity equation in form of the 3rd order polynomial equation, as shown in eq. 7.2.8. Z-axis starts from the bigger end of lower conical section, where R is equal to Dn/2, toward the lower cylindrical section.

32

19.163.81233.3 ⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zzzz RR

RR

RR

WW {7.2.8a}

2))2/tan(5.0( βπ ⋅−=

ZDQW

nz {7.2.8b}

)2

tan(2

β⋅−= Z

DR n

z {7.2.8b}

282


III-166

Q represents wastewater inlet flowrate. Dimensions of the hydrocyclone (Di, Dn, etc.) conform to fig. 7.2.1-1f. To find trajectory, relation between R and Z has to be found. Firstly, consider equation of U and W as follow;

tRUδδ

−= {7.2.9a}

tZWδδ

= {7.2.9b}

From above equations, we have;

WU

ZR=−

δδ {7.2.10a}

WZ

UR δδ=− {7.2.10b}

Integration of eq. 7.2.10b will result in the relation between R and Z, as shown in eq. 7.2.10c.

∫∫ =−LR

R WZ

URZVV

0

δδ {7.2.10c}

RZVV represents any points on the locus of zero vertical velocity (LZVV), which is normally in conical shape. However, since conical angle of the lower cone is very small (1.5o), it can be safely assume that LZVV is cylindrical with the radius of RZVV at the small end on the cone (Z=L). At Z = L, from eq. 7.2.8, RZVV is equal to 0.186(Dn/2). So relation between R and Z or trajectory of oil droplets can be rewritten as shown in eq. 7.2.11. Integration of the equation is transcendental and requires complex numerical method, such as Runge-Kutta. However, it can be done by advanced calculator or computer program, including the program being developed in this research.

∫∫ =−L

R WZ

URDn

0

)2/(186.0 δδ {7.2.11}

d = dc

d > dcd < dc

Z

R

L

Rd

Hydrocyclone wall

LZVV

0.5Dn

0.186Dn

d

ηd

d c

d < d c

Zone 1 Zone 2d = or > d c

100%

Fig. 7.2.1-9 Trajectories of oil droplets and typical efficiency curve

from trajectory analysis model

283


III-167

Efficiency determination

Fig. 7.2.1-9 shows the trajectories of oil droplets in lower conical section of hydrocyclone. As stated before, decanting surface in this case is the locus of zero vertical velocity (LZVV), which can be calculated from eq. 7.2.8, where W = 0. To simplify the calculation, it is assumed that LZVV is cylindrical with R = 0.186(Dn/2).

From the figure, the longest path to reach the LZVV is the path starting at the wall of the hydrocyclone at larger end of the cone to the zero-vertical-velocity point at the lower end of the cone (fig. 7.2.1-9a). The smallest droplet size that can reach this point is called the cut size (dc). The droplets of cut size or bigger are always separated from wastewater stream with 100% removal efficiency (eq. 7.2.12a).

For d ≥ dc %100=dη {7.2.12a}

The smaller droplets can be also separated providing that they enter the cone near to the axis of the hydrocyclone. When uniformly distributed influent flow is valid, which is practically true, the removal efficiency of the droplets smaller than cut size are proportional to corresponding radial distance (Rd) that makes the droplets reach the ZVV point at the lower end of the cone, as shown in eq. 7.2.9b. Typical efficieny curve of the trajectory analysis model is shown in fig. 7.2.1-12b.

For d < dc %100)

2186.0()

2(

)2

186.0(

22

22

⋅−

−=

nn

nd

d DD

DR

η {7.2.12b}

7.2.1.5 Other models

There are several researches that suggests the model to predict the efficiency of hydrocyclone, both theoretical and empirical based, such as Bradley’s, Rietema’s, Dahlstrom’s, Chebelin’s, Plitt’s, Lynch’s, etc. [16],[28]- [34]. However, most of models are developed from solid-liquid hydrocyclone. Some models are developed for specific commercial oil/water hydrocyclone, such as Vertoil’s. So it should be applied only with that specific hydrocyclone. Extrapolation of model is normally not guaranteed.

For THEW’s type hydrocyclone, used by MA in his research, Prof. THEW, himself, and his colleague, COLMAN, have proposed the model for the hydrocyclone (eq. 7.2.13a). However, it is empirical model, which seems to be obtained from curve fitting. The equation for estimation of d75% for THEW’s hydrocyclone, quoted by CHEBELIN [29], is as shown in eq. 7.2.13b (use unit in kg, m, second.)

))19.0

%75(8.1(

1−−

−= dd

d eη {7.2.13a}

5.03

%75)(01.0⎥⎦

⎤⎢⎣

⎡⋅Δ⋅

=QD

d n

ρμ {7.2.13b}

284


III-168


MA’s model was verified by observed data obtained from crude oil/water treatment. Fig. 7.2.1-10 shows the efficiency curve from MA’s trajectory analysis model (eq. 7.2.6 to 7.2.8, 7.2.11), THEW-COLMAN’s model (eq. 7.2.10 and 7.2.11) and observed efficiency [16]. From the figure, it shows that, at droplet size > 20 microns, MA’s and THEW-COLMAN’s models give relatively accurate result (± 10% error). However, at d > d 80%, THEW-COLMAN’s model seems to cause higher degree of error and predict too high value of cut size. This may because the researchers used different assumptions or operating condition to develop their models.

Difference between MA’s model and observed value can be explained by effect of the assumptions used to develop the model as follow;

• Radial velocities of oil droplets are not exactly governed by STOKES law. STOKES law is valid only in laminar flow regime (Re of droplets < 1). However, in some cases, esp. when droplet size is large, Re might be greater than 1. And from very short retention time, the droplets may not reach the terminal velocity, governed by STOKES law. Furthermore, radial velocity of water, which can carry the oil droplets along and add-up to the decanting velocity from, is not accounted. So the value of U used in the model is, generally, lower than actual value. This is the reason why the predicted efficiency is lower than the observed value.

• Effect of eddy current, lift forces, drag forces in vertical and tangential direction is assumed to be negligible and cancelled. Anyway, they, in fact, can cause some discrepancies in the prediction result.

• Effect of hinder settling, esp. near to the center of hydrocyclone

However, Ma’s model is developed from theoretical assumption and does include many parameters, such as viscosity, feed flowrate, and size of the hydrocyclone, etc., which can be used to explain or verify the effect of these parameters to the efficiency. So, it provides valuable tool to designer and will be used in the computer program being developed under the scope of work of this research. Furthermore, the trajectory analysis model is developed under the assumptions that give underestimated result. So the predicted efficiency is on the safe side.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

110%

0.00 10.00 20.00 30.00 40.00 50.00 60.00


Effic

ienc

y (%

)

Thew-Colman's modelObserved dataMa's trajectory analysis

Note: Dn = 0.02 m, Q = 1.943 m3/h, Vi = 7.12 m/s: Wastewater used is crude oil/water emulsion: Δρ = 98 kg/m3, μ = 0.0011 Ns/m2, β = 15o, d = 15 to 50 microns

Fig. 7.2.1-10 Comparison between observed efficiency and predicted efficiency form MA’s model and THEW-COLMAN’s model

285


III-169

There is no tested evidence about the valid range of MA’s model. However, from the assumptions used, the model should be valid for every THEW’s type hydrocyclone as long as operating condition is in valid range of STOKES law (laminar flow region: Re < 1). Anyway, working condition and performance of hydrocyclone are also governed or limited by other parameters or factors. So it is more suitable to state that the model is valid in laminar flow region when other factors also confirm to working limit of hydrocyclone, as described in section 7.2.1.7.


Parameters that affect the performance of hydrocyclone are those presenting in model, as shown in section 7.2.1.4. However there are some factors that have influent on the efficiency of hydrocyclone that can not be expressed in form of numerical function. Influent parameters of hydrocyclone, collected from GPI researches and many literatures, can be summarized as follow;

1. Droplet size (d)

Efficiency of hydrocyclone increases with increasing of droplet size, as clearly shown in fig. 7.2.1-10. So it is recommended to avoid any action that can cause oil drops to break-up into smeller droplets, such as using high-shear feed pump (Ex. high speed centrifugal pump, or regenerative turbine pump). On the contrary, putting some process that can increase the size of droplets at upstream of hydrocyclone, such as high-rate coalescer, will help increasing hydrocyclone performance.

2. Feed flowrate (Q)

From eq. 7.2.4, efficiency of hydrocyclone will increase, if feed flowrate increases for it will make inlet velocity, which is a driving force of hydrocyclone, increase. However, this is true before velocity in the cyclone reaches the value that droplet re-fragmentation occurs. There are evidences [16], [28] that the inlet velocity (Vi) as high as 7 m/s is used without causing any adverse effect on efficiency. This value of Vi can be compared to velocity of 1.7 m/s, using nominal diameter (Dn) of the cyclone. Please note that increasing of flowrate also cause pressure drop to be increased, which effects operating cost. Pressure drop will be described separately in section 7.2.1.8.

3. Shape of hydrocyclone

Shape of hydrocyclone substantially effects its velocity profiles and vortex formation. So it certainly effect its performance. But it can not be expressed in form of numerical equation. Many researchers have studied effect of dimension of each component of solid-liquid hydrocyclone (such as inlet port, overflow, underflow port, vortex finder, etc.). For oil/water hydrocyclone, THEW has thoroughly studied the effect of geometry and proposed the optimized shape of the cyclone, which is used in GPI researches. In fact, we usually design hydrocyclone system from standard commercial product, which, more or less, conform to THEW’s geometry. Anyway, MA [16] had carried out some profound studies of 2 components. The result is as shown below.

• Conical angle θ: Using lower θ (from 20o to 8o) result in decreasing of pressure drop about 11% (from 4.5 to 4.0 bar at 1.6 m3/h, Dn = 16 mm, crude oil/water emulsion) while the efficiency is relatively unchanged.

286


III-170

• Ratio Dn/D: Lower Dn/D from 0.5 to 0.43 can increase efficiency (+5% for Q = 1 to 1.6 m3/h) but pressure drop is also considerably increased (+0.5 bar for Q = 1 to 1.6 m3/h, D = 32 mm, Dn = 16 and 14 mm).

It must be noted that the models in section 7.2.1.4 and 7.2.1.5 are valid only for THEW’s type hydrocyclone only.

4. Inlet oil concentration (Co)

For low concentration oily wastewater (Co < 1%), the concentration has practically no effect on efficiency. For higher concentration, theoretically, it should effect the performance from hinder settling effect and changing of viscosity. However, THEW [28] shows that the efficiency is little affected even at the concentration as high as 10 – 15%.

5. Purge ratio or split ratio(Rf)

For oily/water separation, oil will be purged to outlet overflow port. Purge ratio (Rf) (or split ratio) is the ratio between oil outlet flowrate (overflow) to water feed flowrate (Q). Theoretically, flowrate of overflow should be as close as possible to the flowrate of oil components in the water to get relatively pure separated oil. However, in practice, it is impossible to do so because pressure drop will be too high and some oil will entrain to underflow port. It is recommended to use Rf around 1.8 to 2.0 times of flowrate of inlet oil [28]. For general oily wastewater, which inlet oil concentration less than 5% (by volume), purged oil flow is relatively small compared to the whole flow. So it does not effect velocity profile and efficiency of hydrocyclone. But it effects pressure drop, which will be described in the next section.

6. Wastewater characteristic and presence of surfactant

Wastewater characteristic inevitably affects efficiency of hydrocyclone. The model in section 7.2.1.4 has included general parameters that represent characteristic of wastewater, such as Δρ, μc, etc. So it is useful tool to understand effect of parameters on cyclone performance. Some indirect parameters can be determined if it effects the performance or not by considering its effect on those parameters included in the equation.

For example, presence of salinity will increase density difference, so it favors efficiency. Presence of surfactant will lower interfacial tension, resulting in fragmentation of droplets, so it causes adverse effect on hydrocyclone.

7. Temperature (T)

Proporties of oil and water always change with parameter. So it is inevitable that temperature will effect hydrocyclone efficiency. Rising in temperature make water viscosity decrease, then favors the efficiency. However, it will decrease interfacial tension and density difference, then has adverse effect on the efficiency. So effect of temperature on individual waster will be accounted on these facts. Anyway, for petroleum/water emulsion at normal working condition (<40o C), increasing of temperature, more or less, favors hydrocyclone efficiency.

287


III-171

8. Presence of air core

Air core could form at the axis of hydrocyclone from presence of low-pressure region generated by vortex, especially when it is purged directly to atmosphere. It is also formed by changing in solubility of gas in water from decreasing of pressure. Presence of air core might partially block the overflow port and result in instability of performance. Air core formation can be suppressed by avoiding direct purge to atmosphere and the use of backpressure. However, many system operating with several percent of air core (by volume) still give satisfactory result. So throttling should be made until the efficiency is optimized without over-emphasizing on air core formation. In fact, its formation could not be visible in commercial opaque hydrocyclone.


Pressure drop is very important parameter since it directly relates to driving force of the hydrocyclone, as shown in eq. 7.2.4. Furthermore, purge ratio and back pressure control to suppress the air core is normally be done by valve throttling at the two outlet ports, which directly effects the pressure drop.

Pressure drop can be divided into 2 types, i.e., pressure drop across inlet and overflow port (ΔPo) and pressure drop across inlet and underflow port (ΔPu).

Like the case of efficiency, equations of the pressure drop of hydrocyclone were proposed, based on both theoretical and empirical approach. However, there is no pressure drop model for THEW’s type hydrocyclone, which we knew of. Anyway, many literatures [16], [28] – [34] show that general model of pressure drop relation is as shown in eq. 7.2.14.

)/( 4.2n

xx DQfP =Δ {7.2.14}

From GPI researches, there are sufficient data to formulate semi-empirical models of pressure drop from the general model, as shown below.

For pressure drop (bar) across inlet and overflow port (oil outlet);

1611.0

4

3.2

)1(6.216 ⎟

⎟⎠

⎞⎜⎜⎝

⎛

−⋅=Δ

fno RD

QP {7.2.15a}

Flowrate and hydrocyclone diameter are in m3/s and m, respectively. Rf is lower than 1.0. However, effect of purge ratio is verified from relatively small set of data, so it might cause some error. Firthermore, for low concentration wastewater (Co<5%, Rf <10%), effect or Rf is very small. Eq. 7.2.15a can be rewritten as eq. 7.2.15b, which can be safely used without causing serious error.

17.116 4

3.2

×=Δn

o DQP {7.2.15b}

288


III-172

For pressure drop (bar) across inlet and underflow port (water oultet);

4

2.26.4

nu

D

QP =Δ {7.2.15c}

From above equations, it should be noted that:

• Pressure drop of overflow is always higher than that of underflow. So it is used for pump sizing. If multi-stage hydrocyclone system is used, available pressure for the next stage is equal to (ΔPo-ΔPu). The unit of Q is in m3/s, D in meter.

• Underflow pressure drop is practically independent of split ratio. While overflow pressure drop is slightly affected by the split ratio.

• For oily wastewater treatment, which involves low concentration of oil, characteristic of wastewater does not affect the pressured drops.

General working pressure range of hydrocyclone is up to 6.5 bar. THEW [28] recommended the value of 1-5 bar. Hydrocyclone selection is to be compromised between efficiency, size (capital cost) and pressure drop (operating cost).


Design procedure for hydrocyclone is based upon the equations, shown in the previous sections. To design two-phase hydrocyclone for oily wastewater treatment, the required cut size will be determined first. After that, the size of the hydrocyclone can be preliminary selected. Then, graded efficiency (efficiency of each size of droplet) and then total removal efficiency can be determined. The procedure will be repeated until the optimum size is selected. Calculation in each step is described below.

1 Cut size determination

The cut size can be determined from the degree of treatment required as well as from the limitation of the decanting processes. Cut size determination from degree of treatment is described in chapter 3. For theoretical limitation, since the model is based on STOKES law, it can be applied only when the droplet behavior conforms to the law. Thus, it can not be used with very small droplet sizes for they are subjected to Brownian’s motion and their rising velocity are not governs by STOKES law. There are also some limitations from characteristic of hydrocyclon itself that governs the cut size selection. Generally, recommended smallest diameter that can be separated (but only partially) by hydrocyclone is about 20 microns. For smaller droplets, the efficiency will be very low and erratic.

To design the hydrocyclone by MA’s model, it is recommended to select the cut size that covers majority of oil droplet in the wastewater and provide a safety factor around 10% to 20%, because the model is apt to predict too small cut size. For example, if the desired cut size is 50 microns, it is recommended to select 50(1-0.20) = 40 microns.

2 Hydrocyclone sizing

The size of the hydrocyclone could be roughly estimated by general design criteria, such as maximum working pressure, hydraulic loading, etc. Fig. 7.2.1-1 shows the

289


III-173

relations between flowrate, inlet velocity, velocity at lower conical section (at dia. = Dn) VS. diameter of the cylindrical part of hydrocyclone (D) at Po = 3 and 5 bars, which are generally used as design pressure. For the exact size, the mathematical model can be used. Then the model will be used again to find the graded efficiency of the hydrocyclone. After that, the size may be fine adjusted again to obtain the required efficiency.

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

0 10 20 30 40 50 60 70 80 90 10

Diameter of cylindrical section of hydrocyclone (D) (mm)

Flow

rate

(m3 /h

) and

Vel

ocity

(m/s

)

Q at P = 5 bar

Q at P = 3 bar

Inlet V at P = 5 bar

Inlet V at P = 3 bar

V at Dnat P = 5 bar

at P = 3 bar

0

Fig. 7.2.1-1 Corresponding flowrate and velocities of various sizes of hydrocyclones at overflow pressure drop (Po) = 3 and 5 bars

To size the hydrocyclone and calculate greaded efficiency of the hydrocyclone, the following procedure, based on MA’s model, is recommended.

2.1 Calculate the size of hydrocyclone (Dn)

The size of the hydrocyclone can be calculated by integrating eq. 7.2.11, using the design cut size from item 1. The corresponding R will be equal to (Dn/2). The integration can be done by the use of a scientific calculator, general mathematical software or the computer program developed in the scope of work of this research.

∫∫ =−L

R WZ

URDn

0

)2/(186.0 δδ {7.2.11}

RVdU

c

22

18μρΔ

= {7.2.6}

65.0

2

4

)2(50.0 ⎟

⎠

⎞⎜⎝

⎛

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

=R

D

D

QV n

iπ

{7.2.7b}

32

19.163.81233.3 ⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zzzz RR

RR

RR

WW {7.2.8a}

290


III-174

2))2/tan(5.0( βπ ⋅−=

ZDQW

nz {7.2.8b}

)2

tan(2

β⋅−= Z

DR n

z {7.2.8b}

)2/5.1tan()(25.0

onD

L = {7.2.16}

2.2 Calculate corresponding radial entering distance (Rd)

For the droplet smaller than the cut size, it is necessary to calculate the corresponding Rd that makes the droplet reach the radial distance R = 0.186(Dn/2) at Z = L. To calculate Rd, eq. 7.2.11 need to be integrated again, using every value of “d” that is smaller than the cut size.

3 Outlet concentration and removal efficiency

Graded efficiency

Graded efficiency (or centrifugal efficiency), in this case, represents the efficiency calculated from the trajectory analysis method. For d ≥ dc, graded efficiency will be equal to 100%.

For d ≥ dc %100=dη {7.2.12a}

For the droplet smaller than the cut size, the theoretical graded efficiency can be calculated, using the Rd from item 2.2, by the following equation.

For d < dc %100)

2186.0()

2(

)2

186.0(

22

22

⋅−

−=

nn

nd

d DD

DR

η {7.2.12b}

Outlet concentration

Outlet oil concentration at the underflow (water) and overflow (oil) outlet can be calculated from theoretical efficiency and purged ration as shown in eq. 7.2.17a and b, respectively. The equation is valid only when the purged flow is greater than the oil quantity in the wastewater. Otherwise the oil that can reach the oil core will not be purged entirely and then entrain with the underflow.

)1()1(

f

oddd R

CC

−−

=η

{7.2.17a}

f

oddoverflowd R

CC

η=,

{7.2.17b}

Global efficiency and removal efficiency

Graded efficiency is not the actual efficiency of hydrocyclone because it does not account for purged or split flow. Even though this split flow is small and hardly affects the flow regime in the hydrocyclone, it, somehow, affects the oil concentration calculation. As

291


III-175

described in section 7.2.1.7, the purged flow must be greater than the oil quantity in the wastewater. Otherwise the oil that can reach the oil core will not be purged entirely.

Total removal efficiency (ηt) of the hydrocyclone represents ratio of oil concentration between underflow and inlet flow, which can be calculated from the following equation.

%1001)1(

1 max

min

⋅⋅⋅−

= ∑d

dodd

oft C

CRηη {7.2.18}

Co = Inlet oil concentration of the wastewater Cod = Inlet concentration of the droplet diameter “d” Rf = Split ratio (= Qoverflow/Q)

4 Pressure drop and energy required

The pressure drop (in bar) can be calculated from the following equations. Generally, the overflow pressure drop is greater than that of underflow and will be used as design pressure for feed pump selection. The unit of Q is in m3/s, D in meter.

1611.0

4

3.2

)1(6.216 ⎟

⎟⎠

⎞⎜⎜⎝

⎛

−⋅=Δ

fno RD

QP {7.2.15a}

4

2.26.4

nu

D

QP =Δ {7.2.15c}



The equations described above are developed from the following assumptions and limitations. Thus, it is necessary to ensure that these assumptions are valid when design your hydrocyclone.

1) The model is valid only for THEW’s type hydrocyclone or other cyclones with relative identical geometry.

2) It is recommended to use the model only for droplet diameter of 20 microns or greater. For smaller droplet, it can also be applied, but for comparison only.

3) Eq. 7.2.7a is valid for the hydrocyclone with 2 inlet ports only. If the hydrocyclone has only 1 inlet port, Q in the equation will be modified as shown in eq. 7.2.7a’. However, using 2 inlet ports is recommended for its hydraulics stability. Please note that the size of 2 inlet ports will be smaller than a single inlet port to keep the inlet area constant.

0.65)RnD

)(2iπD

Q2(V = {7.2.7a’}

4) For general oily wastewater with the oil concentration of 5% or less, overflow quantity is small, not greater than 10%. So its effect on velocity profiles and efficiency is small, thus, negligible.

292


III-176

5) From the assumptions used for model development, the model prediction has some discrepancies from the actual value. However, as stated in section 7.2.1.6, it is a useful tool for understanding the effect of each parameter.

6) The hydrocyclone should be designed within the limitations shown in section 7.2.1.7. The size of hydrocyclone should conform to commercial standard sizes.

7) From eq. 7.2.4b, it is recommended to select the smallest possible hydrocyclone. However, the pressure drop, clogging problem from the presence of suspended solids and cost should be taken into account.

8) From the high shear characteristic, it is reasonable to assume that there is no coalescense taking place within the hydrocyclone. Re-fragmentation can be avoided by following the recommended range for pressure and flowrate. So it can imply that the overall size distribution of oil droplets is not affected by the hydrocyclone. This assumption is useful for considering the size distribution of each stage of multi-stage hydrocyclone system.

2. Construction and system integration

Normally, we select the hydrocyclone from commercial products. So we cannot do anything about its design and fabrication. However, it is necessary to study the product to make sure if it conforms to our model limitation.

Hydrocyclone efficiency depends on the sizes of the oil droplets, so the processes that can make the droplets smaller, such as pumping, should be avoid or kept to minimum, at the upstream of the hydrocyclone. Or the hydrocyclone should be installed as close to the wastewater source as possible.

Generally, oily wastewater contains some suspended solids. So the hydrocyclone system is designed as a coupling of soild-liquid and liquid-liquid hydrocyclone, as shown in fig.7.2.3-1. In this case, the feed pressure of the first stage hydrocyclone should be sufficient for the next stage hydrocyclone(s). Installation of booster pump between each stage is possible but not recommended since it will make the oil droplets smaller.

a) schematic diagram b) An example of commercial set

Fig. 7.2.3-1 An example of the coupling between solid-liquid and liquid hydrocyclone for the treatment of oily wastewater containing suspended solids (Source: Ultraspin)

293


III-177

7.2.4 Variations, advantage and disadvantage of hydrocyclone

Variation of two-phase hydrocyclone

There are some variations of liquid-liquid hydrocyclone, for example in flow pattern (counter current, co-current), etc. (see fig. 7.1-1). In any case, the majority of the liquid-liquid hydrocyclone still conforms or relativelty close to THEW’s type. Nowadays, there are several suppliers that commercialize these products, such as Neyrtec, Dorr Oliver, Krebs, Ultraspin, etc. So designers can find further information from these suppliers.

However, there is another type of liquid-liquid hydrocyclone preliminary tested in GPI lab [11] that should be mentioned here, i.e. “the co-current hydrocyclone” (fig. 7.1-1d). Main advantage of this hydrocyclone comes from its shape which is considerable shorter (around 8.5 times of D) than THEW’s type (about 45 times of D). So it requires less installation space. There is no study on comparison between THEW’s and the co-current cyclone efficiencies in GPI lab. However, WANICHKUL [11] proposed useful relation between oil and water outlet velocity of the co-current hydrocyclone. He used an oil extract pump, instead of typical outlet valves, to control oil outlet flowrate. From fig. 7.2.4-1, the effects of relation between water outlet velocity (V1) and oil outlet velocity on oil removal efficiency are summarized below.

Oil outletWater outlet

�

V1

V2

Oil outlet tube, dt

Case I: V1≥V2 and dt > Oil core

Case II: V1<V2 and dt > Oil core

Case III: V1≤V2 and dt < Oil core

Case IV: V1>>V2 and dt < Oil core

Fig. 7.2.4-1 Relation between oil and water outlet velocity in co-current hydrocyclone

• V1 ≥ V2 and the concentric oil outlet tube is greater than diameter of oil core The oil will be largely removed. The efficiency is high.

• V1 < V2 and the oil outlet tube is greater than diameter of oil core The oil will also be largely removed. But some oil will be entrained with the water. The efficiency is slightly lowered.

• V1 ≤ V2 and the oil outlet tube is smaller than diameter of oil core The oil will flow out at both oil outlet and water outlet ports. The efficiency is low.

• V1 >>V2 and the oil outlet tube is smaller than diameter of oil core The efficiency is, somehow, relatively high from the effect of high oil outlet velocity.

These facts are useful for the co-current cyclone operation. However, the efficiency of the co-current hydrocyclone cannot be calculated by MA’s trajectory analysis model. To design this cyclone, designers should consult the manufacturers or find more researches on this type of hydrocyclone.

294


III-178

Advantages: Main advantages of the hydrocyclone include.

• Its compactness and it has no moving part. • Its efficiency increases with flowrate (eq. 7.2.4b). This is the most distinguished

advantage of the hydrocyclone. • No additional chemicals (such as coagulant, floccuclants) required. • The operation is not complicate. • Its modular construction facilitates upgrading or optimization.

Disadvantages: Main disadvantages of the hydrocyclone include,

• It is a pollutant concentrator, rather than a separator. However, GPI lab has developed the oily wastewater treatment system which is the combination of hydrocyclone and coalescer to make it possible to obtain relatively pure separated oil. The details will be discribed in chapter 12.

• The hydrocyclone cannot remove the oil and the suspended solids simultaneously. So it needs at least 2 hydrocyclones of different shape. To overcome this disadvantage, GPI lab has initiated a special hydrocyclone that can separate both the oil and suspended solids at the same time. The details will be described in the following section.

7.3 Three-phase hydrocyclone


As mentioned in the previous section, a limitation of general hydrocyclone is that it cannot be used to separate solid and liquid pollutant from the wastewater simultaneously. Coupling of standard solid-liquid and liquid-liquid hydrocyclones are normally used for this purpose (fig. 7.2.3-1). To obtain these two separations simultaneously, AURELLE and MA [16] had initiated a new type of hydrocyclone, known as “three-phase hydrocyclone”, as shown in fig. 7.3.1-1. The idea behind its geometry is the fusion between liquid-liquid cyclone of THEW and solid-liquid cyclone of RIETEMA. The vortex created by inlet flow will be used as a driving force for the 2 parts of the hydrocyclone. Three-phase hydrocyclone does not require an inter-connecting pipe between 2 cyclones like typical 2-cyclone system (fig. 7.2.3-1). Theoretically, it means the energy loss of the system is reduced. Thus the efficiency should be better. Installation space is also reduced.


DoDDs

DiDu

Dp

L5 L3L1L2

L4

Note: Di/D=0.25 for 1- inlet and 0.175 for 2- inlet, Do/D=0.43,Ds/D=0.28, Du/D=0.19, Dp/D=0.034,

L1/D=0.4,L2/D=5, L3/D=15, L4/D=0.3, Solid-liquid part cone angle=12o, for liquid-liquid part=1.5o

Fig. 7.3.1-1 Three-phase hydrocyclone

295


III-179

Fig. 7.3.1-2a shows the trajectories of oil droplets and suspended solids. The solids will be centrifuged outward to the wall of cyclone and be purged at the annular solid outlet port at the apex of RIETEMA’s conical section. The oil droplets will move toward the axis in the RIETEMA part and flow up into THEW part, where they will join the oil core. The oil core will spirally flow at the axis down to the RIETEMA part again, and then be purged to oil concentric outlet tube.

Generally, working principle of this hydrocyclone is still similar to that of two-phase hydrocyclone. However, The result from MA’s Plexiglas model showed that the vertical velocity in RIETMA part is slightly changed to account for the central oil core, as shown in fig. 7.3.1-2b.

Wastewater inlet

Wastewater inlet

Water inlet

Solid annular outlet port

Concentric oil outlet tube

Fig. 7.3.1-2a Oil drop (sphere) and particle (cube) trajectories in three-phase hydrocyclone

Solid-liquid(RIETEMA) part

Liquid-liquid(THEW) part

Fig. 7.3.1-2b Typical vertical velocity profiles in three-phase hydrocyclone

7.3.1.1 Model for liquid-liquid part

MA reported that the flow pattern and formation of the oil core inside the THEW (oil) part of three-phase hydrocyclone is identical to THEW’s hydrocyclone. But no mathematical model had been proposed. However, from the identical flow pattern compared to THEW’s cyclone, it can be safely assumed that the driving force of this hydrocyclone is identical to THEW’s type hydrocyclone and MA’s trajectory analysis for THEW’s type hydrocyclone should be applicable for this cyclone as well.

From the assumption that two and three-phase hydrocyclone have identical driving force, which is the fraction of their own inlet velocity, we have the following equation;

296


III-180

( ) i(Thew)VThewα3iV3α ⋅=⋅ φφ

( ) ⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛⋅=

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛⋅ 2)i(Thew)π(D

4Q0.52)3iπ(D

4Q3α

φφ

From the geographies of the cyclones (fig. 7.2.1-1f and 7.3.1-1),

Dn in fig.7.2.1.1f = Do in fig. 7.3.1-1 = ND.

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⋅=⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⋅2))

0.5NDπ(0.25(

4Q0.52))

0.43NDπ(0.25(

4Q3α φ

676.02

0.430.50.53α =⎟

⎠⎞

⎜⎝⎛⋅=⋅φ {7.3.1}

The subscript “3φ” and “THEW” represent that the variables belong to three-phase hydrocyclone and THEW’s type hydrocyclone, respectively.

From above equations, MA’s model as shown in section 7.2.2 can be adapted for three-phase hydrocyclone design by changing the value of φ in eq. 7.2.7b from 0.50 to 0.676. Dn in the equations of 2-phase hydrocyclone is replaced by Do in case of 3-phase hydrocyclone. L in those equations is also replaced by L5. The model was verified by the observed data from MA’s research. The result showed that the model could predict the theoretical oil removal efficiency of the three-phase hydrocyclone with ± 20% accuracy.

7.3.1.2 Model for solid-liquid part

MA did not propose model for solid-liquid part either. Anyway, since the geometry of this part of three-phase hydrocyclone is similar to RIETEMA’s cyclone, the assumption that the driving force of this cyclone is identical to standard RIETEMA hydrocyclone should be valid. So the model derived for RIETEMA hydrocyclone, which is general form of solid-liquid hydrocyclone, should be applicable to three-phase hydrocyclone.

RIETEMA [30] defined the performance of the hydrocyclone in the form of d50% and the dimensionless number Cy50, as shown in eq. 5.9. To find graded efficiency besides d50%, correlation of YOSHIOKA and HOTTA [30], for 2% < ηd <98%, may be applied (eq. 5.10).

50CyQcρcη

Δp)4L2)(Lcρss(ρ250d

=⋅−−

{7.3.2}

30.115)50%dd(

e1SSd,η−−

−= {7.3.3}

RIETEMA recommended the value of Cy50 around 3.5 for RIETEMA type hydrocyclone, which MA used as one part of his hydrocyclone. So it is recommended to use Cy50 =3.5 to calculate the efficiency of solid part of three-phase hydrocyclone.

297


III-181


MA proposed the model for pressure drop calculation for his prototype hydrocyclone. For the prototype, he used Do = 14 mm and D = 32 mm. Pressure drops across various ports of the prototype (in bar) can be calculated from the following equations,

2.111.364QwaterΔP = {7.3.4a}

2.340.951QssΔP = {7.3.4b} 03.2140.3 QPoil =Δ {7.3.4c}

The equations are valid only for 14/32-mm. three-phase hydrocyclone. Thus, to extend the valid range of the equations or to develop generalized model. We consider 2 approaches.

• Similarity approach

From TRAWINSKY [33], he suggested that hydrocyclone, like other centrifugal machines, is subject to concept of similarity or affinity law.

Δρ250%d1DΔP −−∝ {7.3.5}

So combination of eq. 7.3.4 and 7.3.5 can be used to predict the pressure drops of any size of three-phase hydrocyclone by calculating the pressure drops of 14/32-mm. hydrocyclone at given flowrate by eq. 7.3.4 first, then use eq. 7.3.5 to find the flowrate at given hydrocyclone diameter and given characteristic of wastewater.

• Empirical approach

Similarity approach is theoretical based. In practice, many factors may cause some discrepancies from theoretical value. To account for these factors, empirical approach is introduced. We base our model on eq. 7.2.14 that we successfully applied for two-phase hydrocyclone.

To develop and verify empirical model, we use the data from MA’s and THEW’s research [16], [18]. The empirical models for predicting of the pressure drop across inlet and various outlet ports are as shown in eq. 7.3.6 (ΔP in Bar, Q in m3/s, D in m.).

4D

2.12Q49.8waterΔP = {7.3.6a}

4D

2.34Q21ssΔP = {7.3.6b}

4D

2.03Q55oilΔP = {7.3.6c}

The equations are in the similar forms to MA’s models. However, substituting the value of D = 32 mm in eq. 7.3.6 will not result in the exact eq. 7.3.4 because eq. 7.3.6 are verified from wider range of data.

298


III-182

Comparisons between the predicted pressure drops of the 2 approaches, using data from [16], [28], are as shown in fig. 7.3.1-3. The graphs show that the two approaches give very accurate results. However, using eq. 7.3.4 and 7.3.5 may cause some difficulty because it requires calculation for d50% first. So using eq. 7.3.6 may be more convenient.

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00


Pred

icte

d pr

essu

re d

rop

(bar

)

= Pressure drop across oil outlet port

= Pressure drop across water outlet port

= Pressure drop across SS outlet port

Sem

i-em

piric

al

Sim

ilarit

y

Fig. 7.3.1-3 Comparison between observed pressure drop and predicted valued from the 2 approaches (similarity and semi-empirical model)


Because the mathematical models of three-phase hydrocyclone is adapted from two-phase hydrocyclone, design procedure as well as design consideration for the three-phase hydrocyclone will be identical to that of the two-phase, as shown in section 7.2.1 and 7.2.2. But the efficiency and pressure drop across solid outlet port of the solid-liquid part will be calculated additionally by eq. 7.3.2, 7.3.3 and 7.3.6b.

7.3.3 Advantage and disadvantage of three-phase hydrocyclone

Advantage: Main advantage of this hydrocyclone is that it can remove solid and oil simultaneously. It can replace typical two-stage system. The installation space is reduced. It also helps saving the piping work

Disadvantage: Various types of three-phase hydrocyclone are developed and patented, such as AURELLE and MA’s and few American designs. However, it is available as custom-made equipment, rather than mass product. So it may not be cheap and may cause some inconvenient in maintenance or part replacement. The oil and solids removal efficiencies are also inter-related for they share the same driving force. So, in some case, to obtain one good removal efficiency (either oil or solid), the other removal performance may not work at the optimized condition. So the flexibility of the system is, somehow, less than that of the coupling between 2 cyclones.

299


III-183


8.1 General

Membrane process is a separation process based, mainly but not entirely, on filtration concept. Physically, membrane is permeable or semi-permeable material, which restricts the motion of certain species [54]. Theoretically, we can always separate one or more components from fluid stream providing that the membrane chosen is suitable for size difference.

Membrane process becomes one of the most promising separation process due to its characteristics that fulfill modern needs, proposed by AURELLE [56], i.e.,

• Its efficient • Its compactness • Its closed-system characteristic, no odor, no noise pollution • Its automatic control capability

Nowadays there are a lot of books, literatures and researches on membrane processes. Membrane manufacturers can also provide useful data from their own experiences. Thus, in this chapter, the main context will emphasize on the researches of GPI lab on the treatment of oily wastewater by membrane processes. Only some basic principles and mathematical model, related to the researches, will be present to provide basic understanding. UF is the main membrane process of the GPI researches on oily wastewater treatment since its separation characteristic practically covers the range of oil droplets found in general oily wastewater, esp. macro- and microemulsion. Thus, majority of this chapter devotes to UF study. For other processes, only some basic knowledge and the significant findings from the GPI researches will be presented.

8.1.1 Classification of membrane processes

There are several types of membrane separation processes. Here will be considered only the pressure-driven membrane processes. Other processes that use other forms of driving forces, such as electrodialysis (electrical potential), dialysis (concentration gradient), pervaporation (concentration and vapor pressure), will not be considered here. Pressure-driven membrane processes can be categorized by their separation characteristics, i.e., microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). An example of relation between material sizes and their corresponding membrane processes is as shown in fig. 8.1.1-1. From the relations shown in the figure, separation characteristic of each membrane process can be summarized as shown in fig. 8.1.1-2. Membrane are classified either by pore size or “molecular weight cut-off” (MWCO). The latter is determined by relation between the sizes of materials to be separated by the membranes and their molecular weights. It should be noted that manufactures, sometimes, use different kinds of material to determine the MWCO, i.e., proteins (globulin, albumin, etc.), saccharides (dextrans), etc. Thus there is some difference in the relation between the pore size and molecular weight. It is recommended to confirm with the manufacturers. The size of material, defined by MWCO, may not be exactly equal to the pore size of the membrane. But it is the size that the membrane can be separated with acceptable efficiency (about 90%), not absolute (100%). So it is usually called “normal MWCO” or “NMWCO”, to distinguish from “absolute MWCO” that which is normally not specified.

300


III-184

MF and UF have relatively large pore sizes, so they work rather like screens or sieve filters. While separation by NF and RO, which have very tiny pore sizes, is not simply by size alone but involves more complex factors, such as osmotic pressure. So we can group the membranes processes into 2 categories, i.e., (1) MF and UF, (2) NF and RO.

Fig. 8.1.1-1 Material sizes and corresponding membrane processes (Source: Osmonics)

Fig. 8.1.1-2 Separation characteristic of membrane processes (Source: Koch membrane system)

301


III-185

8.1.2 Mode of operation of membrane processes

In membrane process, wastewater or mixture that contains the objects or materials to be separated (called “feed”) will be fed to membrane module. Only certain component of feed will be able to pass through the membrane. The component that can pass through the membrane is generally called “permeate” or “filtrate”. The rest, which is retained by the membrane, will remain in the feed side and be called “retentate” or “concentrate”.

Mode of operation, categorized by flow direction of feed, can be generally divided into 2 modes, as shown in fig. 8.1.2-1, i.e.,

• Dead-end mode- Flow direction of feed is perpendicular to the membrane surface. Retained component will accumulate on the membrane surface so its operation is rather like a cake filtration. This cake could obstruct the permeate flow, resulting in lower permeate flowrate. However, because of its simplicity, no complex piping required, it is normally used in slab-scale for preliminary testing of membrane. Anyway some upscale dead-end membrane, such as pleated MF, is also available.

• Cross-flow or tangential flow mode- Unlike dead-end mode, the feed is pumped over the surface of the membrane, resulting in two product streams i.e., permeate and retentate. The later is sometimes recycled to the feed side and fed to the membrane again until it reaches the design concentration, esp. in case that the concentrate is the required product. General commercial membranes are configured to be operated in this mode of operation. It is practically used in real life situation. So, in this chapter, we emphasize only on the cross flow mode.

a) Dead end b) Cross-flow

Fig. 8.1.2-1 Mode of operation of membrane process [11]

8.1.3 Membrane structure

Application of a membrane depends on its mechanism, which, in turn, based upon its structure. Many authors may suggest difference number of membrane structures. By some criteria, some structures may be only sub-types of other structure. However, it can be generally summarized into two major structures, i.e. symmetric and asymmetric membrane.

• Symmetric membranes are the membranes of uniform structure. Their separation mechanism can be compared to in-depth filtration. Microporous membrane is categorized in to this type of the structure. Normally, inorganic membranes, such

302


III-186

as ceramics, microporous glasses, metals, come under this structure. This type of membrane may be produced by these manufacturing processes, i.e.,

• Sintering or stretching • Casting • Phase inversion and etching • Extrusion

• More details can be provided by membrane manufacturers or from literatures and books. Sometimes, symmetric membranes are undergone coating process to achieve some specific properties. In this case, it will become asymmetric membrane with composite structure.

• Asymmetric membranes are the membranes of non-uniform structure, which separation process takes place a thin denser “skin” layer of the membrane. Their separation mechanism can be compared to screening or sieving. They can be produced of a single polymer. In this case, they will be called “integrally skinned”. Sometimes, a thin dense polymer skin is formed on the microporous support. Then, it will be called “composite” or “non-integrally skinned” membrane. Asymmetric membrane may be produced by these manufacturing processes, i.e.,

• Casting of skin and laminating to support film • Dip-coating of skin-forming polymer onto microporous support • Phase inversion • Gas-phase deposition of the barrier layer from a glow-discharge plasma • Interfacial-polymerization of reactive monomers on the surface of support film

a) Symmetric structure b) Asymmetric

Fig. 8.1.3-1 Membrane structures (Source: SCT, Millipore)

8.1.4 Membrane material

Membrane material plays very important role on separation. To obtain efficient separation, the membrane material should posses the following properties;

• Chemical resistance to feed and cleaning agents • Mechanical and thermal stability • High selectivity and permeability • Stable operation

303


III-187

The properties mentioned above depend on types of membrane material as well as feed or, in our case, wastewater. Membrane materials can be divided into 2 groups, i.e.,

• Polymeric material • Inorganic material

Some polymeric membrane materials and their properties are briefly summarized, as shown in table 8.1.4-2. For inorganic materials, they generally include ceramics, glasses and metals. Almost all of inorganic membranes are in tubular form, either single-channel or multi-channel module. For ceramic membranes, they are generally of asymmetric (composite) structure with fine-grain skin, such as titania (TiO2) or zirconia (ZrO2), over porous support, such as alumina (Al2O3) or zirconia. There are many variants of these composite materials, such as zirconia/alumina, titania/alumina, carbon/zirconia, carbon/carbon, etc. For metal membranes, stainless steel is used as support layers with sintered or coated skin, such as zirconia. Inorganic membranes are, sometimes, coated with polymeric skin to produce the membranes of required pore with very durable support layer, which can be re-coated when the skin is damaged. General advantages and disadvantages of inorganic membranes are as tabulated in table 8.1.4-1.

Properties and characteristic of the two types of membranes described in this section can be used as a preliminary guideline for material selection. However, its properties may be modified by modern coating or other process to obtain required characteristic of each application. Furthermore, there are other materials that can be used for membrane production. So it is recommended to consult the manufacturers and perform feasibility test before design the system.

a) Regenerated CA, regenerated PS (above), effect of surface thickness on void formation

b) PP(above) and Polycarbonate from irradiation and chemical process

Fig. 8.1.4-1 Membrane materials (Structures depend on manufacturing processes.) (Source: Millipore, Orelis, WWW.Scienceinafica.co.za)

304


III-188

e) Polysulfone UF d) Ceramic

Fig. 8.1.4-1 Membrane materials (Structures depend on manufacturing processes.) (Source: Millipore, Orelis, WWW.Scienceinafica.co.za) (Con’t)

Table 8.1.4-1 Advantages and disadvantages of inorganic membrane [38]

Advantage Disadvantage

Very good chemical resistance to chlorine, acids, alkalis, common solvents. However, even though the membrane is very durable, chemical resistance of other accessories, such as seals, must be considered.

Initial cost is much more expensive than polymeric membranes. However, lifetime and saving in replacement cost must be taken into account.

Wide range of pH (0.5-13, even 0-14 for some types) and temperature (up to 125o C, even 350o C for some types).

Ceramic membranes are naturally brittle. It could be damaged if dropped or subjected to severe vibration.

High pressure limit (up to 10 bar), which is usually the limit of seals.

Require high tangential velocity (2-6 m/s). So large pumping capacity is required.

Extended operating lifetime, more than 10 years-lifetime is reported.

Narrow range of pore size, normally available only in MF and UF forms.

Backflushing capability allows effective cleaning of membrane.

In general, high permeate flux.

305


III-189

Table 8.1.4-2 Summary on membrane polymeric materials [38], [54], [57]

Material Type of membrane Advantage Disadvantage

Cellulose acetate (CA)

MF, UF, RO Inexpensive Wide range of pore size, Hydrophilicity Good fouling resistance to oil and fat.

Narrow pH (3-6) and temperature range (<40oC), Moderate chlorine resistance (<1 mg/l free Cl) Biodegradable

Polyamide , aromatic (PA)

MF, UF, NF, RO

Wide range of pH (2-12) and temperature (to 70 oC) More permselectivity than CA

Low resistance to free chlorine, esp. in alkali solution Biodegradable.

Polysulfone (PS) MF, UF, RO substrate

Wide range of temperature (up to 75oC –125oC) and pH (1-13) Good chlorine resistance (to 200 ppm) Good chemical resistance to alcohols, acids, halogenated hydrocarbons

Hydrophobicity, Low range of pressure (< 7 bar for plate, <1.7 bar for hollow fiber) Moderate chemical resistance to aromatic hydrocarbons, ketones, ethers, esters

Polyacrylonitrile (PAN)

MF, UF, RO substrate

Hydrophilicity (modified form) Good chemical resistance

Narrow range of pore size Require co-polymer to make less brittle

Polypropylene (PP) MF, UF Inexpensive Wide range of temperature Good chemical resistance

Hydrophobicity

Polytetrafluoro-ethylene (PTFE)

MF, UF Very good chemical and thermal stability

Hydrophobicity Expensive

Polyvinylidene fluoride (PVDF)

MF, UF Very good chemical resistance to chlorine, acid, alkali, common solvents

Hydrophobicity (can be modified to obtain hydrophilicity), Expensive

Note: Bold letters are most usual application. [57]

8.1.5 Membrane module type

There are several types of membrane module available. However, they can be divided into 5 types (fig. 8.1.5-1), categorized by their geometry, i.e.,

• Tubular module • Hollow fiber module • Plate module • Spiral wound module • Other types, such as dead-end pleated module, rotating disc module, etc.

306


III-190

Here will be described only the first 4 types for they are normally applied on oily wastewater treatment. General characteristic of each type of membrane module can be summarized as shown in table 8.1.5-1. Please be noted that technologies are improved everyday. So some data (cost, efficiency, some features, etc) need to be updated regularly.

a) Tubular (Source: Orelis, Koch) b) Hollow fiber (Source: Koch, Romicon)

c) Plate (Source: Millipore, Orelis) d) Spiral wound (Source: Orelis, ALTO japan)

Fig. 8.1.5-1 Membrane modules

307


III-191

e) Pleated module (Source: Vivendi water, Sentry)

f) Rotating disk system (above) and vibrating membrane system (Source: IGB, VSEP)

Fig. 8.1.5-1 Membrane modules (Con’t)

308


III-192

Spir

al w

ound

(fig

. 8.1

.5-1

d)

It ca

n be

vie

wed

as

a ro

lled-

up p

late

mod

ule

arou

nd a

ce

nter

pe

rfor

ated

pi

pe

that

ac

t as

a p

erm

eate

cha

nnel

. Th

e ro

lled-

up m

embr

ane

is

then

pla

ced

in th

e sh

ell.

Equi

vale

nt t

o th

at o

f pl

ate

mod

ule.

But

SS

in f

eed

may

de

posi

t at

th

e ch

anne

l sp

acer

s. So

me

tubu

lar

or

corr

ugat

ed

spac

er

is

bein

g de

velo

ped.

Feed

cha

nnel

hei

ght

varie

s fr

om 0

.5-3

.0 m

m.,

depe

ndin

g on

the

pres

ence

of s

uspe

nded

so

lids.

Re

varie

s fro

m 1

00-1

3,00

0.

800-

1200

Mod

erat

e ($

800-

1,00

0 / m

2 )

Plat

e (f

ig. 8

.1.5

-1c)

A

mod

ule

cons

ists

of

a

flat

shee

t, a

supp

ort

grid

use

d as

a

perm

eate

flo

w c

hann

el a

nd a

m

embr

ane

plat

e.

Abo

ve

the

mem

bran

e is

a

feed

flo

w

chan

nel.

Seve

ral

grou

ps

of

laye

rs m

ay b

e st

acke

d to

form

a

mul

ti ch

anne

l m

odul

e. D

ue t

o its

shap

e, it

is re

lativ

ely

easy

for

on-s

ite re

plac

emen

t.

Feed

is

di

strib

uted

to

ea

ch

rect

angu

lar

feed

flo

w c

hann

el.

The

perm

eate

will

flow

thro

ugh

he m

embr

ane

plat

es a

nd f

low

in

per

mea

tes

chan

nels

to

the

outle

t por

ts.

t Ave

rage

fee

d ch

anne

l he

ight

is

0.5-

1.5

mm

. Su

ppor

t gr

id

of

chan

nel s

pace

r he

lps

prom

otin

g tu

rbul

ence

whi

ch i

ncre

ases

the

pe

rmea

te fl

ux.

Lam

inar

in g

ener

al.

100-

300

Mod

erat

e ($

500-

100

/ m2 )

Hol

low

fibe

r (f

ig. 8

.1.5

-1b)

The

mem

bran

e m

ater

ials

are

ca

st a

s tin

y tu

bula

r sh

ape

like

the

tubu

lar

mod

ule

but

muc

h sm

alle

r and

ther

e is

no

supp

ort

inse

rt. T

hey

are

also

com

e in

th

e fo

rm o

f she

ll-an

d-tu

be.

Like

the

tubu

lar

mod

ule,

fee

d is

dis

tribu

ted

to t

he c

ores

of

the

fiber

s. (O

nly

few

m

anuf

actu

rers

do

vice

ver

sa.)

Ver

y sm

all,

0.2-

3.0

mm

for

nd

MF

and

as s

mal

l as

40

mic

rons

for

RO

. Pr

efilt

er i

s ne

cess

ary.

Bec

ause

of

its s

elf

supp

ort

stru

ctur

e,

Bac

kflu

shin

g is

pos

sibl

e

UF

a

Lam

inar

but

hig

h sh

ear

rate

fr

om it

s sm

all d

iam

eter

(2,0

00-

16,0

00se

c-1).

V o

f 0.5

-2.5

m/s

.

1000

0-20

000

: out

side

–in

RO

15

00-5

000

: ins

ide-

out U

F

Low

[57]

Tub

ular

(fig

. 8.1

.5-1

a)

Poly

mer

ic

mat

eria

ls,

actin

g as

m

embr

ane

surf

ace,

are

cas

t on

th

e in

side

of

po

rous

tu

bula

r in

serts

. For

cer

amic

mem

bran

es,

they

are

usu

ally

mon

olith

and

se

lf-su

ppor

ting.

Th

ey

usua

lly

com

e in

the

for

m o

f sh

ell-a

nd-

tube

. The

out

er s

hell

is a

lso

used

as

per

mea

te c

olle

ctor

.

Feed

is

di

strib

uted

to

ea

ch

mem

bran

es t

ube.

It

flow

s in

side

th

e tu

be a

nd th

e pe

rmea

te f

low

s th

roug

h th

e tu

bula

r wal

ls a

nd o

ut

off t

he o

uter

shel

l at o

utle

t por

t.

Larg

e (≥

12.

5 m

m).

They

can

ha

ndle

w

aste

wat

er

with

re

lativ

ely

larg

e pa

rticl

es.

How

ever

, the

par

ticle

s of

10%

of

tube

dia

met

er o

r lar

ger s

houl

d be

sc

reen

ed o

ut.

It is

bac

kflu

shab

le.

Re

> 10

,000

. Vel

ocity

of 2

-6 m

/s

for U

F.

150-

300

Hig

h ($

1300

–15

00 /

m2 ) [

38]

Tabl

e 8.

1.5-

1 G

ener

al c

hara

cter

istic

of v

ario

us m

embr

ane

mod

ule

[38]

, [54

], [5

7]

Cha

ract

eris

tic

Gen

eral

Flow

pat

tern

Flow

pas

sage

Flo

w re

gim

e

Mem

bran

e a

rea

to

volu

me

ratio

(m2 /m

3 )

Mod

ule

cost

309


III-193

8.2 Ultrafiltration (UF)

8.2.1 Basic knowledge and working principles

8.2.1.1 Pore size and molecular weight cut-off (MWCO) of UF

Ultrafiltration (UF) membrane can retain material of the size larger than 0.001-0.02 microns (1-20 nm) [38], which is the size of finely dispersed oil droplet in emulsions. So the researches in GPI lab emphasized on the application of UF, rather than other membrane processes. UF membranes are normally specified by the molecular weight cut-off “MWCO”. To relate the membrane to MWCO, PORTER [59] proposed the relation based on globular proteins as shown in fig. 8.2.1-1. Please note that if the specified MWCO is not based on the proteins, the graph will not be valid. Hence, the exact pore size information should be confirmed by the manufactures.

1

10

100

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04

Molecular weight cut-off (MWCO) as Kilo Daltan (kd)

Poer

size

(mic

rons

)

Fig. 8.2.1-1 Relation between pore sizes and MWCO [59]

Removal or retaining efficiency of membranes are usually described in the term of “rejection (R)”, which is defined by eq.8.2.1.

R

P

CC

R −=1 {8.2.1}

Where CP = Solute or pollutant concentration in permeate CR = Solute or pollutant concentration in retentate

If the pollutants are completely rejected by the membrane, CP will become zero and R will be equal to 1 or 100%.

Typical characteristic of rejection or removal efficiency of a membrane and molecular weight is a shown in fig. 8.2.1-2. Existing UF membranes often posses “diffuse cut-off”, rather than “sharp cut-off”, characteristic because of its wide distribution of pore sizes. . NMWCO is the molecular weight at the rejection of 90%. This percentage may not be practically exact but it is the generally accepted value. So the pore size shown in fig. 8.2.1-1 is the effective value, corresponding to its NMWCO.

310


III-194

Log (molecular weight)1,000

Rej

ectio

n or

R

emov

al e

ffic

ienc

y

Sharp cut-off

Diffuse cut-off

10,000 100,0000 %

100 %

90 %

NMWCO

Ideal cut-off

Fig. 8.2.1-2 Typical characteristic curves of UF membrane

To select the appropriate membrane to separate the oil droplet or particles from the wastewater, AURELLE summarized and proposed 3 basic considerations as follows,

• Pore size of membrane: To prevent oil droplets to pass through the membrane pore, the size of the pore, firstly, must be smaller than the droplets. From researches in GPI lab, it is recommended that minimum pore size should be 1/4 to 1/3 of average droplet size. Some researches [38] even proposed the ratio of 1/10.

• Characteristic of membrane: for oil/water separation, membrane should be hydrophilic. Membrane material should not react with the wastewater, which can cause pour clogging. Hydrophilic material, such as polyacrylic, cellulose acetate, zirconium oxide, etc., is recommended. Naturally without special treatment or coating, polysulfonate tends to be fouled by oil, resulting in low flux and frequent washing.

• Operating condition: Operating pressure should be less than capillary pressure required to force the oil droplets through the membrane pores. Capillary pressure increases with the hydrophilicity of membrane and decreasing of pore size. However, if pore size and hydrophilicity are carefully selected, the capillary pressure is normally higher than recommended maximum pressure of the membrane. The capillary pressure of oil drop can be calculated by the modified form of the capillary equation (see chapter 2), proposed by NAZZAL (quoted by [10], [11]) as shown in eq. 8.2.2.

⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

−

⎟⎟⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜⎜⎜

⎝

⎛

+−+⎟⎟⎠

⎞⎜⎜⎝

⎛

−−= 1

1/3

)o/wθ3sino/wsinθ(2o/wcosθ3

rdr4

2o/wθ3coso/w3cosθ

ro/wcosθ

o/w2γcapP {8.2.2}

311


III-195

θ

rd

2r

Membrane

Oil droplet

θ

Pore

Where rd = Radius of oil droplet (L) r = Pore radius (L) θ = Contact angle γo/w = interfacial tension of oil and water (MT-2)

Fig. 8.2.1-3 Oil drop at membrane pore

8.2.1.2 Characteristic of cross-flow ultrafiltraton

Because almost all of the ultrafiltration processes in real life situation are in cross-flow mode, characteristics and theories presented in this chapter, then, will be based on cross-flow ultrafiltration. For wastewater treatment process, the aims of the process are (1) to separate the pollutants from the water and (2) to reduce the volume of the retentate as much as possible. So the UF processes are normally designed as a batch system, shown in fig.8.2.1-4, which will be used to describe the characteristics of cross-flow UF in this section. In this section, only characteristic curves are presented but the explanation and the models that govern the characteristics will be described in the next section.

Permeate

Retentate

Membrane

Feed pump

Storage tank

Feed

Po

Pi

Pp

Heat exchanger

Fig. 8.2.1-4 Typical schematic of cross-flow UF for wastwater treatment

1. Permeate flux and transmembrane pressure

From fig. 8.2.1-4, the feed will be pumped by feed pump into the cross-flow UF module. The pressure on the feed side (PF) will be higher than the pressure on the permeate side (Pp), which is normally equal to atmosphere. So the solvent and the solutes that cannot be retain will be forced through the membrane as permeate. Difference between the pressure on feed and permeate sides (eq. 8.2.3) is called the “transmembrane pressure” (Pt). Since the feed inlet pressure (Pi) and outlet pressure (Po) is not identical due to losses in the membrane module, PF is an average value of Pi and Po (eq. 8.2.4). Flowrate of permeate is usually defined as “permeate flux”, which is the flowrate per unit area of the membrane, such as litre per square meter per hour (LMH).

pFt PPP −= {8.2.3}

312


III-196

2oi

FPP

P+

= {8.2.4}

Water flux

If pure water is fed to the membrane, the relation between permeate flux and transmembrane pressure will be or almost linear, as shown in fig. 8.2.1-5a.

Flux of mixture, solution, or wastewater

If the feed concentration is constant, which can be obtained by either recycling the permeate and retentater or not returning both of them back to the feed storage tank, typical characteristic curves of permeate flux and transmembrane pressure will be as shown in fig. 8.2.1-5b. From the figure, the graph can be divided into 2 major zones, i.e.,

• Pressure controlled region- the flux is proportional to pressure. • Mass transfer controlled region- the flux is nearly constant and relatively

independent to pressure.

The graph also shows the effects of various parameters on the flux/pressure relation. Other types of graphs also exist for some types of feed. But they are minority (fig. 8.2.1-5a).

Typical water flux

Typical solution flux


Flux

An example of variants of water flux

An example of variants of solution flux

Fig. 8.2.1-5a Examples of pure water flux and solution flux characteristics

Pressure controlled region

Mass transfer controlled region

Wat

er fl

ux

Higher recirculation flowor cross-flow velocityHiger temperatureLower concentration


Flux

Fig. 8.2.1-5b Typical characteristic curves of water and solution flux VS. transmembrane pressure when feed conc. is constant

313


III-197

2. Permeate flux and feed concentration

Relations between feed concentrations and permeate fluxes in log/normal scale are as shown in fig.8.2.1-6. The typical relation is linear, as shown in curve 1. However, in case of oily wastewater, the curve is proven to be as shown in curve 2 [11], [18]. Other shapes of graphs also exist but are scarcely found.

Log (Concentration)

1 2

Cg case 1

Cg case 2

C’g case 2

Flux

Fig. 8.2.1-6 Characteristic curves of permeate flux and feed concentration

3. Permeate flux and time

Constant feed concentration

Evolutions of flux with time when the feed concentration is kept constant are shown in fig. 8.2.1-7. The flux will decrease rapidly at first, then the rate of decrease will be slower. After that the flux will be relatively constant or steady. The fluxes shown in fig. 8.2.1-4 are the value of flux in this steady zone. The period of time before the flux reaches steady value may take few minutes or several minutes, depending on the type of wastewater, type of membrane and operating conditions, as shown in fig 8.2.1-7. In fig. 8.2.1-7a, the relation that the flux increases with time is scarcely observed. It occurs when the feed concentration is very low and the cross-flow velocity is high.

Flux

Time

Higher cross-flow velocityLower concentration

Time

FluxHigher Pt

a) Effect of velocity and feed concentration b) Effect of transmembtane pressure

Fig. 8.2.1-7 Typical characteristics of permeate flux VS. time for constant feed conc. system

314


III-198

Non-constant feed concentration

Evolution of flux with in this case will be varied with wastewater characteristic, membrane properties and operating condition. For oily wastewater treatment, the retentate is returned to the storage tank where fresh wastewater may and may not be added, as shown in fig. 8.2.1-4. Then it is fed repeatedly to the membrane until it reaches a required concentration, or the flux is unacceptable low. In this case, the concentration of feed, as well as flux, varies with time.

General characteristic of UF of oily wastewater in batch process described above is as shown in fig. 8.2.1-8. At first, the flux drops relatively rapid, then rate of decrease is lower. The flux decreases slower until it reaches a certain concentration. Then rate of decrease in flux is higher until it become about zero.

Time

Flux

Fig. 8.2.1-8 An example of characteristic of flux VS. time for non-constant feed conc. system

8.2.1.3 Flux model in pressure controlled region

From characteristics of UF processes described above, Many researchers had tried to explain them in term of related theories and established mathematical models to predict the characteristics of UF. To predict flux of UF in pressure controlled region (fig.8.2.1-5b), the widely accepted model is the one based on flow through porous channel, which is governed by Hagen-Poiseuille law, as shown in eq. 8.2.5.

μμε ttp P

AP

xd

J ⋅=⋅⎟⎟⎠

⎞⎜⎜⎝

⎛

Δ⋅

⋅=

32

2

{8.2.5}

Where J = Flux (L3L-2T-1, i.e. LMH or gallon per ft2 per day (GFD)) ε = Surface porosity of the membrane dp = Pore diameter (L) Pt = Transmembrane pressure (LT-2M-1) Δx = Length of the channel, in this case, the thickness of “skin” layer of

the membrane (L) μ = Dynamic viscosity of feed (ML-1T-1) A = Membrane permeability coefficient (L)

315


III-199

The model is valid the following assumptions are satisfied [38];

1. The flow is in laminar regime. Re < 2100. It is usually true under certain conditions, such as (1) low pressure (2) low feed concentration, (3) high cross-flow velocity.

2. Density is constant. The liquid is incompressible. 3. The flow is independent of time (steady state condition). 4. The liquid is Newtonian. 5. End effects are negligible.

This equation can also be used to calculate the water flux of the membrane. Normally, it is difficult to determine geometry parameters in the equation, such as ε, Δx, dp. However, these parameters are constant. So the group of parameters in the parenthesis can be determined experimentally by UF test with a known liquid, such as water. After that, it can be applied to other liquids by changing the viscosity.

From the equation, the flux varies linearly with the transmembrane pressure. However, when the pressure is higher, the properties of liquid, such as viscosity, etc., and the effect of mass transfer phenomena is prevailing. Characteristic of UF will change to mass transfer controlled region. Eq. 8.2.4 will be no longer valid.

For the water flux, it is usually governed by eq. 8.2.4. However, deviation may occur in some cases, as shown in fig. 8.2.1-5a. This can be described by the deformation of the membrane pores under high pressure or pore blocking due to trace material in the water. The latter case usually occurs when tap water is used to determine the water flux.

8.2.1.4 Concentration polarization

In mass transfer controlled region, the flux is relatively constant and independent of pressure, as shown in fig. 8.2.1-5b. To explain this, mass transfer phenomena in the UF module should be considered. One of the important transport phenomena is the formation of concentration polarization.

When the mixtures, solution, or wastewater is fed to the UF, particles, solutes, macromolecules or pollutants are rejected by the membrane. These materials, then, tend to accumulate and form a layer of high concentration of the materials at the membrane surface (fig. 8.2.1-9). This layer is called “polarization layer”, “CP layer”, or “gel layer” [38]. However, gel layer is usaually refered to a layer of gel, which concentration is quite constant, while polaization layer is the transition zone which concentration increase from bulk concentration to that of gel. Characteristic of the layer depends on the type of the materials. This layer is believed to hinder and restrict permeate flow by one of the two mechanisms, i.e.,

1. High solute concentration increases the osmotic pressure within the polarization layer. The osmotic pressure will oppose or counter the transmembrane pressure and make the effective driving force of the process decrease. So the flux is decreased.

2. High solute concentration in the polarization layer causes the solute to transport back into the bulk liquid by concentration gradient. At certain concentration, the transport of solute to the layer by convection and the back-transport will counterbalance, resulting in steady flux.

316


III-200

Important notes on polarization layers are,

1. It is assumed to be dynamic. Changes in operating condition, such as cross-flow velocity, pressure, will change the polarization layer. The layer will disappear if the system does not operate. This fact makes it different from the fouling of membrane, which will not disappear unless the membrane is cleaned.

2. It is believed to make the flux pressure-independent in mass transfer controlled region. Change in pressure will change the polarization layer, thus, the osmotic pressure or back-transport will be changed accordingly, resulting in counterbalance of the system. For example, if the pressure is higher, the flux will be briefly higher and then drop back to the previous value.

8.2.1.5 Mathematical model for mass transfer controlled region: film theory

One of the widely used model for flux prediction in mass transfer controlled region is film theory model. It is derived by balancing the convection mass transfer and the concentration-gradient back-transfer. The model is as shown in eq. 8.2.6a

)ln()ln(P

Pg

P

Pg

CCCC

KCCCCDJ

−

−=

−

−=δ

{8.2.6a}

Where J = Flux (L3L-2T-1, i.e. LMH or gallon per ft2 per day (GFD)) D = Diffusion coefficient or diffusivity (L2T-1) δ = Thickness of polarization layer (L) K = Mass transfer coefficient (L3L-2T-1) Cg = Gel concentration CP = Permeate concentration, (= 0 since oil rejection is normally100%) C = Bulk concentration

Gel concentration is theoretically the concentration of gel layer at the surface of membrane. Cg is assumed to be constant for a given solution or wastewater. From fig. 8.2.1-5b, it shows that the flux in mass transfer controlled region still varies with cross-flow velocity. Many researchers including GPI’s [11], [18] proposed that the model could be modified to account for the effect of cross-flow velocity (V) as shown in eq. 8.2.6b. The effect of V on the flux is as shown in fig. 8.2.1-9. The values of k, K, β and Cg depend on the type of wastewater and membrane and are usually obtained by experiment.

Membrane

Polarization layer

Permeate

Bulk concentration CB

Cg

Cp

Gel layer

Cg Log (Concentration)

V1

Flux V2>V1

Fig. 8.2.1-9 Diagram of polarization layer and effect of velocity on flux in film theory

317


III-201

)ln(C

CkVJ gβ= {8.2.6b}

The equations for determining the value of K, based on Buckingham π theory are also widely used (eq. 8.2.6c - e) [38]. However, to use the equations, it is necessary to know the value of Cg and D. The value of D also depends on the type of solution or wastewater.

For turbulent flow, Re>4000 33.08.0Re023.0 ScSh = {8.2.6c}

For laminar flow, Re< 1800, Lv<L, Lc>L

33.033.033.0 )/(Re86.1 LdScSh h= {8.2.6d}

For laminar flow, Lv>L, Lc>L

50.033.05.0 )/(Re664.0 LdScSh h=

{8.2.6e}

Where Sh = Sherwood number = kdh/D Re = Reynolds number = ρdhV/μ Sc = Schmidt number = μρ/D dh = Hydraulic diameter = 4. flow area / wet perimeter (L) Lv = Entrance length, based on velocity profile or the distance from the

entrance of membrane module flow channel that the velocity profile becomes steady (L)

Lc = Entrance length, based on concentration profile (L) L = Length of membrane module flow channel (L)

Eq. 8.2.6c to 8.2.6e are universal models, derived from many assumptions. So they can provide only the approximate value of K. Eq.8.2.6a and b are also developed from concentration-gradient assumption, described in the previous section. So it may not exactly fit real flux/pressure curve. Some researches [54] reported that the concentration at the membrane surface is not constant, but varies with operating condition. However the model is accepted as an effective mathematical tool to estimate the flux in mass transfer controlled region.

8.2.1.6 Resistance model

The porous flow model and film model cannot be used to predict the flux in both pressure and mass transfer controlled region. So other models ware developed to govern the whole range of flux/pressure behavior, such as resistance model and osmotic pressure model. Here will be described resistance model for its mathematical simple that facilitates its use.

For resistance model, the concept of resistance-in-series, like in heat transfer, is adapted. General form of the model is as shown in eq. 8.2.7.

∑=

RP

J t {8.2.7}

ΣR represents the summation of various resistances (see fig. 8.2.1-10) i.e.,

318


III-202

Gel /Polarization layer

Gel resistance (Rg)

Fouling resistance (RF) (Adsorption)

Fouling resistance (RF) (Pore Blocking)Membrane

resistance (Rm)

Membrane

Feed

Permeate

PF

Pp

Fig. 8.2.1-10 Various resistances in UF processes

1. Membrane resistance or intrinsic membrane resistance (Rm): It is the resistance of membrane determined using pure water as the feed. It is the inverse of A in eq. 8.2.5. Rm is theoretically constant and depends only on the type of the membrane.

W

tm J

PR = {8.2.8}

2. Fouling resistance (RF): Membrane fouling, unlike concentration polarization, is characterized by irreversible decline in flux, caused by interaction between feed and membrane as well as deposition and accumulation of some feed components on the membrane surface (external fouling) and/or with in the membrane pores (internal fouling). The nature and extent of membrane fouling as well as evolution of fouling with time are strongly influenced by nature of membrane and characteristic of the feed. Many forms of equation to predict the fouling are proposed, such as Hermia’s [54], Cheryan’s [38], etc. Generally, it can be characterized by powered or exponential equation as examples shown in eq. 8.2.9.

0: >= ntRR noF {8.2.9a}

Or

0: >= neRR ntoF {8.2.9b}

319


III-203

Ro is initial fouling resistance. However, since fouling from feed/membrane interaction is assumed to be because of physicochemical, it will be unaffected by operating condition. So RF in this case can be included in Rm as Rm’, as shown in eq. 8.2.9c. Many researchers [10], [11], [18] studied applications of UF on oily wastewater treatment, using newly prepared cutting oil emulsions. They concluded that fouling resistance of membrane, in this case, scarcely occurs, and if any, can be assumed to be constant without varying with time. So the concept of R’m in eq. 8.2.9c is valid. However, in case of real wastewater, fouling from foreign material deposition may be present Thus RF, in this case, will vary with time and should be observed by pilot-scale test.

Fmm RRR +=' {8.2.9c}

3. Polarization or gel resistance (Rg): This resistance is caused polarization layer. In some literatures, it is divided into 2 separate resistances, i.e., gel resistance and polarization resistance. However, many researches and literatures also prove that using the resistance in a single term of Rg provides sufficiently accurate result. As described in section 8.2.1.4, the resistance is reversible and changes with operating condition. General equation of Rg is as shown in eq. 8.2.10, which α and φ are empirical constants.

tg PVR ⋅⋅= αφ {8.2.10}

Thus, from the equations of various resistance and assumptions stated above, the resistance model of UF on oily wastewater treatment can be rewritten as shown in eq. 8.2.11

tm

t

PVRP

J⋅⋅+

= αφ' {8.2.11}

Eq. 8.2.11 provides a flux/pressure curve that more or less fits the observed data both in pressure and mass transfer controlled region. Like film model, the values of φ and α depend on type of wastewater as well as membrane and have to be obtained by experiment.


Apart from feed concentration, velocity and pressure, of which effects on described in the previous section, important parameters that affect the performance of UF process are summarized below.

1. Viscosity

Viscosity is an important parameter that affects flow regime and shear rate. Thus it affects the polarization layer and make the mass transfer coefficient (k) in eq. 8.2.6 change. Generally, viscosity increases with increasing feed concentration. This can explain the inflection point in flux/concentration curve in fig. 8.2.1-6, which is the case of general oily wastewater. For oil/water mixture, when the concentration reaches certain value, viscosity, thus mass transfer coefficient, changes dramatically (see fig.8.2.2-7). So the graph is shifted from a straight line.

320


III-204

2. Temperature

Since many physical properties, such as viscosity, density, change with temperature, it is inevitably one of the influent parameters on UF performance. Permeate flux generally increases with increasing temperature. So it is beneficial to operate the UF process at the highest possible temperature, providing that there is no unusual effect such as extra fouling due to high temperature precipitation. CHERYAN [38] proposed water flux-temperature correlation as shown in eq. 8.2.12a. In the research on cutting oil emulsion treatment, both fresh and used conditions, WANICHKUL [11] used permeate flux-temperature relation as shown in eq. 8.2.12b. It was proven to be quite accurate. However, it should be noted that higher temperature may also result in lower interfacial tension of feed components, thus lower size of liquid dispersed phase, such as oil droplets in water.

CCC ooo JJJ155025

276.1615.0 == {8.2.12a} ))(0239.0( AB

CBCAeJJ oo

−−= {8.2.12b}

3. Membrane properties

Major membrane properties that affect the performance of UF are as shown below,

• Hydrophilicity: For UF of aqueous feed, generally, the membrane should be hydrophilic to avoid absorption of hydrophobic or amphoteric components in the membrane, which can cause fouling. Hydrophilicity can be roughly determined by measurement of contact angle of oil drop on the surface of membrane, as described in chapter 2. Membrane material can be used to estimate membrane hydrophilicity. However, surface-coating process, which can modify its original properties, should be taken into account.

• Surface roughness: The membrane with smooth surface is likely to foul less. Surface roughness of membrane also affects the contact angle measurement and may cause error in hydrophilicity determination.

• Charge of the membrane: To separate charged particles, the membrane should be of the same charge to take advantage of mutual repulsive force.

• Pore sizes: As described in section 8.2.1.1, the pore sizes of membrane should be suitable for the pollutants to be separated. GPI recommended pores size is around 1/3 to 1/4 of the size of the pollutants (see some exception in section 8.2.3.1). Too large pore size may affect in higher initial flux but it is clogged easily, resulting in lower flux in long term and frequent washing process. Too small pore size also results in low flux and high energy consumption.

• Turbulence: As described before, flux increases with increase in turbulence. Normally, turbulence or flow regime is controlled by cross-flow velocity. However, there are some attempts to enhance the flux by the use of turbulent promoters, such as inserting wire mesh or grid in membrane flow channel. Caution is that it also promotes clogging. Some techniques choose to move membrane, rather than liquid, to increase turbulence, such as rotating disc membrane module.

321


III-205

4. Interaction between feed and membrane

Some components in feed may react or coat membrane of certain material, resulting in adverse affects, such as fouling or decrease in rejection. For example, oils can cause fouling to PVDF membrane for the oil structure is similar to the functional group of the material. So it is recommended to perform UF test with real feedstream before designing the process.

8.2.2 UF process design for oily wastewater treatment

In this section, Design procedure and design consideration as well as some significant findings about application of UF on oily wastewater treatment, based mainly on GPI researches, are described.

8.2.2.1 Membrane selection

Pore size or MWCO: Under proper operating condition, UF membranes with MWCO of 40 to 150 kilo-Dalton were proven to be effectively used to separate oil from secondary and macroemulsion with rejection of 100% [10], [11], [18], [19], [20]. The oily wastewater tested were both in fresh and used condition with the droplet sizes around 0.11-0.21 micron for macroemulsion and 11 microns for the secondary emulsion. For microemulsion (droplet size of 20 –100 nm or 0.02 – 0.1 μ), the membranes of MWCO of 40 – 50 kD can be used with oil rejection of 100%. From GPI researches, it is recommended that pore size should be around 1/3 –1/4 of droplet size to be separated for stabilized emulsion. For non-stabilized emulsion, it is recommended to use maximum pore size of 100 nm. to prevent the oil from passing through the membrane (see section 8.2.3).

Membrane material: Apart from the pore size, membrane material and module are important parameters that affect membrane selection. Every research in GPI lab confirms that hydrophilic material is the most suitable choice for oily wastewater treatment. For polymeric materials, cellulose acetate, acrylics and polyacrylonitrile (PAN) are hydrophilic, while PVDF and polysulfone are generally hydrophobic. In GPI lab, most experiments were carried out by PAN membrane. For inorganic membrane, composite membranes with zirconia (zirconium oxide) and alumina skin also provide good hydrophilicity. However, other factors, such as limiting pressure, chemical resistance, fouling, and cost, should be taken into account for membrane material selection. These data can be obtained from pilot scale test. It must be noted that the membranes of the same pore sizes may give different performance it they are made of different materials.

Membrane module: In GPI lab, all of the UFresearches were based on plate and tubular modules. The tubular modules used in the experiments were inorganic membrane while the plate modules were polymeric types. Thus the result can not be directly compared. However, from many literatures, every type of module, described in section 8.1.5, is reported to be effective for oily wastewater treatment [38]. Hence the flux obtained, fouling and cost should be considered to select the most suitable membrane.

8.2.2.2 Prediction of permeate flux

The size of membrane is determined from the quantity of wastewater, permeate flux and required operating cycle of the system (such as continuous 48 hours, etc.). Permeate flux of UF membrane can be determined based on (1) mathematics models from section 8.2.1, (2)

322


III-206

general design criteria and (3) pilot-scale test result. The latter, if available, is the most significant data for final decision on membrane process design since modular structure of membrane makes it possible to scale-up the process from the test data in relatively linear manner. However, flux result from the models and general criteria acquired from many researches, esp. GPI researches, will be provided here to be a guideline to narrow the choices of membranes to be test or for preliminary technical and economic analysis.

1. General design criteria

Recommended design criteria from various literatures and manufacturers are summarized in table 8.2.2-1. It should be noted that the values in the table are summarized from experiences and researches with various operating conditions and types of wastewater, then, should be used as a guideline. It is recommended to review related researches for the type of wastewater to be treated.

2. UF flux prediction by mathematical models

In case that the data required for mathematical models are available, the models can be used for permeate flux estimation. Many literatures provide the values of important parameters for the models, obtained from various types of wastewater and operating conditions, which can be adapted to design the membrane process for similar wastewater.

Since almost all of GPI’s researches were interpreted using film model and resistance models. In this section is emphasized on these 2 models. Summaries of important parameters for film model and resistance model from GPI’s researches are tabulated in table 8.2.2-2 and 8.2.2-3. Applications of the models are summarized as follows,

(1) Film model: The model (eq. 8.2.6a and b) can be used to predict the flux in mass transfer controlled region, where the flux is theoretically pressure-independent. It can be used to calculate the flux at any wastewater concentration, which is very useful because volume reduction of the wastewater, that makes the concentration change, is one of major objectives of oily wastewater treatment process.

Flux/pressure relation graph for UF of macroemulsion contains an infection point as shown in fig. 8.2.1-6 [11], [18]. In this case, it can be said that there are two Cg. The first one is obtained from extending the steeper part of the graph to cross the X-axis. But it is not the real Cg and used only for flux calculation at lower range of retentate concentration. The real Cg is obtained from the graph after inflection point. For macroemulsion, the real Cg crosses the X-axis at approximately 100% concentration (fig. 8.2.2-1). This can imply that, theoretically, we can use UF to filter it until the retentate becomes water-free oil [18]. However, the flux will decrease and become so low that it may become uneconomic.

323


III-207

Ref

.

[38]

, [59

]

[54]

[38]

[38]

[38]

[38]

[38]

[38]

Rem

ark

Free

oil

mus

t be

rem

oved

fir

st.

Koc

h’s d

ata

Koc

h’s d

ata

The

auth

or

reco

mm

ende

d

GM

pla

nt,

Max

ico

Ret

enta

te

conc

.

40-8

0%

Up

to 4

0% V

oi

l

60%

Up

to 1

00%

30-5

0%

50%

oil,

10%

SS

, 99.

5%

volu

me

redu

ctio

n

Rej

ectio

n or

out

let

oil c

onc.

<10-

50

ppm

<10-

100

ppm

oil

75 p

pm

FOG

100%

re

ject

ion

Flux

/ O

pera

ting

cond

ition

50 L

MH

, 25

o C, 3

.5 b

ar

100

LMH

, 25

o C, 3

.5 b

ar

40 G

FD fo

r sp

iral,

160-

120

GFD

for

disc

, no

othe

r da

ta

spec

ified

Pore

si

ze o

r M

WC

O

50-2

00

kD

Mem

bran

e us

ed

Con

vent

iona

l UF

Koc

h 25

2, 1

” tu

bula

r ,

60 m

2

Koc

h 25

2, 1

” tu

bula

r ,

60 m

2

Mem

brex

ESP

, spi

ral

2-st

ages

syst

em o

f M

embr

ex (s

pira

l w

ound

+ ro

tatin

g di

sc)

Car

bose

p M

9

Inle

t oil/

dro

plet

s si

ze

0.1-

10%

oil

3-5%

oil

1-2%

oil

< 0.

5 %

oil

5-6%

V o

il

1,00

0 pp

m F

OG

, 50

0 pp

m S

S

20 p

pm to

tal H

C,

30 p

pm S

S

Tabl

e 8.

2.2-

1 G

ener

al d

esig

n cr

iteri

a of

UF

proc

ess f

rom

var

ious

lite

ratu

res

Was

tew

ater

Cut

ting

oil w

aste

wat

er o

r st

abili

zed

oily

was

tew

ater

Was

tew

ater

con

tain

ing

spen

t co

olan

t and

lubr

ican

ts

Oily

was

tew

ater

con

tain

ing

emul

sifie

d oi

l

Food

was

tew

ater

, was

tew

ate r

cont

aini

ng su

rfac

tant

, ch

emic

al

Gen

eral

oily

was

tew

ater

Spen

t coo

lant

Aut

omob

ile m

anuf

actu

ring

plan

t was

tew

ater

Efflu

ent o

f bio

logi

cal

treat

men

t of p

etro

leum

oil

refin

ery

324


III-208

Tabl

e 8.

2.2-

2 Su

mm

ary

of p

aram

eter

s of f

ilm m

odel

from

UF

rese

arch

es o

n oi

ly w

aste

wat

er tr

eatm

ent (

refe

renc

e te

mpe

ratu

re =

20o C

)

325


III-209

Tabl

e 8.

2.2-

3 Su

mm

ary

of p

aram

eter

s of r

esis

tanc

e m

odel

from

UF

rese

arch

es o

n oi

ly w

aste

wat

er tr

eatm

ent (

ref.

tem

pera

ture

= 2

0o C)

326


III-210

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

1 10

Retentate concentration (% V of oil)

Perm

eate

flux

(l/m

2 .h)

100

P = 1 bar, V = 1.17 m/s P = 1.5 bar, V = 1.4 m/s

Fig. 8.2.2-1 Examples of flux VS. oil concentration in UF of macroemulsion

Cg Log (C)

VrefFlux VJC,V

JC,Vref

At Pref

C

Pt

Flux JC,V

JC,Vref

Pref

Vref

V

a) Graph from film model a) Graph from resistance model

Fig. 8.2.2-2 Case 1: find flux/pressure relation when k, β, Cg and R’m are known

Table 8.2.2-4 Procedure to predict flux/pressure relation for case 1: Know k, β, Cg and R’m

Item Procedure Variables Equation 1 Find JC,V,Pref C, V

)ln(Pr,, CC

kVJ gBefVC =

{8.2.13a}

2 Find JC,Vref,Pref C,Vref)ln(Pr,, C

CkVJ gB

refefVrefC = {8.2.13b}

3 Find αC ,derived from resistance model at one value of C and two values of V

R’m (or Rm)

)VrefV

ln(

refPmRVC,J

β)refV

V(refPmRVC,Jln

Cα⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

−

−

=

{8.2.13c}

4 Find φC

PrefV,C,JrefPαcV

mRPrefV,C,JrefPC

−=φ

{8.2.13d}

327


III-211

5 Flux/pressure relation at conc. = C

tPαcVCmRtP

tPV,C,Jφ+

= {8.2.13e}

The subscript “ref” represents the parameter at reference condition that the graph 8.2.2-2a is derived from. Normally the reference condition is in mass transfer control region. Otherwise the graph will not be a straight line as shown in the figure. However the equations are proven to provide acceptable result even when the reference pressure is in the transition zone between pressure- and mass transfer control region. In any case, it is recommended to select the reference pressure as high as possible and should be greater than 50% of recommended operating range.

Eq. 8.2.13e can be used to predict flux at concentration “C” and cross-flow velocity “V” over the entire range of working transmembrane pressure “Pt”, usually recommended by membrane manufacturers. Examples of prediction result from this concept are shown in fig. 8.2.2-3. The graphs show that the results are relatively accurate (±10%).

0

20

40

60

80

100

120

140

160

180

0 0.5 1 1.5 2 2.5 3

Transmembrane Pressure (Bar)

Pred

icte

d Fl

ux (l

/ (h.

m2 ))

Observed, C = 2% Predicted, C = 2% Observed, C = 8% Predicted, C = 8% Reference, C = 4%

Fig. 8.2.2-3 Relation between UF permeate flux and Transmembrane pressure at Cref = 4% by volume of oil , V = 1.4

m/s and Predicted relations at C = 2 and 8% (Oil: Elf SeraftA cutting oil macroemulsion, Membrane: IRIS

3042 PAN )

(3.1) Case 2: Know φ, α, Cg and R’m

The procedure to find flux/pressure relation (“JC,V,Pt”) at any concentration and pressure is as described in table 8.2.2-5.

328


III-212

Table 8.2.2-5 Procedure to predict flux/pressure relation for case 2: Know φ, α, Cg and R’m

Procedure Variable Equation

1 Find JCref,V,Pref Cref, V

refPCrefαVCrefm'R

refPVCref,J

⋅+=

φ {8.2.14a}

2 Find JC,V,Pref C,V

)refCgC

ln(

)C

gCln(

VCref,JVC,J =

{8.2.14b}

3 Find JC,Vref,Pref R’m (or Rm), Vref

)refCgC

ln(

)C

gCln(

refPCrefrefV

refC'mR

refP

refVC,J

⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛

+

=α

φ

{8.2.14c}

4 Find αC, φC Use eq. 8.2.13c and d 5 Flux/pressure

relation at conc. = C

tPαcVCmRtP

tPV,C,Jφ+

= {8.2.14d}

The value of Vref in this case can be chosen arbitrarily. For Pref, it is recommend to used the maximum recommended pressure.

(2) Special case: Flux prediction for mixture of cutting oil macro- and microemulsion: Treating Wasted cutting oil emulsion is one of the main applications of UF on only wastewater treatment. Parameters related to treatment of cutting oil macro- and microemulsion are provided in table 8.2.2-2 and 8.2.2-3. However, in real life situation, many mechanical workshops of factories may use more than 1 type of cutting oil to the requirement of individual manufacturing process.

GPI lab has devised an approach to estimate permeate flux of the mixture. The approach is based on the concept that, when 2 types of emulsion are mixed, oil concentration of mixture will be the summation of the 2 initial concentration. Surfactant and co-surfactant in the microemulsion will reduce the droplet size of the macroemulsion in the mixture. However the quantity of the surfactant/co-surfactant will not be sufficient to convert all oil to microemulsion. The resultant flux, then, will be neither the same as the flux of pure microemulsion, nor that of pure macroemulsion. But it will surely fall between these 2 extreme cases. Accurate prediction may require the knowledge of the chemistry of surfactants/co-surfactants.

However, since the excess amount of surfactants/co-surfactants in the mixture is directly proportional to the ratio of the microemulsion presenting in the mixture, it can be estimated that flux of the mixture is the weighted-average of the fluxes of two emulsions.

To estimate the flux of the mixture according to the concept described above, the following procedure is recommended.

329


III-213

• Calculate total oil concentration of the mixture (Coil,mix)

micoilmacoilmixoil CCC ,.. += {8.2.15a} or

micmicmacmacmixoil RCRCC +=, {8.2.15b}

Where Coil,mix, Coil,mac, Coil,mic are oil concentration of the mixture, macroemulsion portion and microemulsion portion, respectively. While Cmac and Cmic represents the concentration of cutting oil concentrate of macro- and microemulsion portions. Rmac and Rmic are the ratio of oil in the emulsion concentrate. For macroemulsion, the majority part of concentrate is oil, Rmac is around 80%. While Rmic is around 20-30%.

For example, when macroemulsion at concentration of 4% by volume of concentrate mixes with microemulsion of 2% by volume of concentrate, if Rmac and Rmic = 80% and 20%, total oil concentration in the mixture (Coil,mix) will be (4x0.8)+(2x0.2) = 3.6% by volume of oil.

• Calculate permeate flexes of whole macroemulsion and whole microemulsion at the oil concentration of Coil,mix (Jmac:Coil,mix and Jmic,Coil,mix) by the procedure decribed in table 8.2.2-4 and 8.2.2-5.

• Estimate the flux of the mixture by the following equation. Error of prediction is around 20%.


mixJ+

= {8.2.15c}

8.2.2.3 Prediction of evolution of flux, wastewater and permeate volume with time and membrane size for batch cross-flow UF system

In real situation, UF process may be designed as batch process, continuous process, single stage process or multi stage process, depending on objectives. However, for wastewater treatment process, UF is normally designed as batch system, as shown in fig. 8.2.1-4. In batch process, the concentration of retentate will be increased up to required limit or as much as possible. Then the process will be stopped and the retentate will be hauled away for final disposal or recycling. After that the batch will start over again.

Process flux will decrease continuously with increase of the concentration. The models and procedure described in the previous paragraph can be used to predict the permeate flux at any concentration. Thus it is possible to estimate evolution of retentate volume, permeate volume, and flux with time. These data is important to design UF process to meet the required operating time. However, it must be noted that predictions are based on the assumption that,

• Flux at any moment is equal to the value calculated by the model described in the previous section.

• There is no fouling from any other foreign materials.

To find permeate volume, we consider that a small volume of permeate (dVol), passing through a UF membrane of area “A” at a small time (dt), will be defined by the following equation.

330


III-214

AdtJ(c)dVol ⋅= {8.2.16}

If the system is operated in the mass transfer controlled region, the film theory (eq. 8.2.6b) can be applied. Eq. 8.2.16, then, can be rewritten as follows,

AdtCgC

lnβkVdVol ⎟⎟⎠

⎞⎜⎜⎝

⎛⋅= {8.2.17}

If we start with initial wastewater volume Volo and concentration Co and the rejection of oil is totally completed, which is true from our every test, the concentration C will be the function of the permeate volume at that moment V.

Vol)o(VoloVoloC

C−

= {8.2.18}

Eq. 8.2.17 can be rewritten as,

AdtoVoloC

gVol)Co(VollnβkVdVol ⎟

⎟

⎠

⎞

⎜⎜

⎝

⎛ −⋅= {8.2.19a}

∫=∫

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ −

t

0AdtβkV

finalVol

oVol

oVoloCgVol)Co(Vol

ln

dVol {8.2.19b}

However, the system may not be operated in the mass transfer controlled region for the whole time. In this case, the function J(c) in eq. 8.2.16 can be calculated by eq 8.2.11. Eq. 8.2.19b is in the form of integration of [1/ln (x)], so it can not be written in general form since the equation will be infinity at x =1. However, a definite integration is possible, using numerical method that can be calculated by computer.

An example of flux/time prediction is shown in fig. 8.2.2-4. The flux is predicted based on the data (Cg, k, φ, α, β, R’m ) of fresh Seraft ABS cutting oil macroemulsion and IRIS 3042 membrane, shown in table 8.2.2-2 and 8.2.2-3. The membrane area used is 1 m2. However, in this example, it is assumed that the system is in mass-transfer controlled region. So eq. 8.2-19b is used.

The predicted flux/time relation is compared with the results from UF test on used cutting oil macroemulsion (Co = 3% V of concentrate or 2.4%V of oil, from Willamette SAS factory) from WANICHKUL’s research [11], as shown in fig. 8.2.2-4. The x-marked circles indicate the observed flux of fresh emulsion. From the graph, it shows that the model can accurately predict flux of fresh (unused) emulsion as the circles are closed to the theoretical flux curve.

Comparing with observed flux of the used emulsion, the graph shows that, at low concentration, theoretical flux is greater than observed value. This is simply because of additional fouling from foreign materials in the emulsion. From the data, the system was undergone rinsing for 5 times, every 300-400 l of permeate volume on the average. Furthermore, at low concentration, the system does not fully reach mass transfer controlled region so the predicted flux from film model is higher than the actual.

331


III-215

However, at high concentration, theoretical flux is, somehow, lower than observed value. This can be explained by partial degradation of the used emulsion. During its working lifetime, cutting oil emulsion will subject to many foreign material and operating factors, such as coated oil on specimen surface, small scraps of specimen, leaked lubricant, and heat. So its quality, as well as its stability, will gradually deteriorate. This is proven by milky appearance of used emulsion, compared with the translucent or transparent characteristic of fresh emulsion. When partial oil is destabilized to be free oil, this means the concentration of oil in emulsion form may be lower than its initial value. At lower concentration, the effect of fouling overwhelms the effect of reduced concentration, so the theoretical flux is higher than the observed value. However, at higher concentration where the concentration effect is stronger, the theoretical flux shows lower value.

Evolution of permeate volume with time, calculated from integration of eq. 8.2.19b is shown in fig. 8.2.2-5. From the observed data, it confirms that the macroemulsion can be ultrafitrated until the retentate is relatively pure oil. In this case, initial volume of 1643 l of 2.4%V of oil is ultrafiltrated to the final volume of 40 l. The theoretical time required to do so is 45 hours, compared to observed value of 34 hours. However, the theoretical volume of permeate is higher until almost at the end of the operation. Evolution of theoretical flux with concentration is shown earlier in fig. 8.2.2-1. It must be noted that eq 8.2.16 to 8.2.19 are based on the assumption that no additional emulsion is added to the storage tank during the operation. If the emulsion is added, eq. 8.2.16 must be modified to account for it.

(Used cutting oil macroemulsion :initial volume 1643 l, final volume 40 l, initial concentration 2.4% by volume of oil (not concentrate): Module UFP10: membrane IRIS 3042)

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 200 400 600 800 1000 1200 1400 1600 1800

Permeate volume (l)

flux

(l/(s

q.m

.h))

Theoretical data at P = 1.0 Bar, v = 1.17 m/s) Observed data Theoretical data at P = 1.5 Bar, v = 1.40 m/s)

= Observed data from UF test of new cutting oil at the same condition

P = 1.0 bar, v = 1.17 m/s P = 1.5 bar, v = 1.40 m/s

1

= Rinse membrane with watern

2

34

5

Fig. 8.2.2-4 Relation between Flux VS. theoretical and observed permeate volume

332


III-216

0

200

400

600

800

1000

1200

1400

1600

1800

0.000 5.000 10.000 15.000 20.000 25.000 30.000 35.000 40.000 45.000 50.000

Time (h)

Perm

eate

vol

ume

(l)

Theoretical data at P = 1.0 Bar, v = 1.17 m/s) Theoretical data at P = 1.5 Bar, v = 1.4 m/s) Observed data

P = 1.0 bar, v = 1.17 m/s P = 1.5 bar, v = 1.40 m/s

Fig 8.2.2-5 Relation between time VS. theoretical and observed permeate volume

Average process flux estimation: Since the equations stated in this section is very complex and could not be solved without calculation or computer program, a rough approach to estimate average flux has been devised and, actually, used even for real design. Average process flux (Javg) can be calculated from the flux at the beginning (Jbegin) and the end of batch (J end). In this case, the fluxes at these only 2 points are required. If the final concentration is relatively Cg, J end can be assumed to be zero. This Javg will be preliminary used to calculate evolution of permeate volume (eq. 8.2.20b) in stead of the integration result from eq. 8.2.19. Observed and calculated average flux from beginning to zero-flux from UF of macroemulsion at 4% of concentrate volume, as shown in the example, are around 47 and 44 LMH, respectively [11].

2

endbeginavg

JJJ

+= {8.2.20a}

Acc. tJVol avgpermeate = {8.2.20b}

These estimated evolution data also provides the idea about the approximate size of membrane required. If the estimated operation time is not suitable, the area of membrane can be adjusted until it results in the required operating time. If should be noted that the predicted result does not include the time for rinsing or cleaning the membrane, which may take as 1-2 hours per cycle, depending on method and cleaning agent.

8.2.2.4 Sizing of related facilities

General components of UF system for wastewater treatment are shown in fig. 8.2.1-4, i.e.,

333


III-217

1. Equalization or storage tank

Here, only single-stage batch process is considered. Details of other types of the processes, such as continuous process, multi-stage process, etc. can be found in many literatures, listed in the bibliography.

For batch process without addition wastewater during UF operation, the volume of the equalization or storage tank is simply equal to the volume of wastewater generated during the operating cycle of the UF process. In this case, the system must have at least 2 tanks. While one tank is being feed to the membrane, another is used to received the incoming wastewater. Then they will be switched over when the wastewater on the first tank is completely treated. Flux/time and wastewater volume/time relations in this case can be directly calculated by the equations described in section 8.2.2.3. Final volume of retentate in this case is not used for volume calculation. It is the simplest method for UF process calculation. However, the volume of the tanks in this case will be relatively large.

For batch process with continuous or intermittent incoming wastewater into the same tank, the volume of the storage tank can be calculated by mass balance. Minimum required storage volume is equal to the maximum difference between accumulated permeate volume and accumulate influent wastewater volume. Graphical presentation for minimum storage volume calculation is shown in fig. 8.2.2-6. Evolution of accumulated influent wastewater depends on influent wastewater flow characteristic, such as constant discharge continuously for 8 hours/ day, etc.

Time

Accumulated volume

Permeate

Influent wastewaterVolume

requiredfor storage tank

Final retentatevolume required

End of batchEnd of filtrationStart cleaning

Influent stops

Fig. 8.2.2-6 Calculation of required storage volume of equalization tank by graphical method

Final retentate volume required can be estimated from recommended Cg for similar wastewater, For evolution of accumulated permeate volume, it is more difficult to achieve. It can be preliminary estimated by the concept described in section 8.2.2.3. However, since the permeate flux depends on retentate concentration, which, in turn, depends on mass balance of recycled retentate, remaining wastewater in the tank and addition influent, eq. 8.2.19b is not valid and needs to be modified. This will complicate the calculation process even more. However, the concept of average process flux, also described in section 8.2.2.3, can be used to simplify the calculation.

334


III-218

2. Feed pump

Here will consider only the system with single feed pump, which is generally used in UF process of wastewater treatment. The feed pump is an equipment that supply wastewater to the membrane as well as maintain recirculation flowrate, thus cross-flow velocity, in UF module.

For the flowrate, the pumping system is sized to cover the maximum recirculation flowarate, which is normally higher than permeate flowrate and capable of turning down to handle the minimum flowrate.

For the pressure, the pump is sized to handle the design working pressure of the system, which may need to be varied to optimize the system performance. There are key values of pressure that should be aware of in pumping system design, i.e.,

• Capillary pressure of the membrane: Transmembrane pressure must be lower than the capillary pressure between oil droplets and membrane, as shown in eq. 8.2.2. So this pressure should be taken into account for pump selection. However, for properly selected membrane, this value is normally higher than recommend maximum operating pressure of the membrane.

• Pressure drop from membrane module friction and piping system: The pressure drop has to be taken into account in pump design. Otherwise it may not be able to provide the design working transmembrane pressure. For feed and return piping system, the pressure drop can be calculated by general head loss equations.

For membrane module, the major loss from wall friction of flow channel in membrane module can be calculated from well-known formula, such as Darcy-Weisbach’s (eq. 8.2.21a).

2

2VDLf=

majorlh {8.2.21a}

For laminar flow, Re64f = {8.2.21b}

For turbulent flow (Colebrook’s equation),

)0.5fRe

2.513.7e/Dlog(20.5f

1

⋅+⋅−= {8.2.21c}

“e/D” is ratio of roughness to diameter. For plate membrane, the flow channel is usually rectangular. D of this channel can be calculated as hydraulic diameter, as shown in eq. 8.2.21d. H and W are height and width of the rectangular channel respectively.

W)2(H4HWD+

= {8.2.21d}

The minor loss can be estimated using standard minor loss equation (hl minor (in metre) = KminorV2/(2g)). The latter becomes the majority of head loss when the recirculation velocity is high.

335


III-219

Even though this pressure drop is not high (normally < 0.5 bar), it may become of significant effect when the UF is operated at low transmembrane pressure, such as 0.5 to1 bar or when the concentration of retentate is high. For oily wastewater, viscosity will vary rapidly in some range of concentration, as shown in fig. 8.2.2-7. This will cause extra pressure drop. So, maximum viscosity expected during the operation should be accounted in pump selection.

(Elf SeralfABS cutiing oil macroemulsion, 20oC, 100% = pure cutting oil concentrate = 80% V(approx) oil )

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

0 10 20 30 40 50 60 70 80 90 10

Oil concentration (% V of cutting oil concentrate in water)

Vis

cosi

ty (c

p)

0

Fig. 8.2.2-7 Relation between oil concentration VS. viscosity of emulsion

For the power requirement of UF system, main power consumption is the power required to maintain transmembrane pressure at require flowrate (or recirculation velocity). This power can be calculated straightforwardly by the basic equation

overallηAVP

overallηQPPower ⋅⋅

=⋅

= {8.2.21e}

Q in this case means recirculation flowrate, not the permeate flowrate. V is the recirculation velocity and A is the flow area of liquid in UF module (the channel between the membrane surface and the UF module wall). P in the equation is pump discharge pressure, which is the summation of transmembrane pressure and other headloss from pipe and values system. It should be noted that transmembrane pressure is average value of the pressure at the inlet and outlet of UF module. Overall efficiency of pump depends on pump type. For progressive cavity pump or progressive screw pump, the efficiency should be around 50-70%.

In case that the data required for mathematical models are available, the models can be used for permeate flux estimation. Many literatures provide the values of important parameters for the models, obtained from various types of wastewater and operating conditions, which can be adapted to design the membrane process for related wastewater.

3. Heat exchanger

For oily wastewater treatment, when the retentate is repeatedly pumped into membrane process, it will accumulate heat loss from feed pumps and friction so it will gain in temperature. If the temperature is too high, it may cause adverse affect to the membrane and

336


III-220

seals. Thus, heat exchanger is required to maintain the temperature of feed stream within the recommended temperature range of membrane or at the design temperature. Heat load of the system depands on piping and components design.

To estimate the maximum heat load, it can be assumed that energy loss from input power supply is converted solely into heat. Thus the heat load can be roughly calculated from modification of eq. 8.2.21e as follows,

PQeat ⋅⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−= 1

overallη1H {8.2.22}

Rising in temperature (ΔT) can be roughly calculated by basic heat capacity equation (Heat = M.C.ΔT). C is specific heat. M is mass flowrate of feed pump. When the heat, ΔT, and the required temperature are known. It is possible to design or select the heat exchanger. But the design of the heat exchanger is a science into itself, which will not be described here. It can be selected from commercial product of specially designed by heat exchangers designers.

8.2.2.5 UF efficiency on oily wastewater treatment

From many researches in GPI lab, it clearly shows that, with appropriate UF membrane pore size hydrocarbon or oil, even in the form of very tiny droplets in macroemulsion and microemulsion, cannot pass the membrane. It can be said that the removal efficiency of UF is 100%. By the way, the efficiency will be lowered after prolonged operation, mainly from normal wear and tear problem. Characteristics of wastewater, such as presence of free-oil or non-stablized oil may lower the oil rejection of the membrane (see section 8.2.3-1). So oil concentration in permeate can be expected in the range of 0-100 ppm, depending on wastewater characteristic and membrane (see table 8.2.2-1).

Even though the oil rejection is generally complete, it does not mean that there is no residual pollutant in the permeate since there may be other components of pollutants in the wastewater that can pass through the membrane, such as surfactants and co-surfactants. Extent of residual pollutants depends on components of the wastewater. Examples of residual pollutants of cutting oil emulsions permeate are shown section 8.2.3. These residual pollutants need to be further treated, such as by RO, distillation, biological treatment, which will be described later in the related chapters.

8.2.3 Design consideration and significant findings from GPI’s researches

In this section, design consideration and significant findings from related researches of GPI’s lab on oily wastewater treatment by UF will be described.

8.2.3.1 Design consideration and significant finding on secondary emulsion treatment

Almost all of the researches on oily wastewater treatment by UF are based on stabilized emulsion since it cannot be separated effectively by other single process. However, SRIJAROONRAT [10] has researched on UF of non-stabilized secondary emulsion. Her study provides useful information to understand the performance of UF when free-oil is present, which is frequently found in general oily wastewater. Significant findings of the research can be summarized as follows,

337


III-221

1. Flux/pressure curve of non-stabilized emulsion shows some decline at high pressure (fig. 8.2.3-1a). For this, SRIJAROONRAT described that it causes by pore clogging of the oil film. Since the oil is not stabilized, it can be coalesced and coated the membrane surface. At high pressure, the oil film is pinned against the membrane and causes the blocking more predominant. Flux was reported to range from several hundreds to few thousands LHM [10]. However, its feasibility should be considered comparing to other possible processes, such as coalescer.

2. If the pore size of membrane is selected by the same criteria as for stabilized emulsion, i.e. 1/3 to1/4 [GPI] or 1/10 [38] of droplet size, the pore will be relatively large, which causes lowering in capillary pressure. Thus the working pressure may high enough to overcome the capillary pressure, resulting in presence of oil in permeate. Permeate flux in this case will be very low. Evidence on coalescing of oil drop in the permeate is also observed. Thus it is recommended to select the pore size of 50-100 nm for non-stabilized emulsion treatment. If the pore size is properly selected, oil rejection will be or almost 100% (fig. 8.2.3-1b)

3. Presence of surfactant causes adverse affect on flux and oil rejection. SRIJAROONRAT showed that, at ratio of surfactant to oil of 2% and Pt = 3 bars, the oil can pass through the membrane and present in the permeate even when the membrane of 100 nm is used. Operating at low-pressure (i.e. 1 bar) help alleviating this problem. Her research can be used to explains why the rejections of UF of real wastewaters are not equal to 100%, unlike in the lab result. This is because free oils or tramp oils are usually present in real wastewater while they are usually not present in synthetic wastewater used in the experiment.

4. Residual pollutant, measured in term of TOD, varies with membrane’s efficiency or oil rejection as well as presence of surfactants/co-surfactants or other soluble components in the wastewater. In case that the surfactants are present, it will tribute to main TOD in the permeate. And if their concentration are high enough, they will make the oil pass through the membrane. The TOD in this case will be very high from presence of oil in the permeate.

Flux of stabilized emulsion

Pt

Flux

Flux of non-stabilized emulsion

Fig. 8.2.3-1a Flux of non-stabilized emulsion

Fig. 8.2.3-1b Photographs of non-stabilized emulsion influent, retentate and permeate (from left) [10]

338


III-222

8.2.3.2 Design consideration and significant findings on macroemulsion treatment

Cutting oil emulsion is the major source of this type of wastewater. All of related GPI’s researches prove that UF process is very efficient for treatment of this wastewater. Design consideration and significant findings from GPI’s lab on macroemulsion treatment are summarized as follows,

1. The value of Cg

Flux/pressure curve of macroemulsion follows the typical pattern as shown in fig. 8.2.1-5b. But for the flux/concentration relation, BELKACEM [18] proven that the relation can be approximated as being linear with an inflection point, as shown in fig.8.2.2-1. The real Cg is about 100% of oil. It means that we can recycle the retentate to the UF module until it is about pure oil. This result was also confirmed by WANICHKUL [11] (see example in section 8.2.2.3). BELKACEM explained the lower values of Cg from other researches by the assumption that they are tested within relatively short periods of time. Thus extension of the results in log/normal scale will intersect the concentration axis at the first virtual Cg (Cg’ in fig. 8.2.1-6 and 8.2.2-1) or somewhere between C’g and Cg.

2. Flux enhancement by salt addition

Flux enhancement by salt addition was confirmed by many researches [11], [18], [20]. The salt used in the experiments was CaCl2. Working principle of the process is that the salt will be added to the wastewater in a certain amount to cause partial, not complete, destabilization of the emulsion. Typical effect of salt concentration on the flux from dead-end batch test at low Pt is as shown in fig. 8.2.3-2a. In cross-flow module, the graph is almost identical but the difference between complete and partial destabilization is relatively smaller and the salt requirement for partial destabilization is slightly higher from the effect of cross-flow to polarization layer formation.

Partial destabilization

CaCl2 concentration

FluxCompletedestabilization

Lower initailoil concentration

Partial destabilization

CaCl2 concentration

FluxCompletedestabilization

a) From dead-end mocule b) From cross-flow module

Fig. 8.2.3-2 Typical relation between CaCl2 concentration and flux at low Pt

The effect of salt that cause changing in stability of the emulsion is reducing the repulsive forces between droplets as well as potential barrier by modification of Zeta potential and conductivity of the solution. This favors coalescence of oil droplets (see fig. 8.2.3-3).

339


III-223

Ca+

Ca+

Oil

Surfactant/Co-surfactant

--

Fig. 8.2.3-3a Partially destabilization by salt and coalesce of oil droplets

Fig. 8.2.3-3b Magnified images (x100) of oil droplets from original (left) and partially destabilized macroemulsion (right) [11]

Fig. 8.2.3-3c Magnified images of new membrane surface (left) and the surface after UF of macroemulsion at 3 bars without salt (middle) and with salt addition (600 mg/l CaCl2) [11]

The amount of salt for partial destabilization is not enough to destabilize all of the oil droplets. But in polarization layer, which the concentration is locally high, oil droplets are closely packed. The effect of this amount of oil is enough to provoke coalescence. The coalesced droplets will become free-oil, floating on the top of storage tank. Thus the concentration of emulsion fed to the membrane is relatively lower than the value calculated from volume reduction (or concentration factor) (fig. 8.2.3-4). So the flux is higher.

The extent of flux increase depends on wastewater characteristic and operation condition as shown in fig. 8.2.3-2 and 8.2.3-4. Flux enhances up to 2-3 time of no-salt condition were reported [18]. For residual salt pollutant, if the optimal dose of salt is added, 20% of the added amount will be found in the permeate. The salt exceeding the optimal dose will pass through the membrane without any retention and found in the permeate. Optimum dose of CaCl2 for macroemulsion, based on Elf Seraft A cutting oil macroemulsion [18], is about 34 mg/l of Ca++ or 125 mg/l of CaCl2 per 1% by volume of oil in wastewater.

340


III-224

Flux of stabilized emulsion

Pt

Flux

Flux of partiallydestabilized emulsion

Theoretical

Observedconcentration

Calculatedconcentration

Free-oil is not separated from storage tank.

Free-oil is separated.

Co

Fig. 8.2.3-4 Relation between fluxes VS. pressure and calculated VS. observed oil concentration in retentate for partially destabilized emulsion

Other significant findings and precautions on flux enhancement by salt addition are summarized below.

• Other types of salt can be used, such as NaCl, ferric salt or aluminium salt, etc. The higher the valence electron of the salt, the better the destabilization as well as the lower the salt concentration required. Thus NaCl is considered not suitable for it required concentration is high and cause high chloride in permeate. For Fe and Al salts, they are more expensive than CaCl2. Furthermore, laws and regulation in many countries limit presence of high-valence ions in effluent.

• Presence of free-oil may cause some problem at high Pt and V, as described in section 8.2.3.1. It can cause pore blocking, resulting in decrease in flux at high Pt (see fig. 8.2.3-4), esp. at high value of V because it can cause high mixing effect and turbulence, so the free-oil in the surface of tank is drawn back with the feed stream to the UF module. This effect is also more eminent when the membrane is hydrophobic (such as PVDF or Polysulfone). Thus, it is recommended to separate this free-oil off the recycled stream, by coalescer.

• Some membrane material, such as ZrO2, may interact with calcium complexes and cause serious decrease in flux. Furthermore, membrane with large pore size (i.e., 150 kD) will be blocked relatively easily by free-oil, as described in section 8.2.3.1.

3. Residual pollutant in permeate

Oil rejection or removal efficiency is about 100%. Theoretical oil concentration in permeate is 0 ppm (see fig. 8.2.3-5). However, actual oil concentration in permeate may greater than 0 ppm, depending on condition of equipment, wastewater characteristic and operating condition, such as presence of free-oil or fouling materials (foulants).

Even though the oil is about completely rejected by UF. Permeate still contains surfactants/co-surfactants and some additives that, naturally, can pass through the membrane.

341


III-225

Some surfactants may react with other substances and form complex or integrates with its species to form micelles that can not pass the membrane. However, it can be safely assumed that the concentration of surfactants/co-surfactants in permeate is equal to free surfactants/co-surfactants in inlet wastewater. For example, Elf Seraft A concentrate contains 10% by weight of surfactant (sodiumsulfone, sodium carboxylate, triethanolamine carboxylate), 5% of co-surfactant (benzylic alcohol), and 4% additives. Its permeate contains pollutant, measured as TOD, around 650 mg/l per 1% by volume of oil in the inlet wastewater. Many researches [10], {11], [18] pointed that TOD removal efficiency of UF for macroemulsion is greater than 95%.

In case that the salt is added to the wastewater, surfactants/co-surfactants will react with the salt to form complex. However, there is not enough data to conclude its effect on residual pollutants in permeate. So, to be on the safe side, it can be assumed that the TOD in this case is identical to the no-salt case. And the salt will also be present in the permeate as described in the previous paragraph.

a) Fresh macroemulsion b) Used macroemulsion c) Microemulsion

Fig. 8.2.3-5 Examples of feed emulsions and their corresponding UF permeates

8.2.3.3 Design consideration and significant findings on microemulsion treatment

Design consideration and significant findings from GPI’s lab on microemulsion treatment are summarized as follows,

1. Flux/pressure relation and Cg

Flux/pressure relation of microemulsion is of typical pattern, like macroemulsion. Even though the droplet size of microemulsion is always smaller that that of macroemulsion, Its flux may be higher or lower than macroemulsion’s, depending on its components. For example, Elf Seraft A macroemulsion (dE 150- 200 nm) has higher flux than Elf Emulself G3 microemulsion (dE = 50 nm.) but lower than Emulself XT (dE = 20 nm) at the same oil concentration and operating condition. For the Cg of microemulsion, there is no experiment that was conducted until the flux was about zero. The maximum value, proposed by BELKACEM and based on Elf Emulself G3, (see table 8.2.2-2), is 45% by volume of oil.

2. Oil rejection and residual pollutants in permeate

Recommended MWCO of UF membrane for microemulsion treatment is around 40-50 kD. Oil rejection, based on fresh emulsion, is 100%. For used emulsion, presence of free-oil or foulant may cause decrease in oil rejection.

342


III-226

Permeate of microemulsion contains much more residual pollutants than that of macroemulsion since it contains very high concentration of surfactants/co-surfactants and additives (around 80% by weight). For example, Elf Emulself G3 microemulsion contains surfactants/co-surfactants and additives around 80-85%. Its permeate contains TOD of 2,660 mg/l per 1% by volume of oil in the inlet wastewater. Many researches [10], {11], [18] pointed that TOD removal efficiency of UF for microemulsion is around 80-90 %. Because of high residual TOD, the permeate need to be further treated by RO or other feasible processes.

Flux enhancement by salt addition had never been studied in GPI’s lab. Theoretically, the effect of salt in this case should be identical to that of macroemulsion. However, since concentrations and types of emulsifier of the two emulsions are greatly different. Salt requirements should be different. Thus it is recommended to perform UF test to find salt requirement, rate of increase in flux and residual salt in the permeate before design.

8.2.3.4 Fouling and cleaning of membrane

Fouling is a major problem of membrane process. Unlike polarization, Decrease in flux from fouling is irreversible. The flux will decrease with time, even when operating condition is constant, and not recover unless proper regeneration or cleaning process is applied. Fouling is represented in the resistance model in the form of “fouling resistance” RF (see section 8.2.1.6). Its mathematical equations are of power or exponential types. The exact equation for each wastewater is achieved only from the experiment.

There are several references, literatures and books about fouling and cleaning of membranes, as listed in bibliography. Here, only fouling and cleaning related to oily wastewater will be mentioned.

1. Typical foulants

Fouling materials, which are usually found in oily wastewater, include.

• Hydrocarbons, fats, oils and grease: These substances are usually listed as a major foulant for UF process. Their presence should be avoided or minimized. However, for oily wastewater treatment, its rejection becomes the objective of the process. Free-oil cause more problems than emulsified oil, as described in section 8.2.3-1. To reduce the fouling from oil, one should start with the right membrane material. Hydrophilic membrane, such as PAN, is recommended. Free oil should be removed from UF feed stream.

• Suspended solids: For cutting oil wastewater or wastewater from mechanical workshops, scraps or small bits of materials from manufacturing processes are major source of SS. Extent of clogging depends mainly on the size of the SS. It is recommended to separate these materials from UF feed steam. Heavy and relatively larger scraps can be settled. For smaller SS, it may be separated by physical processes, i.e. hydrocyclone or prefilter. Generally, the materials, which are larger than 1/10 of membrane flow channel, should be prefiltered.

2. Flux enhancement

Many techniques are invented to slow down fouling process, resulting in longer period of high flux. Examples of these techniques, shown in fig. 8.2.3-6, include,

343


III-227

• Back flushing (sometimes called “backpulsing”, “backwashing”) This method has been tested by SRIJAROONRAT [10], using “Membralox T1-70” inorganic tubular module, active surface of 0.005 m2, equipped with automatic backflushing equipment. After normal UF operating of 1-3 minutes, the permeate of 3 ml is automatically injected in the direction of permeate side to feed side every 0.7 to 5 seconds. For UF of kerosene/water secondary emulsion with backflushing, increases in flux up to 2 times were reported. In present, this technique is commercially available, esp. for ceramic membrane.

• Turbulence promoter: Moving of membrane, such as in rotating disc module, or insertion of wire, mesh into the membrane flow channel can increase turbulence, thus increase permeate flux.

• Other techniques, such as (1) co-current permeate flow, which required a permeate pump, (2) permeate backpressure by throttling of permeate outlet valve to optimize Pt or (3) use of pulsating feed, etc.

Fig. 8.2.3-6 Examples of flux enhance techniques [38]

3. Cleaning of membrane

When flux decreases to an unacceptable level, the membrane needs to be cleaned to restore the flux up to or almost to the original value. Cleaning procedures always involve combination of these means, i.e.

• Chemical, in the form of detergents, acids, alkalis, water or cleaning reagents • Thermal, in the form of heat • Mechanical, in the form of hydrodynamic force.

To remove oil from the membrane, many reagents are recommended such as alkali, alcohol, detergents or even pure water. Selection of cleaning reagent depends on type of product (for manufacturing process), type of foulants, chemical and thermal resistance of membrane material.

For oily wastewater treatment, rinsing with water and/or cleaning with detergents or surfactants are generally used. Rinsing with water is proven to be efficient enough for cleaning UF membrane of used macroemulsion treatment from mechanical workshop [12],

344


III-228

(see fig. 8.2.2-4). Use of surfactants provides better cleaning efficiency pluses that the surfactants will somehow bond to the membrane surface in the form of oleophobic layer consisting of micelles, resulting in more hydrophilicity of membrane. This oleophobic layer, formed by surfactants, will prevent the oil from attaching to the membrane. Thus, it helps reducing fouling, resulting in prolonging period of high flux [18], (see fig. 8.2.3-7).

-

-

- ---

-

Fig. 8.2.3-7 Example of evolution of flux of macroemulsion UF with periodical cleaning with macroemulsion (membrane IRIS 3042, P = 1 bar, V=1.5 m/s. 25oC) and schematic of

interaction between membrane/surfactants [18]

Cleaning of UF membrane by microemulsion

GPI lab had researched on the use of microemulsion as a cleaning reagent [18], [11], since it contains high concentration of surfactants. Recommended cleaning procedure starts with rinsing of the membrane with water for 20-30 minutes, then circulation of the microemulsion for 15 minutes. WANICHKUL [11] recommended another water einsing after the emulsion circulation. Throttle valves in the system should be fully open during the cleaning process. Recommended concentration for the cleaning solution is Around 2% by volume of concentrate. The used cleaning emulsion can be stored and used for several times until it is saturated by oil. The result shows that the procedure can restore the flux up to 98% of original value. Advantages of this concept include;

• Microemulsion is normally available in factories or mechanical workshops. No need to purchase surfactants. Its efficiency is very good. It also helps forming micelles as described in the previous paragraphs.

• The result is relatively identical to the use of surfactants since it can prolong period of high flux as shown in fig. 8.2.3-7.

• For its very high surfactant concentration, it can be reused for several cleaning cycles until it is saturated by oil. Thus this concept is quite economical.

• Used cleaning microemulsion can be treated by chemical breaking, following by UF.

• Apart from the use of commercial microemulsion, BELKACEM also recommended a formula of the special cleaning microemulsion, which generates low-pollutant permeate from its treatment by UF. Its concept is based on the use of low-solubility surfactants/co-surfactants. The

345


III-229

components of the cleaning emulsion are as shown in table. 8.2.3-1. The permeate from treatment of the cleaning emlsion by UF contains TOD of 3 g/l, much lower than that of conventional microemulsion’s permeate. Recommmended concentration to create cleaning solution is 4.5% as volume of the concentrate.

Table 8.2.3-1 composition of the special cleaning microemulsion concentrate [18]

Components Name/ type Percent by weight

Base oil mineral 32% Surfactant Alkyl benzene sulfonate 22% Co-surfactants Alkyl phenol ethoxylate (non-ionic) 22% Oleic acid 5% Anticorrosion Oramide DL810 (by SEPPIC) 8% Bactericide Seppicide HB (by SEPPIC) 4% Antimousse Antimouse 411 2% Water 5%

8.2.3.5 UF test

Since the performance of UF is sensitive to types of membrane and wastewater characteristics, which may varies greatly from source to source, it is strongly recommended to perform UF test to obtain the exact data on UF performance for the wastewater to be treated.

Almost all of membrane manufacturers provide feasibility test service and produce pilot-scale or lab-scale UF test module. The UF test module can be divided into 2 major types, i.e.,

• Dead-end test module, as shown in fig. 8.2.3-7a, b. This type of test module is basically operated in dead-end mode, with some equipment or mechanism, such as mechanical agitator or small recirculation pump, that provide turbulence for polarization control. Driving pressure generally comes from an air compressor or compressed air cylinder. This type of model is inexpensive and takes relatively short test time. However, even though it is equipped with polarization control device, the effect is not identical to the real module, which is operated in cross-flow mode. It is difficult to predict the relation between result from the test module and real cross-flow module, esp. when free-oil is present in the wastewater. So the module is normally used to find preliminary result on flux and fouling of membrane and for comparison purpose between various types of membranes.

• Realistic test module, as shown in fig. 8.2.3-8c, d. This type of test module is practically a scale-down version of the real UF module. So the result from this test module is more realistic and can be straightforwardly scaled-up for design of the full-scale system.

346


III-230

a) Dead- end test module (Source: Millipore)

b) Plate membrane test module (Source: Orelis, Osmonics)

c) Spiral wound test set (Source: Millipore) d) Pilot-scale test set (Source: Millipore)

Fig. 8.2.3-8 Examples of UF test modules

Almost all of the researches on oily wastewater treatment by UF are based on stabilized emulsion since it cannot be separated effectively by other single process. However, SRIJAROONRAT [10] has researched on UF of non-stabilized secondary emulsion. Her study provides useful information to understand the performance of UF when free-oil is present, which is frequently found in general oily wastewater. Significant findings of the research can be summarized as follows,

8.2.3.6 Miscellaneous design consideration

Most of the design considerations for each category of oily wastewater are described in the previous sections. In this section, some general remarks and common design considerations, summarized from synthesis of literatures and GPI researches, will be presented as follows,

• From GPI’s researches as well as other literatures, it is recommended to operate UF system for oily wastewater around a certain flux to prolong the operating time per cycle (before cleaning) and maintain relatively steady flux. This concept is called “critical flux”. At this flux, deposition of foulants can be counterbalanced by shear force from cross-flow velocity. Thus the fouling resistance is relative steady (not changing with time). This can be normally achieved by the use of

347


III-231

moderate pressure (1-1.5 bar). The use of higher pressure may give higher flux but the rate of decrease in flux is higher to, resulting lower overall flux.

• UF performance is sensitive to type of membrane, characteristic of wastewater and operating condition. So it should be very careful to use the data from literatures or researches to design the system. Experimental procedures of the researches should be considered. The common mistakes on the use of researches on UF of oily wastewater treatment are about, (1) pore size, sometimes MWCO and pore sizes are confusing, and (2) concentration of oil, they are reported in various forms (% by volume of oil, % by volume of concentrate in case of cutting oil emulsion, % by weight, etc.), (3) Effect of concentration factor, some experiments carried out by recycling the permeate to the feed stream, so the concentration of feed were constant. The concentration effect is not included.

• The cost of membrane system is high and the performance on secondary emulsion treatment is not better than general separation processes. The UF, then, should be used for some specific wastewater, separating from main stream wastewater.

8.3 Microfiltration (MF)


Microfiltration (MF) membrane can retain material in the range of 0.1 to about 5 microns (100-5,000 nm). It is mainly used as a clarification technique, separating suspended solids from dissolve substance [38]. Size of MF is defined in the form of pore size, rather than MWCO. MF membranes are available in various geometry and materials as UF. Its main applications are separation of suspended particles in biotechnology, food and pharmaceutical industries such as bacteria, red blood cells, latex emulsion, dairy product, etc. Examples of MF membrane are as shown in fig. 8.3.1-1

Tubular module (Source: Koch)

b) Plate module (Source: Millipore)

c) MF membrane structure (Source: Millipore)

Fig. 8.3.1-1 Examples of MF membranes

348


III-232

Theoretically, working principle of MF is relatively identical to that of UF. Theories, models and characteristic curves of UF can be applied to MF. Main difference between MF and UF is the pore size. Since the pore size of MF is relatively large. Its corresponding capillary pressure is relatively low (around 0.1 bar). Thus, its maximum or limiting pressure is much lower than UF. Its pore is about the same size as droplets in macroemulsion, so some oil droplets can pass the MF membrane and be present in permeate. This is the main reason why MF had not been studied seriously on oily wastewater treatment.

However, GPI lab had researched on the feasibility of combination between chemical destabilization and MF for macroemulsion treatments as will be described in the following section.

8.3.2 Significant findings on MF for oily wastewater treatment from GPI researches

As stated above that MF pore size is close to that of oil droplets in macroemulsion, to use MF for this wastewater, the droplet size must be increased. MATAMOROS applied the concept of flux enhancement by salt addition, like the case of UF, to MF. Significant findings from the research are as described below,

1. Recommended pore size is around 0.2 μ for MF (with salt addition) of macroemulsion treatment. The use of too large pore results in very low or zero oil rejection. Both oil and water can pss through the membrane. The use of smaller pore is not possible, since its maximum pressure, calculated from pore size, is not enough to drive the separation process. Both water and oil cannot pass the membrane. If we increase the pressure, it can force the oil through the membrane, resulting in very low or zero oil rejection anyway.

2. Quantity of salt required is much higher (about 3 times) than that of UF and relatively close to the quantity required for total destabilization. Optimum dose of CaCl2 for macroemulsion, based on Elf Seraft ABS cutting oil macroemulsion [20], is about 95 mg/l of Ca++ or 350 mg/l of CaCl2 per 1% by volume of oil in wastewater.

3. Result from dead-end test module of MF with salt addition shows very high flux, relatively close to that of water flux. However, in case of cross-flow module, the initial flux is about identical to the water flux at the beginning of the operation. Then, the flux decreases rapidly within very short time (few hours), more rapid than that of UF. The steady flux is relatively low and may not be higher than that of UF’s flux (see fig. 8.3.2-1). Steady flux, based on Seralf ABS 4% at 0.1 bar, is around 25 LMH.

349


III-233

Fig. 8.3.2-1 Evolution of flux from MF (with salt addition) of macroemulsion [20]

4. The difference between dead-end and cross-flow model may be caused difference in test volume since the dead-end module is much smaller than the pilot scale cross-flow module.

5. Rapid decrease in flux can be explained by deposition and coalesce of free-oil at the surface as well as within the pore size of membrane since the pore size of MF is relatively large.

6. When the membrane is properly selected, oil rejection is about 100% at low operating pressure (about 0.05-0.1 bar). Increase in pressure beyond the capillary pressure results in presence of oil in permeates. Since the pressure is much lower than (about > 10 times lower) that of UF, energy consumption of MF is very low, compared to UF.

7. Residual TOD comes from the surfactants/co-surfactants. Concentration of residual salt and TOD are relatively identical to that of UF with salt addition, described in section 8.2.3.2.

8. To reduce fouling, MATOMAROS recommends the use of other separation process, such as coalescer or hydrocyclone, to separate the free-oil from MF feed stream.

9. Since quantity of salt added into the wastewater is high, the emulsion is largely destabilized already since it is in the storage tank, indicated by presence of oil layer on the surface. Thus, the use of MF should be compared to the combination of destabilization and other basic process, such as decanter, coalescer.

350


III-234

8.4 Reverse osmosis (RO)

Reverse osmosis is the ultimate pressure driving membrane process that capable of retaining ionic-range materials, the range of 0.1 nm. Structure of RO membrane is dense, compared to porous structure of UF and MF’s (see fig 8.4-1). Its main application is production of ultra pure water for many industries, such as, pharmaceutical, electronic, food, and nuclear (water for reactor). RO membranes are available in various materials and modules, i.e. plate, etc., as described on section 8.1.

Cutaway-view of RO module (Source: Desal)

RO pure water system (Source: Koch)

Fig. 8.4-1 Examples of RO membranes


Working principles of RO is based on the concept of osmosis. Osmosis phenomenon can be explained by considering the system of 2 compartments, separating by dense membrane, as shown in fig. 8.4.1-1. Each compartment contains the solution at different concentration.

Osmosis phenomenon is defined by mass transfer of the solvent through the membrane, from the diluted solution to the concentrated one. Increase in pressure on the concentrated side can lower or stop the mass transfer. To stop the mass transfer, the pressure required shall be equal to “osmotic pressure”, as shown in fig. 8.4.1-1b. If the applied pressure is higher than the osmotic pressure, the solvent will inversely travel from concentrate side to the diluted side (fig. 8.4.1-1c). This is the phenomenon taking place in RO and give the process its name.

DC

Posmotic

DC

P = Posmotic

DC

P = Posmotic

a) Osmosis phenomenon b) Osmotic pressure c) Reverse osmosis

Fig. 8.4.1-1 Working principles of reverse osmosis

351


III-235

Osmotic pressure is the function of concentration and usually defined by Van’t hoff as shown in eq. 8.4.1.

...33

221 +++= CACACAπ {8.4.1}

Where π = Osmotic pressure C = Concentration of solution An = Virial coefficients of the solution

The value of A1, A2, A3, etc. depends on type of solution. The virial coeeficients of large molecules or materials are relatively small. Thus the osmotic pressure is low and usually negligible, as in case of MF and UF. For low molecular-weight molecules, supposed to be separated by RO, the values of high-ordered A are high so the osmotic pressure becomes significant.

Mathematical model for RO

Mathematical model of RO can be written in the form of resistance model, as shown in eq. 8.4.2. For RO, transmembrane pressure is countered by osmotic pressure, so the effective driving force for the process is the difference between the two pressures. RO resistance mainly consists of the membrane resistance. Mass transfer resistance in the boundary layer may be accounted in some cases.

M

Mt

RP

Jπ−

= {8.4.2}

Where J = Permeate flux πΜ = Osmotic pressure, calculated from the concentration at the

membrane surface RM = Membrane resistance

RO may find its application in industrial water reuse process. However, due to its minuscule pore size and very high osmotic pressure, pressure requirement, thus energy consumption, of RO is very high. So it is scarcely used on general wastewater treatment. In GPI lab, RO had been studied for its performance on treatment of highly polluted permeate from UF of wasted emulsion. The studies will be summarized in the following section.

8.4.2 Significant findings on RO for oily wastewater treatment from GPI’s researches

BELKACEM [18], MATAMOROS [20] and WANICHKUL [11] had studied on application of RO on treatment of highly polluted permeates from treatment of wasted cutting oil emulsion. Significant findings of the researches are as summarized below,

1. RO is proven to be an efficient process for treatment of UF permeate of macro-or microemulsion, which contains very high concentrations of surfactants/co-surfactants that are not retained by the UF membrane. When retentate and permeate are recycled, relation between flux and pressure from RO of microemulsion’s permeate from UF is linear up to the maximum test pressure of 60 bars as shown in fig. 8.4.2-1a.

352


III-236

Pt

Flux

Lag due to osmotic pressure

Min recommended pressure

Log (conc.)

Flux

Fig. 8.4.2-1a Typical relation between flux and pressure of RO

Fig. 8.4.2-1b Typical relation between flux and log of concentration

2. Evolution of RO flux from treatment of microemulsion’s and macroemulsion’s UF permeate, both in used and fresh condition, are relatively identical as shown in fig. 8.4.2-2. The graph is for RO operation in batch process (see fig.8.2.1-4) without return of permeate so effect of factor of concentration is already included.

0

20

40

60

80

100

120

140

160

180

0 2000 4000 6000 8000 10000 12000 14000

Time (sec)

Flux

(L/(m

2.h)

)

Microemulsion Macroemulsion Used macroemulsion

Fig. 8.4.2-2 Examples of flux evolution from RO of the UF permeates of various emulsions (the RO permeate are not recycled) [11]

3. Flux decreases with increase in concentration factor (fig. 8.4.2-1b). There is no clear evident if the relation between flux and factor of concentration will have inflection point like that of UF of macroemulsion (fig. 8.2.1-6) since the experiments were usually stopped before the zero-flux was reached. The maximum factor of concentration (or volume reduction factor) ever tested in GPI lab is 2.5 (at 25 bars, 30oC, V = 1.8 m/s)[18].

4. BELKACEM [18] reported that permeate’s TOD generally increases with time or concentration factor (shown in fig. 8.4.2-3a). However, WANICHKUL showed [11] that TOD may decrease with time or concentration factor, as shown in fig. 8.4.2-4. Decrease in permeate TOD, in this case, may be explained by

353


III-237

formation of micelles on the membrane surface when the concentration of feed is continuous increased along the process time until it reaches the critical micelle concentration (CMC) [11]. These micelles cannot pass through the membrane so the quantity of pollutant in permeate is decreased. Formation of micelles depend on type and concentration of surfactants/co-surfactants in the permeate, as well as interaction between those pollutants and other material such as calcium that may cause formation of complexes as well as electrical charge of membranes. Thus, exact result should be obtained from RO test.

5. Typical relation of rejection or removal efficiency with pressure, based on salt, is as shown in fig. 8.4.2-3b [59]. But MATAMOROS [20] indicates that, for UF permeate of microemulsion, the rejection decreases at high pressure (the dashed line in fig. 8.4.2-5b). There is not sufficient data to evaluate the cause of this decrease. However, it is recemmeded to use moderate transmembrane pressure (about 20-25 bars) to obtain good rejection.

Concentration factor

Rejection andTOD permeate

1

Rejection

TOD

Rejection

Min recommended pressure

Pt

Fig. 8.4.2-3a Relation between rejection, TOD of permeate and concentration factor

Fig. 8.4.2-3b Relation between rejection and transmembrane pressure

0

1

2

3

4

5

6

0 2000 4000 6000 8000 10000 12000 14000 16000

Time (sec)

TOD

of p

erm

eate

(mg/

l)

Microemulsion Macroemulsion Used macroemulsion

Fig. 8.4.2-2 Examples of permeate TOD evolution from RO of the UF permeates of various emulsions (the RO permeate are not recycled) [11]

354


III-238

6. Guidelines for RO design or evaluation, summarized from various GPI’s lab, are as shown in table 8.4.2-1. Membrane material used is polyamide. MWCO is around 100-150 Daltons.

8.5 Nanofiltration (NF)

Nanofiltration is an intermediate membrane process between ultrafiltration and reverse osmosis. It is a relatively new technology, compared to others membrane processes. NF membrane pore sizes are between those of UF and RO. It can retain the meterials of the size around 100 – 600 nm. Thus it can be used to separate dissociated form of a compound from the undissociated form [38]. For example, lactic, citric, and acetic acids can pass through to NF at low pH but are rejected at high pH when in their salt forms. NF is generally used in many industries, such as biotechnology, food, and drinking water as well as in environmental applications. Apparently, structure and module of NF are relatively identical to those of RO.


Separation in NF membrane is based on both size difference, like UF, and diffusion mechanism, like RO. Generally, NF membranes are charged so they can be used for selective separation of charged materials. Uncharged membranes are available but very rare and not popular. Separation mechanisms in NF membrane are not well understood. But they are generally described in 2 approaches, i.e. ionic exclusion and, like RO, solution-diffusion.

For ionic exclusion approach, separation of charged materials by NF depends on the charges of ions in solution to be separated and of membranes. Ions of the same charge as the membrane will be pushed while the counter-charged ion will be attracted by the membrane charges. For uncharged membrane, the modified form of solution-diffusion model, as shown in eq. 8.4.2-2, was used to explain the separation mechanisms in the NF membranes by some researchers. Applications of NF on oily wastewater treatment are known of but they are privately designed and there are not many publications on the issue.

In GPI lab, MATAMOROS [20] performed the feasibility study on application of NF for cutting oil emulsion treatment. The result will be summarized in the next section.

8.5.2 Significant findings on NF for oily wastewater treatment from GPI’s researches

MATAMOROS [20] had studied the performance of NF on the treatment of cutting oil emulsions as well as permeates from UF of cutting oil emulsions, using cellulose membranes, i.e. Desal 5 (600 Da), SV10 (100-300 Da) and SG15 (2000 Da). Significant findings of the research are as summarized below,

1. From the research, NF is proven to be an efficient process for treatment of UF permeate of macro-or microemulsion, which contains very high concentrations of surfactants/co-surfactants that are not retained by the UF membrane. Rejection is slightly lower than that of RO. But the energy consumption is about half of RO’s, since its pressure range is around 4 –20 bars, compared to the range of 20 – 60 bars for RO. Evolution of flux with time and with pressure for NF of permeate are relatively identical to that of RO (fig. 8.4.2-1 and 8.4.2-2).

355


III-239

Tabl

e 8.

4.2-

1 Su

mm

ary

of R

O d

ata

on o

ily w

aste

wat

er tr

eatm

ent

356


III-240

2. NF performances on macro- and microemulsion treatment were also studied. The result showed that, at constant feed concentration, relation between pressure and flux of NF for macro- and microemulsion treatment are divided into 2 regions i.e. pressure controlled and mass transfer controlled region (fig 8.5.2-1), like that of UF.

3. Relation of flux and theoretical concentration of feed, when permeate is not recycled, is as shown in fig. 8.5.2-2. The relation is not linear so it is not governed by film model (section 8.2.1.5). This can be explained by destabilization in-situ of emulsified oil during NF operation, justified from presence of oil layer in the storage tank. Destabilization of oil, like the case of UF with salt addition, makes the real feed concentration lower than theoretical value.

4. Results of MATAMOROS are as shown in table 8.5.2-1. From the table, TOD removal efficiencies of UF are always higher than UF. This means the oil is completely separated and the surfactants/co-surfactants are retained with higher efficiency. However, it should be noted that the results are based on fresh emulsion. For used emulsion, result may differ and should be obtained by pilot test.

Fig. 8.5.2-1 Relation of flux and transmembrane pressure of NF for macroemulsion (Elf Seraft ABS) and microemulsion treatment (Elf Emulself G3 EAB), using desal5

membrane at 20oC [20]

357


III-241

Fig. 8.5.2-2 Relation of flux and theoretical feed concentration of NF for macro- and microemulsion treatment [20]

8.6 Comparison of membrane processes on emulsion treatment

Major objective of GPI’s researches on oily wastewater treatment by membrane process is the treatment of stabilized emulsion. From the previous sections, they show that performance of each process or combinations of processes on emulsion treatment are different. To compare the performance of processes or combinations of membrane processes, MATAMOROS had performed pilot test of various processes using the same macroemulsion. His results are tabulated in table 8.6-1. However, It should be noted that,

1. The results are based on specific operating conditions, types of wastewater and membrane. Since performances of membrane processes are sensitive these parameters, the results can be used as a guideline for preliminary evaluation only.

2. Fresh emulsions were use so fouling due to foreign materials was not included.

3. Evolutions of flux with time are not present. In real design, this data is important for membrane sizing and determination of operating and cleaning cycle. For example, NF and MF+CaCl2 may be very interesting from their relative low energy consumption. But they may foul easily, resulting in low flux for long-tern operation and more frequent cleaning.

4. Even though the efficiency of each process is very high. It does not mean that the permeate always conforms to the related standard. For example, permeate’s TOD from RO of UF+RO of macroemulsion at 4% by volume of oil is around 1,250 mg/l, which is still high and may need further treatment or to be mixed with relatively diluted wastewater and treat by conventional treatment system.

5. From the previous sections, it is obvious that membrane process is very versatile and efficient process. But its performance is sensitive to many parameters. So it is, again, strongly recommenced to perform pilot test before design the real membrane process.

358


III-242

Tabl

e 8.

5.2-

1 Su

mm

ary

of N

F da

ta o

n oi

ly w

aste

wat

er tr

eatm

ent

359


III-243

Table 8.6-1 Comparison of membrane processes on cutting macroemulsion treatments (based on Elf Seraft ABS at 4% by V of oil)

MF+CaCl2 UF UF+RO UF+NF NF

Membrane MS11107 Iris3042 Iris3042 +MS10 Iris3042 +Desal5

Desal5

Global TOD removal efficiency (%)

97.2 95.8 99.1 97.9 97.4

TOD of final permeate (g/l)

2.76 4.5 0.92 2.2 2.7

Pt (bar) 0.1 1 1+30 1+10 10

Flux (LMH) 24.9 52.92 52.92/42.12 52.92/30.96 27

Specific energy consumption (KWh/m3)

Not available 3.2 3.2+40.2 3.2+22.8 22.2

Remark See note 1 See note 2

Note 1 Initial flux of MF is very high, compared to that of water, but the membrane is rapidly clogged by free-oil. The flux, thus, drops sharply. The final flux is relatively low.

2 Evolution of time was not present in the original research. But the membrane tends to be clogged by free-oil from destabilization in-situ.

3 No comparison data of microemulsion treatment was proposed.

360


III-244

[This page is intentionally left blank.]

361


III-245

Chapter 9 Thermal processes

9.1 General

Thermal processes are separation processes that involve changing of phases of the materials to be separated. For distillation, it involves liquid and gas. For crystallization and zone refining, they involve liquid and solid. Since the processes involve phase changing, they inevitably consume high energy. This is the reason why they are scarcely used for general wastewater treatment. However, it may become economical alternative if the product from the treatment can be recycled or relative easier for ultimate disposal.

Oily wastewater is actually binary system, containing, mainly, 2 immiscible liquids i.e. oil and water. So it could be separated by thermal processes. In GPI lab, there are some researches on applications of thermal processes on oily wastewater treatment. Most of all were based on distillation. Only few were based on crystallization. So, in this chapter will be emphasized on distillation, which can be divided into 2 major types, i.e. conventional distillation and enhanced distillation, called “heteroazeotropic” distillation.

It must be noted that the researches are based on lab-scale experiments to study the theoretical concepts that underline process operation. In real processes, the distillations are usually carried out in distillation columns, which their designs are a major science into itself. So it will not be mentioned here. However, the researches can be used as a guideline to understand and evaluate the feasibility of the processes.

9.2 Basic knowledge on distillation

Since distillation always involves liquid and vapor phase of mixtures, it can be described by the concept of vapor/liquid equilibrium. The mixtures may have two or more components. However, to provide basic knowledge of the process, it is sufficient to simply consider the mixture of 2 components, called “binary” mixture.

9.2.1 Basic knowledge on vapor/liquid equilibrium of mixtures

Since distillation always involve the mixture of two or more components, it is necessary to know a certain number of variables to describe or characterize the equilibrium stage of the system. The number of required variables, called “degree of freedom” is calculated by the phase rules, as shown in eq. 8.2.1.

NF +−= π2 {9.2.1}

Where F = Degree of freedom π = The number of phases, for vapor/ liquid system, π = 2 N = The number of species in the mixtures

For example, to describe a binary mixture system, required degrees of freedom are 2-2+2 = 2, which are normally pressure and temperature. Generally, distillation processes operate at constant pressure, so operating parameter that is used to control the process is temperature.

To understand vapor/liquid equilibrium of a binary system, consider a binary mixture of specie A and B in the container, as shown in fig. 9.2.1-1. Assume that overall pressure of the system is constant at 1 atm. When the mixture is heated, temperature of the mixture will

362


III-246

increase, as shown in fig. 9.2.1-2. When it reaches a certain temperature, called “bubble point”, the first bubble of vapor will appear. From this point, the mixture will exist in both vapor and liquid phases. Unlike a pure substance, temperature of the system during evaporation period continuously changes, rather than being constant. When the temperature reach a certain value, called “dew point”, the last drop of liquid will disappear.

Liquid phase

Vapor phase

P = constant

HeatSpecie BSpecie A

Time

T

Dew point

Liquid Liq Gas Gas

Bubble point

P = const.

Fig. 9.2.1-1 Diagram of a binary mixture system

Fig. 9.2.1-2 Evolution of temperature of a binary mixture at a constant P

Dew temperature and bubble temperature of a binary mixture varies with pressure. An example of relation between pressure and temperature is shown in fig. 9.2.1-3. Characteristic of P-T curve depends on types of components in the mixture.

Critical locus

T

P

C

C

Bubble curveDew curve

Methanol/Benzene mixture

T

P

0% Methanol

100% Methanol

Fig. 9.2.1-3 Examples of P-T diagram

At a binary system at constant pressure, the dew- and bubble points of the system vary with composition of the mixture. Compositions of the mixture are usually reported in the from of molar fraction “x” and “y”. The value “xi” represents the ratio of moles of species i in liquid phase (ni,l) to summation of moles of every species in liquid phase (nl) (eq. 9.2.2). On the other hand, the value of “yi” is the ratio of moles of species i in vapor phase (ni,v) to summation of moles of every species in vapor phase (nv) (eq. 9.2.3).

l

lii n

nx ,= , ∑ =

iix 1 {9.2.2}

v

vii n

ny ,= , ∑ =

iiy 1 {9.2.3}

363


III-247

When the bubble points at various values of x1 are plotted, the curve is called the “bubble curve” or “saturated liquid line”. On the other hand, when the dew points at various values of yi are plotted, the curve is called the “dew curve” or “saturated vapor line”. Generally, dew curve and bubble curves are plotted on the same coordinate. They can be plotted both on T-x-y scale and P-x-y scale as shown in fig. 9.2.1-4a and b. Since distillation process usually operates at a constant P, T-x-y curve is normally used. Relations between pressure, temperature, x and y can be combined into a three-dimension coordinate, resulting in P-T-x-y diagram as shown in fig. 9.2.1-5.

xA = 1yA = 1

T P = const.

Pure B Pure AxA = 0yA = 0

xA , yA

C

C C


xA = 1yA = 1

P T = const.


xA , yA

C

C

C


a) T-x-y diagram b) P-x-y diagram

Fig .9.2.1-4 Examples of T-x-y and P-x-y diagrams

To understand phase changing taking place in distillation process, we will consider a P-x-y diagram as shown in fig. 9.2.1-6. From the figure, the points on the dew curve represent saturated vapor. the area above the dew curve (T-yi) is of superheated vapor. The points on the bubble curve represent saturated liquid. The area below the bubble curve (T-xi) is of subcooled liquid. The area between the 2 curves is the two-phase region. The two curves meet at the edges of the diagram where saturated liquid and saturated vapor of the pure species coexist.

P

T

1

0

yA

xA

Critical

Pure

A

Pure B

xA = 1yA = 1

T

P = const.


yA = 0.4

= 0.6

Superheated vapor

= 0.8 xA xA

a

bb’

c’ c

yA = 0.6

Subcooled liquidBubble curve

Dew curve

Fig. 9.2.1-5 An example of P-T-x-y diagram Fig. 9.2.1-6 An example of T-x-y diagram

364


III-248

From fig. 9.2.1-6, if we heat the subcooled liquid of 60 mol-% of specie A and 40 mol-% of specie B (so xA = 0.6, xB = 1-0.6 = 0.4) under a constant P from point a, the first bubble will appear when the temperature reaches the bubble curve at point b. The molar fraction of specie A in this very first bubble will be determined by drawing the horizontal line from point b to meet the dew curve at point b’. If we continue heating, more vapors will appear while the quantity of liquid will decrease. The compositions of liquid will follow the bubble curve from point b (xA = 0.6) to c (xA ≈ 0.8) and the composition of vapor will follow the dew curve from b’ (yA ≈ 0.8) to c’ (yA = 0.6), respectively. When the temperature reaches point c, the last drop of liquid disappears. The mixture will become entirely vapor with yA = 0.6. Continuing heating, the vapor will have higher temperature and become superheated vapor. Cooling of vapor can be explained by the same manner described above.

9.2.2 Equilibrium of various mixtures

Characteristics of vapor/liquid equilibrium curves depend on types of components of the mixture, pressure and temperature. Examples of various types of vapor/liquid equilibrium are as shown in fig. 9.2.2-1. The first 3 curves are of miscible mixtures. For some mixtures, the T-x-y diagrams present a maximum or minimum point where the boiling liquid at this point produces a vapor of exactly the same composition. This point is called the “azeotrope”.

At some conditions, liquid/liquid equilibrium (LLE) coexists with vapor/liquid equilibrium (VLE), this give rise to vapor/liquid/liquid equilibrium (VLLE) [61], as shown in fig. 9.2.1-1 d to f. The state that VLLE is present is called “heterogeneous azeotropic” or “heteroazeotropic”. As stated before, VLLE varies with intensive properties (P,T) and may become only VLE at certain conditions as shown in fig. 9.2.2-2.

xA = 1yA = 1

T


xA , yA

Locus of azeotropeHigh P

Low P

Fig. 9.2.2-2 Examples of evolution of VLE and VLLE with pressure

365


III-249

Example

Methanol-Water

Ethanol-Water

Acetone-Chloroform

N-butanol-Water

Decane-Water

Polypropylene oxide-Water

Azeo

Azeo

Azeo

Azeo

Azeo

Azeo

AzeoAzeo

V

V

L+V

L

L+V

L

V

L+V

L

V

L+V

L

2 L

L

V

L+V

2 L

L+V

2 L L

L

L+V

V

T Y

XX , Y

T Y

XX , Y

T Y

XX , Y

T Y

XX , Y

T Y

XX , Y

T Y

XX , Y

Type

Without azeotrope

Minimum-temperature azeotrope

Maximum-temperature azeotrope

Heteroazeotrope (Partially miscible)

Demixtion

Heteroazeotrope (Immiscible)

Fig. 9.2.2-1 Examples of various types of vapor/liquid equilibrium

366


III-250

9.3 Heteroazeotropic distillation of oily wastewater

Heteroazeotopic distillation, in this chapter, is technically referred to the distillation process that a proper substance, called entrainer, is added to the feed to ensure the formation of heteroazeotrope. It does not include the process that the azeotrope occurs naturally. However, their working principles are identical. Details of the heterotropic distillation process are described below.


Since hydrocarbons or oils have very low water-solubility, it is generally acceptable to assume that oily wastewater, which is actually the oil/water mixture, is immiscible binary mixture. Typical characteristic of T-x-y diagram at a constant pressure (called “isobar diagram”) of oily wastewater is as shown in fig. 9.3.1-1. Generally, hydrocarbons have higher boiling point than that of water.

A

xw = 1yw = 1

T P = const.

Pure Oil Pure water (W)

xw = 0yw = 0

Oil +V

xw , yw

Boiling Tof B

Vapor (V)

W + V

Oil+water

xw = xH yw = yH

Boiling Tof Water

THB C

D

xw,1

xw,2

xw,3 xw,4=1

xw,5=1xw,6=1

yw,1’ to yw,4 = yH

yw,5

yw,6

xw,1’

Azeotrope (H)

Fig. 9.3.1-1 Typical isobar diagram oily wastewater

From the diagram, it shows that the wastewater can be boiled at the temperature lower than that of pure water and pure oils. Heteroazeotropic distillation, which is the distillation when the VLLE is present, makes use of this fact so it requires lower energy than classical distillation.

From the figure, the bubble curve in this case will look like a figure “U” with square legs (Line ABHCD). If the original wastewater contain 70 mol-% of water (xw,1 = 0.70), it will boil at the temperature TH. The composition of this very first bubble can be determined by drawing the horizontal line to meet the dew curve (Curve AHD). In this case, they will intersect at point H. Thus the molar fraction of water (yw) in the bubble is that of azeotrope (yH, xH). Continuing heating, the liquid volume gradually decreases and its composition gradually change from xw,1 to x.w,2, xw,3, etc. However, by drawing the horizontal lines from these values of xw, the compositions of yw are always equal to yH. During this period the temperature is automatically constant at TH. When xw finally reaches 1, this means the residue becomes pure water. All of oils are separated from water and in vapor phase.

After this, the temperature rises until it reaches the boiling point of pure water. The value of xw is constant at 1. The value of yw increases from yH to yw,4, yw,5, etc. since the pure water continues boiling at add more steam to the overall vapor phase.

367


III-251

The vapor phase is usually condensed to recover oil vapor. If the vapor is condensed while xw < 1, the composition of vapor will be constant at yH. From fig. 9.3.1-1,these vapors will condense at the temperature TH, resulting in the distillate of oil and water at xw = yH.

From GPI’s research by TOULGOAT [19], the distillate will not be formed as an emulsion if the hydrocarbons have high vapor tension and/or their water solubility is very low and not sensitive to temperature change. This condition is usually satisfied for oily wastewater treatment polluted by petroleum-base oil or for the carefully selected entrainer (described in section 9.3.4). Thus the distillate from heteroazeotropic distillation of oily wastewater is usually in the form of 2 separate layers of water and oil.

9.3.2 Raoult’s law and Dalton’s law

To define dew curve and bubble curve of a binary mixture, the following equations are generally used, i.e. eq. 9.3.1 and eq. 9.3.2 (Dalton’s law)

satiii Pxp = {9.3.1}

Pyp ii = {9.3.2}

Where pi = Partial pressure of specie i PP

sat = Vapor pressure of specie i P = Pressure of overall system, which is normally kept constant, e.g. 1 bar

From eq. 9.3.1 and 9.3.2, it gives rise to Raoult’s law (eq. 9.3.3). The equation is valid under the assumption that the vapor phase is an ideal gas and the liquid phase is an ideal liquid.

satiii PxPy = {9.3.3}

If the system is not ideal system, eq. 9.3.3 will be modified as shown in eq. 9.3.4 (modified Raoult’s law).

satiii

satii fxPy γφ = {9.3.4}

γi is called “activity coefficient”, which is normally obtained from experiments. If γ = 1, this means the system is ideal. The equation will become Raoult’s law. The vapor pressure is replaced by the term “fugacity” (f) and “fugacity coeficienct“ (φ). Relation between Psat, f and φ are as shown in eq. 9.3.5.

sati

satisat

i Pf

=φ {9.3.5}

For application of oily wastewater treatment, the use of eq. 9.3.3 is proven to be accurate enough [24], [11]. These equations will be used to determine the dew curve and bubble curve, as well as azeotropic point of oily/water mixture, as described in the next section.

368


III-252

9.3.3 Calculation of azeotropic temperature and composition, dew curve and bubble curve.

To calculate the isobar diagram of VLLE of oily wastewater or oil/water mixture, the following procedure, based on the theories described in the previous section, is recommended.

1. Find the azeotropic temperature (TH) and bubble curve (T-x)

For immisible bunary mixture of A and B, it can be imagined that the two liquids are separately located in the container. Each liquid exerts the corresponding vapor pressure. Thus the vapor pressure of the system is the summation of the vapor pressures of every species, as shown in eq. 9.3.6. On the bubble curve, the mixture will boil when the vapor pressure of the mixture is equal to the overall pressure of the system (P) (eq. 9.3.6).

PPPP satB

satA

satBA =+=+ {9.3.6}

To find the temperature TH that corresponds to eq. 9.3.6. Relation between vapor pressure and temperature of both liquids must be known. For general liquids, their properties can be found in many references, e.q. Perry’s [2]. If the relations are provided in the form of data table, it may be more convenient to use graphical method to find TH, as shown in fig. 9.3.3-1. Curves of liquid A and B are drawn, using the data from the reference. Curve of the mixture A+B is obtained by the sum of vapor pressure of A and B at the same temperature. This curve is the relation between azeotropic temperature (TH) and pressure. Thus TH at any given operating pressure of distillation process can always be found.

TH Temperature

Pressure

Pure A

Pure B

A+B

Pdesign

Fig. 9.3.3-1 Graphical method to find heteroazeotropic temperature

Sometimes, Relation between Psat and T of pure liquid is given in the form of equation, called “Antoine equation” (eq. 9.3.7).

CTBAP Tsat

−−=)ln( , {9.3.7}

T is temperature. A, B and C are numerical constants, depending on the types of liquids. These constants can be also found in many references. From Antoine equations of the two liquids, eq. 9.3.6 can be rewritten in the form of T. After solving the equation at a given P, the resulting T is the azeotropic temperature TH. The bubble curve of an immiscible binary mixture can de drawn as a straight horizontal line at the temperature TH on the T-x,y diagram.

369


III-253

2. Find the Dew curve (T-y)

At every point on the dew curve, the liquid A and B will condense so, Tsat

AA Pp ,= {9.3.9} Tsat

BB Pp ,= {9.3.11}

From eq. 9.3.2, the values of yA at any T (yA,T) in the region A-V (fig. 9.3.1-1), which A will condense, can be calculated from the following equation.

PPyPPyp

TsatAT

ATsat

ATA

TA

,, =→== {9.3.12}

In the same manner, the values of yA at any T (yA,T) in the region B-V (fig. 9.3.1-1), which B will condense, can be calculated from the following equation

PPy

PPyPPyp

TsatBT

A

TsatBT

BTsat

BTB

TB

,,, 1−=→=→== {9.3.13}

By varying the values of T, yA at various value of T can be calculated. Plotting the values of yA and T on T-x-y diagram result in a dew curve. The curve will be calculated easily by graphical method as shown in fig. 9.3.3-2. Generally, specie A represents water. So the value of x and y represent the molar fraction of water in liquid and vapor phase, respectively.

T1 T2

Psat,T1B

Temperature

Pressure

A (Pure water)

B (Pure oil)

Water+oil

Psat,T1A

Psat,T2A

Psat,T2B

Pure water (A)

Temperature

Pure oil (B) x,y yH

TH

T1

T2

yB 1 yB 2 yA 2

yA 1

Calculate yA and yB by eq.9.3.12 and 9.3.13, then, plotT,yA and T,yB to obtain dewcurves

Fig. 9.3.3-2 Graphical method to find dew curve

3. Find heteroazeotropic composition (xH, yH)

From eq. 9.3.2, azeotropic composition (yH) in the form of molar fraction of “A” can be calculated from the following equation,

Pp

y AH = {9.3.8}

At TH, liquid A (and B) will condense so, HTsat

AA Pp ,= {9.3.9}

370


III-254

Thus,

HH

HH

TsatB

TsatA

TsatA

TsatA

H PPP

PP

y ,,

,,

+== {9.3.10}

This means that, at azeotropic point, liquid phase will contain (100yH) mol-% of specie A and 100(1-yH) mol-% of specie B. For the vapor phase, it will also contain (100yH) mol-% of specie A and 100(1-yH) mol-% of specie B. Heteroazeotropic temperature (TH) and composition (yH) of common hydrocarbons, calculated from theoretical equation described above, are tabulated in table 9.3.3-1. Observed values of yH from LUCENA [24] are also presented in the table. It shows that the theoretical equations can be used to predict the characteristic of real process with high accuracy.

Table 9.3.3-1 Heterotropic temperature and composition from various hydrocarbons [24]

Extractant Molecular weight (g/mol)

TH (deg. C)

yH (by molar)

yH (by volume)

y H observed (by volume) [24]


9.3.4 Application of heteroazeotropic distillation on treatment of inverse emulsion or concentrated oily wastewater

In spite of its lower energy consumption than the classical distillation, heteroazeotropic distillation still consumes relatively high energy. So it may not be economical to treat general wastewater by this process. However, for the wastewater containing high concentration of oil or inverse (water in oil) emulsion (e.g. slops from refineries), the portion of water to be separated is relatively small and the residue from distillation process, which is water-free oil, become valuable for it can be re-processed or recycled. In these cases, heteroazeotropic distillation may become economically feasible.

GPI had studied applications of heteroazeotropic distillation on treatment of slop [24] and retentate from UF of cutting oil emulsion [11]. Significant findings from the researches are summarized as follows,

1. Addition of entrainer (or extractant)

Even though slop or retentate is naturally oil-water mixture, it may or may not contain a component that forms an azeotrope. Sometimes, in self-entraining mode, obtained

371


III-255

distillate is still in the form of emulsion. So to ensure the formation of azeotrope and readily separated distillate, specific hydrocarbons, called entrainer or extractant, will be added to the wastewater. So the heteroazeotropic distilltation, in this chapater, means the distillation process that entrainer is added to promote the formation of azeotrope. This entrainer will extract the water from the wastewater when the temperature of the system reaches the azeotropic temterature (TH). The vapor, evaporating at TH, will contain (100yH) mol-% of water and 100(1-yH) mol-% of entrainer. Performance of entrainer depends on its value of yH. The higher the value of yH, the better the performance of entrainer. Since it means that only small amount of entrainer is required to separate relatively large amount of water from the waste. For properly selected entrainer, the distillate is condensed to form two separate layers of water and entrainer, which can be reused.

2. Characteristic of the process

Experiments in the lab were based on simple distillation apparatus, as shown in fig. 9.3.4-1. Operating pressure of the process is ambience. Typical evolution of temperature during the heteroazeotropic distillation process is as shown in fig. 9.3.4-2 [11].

Distillate

Heat exchangerTI

Feed container

Heater

Thermometer

Fig. 9.3.4-1 Examples of lab-scale apparatus for heteroazeotropic distillation [11], [24]

0

20

40

60

80

100

120

140

160

180

200

0 10 20 30 40 50 60

Time (minute)

Tem

pera

ture

(o C)

0

15

30

45

60

75

90

105

120

135

150

Acc

umal

ated

dia

tilla

te v

olum

e (m

l)

Temperature Distillate volume

Heteroazeotropic distillation

Classical distillation

Fig. 9.3.4-2 Example of evolution of temperature with time from the treatment of UF permeate of cutting oil emulsion, using decane as entrainer [11]

372


III-256

From the figure, the temerature will rise from ambience to azeotropic temperature (TH), and then remain constant until the water is totally separate. After that, the temperature will rise up to the boiling point of the entrainer. From the example, the entrainer is decane, theoretical TH (table 9.3.3-1) is 97.6oC, relatively identical to the observed value. Then the temperature rises to 174oC, which is corresponding to the boiling point of decane. The evolution of temperature is natural, without controlling of any kind. It indicates that the process is relative easy to operate.

3. Performance of the process

Condensation of the vapor phase results in the distillate consisting of 2 separate layers of entrainer (upper layer) and water (lower layer). The residue is dehydrated hydrocarbons, which is very fluid, compared to mayonnaise-like, viscous slop or UF retentate. The pictures of feed, residue and distillate of the slop and UF retentate are as shown in fig. 9.3.4-3.

a) b) c) d) e) Slop (a), distillate and residue (b), magnified pictures of slop (c)

and residue (d) UF retentate (30% vol of oil),

residue and distillate (e)

Fig. 9.3.4-3 Pictures of feed, residue and retentate of slop and UF retentate of used macroemulsion (30% by volume of oil) [24], [11]

The entrainer can be simply decanted and then reused for the next distillation cycle. For the water, even though its appearance is transparent and seems clean, it may contain some soluble pollutants, such as some volatile chemicals or surfactants, depending on characteristic of the feed. WANICHKUL [11] reported that TOD of the water from UF retentate of cutting oil emulsion is around 1,000 – 2,800 mg/l. In this case, it needs to be further treated.

In case of slops, the residue can be sent back to refinery process for reproduction. For UF retentate, the residue, which is mainly base oil from cutting oil emulsion, has high calorific value so it can be reused. Generally, mechanical workshops or cutting oil users will send their UF retentates to central treatment facilities, where they can be treated more economically. Heteroazeotropic distillation can be used in such treatment facilities to treat these kinds of wastes.

4. Performance of entrainers

As stated before, the performances of entrainers depend on their capability to extract water from wastewaters, represented by their corresponding yH. The higher the yH, the

373


III-257

better the extracting performance. LUCENA had proven that the value of theoretical yH is close to the observed value, as shown in table 9.3.3-1. Apart from pure hydrocarbons, he also tested some general commercial hydrocarbons. The results are tabulated in table 9.3.4-1. From the result, kerosene and gasoline are suitable to use as entrainers.

Table 9.3.4-1 Water extracting performance of various commercial hydrocarbons [24]

Name Ratio of water in the distillate or yH (by volume) Remark

Kerosene 0.635 Gasoline 0.75 BTX cut (or fraction) 0.134 “Charge réformat “oil 0.15, 0.23, 0.455 See note 1 “Cœur FCC” oil 0.08, 0.144, 0.343 See note 1

Note 1. These oils contain several fractions of hydrocarbons so they start extracting water with the lowest-yH hydrocarbon. When it is used up, the next higher-yH hydrocarbons are used.

9.3.5 Application of heteroazeotropic distillation on treatment of the wastes polluted by trace hydrocarbons: Steam stripping

Steam stripping is another, in effect the reverse, form of heteroazeotropic distillation that the small amount of relatively volatile materials, such as volatile hydrocarbons, hydrogen sulfide or ammonia, are removed from large amount of less volatile material or wastewater. In this case, water will be used as an entrainer to extract the pollutants. For wastewater, even though it contains water and has self-entraining property, steam is normally used to heat the wastewater and reduce the effective pressure in the distillation apparatus to save the energy required. Stripping with inert gas is also available.

Stripping reactor is generally a packed column to provide efficient vapor-liquid contact. The waste is fed at the top of the column while the steam enters at the bottom. The quantity of steam required to remove pollutants depends on the types of the pollutants. The vapor of steam and pollutants flows out at the top of the column to further process, depending on the type of pollutants. Condensate of oil and water is sent to oil separation process for recovery of oil. Hydrogen sulfide or ammonia may be sent to flare of furnace or disposed off, if possible. It should be noted that steam stripping is based on different concept from air stripping, which is based on solubilty of gas governed by Henry’s law.

9.3.6 Design calculation and design considerations

9.3.6.1 Quantity of entrainer required

As stated before that the column design will not be discussed here, the major calculation of the process is quantity of entrainer required to separate the pollutants. It can be divided into 2 cases as follows,

374


III-258

1. Heteroazeotropic distillation of concentrated oily wastewaters or slops

In this case, the entrainer is selected hydrocarbons and the pollutant is water. Theoretical quantity of entrainer can be calculated from yH of the entrainer and quantity of water to be removed by the following equation.

H

Hwaterentrainer y

yVolumeFSVolume

)1(.

−⋅= {9.3.11}

The yH of various entrainers are listed in table 9.3.3-1 and 9.3.4-1. If other entrainer will be used, its yH can be determined from experiment or calculated by the procedure shown in section 9.3.3. If the theoretical value of yH is used, S.F of 1.05 to 1.2 is recommended [24].

2. Steam stripping

In this case, the entrainer is steam and the pollutant is volatile material in the wastes. Theoretical quantity of steam can be calculated from quantity of pollutants to be removed and their corresponding yH by the following equation.

)1(. tan

H

Htspollusteam y

yVolumeFSVolume−

⋅= {9.3.12}

Please note that the calculated quantity of steam is the quantity required for heteroazeotropic distillation process only. Additional heat (sometimes, also in the form of steam) may require to raise the temperature of the system up to the design point.

9.3.6.2 Design considerations

From GPI’s researches, design considerations can be summarized as follows,

1. In case of treatment of slop or concentrated wastewater, quantity of water should be determined before addition of entrainer. Excess dosage of entrainer results in more energy consumption. Since the temperature required to recover this excess entrainer will be equal to the boiling point of the entrainer, not the azeotropic temperature (see fig.9.3.4-2). Otherwise, this excess part will be wasted with the residue and not be recycled.

2. If possible, it is recommended to perform lab-scale tests or pilot-test to evaluate the type of entrainer, its required quantity and the efficiency of the process, such as TOD of distilled condensate, before design the real process. Since some unknown factors, e.g. presence of surfactants, volatile substances, etc., may affect the performance of the process.

3. For steam stripping, the components of pollutants, sometimes, may be unknown. Thus it is strongly recommended to perform lab-scale tests or pilot tests. To evaluate the feasibility of the system and the quantity of steam required.

4. The process operates at high pressure. Thus presence of some chemicals, such as hydrogen sulfide, may give rise to high-corrosive environment. So these precautions need to be taken into account in the distillation reactor design.

375


III-259

5. The distillation system is generally costly and consumes high energy. Economic analysis of the system should be performed, compared to other competitive processes, e.g. stripping VS. chemical treatment or solvent extraction or adsorption.

9.4 Classical or conventional distillation of oily wastewater


Classical or conventional distillation in this chapter is referred to the distillation process that no entrainer is added to the feed to force the formation of azeotrope. Working principle of classical conventional distillation is similar to that of the heteroazeotropic except that there is no addition of entrainer to the feed. However it may contain azeotrope in case that there are some components in the feed that can act as entrainer.

If the azeotrope is not naturally formed, the boiling point in this case will not be lowered by azeotropism. So, more energy is required to rise the temperature up to operating value. In case of oily wastewater, the temperature is around than 100oC, which is the boiling point of the water at 1 atm. Since it consumes high energy, its application on wastewater treatment is limited. Distillation under reduced pressure is reported to be used for emulsion treatment in Germany. In GPI lab, WANICHKUL [11] had studied the application of distillation on oily wastewater, as will be described in the next section.

9.4.2 Significant findings on classical distillation for oily wastewater treatment from GPI’s researches

WANICHKUL [11] had divided his research into 2 parts, i.e. (1) distillation of stabilized emulsion and (2) distillation of permeate from UF of stabilized emulsion. Significant findings from the research are as summarized below,

9.4.2.1 Distillation of stabilized emulsion

WANICHKUL [11] had studied the performance of distillation on stabilized emulsions, using both macro- and microemulsion. The result shows that,

Distillation of cutting oil macroemulsion

1. The evolution of temperature of macroemulsion (based on Elf Seraft ABS 4% by volume of concentrate) is as shown in fig. 9.4.2-1. From the figure, the temperature rises from the initial value to 92oC and remains constant throughout the experiment. This can be explained that the process contains a self-induced azeoptrope since the boiling point of the mixture is lower than the boiling point of water (100oC).

376


III-260

0

50

100

150

200

250

300

0 20 40 60 80 100 120 140 160

Time (minute)

Tem

pera

ture

(o C) ,

Dis

tilla

te v

olum

e (m

l).

0

3

6

9

12

15

18

TOD

of d

istil

late

(g/l)

Temperature Accumulated distillate volume TOD of water part TOD of oil+water part

Fig. 9.4.2-1 Evolution of temperature, volume of distillate, TOD of mixed distillate and TOD of water part of the distillate (Based on Elf Seraft ABS, 4% by volume of concentrate) [11]

2. Other evidence for self-induced azeotropism is that the distillate consists of two separate layers of oil and water, as shown in fig. 9.4.2-2. These two parts are readily separated. The oil part is mainly dehydrated hydrocarbons portion in the cutting oil and can be reused or recycled.

Fig. 9.4.2-2 The feed, residue and distillate from distillation of the macroemulsion [11]

3. Some pollutants are found in the water part of the distillate. Evolution of TOD in the distillate is very complex, resulting from complicate ingredient of the emulsion, as shown in fig. 9.4.2-1. From the experiment, TOD of the water part is around 2,000-7,000 mg/l. TOD removal efficiency is around 96%, based on cutting oil TOD of 120,000 mg/l.

4. The efficiency of the process is close to that of UF (98%) but the energy consumption distillation process is much higher. UF, generally, requires energy around 10-20 kWh.m-3 while energy requirement of distillation is around 625 kWh.m-3 [11]. Thus if the treatment of permeate is not concerned, UF seems to be more feasible than distillation for macroemulsion treatment.

377


III-261

5. From complexity of cutting oil formulation, x-y diagram of the mixture can not be established. But the result can be used as a guideline to evaluate the feasibility of distillation on this kind of application

Distillation of cutting oil microemulsion

1. The evolution of temperature of microemulsion (based on Elf G3 EAB 4% by volume of concentrate) is as shown in fig. 9.4.2-3. From the figure, the temperature rises from the initial value to 99oC and remains constant throughout the experiment. So the process also contains an azeotrope.

0

20

40

60

80

100

120

0 10 20 30 40 50 60 70 80 90

Time (minute)

Tem

pera

ture

(o C) ,

Dis

tilla

te v

olum

e (m

l).

0

3

6

9

12

15

18

TOD

of d

istil

late

(g/l)

Temperature Accumulated distillate volume TOD of water part TOD of oil+water part

Fig. 9.4.2-3 Evolution of temperature, volume of distillate , TOD of mixed distillate and TOD of water part of the distillate (Based on Elf G3 EAB, 4% by volume of concentrate) [11]

2. The distillate also consists of the layers of oil and water, which are readily separated (fig. 9.4.2-4). Evolution of TOD is complex, like the case of the macroemulsion. TOD of the water part of the distillate varies from 1,200 to 2,200 mg/l. TOD removal efficiency is around 98%, based on cutting oil TOD of 75,000 mg/l. The efficiency is higher that that of UF (approx. 85%)

Fig. 9.4.2-4 The feed, residue and distillate from distillation of the microemulsion [11]

3. Energy consumption for microemulsion is relatively similar to that of macroemulsion (approx. 625 kWh/m-3) . So it consumes much more energy than UF ( < 20 kWh/m-3) [11].

378


III-262

Distillation of permeate from UF of macroemulsion

1. The permeate from UF of used macroemulsion was used. Inlet TOD is about 4,400 mg/l. The permeate boils at 100oC so there is no azeotrope. Like the case of emulsions, the x-y diagram for distillation of the permeate can not be established, resulting from its complex ingredients.

2. The distillate is clear and contains. Evolution of TOD is shown in fig. 9.4.2-5. Final TOD of the distillate at the end of the experiment is around 100-500 mg/l. It may still exceed the discharge limit of some effluent standards. TOD removal efficiency is around 90%.

0

50

100

150

200

250

300

0 20 40 60 80 100 120 140 160

Time (minute)

Tem

pera

ture

(o C),

Dis

tilla

te v

olum

e (m

l).

0

0.2

0.4

0.6

0.8

1

1.2

TOD

of d

istil

late

(g/l)

Temperature Distillate volume Series3

Fig. 9.4.2-5 Evolution of temperature, volume of distillate, and TOD of distillate (Based on UF of used macroemulsion, TOD of feed =4400 mg/l) [11]

379


III-263

Chapter 10 Chemical treatment processes

10.1 General

Naturally, oils or hydrocarbons tend to separate from water. In their natural form, separation of oil droplets is governed by their size. Large oil droplets separate from water within a short time while small droplets take longer time to be decanted or separated. Efficiency of most of physical separation processes aforementioned decreases with the decrease of oil droplet sizes.

Furthermore, if the droplet sizes are small enough (< 5 microns), the oil droplets will subject to Brownian motion. Thus theirs rising velocity are no longer governed by STOKE’s law and can be negligible. These droplets will remain dispersing in the water without decanting. (In fact, they may probably decant but take very long time, such as years.) In this case, it can be said that the oil is stabilized. The oil/water mixture is called stabilized emulsion or stable emulsion.

In some applications, it is necessary to disperse oil into water phase and keep it in the stabilized form. Examples for these applications include some dairy product processing, washing process using detergents or the use of cutting oil emulsion. In case of machanical workshops, the stabilized mixture of oil and water, called “cutting oil emulsion” is needed to perform very important roles in lubrication of machine tools and specimens, cooling, washing away the scraps and impurities away and protecting the tools from corrosion. The oil has to be dispersed homogeneously in the water to provide consistent properties. For washing, greasy dirt on the clothes needs to be removed and suspended in wash water without coming back to the clothes again. To make the oil stable, apart from oil and water, the third components called “surface active agents” is always required.

When this stabilized emulsion is no longer suitable to use and then wasted, it will become one of oily wastewater that is most difficult to treat. Thermal and membrane processes are proven to be applicable for this kind of wastewater.

STOKE’s law-based physical separation processes (e.g. decanter, coalescer, DAF), can not be directly used in this case since the wastewater contains very small and stable droplets. To use these physical processes, the wastewater needs to be undergone necessary chemical treatment processes.

Main objectives of chemical treatment processes for stabilized emulsion treatment are,

• To destabilize the emulsion and

• To make the oil drops ready to separate from the water, e.g. to increase the size of the droplets.

GPI lab had studied chemical processes for oily wastewater treatment, which can be summarized as shown in the following sections.

380


III-264

10.2 Basic knowledge

10.2.1 Stability of the emulsion

Stability of emulsion refers to the ability of the emulsion to maintain its properties, esp. its dispersion. Stable or stabilized emulsion can, practically, maintain its oil dispersion without changing with time. So its required qualities, such as cooling or lubricating capacity, are relatively consistent. Stability of emulsion are generally based on 2 factors,

• Droplet size: The oil droplet size must be very small. So they are not decanted naturally or by STOKE’s law-based separation processes.

• Resistance to coalesce: The oil droplets must have good resistance to coalesce. So their size distribution remain relatively constant.

From section 2.2.1, the droplet size is proportional to effective work and inverse of interfacial tension (1/γo/w). Thus, to obtain stabilized emulsion, it is reasonable to decrease the interfacial tension between oil and water and prevent the interaction between oil droplets. However, low interfacial tension is not the only factor that guarantees the stability of the emulsion. AURELLE and ZHU [21] proposed the properties that give rise to stable emulsion, which will be described in section 10.2.3. To achieve these properties, the third component, called “surface-active agents” or “surfactants”, is required. Details on surfactants are described in the following section.

10.2.2 Surface-active agents

Surface-active agents, or surfactants, are practically the materials that are soluble both in oil and water. They usually localize themselves and form a layer (generally, monomolecular layer) at the surfaces or interfaces. This phenomenon is called surface activity, which give the materials their names. Chemically, they always comprise of large polar functional groups. One end of the molecules that is soluble in oils is usually a long chain of aliphatic or aromatic or both forms of organic groups with 8 to 18 carbon atoms. Being a long chain, this hydrophobic part is usually called “tail” of the surfactants. Another end, which is water soluble or hydrophilic, is usually called “head” of the surfactant. Thus the symbol of surface-active agents is usually drawn as a circle with long tail as shown in fig. 10.1.2-1.

or

Hydrophilic “head”

Hydrophobic“tail”

-

+ + -

Anionic

Cationic Amphoteric

Nonionic

-

-

-

Oil

- Micelle

--

Fig. 10.2.2-1 Symbols of surface active agent and its localization at oil/water interfaces

381


III-265

Types of surface-active agents or surfactant are categorized by charge of their head as follows,

1. Anionic surfactant is the surfactant that ionizes to yield a positive charge, free ion and a negative charge (surface active ion) which localizes at the interface. Its symbol contains a “minus” sign in the head. This type of surfactant is relatively cheap and widely used in industries. Common anionic surfactants are, i.e.,

• Soaps are usually sodium or potassium salt, derived from fats and oils by saponification (hydrolysis with presence of alkaline agent) with sodium hydroxide.

• Sulfate surfactants are salts of sulfated alcohol, such as dodecyl or lauryl alcohol. General formula of these surfactants is in the form of R-OSO3-M+, where M+ represent positive-charge ion, such as sodium or potassium.

• Sulfonate surfactants are sulfonated compound, mainly derived from esters, amides and alkylbenzenes. An example of these surfactants is sodium alkylbenzene sulfonate. General formula is R-SO3-M+.

2. Cationic surfactant is the surfactant that the surface-active part is cation. It is usually salt of quaternary ammonium hydroxide, which its hydrogen of the ammonium ion have been replaced with alkyl groups. Cationic surfactants are quite expensive. But they are noted for their disinfecting (bactericidal) property [1]. The symbol of the surfactant contains “plus” sign.

3. Non-ionic surfactant is the surfactant that does not ionize and have to depend on groups in the molecule to make it soluble [1]. The groups are usually polymers of ethylene oxide (C2H4O). The symbol of the surfactant, in this case, contains no sign. Examples of the surfactants are polyethyleneglycol mono-oleate, nonylphynol ethoxylene of ethylene oxides. The surfactants are noted for adjustable hydrophil-lipophil properties.

4. Amphoteric surfactant is the surfactant that contains both positive and negative surface-active part. Its symbol contains 2 circles, one with minus sign, another with plus sign.

In production of industrial emulsions, esp. manufacturing of cutting oil, surface-active agents are divided into 2 types, i.e.,

1. Surfactants are the main surface-active agents used to lower the oil/water interfacial tension, thus, stabilize the oil droplets. They are usually of anionic type. Thus they forms anionic stabilized emulsion. However, nonionic emulsions are also available.

2. Co-surfactants are normally nonionic surfactants such as fatty alcohol. Their localization at the surface of oil droplets among the main surfactants helps reducing the repulsive force between the ionic heads of surfactants. Thus they give rise to smaller droplet sizes and more stable emulsion.

Effect of surfactant: Surfactants lowers the oil/water interfacial tension (as well as surface tension of water, in case of air/water interface) by the mechanism described in the next section. Relation between interfacial tension and concentration of surfactant is as shown

382


III-266

in fig. 10.2.2-3. From the graph, interfacial tension decreases rapidly at the beginning. Then the rate of decrease lowers until the concentration of surfactant reached a certain value, called “critical micelle concentration” (CMC). After that, the tension remains relatively constant. This can be explained by the formation of groups of surfactant molecules called “micelle” (fig 10.2.2-1). These surfactants will no longer localize at the surface of droplets, so they have no effect on the tension.

Time

γow or

CMC

γw

Fig. 10.2.2-2 Diagram of the electrical double layer

10.2.3 Important properties to obtain stable emulsion

As mentioned in section 10.2.1, emulsion must have one or more of the following properties, proposed by AURELLE and ZHU [21], to become stable emulsion.

1. Thermodynamic stability

Normally, interfacial tension of oil is positive. To increase the stability of oil droplets, the interfacial tension should be lower to increase the area, thus decrease the diameter of the droplets. Addition of surfactant can lower the interfacial tension by its localization at the droplet surface. The surfactant will try to stretch or increase the surface of oil droplets as much as possible in order to locate itself at the surfaces. It results in virtual force (p) that tries to stretch the surface, countering to the interfacial tension (γo/w) that tries to contract the surface (see fig. 10.2.3-1). Effective interfacial tension is the difference between these two forces, as shown in eq. 10.2.1.

pp

γowγow

Surfactants, co-surfactants

Droplet surface

Water

Oil

γow = 0 γow = 0γow < 0

Coalesce Redistribute

Fig. 10.2.3-1 Interfacial of oil and water with the presence of surfactants

Fig. 10.2.3-2 Coalescence and redistribution of droplets in thermodynamiclly stabilized

emulsion

383


III-267

poriginalowow −= γγ {10.2.1}

If the surfactant concentration is high enough, it will lower the tension until it becomes zero. At thermodynamic equilibrium condition, the system of these small droplets has zero energy. When they coalesce, the sum of surface area will decrease so the equilibrium is disturbed. The energy, thus the tension, will become negative. So they will spontaneous redistribute to their original small diameters to maintain the equilibrium, as shown in fig. 10.2.3-2.

2. Dynamic stability

From thermodynamic point of view, the system that has zero energy is very stable. But, actually, some systems that have positive energy are also found very stable. So, besides thermodynamic stability, the emulsion must have dynamic stability. This stability comes from 2 equally important factors or resistance, i.e. electrical and mechanical resistances or barriers.

2.1 Electrical barrier

Electrical characteristic of charged particles (in this case, oil droplets) can be explained by the double layer theory (proposed and developed by Helmholtz, Stern, Gouy and Chapman) or the DLVO theory (proposed by Derjaguin, Landan, Verwey and Overbeek). To describe a charge particle, the two theories are relatively similar and characterize the charged particle as shown in fig. 10.2.3-3.

The original charge of oil droplet is normally negative, acquired by negative ion adsorption. From its electrical charge, the oil droplet can attract ions of the opposite charges (counter ions) to surround it. However the counter ions, in this case, are positively charged ions, which are usually surrounded by molecule of water. So they can come close to the oil droplet only at a certain distance called “Stern layer thickness” (Ω). The stern layer is the inner layer of the double layer, according to the theories. Other counter ions that locate outside the stern layer will more dense near the surface and then will thin out until their concentration are equal to that in bulk liquids. This outer layer is called the diffused layer. Outside the diffuse layer, the effect from droplet charge is negligible.

Oil w/surfactants

++

+++ +

+

+

++

- -

--

-

--

-

- ----

- -- --

--

++

+

+

++

+

+

+

+

++

++

+

+

++

+

+

-+

-+

+-

+

+

+

-

+

++

+

++

- -

Stern layer

Diffuse layer

Positive charge ion withwater solvation

Stern potentialZeta potential

---

---

-

--

+

+

+

++

++

++

+

+

+

+

+

+

+

+

+

+

+

-

+

Stern layer

Diff

use

laye

r

Plan

e of

shea

r -

-

+

Oil

w/ a

dsor

bed

surf

acta

nts a

t th

e su

rfac

e

Distance

Voltage

Fig. 10.2.3-3 Diagram of the electrical double layer

384


III-268

Electric force from the charge of the droplet can be measure by its movement when it is placed in electrical field (Electrophoretic). For the negatively charged particle, it will move toward anode. The movement can be transformed to the value of electrical voltage. This voltage is called Zeta potential (Z). The higher the value, the greater its force to repulse co-ions. So the droplets can not some close to each other.

2.2 Dynamic barriers

Some systems of low Z are found very stable too. This can be explained the presence of dynamic barriers. The film of surfactants on the surface of droplets is relatively rigid. Even the droplets collide, the film will not rapture and can prevent coalescence. To increase the rigidity of film, Non-ionic co-surfactants are added so their molecules can locate tightly among charged surfactants, resulting in a rigid film. This is the reason why the co-surfactants or muti-surfactants are used in production of cutting oil emulsion.

10.2.4 Destabilization of emulsion

From the previous sections, the properties to obtain stable emulsion are shown. So, to destabilize (or “crack” or “break”) the emulsion, those factors will be eliminated or minimize, i.e.,

• Increase of interfacial tension to eliminate thermodynamic stability

• Minimize or elimination of surfactant films around the droplets

• Reduction of charge of the droplets to eliminate or minimize electrical barriers. When the droplets can come close to each other at some certain distance. Attractive force between molecules (called “Van Der Waal force”) overcomes repulsive force from electrical charges. So the net force will be attractive (fig. 10.2.4-1) and the droplets move toward each other and have a chance to coalesce.

Oil droplets+

+

+++ +

+

+

++ - - --- -

--

-

-

-

---

- -- --

-

++

+

+

++

+

+

+

++

Oil droplets+

++

+++ +

+

+

++ - - --- -

--

-

-

-

---

- -- --

-

++

+

+

++

+

+

+

+

++Van Der Waal

Attraction(Short range)

Electrical repulsion

(Longer range)

Attractive force

Distance

Attrative force prevails

Repulsive forceRepulsive force

Resulting force

Val Der WaalForce

Fig. 10.2.4-1 Force diagram of oil droplets and relation of repulsive, attractive and resulting force with the distance between oil droplets

385


III-269

10.2.4.1 Destabilization mechanisms

To achieve above goals, one or more of the following destabilized mechanisms is needed.

1. Reduction of diffuse layer thickness

When counter-ions are added into the wastewater, these ions will be attracted by the droplet charges. Thus they will surround more tightly near the droplets and reduce the diffuse layer thickness. This effect results in reduction of Zeta potential. So the droplets can come closer to each other and have a chance to coalesce. For general oily wastewater and cutting oil wastewater, droplets have negative charges. So counter ions in these cases are positive charges, e.g. Na+, H+ in the forms of NaCl or H2SO4, etc. The counter-ions can be added until the system reaches iso-electric condition (potential = 0). However, this mechanism cannot reverse the droplet charges, no matter how many ions are added.

2. pH adjustment

If soaps are used as surfactants, such as some microemulsions, the acid added will neutralize NaOH used in saponification, thus shift the equilibrium of hydrolysis of fatty acid. So the soaps transform back to fatty acid and loss surfactant effects. The emulsion, then, is destabilized.

3. Precipitation of surfactants

Since emulsion stability is based on presence of surfactants, precipitation of surfactants can certainly destabilize the emulsion. Addition of some chemicals can react with the surfactants, resulting in complexes of no surfactant property. Bivalent or multivalent salts are used to precipitate the surfactants, such as CaCl2, MgCl2, MgSO4, alum (aluminium sulfate) or ferric chloride. Generally, the higher the valence, the better the efficiency to precipitate the surfactants and the smaller dosage of the salts required. However, precipitation efficiency also depends on the types of salts and surfactants used in the emulsion, which should be verified by jar test. Efficient salt for one surfactant may ineffective for other types of surfactants.

4. Sweep coagulation

Some metal salts can form complexes with other ions in the water, such as hydroxide. These complexes can trap the oil droplets, thus the droplets can be separated from the emulsion. These types of chemicals are normally multivalent metal salts, such as alum.

5. Adsorption and charge neutralization

Addition of surfactants that have the opposite charges to those used in the emulsion may result in destabilization of oil droplets by adsorption of the new surfactants and neutralization of the existing charges. But it must be noted that, if overdose, this mechanism will cause reversal and re-stabilization of emulsion.

6. Bridging

There are several commercial chemical products that can be used to destabilize the emulsion. From their molecule structure and properties, they may trap the oil droplets by

386


III-270

their bridging properties or adsorb the oil droplets and from scum or sludge. They may contain the salts that can destabilize the emulsion by the mechanism mention above. Examples of these chemicals are the products of DEGUSSA and Primafloc.

10.2.4.2 Coagulants and destabilization chemicals

Chemicals or materials generally used to archive the destabilization mechanisms described above include,

1. Monovalent electrolytes

Examples of this type of salt are NaCl and H2SO4. Main destabilization mechanism is reduction of diffuse layer. Thus required dosage is quite high in order to provide sufficient concentration of positive ions in the entire emulsion to destabilize the droplets. For certain types of surfactants, e.g. soaps, acid can cause destabilization be neutralization of saponification process that gives rise to the soaps. In this case, required concentration is much lower. Main disadvantage is the formation of saline or acid pollutants after destabilization. So it seems like the pollution changes from oily waste to saline or acid wastes.

2. Bivalent electrolytes

Examples of this type of chemicals are CaCl2, Ca(COOH)2 (Calcium formiate), MgSO4 and MgCl2. Main destabilization mechanism is precipitation of surfactants. Free surfactants in water will react with Ca or Mg ions and form complexes. Equilibrium between adsorbed, ionized surfactants on the droplet surfaces and free surfactants is shifted. So ionized surfactants will reverse into free surfactants thus, reduce the stability of droplets. This effect is practically governed by solubility product of the surfactants. Required dosage in this case is lower than that of monovalent ones. If inorganic electrolytes are used, they may also cause high saline waste, depending on required dosage. Organic salts, such as calcium formiate (Ca(COOH)2) may be more interesting since the residual pollutants is (COOH)2 which is organic and biodegradable.

3. Multivalent electrolytes

Examples of this type of chemical are ferric chloride (FeCl3) and alum. They are generally more effective in destabilization than the previous two chemicals. But it may not be used with some surfactants, such as certain types of soaps. Main destabilization mechanisms are combination between precipitation of surfactants as well as sweep coagulation. So the actual dosage is lower than that calculated from solubility product alone and usually lowest among the first three electrolytes

4. Surfactants of opposite charge

Examples of cationic surfactants that may be used for emulsion destabilization are N-cetylpyridinium chloride and salts of quaternary ammonium hydroxide. Main destabilization mechanism is adsorption and charge neutralization. Overdose must be avoided to prevent charge reversal and re-stabilization.

387


III-271

5. Commercial adsorbents

Oil destabilization and absorbents are commercializes. Destabilization mechanisms and efficiency vary with its components and emulsion ingredient.

10.2.4.3 Destabilization of emulsion stabilized by nonionic surfactants

The mechanisms stated above are generally based on emulsions stabilized by ionic surfactants. In case of those stabilized by nonionic surfactants, the mechanisms that involve reduction or elimination of electrical barrier are invalid since the droplets carry, practically, no charge. Main destabilization mechanism in this case must base on precipitation of the surfactants to form insoluble complexes. ZHU proposed original chemical destabilization method of the emulsion stabilized by nonionic surfactants of fatty acid based by addition of anionic surfactants (such as alkylsulfate of fatty alcohals) and cationic trivalent electrolytes (such as alum, ferric chloride).

10.2.4.4 Dosage, efficiency and residual pollution

ZHU had tested various type of chemicals to destabilize many kinds of emulsion, both macro- and microemulsion, both fresh and used conditions. Efficiency, residual pollution and examples of optimum dosage are summarized as follows,

1. The emulsions and destabilization chemicals were tested in lab scale by simple mixing and decanting.From ZHU’s criteria, chemicals are considered to be effective destabilization reagents for that emulsion when the oil is separated in the form of free oil at the surface within 1 hours. However, in effective cases, the decanting is carried out within relatively short time, less than 20-30 min. Oil removal efficiency varies from 0 (cannot destabilized) up to 40-70% or 99%, depending on types of destabilization chemical and emulsion.

2. Residual pollutant after decanting of free oil is mainly co-surfactants, esp. in the form of fatty alcohol, which is highly soluble. Concentration of this soluble pollutant depends on ingredient of emulsion treated and initial concentration of concentrate (in case of cutting oil emulsion) in the emulsion. This pollutant contains high TOD and must be further treated, as previously discussed in chapter 8 “Membrane processes”.

If salts, e.g. NaCl, etc. or acids are used to destabilization, they also cause residual pollutants in form of saline and acid. If the residual TDS concentration is high, it may not conform to effluent standard. This is the main disadvantage of destabilization by salt. Effluents from destabilization with acids, ferric chloride or alum have low pH and must be neutralized before discharge.

3. There are no universal chemicals and dosages valid for every emulsion. Types of effective chemicals, optimum dosages, and residual pollutants level must be evaluated first in lab scale before design the full-scale chemical process. However, test result from fresh emulsion can generally be applied for used emulsion. [21]. Results from ZHU’s study on destabilization of certain emulsions are shown in table 10.2.4-1.

4. Increase of temperature, generally, helps improving destabilization efficiency by [21],

388


III-272

• Increase of Brownian motion that favors collisions of droplets • Decrease of water viscosity that favors draining of water film around

droplets • Partial dehydration that helps weaken mechanical barriers of droplets

Table. 10.2.4-1 Results from ZHU’s research on destabilization of various emulsions

A) Emulsions tested

Surfactant Co-surfactants Type/ appearance %

Oil Name % Name %

% Addi-tive

% Water

A Milky anioninc macroemulsion

80 Sodium alkylbenzene sulfonate, sodium carboxylate, triethanolamine carboxylate

11 Benzylic alcohol, PEG ester.

4 5

B Milky anioninc macroemulsion

7 Soduim sulfonate, alkylphosphate

33 Butyldiglycol 4

C Transparent anioninc microemulsion

25 Mixed soaps of mono- and diethanolamine of fatty acids

38 Butyldiglycol 8 2 27.5

D Transparent anioninc microemulsion

6 Amine caboxylate and K alcanolamine

30 Butyldiglycol 12

E Milky nonioninc macroemulsion

80 PEG mono-oleate, nonylphenolethoylate (8 moles of C2H4O)

20

Note: PEG = Pltyethylene glycol

B) Results on destabilization of emulsion A at conc. of 2% by volume of concentrate, 20oC

Dosage Chemicals

g/l meq/l

Oil removal eff. (%)

TOD removal eff. (%)

Effluent TOD (g/l)

HC in effluent (mg/l)

pH γo/w dyn/cm

H2SO4 18.3 355 97.1-97.5 99.5-99.9 1.21-1.38 470-543 0.5-0.9

NaCl 20 339 96.3-96.6 99.6-99.7 1.62-1.81 630-705 8.3-8.5

CaCl2 1.5 20 96.3-96.5 99.6-99.9 1.67-1.81 658-705 7.6-8.0 1.4

MgCl2 1.8 17 96.3-96.6 99.6-99.8 1.62-1.79 630-696 8.3-8.4

MgSO4 3 24 96.3-96.6 99.4-99.6 1.65-1.79 639-696 8.3-8.5 1.0

FeCl3 0.6 11 96.4-96.9 99.1-99.5 1.52-1.75 620-714 2.1-2.5 1.8

Alum 1 9 96.2-96.7 99.4-99.7 1.60-1.84 724-855 3.8-3.9 2.1

Ca(COOH)2 1.5 23 90.8-95.9 1.99-4.45 602-639 7.4-7.5

389


III-273

C) Results on destabilization of emulsion B at conc. of 2% by volume of concentrate, 20oC

Dosage Chemicals

g/l meq/l



Effluent TOD (g/l)

HC in effluent

(g/l) pH γo/w

dyn/cm

H2SO4 13.7 265 80.5-82.5 6.10-6.69 3.92-4.37 1.8-2.4

NaCl 20 339 75.0-76.5 8.50-9.00 5.27-5.60 8.4-5.8

CaCl2 2 27 80.5-81.0 6.70-6.90 4.26-4.67 8.0-8.2

MgCl2 2 19 77.5-80.0 7.10-8.10 4.48-5.04 8.3-8.6

MgSO4 3.5 28 78.5-79.5 7.30-7.70 4.59-4.82 7.6-8.2

FeCl3 1.25 23 80.0-81.0 6.70-7.10 4.26-4.48 2.7-3.8

Alum 3 27 80.0-81.0 6.70-7.10 4.26-4.48 4.3-4.6

Ca(COOH)2 2 30 73.5-78.0 7.90-9.60 4.59-5.72 8.3-8.4

D) Results on destabilization of emulsion C at conc. of 2% by volume of concentrate, 20oC

Dosage Chemicals

g/l meq/l



Effluent TOD (g/l)

HC in effluent

(g/l) pH γo/w

dyn/cm

H2SO4 3.7 71 76.3-87.3 3.90-7.70 2.60-4.90 0.6-2

NaCl 60 1016 10.5-71.5 9.30-9.60 5.80-6.00

CaCl2 2 27 70.0-71.0 9.40-9.80 5.92-6.12

MgCl2 2.5 24 69.5-70.3 9.70-9.90 6.07-6.22

Alum 6.2 55 71.0-71.5 9.30-9.40 5.81-5.92

Ca(COOH)2 3 46 64.0-70.0 9.80-11.8 6.12-7.34 3.9-4.2

E) Results on destabilization of emulsion D at conc. of 2% by volume of concentrate, 20oC

Dosage Chemicals

g/l meq/l



Effluent TOD (g/l)


pH γo/w dyn/cm

H2SO4 1.83 35 54.0-58.0 11.3-12.6 8.71-9.54 1.9-3.5

NaCl 50 947 46.5-53.0 13.0-15.1 9.75-11.1 9.2

CaCl2 2 27 50.5-52.0 13.3-13.8 9.95-10.3 8.7-8.8

MgCl2 2 19 53.5-55.5 12.2-12.8 9.23-9.64 8.7

MgSO4 2.5 20 53.5-54.5 12.5-12.8 9.44-9.64 8.7-8.8

Alum 4 36 52.5-57.0 11.6-13.2 8.92-9.85 4.2-6.2

Ca(COOH)2 1.5 23 43.0-52.0 13.3-16.3 9.95-11.8 7.7-8.8

390


III-274

F) Results on destabilization of emulsion E at conc. of 2% by volume of concentrate, 20oC

Anionic surfactants

(%) Chemicals (g/l)

Cmin Cmax



Effluent TOD (g/l)


pH γo/w dyn/cm

Ferric chloride

0.6 0.6 0.8 87.5-89.5 4.85-5.80 1.92-2.29 3.0-3.1

1.0 0.6 1.5 87.1-97.3 1.14-1.32 0.46-0.53 2.7-3.2

1.2 0.6 2.0 96.8-97.7 1.05-1.46 0.42-0.59 1.7-3.0

1.5 0.6 3.0 92.3-97.0 1.37-3.60 0.55-1.42 2.6-3.1

2.0 0.6 3.5 95.2-97.4 1.19-2.22 0.48-0.89 2.7-2.8

Alum

2 4 5 79.5-84.5 7.15-9.40 2.84-3.76 3.9

3 4 8 93.0-98.0 0.92-3.30 0.37-1.28

4 4 11 93.3-97.7 1.08-3.10 0.43-1.23 4.0-4.1

6 4 16 92.5-98.0 0.92-3.35 0.37-1.37 4.0-4.7

Note: The anionic surfactant used in the experiments was Melanol CL30 (for FeCl3) and Melanol V90 (for alum). They are products of SEPPIC, contains 30 and 90% of fatty acid alkylsulfate, respectively.

10.3 Process design

Even though there are many types of destabilization chemicals, from the point of view of process design, the reactions require the same kind of reactors. The reactors in this case are actually the same as those required for coagulation-flocculation process in potable water treatment. In fact, destabilization process can be counted as coagulation–flocculation process. Generally, the process is divided into 2 steps, i.e.,

• Mixing or rapid mixing • Flocculation

Design of each step is described below,

M

M

To separationprocess

FlocculatorRapid mixing

DestabilizationchemicalsEmulsion

Fig. 10.3-1 Schematic diagram of chemical process for emulsion destabilization

391


III-275

10.3.1 Rapid mixing

Rapid mixing is designed to provide complete mixing between wastewater and destabilization chemicals. There are 2 main types of mixing methods, i.e. hydraulic (e.g. weir, flume, static mixer) and mechanical (e.g. pumping, mechanical mixer). Main design criterion of this step is described in the form of velocity gradient (G). Recommend velocity gradient is 300 -1500 s-1[46]. In this chapter we will emphasize on the mixing tank, which is generally used in industries. The value of 100-300 s-1 is common for mixing tank. Recommended detention time (τ) is 5-20 second. However, if wastewater flowrate is low, the detention time will be controlled by the smallest size that can be constructed, which may be greater 20 s. There is practically no adverse affect about this except capital cost and energy cost is higher.

Important equations to design mixing tank are as shown in eq. 10.3.1 and 10.3.2.

5.0

⎟⎟⎠

⎞⎜⎜⎝

⎛=

VPG

μ {10.3.1}

Where G = Velocity gradient (t-1, normally in second –1) μ = Dynamic viscosity of wastewater, normally about the same as

water P = Mixing power (ML2s-3, normally in watt) V = volume of the mixing tank (L3)

Power required for mechanical mixers

ρ53 DnNP p= {10.3.2}

Where Np = Power number, dimensionless ρ = Density of wastewater, normally about the same as water n = Impeller rotating speed (rev/s) D = Impeller diameter (m)

Since the mixing tank is usually small, the impeller used is usually propeller or pitched turbine (fig. 10.3.1-1). The value of Np for propeller and pitched turbine varies with Reynolds number as shown in fig. 10.3-1.2) [63]. Reynolds number in this case is calculated from rotation speed (n) and impeller diameter (D) as shown in eq. 10.3.3.

μρ 2

Re nD= {10.3.3}

Fig. 10.3.1-1 Static mixer (leftmost) and impeller types (from left): flat turbine, pitched blade turbine, propeller type (Source: Memko, Sharp mixers)

392


III-276

Fig. 10.3.1-2 Relation between Np and Re [63]

Calculation procedure is as shown below,

1. Select required gradient (G) and detention time (τ).

2. Calculate required volume (V) from τ and wastewater flowrate (Q). Geometry of the tank should be cubic.

τ/QV = {10.3.3}

3. Calculate require power (P) by eq. 10.3.1.

4. Calculate n and D for required power by eq. 10.3.2. Impeller diameter should be around ½ of tank width. Mechanical mixers are commercially available, thus it can be selected from manufacturer catalog to suit the required tank volume.

10.3.2 Flocculator

Floculator is designed to provide mildly mixing to make the destabilized droplets or flocs collide and coalesce into bigger drops of flocs. There are many variants of flocculator, also categorized mainly by their mixing methods, e.g.. Baffle type, mechanical mixer type. However, for industries, mechanical mixers are widely used for its flexibility (variable speed drive can be used) and reliability. The industries always have personals that can due with O&M of these machines. So there is no problem form maintenance point of view. So we will emphasize on the mechanical mixing flocculation tank. Design criteria of flocculation tank are based on velocity gradient (G) and detention time (τ). Recommended values are tabulated in table 10.3.2-1

393


III-277

Table 10.3.2-1 Recommended value of gradient and detention time

Parameter Range Ref.

Velocity gradient (G) 20-80 s-1 [46]

20-75 s-1 [41]

20-150 s-1 Kawamura S.

Detention time (τ) 10-30 min [46]

20-30 min Kawamura S.

Product of G.τ 23,000-210,000 [41]

For mechanical mixing flocculation tanks, they are usually divided into 3 compartments of the same volume to avoid short circuit. The gradient will vary from the highest value at the first compartment to the lowest at the final compartments (e.g. 75 s-1 to 30-40 s-1 t0 20-30 s-1). There are 2 major types of mechanical mixers used in the flocculation tanks, i.e. impeller type, like the rapid mixing tank, and paddle type, as shown in fig. 10.3.2-1.

Fig. 10.3.2.1 Paddle type mixers (Source: Norfolk WTP, Aqua Pak)

Design procedure of floccuclation tank is relatively identical to that of the rapid mixing tank. For paddle type mixer, its sizing can be calculated by eq. 10.3.4 [64].

μρ

VCnAv

G d

2

3

= {10.3.4}

Where G = Velocity gradient (t-1, normally in second –1) n = The number of paddle blades A = Surface area of one paddle (m2) v = Tip speed of paddle (m/s) Cd = Drag coefficient (normally = 0.6) V = Tank volume (m3) ρ = Density of wastewater, normally about that of water, (kg/m3) μ = Dynamic viscosity of wastewater, normally about that of water

(kg/(m.s))

394


III-278

10.4 Design consideration

Process design of the chemical treatment is relatively simple. However, it must be noted that,

1. There are no universal chemicals and dosages valid for every emulsion. Types of effective chemicals, optimum dosages, removal efficiency and residual pollutants level must be evaluated first in lab scale (e.g. jar test) before design the full-scale chemical process.

Fig. 10.4-1 Jar test equipment (Source: ECE engineering)

2. Normally, there are residual pollutants in the form of soluble co-surfactants, as well as residual salt or acid pollutants from destabilization chemical, which need to be further treated. The extents of residual pollutants depend of type of destabilization chemical as well as ingredient and initial concentration of emulsion.

3. In case of effective destabilization, destabilized oil drops can be decanted within relatively shorts time (20 min to less than 1 hour). For mathematical point of view, it is reasonable to use this decanting time to find the diameter of destabilized droplets. The recommended value is 200 microns (see fig. 10.4-2), which can be used for further calculation of the downstream separation process, such as DAF. For the non-stabilized oil droplets (since oil removal efficiency is < 100%), we can not be sure of its size distribution. To be on the save size, it may be estimated that the size distribution or granulometry of non-stabilized oil droplets remain the same as that before chemical treatment.

395


III-279

Fig. 10.4-2 Example of stabilized emulsion before (right) and after chemical treatment (left)

396


III-280


11.1 General

From the previous section, various oily wastewater treatment processes are described. At their best conditions, most processes can treat the wastewater at the oil removal efficiency more than 90%. However, in cases of high concentration wastewater, residual pollutants from those processes may yet conform to related effluent standards. Furthermore, there are some residual pollutants that are not exactly oils or hydrocarbons but come from other components usually present in emulsion or oily wastewater. These pollutants are mainly surfactants, co-surfactants, esp. in the form of fatty alcohol. Some treatment processes also add residual pollutants to the effluent, such as salt, TDS, or acid from chemical destabilization processes. Thus these effluents need to undergo finishing treatment processes to ensure that their qualities meet the effluent standard.

There are 2 common finishing processes, generally used to treat the effluents of oily wastewater from physico-chemical or physical treatments, i.e. biological treatment and adsorption.

These processes are extensively studied for long times. Details of the processes are available in many sources, including books, journals, literatures, as well as references from users or system manufacturers. For biological treatment, it is actually a science by its own right. Moreover, the researches of GPI lab, directed by Prof. AURELLE do not emphasize on these processes. Thus, in this chapter, they are only briefly described to fulfil the whole content of oily wastewater treatment processes.

11.2 Biological treatment

11.2.1 Basic knowledge

Biological treatment processes are referred to the treatment processes that use microorganisms to eliminate pollutants in wastewater. The pollutants may be utilized by the microorganisms as substrates (biodegradation), which is, then, transformed to new cells and non-pollutant substances, e.g. CO2, H2O, etc. Sometimes, the pollutants are trapped (sorption) in flocs of microorganisms. Biological reactors are designed to provide controlled environment to suit the microorganisms for their growth and utilization of the pollutants. The reactors are also designed to prevent most of the microorganisms from carrying over with effluent, since their presence in the effluent can be considered a kind of pollutants as well.

Biological treatment processes can be categorized by many criteria. According to the types of microorganisms, they are generally divided into 2 types, i.e.,

• Aerobic processes, which microorganisms depend on aerobic respiration. In these processes, oxygen is an important factor for it is used as the electron acceptor in the respiratory metabolism of microorganisms

• Anaerobic processes, which microorganisms do not depend on oxygen. The microorganisms in these cases generate energy by fermentation metabolism that does not require oxygen as an electron acceptor. Some anaerobic microorganisms use other substances as electron acceptors such as nitrate, sulfate, or carbondioxide.

397


III-281

From the point of view of reactor design, biological processes are generally divided into 2 major types, i.e.,

• Suspended growth, which microorganisms are suspended within the water. The major suspended growth aerobic system is the activated sludge process (AS). General schematic diagram of AS is as shown in fig. 11.2.1-1. The microorganisms are controlled to form flocs that can be easily separated from the water in sedimentation tanks.

• Attached growth, which microorganisms are mainly attached to solid surfaces in the biological reactors. Examples of these processes are trickling filter, biocontact reactor and anaerobic filter.

MInlet wastewater

Q, So

X

QR, XR

Se

Effluent

QW, XRWasted sludge

Sedimentation tankor clarifier

Biological reactoror aeration tank

Fig. 11.2.1-1 General schematic diagram of activated sludge

11.2.1.1 Mechanisms of pollutant removal by biological processes

Removal of pollutants in biological treatment are complex and consists of many phenomena, such as removal of suspended solids by enmeshment in the floc, physico-chemical adsorption of colloidal material and biosorption or biodegradation by microorganisms [51]. Many researches had studied these mechanisms and proposed many mathematical models to predict the performance of the processes.

Generally, pollutants are measured in the form of the oxygen demand required to oxidize them. If the oxygen demand is based on biological reaction, it is called biochemical oxygen demand (BOD). If it is based on chemical reaction, such as oxidation with dichromate, it is called chemical oxygen demand (COD). However, pollutant removal can be written in the form of general mathematical model as shown below [51],

n

on S

SXkdtdS

⎟⎟⎠

⎞⎜⎜⎝

⎛−= {11.2.1}

Where S = COD (or BOD) concentration at time “t” (ML-3, normally in mg/l) So = Initial COD (or BOD) concentration (t = 0) (ML-3, normally in

mg/l) t = Time (T, normally, in day) kn = Rate coefficient (T-1, normally, in day-1)

398


III-282

X = Effective biomass concentration. In case of As, X represents MLVSS (Mixed Liquor (the mixture in aeration tank) Volatile Suspended Solids) (ML-3, normally in mg/l)

n = Function power order (e.g. n = 0 zero order, n = 1 first-order, etc.)

For completely mix reactor where BOD reduction ratedecreases with time since readily biodegradable is gradually used up. Eckenfelder [51] proposed the simplified model as follows,

SS

ktXSS eeo =⋅− {11.2.1}

Where Se = COD (or BOD) concentration of effluent (ML-3, normally in mg/l)

Biodegradability of substance is presented by its rate coefficient (k). The higher the value of k, the better the biodegradability. Examples of rate coefficient for some types of wastewater are listed in table 11.2.1-1

Table 11.2.1-1 Rate coefficient for selected wastewaters [51]

Wastewater k (d-1) Temperature (oC)

Domestic wastewater (soluble) 8.0 20

Potato processing 36.0 20

High nitrogen organics 22.2 22

Organic phosphates 5.0 21

Cellulose acetate 2.6 20

Vegetable tannery 1.2 2.0

11.2.1.2 Biodegradability of oily wastewater

Hydrocarbons or oils are considered biodegradable but the degree of biodegradability of each type of hydrocarbons are different, depending on their characteristics, such as molecular structure. Generally, small molecular hydrocarbons, such as hexane, octane, are readily biodegradable while complex or large molecular hydrocarbons, such as dodecane, are considered low- or non-biodegradable.

Examples of biodegradable organic compounds include common aliphatic alcohols, aliphatic aldehydes, phenols, and aliphatic esters. Surfactants, such as alkyl benzene sulfonates and glycols, are biodegradable. These surfactants/co-surfactants are main residual pollutants in effluent from physical and physico-chemical treatment of oil wastewater. So it is feasible to treat these effluent by biological processes.

399


III-283

Examples of substances generally resistant to biological degradation includes some hydrocarbons, esp. long chain aliphatic and aromatics, ethers, tertiary aliphatic alcohols, and tertiary aliphatic sulfonates.

However, biodegradability also depends on reactor design and cultures (or species) of microorganisms. Moderate biodegradable substances may be treated if the reactor is properly design, e.g. the solid residence time (or sludge age) is long enough. Some special cultivated bacteria can decompose oil effectively. For example, commercial mixture containing certain emulsifier and microorganisms are used to treat sea oil spill.

On the other hand, hydrocarbons, mainly in the form of free oil, are considered to be biotoxic substances for microorganisms. Free oils can cover the surfaces of floc and water, thus disturb the oxygen transfer. Emulsified oil cause much less problem and usually biodegradable. Presence of some hydrocarbons at some concentration can cause adverse effect on the performance of the biological treatment process. Biodegradability and biotoxicity of certain substance are presented in table 11.2.1-2. EC50% means effective concentration of the corresponding substances that, if present in the wastewater, reduces the performance of the process to 50% of the maximum or nominal value. Nitorsomonas and heterotrophs represent the species of bacteria for nitrification (nitrogen removal) and generally organic oxidation (BOD removal), respectively.

It should be noted that biodegradability, shown in the table, is expressed in the form of ratio of BOD/TOD. TOD means total oxygen demand. This parameter can be conveniently measured by automatic sensor. So it is widely used in process control because it can be measured in real-time mode. TOD value includes non-organic oxygen demand such as ammonium. Thus it is usually higher than BOD. In laboratory, some researches, including GPI’s, presented the oxygen demand in the form of TOD. When relation between BOD and TOD is known (table 11.2.1-2), it can be used to convert TOD, shown in the researches, to BOD for biological process design.

Some toxic substances, esp. heavy metals, may present in oil products. API recommended the toxic threshold of some metals for biological system as shown in table 11.2.1-3.

400


III-284

Table 11.2.1-2 Biodegradability and biotoxicity data [51]

401


III-285

Table 11.2.1-2 Biodegradability and biotoxicity data [51] (Cont.)

402


III-286

Table 11.2.1-2 Biodegradability and biotoxicity data [51] (Cont.)

403


III-287

Table 11.2.1-3 Concentration of certain metals affecting biological systems [45]

Conc. (mg/l) Chromium (6) Copper Nickel Zinc

Threshold 10 1 1.0-2.5 5.0-10.0

Harmful 500 75 50-200 160

11.2.2 Design consideration and significant finding on biological treatment for oily wastewater from GPI’s researches

GPI’s lab had also studied on feasibility of biological process for treatment of residual pollutants from oil/water separation processes, discussed in the previous chapters [11]. Design consideration and some significant findings on biological treatment of oily wastewater are summarized below,

1. Maximum inlet concentration of oil and expected effluent concentration

It may not be possible to designate the exact values of maximum inlet oil concentration and expected effluent concentration since there are many parameters involved, such as type of oil, reactor design, etc. Furthermore, biological process is noted for its versatility. If the reactor is properly acclimated, the microorganisms can adapt to use the existing pollutants as their main substrate. So even hydrocarbons may be treated. Apart from the data in section 11.2.1.2, some case studies are summarized from many sources to provide some ideas about the performance of biological treatment of oily waster, as shown in table. 11.2.2-1.

Table 11.2.2-1 Case studies on biological treatment of oily wastewater [66]

Process Inlet oil concentration (mg/l) Effluent oil (mg/l)

Activated sludge (AS) 5-100 5-40

Aerated lagoon 5-100 5-40

Stabilization pond 5-100 5-50

Chemical treatment + DAF +AS 5-100 2-20

Using data from several AS plants for petroleum refineries, API [45] reported that BOD and COD removal efficiency is around 50-87% and 60-70%, respectively. MLSS ranged from 1200 to 5000 mg/l. Retention time varied from 2.8 - 29 h. Eckenfelder [51] showed that the biological processes were used effectively in petroleum refinery wastewater treatments. He reported that, with BODin = 138-575 mg/l and CODin = 275-981 mg/l, effluent COD is around 42-106 mg/l and rate coefficient are 1.11-1.7 d-1 and 2.74-7.97 d-1, based on BOD and COD respectively. It shows that, with properly designed reactor, biological process can be used to treat the oily wastewater to meet effluent standard.

2. Treatment of surfactants/co-surfactants

For the treatment of surfactant/co-surfactants, which is the main residual pollutant in permeate of membrane process and other separation processes, WANICHKUL [11] had performed the biodegradability test of the permeate from UF of cutting oil emulsion

404


III-288

(TOD > 3600 mg/l). The result showed that, after 4 days acclamation period, total organic carbon (TOC) removal from 5-hour biodegradation test is 82-90% at the fifth and the sixth day. Thus it can be concluded that biological process is feasible to treat the surfactants-rich effluent from oil/water separation.

3. General design consideration

General design consideration on biological treatment design for oily wastewater, summarized from many literatures, include.

• For industrial wastewater, since oily wastewater is usually only one portion of the whole wastewater, it is recommended, if possible, to treat the oily wastewater separately by physical or physico-chemical process. After that the effluent from those process may be sent to mix with relatively biodegradable wastewater, e.g. canteen wastewater, before sending to biological treatment. This will prevent adverse effect on the biological process by the presence of oil. It also helps saving the energy required for aeration system to cover high oxygen demand of oil. Furthermore, these domestic wastes can provide nutrient (e.g. nitrogen and phosphorus) required for biological treatment, which are usually not present in oily wastewater from industrial processes. General ratio of BOD to nutrients is BOD: N: P = 100: 5: 1. If nutrient quantities in the wastewater are not sufficient, they must be added in the form of chemicals, e.g. urea, anhydrous ammonia and phosphate.

• Since hydrocarbons or oils have very high oxygen demand. Their presence must be taken into account for aeration or oxygenation system in aerobic process, otherwise it may cause some shortage in oxygen level in the reactor, resulting in low performance or, even, plant failure.

11.3 Adsorption

Adsorption process is referred to the process that uses special material that is capable to adsorb molecules or colloids into its surface. This special material is called “adsorbent”. The molecules or colloids adsorbed are called “adsorbate”. For its ability to remove dissolved matters and molecules that somehow remain in the effluent of main treatment processes, the adsorption process is usually used as a tertiary treatment or polishing process, which is the last process before the effluent being discharged. It also used in recycled water treatment system.

Adsorption phenomenon is caused by non-equilibrium of surface force field of the adsorbent. Thus it tends to adsorb the molecules or colloids into itself in order to gain self-equilibrium. Attractive forces between adsorbent and adsorbate are Van Der Waal force and chemical bonds.

Major characteristic of adsorbent is its high surface area, which is the result from its highly porous structure. This characteristic is indicated by a parameter called “specific surface area”, normally on m2/g. Adsorbates are actually adsorbed to the surface of the material, both external surface and pore surfaces. When the adsorbent is saturated by adsorbates, it loses its adsorptive capacity. Some adsorbents can be regenerated, normally by heat, to recover its adsorptive capacity by removing of the adsorbates from its pores and surfaces.

405


III-289

Water fi

lm

Pore diffusionAdsorption to surface

Film di

ffusio

n

Fig. 11.3-1 Schematic diagram of adsorption

Adsorbents can be divided into 3 major types, i.e.,

• Natural adsorbents, such as natural clays, bone char, activated silica. Important limitation of these adsorbents is that they can adsorb only some kinds of molecules.

• Synthetic adsorbents, such as ion exchange resins. Thy have moderate specific surface area (300-500 m2/g). Its main advantage is that it can be regenerated relatively easily and less expensive, such as regeneration by NaCl.

• Activated carbon (AC). AC can be classified as synthetic adsorbents. But, from its very high specific surface area (600-1100 m2/g), it is widely used. So it is classified as its own class. Adsorption process, described in this chapter and many literatures, is usually referred to activated carbon adsorption process.

11.3.1 Activated carbon (AC)

Activated carbon can be made of various materials, such as coconut shells, coals, bones, etc. These materials will be subjected to dehydration process (at low heat) then carbonization (heat at 400-600 oC), following by activation (heat at 750-950 oC) to eliminate tar in its pore. AC can be divided into 2 type, i.e., powder activated carbon (PAC) and granular activated carbon (GAC). AC is the most popular adsorbent for its adsorption efficiency and very specific surface area. Examples of PAC and GAC properties are as shown in table 11.3.1-1. From the table, Iodine number is the parameter that indicates performance of AC on adsorption of small molecules. The higher the number, the better the performance.

Table 11.3.1-1 Examples of PAC and GAC properties

Property GAC PAC

Specific surface area (m2/g) 600-1100 600-1100 Particle size 0.55-1.0 mm 100 – 325 mesh (< 0.15 mm) Apparent density 430-600 kg/m3 520-650 kg/m3

Specific gravity 1.30-1.55 1.4-1.5 Iodine number 850-1050 700-900

406


III-290

AC is capable to adsorb many kinds of substances, including heavy metals, phenol, surfactants and hydrocarbons. Thus it is widely used as adsorbent material for polishing process of industrial effluents. Absorption capacities of some hydrocarbons are listed in table 11.3.1-2. However, AC cannot absorb very small molecules (atoms of C < 3), such as small molecular of organic acid and some alcohol. Anyway, they are readily biodegradable substances, which can be easily eliminated by biological process.

For oily wastewater treatment polishing process, AC is usually used in the form of GAC filter. Its hydraulic working principle is similar to that of sand filter or granular bed coalescer, described in chapter 5. Thus, in this chapter, only this application will be described.

11.3.2 Basic knowledge

11.3.2.1 Isotherm diagram

The adsorptive capacity (q) at a specific temperature is normally presented in the form of graph, called isotherm diagram (fig. 11.3.2-1). The graph is normally obtained by experiment. In the experiment, the water to be treated (initial concentration = Co) and the adsorbents are put in contact in a constant stirred reactor for a sufficient time to approach equilibrium. Concentration will decrease from Co to equilibrium concentration (Ce).

1. Freundlich model, as shown in eq. 11.3.1

)/1( neCkq ⋅= {11.3.1}

Where q = Adsorptive capacity (amount of adsorbates adsorbed per unit weight of adsorbent) (M/M)

k,n = Empirical constants Ce = Equilibrium concentration (ML-3)

2. Langmuir model, as shown in eq. 11.3.2

e

e

bCCab

q+⋅

=1

{11.3.2}

Where a, b = Empirical constants

It is recommended to fit the empirical isotherm with the both models and use the one that give more accurate correlation.

407


III-291

Table 11.3.1-2 Adsorptive capacity of AC for some hydrocarbons [65]

0

50

100

150

200

250

300

0 1000 2000 3000 4000 5000 6000

Equillibrium concentration (Ce), mg/l

Ads

orpt

ive

capa

city

(q),

mg/

g

Fig. 11.3.2-1 Examples of adsorption isotherm diagrams

408


III-292

11.3.2.2 Process analysis for GAC filter

In GAC filter (or contactor), as wastewater passes through the bed at a design rate, the pollutant will be adsorbed and its concentration will be reduced as shown in fig. 11.3.2-2. The height of the bed in which adsorption occurs is called the mass transfer zone (MTZ). After of MTZ, the pollutant is practically eliminated, so there is no adsorption in the lower bed under MTZ. Upper portion of bed is exposed to the pollutant first so this portion is the first to be saturated. After it is saturated, it no longer has adsorptive capacity. The adsorption will occur at the lower portion of bed instead. Thus it seems that the MTZ moves along the depth of the bed as shown in fig. 11.3.2-1. When it move to the bottommost of the bed, it signifies that the bed is expired. Pollutant concentration no longer meets the required value. Thus the bed should be replaced.

Ha

CCoCe

H

t1

CCoCe

H

< t2

Satu

red

zone

HaMTZ

CCoCe

H

< t3

Ha

CCoC = Ce

H

< t4

Bed needs to be replaced.

Outlet C does notmeet requirement.

GAC bed.HT

Feed

Outlet

Fig. 11.3.2-2 Evolution of pollutant concentration along the bed depth

MTZ is obtained from experiment. The wastewater will be fed to a GAC bed at a constant rate. Effluent concentration and feed quantity of wastewater is collected and plotted as shown in fig. 11.3.2-3. Then, the bed height of MTZ (or adsorption height (Ha)) can be calculated by eq. 11.3.3, which is derived based upon the symmetry between C/V curve and C/H curve.

Vc

C

Co

CeH

C

Co

Ce

Ha VaVb Ve

Qa Qa

Fig. 11.3.2-3 C/V curve and C/H curve

409


III-293

)( ce

aca VV

VHH

−= {11.3.3}

Where Hc = Bed height corresponding to the concentration “C” (L) Vc = Accumulated feed volume corresponding to the concentration “C”

(L3) Ve = Accumulated feed volume at equilibrium (L3) Va = Accumulated feed volume required to saturate the MTZ (L3)

From fig.11.3.2-2, when the MTZ moves to the bottommost of the bed, it does not mean that the whole bed is saturated. Only the portion above the MTZ is saturated. The bed is the MTZ is partially saturated otherwise the effluent concentration will not meet the design value. This means the bed in the MTZ cannot be fully utilized up to it capacity. If the MTZ is large, this means partially utilized zone is large too. The available capacity left form utilization in the MTZ, represented by a toned area in fig. 11.3.2-3, can be calculated eq. 11.3.4.

∫ −=Ve

Vbca dVCCQ )( 0

{11.3.4a}

∫ −=Ve

Vbc

aa dVCC

mq )(1

0 {11.3.4b}

Where Qa = Total available (or left) capacity of absorbent in MTZ (M) Co = Initial concentration (ML-3) qa = Available adsorption of MTZ (M/M) ma = Mass of absorbent in MTZ (M)

C/V curve and C/H curve vary with empty bed velocity (or flowrate per unit area of bed), as shown in fig. 11.3.2-4. It means that the value of Ha and Qa also vary with empty bed velocity (v). If wastewater flowrate and empty bed velocity (v) are specified and the C/H or C/V at that “v” and isotherm diagram are known from experiments, total bed height (HT)at any given total operating period (tT) before the adsorbent needs to be replaced can be calculated from basic mass balance equation, as shown in eq. 11.3.5.

abTeoo QqAHCCV −=− ρ)( {11.3.5a}

TTo tAvtQV ⋅⋅=⋅= {11.3.5b}

Where V0 = Total accumulated feed volume before the bed needs to be replaced (L3)

tT = Total operating time before bed replacement ( or regeneration) (ML-3)

A = Cross section area of GAC bed (L2) HT = Total Bed depth (L) ρb = Bulk (or apparent) density of bed (ML-3) v = Empty bed velocity (LT-1)

410


III-294

The value of q is obtained from isotherm diagram while the value of Qa is calculated by eq. 11.3.4 from C/V relation obtained from experiments. Empty bed velocity can be chosen arbitrarily but it will affect the value of Qa as described on the previous paragraph. Recommended value of v is around 2.5 m/h per every 0.30 m of bed height (e.g. v = 7.5 m/h for bed height of 0.90 m.). Total operating time (tT) can be chosen to match main manufacturing process plant shutdown, in case of industrial wastewater, such as 1 year, etc. Recommended retention time is not less than 5 min.

Since the bed is not totally saturated, effective saturation of bed (E) can be calculated by eq. 11.3.6a and b. The value of E is normally between 0.5 – 0.95 (average 0.75).

))1(1(100(%)HH

eE a−−⋅= {11.3.6a}

ao

a

VCQ

e −=1 {11.3.6b}

From the concept of E, eq. 11.3.5a can be rewritten in the form of E, as shown in eq. 11.3.6c

qEAHCCV bTeoo ρ=− )( {11.3.6c}


Important parameters that affect the performance in adsorption process include,

1. Types of adsorbate: Adsorptive capacities of AC for each molecule are different. If possible, isotherm test on the pollutants to be removed should be performed.

2. Specification of carbon: Adsorptive capacity depends mainly on surface area of AC. The higher the surface, the better the capacity.

3. Temperature: Since the process is exothermal, adsorptive capacity will decrease with increase of temperature.

4. pH: pH affects ionization of molecules. Thus it may affect the ionization of the molecule to be adsorbed.

5. Turbulence: Adsorption also depends on mass transfer of adsorbates to the surface. So turbulence of reactor affects the performance. In GAC filter, film diffusion is usually rate determining step. Design parameters such as empty bed velocity can be optimized in lab-scale test to obtain good performance. Generally, recommended design criteria is proven to provide good result.

411


III-295


11.3.3.1 GAC filter sizing

To calculate the size of GAC filter, the following procedure is recommended,

1. Find the reference about isotherm and relation between Qa and V at various values of v. ZHU recommended these relation for some common co-surfactants, usually found in effluent of cutting oil emulsion after physical or physico-chemical processes, as shown in table 11.3.3-1.

2. Select design empty bed velocity (v) and total operating time (tT)

3. From required wastewater flowrate (Q), calculate cross section area of GAC filter (A) from

vQA /= {11.3.7}

4. Calculate q and Qa at the selected value of v, if the relation q VS. v and Qa VS. v are known from item 1.

Table 11.3.3-1 Adsorption isotherm and MTZ data of some co-surfactants [21]

Co-surfactants Relation

Benzylic alcohol

Adsorption isotherm [Lungmuir model], in mg/g (Ce in mg/l) e

e

CCq 31077.81

19.2−×+

=

Height of mass transfer zone, in cm (v in m/h) vaH 112.01026.6 ×=

Available capacity of the bed in MTZ (by volume) AQ va ××= 078.01083.3

Butyldiglycol

Adsorption isotherm [Lungmuir model], in mg/g (Ce in mg/l) e

e

CC

q 31016.4139.1

−×+=

Note: q in mg/g, Ha in cm, Qa in cm3, A in cm2, C in mg/l

5. Calculate Vo and HT from eq. 11.3.5.

6. When relation between Qa and v or Ha and v are not available. It can be assumed that Qa = 0, replace q in eq. 11.3.5 with q/ S.F, then use procedure in step 2 to step 5 to calculate HT. The value of safety factor is between 1.3 to 2. However, it is recommended to perform lab-scale to obtain the data about Qa and Ha.

7. In case that the isotherm data is not available, it is strongly recommended to perform lab-scale test. The data of other substances of the same kind can be used to estimate the size of the filter, but for preliminary evaluation only.

412


III-296

11.3.3.2 Headloss

Headloss of the filter can be calculated in the same manner as that of sand filter or granular coalescer. The size of GAC is around 0.5-1.0 mm, about identical to that of coalescer bed. Thus headloss of GAC bed should be calculated using the equation (Koseny-Carman) and other data of the coalescer bed, e.g. porosity data, described in chapter 5.

11.3.3.3 Filter construction and accessories

GAC filter construction and accessories are relatively identical to those of sand filter. So feed pumps, piping, underdrain and backwash system can be designed in the same way as for sand filter. However specific gravity of GAC is around that of anthracite (1.3 –1.5), which is lower than sand’s. Thus recommended backwash rate is around 20 – 50 m/h. Bed expansion is about 50%. It should be noted that wet GAC is corrosive. Rubber liner or equivalent anti-corrosion material must be used.

a) Granular activated carbon b) Powder activated carbon

c) AC plant (Source: Carbonel) d) Example of GAC filter (Source: NFS)

Fig.11.3.3-1 Examples of AC and GAC filter

413

Chapter 12 Guideline for treatment process selection and examples

III-297

Chapter 12 Guideline for treatment process selection and examples of treatment processes for certain oily wastewaters

12.1 Guideline for treatment process selection

From detail of each process, described before in the previous chapters, it is clearly shown that each process has its own limitation and can be effectively used on some range of oily wastewater. In this case, the treatment system will consist of several processes connecting together as a process train. On the other hand, for some ranges of oily wastewater, there is more than one process that can be applied to treat them. To design practical wastewater treatment system, designers should understand the nature and limitation of each process, both from technical and economic points of view.

It must be noted that treatment process selection and design is a state-of-art. It is very difficult to decide that which one is the most suitable one. Normally, engineers base their design on their experience. In some situation, it is still difficult to choose the best process train because there are more than one applicable process. In this case, to select the most suitable process, advantages, disadvantages and economic constraints and theoretical comparison of removal efficiency between each feasible process shall be taken into account. Concept of least cost method can be used, as described previously in chapter 2.

In previous chapters, many researches, covering almost all of the possible range of oily wastewater and their treatment processes, have been reviewed. These researches can be summarized to formulate a guideline for process selection and to develop a computer program for design and simulation of oily wastewater treatment process.

The guideline, categorized by the types of oily wastewater, can be summarized as shown in table 12.1-1 and section 12.1.1 to 12.1.4.

12.1.1 Oil film

The oil film is the simplest form of oily wastewater. Technically, the oil in this case is already separated from the water and presents in the form of 2 separate layers of oil and water, rather than homogeneous mixture of oil and water. This type of oily wastewater might be raw wastewater from the source or the result from other treatment processes, such as decanted oil from decanter or destabilized oil retentate from UF process.

Oil layer can be removed or skimmed from water surface in several ways. The simplest form of skimmer is overflow device, such as weir, bell mouth pipe. GPI lab has developed oil disk skimmer and oil drum skimmer based on surface chemistry concept, as described in chapter 3. These types of skimmers have good selectivity for they recover relatively water-free oil.

To select the skimmer, one should consider:

• Usage of skimmed oil: If the water-free skimmed oil is required, the skimmers with good oil-water selectivity, such as drum skimmer or disk skimmer, may be required to assure efficient separation.

414


III-298

Tabl

e 12

.1-1

Gui

delin

e fo

r oily

was

tew

ater

pro

cess

sele

ctio

n (b

ased

on

GPI

’s re

sear

ches

)

415


III-299

• Geometry or size of the decanter or oil retaining vessel: If the size of the tank is not so large. Simple hydraulic device, such as bell mouth, can work effectively. To enhance its oil-water selectivity, the thickness of oil layer can be increased to avoid carry-over of water with the skimmed oil. In case of small tank, this can be done because the open surface of tank is small. So loss of volatile oil in this case is quite acceptable. However, it may not be practical for large tanks.

• Economic point of view: investment and operating cost of skimmer and recovered benefit from usage of skimmed oil and other related cost and benefit (if any) should be taken into account.

12.1.2 Primary emulsion

For primary emulsion, the oil droplets in the wastewater are relatively big and can be separated by natural or non-accelerated process, i.e. decanter However, besides the oil droplet size, presence of suspended solids in the wastewater is an important factor that has to be taken into account. Guideline for treatment process selection in this case, categorized by presence of suspended solids, will be as described below.

• Primary emulsion without presence of suspended solids: Examples of this case are condensate and process waters from industries. In this case, both simple decanter and lamella decanter are applicable. Without presence of suspended solids, there is no risk of clogging. So, compact decanter or closely inserted lamella decanter, such as Spiraloil, can be used.

• Primary emulsion containing coarse suspended solids: Example of this case is wastewater from general washing, etc. In order to consider if the suspended solids are classified as coarse suspended solids, the rising velocity of droplet and the settling velocity of the solids will be compared. If the settling velocity of solids is greater than the rising velocity, the suspended solids, then, tend to settle within the oil separator. In this case, the suspended solids are considered as “coarse” suspended solids. For API tank, design cut size of oil droplet is around 150 microns, then the suspended solids that can be classified as “coarse” will be around 50 microns in diameter. For this type of wastewater, API tank, inclining tube settler or lamella decanter with large spacing between plate are recommended for they can serve 2 purposes, i.e., solids and oil separation. Closely inserted decanter is not recommended for risk of clogging. Dissolved Air Flotation is also a good choice in this case.

• Primary emulsion containing fine suspended solids: Example of this case is stormwater. If these solids are fine, it will not settle in the oil separator. In this case, theoretically, it can be treated in the same way as condensate’s. However, the solid separation process should be provided. DAF is recommended in this case because it can separate both oil and solids alike.

Among these 3 categories, the second or combination between the second and the third categories are more common. Other accelerated processes, such as hydrocyclone, are also applicable. But it requires further treatment process, such as coalescer, to treat the separated oil. Thus it may not be economical choice when simple process can do the job as well. In effect, when the land is not the restricting condition, the process with relatively low loading rate, such as decanter or DAF, may be useful. Because they can serve as both solid

416


III-300

and oil separation, as stated above. Their voluminous size can be used to dampen any transient conditions occurring. And they can always be upgraded or changed to other processes that require smaller footprint in the future. Membrane process and thermal process are applicable but not recommended for economic reason. After oil separation, the wastewater may need finishing processes (e.g. biological treatment and carbon adsorption), depending on amount of residual pollutants.

12.1.3 Secondary emulsion

For secondary emulsion, the oil droplet sizes in the wastewater are between 5 to 100 microns. However, since lower limitations of some treatment processes are around 20 microns, it is reasonable to divide the emulsion into 2 groups, i.e. (1) droplet size between 20-100 microns and (2) droplet size from 5-20 microns. In both groups, the treatment processes are still affected by presence of suspended solids. Guideline for treatment process selection, categorized by presence of suspended solids, will be as described below.

12.1.3.1 Droplet size between 20-100 microns

• Emulsion without presence of suspended solids: Examples of this case are condensate and oil-contaminated process waters from industries. In this case, decanter alone is unable or uneconomical to completely treat this emulsion. However, lamella decanter and spiraloil may be used as preliminary treatment (but not necessary) since it can handle the upper range of oil droplets in the emulsion. Lamella decanter is the better choice between the two because it is more voluminous so it can handle some strayed solids, if any, or equalize some shock load better than spiraloil. Every type of coalescers and hydrocyclones can treat this group of emulsion effectively. However, for hydrocyclone, it always requires further process to treat the separated oil. Combination of hydrocyclone/ coaleser is recommended. Hydrocyclone can reduce the amount of wastewater to be treated. Then the concentrated wastewater is sent to a small coalescer since it handles only small portion of wastewater.

• Emulsion containing coarse suspended solids: Example of this case is wastewater from general washing, etc. In this case, Upflow granular bed coalescer is not recommended since it can be clogged and its cleaning is very difficult. Downflow granular bed coalescer, fibrous bed coalescer (only brush type) and hydrocyclones are applicable. Solids can be separated by API tank or lamella separator, which can be used as preliminary treatment process otherwise specific solid separation process must be provided. DAF is recommended for its solid and oil separation capability.

• Emulsion containing fine suspended solids: Example of this case is stormwater. In this case, theoretically, it can be treated in the same way as condensate’s. However, the solid separation process should be provided. DAF is recommended in this case because it can separate both oil and solids alike.

12.1.3.2 Droplet size between 5 - 20 microns

In this range of droplet size, these two processes are not applicable. Fibrous bed coalescer or hydrocyclone alone still cannot treat this emulsion. Recommended process is DAF or combination of coalscer/ hydrocyclone/ coalescer. For the latter case, the first coalscer will partially coalesce the oil droplets to bigger sizes (around 10-20 microns), which

417


III-301

are sufficient to separate by the hydrocyclone. Concentrated wastewater from the hydrocyclone will be treated by the downstream coalescer. In case that SS is present, solids separation process must be provided

Membrane process and thermal process are applicable but not recommended for economic reason. After oil separation, the wastewater may need finishing processes (e.g. biological treatment and carbon adsorption), depending on amount of residual pollutants. It is possible that surfactants may be present in this type of wastewater.

Effect of surfactants on the performance of each process is provided in the previous chapters. When the efficiency equation includes the interfacial tension, efficiency of process with presence of surfactants can be readily calculated. If not, its effect can be compensate by applying some safety factors.

12.1.4 Macroemulsion and microemulsion

For macro- and microemulsion, the oil droplet sizes in the wastewater are between 0.1 to 5 microns and 10 – 60 nm, respectively. At these ranges of dropets, Brownian motion is predominant. Furthermore the droplets are stabilized by surfactants so they cannot coalesce. STOKE’s law-based process without destabilization of emulsion are completely not applicable. Recommended treatment processes are divided into 3 groups, i.e.,

• Chemical destabilization process and decanter or DAF: Major step of this process train is destabilization of the emulsion. Detail of this process is described in chapter 10. After destabilization, the droplets can coalesce or flocculate, then be separated by decanter of DAF. The use of coalescer and hydrocyclone may not provide further advantage since the interfacial after destabilization is still low which is unfavorable for these two processes. Precaution for this process train is the residual pollutants, depending on characteristic of wastewater and destabilization chemicals. Finishing process is normally required.

• Membrane process: Membrane processes, esp. UF, are capable of treating these emulsions. However, residual pollutant, esp. co-surfactants, concentration are always high. Permeate from UF of emulsion treatment can be sent to RO to remove these soluble pollutants. Anyway, RO effluent still contains relatively high residual pollutants and needs further polishing treatment, such as biological treatment or carbon adsorption.

• Thermal process: the emulsion can also be treated by classic distillation. However energy cost will be high. Again, dissolved residual pollutant can be found in the distillate. So it needs polishing process.

12.1.5 Concentrated oily wastewater or refinery slops

Concentrated oily wastewater in this case refers mainly to retentate from membrane process of stabilized emulsion treatment, such as cutting oil treatment. Refinery slops refers to residue from refinery process in the form of viscous, concentrated oily wastewater, Slops usually compose of 40-80% of water, 20-50% of hydrocarbon and 1-10% of suspended solids. Because of their high concentration and viscous characteristic, they cannot be treated effectively by normal physical processes. To treat these wastes, GPI lab recommends the use of heteroazeotropic distillation (chapter 9). The process provide the opportunity to recycle the oil, esp. in case of slops, since the separated oil is relatively water-free and ready to be sent

418


III-302

back to refinery process. In case of retentate, the separated oil, also water-free, can be disposed of as used oil or reused in fuel for incinerator.

12.2 Examples of treatment processes for certain oily wastewaters

Apart from detailed studies on each process, GPI lab also conducted feasibility studies on treatment process trains of certain oily wastewaters, i.e. non-stabilized emulsion and cutting oil emulsion. Recommended processes trains for the two wastewaters, based on GPI’s researches, are summarized as follows,

12.2.1 Treatment of cutting oil emulsion

Cutting oil emulsion is one of the most difficult-to-treat wastewater. From it components, it contains relatively high concentration of oil and surfactants. Concentrate of macroemulsion contains around 80 % of oil and 15 % of surfactants/co-surfactants. For microemulsion, it contains 20-30 % of oil, 40-50 % of surfactants/co-surfactants and 5-10% of water. Generally, the concentrates are mixed with water at 2-6 % by volume of to prepare cutting oil emulsion. Oil and surfactants/co-surfactants contribute to very high concentration of oxygen demand. Oil droplets are very small and very stable. Thus they cannot be treated by STOKE’s law-based processes alone.

WANICHKUL [11] had compared the performance of many possible process trains, e.g. combination of various membrane processes, thermal processes and biological processes. The result showed that the feasible treatment process train, as shown in fig. 12.2.1-1, should consist of the following processes, i.e.,

1. Ultrafiltration: UF is used to treat wasted cutting oil emulsions from sources. It is recommended to perform lab-scale or pilot-scale test to find the optimum type of membrane and operating condition before design the real system. Operation at mid- or lower side of recommended pressure range might result in less clogging and high average long-term flux. In case of macroemulsion, partially destabilization by addition of salt, as shown in section 8.2.3.2, may enhance permeate flux. But a skimmer should be provided to remove destabilized oil otherwise it will clog the membrane, resulting in lower flux. Outlet oil concentration of 0-10 mg/l can be expected.

2. Reverse osmosis: RO is used to remove major part of residual pollutants, mainly dissolved co-surfactants, from UF permeate. At optimum operating condition, which depends on types of emulsion and membrane, TOD removal efficiency around 80-90% can be expected. However, since the inlet TOD is very high, esp. in the case of microemulsion, RO permeate may still contain high residual pollutants and need polishing treatment. Retentate from RO can sent to treat together with the UF retentate.

3. Heteroazeotropic distillation: It is used to treat the retentate from UF and RO. With proper entrainer, residue from the distilltion will contain relatively water-free oil, which is the base oil in cutting oil concentrate. The residue can be disposed as oil, not waste, or can be recycled or used in an incinerator. The distillate is 2 separate layers of entrainer, which can be recyled for the next distillation cycle, and oil which somehow still contain some dissolved pollutants. It is recommended to send this distillate to mix with the UF permeate for treatment by RO.

419


III-303

Heteroazeotropic distillation

M

X

Return sludge

Effluent

Wasted sludge

ClarifierAeration tankDistillate

TI

Heater

Permeate

Retentate

Membrane

Feed pump

Storage tank Feed

Po

Pi

Pp

Heat exchanger R

O P

erm

eate

Retentate

Membrane

Feed pump

Feed

Po

Pi

Pp

Heat exchanger

Effluentdischared(If possible)

Inlet wastewater

Entrainer

Heat exchanger

Biological treatment

Ultrafiltration Reverse osmosisWas

ted

rete

ntat

e

Was

ted

rete

ntat

e

Fig. 12.2.1-1 Schematic diagram of cutting oil emulsion treatment system

4. Biological treatment: Distillate and RO permeate are treated by biological treatment before discharge to receiving water. WANICHKUL showed that the biological treatment can use dissolved pollutants in RO as main substrate with TOC removal efficiency more than 90%. So higher efficiency can be expected if the permeate is mixed with the wastewater of more biodegradability, such as domestic wastewater from office or canteen.

12.2.2 Treatment of non-stabilized secondary emulsion

This emulsion can be treated with very high efficiency by various methods. So the efforts to enhance the performance are based mainly on minimization of process footprint. Conventional oily wastewater treatment system is usually based on DAF, as shown in fig. 12.2.2-1, which is also effective when surfactant is present.

However, in case of non-stabilized wastewater, GPI lab proposes the compact treatment process train for this emulsion (fig. 12.2.2-2), consisting of the following processes.

1. Upstream in-line fibrous bed coalescer: The coalescer can partially coalesce oil droplets of diameter into bigger oil droplets. The fibrous bed is used since it is hardly clogged by suspended solids. Empty bed velocity of the coalsecer is reported to be as high as 130 m/h or more [11].

2. Hydrocyclone: The hydrocyclone will concentrate the partially coalesced oily wastewater from the upstream coalescer. Since the oil droplets are partially coalesced, thus bigger. The efficiency of cyclone is improved. The water can

420


III-304

be discharged or sent to further treatment if some SS or other pollutants are still present. Concentrated oil will be sent to downstream coalescer.

3. Downstream fibrous bed coalescer: The coalescer receives the concentrated or purged oily wastewater from the cyclone. Since the quantity of wastewater is much reduced, the coalescer can be designed and operated at low loading rate to ensure high efficiency without consuming unacceptable footprint. The oil will be separated from the water and then reused or disposed of as oil. The water can be discharged or sent to further treatment if SS or other pollutants are still present.

Inlet wastewater

M

X

Return sludge

Effluent

ClarifierAeration tank

API tank

Biological treatmentDAF

To sludge treatment

To sludge treatment

Tosludge

treatment

Saturator

Primary sedimentation tank

OilOil

Oil

Skimmer

Skimmer

Wasted sludge

GAC filter(if required)

M M

Chemicals

Chemical treament

Fig. 12.2.2-1 Schematic diagram of conventional oily wastewater treatment system

Inlet wastewater

M

X

Return sludge

Effluent

ClarifierAeration tank

Biological treatmentPrimary sedimentation

tank or API tank

Oil

Wasted sludge

GAC filter(if required)Oil

Fibrous coalescer/Hydrocyclone/

Fibrous coalescer

Fig. 12.2.2-2 Schematic diagram of compact oily wastewater treatment system for Non-stabilized secondary emulsion

421

Part IV Computer program development


IV-i

Contents

Page

Part IV Computer program development Chapter 1 Program overview

1.1 Introduction IV-2 1.2 Conceptual design of the program IV-2

1.2.1 E-book mode IV-3 1.2.2 Recommendation mode IV-5 1.2.3 Design mode IV-6 1.2.4 Analysis mode IV-7

1.3 Development tools IV-11 1.3.1 Main development software package IV-11 1.3.2 Special graphic user interface (GUI) component IV-12 1.3.3 The third party software IV-12

1.4 Program architecture IV-13 1.4.1 Forms IV-13 1.4.2 Modules IV-16 1.4.3 Modules IV-16 1.4.4 Class modules IV-16 1.4.5 Add-in project IV-16

1.5 Program development IV-17

Chapter 2 Program reference and user manual 2.1 Overview of the program IV-15

2.1.1 Main program IV-18 2.1.2 Project window IV-20 2.1.3 E-books worksheet IV-20 2.1.4 Recommend worksheet IV-21 2.1.5 Design worksheet IV-23 2.1.6 Analysis mode IV-24 2.1.7 Warning dialog box IV-26

2.2 Program capability IV-26 2.3 Program limitation IV-26 2.4 System requirement IV-28 2.5 User instruction IV-28

2.5.1 Program Installation IV-28 2.5.2 Starting the program IV-29 2.5.3 Using E-book mode IV-29 2.5.4 Using Recommend mode IV-29 2.5.5 Using Design mode IV-31 2.5.6 Using Analysis mode IV-33 2.5.7 Printing and file operation IV-36

2.6 Upgrading procedure and recommendation for further IV-37 development 2.6.1 Upgrading procedure IV-37 2.6.2 Recommendation for further development IV-39


IV-ii

Contents

Page

Chapter 3 Process references 1) Drum skimmer IV-42 2) Disk skimmer IV44 3) Simple decanter IV-46 4) Compact decanter IV-49 5) Customized decanter IV-52 6) Granular bed coalescer IV-55 7) Brush type bed coalescer IV-59 8) Dynamic fibrous bed coalescer IV-63 9) Metal wool bed coalescer IV-67 10) Dissolved air flotation IV-71 11) Two-phase hydrocyclone IV-77 12) Three-phase hydrocyclone IV-81 13) Ultrafiltration IV-85 14) Reverse osmosis IV-89 15) Heteroazeotropic distillation IV-92 16) Stripping IV-94 17) Chemical destabilization, coagulation-flocculation IV-96 18) Biological treatment IV-99 19) GAC filter IV-101 20) Customized concentrator IV-105 21) Customized oil separator IV-107 22) Customized inline concentrator IV-109 23) Inlet IV-111 24) Outlet IV-112 25) Flow merge IV-113 26) Flow split IV-115


IV-iii

Table Page

Table 2.2-1 The processes included in the GPI program IV-27 Table 1-1 Related parameters of drum skimmer IV-42 Table 1-2 Influent parameters for drum skimmer IV-43 Table 2-1 Related parameters of disk skimmer IV-44 Table 2-2 Influent parameters for disk skimmer IV-44 Table 3-1 Related parameters of simple decanter IV-47 Table 3-2 Influent parameters for simple decanter IV-48 Table 4-1 Related parameters of compact decanter IV-50 Table 4-2 Influent parameters for compact decanter IV-51 Table 5-1 Related parameters of customized decanter IV-53 Table 5-2 Influent parameters for customized decanter IV-54 Table 6-1 Related parameters of granular bed coalescer IV-56 Table 6-2 Influent parameters of granular bed coalescer IV-58 Table 7-1 Related parameters of brush type bed coalescer IV-60 Table 7-2 Influent parameters of brush type bed coalescer IV-62 Table 8-1 Related parameters of dynamic fibrous bed coalescer IV-64 Table 8-2 Influent parameters of dynamic fibrous bed coalescer IV-66 Table 9-1 Related parameters of metal wool bed coalescer IV-68 Table 9-2 Influent parameters of metal wool bed coalescer IV-70 Table 10-1 Related parameters of metal wool bed coalescer IV-74 Table 10-2 Influent parameters of DAF IV-76 Table 11-1 Related parameters of two-phase hydrocyclone IV-79 Table 11-2 Influent parameters of two-phase hydrocyclone IV-80 Table 12-1 Related parameters of three-phase hydrocyclone IV-83 Table 12-2 Influent parameters of three-phase hydrocyclone IV-84 Table 13-1 Related parameters of ultrafiltration IV-87 Table 13-2 Influent parameters of three-phase hydrocyclone IV-88 Table 14-1 Related parameters of RO IV-90 Table 15-1 Related parameters of heteroazeotropic distillation IV-93 Table 16-1 Related parameters of stripping IV-95 Table 17-1 Related parameters of chemical destabilization IV-97 Table 18-1 Related parameters of biological treatment IV-100 Table 19-1 Related parameters of GAC filter IV-103 Table 20-1 Related parameters of customized concentrator IV-106 Table 21-1 Related parameters of simple decanter IV-108 Table 22-1 Related parameters of chemical destabilization IV-110


IV-iv

Figure Page

Fig. 1.2-1 Flow chart of the program under the scope of work of this thesis IV-4 Fig. 1.2.1-1 Detailed flowchart of e-book mode IV-5 Fig. 1.2.2-1 Detailed flowchart of recommendation mode IV-6 Fig. 1.2.3-1 Detailed flowchart of design mode IV-8 Fig. 1.2.4-1 Detailed flowchart of design mode: Data input operation IV-9 Fig. 1.2.4-2 Detailed flowchart of design mode: Analysis and result display operation IV-10 Fig. 1.2.4-3 Detailed flowchart of design mode: file management operation IV-11 Fig. 1.4-1 Structure of the program IV-13 Fig. 1.4.1-1 Main form (under construction) IV-14 Fig. 1.4.1-2 Project form IV-15 Fig. 1.4.1-3 Input data form IV-15 Fig. 2.1.1-1 Graphic user interface of the main program IV-20 Fig. 2.1.3-1 Graphic user interface of the E-book worksheet IV-21 Fig. 2.1.4-1 Input screen ( a tabbed worksheet in project window) IV-22 Fig. 2.1.4-2 Result screen. The program will put tick marks in the first column. IV-22 Fig. 2.1.4-3 Flowchart of recommendation logic IV-23 Fig. 2.1.5-1 User interfaces of design worksheet IV-24 Fig. 2.1.6-1 User interface of analysis worksheet IV-25 Fig. 2.1.7-1 Example of warning dialog box IV-26 Fig. 2.5.3-1 Using E-book mode IV-29 Fig. 2.5.4-1 Using recommend mode: Input data screen IV-30 Fig. 2.5.4-2 Using recommend mode: Result window IV-30 Fig. 2.5.5-1 Using Design mode: Step 1 “Wastewater data input” IV-31 Fig. 2.5.5-2 Using Design mode: Step 2 “Process selection” IV-32 Fig. 2.5.5-3 Using Design mode: Step 3 “Process data input” IV-32 Fig. 2.5.5-4 Using Design mode: Step 4 “Result” IV-33 Fig. 2.5.6-1 Using Analysis mode: Draw the schematic diagram IV-34 Fig. 2.5.6-2 Using Analysis mode: Process data input IV-35 Fig. 2.5.6-3 Using Analysis mode: Viewing result in the diagram IV-36 Fig. 2.5.6-4 Using Analysis mode: Exporting the result to excel IV-36 Fig. 2.6.1-1a Configuration of the database IV-37 Fig. 2.1.6-1b Examples of existing database IV-38 Fig. 2.1.6-2 Example on database upgrading. When new category is added in the IV-38

database (left), Related field in the program will be automatically updated (right)

Fig. 1-2a Icon of drum skimmer IV-43 Fig. 1-2b Graphical diagram of drum skimmer IV-43 Fig. 2-1a Icon of disk skimmer IV-45 Fig. 2-1b Graphical diagram of disk skimmer IV-45


IV-v

Figure (Con’t)

Page

Fig. 3-1a Icon of simple decanter IV-47 Fig. 3-1b Graphical diagram of simple decanter IV-47 Fig. 4-1a Icon of compact decanter IV-51 Fig. 4-1b Graphical diagram of compact decanter IV-51 Fig. 5-1a Icon of customized decanter IV-53 Fig. 5-1b Graphical diagram of customized decanter IV-53 Fig. 6-1a Icon of granular bed coalescer IV-57 Fig. 6-1b Graphical diagram of granular bed coalescer IV-57 Fig. 7-1a Icon of brush type bed coalescer IV-60 Fig. 7-1b Graphical diagram of brush type bed coalescer IV-60 Fig. 8-1a Icon of dynamic fibrous bed coalescer IV-64 Fig. 8-1b Graphical diagram of dynamic fibrous bed coalescer IV-64 Fig. 9-1a Icon of metal wool bed coalescer IV-69 Fig. 9-1b Graphical diagram of metal wool bed coalescer IV-69 Fig. 10-1a Icon of DAF IV-75 Fig. 10-1b Graphical diagram of DAF IV-75 Fig. 11-1a Icon of 2-phase hydrocyclone IV-79 Fig. 11-1b Graphical diagram of 2-phase hydrocyclone IV-79 Fig. 12-1a Icon of 3-phase hydrocyclone IV-83 Fig. 12-1b Graphical diagram of 3-phase hydrocyclone IV-83 Fig. 13-1a Icon of ultrafiltration IV-88 Fig. 13-1b Graphical diagram of ultrafiltration IV-88 Fig. 14-1a Icon of RO IV-91 Fig. 14-1b Graphical diagram of RO IV-91 Fig. 15-1a Icon of heteroazeotropic distillation IV-93 Fig. 15-1b Graphical diagram of heteroazeotropic distillation IV-93 Fig. 16-1a Icon of stripping IV-95 Fig. 16-1b Graphical diagram of stripping IV-95 Fig. 17-1a Icon of chemical destabilization IV-98 Fig. 17-1b Graphical diagram of chemical destabilization IV-98 Fig. 18-1a Icon of biological treatment IV-100 Fig. 18-1b Graphical diagram of biological treatment IV-100 Fig. 19-1a Icon of GAC filter IV-103 Fig. 19-1b Graphical diagram of GAC filter IV-103 Fig. 20-1a Icon of customized concentrator IV-106 Fig. 20-1b Graphical diagram of customized concentrator IV-106 Fig. 21-1a Icon of customized separator IV-108 Fig. 21-1b Graphical diagram of customized separator IV-108


IV-vi

Figure (Con’t)

Page

Fig. 22-1a Icon of inline concentrator IV-110 Fig. 22-1b Graphical diagram of inline concentrator IV-110 Fig. 23-1a Icon of inlet IV-111 Fig. 24-1a Icon of outlet IV-112 Fig. 25-1a Icon of flow merge IV-114 Fig. 26-1a Icon of flow split IV-115 Fig. 26-1b Graphical diagram of flow split IV-115


IV-1


In this part, we will use the data reviewed and verified in Part 1 and Part 2 as well as the text book composed in Part 3 as main theories to develop a computer program for design, calculation and simulation of oily wastewater treatment processes.

422

Chapter 1 Program overview

IV-2


1.1 Introduction

The final objective of this thesis is to develop a computer program for design, calculation and simulation of oily wastewater treatment process or process train, which includes every process reviewed in the previous three parts. The program is developed based on 4 major aims, i.e.,

• To provide background knowledge on oily wastewater treatment and working principle, design formula and design consideration on each oily wastewater treatment process.

• To provide preliminary guideline on oily wastewater treatment process selection to suit the wastewater to be treated.

• To calculate or size the oily wastewater treatment process under the design condition given by user, based on GPI researches under direction of Prof. AURELLE reviewed in Part 1 to 3.

• To simulate the oily wastewater treatment process train designed by user at any operating condition.

In this chapter, development procedure and details of the program in development stage are described. Details and user manual of the finished program will be described in chapter 2.

To develop the software under the scope of work of this thesis, the following procedure is used, i.e.,

1. Conceptual design of the program

The outline of software is preliminary designed to fulfil the 4 aims described above. Conceptual design of the program will be described section 1.2.

2. Review of related data

Related data can be divided into 2 parts, i.e. data on oily wastewater treatment and data relating to program development. The former is thoroughly reviewed in the first 3 part of this report. The latter will be described in section 1.3.

3. Program development and debugging

Detail of each step is summarized as follows.

1.2 Conceptual design of the program

To fulfil the aims described above, the program is divided into 4 major modes, i.e.,

• E-book mode: provides background knowledge and useful database about the oil pollution and the treatment processes,

423


IV-3

• Process recommendation mode: provides recommendation to narrow the range of feasible processes for any input influent,

• Design (or calculation) mode: used for sizing the process unit,

• Analysis mode: allows users to integrate any separation processes, included in the program database, to build their own treatment process train. And the program will simulate the process train to forecast the efficiency of each unit.

Framework of the program is as shown in the flowchart in fig. 1.2-1. The program is developed as a Windows-based program to make use of many standard and useful features of Windows and to facilitate link to other common Windows-based program, such as MS Office software package. It also operates in multi-tasking mode so users can toggle freely from one mode to another and can work on more than one project (meaningly, can open more than 1 window) at a time.

Conceptual design and features of each mode are described in the following sections.

1.2.1 E-book mode

E-book mode is one part of the program that is designed to provide knowledge and useful database about the oil pollution and the treatment processes. The program in this mode is designed to provide the following features,

1. Facilitate access or selection of the information. List of e-book files and their brief description about the context of the file is provided. After access to the file, detailed contents of each file are also provided to assure quick access to the required information.

2. E-book and help files are developed in general file formats, i.e. Acrobat and MS office software package, to take advantage of useful features of these software, such as browsing, word searching, printing, copy/paste capability, etc. Moreover, users are usually familiar with these file formats so they can readily handle the files.

3. Facilitate upgrading of the program to cover more processes in the future. The program is versatilely designed so it can be upgraded. List of e-book files is not fixedly coded but linked to a database that can be easily added or editing. Once that user adds new e-book file names, their short description text, and their location (path and directory), the program will automatically include them in the interface (screen) of e-book mode, ready to use.

The detailed flowchart of e-book mode is as shown in fig. 1.2.1-1.

424


IV-4

Proc

ess u

nit

libra

ry

Ope

n fil

e m

anag

emen

t w

indo

w

Inpu

t w

aste

wat

er

char

acte

ristic

Sele

ct sa

ved

file

Fig.

1.2

-1 F

low

cha

rt o

f the

pro

gram

und

er th

e sc

ope

of w

ork

of th

is th

esis

Star

t

Sele

ct m

ode

Ope

n E-

book

win

dow

Sele

ct

E-bo

ok

Dis

play

E-b

ook

(Acr

obat

)

E-bo

ok/H

elp

libra

ry

Ope

n R

ecom

men

datio

n w

indo

w

Inpu

t w

aste

wat

er

data

Rec

omm

end

appl

icab

le

proc

esse

s

Ope

n D

esig

nw

indo

w

Sele

cttre

atm

ent

proc

ess

Dis

play

ca

lcul

atio

n re

sult

Inpu

t re

quire

d da

ta

Ope

n A

naly

sis

win

dow

Cre

ate

or e

dit

pro

cess

trai

n

Ope

n fil

e

Ope

n ne

w

or se

lect

edfil

e

Inpu

t siz

e of

eac

h un

it p

roce

ss

Dis

play

ca

lcul

atio

n re

sult

Prog

ram

cal

cula

te

the

perf

orm

ance

of

pro

cess

esPr

int o

ut

Acr

obat

Ope

n M

S Ex

cel &

au

tom

atic

ally

cr

eate

wor

kboo

k

Con

tinue

Fi

nish

Prin

t/Exp

ort

Exis

ting

New

(def

ault)

All m

ode

Onl

y fo

r sim

ulat

ion

No

No

Yes

425


IV-5

Start

Select mode

Select E-book

tab

Select E-book by

double-click

Display E-book (Acrobat)

E-bookdata in

srachu.mdb

Print out by the software

Acrobat,Close the software

and return to E-book worksheet

Other E-book

Finish

Print/ExitPrint

Exit

No

Yes

E-book worksheet activated

Open the E-book fileby suitable software,

e.g. Acrobat

User can use all fuction of the software, e.g

search, copy, print, etc.E-bookfiles in

C:/../EBooks

Select other tab or quit the program

E-book

To other modes

Fig. 1.2.1-1 Detailed flowchart of e-book mode

1.2.2 Recommendation mode

Recommendation mode is one part of the program that is designed to provide recommendation to narrow the range of feasible processes for any input influent. The program in this mode is designed to provide the following features,

1. Facilitate data input. Interface of this part is designed to be interactive to prompt user to input all necessary data required for decision.

2. Provide help to guide the user through each input step. It is usually that the users do not know some required wastewater information, such as which categories of oily wastewater they have. Thus the interface of recommendation mode is provided with links to these data in e-book mode.

3. Provide recommendation about the feasible processes in user friendly form as a table with important precautions provided.

The detailed flowchart of recommendation mode is as shown in fig. 1.2.2-1.

426


IV-6

Start

Select mode

Select Recommend

tab

Input required data, follow on-screen

instruction

Display result as new window

Required parameters

data in srachu.mdb

Print out by the program

Finish

Print/ExitPrint

Exit

No

Yes

Recommend worksheet activated

Program performs logic calculation

E-bookfiles in

C:/../EBooks


Recommendation

To other modes

Click calculate botton to start

calculation

ContinueNeed help on some parameters

Click recommend

parameter / reference

Display E-book (acrobat, excel)

Acrobat,MS office

Open the E-book fileby suitable software, e.g. Acrobat, excel

Print/Exit

Recommend parameter subroutine

New/Load/save file subroutine

(fig. 1.2.4-3)


(fig. 1.2.4-3)

Fig. 1.2.2-1 Detailed flowchart of recommendation mode

1.2.3 Design mode

Design mode is one part of the program that is designed for sizing the process unit. The program in this mode is designed to provide the following features,

1. Calculate any required parameters of considering unit process when the rest data are provided. For example, users can provide the program with geometry of the process to find the efficiency or vice versa.

427


IV-7

2. Provide help to guide the user through each input step in case that some data is not known. So the interface of this mode also provides links to related data files.

3. This mode is provided with printing capability so users can print out the data and result as a hard copy. Moreover, the data can be saved for further use.

The detailed flowchart of design mode is as shown in fig. 1.2.3-1.

1.2.4 Analysis mode

Analysis of simulation mode is one part of the program that is designed to allow users to integrate any separation processes, included in the program database, to build their own treatment process train. And the program will simulate the process train to forecast the efficiency of each unit process. The program in this mode is designed to provide the following features,

1. Allow users to freely integrate any unit processes into a process train in the form of graphical schematic diagram.

2. The program will provide with user-friendly graphic editing interface with all common useful features, such as drag/drop, automatic snap-to-object connectors, basic drawing and text tools. This kind of interface is a kind of common feature in well-known software, such as MS PowerPoint, and other simulation software (e.g. Superpro, etc.). So it can be readily handle by users.

3. Provide help to guide the user through each input step in case that some data is not known. So the interface of this mode also provides links to related data files.

4. The program can calculate graded efficiency, total efficiency and other necessary components of every process, reviewed in Part 3, e.g. pressure drops for hydrocyclone, or saturator power requirement for DAF.

5. This mode is provided with printing capability so users can print out the data and result as a hard copy for further use.

6. The data and result can also be exported to MS Excel to use its efficient and useful features, such as formatting, chart building, etc., for advance display and printing of the data and the result. It also helps saving the time for developing these existing features.

7. The data and result can be saved for further reviewing in the future.

The detailed flowcharts of analysis mode, categorized by groups of operations, are as shown in fig. 1.2.4-1 to 1.2.4-3. They consist of,

• Data input operation

• Analysis and result display operation

• File management operation

428


IV-8

Start

Select mode

Select Design

tab

Input wastewater data, follow on-

screen instruction

Required parameters

data in srachu.mdb

Print out by the program

Finish

Print/ExitPrint

Exit

No

Yes

Designworksheet activated

Program perform calculation /send result

to Step 4 /


Design

To other modes

Click calculate botton/ step 4 tab

Edit/Continue

Need help on some parameters

Select step

Recommend paremeter subroutine

(see fig. 1.2.2-1)

select required process by click

at category and process

Step 1: WWdata input screen

activated

Step 2:select process screen

activated

Input required data, follow on-

screen instruction

Step 3:Inputprocess data

screen activated

Step 4 : Result screen

activated

Display resultparameters

for resultin

srachu.mdb

Edit

Edit

Edit


(fig. 1.2.4-3)

Yes

Built-in processdata in

srachu.mdb

A

A

No

Yes

NoYes

Help can beaccessed at alltime


(fig. 1.2.4-3)

Required parameters

data in srachu.mdb

Fig. 1.2.3-1 Detailed flowchart of design mode

429


IV-9

Start

Select mode

Select Analysis tab

Analysisworksheet activated

Analysis

To other modes

Need help on some parameters

Recommend paremeter subroutine

(see fig. 1.2.2-1)


(fig. 1.2.4-3)

A

A

Help can beaccessed at alltime

select required process by click at process name

Program loads the icon of the process

to the screen

move the mouse over the icon until the

“hand” cursor appears at the required position

More process

No

Yes

To connect the icons

Click mouse, hold down, move the mouse to another

required position

To edit connector or icon position

Edit

Click mouse to select the item

Click at the position on the connector to

be edited, hold down &move to new location

Double click at the icon

Display input data form

of the process

Input required data, follow on-

screen instruction

Required parameters

data in srachu.mdb

Select operationTo enter/ edit data of the process

No

Click OK to close the form

Yes

Click at required

drawing toolbutton

Click and hold down mouse at require position to place

drawing object (line, text, circle, etc.)

To add drawing object

To other operations

Icons in C:/../Unitp

rocess

Fig. 1.2.4-1 Detailed flowchart of analysis mode: Data input operation

430


IV-10

Click calculate button

Program sends results to result screen

No

Yes

To view resultin the program

Edit

No

Yes

To other operations

Formula inUnitProcessLib

Program validates the process train

Pass

Display warningmessage

Program calculatesthe result

Select operation

Double click at iconon the process train

Display input data form

of the process

Click at result tab

Display result formof the process

Click OK to close

To print by the program

Click print buttonor “Impremer”

in “Fichier” menu

Display print form

Click “Print”

Print out

Click OK to close

To export to Excel

Click “Exporter”in “Donnee”

Program open Excel andexport data and result

User can perform all Excel operation, even save the file in *.xls.Operation does notaffect the program

Close excel or click program

window to return

To edit data

Double click at the icon

To datainputfig. 1.2.4-1

Result parameters

fromsrachu.mbd

From Input dataFig. 1.2.4-1

Fig. 1.2.4-2 Detailed flowchart of analysis mode: Analysis and result display operation

431


IV-11

Start

Select operation

Select “Ouvrir”

Display file management

form

Program perform requested operation

To other operations

Select “Nouveau” Select “Supprimerde ficher”

Select “Enregistrer” Select “Fermer”

New file Select filename

Yes


form


form


form

Select filename Prompt to save

Program open new project

Program open the file

Click “Fichier” menu

No

Save

Select filename

No Yes

No

Fig. 1.2.4-3 Detailed flowchart of file management operation

1.3 Development tools

Related data for program development are reviewed. Tools and programming techniques are evaluated and carefully selected to suit the objective of program development. Evaluation criteria include,

• Capability of development tools to fulfil program features described in section 1.2,

• Developing time of each tool. Suitable tool should provide ready-to-use features required without extra code writing,

• Availability of the tools. The possible tools are firstly considered from the development packages that are readily available without additional procurement. Or they should be free-ware so it can be used without license problem.

From the criteria mentioned above, the program in this thesis is developed using the following development tools, i.e.,

1.3.1 Main development software package

The program is developed using Microsoft Visual Basic programming language, Version 6.0 (VB 6.0). The software is used under the permission of the software licensee, Progress Technology Consultants Co., Ltd. (Thailand) (PTC). The software features ready-to-use tools to create Windows-based application, such as builders of windows, forms and

432


IV-12

other user interfaces, e.g. input boxes, buttons, database link engines, etc, which fulfil every development requirement. It has event-driven feature, such as start working by mouse click or after enter the data, that is suitable for development of interactive graphic user interface.

1.3.2 Special graphic user interface (GUI) component

Graphic user component in analysis mode is developed on a special GUI component, FlowChartX version 3.2, the product of MindFusion limited. The component is actually an ActiveX component that can be used as an add-on of MS VB 6.0. It features presenting and producing of drag/drop graphic image, connectors, basic drawing tools and useful VB 6.0 compatible methods and properties. The software used in our program is free demo edition, which, somehow, provides sufficient features for our scope of work.

1.3.3 The third party software

Apart from its own VB program file, the program in this scope of work makes use of the third party software for their useful features, i.e.,

1. Microsoft Excel

Microsoft Excel is one of the most famous spreadsheet software with a lot of versatile and efficient features, e.g. calculation capability, chart builders, macro programming, searching, editing, formatting and printing tools. Thus in our program, recommend parameters, such as interfacial tension data, etc, are in excel file formats (*.xls) to facilitate searching or copying the data to the program. In analysis mode, the result can be exported to Excel to use its graph builder and formatting/printing facilities. Thus, to use the program of this thesis, MS Excel is required otherwise some help file will not be available and some results will not be exported. The program is compatible to MS excel 2000 or newer.

2. Adobe Acrobat reader

As described in section 1.2, e-book files are in acrobat format (*.pdf) to make use of its efficient searching and categorizing features. Other useful features are that it can operate in any platforms, e.g. Windows, Macintosh, etc., and it eliminates the problem about the font so the file is always readable on every computer. Acrobat reader is free ware so everybody can have a copy without purchasing. The e-book files are created by Adobe Acrobat professional version 7.0, under the permission of the software licensee, Progress Technology Consultants Co., Ltd. (Thailand) (PTC). To use the program of this thesis, Acrobat reader is required otherwise some e-book files will not be available. It is recommended to use Acrobat reader 6.0 or newer. The older version may be used but there may be a problem about using index and fonts, which makes some letters unreadable.

3. Microsoft Access

MS Access is used as open data source connectivity (ODBC). It allows the main program to connect to MS Access database file format (*.mdb), which is used as principle database of the program. To use the program, it requires only MS Access driver, which is usually provided with Windows OS. The program will automatically acquire the driver from the Windows in setup procedure of the program. However, if users want to change the database, such as addition of more process, or e-book files, they will need full MS Access program. The database is compatible to MS Access 2000 or newer.

433


IV-13

1.4 Program architecture

In this section, program architecture will be described to provide the idea about how the program is developed and relation between components of the program. This information is also helpful for further program development in the future.

Major program file is a Visual Basic program group named “prjThesis” (prjThesis.vbp). To fulfil the conceptual design described in section 1.2, program architecture of “prjThesis”, based on VB 6.0, is divided into following components, i.e. (see fig 1.4-1).

Fig. 1.4-1 Structure of the program

1.4.1 Forms

Forms are practically the graphic interface of the program. They will consists of many controls (such as buttons, menu, etc.), which, in turns, consists of many source codes to handle activity or event happening to the interface, such as button clicking. The program is divided into 3 major forms, i.e.,

1.4.1.1 Main form

Main form (frmain.frm) is the major form of the program that includes a workspace of the program as well as command button and the menu that control basis operation and file management operation. It is the first form that is open when the program is run. The interface of the form is as shown in fig. 1.4.1-1.

434


IV-14

Fig. 1.4.1-1 Main form (as seen in source code)

1.4.1.2 Project form

Project form (frDesigngrid.frm) is the form that is used as a workspace for operation of the program in every mode (recommendation, design, anlysis and e-book.) for one project. It is divided into 4 tabbed worksheets that represent 4 modes of operation. Users can freely toggle between sheets to work in each mode. The data keyed in one worksheet still remains while working in other worksheets. Data in each field in this form will be saved as encrypted data in the preset database file of the program. Each tabbed worksheet will contain controls and corresponding source code required for operation in each mode, as shown in fig. 1.4.1-2. The texts display in each fields are linked to a database file (srachu.mdb)

1.4.1.3 Input data form

Input data form (frInputdata.frm) is the form that is used for data input in analysis mode. Unlike the first two forms, it is normally not shown on the screen. The form will appear when a new unit process is inserted to the schematic diagram in the analysis mode or by double click at the process icon in the diagram. It is also used to display the result of each process in the analysis mode after the calculation is finished.

However, Other miscellaneous forms are also developed to fulfil purposes other than process calculation, such as frPrintMain.frm for “printing” task, frOpenDialog.frm for “Open file” task, etc.

435


IV-15

Fig. 1.4.1-2 Project form

Fig. 1.4.1-3 Input data form

436


IV-16

1.4.2 Modules

Modules are the source codes that govern the overall operation of the program, which is not linked to events. The major module is Main module (modMain.bas). It governs basic tasks of the program, such as starting the program, file path setting, etc.

There are also other supplementary modules, i.e. modDatefunctions.bas, modDBManager.bas and modUtil.bas. These modules are the customized source codes that govern basic file operations (such as open and save file, link to database) which are written by AWS Co., Ltd. (Thailand), system service provider of PTC. These modules are used under the permission of the company. The use of the modules allows us to save the time to develop the codes for these fundamental tasks.

1.4.3 User controls

User controls are special or add-in components besides standard components that come with standard VB development packages. In our case, one user control component’ called “ctlGrid”, is added to use in various input and output form in our program.

1.4.4 Class modules

Class module is special module which can be developed to have its own customized properties, e.g., in our case, “.isdroplet” (used to check if the process has its droplet data or not), etc. It greatly facilitates program development. In our program, class modules are used to control link of each type of data, i.e. process category name, unit process name and variables, etc., between the database file and the corresponding fields in the forms. For examples, it will link the file names and short descriptions of e-books file to display in grid table in the e-book form. Lists of class modules are shown in fig.1.4-1.

1.4.5 Add-in project

To facilitate upgrading of the program and versatile operation of the program, the sub-programs used for calculation of each unit process are separated as another Visual Basic project, namely “UnitProcessLib” (UnitProcessLib.vbp). It consists of module and class modules, i.e.,

1.4.5.1 Module

There is two modules in this add-in project. The first module is called “ModUNPFunction” (modUNPFunction.bas). It is used to manage operation in the analysis workspace, such as load the picture of selected unit process and run the tag number for each unit process. It can handle more addition unit process without any modification, in case that more processes are added in the future.

The second module is “ModExcelFunction”(ModExcelFunction.bas). It is used to control excel operation, such as calculation of some processes, import & export data between excel and our program.

1.4.5.2 Class modules

There are a lot of class modules in the add-in project. Each class module devotes to each unit process (see fig. 1.4-1). Thus, if more processes are to be added, all that have to do

437


IV-17

about coding is add the corresponding modules for those processes and add their variables and corresponding help or e-book file data in the database file “srachu.mdb”. The rest of the program needs no modification.

To develop new class modules for new processes, one of the old class modules can be used as a prototype. Developers just change the equation and solving procedure (such as iteration, etc.) to suit the new process and then add them into the add-in project. Details and equations used in calculation of each process are described in chapter 2.

1.5 Program development

After the conceptual design is set, all components and source codes of the program are developed in accordance with the flowcharts aforementioned.

When all components described above are completely developed, the program then undergoes debugging process to iron out any syntax errors and error from wrong coding.

After that, the Visual Basic source codes are complied to make a ready-to-use or executable file and setup file, using VB compiler and the third party VB setup file builder, called “Wise Installation system”. User will not see the components or source codes described above. They will see only a group of executable files and supplementary files. Details and user manual of the compiled program will be described in chapter 2.

The “setup” file, as well as the source code, of the program is submitted to Prof. AURELLE, thesis director. So it may be available upon request. For more information, please contact Prof. AURELLE or [email protected].

The copyright of the program, as well as the textbook shown in Part 3 of this thesis and as e-book files in the program, is registered with the Copyright office, Department of Intellectual Properties, Thailand and under international agreement on copyright (TRIPs).

438

Chapter 2 Program reference and user manual

IV-18


2.1 Overview of the program

The program developed under the scope of work of this thesis is the computer program that can calculate the size of processes and simulate or analyze the performance of process train, defined by users, without limitation on the number and configurations of the process. The program is intended to be upgradable to cover more process in the future. However, in this scope of work, the program covers only wastewater treatment processes, based on GPI’s researches under the supervision of Prof. AURELLE. Thus, it is named “GPI” program.

GPI program is developed in accordance with the conceptual design described in the previous chapter. Thus, GPI program can be divided into 4 main modes, i.e.,

• E-book mode: provides background knowledge and useful database about the oil pollution and the treatment processes,

• Process recommendation mode: provides recommendation to narrow the range of feasible processes for any input influent,

• Design (or calculation) mode: used for sizing the process unit,

• Analysis mode: allows users to integrate any separation processes, included in the program database, to build their own treatment process train. And the program will simulate the process train to forecast the efficiency of each unit.

However, from the user point of view, it is more convenient to describe each mode of program by the screens or graphic user interfaces (GUI). From this, the program can be divided into 2 major parts, i.e. main program and project window. Their relations can compare to those of the main Excel program and its workbook (which contains many worksheets). Details of each part of completely developed program are shown in section 2.1.1 to 2.1.4. They will describe the features of each part and show that how the conceptual design is transformed into real working program.

2.1.1 Main program

Main program is the main graphic user interface (see fig. 2.1.1-1) including main menu and a blank area that are used as workspaces for project window(s). The main program is developed from the common tasks of each mode, i.e.

• File management operation (open, save, close, etc.) of every mode,

• Basic edit tool (such as cut, copy, paste) for certain components,

• Print and export operation of every mode,

• Access to E-books and help that can be commonly accessed by any modes of program at any time.

Main components of this GUI is menu and tool bars. The menu, written in French language, provides common basic functions for all modes, i.e.,

439


IV-19

1. Fichier (File) menu: consists of the following functions or operations, i.e.,

• Nouveau (New): open new (blank) project window. User can use more than 1 project window independently at the same time. Change active project window from one to another can be done by click at any area on the new active window, like standard Windows-based softwares.

• Ouvrir (Open): open existing project, which can be save as encrypted data in database of the program.

• Fermer (Close): close active project.

• Enregistrer (Save): save active project as encrypted data in database of the program.

• Supprimmer de fichier (Delete file): delete unwanted project from the database.

• Imprimer (Print): print selected screen to printer in built-in preset format. User cannot alter the format. However, data in analysis mode can be exported (see menu “donneé’”) to Excel for advance presentation or report formatting.

• Quitter: exit the program.

2. Edition (Edit) menu: consists of the standard edit functions, i.e., resto (undo), redo (redo), couper (cut), coupier (copy) and coller (paste) for some applicable fields.

3. Donneé (Data) menu: consists of data transfer functions, i.e.,

• Exporter d’image (Export picture): export active process train diagram (only in Analysis mode) as a bitmap file, compatible with presentation or graphic editing software, such as PowerPoint, or PhotoShop.

• Exporter (Export): export active input data and result to MS Excel.

4. Aide (Help) menu: consists of help and information functions, i.e.,

• Utilisation du programme (help contents and index): open program help file, which is exactly brought from the context of section 2.2 in this chapter.

• Documentation électronique (E-book): open E-book mode. Clicking at E-book tab in a project window also gives the same operation.

• A propos de (About): open program information (names, version, etc.) dialog box.

440


IV-20

Tool barMain menu

Main program

Project window

Fig. 2.1.1-1 Graphic user interface of the main program

2.1.2 Project window

Project window is the window containing tabbed worksheets that contribute to each mode of program, i.e., recommend, design, analysis and e-book. User can open more than 1 project window at a time (see fig. 2.1.1-1). These windows operate separately. This part of program are combination of the 4 modes of the program as described in the flowcharts in section 1.2, except the parts that are separated to be a main program.

2.1.3 E-books worksheet

E-books worksheet is one of the tabbed worksheets in a project window. It is actually customized “open file” window that can show the list of built-in e-books or help files with their short description or synopsis of the files to inform the users about the content of each file. Double clicking at the file name or clicking at open button after file name selection will

441


IV-21

open the selected e-book (see fig. 2.1.3-1). E-book worksheet is developed under the conceptual design, shown in fig. 1.2.1-1.

The help or e-book file names and short descriptions are stored in a database, namely “srachu.mdb”. If there are more help files or e-books to be added into the program. Users can simply input the filenames and their short description (as normal text data) into the database file, using MS Access, and copy the new help file or e-books into the existing “Ebooks” directory.

Overview of oily wastewater treatment

DecanterSTOKE’s law, Interfacial tension, Capillary API, PPI, Spiraloil,

Fig. 2.1.3-1 Graphic user interface of the E-book worksheet

2.1.4 Recommend worksheet

Recommend worksheet is one of the tabbed worksheets in project window (fig. 2.1.4-1). It can provide recommendation about possible or potential processes (based on guideline summarized from GPI’s researches in Part 3, chapter12), corresponding to input wastewater data. Recommend worksheet is developed under the conceptual design shown in fig. 1.2.2-1.

Users just answer the preset questions to provide the closest description to the wastewater to be treated. The questions start at the topmost frame, while the rest frames are still disabled to avoid confusion. The frame corresponding to the previous answer will be shown next. Fig. 2.1.4-2 is the retouched screen to show all frames at the same time. After clicking calculation button, the program will display the result screen (fig. 2.1.4-2) and put tick mark ( ) in the first column to indicate the categories of the wastewater to be treated.

This mode of operation is designed for oily wastewater treatment process only since decision logic is relatively complex so it needs fixed source code. The logic used to recommend feasible processes is as shown in flowchart (fig. 2.1.4-3).

Thus, to upgrade the program to cover other process categories (such as air pollution treatments) in the future, it can be done by addition of checklists in the form of help files or e-books in e-book worksheet instead. It may not be interactive but it would be equally effective.

442


IV-22

Fig. 2.1.4-1 Input screen ( one of tabbed worksheets in project window)

Fig. 2.1.4-2 Result screen. The program will put tick marks in the first column.

443


IV-23

Finish

By source/granulometry

Categories

Start

Select categories

Source / granulometry

Display result as new window

Program puts tick mark in

corresponding cells

d>100μ D=1-100μ

D<1μ Oil layer

WW contains oil layer.

WW contains primary emulsion.

WW contains secondary emulsion.

WW contains macro/micro emulsion.

Oil/HC, sheared by pump

Storm water/ condensate

Wash water w/surfactants

Oily WWw/o surfactants

Slop/ retentate/ conc. oily WW

Cutting oil/ stabilized emulsion

Oil lyer + Primary +secondary emulsion.



Primary +secondary emulsion.

Slop/ retentate/ conc. oily WW

Macro/microemulsion

KnowncategoriesJu

st sh

ow

the

guid

elin

e

By source

By granulometry

Fig. 2.1.4-3 Flowchart of recommendation logic

2.1.5 Design worksheet

Design worksheet is one of the tabbed worksheets in project window. It is used to calculate any parameter providing that the remaining parameters are given, e.g. find efficiency from the given size of process at the given wastewater data. Thus its operation can compare to programmable calculator. Design mode, in this thesis, is developed for oily wastewater treatment processes and can be expandable to cover other types of processes in the future (by others). Design mode worksheet is also divided into 4 tabbed worksheets, devoting to each design step, i.e.,

1. Step 1: input wastewater properties. (fig. 2.1.5-1a) Required data includes influent parameters on oily wastewater treatment process, i.e, oil concentration, physical properties, such as densities and viscosities of oil and water, as well as size distribution data (granulometry). In case that some parameters are not known, user can consult built-in parameter database, by clicking “recommend parameters” button. The database is in MS Excel format, to facilitate additional, editting and cut/paste operation. The files are stored in directory “RefFiles”.

2. Step 2: select the process to be calculated (fig. 2.1.5-1b). Oily wastewater treatment processes included in the program are based on every process reviewed and described in Part 3, i.e.,

• Skimmers • Decanters • Coalescers • Dissolved air flotation • Hydrocyclones

• Membrane processes • Heteroazeotropic distillation • Chemical process

(Destabilization)

444


IV-24

Equations and guide on data input are described in chapter 3. The program is versatilely designed so it can be upgraded to include more process categories, such as air pollution treatments, various biological treatments, etc., in the future. Details will be described in chapter 3.

3. Step 3: specify the parameter to be calculated and input the rest parameters (fig. 2.1.5-1c).

4. Step 4: display the result (fig. 2.1.5-1d). The result will be displayed after user clicks at the “calculate” button. If the data is changed, the result will be cleared to avoid confusion or using the wrong set of result.

If the design worksheet is activated (tab “design” is clicked), the data and the result can be printed in the preset format by selecting “Imprimer” in “Fichier” menu. Recommend worksheet is developed under the conceptual design shown in fig. 1.2.3-1.

Step 1 “Wastewater data input” Step 2 “Process selection”

Step 3 “Process data input” Step 4 “Result”

Fig. 2.1.5-1 User interfaces of design worksheet

2.1.6 Analysis mode

Analysis mode is one of the tabbed worksheets in project window. It provides users with capability to analysis the performance of process trains of any combinations of unit processes. The process train will be represented by graphical schematic diagram as shown in

445


IV-25

fig. 2.1.6-1. User can select the process category in the top left list box. Then, the program will display list of available process in that category in the lower list box. In this thesis, there is only one category available, namely “oily wastewater treatment”.

User can insert required process by double clicking at the name of the process. Then, the program will insert the icon (picture) of the process into the graphic-editing screen. The icons can be move freely by normal drag/drop method. Each icon will be automatically labeled by a code, indicating its type of process and number, such as SD-01 for simple decanter no. 1. These tag numbers are useful for calculation of the program.

The program is equipped with automatic connector capability that can connect the processes just by mouse clicking and dragging. Basic graphic tools, such as text box, line, rectangular, circle, are provided. Users can add/edit these items into the diagram without affect on calculation of the program.

Sizing of each process is specified by double clicking at the icon. Data input screen will be displayed. After the schematic diagram and data input are ready, user can start calculation by clicking at “calculation button” or “calcul” menu. The program will validate the data and prompt for correction in case of error until it is cleared. Then it will analyze the process to find efficiency of the process as well as other important parameters of each process, such as pressure drop for hydrocyclone, etc. The details will be described in chapter 3. The result will be displayed within the program or exported to Excel. User instructions on drawing the diagram, calculation, etc. are described in section 2.5.

Graphic editing area

Basic drawing tool barCalculation button

Input and result screen(displayed when double clicking at the icon)

Category selection

Process selection

Fig. 2.1.6-1 User interface of analysis worksheet

446


IV-26

2.1.7 Warning dialog box

Warning dialog box is the text box that is displayed when the program finds some error, e.g. in data typing, or wants to prompt users on some events or information, e.g. SD-01: specified droplets are beyond limit of the process. Example of warning dialog box is as shown in fig. 2.1.7-1. It can be closed by clicking OK or close (X) button at its corner.

Fig. 2.1.7-1 Example of warning dialog box

2.2 Program capability

Capability of program, from the point of view of process calculation, can be divided into 2 parts, in accordance with 2 main modes of the program, i.e. design and analysis mode.

The processes included in the program are mainly the treatment processes that are reviewed in Part 3. However, there are some additional processes that are included as logical module in the analysis mode, e.g. inlet module and outlet module, or to represent a customized process that is not included in those of Part 3. All processes included in design and analysis mode are summarized in table 2.2-1. Equations and related parameters, as well as guideline for data input, for each process are described in chapter 3 “Program references”.

2.3 Program limitation

Since the program is newly developed, it has some limitations, based on programming techniques, limitation from scope of work of the thesis, as well as attempts to generalize the program to support upgrading in the future. General limitation of the program are as summarized below.

1) In this scope of work, GPI program includes only oily wastewater treatment process.

2) The unit of every parameter (e.g. kg, m. second, etc.) is fixed to lessen the complication of the program.

3) The program includes only calculation of oil separation. Other pollutants, such as suspended solids, are not taken into account. However, from its open architecture, it can be upgraded to cover other treatment or other type of process in the future.

4) The process train in the program can contain only 1 outlet since it is an important logical module, used in program loop. In case that there is more than 1 outlet in the real process train, user can combine those outlets, using flow merge module, into 1 outlet. However, to do so, it does not affect calculation result.

447


IV-27

Table 2.2-1 The processes included in the GPI program

Name Design mode Analysis mode

Skimmer 1 Drum skimmer Yes Yes 2 Disk skimmer Yes Yes Decanter 3 Simple decanter Yes Yes 4 Compact decanter Yes Yes 5 Customized decanter Yes Yes Coalescer 6 Granular bed coalescer Yes Yes 7 Brush type bed coalescer Yes Yes 8 Dynamic fibrous bed coalescer Yes Yes 9 Metal wool bed coalescer Yes Yes Dissolved air flotation

10 Dissolved air flotation Yes Yes Hydrocyclone

11 2-phase hydrocyclone Yes Yes 12 3-phase hydrocyclone Yes Yes Membrane processes

13 Ultrafiltration Yes Yes 14 Reverse osmosis Yes Yes Thermal processes

15 Heteroazeotropic distillation Yes Yes 16 Stripping Yes Yes Chemical process

17 Chemical destabilization Yes Yes Finishing processes

18 Biological process Yes Yes 19 GAC filter Yes Yes Customized process

20 Customized oil concentrator Yes 21 Customized oil separator Yes 22 Inline oil concentrator Yes Logical module

23 Inlet Yes 24 Outlet Yes 25 Flow merge Yes 26 Flow split Yes

448


IV-28

5) Some influent parameters can not be interpreted in the form of equation, such as tortousity of bed in coalescer process, thus can not be accounted in calculation. However, effect of these parameters can be compensated by the use of correction factor (CF).

6) The program requires third party program to operate, such as MS excel, Acrobat. Some of these programs are not free-ware and may cause some expense if they are not readily available.

7) From the limitation of VB 6.0, its control components, such as text box, menu, table, etc., support only one font, which is Windows (English version) default font (as seen in the windows explorer and menu of every program in the Windows). So we cannot use symbols, Greek letters or the letters with accent, such as γ, τ, é, etc. in the program.

2.4 System requirement

To use the program, these hardware and software, based mainly on the requirement of Windows, VB and Excel, are needed.

Hardware

1. Computer and processor: PC with 300 megahertz or higher processor clock speed recommended; 233 MHz minimum required (single or dual processor system);* Intel Pentium/Celeron family, or AMD K6/Athlon/Duron family, or compatible processor recommended.

2. Memory: 128 MB of RAM or greater (minimum of 256 MB is recommended.) 3. Hard disk: 50 MB of available hard-disk space 4. Display: Super VGA (800x600) or higher resolution 5. CD-ROM or DVD drive

Software

1. Windows 2000 with service pack 3 (SP3), Windows XP or higher 2. Adobe Acrobat reader 6.0 or higher 3. Microsoft Excel 2000 or higher 4. Microsoft Access 2000 or higher (Optional for database editing)

2.5 User instruction

GPI program is Windows-based program so it features familiar graphic user interface and basic Windows function, e.g. drag/drop, etc. Thus it is quite ready to use for Windows users. In this section, instructions for using customized features of the program are described

2.5.1 Program Installation

Setup files of the GPI program are contained in a CD. To install program;

1. Insert CD into a drive

2. The program will run setup program automatically. If auto run is not operated,

449


IV-29

2.1 Use any browsing programs, such as window explorer, to browse the contents on the CD and then <double click> at “Setup.exe”.

2.2 <Click> Windows’s “Start” button, <Click> “Run” Then type “<your CD- ROM drive>:setup”, such as “d:setup”

3. Follow on-screen instruction.

4. Set program will install GPI program to the computer and automatically create an icon on the desktop and add the program into the Programs menu of the Windows.

2.5.2 Starting the program 1. <click> Windows’s “Start” button, then <click> “Programs” and <Click> the

program name.

2. Or <double click> at the shortcut

on the desktop (created by user).

3. Or <click> Windows’s “Start” button, then <click> “Programs” and <Click> the program name.

4. The Main form and a new project window will appear on the screen. The program is ready to use.

2.5.3 Using E-book mode

Fig 2.5.3-1 Using E-book mode

13

2, 4

1. <Click> “E-book” tab or “documentation électronique” in “Aide” menu.

2. <Click> at the name of an e-book to be open

3. <Click> “Open” button

4. Or <Double click> at the name

5. The program will open Acrobat and display the selected file.

2.5.4 Using Recommend mode

1. <Click> “Recommend” tab

2. <Click> a radio button to select the suitable answer for your wastewater in the topmost frame

450


IV-30

Fig. 2.5.4-1 Using recommend mode: Input data screen

1

3

2

4

5

Fig. 2.5.4-2 Using recommend mode: Result window

451


IV-31

3. The frame corresponding to your first answer will be displayed. <Click> another radio button. Other frames will be disabled to avoid confusion.

4. <Click> suitable list box. You can select more than one box.

5. When data input is sufficient, “Calculate” button will be enabled. <Click> the button to start calculation and display the result.

6. Result window (fig.2.5.4-2) is shown. Your wastewater is indicated by tick marks. <Click> “X” button to close the result. To start over, go back to step 1.

2.5.5 Using Design mode

1. <Click> “Design” tab

2. <Click> “Step 1” tab. Wastewater data input screen is shown.

3. Input the data in input fields (text boxes and table) by <Click> at the selected field, then type your data

4. To move to next field, user arrow keys (<↓>,<↑>) or <enter> or <click> at the next field

5. When data in step 1 is complete, <Click> “Step 2” tab or <click> “next” button. Process selection data is shown.

Fig. 2.5.5-1 Using Design mode: Step 1 “Wastewater data input”

3

2

1

54

6. <Click> at required category name and process name. The text boxes on the left hand will display short description of the selected category and process. In this thesis, these are only 1 category, i.e. oily wastewater treatment.

452


IV-32

Fig. 2.5.5-2 Using Design mode: Step 2 “Process selection”

1

6 6

6

7

Fig. 2.5.5-3 Using Design mode: Step 3 “Process data input”

7

8

9

453


IV-33

7. When selection is done, <click> “Next” button or “Step 3” tab. Process data input screen is shown.

8. Input the data in input fields (text boxes and table) by <Click> at the selected field, then type your data. To move to next field, user arrow keys (<↓>,<↑>) or <enter> or <click> at the next field. The field that we want to find the answer must be left blank. <Click> “Recommend parameters” or “Reference” for more information on the value of each parameter and related equations and limitation of each process.

9. When data in step 3 is complete, <Click> “Calculate” button. The program will calculate the result and put it in the result screen “Step 4”. If the data is not sufficient or redundant, the program will not calculate and will display a warning.

10. After calculation, the result screen (step 4) will be automatically shown.

11. To edit some data in each step, <click> the required tab or <click> “Back” and “Next” button. When the data is edited, the result is cleared to avoid confusion.

Fig. 2.5.5-4 Using Design mode: Step 4 “Result”

10

11

2.5.6 Using Analysis mode

1. <Click> “Analysis” tab

2. <Click> at require category name in the upper left box. List of processes in that category will be shown in the lower left box. In this thesis, there is only 1 category, i.e. oily wastewater treatment.

454


IV-34

1 5

Drag and drop to move

<Double click> to insert

52

4

3

Fig. 2.5.6-1 Using Analysis mode: Draw the schematic diagram

3. Input your process train by drawing the schematic diagram, using the unit processes provided in the lower left box on the screen. To insert an icon or graphic representative of the process, <Double click> at the process name. The corresponding icon is shown in the graphic editing area.

4. To move the icon to proper location, move the cursor over the icon until it turns

into a “4-direction arrow” ( ) sign. Then <click> and hold the mouse button to drag and release the button to drop the icon at new location.

5. Insert other processes using step 3 and 4 until all required processes are inserted. To connect the processes together, move the cursor over the icon until it turns into a “hand” ( ) sign. Connecting points of the process will be shown as small red squares. Move the “hand” to the required connecting point, <click> the mouse, hold the button to a required connecting point of another process to be connected and release the mouse. The program will draw the connecting line automatically. Users will actually see the line while they hold and move the mouse. User can add non-active graphic, e.g. text box, line, etc. into the diagram, using drawing tools.

6. After connecting the processes, User must input the size of each process. To input or edit the data of a process, <double click> at its icon. Process data input/ result screen is shown. However, the result tab will be disabled unless the calculation is completed. The process train must contain only 1 outlet module. But there can be several inlet modules. See chapter 3 for details and restriction of each module.

455


IV-35

Fig. 2.5.6-2 Using Analysis mode: Process data input

7

10

6

7. After data input procedure is completed. <Click> “Calculate” button or “Calcul Now” in “Calcul” menu to start caluculation. The program will validate the data and display warning to prompt users to correct the errors. After the errors are cleared, the program will calculate the result of each process. In analysis mode, the program can calculate process performance (e.g. graded efficiency, outlet concentration of oil, etc.) only. It can not be used for sizing the process, as in the design mode.

8. To view the result, <Double click> at required process icon. The data input/ result will be displayed. After calculation in step 8, the “Result” tab is enabled. <Click> the tab to view the result of the process. The result tab is just the same as that of design mode, shown in fig. 2.5.5.4.

9. Total outlet concentration of each process will also be shown as a text box under each icon for evaluation of the performance at-a-gaze (fig. 2.5.6-3)

10. To display result in more complex format, it must be exported to Excel. To do so, <Click> “Exporter” in “Donneé” menu. The program will open Excel and automatically create a workbook. Each worksheet in the workbook devotes to result of each process (fig 2.5.6-4). User can also save this excel file. User can formulate their own formats of result such as building graph showing graded efficiency of every process, etc., in Excel. The diagram can also be exported to any Windows-compatible software, such as PowerPoint, PhotoShop, etc., by <click> at “Exporter d’image” in “Donneé” menu.

456


IV-36

Fig. 2.5.6-3 Using Analysis mode: Viewing result in the diagram

Fig. 2.5.6-4 Using Analysis mode: Exporting the result to excel

2.5.7 Printing and file operation

1. Every screen in the program can be printed in pre-set format. To print, <click> “Imprimer” in “Fichier” menu. Print dialog box will be shown. User can select the mode(s) that they want to prink, then click “OK”.

2. Data and result of design and analysis can be saved in the database. To conduct file operation (new, open, save, delete, close), <click> required operation in “Fichier” menu. File operation dialog box will be shown.

457


IV-37

2.6 Upgrading procedure and recommendation for further development

As described in previous section that the program is designed to facilitate upgrading in the future, in this section, upgrading procedure and recommendation for further development are described.

2.6.1 Upgrading procedure

Upgrading procedure can be divided into 2 major parts, i.e. upgrading of database and source code upgrading.

2.6.1.1 Upgrading of the database

Every data related to process calculation, as listed below, is linked to the database, namely “srach.mdb”, which is automatically installed in setup procedure.

1. Process categories (see fig. 2.5.5-2, and 2.5.6-1)

2. Unit processes of each process category, as shown in table 2.2-1 and sub-routine names (called “class” as described the previous chapter)

3. Wastewater input data for each process category (see fig. 2.5.5-1 and 2.5.6-2)

4. Input data for each unit process, as well as its icon file and help file (see fig. 2.5.5-3 and 2.5.6-2)

5. Output data for each process (see fig. 2.5.4-4 and chapter 3)

6. E-book (see fig. 2.5.3-1)

These data are in the form of tables in the database file, as shown in fig. 2.6.1-1. If users want to add new process or process categories, it can be easily done by add related data into the tables. The additional data will be automatically displayed in the related screens of the program without any additional coding, as shown in fig. 2.6.1-2. This procedure does not require Visual Basic development package.

Process category:

Table name:tbProcessCategory

Unit process:

Table name:tbUnitProcess

Input/output datafor oily wastewater

category :

Table name:tbVariableDef

Standard result(such as Cout, etc. :

Table name:tbStdResult

Unit process:

Input data for otherwastewatercategory :

Standard result forthat category

For future upgradingExisting

Ebook file:

Table name:tbEBook

Fig. 2.6.1-1a Configuration of the database

458


IV-38

Fig. 2.1.6-1b Examples of existing database

CAT000001CAT000002 Air pollution treatment

Air pollution treatment

Fig. 2.1.6-2 Example on database upgrading. When new category is added in the database (left), Related field in the program will be automatically updated (right)

2.6.1.2 Source code updating

Even though the program’s interfaces can be easily upgraded by updating the database file. To add sub-routines (or classes) of new processes into the program, the new classes have to be written in Visual Basic. The simplest way to write a new class is to copy one of the old class and change the formula or equation to suit the new process. To do so, Visual basic development package and the original source code of any process is required. Developers can define their own variable easily in the database file. So, in their new classes, they can use these variables in the new equations.

459


IV-39

After writing the new classes in VB, the classes need to be recompiled into executable file. Then, the file will be copied to replace the old compiled class file without reinstallation of the whole program.

The reason that we developed the program in the manner that it requires source code for updating is that we can control the development of the program to proper developers. The source code of the program is submitted to thesis director, Prof. AURELLE.

2.6.2 Recommendation for further development

Since this program in newly developed so it have some limitations as discussed in section 2.3. However it may have prospect for further development, at least, in the following approach.

1. To upgrade to include more process categories,

2. To add a subroutine for unit conversion (metric, US customary, etc.), which would be very useful in case that some users may be familiar with different kind of unit.

3. To develop or include some advanced components, which may be present in the newer version of VB, to enhance the performance of the program or make it operate in stand-alone mode without requiring third-party software (such as Excel). It is also useful to enhance its text display capability so it will be able to display Greek letter, subscript, or superscript.

4. Recommend mode should be enhanced by replacing the fixed-coded source codes with sub-program written with logical programming language, such as Prolog.

460

Chapter 3 Process references

IV-40


Process references, which include the parameter to be calculated, equations used and specific constraints of each process, are described in the following sections. This information is included in the help file of the program, (<Click> “Utilisation du program” in “Aide” menu), or (<Click> “Reference” button in any input forms or user interfaces).

Common parameters

Common parameters or variables that are used for every process module are summarized as follows,

Parameter Description Variable name in the program

Unit used in the program

Cod Granulometry or oil droplet size distribution : graded concentration (concentration of oil at each droplet size)

Cind mg/l

Concentration of inlet oil in the form of oil layer or film

Cinlayer mg/l

Co Total oil inlet concentration Co mg/l d Granulometry or oil droplet size

distribution : droplet size din micron

Q Wastewater flowrate Qin m3/h T Temperature temp Celcius ρc Dynamic (or absolute) viscosity of

continuous phase, which is water, for oily wastewater

Denc Kg/m3

ρd Density of dispersed phase, in this case, oil Dend Kg/m3 μC Dynamic (or absolute) viscosity of

continuous phase, which is water, for oily wastewater

Muc N.s/m2

(= 1000 cp)

μd Dynamic (Absolute) viscosity of dispersed phase, in this case, oil

Mud N.s/m2

(= 1000 cp) γow Interfacial tension between oil and water gow kg/s2 or N/m

(= 1000 dyne/cm)

Values of these parameters will be given by user via inlet module (See section 23). They will be used to calculate the results, which are usually divided into 2 streams, i.e., water outlet and oil (or concentrated oily wastewater outlet. Generally, the result will contain the parameters mentioned above. But the values will be changed in accordance with corresponding model or equations of each process. Then they will be sent as input data for downstream process until the end of process train (outlet module).

General remarks and precautions of the program

General remarks and precautions of the program are as shown below,

• The program is valid only when the oil is lighter than water.

461


IV-41

• The processes available in the program generally consists of 1 inlet port and 2 outlet ports, i.e., outlet port of treated water, called “water outlet port”, and outlet port of separated pure oil or concentrated oily water, called “Oil outlet port”.

• Graded concentration is specified as quantity of oil (mg) per volume of wastewater (= volume of oil+water).

• Graded efficiency (ηd) in the program is based solely oil mass removal, regardless of flow splitting effect. To calculate outlet concentration, flow splitting between oil outlet port and water outlet port of each process will be taken in to account. Thus, in some processes, outlet oil concentration is not exactly equal to (1-ηd)Cod.

• Total efficiency (ηt) in the program can be generally used to calculate total inlet concentration. Total outlet oil concentration is equal to (1-ηt)Co in some processes. However, there may be some exceptions in certain processes. See reference of each process for more information.

• The pictures or icons of processes in analysis mode generally contain 3 connections or hot spots (there may be only 1 or 2 spots in some processes.) The black connection represents inlet point of the process. The red one represents oil outlet port and the blue one represents treated water outlet port.

• In E-book mode, the e-book files are designed for a fixed directory, named c:/SR/E-book. The directory contains an Acrobat index file “E_book_oily_wastewater”. This index facilitates users to search for any words or topics in all e-book files in the directory. However, it is developed by Acrobat 7.0, it may not be run on some old versions of Acrobat reader. In this case, it is recommend to set option of the option “All PDF documents in” in Acrobat’s search window to the e-book directory. It can substitute the use of the index file.

• The program is linked to MS Excel for parameter recommendation, calculation of some processes and for result display. In case that, Excel security is set to “high level”, it will happen that the program is not allowed to link to Excel. To solve or avoid this problem, you can choose one of these options,

• Set security of Excel to medium or low. (in Menu “Tools”>Options>Security).

• If high or highest security level is needed, try setting the macro in the file “Process_calculation.xls” as a trusted source (in Menu “Tools”>Options>Security). Consult your corresponding Excel’s help file for more information.

462


IV-42

1) Drum skimmer

1. Process abbreviation: DRM

2. Process description : Oleophilic oil drum skimmer

3. Reference : Part 3, chapter 3

4. List of outputs and related equations

4.1 Graded efficiency: 100% for oil film or layer and 0% for droplet.

4.2 Total efficiency: 100% for oil film or layer and 0% for droplet.

4.3 Graded outlet oil concentration in water outlet flow: This parameter is not valid for this process.

4.4 Total outlet oil concentration in water outlet flow: This parameter is not valid for this process.

4.5 Graded outlet oil concentration in oil outlet flow: the outlet oil in the oil outlet flow is in pure condition (100% oil).

4.6 Total outlet oil concentration in oil outlet flow: the outlet oil in the oil outlet flow is in pure condition (100% oil).

4.7 Inlet flow: the process is valid only for separated oil flow from upstream process.

4.8 Water outlet flow: This process does not have this property.

4.9 Oil outlet flow: flowrate or oil productivity (m3/h) is governed by the following equation.

0.514g

L0.486oν

1.541N1.5413.035DP ⋅= CF {1.1}

5. Related parameters: Related parameters are as summarized in the following table.

Table .1-1 Related parameters of drum skimmer



CF Efficiency correction factor, CF ≤ 1 CF D Diameter of skimmer D M g Gravitational acceleration Constant = 9.81 m/s2 L Length of the skimmer L M N Rotational speed of the skimmer N Rev/s P Oil productivity of the skimmer Prod m3/h γo Superficial tension of oil go kg/s2 or N/m

(= 1000 dyne/cm) νo Kinematics viscosity of oil (used μ/ρ) m2/s

463


IV-43

6. Related graphics: Icon in analysis mode and graphic diagram for input screen for this process are as shown below.

Drum Skimmer (DRM-??)

Scrapper

Oil layer

D

L

Fig. 1-1a Icon of drum skimmer Fig .1-1b Graphical diagram of drum skimmer

7. Constraints and limitations

7.1 Superficial tension of oil is in the range of 27 – 34 dynes/cm, which practically covers all common oil.

7.2 Capillary number (Ca = μo V/γo) is in the range of 0.2 – 1.0.

7.3 Oil density is around 790 – 830 kg/m3. Oil dynamic viscosity (μ) tested is between 1.35x10-3 to 291x10-3 (N.s)/m2 (1.35-291 cp).

7.4 Peripheral or tip velocity should not be greater than 0.8 m/s. To avoid water entraining, velocity of 0.44 m/s or less is recommended.

7.5 Recommended minimum immersion depth is 1.0-2.0 cm.

7.6 Drum skimmer surface used for model development is polypropylene. But it is proven to be valid for SS, PVC, and PTFE.

7.7 Skimmer can only be connected to pure oil outlet flow of upstream process.

8. Influent parameters: Effects of certain parameters, when they are increased, on the oil removal performance of the process are summarized as follows. For more details, see chapter 3, Part 3.

Table 1-2 Influent parameters for drum skimmer

Parameter Effect on process performance if the parameter is increased

Length of drum + is Diameter of skimmer + is Oil viscosity + Velocity ± Present of surfactants -

Note: “+” means “increases performance”, “-“ means “decreases performance”, “±” means “have both positive and adverse affects”, “is” means “effect is insignificant”.

464


IV-44

2) Disk skimmer 1. Process abbreviation: DSK

2. Process description : Oleophilic oil disk skimmer



4.1 Graded efficiency: 100% for oil film or layer and 0% for droplet. 4.2 Total efficiency: 100% for oil film or layer and 0% for droplet. 4.3 Graded outlet oil concentration in water outlet flow (Cd): This parameter is

not valid for this process. 4.4 Total outlet oil concentration in water outlet flow (C): This parameter is

not valid for this process. 4.5 Graded outlet oil concentration in oil outlet flow (Coild): The outlet oil in

the oil outlet flow is in pure condition (100% oil). 4.6 Total outlet oil concentration in oil outlet flow(Coil): The outlet oil in the oil

outlet flow is in pure condition (100% oil). 4.7 Inlet flow: the process is valid only for separated oil flow from upstream

process. 4.8 Water outlet flow: This process does not have this property. 4.9 Oil outlet flow: flowrate or oil productivity (m3/h) is governed by the

following equation.

0.332g

1.17I0.452oν

1.212N1.2581.328DP ⋅⋅= nCF {2.1}

5. Related parameters: Related parameters are as summarized in the following table.

Table 2-1 Related parameters of disk skimmer



CF Efficiency correction factor, CF ≤ 1 CF D Diameter of skimmer D m g Gravitational acceleration Constant = 9.81 m/s2 I Immersion depth of the skimmer, I ≤ D/2. I m n The number of disk num N Rotational speed of the skimmer N Rev/s P Oil productivity of two sides of the

skimmer Prod m3/h

γo Superficial tension of oil go kg/s2 or N/m (= 1000 dyne/cm)

νo Kinematics viscosity of oil (used μ/ρ) m2/s

465


IV-45


Disk

Disk Skimmer (DSK-??)

I

ScrapperD

Fig. 2-1a Icon of disk skimmer Fig. 2-1b Graphical diagram of disk skimmer

7. Constraints and limitations.

7.1 Superficial tension of oil is in the range of 27 – 34 dynes/cm, which practically covers all common oil.

7.2 Capillary number (Ca = μo V/γo) is in the range of 0.04 – 3.6. 7.3 Oil density is around 790 – 830 kg/m3. Oil dynamic viscosity (μ) tested is

between 1.35x10-3 to 291x10-3 (N.s)/m2 (1.35-291 cp). 7.4 Peripheral or tip velocity should not be greater than 1.13 m/s. To avoid

water entraining, velocity of 0.5 m/s or less is recommended. 7.5 Disc skimmer surface used for model development is PVC. But it is proven

to be valid for SS, PP, and PTFE. 7.6 Skimmer can only be connected to pure oil outlet flow of the upstream

process.


Table 2-2 Influent parameters for disk skimmer


Immersion depth + Diameter of skimmer + is Oil viscosity + Velocity ± Present of surfactants -


466


IV-46

3) Simple decanter

1. Process abbreviation: SD

2. Process description: API tank or any decanter that does not have any oil interceptors installed in the tank so the rising distance of oil drops is equal to the water depth.



4.1 Graded efficiency (ηd): Graded efficiency of the process is governed by the following equation.


%100⋅= CFdη {3.1}


%100⋅⋅=dc

dd U

UCFη {3.2a}

For oil droplet size, d ≤ 20 microns

%0=dη {3.2b}

Rising velocity (U), m/s, of the droplet “d” or of the cut size “dc” is calculated from the following equations.

cd

dgU

μρ

18)10( 26−⋅⋅⋅Δ

= {3.3}

⎟⎠⎞

⎜⎝⎛

⋅=

SQUdc 3600

{3.4}

Cut size is calculated by the following equation.

62/1

103600.

18⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛⋅⋅Δ

=Sg

Qd c

c ρμ micron {3.5}

4.2 Total efficiency (ηt):

( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {3.6}

4.3 Graded outlet oil concentration in water outlet flow (Cd):

)1( dodout

d CQQC η−⋅⋅= mg/l {3.7}

4.4 Total outlet oil concentration in water outlet flow (C ):

∑=max

min

d

ddCC mg/l {3.8}

4.5 Graded outlet oil concentration in oil outlet flow (Coild): The outlet oil in the oil outlet flow is in pure condition (100% oil).

467


IV-47

4.6 Total outlet oil concentration in oil outlet flow (Coil): The outlet oil in the oil outlet flow is in pure condition (100% oil).

4.7 Inlet flow (Q): Inlet flow of the process is equal to the outlet flow of the upstream process.

4.8 Water outlet flow (Qout):

)1000//)(1( oiloout CCQQ ρ−−= m3/h {3.9}

4.9 Oil outlet flow(Qoil):

outoil QQQ −= m3/h {3.10}

4.10 Customized output: 4.10.1 Theoretical cut size (dc): cut size is calculated by eq. 3.5.

5. Related parameters: Related parameters are as summarized in the following table

Table 3-1 Related parameters of simple decanter



dc Cut size of the decanter dc micron S Bottom projection area of the tank S m2 Ud Rising velocity of the droplet diameter “d” Ud m/s Udc Rising velocity of the droplet at cut size “dc” Udc m/s Δρ Difference between density of dispersed

phase and continuous phase Denc-Dend Kg/m3

ρc Density of continuous phase, in this case, water

Denc Kg/m3

μC Dynamic (or absolute) viscosity of continuous phase, which is water, for oily wastewater

Muc N.s/m2

(= 1000 cp)

ρd Density of dispersed phase, in this case, oil Dend Kg/m3


Simple decanter (SD-??)

Ud

V

Q

d = cut size

d < cut size

d > cut size

Influent

L

Oil droplets

Separatedoil layer

S

Fig. 3-1a Icon of simple decanter Fig. 3-1b Graphical diagram of simple decanter

468


IV-48


7.1 Reynolds number of droplet, Re, is between 10-4 to 1, which is the range that STOKES law is valid.

c

dc dUμ

ρ ⋅⋅=Re {3.11}

7.2 The oil droplets are uniformly distributed across the cross section area of the tank, which can be achieved by proper design of inlet chamber.

7.3 ηd of oil layer is 100%.

7.4 The oil droplet is spherical, which is normally true.

7.5 For droplets smaller than 20 microns, they are subjected to Brownian motion and cause error in the prediction of the efficiency. So it is recommended to avoid using the decanter for the wastewater with the majority part of oil droplets smaller than 20 microns. However, if these small droplets are the minority part of pollutants, the models can be used to predict the efficiency without any harm because its prediction is usually lower than observed value, thus make the prediction result on the safe side.

7.6 Oil outlet point of the process can only be connected to the skimmers only.


Table 3-2 Influent parameters for simple decanter


Hydraulic loading rate -

Droplet diameter +


469


IV-49

4) Compact decanter

1. Process abbreviation: CD

2. Process description: Parallel plate interceptor, lamella separator, or other types of plate insertion decanter that the rising distance of the droplets is equally divided and can be clearly specified.





%100⋅= CFdη {4.1}


%100⋅⋅=dc

dd U

UCFη {4.2a}


%0=dη {4.2b}

Rising velocity (U) of the droplet “d” or of the cut size “dc” is calculated from the following equations.

cd

dgU

μρ

18)10( 26−⋅⋅⋅Δ

= m/s {4.3}

⎟⎟⎠

⎞⎜⎜⎝

⎛

⋅+⋅=⎟⎟

⎠

⎞⎜⎜⎝

⎛⋅

=αcos)1(36003600 NS

QS

QUpd

dcm/s {4.4}

Cut size (micron) is calculated by the following equation.

62/1

10cos)1(3600

18⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛⋅+Δ⋅

=αρ

μNgS

QdP

cc

micron {4.5}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {4.6}


)1( dodout

d CQQC η−⋅⋅= mg/l {4.7}


∑=max

min

d

ddCC mg/l {4.8}


470


IV-50




)1000//)(1( oiloout CCQQ ρ−−= m3/h {4.9}



4.10 Customized output:

4.10.1 Theoretical cut size (dc): cut size is calculated by eq. 4.5. 4.10.2 Foot print area of the decanter (S): It is assumed to be identical to

Sp.


Table 4-1 Related parameters of compact decanter



dc Cut size of the decanter dc micron N The number of plates N Sp Area of one inserted plate (only 1 side and

measured perpendicularly to the plate surface.)

Sp m2

Ud Rising velocity of the droplet diameter “d” Ud m/s Udc Rising velocity of the droplet at cut size “dc” Udc m/s α Inclination of plate from horizontal axis (0-

90 degree) incl degree

Δρ Difference between density of dispersed phase and continuous phase

Denc-Dend Kg/m3


Denc Kg/m3


Muc N.s/m2

(= 1000 cp)



471


IV-51

Compact

Compact decanter (CD-??)

QInfluent

H


Sp

Fig. 4-1a Icon of compact decanter Fig. 4-1b Graphical diagram of compact decanter



c

dc dUμ

ρ ⋅⋅=Re {4.11}


7.3 The oil droplet is spherical, which is normally true. ηd of oil layer is 100%.




Table 4-2 Influent parameters for compact decanter


Hydraulic loading rate - Droplet diameter + Inclination of plates (0o = horizontal axis) - but helps draining the sludge from

the plate surfaces


472


IV-52

5) Customized decanter

1. Process abbreviation: CTD

2. Process description: Decanter that decanting area or oil interception area is specified. Decanting area is the area of every surface that can intercept oil droplets, regardless of rising distance of oil droplets. However, majority of rising paths should be identical. An example of the decanters is closely inserted plate interceptor, e.g. “Spiraloil”.





%100⋅= CFdη {5.1}


%100⋅⋅=dc

dd U

UCFη {5.2a}


%0=dη {5.2b}

Rising velocity (U) of the droplet “d” or of the cut size “dc” is calculated from the following equations.

cd

dgU

μρ

18)10( 26−⋅⋅⋅Δ

= m/s {5.3}

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

=d

dc SQU

3600 m/s {5.4}

Cut size is calculated by the following equation.

62/1

103600

18⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛Δ⋅

=d

cc gS

Qdρμ micron {5.5}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {5.6}


)1( dodout

d CQQC η−⋅⋅= mg/l {5.7}


∑=max

min

d

ddCC mg/l {5.8}

473


IV-53





)1000//)(1( oiloout CCQQ ρ−−= m3/h {5.9}




4.10.1 Theoretical cut size (dc): cut size is calculated by eq. 5.5.


Table 5-1 Related parameters of customized decanter



dc Cut size dc micron Sd Decanting area ( the area of all surfaces in the

decanter that can intercept oil droplets) Sd m2

Ud Rising velocity of the droplet diameter “d” Ud m/s Udc Rising velocity of the droplet at cut size Udc m/s Δρ Difference between density of dispersed phase

and continuous phase Denc-Dend Kg/m3


Denc Kg/m3

μC Dynamic (or absolute) viscosity of continuous phase, which is water

Muc N.s/m2

(= 1000 cp) ρd Density of dispersed phase, in this case, oil Dend Kg/m3


Custom

Customized decanter (CTD-??)

QInfluent

Sd

Fig. 5-1a Icon of customized decanter Fig. 5-1b Graphical diagram of customized decanter

474


IV-54



c

dc dUμ

ρ ⋅⋅=Re {5.11}



7.4 The oil droplet is spherical, which is normally true.



7.7 Loading rate of the decanter can be as high as 14.1-54 m/h.

7.8 Calculated efficiencies of closely inserted plate interceptor are usually lower than observed values. Prediction of cut size is relatively accurate so the difference of ± 20% can be expected.


Table 5-2 Influent parameters for customized decanter


Hydraulic loading rate - Droplet diameter + Inclination of plates (0o = horizontal axis) - but helps draining the sludge from the plate

surfaces Decanting area + Rising distance of oil - but too small distance may cause clogging. Presence of surfactants - for closely inserted plate interceptor


475


IV-55

6) Granular bed coalescer

1. Process abbreviation: GCC

2. Process description: Granular bed coalescer in this program is based on upflow pattern. However, it can be applied to downflow coalescer since the rising velocity of droplet are usually small, compared to flow velocity. So relative velocity between oil droplets and bed can be safely assumed to equal flow velocity.




%100)()()()()10(58.0 09.009.008.012.02.06

⋅Δ⋅

⋅= −−

cc

d

ow

cd

dpVdpH

dpdCF

ρρ

μμ

γρ

η {6.1}

And %100≤dη {6.2}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {6.3}


)1( dodout

d CQQC η−⋅⋅= mg/l {6.4}


∑=max

min

d

ddCC mg/l {6.5}





)1000//)(1( oiloout CCQQ ρ−−= m3/h {6.6}




4.10.1 Theoretical cut size (dc): Cut size is calculated from eq. 6.1 by assuming that the graded efficiency is 100%.

476


IV-56

4.10.2 Range of pressure drop (P): Pressure drop is calculated from Kozeny-Carman equation, as shown in eq. 6.8. Recommended minimum and maximum value of porosity are 0.13 and 0.23, respectively.

32

2)1(180ερεμ

⋅⋅⋅−

=dpgVHP

c

c m {6.8}


Table 6-1 Related parameters of granular bed coalescer



dc Cut size dc micron dp Diameter of collector or bed material dpc m H Bed height (height of granular bed) H m P Pressure drop across the bed Pdropmin,

Pdropmax m

V Flow velocity or empty bed velocity V m/s ε Porosity or void ratio of the bed (= void

volume / total volume) Constant 0.13 and

0.23

γow Interfacial tension between oil and water gow kg/s2 or N/m (= 1000 dyne/cm)

Δρ Difference between density of dispersed phase and continuous phase

Denc-Dend Kg/m3


Denc Kg/m3


Muc N.s/m2

(= 1000 cp)

μd Dynamic (or absolute) viscosity of dispersed phase, which is oil, for oily wastewater

Mud N.s/m2

(= 1000 cp)

ρd Density of dispersed phase, in this case, oil

Dend Kg/m3


477


IV-57

Granular bed coalescer(GCC-??)

Collector size = dpVoid ratio = ε

Bed height= H

InfluentFlow = Q

Effluent

Oil outlet

Velocity = V=Q/A

Cross section area= A

Fig. 6-1a Icon of granular bed coalescer Fig. 6-1b Graphical diagram of granular bed coalescer


7.1 Coalescer bed shall be oleophilic and relatively spherical in shape.

7.2 Tested size of bed media (dp) is between 0.20 – 1.0 mm. The larger the media size, the lower the efficiency.

7.3 Tested range of bed height (H) of the model is between 1 to 10 cm. However, bed height as low as 1 cm is not recommended. The greater the bed height, the safer the coalescer operation. However, it also results in higher pressure drop.

7.4 The velocity (V) should be in the range of 0.09 to 0.54 cm/s or 3.2 to 19.4 m/h.

7.5 Tested interfacial tension (γow) is between 11 to 42 dyne/cm or 0.011 to 0.042 N/m. (i.e. T.I.O.A, Heptane, Anisole, Toluene and Kerosene)

7.6 Different density between dispersed phase (oil) and continuous phase (water) (Δρ) is between 80 to 315 kg/m3 (approx.).

7.7 The equation is valid for droplet size (d) of 10 microns or bigger. For smaller droplets, result from the equation may not be accurate because it is beyond the data range that has been used to verify the model.

7.8 The model is valid for inlet concentration between 100 to 1,000 mg/l. At higher concentration, mousse or jet formation may occur, resulting in unpredictable decreasing of the efficiency.


7.10 Within the valid range of model, error in efficiency prediction is less than ±10%.


478


IV-58

7.12 Recommended minimum and maximum value of porosity are 0.13 and 0.23, respectively. For pressure drop calculation, since wastewater is rather diluted, it is assumed that density and viscosity of water is used in the equation.


Table 6-2 Influent parameters of granular bed coalescer


Empty bed velocity - Droplet diameter + Collector or bed material diameter - Hydrophilicity of bed - Surface roughness of bed + Bed height + Ratio of oil to water - Presence of surfactants - Temperature +


479


IV-59

7) Brush type bed coalescer

1. Process abbreviation: BCC

2. Process description: Brush type bed coalescer refers to the coalescer whose fibrous elements are in aligned neatly in radial direction, like bottlebrush.




( ) %100)(1)()10()(5.104 694.035.018.018.06

77.0 ⋅−⋅

⋅= −−

−

DH

Dd

DdVDCF F

c

cd ε

μρη {7.1}

And %100≤dη {7.2}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {7.3}


)1( dodout

d CQQC η−⋅⋅= mg/l {7.4}


∑=max

min

d

ddCC mg/l {7.5}





)1000//)(1( oiloout CCQQ ρ−−= m3/h {7.6}





4.10.2 Pressure drop (P): Pressure drop is calculated from Hazen –William equation, using equivalent length of the coaleser. The equivalent length is assumed to be 5 times of the actual length of coalescer column.

480


IV-60

167.185.1582.6 ⎟

⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛=

DH

CVPHW

m {7.8}


Table 7-1 Related parameters of brush type bed coalescer



CHW Hazen William’s C, depanding on material. C of 130 is recommended for steel column.

Constant = 130

dc Cut size dc micron D Diameter of coalescer column or of brush,

which should be relatively closed. Dcc m

dF Diameter of fiber element dF m H Bed height (height of granular bed) H m P Pressure drop across the bed Pdrop m V Flow velocity or empty bed velocity By calculation m/s

ε Porosity or void ratio of the bed (= void volume / total volume) (must be < 1)

Void

ρc Density of continuous phase, in this case, water Denc Kg/m3


Muc N.s/m2

(= 1000 cp)



Brush type bed coalescer(BCC-??)

Effluent

Oil outlet

Fiber element size = dFVoid ratio = ε

Bed height= H

InfluentFlow = Q

Velocity = V=Q/A


Fig. 7-1a Icon of brush type bed coalescer Fig. 7-1b Graphical diagram of brush type bed coalescer

481


IV-61


7.1 48 < Re < 1100. Re is the term (ρcVD/μc) in the equation. 1 < H/D < 10.

7.2 Diameter of coalescer bed (D) tested is around 5.0 cm. Using bigger coalescer diameter may cause deflection at the tips of fibers because of longer overhung length, which may cause channeling of untreated wastewater and error in efficiency calculation.

7.3 The model is valid for droplet size (d) of 1 micron and greater.

7.4 Empty bed velocity (V) is between 0.5 to 5.0 cm/s (1.8 to 180 m/h). However available raw data used to verify the model is between 0.5 to 2.0 cm/s. Using velocity > 2.0 cm/s may cause unpredictable error on calculated efficiency.

7.5 Fiber size (dF) is between 40 to 200 microns. However available raw data used to verify the model is between 100 to 200 microns. Using fiber size < 100 microns may cause unpredictable error on calculated efficiency.

7.6 Void ratio of the bed (ε) is around 0.845 to 0.96.

7.7 The model is valid for droplet size (d) of 1 microns and greater.

7.8 The model is verified at inlet oil concentration up to 1000 mg/l. Applying the model to the concentration > 1000 mg/l will cause underestimation of predicted efficiency.

7.9 The beds used in these researches vary from “bottle brush” type, simple spiral type and combination of internal bed of “simple spiral” and concentric “coil spring–like” external bed with the tip of the fibers pointed to the centerline. However, they are all oleophilic. There is some difference in efficiency between each type, but there is too few data to make a conclusion. However, because of its rigidity, the “simple spiral in coil spring- like” bed tends to operate more stable without decreasing in efficiency with time, while others tend to be deflected by weight of accumulated oil drops.


7.11 Within the valid range of model, error in efficiency prediction is less than ±20%.



482


IV-62

Table 7-2 Influent parameters of brush type bed coalescer


Empty bed velocity - Droplet diameter + Diameter of fiber element - Coalescer diameter - Hydrophilicity of bed - Porosity - Bed height + Tortousity of bed + Presence of surfactants - Temperature +


483


IV-63

8) Dynamic fibrous bed coalescer

1. Process abbreviation: DCC

2. Process description: Dynamic fibrous bed coalescer is actually the brush type bed coalescer whose bed is rotated by outside prime mover.




( ) %100)()(1)()()(76.1 53.035.035.058.058.021.0 ⋅⋅

−⋅= −−

VND

DH

Dd

DdVD

CF F

c

cd ε

μρ

η {8.1}

And %100≤dη {8.2}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {8.3}


)1( dodout

d CQQC η−⋅⋅= mg/l {8.4}


∑=max

min

d

ddCC mg/l {8.5}





)1000//)(1( oiloout CCQQ ρ−−= m3/h {8.6}





4.10.2 Pressure drop (P): Pressure drop is calculated from Hazen –William equation, using equivalent length of the coaleser. The

484


IV-64

equivalent length is assumed to be 5 times of the actual length of coalescer column.

167.185.1582.6 ⎟

⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛=

DH

CVPHW

m {8.8}


Table 8-1 Related parameters of dynamic fibrous bed coalescer




Constant = 130

dc Cut size dc micron D Diameter of coalescer column or of brush,


dF Diameter of fiber element dF m H Bed height (height of granular bed) H m N Rotating speed of bed N Rev/s P Pressure drop across the bed Pdrop m V Flow velocity or empty bed velocity By calculation m/s ε Porosity or void ratio of the bed (= void

volume / total volume) void

ρc Density of continuous phase, in this case, water Denc Kg/m3 μC Dynamic (or absolute) viscosity of continuous

phase, which is water, for oily wastewater Muc N.s/m2

(= 1000 cp) ρd Density of dispersed phase, in this case, oil Dend Kg/m3


Dynamic bed coalescer(DCC-??)

InfluentEffluent

Oil outlet

Fiber element size = dFVoid ratio = ε

Bed height= H

N rev/s

InfluentFlow = Q

Velocity = V=Q/A


Fig. 8-1a Icon of dynamic fibrous bed coalescer

Fig. 8-1b Graphical diagram of dynamic fibrous bed coalescer

485


IV-65


7.1 52 < Re < 1164. Re is the term (ρcVD/μc) in the equation.

7.2 1 < H/D < 2. Using H/D > 2 in the model can be also applied for comparison purpose only. However, the maximum H/D is 6.

7.3 Rotating speed of the bed (N) is between 0.167 to 3.33 rps (10 to 200 rpm). Please note that N is in the form of revolution per unit time, not radian per unit time). Recommended minimum rotating speed is 75 rpm. Using lower speed may not provide any additional benefit over simple fibrous bed coalescer because the effect of rotating on interception probability may be cancelled out by the shear effect, which causes fragmentation of oil drops.

7.4 Empty bed velocity (V) is between 0.1 to 1.1 cm/s (3.6 to 39.6 m/h).

7.5 Diameter of fiber (dF) is around 100 to 300 microns

7.6 Diameter of coalescer bed (D) is not greater than 11.5 cm. Using bigger diameter may cause deflection at the end of fibers from longer overhung lengths, which may cause channeling of untreated wastewater and error in calculation.

7.7 It is recommended to use the model only for the droplet size (d) of 10 microns or greater. For smaller droplet, the model can also be applied, but for comparison purpose only.

7.8 The beds, used in the experiment, are “bottle brush” types, made of oleophilic polyamide or polypropylene with stainless steel shaft. The area of shaft should not be greater than 30% of cross section area of the bed otherwise it will affect the efficiency.

7.9 Void ratio of the bed (ε) is around 0.845 to 0.96


7.11 Within the valid range of model, error in efficiency prediction is less than ± 10%.



486


IV-66

Table 8-2 Influent parameters of dynamic fibrous bed coalescer


Empty bed velocity - Droplet diameter + Diameter of fiber element - Coalescer diameter - Hydrophilicity of bed - Rotating speed + within the upper limit Porosity - Bed height + Tortousity of bed + Presence of surfactants - Temperature +


487


IV-67

9) Metal wool bed coalescer

1. Process abbreviation: WCC

2. Process description: Metal wool bed coalescer is one of fibrous bed coalescer that uses disorderly (random) or woven fiber material, like steel wool for kitchen use, as a bed. The bed is actually not necessary to be of metal. It is just named to refer to general appearance of bed.




%100)()()()(35.3 36.003..003.023.0 ⋅⋅= −−

DH

Dd

DdVDCF F

c

cd μ

ρη {9.1}

And %100≤dη {9.2}


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {9.3}


)1( dodout

d CQQC η−⋅⋅= mg/l {9.4}


∑=max

min

d

ddCC mg/l {9.5}





)1000//)(1( oiloout CCQQ ρ−−= m3/h {9.6}





488


IV-68

4.10.2 Pressure drop (P): Pressure drop is calculated from Hazen –William equation, using equivalent length of the coaleser. The equivalent length is assumed to be 5 times of the actual length of coalescer column.

167.185.1582.6 ⎟

⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛=

DH

CVPHW

m {9.8}


Table 9-1 Related parameters of metal wool bed coalescer




Constant = 130

dc Cut size dc micron D Diameter of coalescer column or of bed,


dF Diameter of fiber element dF m H Bed height (height of granular bed) H m P Pressure drop across the bed Pdrop m V Flow velocity or empty bed velocity By calculation m/s ρc Density of continuous phase, in this case,

water Denc Kg/m3


Muc N.s/m2

(= 1000 cp)



489


IV-69

Metal wool bed coalescer(WCC-??)

Collector size = dpVoid ratio = ε

Bed height= H Effluent

Oil outlet

InfluentFlow = Q

Velocity = V=Q/A


Fig. 9-1a Icon of metal wool bed coalescer Fig. 9-1b Graphical diagram of metal wool bed coalescer


7.1 The beds used in the experiment are highly disorderly bulk of stainless steel fiber, dF = 75 microns, and steel wool, dF = 40 microns (see fig. 5.4.1-4). However, only the latter case, which raw experimental data is available, is used to develop the model. The minimum size of oil droplet tested is 1 micron.

7.2 Tested Reynolds number is between 840 to 2470.

7.3 Porosity of the bed (ε) is around 0.95. But there is not sufficient data to include it into the model.

7.4 Diameter of the coalescer (D) = 5 cm. However, according to rigidity and, controversially, density uniformity of this type of bed, the diameter in the model should be scaled up or down without causing serious error.

7.5 Height of the coalescer bed (H) is between 0.07 to 0.21 m.

7.6 Velocity (V) is between 1 to 2.5 cm/s or 36 to 90 m/h.

7.7 Inlet concentration is around 1000 mg/l.


7.9 Within the valid range of model, error in efficiency prediction is less than ± 10%.



490


IV-70

Table 9-2 Influent parameters of metal wool bed coalescer


Empty bed velocity - Droplet diameter + Diameter of fiber element - Coalescer diameter - Hydrophilicity of bed - Porosity not sufficient data to verify but should be - Bed height + Tortousity of bed + but can cause clogging Presence of surfactants - Temperature +


491


IV-71

10) Dissolved air flotation

1. Process abbreviation: DAF

2. Process description: Dissolved air flotation (DAF) in this case does not include the coagulation/flocculation process, which is separated into another module.



4.1 Graded efficiency (ηd): Graded efficiency of the process is governed by SIEM’s model (INSA thesis 1983) and its scale-up procedure, derived in the scope of work of this program.

SIEM’s model

%100)1())(

23(

,

exp

⋅−⋅=Φ

−bd

HAV

refd eCFαη

η {10.1}

5919.0exp )(009005.0)( theoηαη = {10.2}

diffIntsedtheo ηηηη ++=

26

)10(23

bInt d

d −⋅=η {10.3}

rcsed V

dgμ

ρη18

)10( 26−⋅Δ= {10.4}

3/2))273((9.0brc

Diff ddVTK

μη +

= {10.5}

c

bwaterairbr

gdUVμ

ρ18

2/Δ

== {10.6}

And %100≤dη {10.7}

The model is valid only when Φ/AV = 0.0516. However H can vary.

Scale-up from SIEM’s condition by population balance method

1) Calculate the reference efficiency (ηd,ref) of the model at required height (Hreq), average bubble diameter (db) and droplet sizes (d) using eq. 10.1 to eq.10.7. Use Φ/AV = 0.0516 in order that the operating condition of SIEM still holds.

2) Scale up the area from 0.01767 m2 (Amodel) to required area (Areq). At this step, SIEM’s operating condition still holds. So efficiency from above equations remains the same. This required area could be approximated from recommended hydraulic loading rate (Vreq) and ratio of pressurized water to wastewater ((Qpw/Q)req). Refer to reference for more detail.

req

reqpw

req

req V

QQQ

A⎟⎠⎞

⎜⎝⎛ +

=)(1

{10.8}

492


IV-72

3) Find Φref, corresponding to the area Areq, by following equation,

01767.0102.4 7

modmod

reqel

el

reqref

AAA −×

=Φ⋅=Φ m3/s {10.9}

4) Find τref, corresponding to the height Hreq, by following equation,

36001

00046.0modmod

⋅=⋅= reqel

el

reqref

HHH

ττ Hour {2.610.10}

Change Φ and τ from SIEM’s condition by population balance model

5) From population balance method, calculate κ2,ref corresponding to Areq, Hreq, τref and Φref from the reference efficiency (from item 1)) by the following equations. Please note that, at this point, SIEM’s condition still holds. κ2,req has to be calculated separately for each droplet diameter.

refref

refdref τ

ηκ

⋅Φ

−−=

)1ln( ,,2

{10.11}

Or )(

,,21 refrefreferefd

τκη Φ−−= {10.12}

6) Find Φreq from required ratio of pressurized water to wastewater (see reference for the recommended value) by following equations.

3600QR

VVQ

VV

pw

airpw

pw

air ⋅⋅=⋅=Φ m3/s {10.13a}

Then, from Henry’s law QRHKmolmTPPy atm ⋅⋅⋅⋅×⋅+⋅−⋅=Φ − ))/(10082.0()273()( 33 {10.13b}

For air, y = 1. H is Henry’s constant as molair/(m3 water.atm). Henry’s constant of air can be calculated from the following equation.

123 10)472.122745.00039.000002.0( −⋅+⋅−⋅+⋅−= TTTH {10.14}

7) Choose τreq from recommended criteria (see reference).

8) To change Φ and τ from SIEM’s, the following procedure is recommended and precautions should be noted.


In this case, κ2 is assumed to remain the same and be equal to κ2,ref. Φreq and τref are used. The efficiency can be estimated by eq. 10.15a.

)( ,21 refreqrefedτκη Φ−−= {10.15a}

The calculated efficiency will be lower than the real value.


In this case, κ2 = κ2,ref. Φreq and τreq are used. The efficiency can be calculated by eq. 10.15b. The calculated efficiency will be lower than the real value.

)( ,21 reqreqrefedτκη Φ−−= {10.15b}

493


IV-73


Like the former case, the efficiency can be calculated by eq. 10.15b.


This case is not feasible because it means that we have to decrease wastewater flowrate and increase pressurized water flowrate.

There is no obvious limit for the 4 adaptations, shown above. However it is recommended to use the values of each parameter (d, db, C, etc.) within general range (see reference).

It must be noted that the graded efficiency in this case is independent of source of water that is used as pressurized water.

4.2 Total efficiency (ηt): The value depends on the source of water that is used as pressurized water. In the program, it is assumed that DAF effluent is used to prepare pressurized water.

%100)(⋅

−=

o

ot C

CCη {10.16a}

4.3 Graded outlet oil concentration in water outlet flow (Cd): In the program, it is assumed that DAF effluent is used to prepare pressurized water. Thus residual oil in the effluent is recycled. So oil of certain droplet sizes may accumulate in the system, depending on their ηd.

diloddout

d CQQC ,)1( η−⋅= mg/l {10.17}

t

od

d

rd

dilod QQC

RR

RR

C ⋅−−

+

−−+=

−

1)1(1

)1))1(1

(( 1max

,η

η mg/l {10.18}

The rmax is an assumed number of loop, which effluent is recycled until the concentration reaches steady stage. Ideally, rmax must approach infinity. However, the value of rmax in the program is set at 30, which is generally sufficient to make the equation reach steady stage (esp. when ηd > 20%). If the value of Cod,dil from r = 30 differs from Cod, dil at r = 29 more than 10%, the program will display the warning box, showing that the oil may accumulate in the system.


∑=max

min

d

ddCC mg/l {10.19}




494


IV-74

4.8 Water outlet flow (Qout): )1000//)(1( oiloout CCQQ ρ−−= m3/h {10.20}


oiloil QQQ −= m3/h {10.21}


4.10.1 Ratio of air/oil (W/W)

10003600

0

⋅⋅⋅Φ

=CQW

W airreq

oil

air ρ kg air / kg oil {10.22}

4.10.2 Theoretical energy (watt) required for air compressor:

pwatmcomp

air QairMW

airConcPPTRPower ⋅⋅

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡−⎥

⎦

⎤⎢⎣

⎡⋅+

=

−

)1000/)(()(1

4.0)273(

)4.1

14.1(

η{10.23}

Rair = universal gas constant (8.314)

1000)().( HairMWPyairConc ⋅⋅⋅

= mg/l {10.24}

4.10.3 Theoretical energy required for pressurized water pump:

pump

atmc

pump

atmpw PPgQRPPQPower

ηρ

η ⋅⋅−⋅⋅⋅⋅

=⋅

−⋅=

360010)(

3600)( watt {10.25}

P is measured in atm.

4.10.4 Hydraulic loading rate

ARQV )1( +

= m/h {10.25}

4.10.5 Retention time (based on total flow (Q+Qpressurized water):

60⋅=VHτ min {10.25}


Table 10-1 Related parameters of DAF



A Cross section area of DAF column A m2 db Diameter of bubble db m H Column height DAFH m P Absolute pressure of saturator DAFP atm R Recycled ratio (R = Q pressurized water/Q) R

ηcomp, ηpump Efficiency of air compressor and pressurized water pump (always < 100%, 60-70% is recommended)

Constant = 70%

Note: Other parameters are internally used. Users do not need to input. For list of others parameters, see reference.

495


IV-75


Dissolved air flotation(DAF-??)

Pressurized water flow = Qpw

Oil outlet

Effluent

Influent

H

Area = A

Fig. 10-1a Icon of DAF Fig. 10-1b Graphical diagram of DAF


7.1 Inlet oil concentration should not be greater than 1,200 mg/l (before dilution) or 435 mg/l (after dilution). Using the model with higher oil concentration will result in underestimating of efficiency.

7.2 SIEM’s model is tested at the following operating condition;

• Φ/AV = 0.0516. Only this value must be used in the equations. As long as this value is fixed, SIEM’s operating condition still holds and the model is still valid.

• Retention time, based on total flowrate (Qt), is around 25 minutes. • Droplet diameter (d) tested is between 2 to 40 microns. • Diameter of air bubbles (db) varies from 15 to 130 microns. Tested

average bubble diameter is 70 microns, which is used to verify the model, and standard deviation of bubble diameters is 34.5 microns. The range of bubble sizes is common for commercial pressurized water system or saturator. The pressure of the test system is 4 atm (absolute).

• Tested air flowrate (Φ) is 0.42 cm3/s (4.2e-7 m3/s). • Tested wastewater flowrate (Q) is 3.9 cm3/s (3.9e-6 m3/s) • Tested water depth (H) is 0.70 m. H can be freely changed as long as

(Φ/AV) is fixed. Anyway, H around 1.8 to 2.7 is recommended by API. • Diameter of flotation column is 0.15 m Cross section area of column

(A) is 0.01767 m2. • Ratio of pressurized water to wastewater (Qpw/Q) is 1.76. • Air to pollutants ratio used is around 0.12 kg. air/ kg. oil. • Ratio of number of bubble/ oil droplet tested is around 1.4 oil droplet/ 1

air bubble. • Hydraulic loading rate or flow velocity (V), based on Qt, is 1.6 m/h

7.3 Within the valid range of model, error in efficiency prediction is less than ± 20%. If SIEM’s condition does not holds and procedure described in item 8) is used, error in efficiency prediction depends on difference between

496


IV-76

SIEM’s and actual condition. However, the predicted efficiency is usually lower than actual value.



7.6 It is assumed that DAF is operated at Patm. Thus Equations and some constants, e.g. air density, are fixed and valid for Patm only.


Table 10-2 Influent parameters of DAF


Retention time + Droplet diameter + Diameter of bubble - Air flowrate + Column height + Presence of transfer compound ± see reference Presence of surfactants - Turbulence in column ± see reference


497


IV-77

11) Two-phase hydrocyclone

1. Process abbreviation: 2CY

2. Process description: The process is based on THEW type, liquid-liquid hydrocyclone.




For d ≥ dc, %100=η {11.1}

For d < dc,

%1002)

2nD(0.1862)

2nD((

2)2nD(0.1862

d(R

dη ⋅−

−= {11.2}

Rd (Radial distance from axial axis to entry position of droplet “d” (Z=0))can be calculated from the following equations. If Rd in eq. 11.3a is equal to Dn/2, the corresponding “d” will be equal to dc.

∫=∫L

0 WdZdR

zvvR UdR {11.3a}

RV 2

18μ

2)6-10Δρ(dU ⋅⋅

= m/s {11.3b}

0.65)RnD

)(2iπD

Q(V = m/s {11.3c}

319.1

263.81233.3 ⎟⎟

⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zRR

zRR

zRR

zWW {11.3d}

2))2

tan(Znπ(0.5D

QzW β

⋅−= m/s {11.3e}

)2

tan(2

β⋅−= Z

DR n

z m {11.3f}

)2/5.1tan()(25.0

onD

L = m {11.3g}

For THEW’s type hydrocyclone, when Z = L:

)2/(186.0 nVZZ DR = {11.3f}


∑ ⋅−

=max

min

%100)1(

1 d

d f

odd

ot R

CC

ηη {11.4}

498


IV-78


)1()1(

f

oddd R

CC

−−

=η mg/l {11.5}


∑=max

min

d

ddCC mg/l {11.6}

4.5 Graded outlet oil concentration in oil outlet flow (Coild): Unlike other true separation process, hydrocyclone only concentrates oil. Thus concentrated oily wastewater, not pure oil, is obtained at oil outlet (overflow) port.

f

oddoverflowdoild R

CCC η== ,

mg/l {11.7}

4.6 Total outlet oil concentration in oil outlet flow (Coil):

∑=max

min

d

doildoil CC mg/l {11.8}



)1( fout RQQ −= m3/h {11.9}


foil RQQ ⋅= m3/h {11.10}


4.10.1 Theoretical cut size (dc): Cut size is calculated from item 4.1. 4.10.2 Pressure drop (P): Pressure drops across inlet/overflow (Po) and

inlet/underflow (Pu) ports of 2-phase hydrocyclone are governed by the following equations.

1611.0

4

3.2

)1(6.2)3600/(16 ⎟

⎟⎠

⎞⎜⎜⎝

⎛

−⋅=Δ

fno RD

QP bar {11.11a}

4

2.2)3600/(6.4n

u DQP =Δ bar {11.11b}


499


IV-79

Table 11-1 Related parameters of two-phase hydrocyclone



dc Cut size dc micron Dn Nominal diameter of hydrocyclone Dn m Rf Split ratio (Qoverflow/Q), Rf < 1.

Recommended value of Qoil is 1.8 to 2 times of inlet oil flowrate.

Rf

Note: Other parameters are internally used. Users do not need to input. For list of others parameters, see reference.


Di

L1

D

Dn

Ds

L3

Doθ

β

L

Oil out

Water out

Water in

2-phasehydrocyclone

(2CY-??) Dn/D=0.5, Ds/D=0.25, Do/D<0.05, L1/D=1, L3/D=15-20 , β=1.50o, θ=20o,Di/D=0.25 for 1 inlet and 0.175 for 2 inlet ports, total length/D =45 (approx.)

Fig. 11-1a Icon of 2-phase hydrocyclone

Fig. 11-1b Graphical diagram of 2-phase hydrocyclone


7.1 The model is valid for THEW hydrocyclone or other cyclones with relative identical geometry.

7.2 It is recommended to use the model only for droplet diameter of 20 microns or greater. For smaller droplet, it can also be applied, but for comparison only. In analysis mode and design mode when “Find efficiency” is selected, efficiencies of droplets smaller than 20 microns are assumed to be 0%. ηd of oil is 100%.

7.3 The equations are valid for the hydrocyclone with 2 inlet ports only.

7.4 Overflow quantity (Qoveflow) is usually small, not greater than 10%. Its effect on velocity profiles and efficiency is small, thus, negligible. Recommended Qoveflow is 1.8 to 2 times of inlet oil flowrate.

7.5 The model tends to predict lower efficiency when d < d80% and higher efficiency when d > d80%. Error of cut size prediction is around 10-20%,

500


IV-80

e.g. if predicted cut size is 50 microns, observed cut size should be around 40 – 45 microns. For more details, see reference.

7.6 Graded efficiency (ηd) in this case is based on ratio between outlet oil quantity at water outlet port to that of inlet wastewater. Effect of split flow is not yet taken into account.


Table 11-2 Influent parameters of two-phase hydrocyclone


Flowrate + Droplet diameter + Split ratio is Inlet oil concentration is Pressure drop + Presence of surfactants - Temperature +


501


IV-81

12) Three-phase hydrocyclone

1. Process abbreviation: 3CY

2. Process description: The process is based on 3-phase hydrocyclone, initiated by MA and AURELLE. Geometry of liquid-liquid separation of the hydrocyclone is based on Thew type. For solid-liquid, it is based on Rietema type. It can be used for simultaneous separation of oil and suspended solids from wastewater. Calculation procedure and equation are relatively identical to that of 2-phase cyclone with only little modification. Solid removal efficiency is not included in this program.



4.2 Graded efficiency (ηd): Graded efficiency of the process is governed by the following equations. Dn in this case is equal to Do and L is equal to L5 (See graphical diagram).

For d ≥ dc, %100=η {12.1}

For d < dc,

%1002)

2nD(0.1862)

2nD((

2)2nD(0.1862

d(R

dη ⋅−

−= {12.2}

Rd (Radial distance from axial axis to entry position of droplet “d” (Z=0))can be calculated from the following equations. If Rd in eq. 12.3a is equal to Dn/2, the corresponding “d” will be equal to dc.

∫=∫L

0 WdZdR

zvvR UdR {12.3a}

RV 2

18μ

2)6-10Δρ(dU ⋅⋅

= {12.3b}

0.65)RnD

(2iD

4π(Q/2)0.676V

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

= {12.3c}

319.1

263.81233.3 ⎟⎟

⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛−+−=

zRR

zRR

zRR

zWW {12.3d}

2/2))tan(Znπ(0.5D

QzW

β⋅−= {12.3e}

)2

tan(2

β⋅−= Z

DR n

z {12.3f}

)2/5.1tan()(25.0

onD

L = {12.3g}

502


IV-82

For THEW’s part hydrocyclone, when Z = L:

)2/(186.0 nVZZ DR = {12.3f}


∑ ⋅−

=max

min

%100)1(

1 d

d f

odd

ot R

CC

ηη {12.4}


)1()1(

f

oddd R

CC−

−=

η mg/l {12.5}


∑=max

min

d

ddCC mg/l 12.6}

4.5 Graded outlet oil concentration in oil outlet flow (Coild): Unlike other true separation process, hydrocyclone only concentrate oil. Thus concentrated oily wastewater, not pure oil, is obtained at oil outlet (overflow) port.

f

oddoild R

CC η= mg/l {12.7}


∑=max

min

d




)1( SSfout RRQQ −−= m3/h {12.9}


foil RQQ ⋅= m3/h {12.10}


4.10.1 Theoretical cut size (dc): Cut size is calculated from item 4.1. 4.10.2 Purged flow for SS removal:

SSss RQQ ⋅= {12.11} 4.10.3 Pressure drop (P): Pressure drops across inlet/overflow (Po) and

inlet/underflow (Pu) ports of 2-phase hydrocyclone are governed by the following equations.

4D

2.12(Q/3600)49.8waterΔP = bar {12.12a}

4D

2.34(Q/3600)21ssΔP = bar {12.12b}

4D

2.03(Q/3600)55oilΔP = bar {12.12c}

503


IV-83

Please not that RSS is not included in the model so the pressure drop for SS port and water port are estimated value only. But for oil port, Rf is usually small so ΔPoil is relatively accurate and hardly affected by Rf.


Table 12-1 Related parameters of three-phase hydrocyclone



dc Cut size dc micron

D Diameter of the biggest cylinder of hydrocyclone (refer to related graphic)

D3cy m

Rf Split ratio (Qoil /Q), Rf < 1. Recommended value of Qoil is 1.8 to 2 times of inlet oil flowrate.

RSS Split ratio (QSS/Q). Recommended value is 0.2.

Note: 1. Other parameters are internally used. Users do not need to input. For list of others parameters, see reference.

2. Rss is calculated by Yoshioka and Hotta relation [30]: 1-Rf=0.95/((Du/Ds)4+1).



DoDDs

DiDu

Dp

L5 L3L1L3

L4

3-phasehydrocyclone

(3CY-??)

Note: Di/D=0.25 for 1- inlet and 0.175 for 2- inlet,

Do/D=0.43,Ds/D=0.28, Du/D=0.19, Dp/D=0.034,

L1/D=0.4,L2/D=5, L3/D=15, L4/D=0.3, Solid-liquid part cone

angle=12o, for liquid-liquid part=1.5o

Fig. 12-1a Icon of 3-phase hydrocyclone

Fig. 12-1b Graphical diagram of 3-phase hydrocyclone


7.1 The model is valid for THEW hydrocyclone or other cyclones with relative identical geometry.

504


IV-84

7.2 It is recommended to use the model only for droplet diameter of 20 microns or greater. For smaller droplet, it can also be applied, but for comparison only. In analysis mode and design mode when “Find efficiency” is selected, efficiencies of droplets smaller than 20 microns are assumed to be 0%. ηd of oil is 100%.

7.3 The equations are valid for the hydrocyclone with 2 inlet ports only.

7.4 Purged quantity at oil outlet port (Qoil) is usually small, not greater than 10%. Its effect on velocity profiles and efficiency is small, thus, negligible.

7.5 The model tends to predict lower efficiency when d < d80% and higher efficiency when d > d80%. Error of cut size prediction is around 10-20%, e.g. if predicted cut size is 50 microns, observed cut size should be around 40 – 45 microns. For more details, see reference.

7.6 SS removal coefficient is not calculated since it is not major scope of work of this thesis so it cannot connect to other model. Only purged flow of SS port is calculated

7.7 RSS is not included in the model so the pressure drop for SS port and water port are estimated value only. But for oil port, Rf is usually small so ΔPoil is relatively accurate and hardly affected by Rf.

7.8 Graded efficiency (ηd) in this case is based on ratio between outlet oil quantity at water outlet port to that of inlet wastewater. Effect of split flow is not yet taken into account.


Table 12-2 Influent parameters of three-phase hydrocyclone


Flowrate + Droplet diameter + Split ratio (Rf) is Inlet oil concentration is Pressure drop + Presence of surfactants - Temperature +


505


IV-85

13) Ultrafiltration

1. Process abbreviation: UF

2. Process description: Flux calculation of UF in the program is based on film model and resistance model. Flux prediction is available in DESIGN mode only.



4.1 Graded efficiency (ηd): For stabilized emulsion wastewater, oil removal efficiency of UF can be safely assume to be 100%. The program will use the value of CF.100%.

%100⋅= CFdη {13.1a}

For wastewater containing free oil or non -stabilized emulsion, the efficiency is lower but, somehow, unpredictable. Anyway, for such wastewater, the use of UF is not necessary. However the program will, nonetheless, assume that the efficiency is CF.100%.

4.2 Total efficiency (ηt): Total efficiency falls into the same case as that of hydrocyclone. For UF, users normally set their final volume of filtrated wastewater or retentate (Volretentate), which will be used for total efficiency calculation. UF calculation in the program is based on batch system without additional feed during operation period. So the concentration depends on initial wastewater volume (Volo) and feed duration (tf)

∑∑ ⋅−

−=⋅

−

−=

max

min

max

min

%100)11(

)1(1%100)1(

)1(1 d

d

odd

o

d

d

o

retentate

odd

ot

F

CC

VolVol

CC

ηηη{13.1b}

F is factor of concentration, which is specified by user.


)1(

)1(

o

retentate

oddd

VolVol

CC

−

−=

η mg/l {13.2}


∑=max

min

d

ddCC mg/l {13.3}

4.5 Graded outlet oil concentration in oil outlet flow (Coild): Unlike other true separation process, UF only concentrate oil. Thus concentrated oily wastewater, not pure oil, is obtained as retentate. Granulometries of the retentate were not studied. Coalescence and partial destabilization, which leads to formation of oil film, were reported. However, to be on the safe side, the program will assume that there is no coalesce occurring.

FCVolVol

CC oddoretentate

oddoild ηη

==)/(

mg/l {13.4}

506


IV-86


∑=max

min

d



4.8 Water outlet flow (Qout): Actual water outlet flow from the membrane is the product of permeate flux, which varies with time, and membrane area. However, since the process is usually designed as batch system, outlet flow can be chosen freely, providing that permeate storage tank is big enough. User can freely specify permeate discharge time (td). To imitate a continuous process, it can be assumes that default permeate discharge duration (td) is equal to wastewater inlet duration.

)11()(

FtQ

tVoltQ

Q fd

retentatefout −⋅⋅=

−⋅= m3/h {13.6}

4.9 Oil outlet flow (Qoil): Like water outlet flow, oil outlet flow or retantate flow depends on its discharge duration (tdoil). This duration should be equal to that of tf and td to fully imitate continuous process.

doil

fdoilretentateoil tF

tQtVolQ

⋅⋅

== / m3/h {13.7}

4.10 Customized output: 4.10.1 Permeate flux and volume evolution: Flux varies with time since

concentration of the feed (actually, retentate) increases from continuous loss of its permeate. The program can calculate the flux under the assumption that there is no fouling, which is practically proven in case that suitable pretreament is provided. Flux will be calculated using 2 models, i.e. Film model and Resistance model, and the related equations,

)ln(C

CkVJ gβ= l/(m2-h) {13.8}

tm

t

PVRP

J⋅⋅+

= αφ' l/(m2-h) {13.9}

))(0239.0( ABCBCA

eJJ oo−−= {13.10}

The values of K, Cg, etc. for certain wastewaters are provided in help file. If two types of wastewater with different values of k, Cg, etc.(such as microemulsion mixes with macroemulsion.) are encountered, the flux will be calculated using weight average method (eq. 13.11). This calculation exists only in design mode.


mixJ+

= {13.11}

Evolution of permeate flux and volume is governed by the following equation.

AdtJ(c)dVol ⋅= {13.12}

507


IV-87

Integration of eq. 13.11 is carried out using step method by dividing total permeate volume (eq. 3.12) into 100 parts.

4.10.2 Time required to complete one batch for a specified amount of wastewater: It is also governed by eq. 13.12. It must be noted that the program can calculate the flux/time evolution under the condition that there is no additional inlet wastewater during UF operation.


Table 13-1 Related parameters of ultrafiltration



A Membrane area AreaUF m2 C’g, C’g2 Gel concentration for low range of

concentration (if any) of 1st and 2nd emulsion Cg1low, Cg2low % oil volume/total

volume Cg, Cg2 Real gel concentration of 1st and 2nd emulsion

(see reference) Cg1High, Cg2High

% by volume of oil

Cref1, Cref2 Reference concentration of 1st and 2nd emulsion (conc. that is used to acquire the value of model constants)

Cref1, Cref2 % by volume of oil

Co1/(Co1+Co2)

Oil concentration ratio of wastewater no.1 to total oil concentration[2], always < 1

Rofoil1 in the unit of bar, (m/s) and l/h-m2

k’1, k’2 Constant for film model for low range of oil concentration of 1st and 2nd emulsion

Kreflow1, kreflow2

in the unit of bar, (m/s) and l/h-m2

k1, k2 Constant for film model for high range of oil concentration of 1st and 2nd emulsion

Krefhigh1, krefhigh2


Pt Transmembrane pressure UFP bar Q Flowrate of inlet wastewater Q m3/s

Rm (or R’m) Modified membrane resistance (= membrane resistance + fouling resistance (if any))

Rm in the unit of bar, (m/s) and l/h-m2

td Duration for discharge the permeate (after UF operation is finished), recommended to be equal to tf.

td h

tdoil Duration for discharge the retentate (after UF operation is finished), recommended to be equal to tf.

tdoil h

tf Duration to fill the storage tank by inlet wastewater before the wastewater is stopped and UF operation starts.

tf h

V Recirculation velocity UFV m/s F Concentration factor (Volo /Volretentate) ConcF

α1, α2 Constant for resistance model of 1st and 2nd emulsion

alpharef1, alpharef2


β1, β2 Constant for film model of 1st and 2nd emulsion bref1, bref2 in the unit of bar, (m/s) and l/h-m2

φ1,φ2 Constant for film model of 1st and 2nd emulsion Phiref1,Phiref2 in the unit of bar, (m/s) and l/h-m2


508


IV-88


Ultrafiltration(UF-??)

UF

Permeate

Retentate

Membrane

Feed pump

Storage tank

Feed

Po

Pi

Pp

Heat exchanger

Pt = ((Pi+Po)/2)-Pp

V

No influent added during filtration process

Fig. 13-1a Icon of ultrafiltration Fig. 13-1b Graphical diagram of ultrafiltration


7.1 Flux calculation of mixed wastewater exists only in design mode.

7.2 The process can be used for microfiltration and nanofiltration if they can be represented by film model and resistance model.

7.3 The program can calculate only batch operation.

7.4 Accuracy of the models are governed by that of constants used, and deviation of real operation from their assumption, thus unpredictable. If possible, it is recommended to perform UF test, using real wastewater.


Table 13-2 Influent parameters of ultrafiltration


Viscosity of feed -

Temperature +

Membrane properties ±

Reaction between feed and membrane ±


509


IV-89

14) Reverse osmosis

1. Process abbreviation: RO

2. Process description: RO in this case is used as a finishing treatment process to treat the effluent of UF or distillation only. Its main purpose is to remove soluble pollutants, normally surfactants/co-surfactants, left after oil separation. Furthermore, process calculation is based mainly on specific data from manufacturer. Thus, the module exists only in ANALYSIS mode.



4.1 Graded efficiency (ηd): It is assumed to be 100%. CF is not allowed in this process.

4.2 Total efficiency (ηt): It is assumed to be 100%.


Cd = 0 mg/l


C = 0 mg/l

4.5 Graded outlet oil concentration in oil outlet flow (Coild): Unlike other true separation process, RO only concentrate oil. Thus concentrated oily wastewater, not pure oil, is obtained as retentate. The program will assume that droplet sized and numbers are not changed.

FCVolVol

CC oddoretentate

oddoild ηη

==)/(

mg/l {14.1}


∑=max

min

d



4.8 Water outlet flow (Qout): Actual water oultet flow from the membrane is the product of permeate flux, which varies with time, and membrane area. However, since the process is usually designed as batch system, outlet flow can be chosen freely, providing that permeate storage tank is big enough. User can freely specify permeate discharge time (td). To imitate a continuous process, it can be assumes that default permeate discharge duration (td) is equal to wastewater inlet duration.

)11()(

FttQ

tVoltQ

Qd

f

d

retentatefout −⋅

⋅=

−⋅= m3/h {14.3}

4.9 Oil outlet flow (Qoil): Like water outlet flow, oil outlet flow or retantate flow depends on its discharge duration (tdoil). This duration should be equal to that of tf and td to fully imitate continuous process.

doil

fdoilretentateoil tF

tQtVolQ

⋅⋅

== / m3/h {14.4}

510


IV-90


4.10.1 TOD of water outlet flow (TODwater): Since characteristic of inlet wastewater for RO varies only slightly. TOD removal efficiency (ηTOD) is within the certain range around 70-90% (see reference). So the program will calculate the TOD from TODo and the efficiency, using the following equation.

)/11(1)1())((

FmgmgCTODTOD TOD

oil

TODoowater −

⋅−⋅⋅−= η {14.5}

4.10.2 TOD of oil outlet flow (TODoil):

FF

TODTODTOD waterooil ⋅−⋅−= ))11(( {14.6}


Table 14-1 Related parameters of RO



mgTOD/ mgoil

Unit TOD of oil. For more information, see reference

unitTOD

td Duration for discharge the permeate (after UF operation is finished), recommended to be equal to tf.

td h

tdoil Duration for discharge the retentate (after UF operation is finished), recommended to be equal to tf.

tdoil h

tf Duration to fill the storage tank by inlet wastewater before the wastewater is stopped and UF operation starts.

tf h

TODo TOD of the inlet wastewater TODin mg/l F Concentration factor (Volo /Volretentate) ConcF

ηTOD TOD removal efficiency. It includes only soluble TOD. TOD from oil is not included.

effTOD %



511


IV-91

RO

Reverse osmosis(RO-??)

Permeate

Retentate

Membrane

Feed pump

Storage tank

Feed

Po

Pi

Pp

Heat exchanger

Pt = ((Pi+Po)/2)-Pp

V

No influent added during filtration process

Fig. 14-1a Icon of RO Fig. 14-1b Graphical diagram of RO


7.1 The process is included in the program to fulfil the entire process train of oily wastewater treatment only. Thus, its calculation is based on data from researches, rather than mathematical model.

7.2 The process should be connected only to the water outlet port of UF or distillation since it can used only with oil-free water or water with only trace of oil. If connected to other unit, the program will display warning.

7.3 The program can calculate only batch operation.

7.4 Calculated TOD is an internal result of the model. It cannot be exported to other process because of programming limitation.

7.5 This module is available only in ANALYSIS mode.

512


IV-92

15) Heteroazeotropic distillation

1. Process abbreviation: HD

2. Process description: Heteroazeotropic distillation is an enhanced distillation process by addition of certain chemical, usually hydrocarbons, as an entrainer to lower the boiling point of the system and extract the water from the waste to be treated. The process can be used to treat slop or UF retentate from cutting oil treatment or concentrated oily wastewater.


4. List of outputs and related equations 4.1 Graded efficiency (ηd): It is assumed to be 100%. CF is not allowed in this

process. 4.2 Total efficiency (ηt): It is assumed to be 100%. 4.3 Graded outlet oil concentration in water outlet flow (Cd):

Cd = 0 mg/l 4.4 Total outlet oil concentration in water outlet flow (C ):

C = 0 mg/l 4.5 Graded outlet oil concentration in oil outlet flow (Coild): Outlet oil is in the

form of water-free oil (100% oil). 4.6 Total outlet oil concentration in oil outlet flow (Coil): Outlet oil is in the

form of water-free oil (100% oil). 4.7 Inlet flow (Q): Inlet flow of the process is equal to the outlet flow of the

upstream process. 4.8 Water outlet flow (Qout): Water outlet flow is equal to water content of inlet

wastewater under the assumption that size of the process is sufficient to handle inlet flowrate on real time basis.

)1000//)(1( oiloout CCQQ ρ−−= m3/h {15.1} 4.9 Oil outlet flow (Qoil): Oil outlet flow is equal to oil content of inlet

wastewater.

outoil QQQ −= m3/h {15.2} 4.10 Customized output:

4.10.1 Required quantity of entrainer: Quantity of entrainer depends on the type of entrainer and the quantity of water to be removed, which can be calculated by the following equation.

H

Houtentrainer y

yQQ )1( −= m3/h {15.3}

The values of yH (azeotropic composition) of certain entrainers are provided in the help file.


513


IV-93

Table 15-1 Related parameters of heteroazeotropic distillation



yH Azeotropic composition (Ratio of volume of water in ditillate to total distillate volume)

yH Vol/vol

Note: 1 Other parameters are internally used. Users do not need to input. For list of others parameters, see reference.


Aze

otro

pic

Heterozeotropic distillation (HD-??)

2 ph.vapor

Bubble curve

Pure H2O

Temperature






TH

Fig. 15-1a Icon of heteroazeotropic distillation

Fig. 15-1b Graphical diagram of heteroazeotropic distillation


7.1 Calculation is based on theoretical equation, regardless of distillation column design.

7.2 There is some TOD present in the distillate, caused by volatile pollutants, which depends on wastewater characteristic. Thus it can not be accurately predicted. Distillate TOD of 2000 –3000 mg/l is recommended for distillation of UF retentate from cutting oil treatment.

7.3 Entrainer can be continuously reused.

7.4 If possible, pilot-test should be conducted to verify the value of yH and observe other problems that may be present, such as release of hydrogen sulfide.

514


IV-94

16) Stripping

1. Process abbreviation: ST

2. Process description: Stripping is reversed version of heteroazeotropic distillation. It is an enhanced distillation process by addition of water, as an entrainer to lower the boiling point of the system and extract the hydrocarbon from the waste to be treated. The process can be used to treat water with trace volatile hydrocarbons.


4. List of outputs and related equations 4.1 Graded efficiency (ηd): It is assumed to be 100%. CF is not allowed in this

process. 4.2 Total efficiency (ηt): It is assumed to be 100%. 4.3 Graded outlet oil concentration in water outlet flow (Cd):

Cd = 0 mg/l 4.4 Total outlet oil concentration in water outlet flow (C ):

C = 0 mg/l 4.5 Graded outlet oil concentration in oil outlet flow (Coild): Outlet oil is in the

form of water-free oil. 4.6 Total outlet oil concentration in oil outlet flow (Coil): Outlet oil is in the

form of water-free oil. 4.7 Inlet flow (Q): Inlet flow of the process is equal to the outlet flow of the

upstream process. 4.8 Water outlet flow (Qout): Water outlet flow is equal to water content of inlet

wastewater under the assumption that size of the process is sufficient to handle inlet flowrate on real time basis.

)1000//)(1( oiloout CCQQ ρ−−= m3/h {16.1} 4.9 Oil outlet flow (Qoil): Oil outlet flow is equal to oil content of inlet

wastewater.

outoil QQQ −= m3/h {16.2} 4.10 Customized output:

4.10.1 Required quantity of entrainer: Quantity of entrainer, in this case, water, depends on the type and the qauntity of hydrocarbon to be removed, which can be calculated by the following equation.

)1( H

Hoilsteam y

yQQ−

= m3/h {16.3}

The values of yH (azeotropic composition) of certain hydrocarbons are provided in the help file.


515


IV-95

Table 16-1 Related parameters of stripping



yH Azeotropic composition (Ratio of volume of water in ditillate to total distillate volume). The value for various hydrocarbons are recommended in the help file.

yH Vol/vol



Strip

ping

Stripping (ST-??)

2 ph.vapor

Bubble curve

Pure H2O

Temperature






TH

Fig. 16-1a Icon of stripping Fig. 16-1b Graphical diagram of stripping


7.1 Calculation is based on theoretical equation, regardless of distillation column design.

7.2 The quantity of steam calculated by the program is not included the steam for heating requirement.

7.3 Entrainer ( water) can be continuously reused.

7.4 If possible, pilot-test should be conducted to verify the value of yH and other problems that may be present.

516


IV-96

17) Chemical destabilization, coagulation-flocculation

1. Process abbreviation: CH

2. Process description: Calculation is based on reactor design, which can be applied to any type of chemical or coagulant. Mixing is provided by mean of mechanical mixers. The process is classified as inline concentrator and consists of 1 mixing tank and 3 flocculation tanks. Droplets smaller than 200 microns are assumed to successfully coalesce to form droplets of 200 micros. For bigger oil droplets, they are assumed to remain the same.



4.1 Graded efficiency (ηd): This parameter is actually not valid since no oil is removed from the wastewater. However, the term ηd in this case represents removal efficiency due to coalesce. From this definition, the value of ηd are as follows,

For d < 200 microns, ηd = CF.100%

For d = or > 200 microns, ηd = 0%

However, for d=200 microns, its concentration (if already existed) will increase since coalesced droplets are assumed to be of this size.

4.2 Total efficiency (ηt): This parameter is not valid since no oil is removed.

4.3 Graded outlet oil concentration in water outlet flow (Cd): It is assumed that all oil droplets are coalesced or flocculated to form big oil drop of 200 microns in diameter. Thus

Cd = Co for d > 200 microns and Cd = 0 for other values of “d”.

For d= 200 microns, its Cd is summation of its initial concentration and the sum of concentration of droplets smaller than 200 microns.


C = Co mg/l

4.5 Graded outlet oil concentration in oil outlet flow (Coild): This parameter is not valid since no oil is removed.

4.6 Total outlet oil concentration in oil outlet flow (Coil): This parameter is not valid since no oil is removed.


4.8 Water outlet flow (Qout): Outlet flow of the process is equal to the inlet flow.

4.9 Oil outlet flow (Qoil): This parameter is not valid since no oil is removed.


4.10.1 Retention time:

60⋅=QVτ min {17.1}

517


IV-97

4.10.2 Product between G and τ

4.10.3 Power required (P) or Velocity gradient (G) of each tank: G and P of each tank can be calculated by the following equations, depending which value is specified by users.

5.0

⎟⎟⎠

⎞⎜⎜⎝

⎛=

VPG

μ {17.2}

For turbine mixer, ρ53 DnNP p= {17.3a}

The program features built-in Np for pitched blade turbine mixer and propeller mixer. For paddle mixer,

3)(vANCdP c ⋅⋅⋅⋅= ρ {17.3b} If turbine mixer option is selected, the program will calculate both pitch blade and propeller type.


Table 17-1 Related parameters of chemical destabilization



A Area of one paddle blade (single side) AreaPaddle m2 Cd Drag coefficient of paddle blade Constant = 0.6 D Diameter of impeller of mixer DMixer m G Velocity gradient Grapid,

Gfloc1,2,3 Sec-1

n Rotating speed of mixer RPS1-4 Rev/s N Number of blades NumBlade Np Power coefficient for turbine mixer Built-in P Power required for mixer1 of mixing tank and

mixer 2,3,4 of flocculation tanks Prepid,

Pfloc1,2,3 Watt

v Tip speed of paddle (= 2πn (D/2)) VolRapid, VolFloc1,2,3

m3

V Volume of mixing tank and flocculation tanks VolRapid, VolFloc1,2,3

m3

ρc Density of continuos phase, in this case, water

Denc Kg/m3



518


IV-98

Chemical destabilizaion

(CH-??)

M

MEffluent

FlocculatorRapid mixing

DestabilizationchemicalsInfluent

G = 100-300 s-1

G = 50 s-1 G = 30 s-1 G = 20 s-1

The values of G in the figure are general values.

Fig. 17-1a Icon of chemical destabilization

Fig. 17-1b Graphical diagram of chemical destabilization


7.1 The program assumes that destabilization is successful and the oil droplets coalesce to form bigger drops that can be decanted within 20 min. to 1 hour. So the final droplet size of 200 microns is proposed, Users can change this value in the input screen.

7.2 Dosage of chemical varies with wastewater characteristic, type of chemicals used. Thus it cannot be calculated and must be confirmed by jar test.

7.3 For ANALYSIS mode, The configuration of the process is fixed, i.e. 1 mixing tank and 3 flocculation tanks, all with mechanical mixers.

7.4 For DESIGN mode, the program can be used for sizing 2 types of mixers, i.e. turbine and paddle mixers. Turbine mixers are also divided into 2 cases, i.e. pitched blade turbine and propeller. However, they will be calculated simultaneously by the program.

519


IV-99

18) Biological treatment

1. Process abbreviation: Bio

2. Process description: Biological process in this case is used as a finishing treatment process to treat the effluent of other process. It is included in the program to fulfil the entire process train of oily wastewater treatment. The process is not intended to refer to any specific type of biological system. Its calculation is based on user-specified total TOD and oil removal.



4.1 Graded efficiency (ηd): This parameter is equal to total efficiency.

4.2 Total efficiency (ηt): This parameter is specified by user.


oddd CC )1( η−= mg/l {18.1}


∑=max

min

d

ddCC mg/l {18.2}

4.5 Graded outlet oil concentration in oil outlet flow (Coild): This parameter is not valid since oil is not removed but destroyed or transform into other forms.

4.6 Total outlet oil concentration in oil outlet flow (Coil): This parameter is not valid since oil is not removed but destroyed or transform into other forms.




4.10 Customized output: 4.10.1 TOD of water outlet flow (TOD): It can be calculated by the

following equation. TODo, unit TOD and ηTOD are given by users.

)1())()1(( TODoil

TODoto mg

mgCTODTOD ηη −⋅⋅−−= mg/l {18.1}


520


IV-100

Table 18-1 Related parameters of biological treatment



mgTOD

/mgoil Unit TOD of oil. For more information, see reference file

unitTOD

TODo TOD of the inlet wastewater TODin mg/l ηt Total oil removal efficiency. efft %

ηTOD TOD removal efficiency. It includes only soluble TOD. TOD from oil is not included.

effTOD %



Biologicaltreatment(Bio_??)

Influent

Effluent

Biological reactor

Clarifier(if any)

Wasted biomass(if any)

Fig. 18-1a Icon of biological treatment Fig. 18-1b Graphical diagram of biological treatment


7.1 The process is included in the program to fulfil the entire process train of oily wastewater treatment only. Thus, its calculation is based on data from researches, rather than mathematical model.

7.2 The module is available only in ANALYSIS mode.

7.3 Flowrate of wasted sludge is assumed to be negligible. Thus outlet flow is equal to inlet flow.

7.4 It is assumed that there is no coalescence taking place within the process.

7.5 Calculated TOD is an internal result of the model. It cannot be exported to other process because of programming limitation.

521


IV-101

19) GAC filter

1. Process abbreviation: GAC

2. Process description: GAC filter in this case is used as a finishing treatment process to treat the effluent of other process. It is included in the program to fulfil the entire process train of oily wastewater treatment. Its calculation is based on user-specified TOD removal. For oil removal, the efficiency is assumed to be 100%. Since its main objective is removal of residual pollutant, its inlet oil concentration should be not greater than 50 mg/l. If the inlet wastewater contains higher oil concentration, the program will display warning.



4.1 Graded efficiency (ηd): It is assumed to be 100%.

4.2 Total efficiency (ηt): It is assumed to be 100%.


Cd = 0 mg/l


C = 0 mg/l

4.5 Graded outlet oil concentration in oil outlet flow (Coild): This parameter is not valid since oil adsorbed into the bed.

4.6 Total outlet oil concentration in oil outlet flow (Coil): This parameter is not valid since oil is adsorbed into the bed.



4.9 Oil outlet flow (Qoil): This parameter is not valid.


4.10.1 Adsorptive capacity (q): The program can predict adsorptive capacity, using 2 well-known Isotherm models, i.e.,

• Lungmuir’s model: po

po

bCaC

q+

=1

mg/g {19.1a}

• Frendlich’s model: )/1( npoCkq ⋅= mg/g {19.1b}

A, b, k and n are numerical constant, acquired from experiment. User can select the model from on-screen option button.

4.10.2 Bed life span or total operation time before bed replacement (tT): When isotherm (q VS. Cp relation) and mass transfer zone data (Qa, Ha or E as the functions of Cp) are available;

522


IV-102

)( ppo

abT CCQ

QqHAt

−−⋅⋅

=ρ {19.2a}

Or,

)( ppo

bT CCQ

qHAEt

−⋅⋅⋅

=ρ {19.2b}

)/(1010011 AQdc

HE ⋅⋅⋅

⋅−= {19.3}

E is effective saturation of bed. The recommended valued is around 50-95% (average 75%). In the program, eq. 19.2b is used. H and V are in m and m/h, respectively.

4.10.3 Hydraulic loading rate:

AQV = m/h {19.4}

4.10.4 Retention time (τ):

60)/(

⋅=AQ

Hτ min {19.5}

4.10.5 Headloss (P): Head loss of the bed can be calculated from Kozeny-Carman’s equation. Built-in minimum and maximum void ration is 0.18 to 0.25, respectively.

32

2)1)(60/(180ερ

εμ⋅⋅⋅

−=

dpgVHP c {19.3}


523


IV-103

Table 19-1 Related parameters of GAC filter



A Cross section area of bed A m2 Cp Required outlet pollutant concentration Polout mg/l Cpo Inlet concentration of pollutant Polin mg/l dp GAC particle size Dpc m

Ε Effective saturation of bed, E < 100% GACE % H Bed height GACH m P Pressure drop Pdropmin,

Pdropmax m

q Isotherm data (mass of adsorbate/ mass of adsorbent)

GACq mg/g

V Empty bed velocity or hydraulic loading rate GACV m/h a, b Empirical constant for Lungmire’s model Cpo in mg/l m/h k, n Empirical constant for Freundlich’s model Cpo in mg/l m/h c, d Empirical constant for mass transfer zone data Cpo in mg/l m/h

ε Bed porosity or void ratio, recommended value is 0.18 – 0.25.

Void

ρc Density of continuous phase Denc kg/m3

ρb Bulk density of bed (430 – 600 kg/m3) DenGAC kg/m3



GAC filter(GAC-??)

Ha

CCoCe

H

t = 0

Satu

red

zone

MTZ

CCoCe

H

t = tT

Ha

Bed needs to be replaced.

GAC bed.HT

Influent

Effluent

Velocity=V

Fig. 19-1a Icon of GAC filter Fig. 19-1b Graphical diagram of GAC filter

524


IV-104


7.1 Calculation is based on research data, given by users.

7.2 If possible, it is recommended to conduct is lab test, using real wastewater, to find the exact data about isotherm adsorptive capacity (a, b, k, n), and mass transfer zone (c and d) data of GAC bed.

525


IV-105

20) Customized concentrator

1. Process abbreviation: CC

2. Process description: The process works as an oil concentrator, e.g. hydrocyclone. But users are allowed to specify graded efficiency of the process.

3. Reference : None


4.1 Graded efficiency (ηd): User-defined in the form of a table. Graded efficiency of the droplet size that falls between 2 specified values will be obtained by interpolation or extrapolation.


∑ ⋅−

−=

max

min

%100)1(

)1(1 d

d f

odd

ot R

CC

ηη {20.1}


)1()1(

f

oddd R

CC

−−

=η mg/l {20.2}


∑=max

min

d

ddCC mg/l {20.3}

4.5 Graded outlet oil concentration in oil outlet flow (Coild):

f

oddoild R

CC η= mg/l {20.4}


∑=max

min

d

doildoil CC {20.5}



)1( fout RQQ −= m3/h {20.6}


foil RQQ ⋅= m3/h {20.7}

4.10 Customized output: None


526


IV-106

Table 20-1 Related parameters of customized concentrator



Rf Split ratio (Qoil/Q) Rf d, ηd Specified droplet size and graded efficiency

of the process Customdin, Customeffd

Micron, %


Customconcentrator

(CC-??) Oil outlet

Influent

Effluent

Fig. 20-1a Icon of customized concentrator

Fig. 20-1b Graphical diagram of customized concentrator

527


IV-107

21) Customized oil separator

1. Process abbreviation: CS

2. Process description: the process works as an oil separator, e.g. decanter. But users are allowed to specify graded efficiency of the process.

3. Reference : None


4.1 Graded efficiency (ηd): User-defined in the form of a table. Graded efficiency of the droplet size that falls between 2 specified value will be obtained by interpolation.


( )%100

max

min ⋅⋅

=∑

o

d

dodd

t C

Cηη {21.1}

4.3 Graded outlet oil concentration in water outlet flow(Cd):

)1( dodout

d CQQC η−⋅⋅= mg/l {21.2}


∑=max

min

d

ddCC mg/l {21.3}





)1000//)(1( oiloout CCQQ ρ−−= m3/h {21.4}





528


IV-108

Table 21-1 Related parameters of customized separator



d, ηd Specified droplet size and graded efficiency of the process

Customdin, Customeffd

Micron, %


Custom separator(CS-??)

Influent

Effluent

Oil outlet

Fig. 21-1a Icon of customized separator Fig. 21-1b Graphical diagram of customized separator



529


IV-109

22) Customized inline concentrator

1. Process abbreviation: IC

2. Process description: the process work as an inline oil concentrator, e.g. coagulation-flocculation, or inline coalescer. But users are allowed to specify graded efficiency of the process.



4.1 Graded efficiency (ηd): This parameter is actually not valid since no oil is removed from the wastewater. However, the term ηd in this case represents removal efficiency due to coalesce, defined by user in the form of a table. Graded efficiency of the droplet size that falls between 2 specified value will be obtained by interpolation or extrapolation. However the concentrated oil is not separated from the wastewater but assumed to coalesce to form bigger oil drops of a specified size (dcoalesce). From this definition, the value of ηd are as follows,

For d < dcoalesce, ηd is as specified by user. The rest of the droplets that are not separated will partially coalesce to form bigger droplets at the size of dcoalesce.

For d = or > dcoalesce, ηd is as specified by user.

However, for d = dcoalesce, its concentration (if already existed) will increase since coalesced droplets are assumed to be of this size.

4.2 Total efficiency (ηt): This parameter is not valid since no oil is removed.

4.3 Graded outlet oil concentration in water outlet flow (Cd): It is assumed that separated oil droplets coalesce to form big oil drop which is classified as oil layer.

And as described above, it is assumed that there are partially coalesces of the small oil droplets. These droplets will coalesce and form bigger oil droplets “dcoalesce”. Or all droplets smaller than dcoalesce will coalesce to form the droplet size “dcoalesce”.

Thus, for d < dcoalesce,

0=dC mg/l {22.1a}

For d = dcoalesce,

∑ −=coalsced

dddd CC

min

)1( η mg/l {22.1b}

For d > dcoalesce,

)1( dodd CC η−⋅= mg/l {22.1c}

For oil layer,

∑+=max

min

,

d

ddlayerolayer CCC mg/l {22.1d}

530


IV-110


C = Co mg/l

4.5 Graded outlet oil concentration in oil outlet flow (Coild): This parameter is not valid since no oil is removed.

4.6 Total outlet oil concentration in oil outlet flow (Coil): This parameter is not valid since no oil is removed.






Table 22-1 Related parameters of customized inline concentrator



dcoalesce Diameter of partially coalesced oil drop from the process

customDcoalesce micron

d, ηd Specified droplet size and graded efficiency of the process

Customdin, Customeffd

Micron, %


Inline concentrator(IC-??)

Influent

Effluent

Fig. 22-1a Icon of inline concentrator Fig. 22-1b Graphical diagram of inline concentrator

531


IV-111

23) Inlet

1. Process abbreviation: IN

2. Process description: This module is used to input the wastewater into a process train. There can be more than one input module in a process train.

3. Reference : None

4. Input parameters



Cod Granulometry or oil droplet size distribution : graded concentration (concentration of oil at each droplet size)

Cind mg/l

Concentration of inlet oil in the form of oil layer or film

Cinlayer mg/l

Co Total oil inlet concentration Co mg/l d Granulometry or oil droplet size distribution :

droplet size din micron

Q Wastewater flowrate Qin m3/h T Temperature temp Celcius ρc Dynamic (or absolute) viscosity of continuous

phase, which is water, for oily wastewater Denc Kg/m3

ρd Density of dispersed phase, in this case, oil Dend Kg/m3 μC Dynamic (or absolute) viscosity of continuous

phase, which is water, for oily wastewater Muc N.s/m2

(= 1000 cp) μd Dynamic (Absolute) viscosity of dispersed

phase, in this case, oil Mud N.s/m2

(= 1000 cp) γow Interfacial tension between oil and water gow kg/s2 or N/m

(= 1000 dyne/cm)

5. Related graphics: Icon in analysis mode for this process are as shown below.

InletInlet

Inlet(IN-??)

Fig. 23-1a Icon of inlet


6.1 There can be more than one inlet module in a process train.

6.2 Graded concentration is specified as quantity of oil (mg) per volume of wastewater (= volume of oil+water).

532


IV-112

24) Outlet

1. Process abbreviation: -

2. Process description: This module is used to specify the outlet point of the process train. It is also used as a logical checkpoint of the program to perform calculation. Thus there can be only one outlet in a process train.

3. Reference : None


OutletOutlet

Outlet

Fig. 24-1a Icon of outlet


5.1 There can be only one outlet in a process train.

5.2 Location of the outlet in the process train does not need to be an actual location. However, it is recommended to place it at the end of main process stream for faster calculation.

533


IV-113

25) Flow merge

1. Process abbreviation: FM

2. Process description: This module is used to combine or merge any 2 streams of wastewater together.

3. Reference : None

4. Output parameters:

4.1 Wastewater flowrate (Q): It is calculated from the summation of the flowrate of the two streams.

4.2 Granulometry or size distribution of oil droplets in the wastewater: It is calculated from weight average value, as shown in the following equation.

21

22,11,

QQQCQC

C ddd +

+= {25.1}

4.3 Total oil concentration (C):

∑=max

min

d

ddCC {25.2}

4.4 Oil density (ρd): as kg/m3. It is assumed to be equal to the weight average value, as shown in the following equation.

2211

222,111,

QCQCQCQC dd

d ++

=ρρ

ρ {25.3}

4.5 Water density (ρc):

21

2211

QQQQ cc

c ++

=ρρρ {25.4}

4.6 Oil viscosity (μd): as N.s/m2. It is assumed to be equal to the weight average value, as shown in the following equation.

2211

222,111,

QCQCQCQC dd

d ++

=μμ

μ {25.5}

4.7 Water viscosity (μc):

21

2211

QQQQ cc

c ++

=μμμ {25.6}

4.8 Temperature (T):

21

2211

QQQTQTT

++

= Celcius {25.7}

4.9 Interfacial tension (γow): It is assumed to be equal to the weight average value, as shown in the following equation.

2211

222,111,

QCQCQCQC owow

ow ++

=γγ

γ {25.8}

5. Related graphics: Icon in analysis mode for this process are as shown below.

534


IV-114

2

1

Flow merge(FM-??)

Fig. 25-1a Icon of flow merge

535


IV-115

26) Flow split

1. Process abbreviation: FS

2. Process description: This module is used to divide or split a stream of wastewater into 2 streams.

3. Reference : None

4. Output parameters:

4.1 Wastewater flowrate (Q1 and Q2):

QRQ f=1 {26.1a}

QRQ f−= 12 {26.1b}

4.2 Granulometry or size distribution of oil droplets in the wastewater:

oddd CCC == 2,1, {26.2}

4.3 Total oil concentration (C):

oCCC == 21 {26.3}

4.4 Oil density (ρoil):

oiloiloil ρρρ == 2,1, {26.4}

4.5 Temperature (T):

TTT == 21 {26.5}

4.6 Interfacial tension (γow): It is assumed to be equal to the weight average value, as shown in the following equation.

owowow γγγ == 2,1, {26.6}


2

1

Flow split(FS-??)

2

1

InfluentQ1 = RfQ

Q2 = 1- RfQ

Fig. 26-1a Icon of flow split Fig. 26-1b Graphical diagram of flow split

536

Part V General conclusion

General conclusion

V-1

General conclusion

One of the main objectives of this thesis is to review all of the researches on various oily wastewater treatment processes, directed by Prof. AURELLE and summarize and establish general theory or mathematical models that govern the performance and design of such processes.

This objective is accomplished in Part 1 to Part 3 of this thesis. Every unit processes for oil separation, studied in GPI lab, had been revised and their corresponding mathematical models as well as design limitation and influent parameters had been found. These processes revised under the scope of work of this thesis include,

1. Oil skimmer

This equipment is developed to selectively remove oil layer from the water surface without carrying over the water with it. It is found that the key to achieve good oil selectivity depends on the surface energy or critical surface tension of skimmer material. The material with low critical surface tension is suitable to use as skimmer material. Mathematical models of 2 types of skimmer, i.e. drum skimmer and disk skimmer are verified.

2. Decanter

Its underlining theory, namely STOKE’s law, is quoted. General mathematical models for 3 different varieties of the processes, i.e. simple decanter (e.g. API tank), parallel plate interceptor (e.g. PPI and lamella separator), and compact decantor (e.g. Spiraloil) are proposed and verified.

3. Coalescer

Theoretical model, proposed by AURELLE, based on filtration model, is reviewed. The model provides clear ideas about influent parameters on coalescer performance. Several initiative ideas for coalescer performance enhancemen, e.g. the use of guide to increase oil loading of the coalescer and the use of mix bed material to treat mixture of direct/inverse emulsion simultaneously are realised. Finally, the empirical models, which are verified by relatively wide range of data, are proposed for 4 varieties of coalescer, i.e. granular bed coalescer, brush type bed coalescer, dynamic fibrous bed (rotating brush) coalescer, and disorderly fibrous bed (metal wool) coalescer. Headloss equations are also proposed in this thesis.

4. Dissolved air flotation

Mathematical model based on filtration concept, proposed by SIEM [12] is reviewed. The model is useful for understanding effects of influent parameters on DAF performance. Extension equations of SIEM’s model, based on population balance theory, is established within this thesis to extend the valid range of SIEM’s model to cover the range of oily wastewater frequently found. Equations for saturator or pressurized water system design are also proposed and verified within this thesis.

5. Hydrocyclone

Trajectory analysis model for 2-phase (liquid-liquid) hydrocyclone, the new concept based on STOKE’s law, is proposed and verified. The model can actually predict

537

General conclusion

V-2

graded oil removal efficiency of any sizes of oil droplet in the wastewater by theoretical based equation, unlike other models which are based on curve fitting and similarity concepts. Thus it is very useful to understand the effect of related parameters on hydrocyclone performance. Models of three-phase hydrocyclone, GPI innovation for simultaneous removal of oil, suspended solids amd water, are also established within the scope of work of this thesis. The model is also based on trajectory analysis concept. Finally, headloss equations for 2- and 3-phase hydrocyclone are established in this thesis.

6. Membrane processes

Applications of membrane processes, i.e. microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO) on oily wastewater treatment in GPI lab had been reviewed. Several useful facts from those researches are realized and calculation techniques are established in this thesis, i.e.,

• Flux enhancement for UF on cutting oil emulsion treatment by partial destabilization. Additional of salt, in lower amount than that for total destabilization, into the feed helps increasing permeate flux, thus saving on energy consumption and membrane size. The downside of this technique, concluded in this thesis, is that the destabilized oil can clog the membrane if it is not properly partition or removed from recirculated feed stream.

• Cleaning of UF membrane after treatment of cutting oil by cleaning microemulsion. It can effectively wash accumulated oil or foulants from the membrane. It can be reused for many times until it is saturated by oil.

• Technique to extend the UF performance data at 1 condition to cover other condition, based on combination of film model and resistance model. This technique is proposed and verified in this thesis. It is useful when only limited data on the wastewater is available. It allows us to estimate evolution of flux and permeate volume with time, which is useful for process design.

• Technique to predict flux of mixture of two different emulsion. This technique is also proposed and verified in this thesis. It is useful when mixed emulsion can be expected and its ratio is likely to vary.

7. Thermal process

The main interest of this type of process is heteroazeotropic distillation (HD). Application of HD to remove water content from refinery slops and UF retentates from UF of cutting oil emulsion is reviewed. The process is achieved by addition of certain chemical that promotes azeotropic formation (called entrainer), usually hydrocarbons, into the wastewater. It will lower the boiling temperatur of the system, thus save the energy. These applications provide prospects on re-value these supposed-to-be-wastes. Its reverse application, namely stream stripping (for removing volatile substance from water by addition of stream), is also realized. Theoretical data to calculate required amount of entrainer are also proposed.

8. Chemical process

Destabilization (or breaking or cracking) mechanism of emulsion by addition of various chemicals, i.e. mono valence salt, bivalence salts, polyelectrolytes, acid and special absorbates are reviewed. Equations for mixing tank and mixer designs are also reviewed and included in the program.

538

General conclusion

V-3

9. Finishing processes

Two widely used finishing processes, i.e., biological treatment and carbon adsorption, is realized. Useful data on biological process, related to oily wastewater treatment, and design equations of GAC filter, as well as absorptive capacity of certain chemicals, are reviewed and included in the program.

To complete the first objective, guideline on oily wastewater treatment process selection and recommended treatment processes for certain oily wastewater are proposed.

The final objective of the thesis is to develop the program for calculation, design and simulation of wastewater treatment process train in order to value and make use of the know-how and significant finding from the researches by presenting them in the form of user-friendly program. To fulfil this objective, the program, namely GPI program, is developed. It is divided into 4 major modes, i.e.,

• E-book mode: provides background knowledge and useful database about the oil pollution and the treatment processes. Actually, the textbook in part 3 is transformed into e-book files used in this mode,

• Process recommendation mode: provides recommendation to narrow the range of feasible processes for any input influent. Selection criteria are as proposed in the guideline in Part 3, chapter 12.

• Design (calculation) mode: used for sizing the unit process. The models used for calculation are as summarized in Part 3.

• Analysis (simulation) mode: allows users to integrate any separation processes, included in the program database, to build their own treatment process train. And the program will simulate the process train to forecast the efficiency of each unit.

The program is developed to be upgradable. Its architecture consists of the database in the form of common text database file and sub-programs. To upgrade the program, it can be done conveniently by adding the data, such as name of new process, its related parameter needed for calculation, into the database. The program will link the new process into the graphic user interface automatically. For sub-program for calculation of the new process, it can be separately developed using Visual Basic programming language. The easiest way is to copy the source code of an existing process and change the equation to suit the new process. After compilation in to an executable file, it can be copied to replace the old GPI program file without re-installation of the program. So the program could be further developed to cover more researches and processes in the future.

539

Reference

Reference 1 Sawyer C., Mccarty P., Parkin G.

Chemistry for environmental engineering and science Book, 5th edition, Mcgraw-Hill, 2003

2 Perry R., Green D. Perry's chemical engineering's handbook Book, 7th edition, Mcgraw-Hill, 1997

3 Aurelle Y. Contribution a l'etude des mechanismes fondamentaux de la coalescence des emulsions sur lit granulaire Thèse de doctuer d'etat, Université Pual Sabatier,1980

4 Cherid S. Conception et etude de nouveaux separateurs lamellaires hydrocarbon-eau de type "SPIRALOIL" Thèse de doctorat, INSA-Toulouse, 1986

5 Thangtongtawi S. Contribution a l'etude et au dimensionnement de recuperateurs de surface du type tambours et disques deshuileurs Thèse de doctorat, INSA-Toulouse, 1988

6 Sanchez Martinez J.C. Application de la coalescence a l'extraction liquide-liquide Thèse de doctorat, INSA-Toulouse, 1982

7 Darme C. Separation par coalescence des emulsions hydrocarbure/eau stabilees par des agents tensio-actifs Thèse de doctorat, INSA-Toulouse, 1983

8 Tapaneeyangkul P. Etude et modilisation d'une nouvelle generation de coalesceurs liquide-liquide: "Le coalesceur a garnissage fibreux dynamique" Thèse de doctorat, INSA-Toulouse, 1989

9 Damak L. Nouveaux separateurs d'emulsions huile/eua chargee en matiers en suspension - Separateur a inversion de phase - coalesceur granulaire a lit pulse Thèse de doctorat, INSA-Toulouse, 1992

10 Srijaroonrat P. Nouvelles techniques de traitement des emulsion hydrocarbure-eau non stabilisee: ultrafil tration et couplage coalesceur-hydrocyclone Thèse de doctorat, INSA-Toulouse, 1998

11 Wanichkul B. Etude des potentialities de nouveaux procedes de traitement d'emulsions hydrocarbure-eau: ultrafiltration, distillation et couplage coalesceur-hydrocyclone Thèse de doctorat, INSA-Toulouse, 2000

540

12 Siem N. Contribution a l'etude des separations hydrocarbure-eau par flottation Thèse de doctorat, INSA-Toulouse, 1983

13 Aoudjehane M. Traitement par flottation des effluents aqueux charges en hydrocarbures Thèse de doctorat, INSA-Toulouse, 1986

14 Dupre V. Etude des mecanismes et de la modilisation procedes de flottation a air dissous Thèse de doctorat, INSA-Toulouse, 1995

15 Ponasse M. Elements theoriques et experimentaux en vue de l'amelioration des installations de flottation a air dissous. Developpement de nouveaux flottateurs Thèse de doctorat, INSA-Toulouse, 1997

16 Ma B. Epuration des eaux residuraires de l'industrie petreliere Par hydrocyclonage

Thèse de doctorat, INSA-Toulouse, 1993

17 Cazal E. Etude des potentialities des hydrocyclones dans domaine du traitement des eaux residuaire urbains et des eaux de ruissellement Thèse de doctorat, INSA-Toulouse, 1996

18 Belkacem M. Nouvelle methodologie dans le traitement des huiles de coupe par ultrafiltration Thèse de doctorat, INSA-Toulouse, 1995

19 Toulgoat K. Etude des macanismes de formation des emulsions thermiques; potentialites de separation par ultrafiltration Thèse de doctorat, INSA-Toulouse, 1996

20 Matamoros H. Procedes membranaires pour le traitment des emulsions stabilisees de type fluid de coupe Thèse de doctorat, INSA-Toulouse, 1997

21 Zhu S. Etude des traitement physico-chimiques d'epuration des emulsion d'huile de coup; Influence de leur formulation Thèse de doctorat, INSA-Toulouse, 1990

22 Yang C. Developpement de nouvelles formulations de fluides de coupe peu polluants, mise au point de techniques de traitement adaptees Thèse de doctorat, INSA-Toulouse, 1993

23 Lorain O. Contribution a l'etude du traitement des eaux par congelation : potentialites et applications Thèse de doctorat, INSA-Toulouse, 2002

541

24 Lucena E. Nouveau procede de traitement des slops de l'industrie petroliere par distillation heteroazeotropique: comparaison avec d'autres procedes Thèse de doctorat, INSA-Toulouse, 2004

25 Puprasert C. Contribution a la mise au point d'application specifique des hydrocyclones en traitement des eaux Thèse de doctorat, INSA-Toulouse, 2004

26 Aurelle Y. Contribution a l'etude du traitement des eaux polluees par des hydrocarbures emulsionnes par coalescence sur resines oleophiles Thèse de doctorat, Universite Pual Sabatier de Toulouse, 1974

27 Tan Prapon Theoritical study of Coalescer and its application to treatment of wastewater for vegetable oil industry Master degree thesis, Chulalongkorn University, Thailand, 1984

28 Thew M.T. Hydrocyclones for liquid-liquid separation Lecture on the intensive short course, University of Bath, September 1984

29 Chebelin T. Separation compacte: modilisation des hydrocyclones Document de stade de fin d'etude, INSA-Toulouse, 1984

30 Bradley D. The hydrocyclone Book, 1965

31 M.a B., Aurelle Y., Seureau J. Three phase hydrocyclone for simultaneous separation of solids from liquid-liquid mixtures Publication, source unknown

32 Ma B., Aurelle Y., Wolbert D. Efficiency estimation of liquid-liquid hydrocyclones using trajectory analysis AIChE journal, Vol. 41, No. 6, 1995

33 Trawinski H. Hydrocyclones Publication, source unknown

34 Leppinen D.M., Dalziel S.B., Linden P.F. Modeling the global efficiency of dissolved air flotation Water science and Technology, Vol. 43 No. 8, 2001

35 Yao K., Habibian M.T., O'melia C.R. Water and wastewater filtration: Concepts and application Environmeantal sciences and Technology, No. 11, 1971

36 Henke L. Osmosis, RO and Filtration: What they have in common? Asian Water, Vol. 19 No.5, June, 2003

542

37 Mutsui Y., Fukushi K., Tambo, N. Modeling, simulation and operational parameters of dissolved air flotation J. water SRT, No. 47, 1998

38 Cheryan M. Ultrafiltration and microfiltration handbook Book, Technomic, 1998

39 Pushkarev V.V., Yuzhaninov A.G., Men S.K. Treatment of oil containing wastewater Book, Allerton press inc., 1983

40 Shaw D. Introduction to colloid and surface chemistry Book, 4th edition, Butterworth-Heinemann Ltd., 1992

41 Aurelle Y. Physical and chemical treatment techniques Paper on short course seminar: Faculty of engineering, Chulalonkorn university, Thailand Book, Chulalongkorn University, September 1985

42 Aurelle Y. Treatments of oil containing wastewater Chulalongkorn University

43 Parker H.D, Pitt G.D. Pollution control instrument for oil and effluents Book, Graham & Trotman, 1987

44 Happel J. Viscous flow in multiparticle systems : slow motion of fluids relative to beds of spherical particles AIChE journal, Vol. 4 No. 2, 1985

45 American petroleum institute (API) Manual on disposal of refinery wastes: Volume on liquid wastes Book, 1st edition, 1969

46 Metcalf & Eddy Wastewater engineering: Treatment, disposal, reuse Book, 3rd edition, Mcgraw-Hill, 1991

47 Jones H.R. Pollution control in petroleum industry Book, Noyes data corperation , 1973

48 Kiuru H.J. Development of dissolved air flotation technology from the first generation to the newest (third) one (DAF in turbulent flow conditions) Water science and Technology, Vol. 43 No.8, 2001

49 Degrémont Wastewater treatment handbook Book, 5th edition, John Wiley & son, 1979

50 Luthy R.G., Selleck R.E., Galloway T.R. Removal of emulsified oil with organic coagulants and dissolved air flotation J. - Water Pollut. Control Fed., Vol. 50:2, 1978

543

51 Eckenfelder W. Jr. Industrial water pollution control Book, 3rd edition, Mcgraw-Hill, 2000

52 Souzaa F. J., Silveira N. A. Preliminary results of large eddy simulations of hydrocyclone Engenharia Térmica (Thermal Engineering), Vol.3 No.2, December 2004

53 Svarovsky L., Thew M. Hydrocyclones: Analysis and applications Book, Kluwer academics publishers, 1992

54 Scott K., Hughs R. Industrial membrane separation technology Book, 1st edition, Blackie academic & professional, 1996

55 Fox R.W., Mcdonald A.T. Introduction to fluid mechanics Book, 3rd edition, John Wiley & son, 1985

56 Jain R.K., Aurelle Y., Cabassud C., Roustan M., Shelton S.P. Environmeantal technologies and trends

Book, Springer, 1997

57 Judd S., Jefferson B. Membranes for wastewater recovery and re-use Book, Elsevier , 2003

58 Murkes J., Carlsson C.G. Crossflow filtration Book, John Wiley & son, 1988

59 Porter M. Handbook of industrial membrane technology Book, Noyes publications, 1990

60 Null H. Phase equilibrium in process design Book, John Wiley & son, 1970

61 Smith J.M.,Van Ness H.C., Abbot M.M. Introduction to chemical engineering thermodynamics Book, 6th edition, Mcgraw-Hill, 2001

62 Lucena E., Verdun P, Aurelle Y. Nouveau procédés de valorisation des slops de raffineries et dechets huileux par distillation heteroazeotropique Publication, source unknown

63 US department of commerce, National technical information service Manual of practice for wastewater neutralization and precipitation

book, August 1981

64 Bellivier D., Cazes C. Dimensionnement d'une filière de traitement d'eau de gisement Projet calcul. 5éme annee, INSA Toulouse, 2001

544

65 Tuntulawes M. Potable water engineering Book, 1st edition (Thai version), 1984

66 Cheng J. W., Schechter, R., Malina, J., Gloyna, E. Separation of free oil following coalescer Technical report CRWR-143 EHE 77-01, University of Texas at Austin, January 1977

67 Schweitzer P. A. Handbook of separation techniques for chemical engineering Book, McGraw-Hill, 1979

68 Rachu S. Computer program development for hydraulic design and analysis of wastewater treatment plants Master degree thesis, Chulalongkorn university, Thailand, 1997

69 Pakdeewatanakul K., Krue-usaha C. Visual Basic 6: Programmer edition (Thai version) Book, D.K. book house (Thailand), 2000

70 Webb J., Jacobs R. L. Using Visual Basic for Applications: Excel edition Book, Que Corp., 1994

545

Annexe

Annex A.1.1 Comparison between calculated efficiency and experimental resultfor simple spiral "Spiraloil" decanterRef. [4] pp.79

Case Droplet diameter

Empty bed

velcity

Calculated removal efficiency

Observed efficiency from experiment(by

droplet size)

Difference in calculated and

observed efficiencyd V ηd ηd observed

micron m/s % %1 1.8 0.4 1.04% 0.00% -1.04%

2.3 0.4 1.71% 0.00% -1.71%2.95 0.4 2.81% 0.50% -2.31%3.7 0.4 4.42% 21.42% 17.00%

4.65 0.4 6.99% 51.80% 44.81%5.9 0.4 11.25% 59.92% 48.67%

7.45 0.4 17.95% 68.96% 51.01%9.45 0.4 28.57% 82.62% 54.05%

11.85 0.4 45.41% 93.20% 47.79%14.95 0.4 72.29% 97.98% 25.69%18.85 0.4 100.00% 99.35% -0.65%23.7 0.4 100.00% 100.00% 0.00%

29.85 0.4 100.00% 100.00% 0.00%37.65 0.4 100.00% 100.00% 0.00%47.5 0.4 100.00% 100.00% 0.00%55.5 0.4 100.00% 100.00% 0.00%

2 1.8 0.8 0.52% 0.00% -0.52%2.3 0.8 0.85% 0.00% -0.85%

2.95 0.8 1.40% 0.00% -1.40%3.7 0.8 2.21% 0.00% -2.21%

4.65 0.8 3.49% 1.70% -1.79%5.9 0.8 5.62% 14.25% 8.63%

7.45 0.8 8.96% 48.01% 39.05%9.45 0.8 14.26% 49.72% 35.46%

11.85 0.8 22.67% 72.10% 49.43%14.95 0.8 36.09% 85.20% 49.11%18.85 0.8 57.37% 94.00% 36.63%23.7 0.8 90.70% 98.90% 8.20%

29.85 0.8 100.00% 100.00% 0.00%37.65 0.8 100.00% 100.00% 0.00%47.5 0.8 100.00% 100.00% 0.00%55.5 0.8 100.00% 100.00% 0.00%

3 1.8 1.6 0.26% 0.00% -0.26%2.3 1.6 0.42% 0.00% -0.42%

2.95 1.6 0.70% 0.00% -0.70%3.7 1.6 1.10% 0.00% -1.10%

4.65 1.6 1.74% 0.00% -1.74%5.9 1.6 2.81% 0.00% -2.81%

7.45 1.6 4.48% 0.00% -4.48%9.45 1.6 7.23% 0.00% -7.23%

11.85 1.6 11.33% 5.40% -5.93%14.95 1.6 18.04% 26.20% 8.16%18.85 1.6 28.35% 54.80% 26.45%23.7 1.6 45.35% 80.20% 34.85%

29.85 1.6 71.94% 95.30% 23.36%37.65 1.6 100.00% 98.80% -1.20%47.5 1.6 100.00% 99.90% -0.10%55.5 1.6 100.00% 100.00% 0.00%

546

Annex A1.2 Comparison between calculated efficiency and experimental result for mixed spiral "Spiraloil" decanterRef. [4] pp.88

Case Droplet diameter

Empty bed

velcity

Calculated removal efficiency

Observed efficiency from experiment(by

droplet size)

Difference in calculated and

observed efficiencyd V ηd ηd observed

micron m/s % %1 1.80 0.5 1.88% 52.22% 50.34%

2.30 0.5 3.08% 58.83% 55.75%2.95 0.5 5.07% 67.54% 62.47%3.70 0.5 7.97% 72.08% 64.11%4.65 0.5 12.60% 79.89% 67.29%5.90 0.5 20.29% 84.97% 64.68%7.45 0.5 32.35% 89.00% 56.65%9.45 0.5 51.50% 94.01% 42.51%

11.85 0.5 81.85% 97.02% 15.17%14.95 0.5 100.00% 98.22% -1.78%18.85 0.5 100.00% 99.40% -0.60%23.70 0.5 100.00% 100.00% 0.00%29.85 0.5 100.00% 100.00% 0.00%37.65 0.5 100.00% 100.00% 0.00%47.50 0.5 100.00% 100.00% 0.00%53.00 0.5 0.00%

2 1.80 1.52.30 1.52.95 1.53.70 1.54.65 1.5 4.20% 43.18% 38.98%5.90 1.5 6.76% 61.82% 55.06%7.45 1.5 10.78% 67.27% 56.49%9.45 1.5 17.17% 71.82% 54.65%

11.85 1.5 27.28% 75.45% 48.17%14.95 1.5 43.18% 77.27% 34.09%18.85 1.5 70.00% 84.54% 14.54%23.70 1.5 100.00% 92.73% -7.27%29.85 1.5 100.00% 100.00% 0.00%37.65 1.5 100.00% 100.00% 0.00%47.50 1.5 100.00% 100.00% 0.00%55.50 1.5

547

Annex A2.1 Comparison between calculated efficiency and experimental result for granular bed coalescer

Ref. [3] pp.285 & 321, [9], pp.174Case Total

hydrocarbon concentration

Droplet diameter

Bed media

diameter

Bed height

Dispersed phase density

Continuous phase density

Density difference

Dynamic viscosity

of dispersed

phase

Dynamic viscosity of continuous

phase

Empty bed

velcity

Interfacial tension

Calculated removal

efficiency from

DAMAK's model

Observed efficiency

from experiment(by droplet size)

Co d dp H ρd ρc Δρ μd μc V γo/w ηd ηd observed

mg/l m m m kg/cu.m kg/cu.m kg/cu.m (N.s)/m2 (=1000cP

(N.s)/m2 (=0.1000cP)

m/s N/m (=1000dyn/cm

% %

1 1000 41.00 0.37 0.01 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.79% 95.00%

1000 41.00 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.70%

1000 41.00 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 41.00 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 32.65 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.99%

1000 32.65 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 32.65 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 25.90 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.00%

1000 25.90 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 25.90 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 20.57 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.18% 98.00%

1000 20.57 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 20.57 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 16.36 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 94.74% 96.00%

1000 16.36 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 16.36 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 12.99 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 90.47% 94.30%

1000 12.99 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 96.19% 99.70%

1000 12.99 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 10.30 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 86.37% 92.00%

1000 10.30 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 91.83% 98.70%

1000 10.30 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.79% 100.00%

2 1000 41.00 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.60%

1000 41.00 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 41.00 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.40%

1000 32.65 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.00%

1000 32.65 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 32.65 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.40%

1000 25.90 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.60%

1000 25.90 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 25.90 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.00%

1000 20.57 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.18% 97.00%

1000 20.57 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 20.57 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.60%

1000 16.36 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 94.74% 94.60%

1000 16.36 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 99.60%

1000 16.36 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.60%

1000 12.99 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 90.47% 92.60%

1000 12.99 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 96.19% 99.20%

1000 12.99 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 98.60%

1000 10.30 0.37 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 86.37% 88.80%

1000 10.30 0.37 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 91.83% 97.40%

1000 10.30 0.37 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.79% 98.20%

3 1000 41.00 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 93.83% 97.00%

1000 41.00 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.76% 99.00%

1000 41.00 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 32.65 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 89.66% 93.00%

1000 32.65 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 95.32% 97.00%

548




Droplet diameter

Bed media

diameter

Bed height



Density difference

Dynamic viscosity

of dispersed

phase


phase

Empty bed

velcity

Interfacial tension

Calculated removal

efficiency from

DAMAK's model

Observed efficiency




(N.s)/m2 (=0.1000cP)


% %

1000 32.65 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 100.00% 100.00%

1000 25.90 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 85.60% 90.20%

1000 25.90 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 91.01% 96.40%

1000 25.90 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 98.90% 100.00%

1000 20.57 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 81.74% 83.20%

1000 20.57 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 86.91% 94.40%

1000 20.57 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 94.45% 99.40%

1000 16.36 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 78.08% 74.80%

1000 16.36 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 83.02% 91.20%

1000 16.36 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 90.22% 98.00%

1000 12.99 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 74.56% 61.00%

1000 12.99 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 79.28% 87.20%

1000 12.99 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 86.15% 96.60%

1000 10.30 0.60 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 71.18% 58.20%

1000 10.30 0.60 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 75.68% 84.00%

1000 10.30 0.60 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 82.25% 93.40%

4 1000 41.00 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 85.82% 89.60%

1000 41.00 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 91.25% 98.00%

1000 41.00 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 99.16% 100.00%

1000 32.65 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 82.00% 85.20%

1000 32.65 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 87.18% 93.20%

1000 32.65 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 94.75% 100.00%

1000 25.90 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 78.29% 75.80%

1000 25.90 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 83.24% 96.20%

1000 25.90 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 90.46% 99.60%

1000 20.57 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 74.76% 61.00%

1000 20.57 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 79.49% 95.40%

1000 20.57 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 86.38% 99.00%

1000 16.36 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 71.42% 56.00%

1000 16.36 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 75.93% 90.80%

1000 16.36 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 82.52% 96.60%

1000 12.99 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 68.20% 42.60%

1000 12.99 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 72.51% 83.20%

1000 12.99 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 78.80% 90.80%

1000 10.30 0.75 0.03 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 65.10% 24.60%

1000 10.30 0.75 0.05 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 69.22% 78.80%

1000 10.30 0.75 0.1 793.4 998 204.6 1.07E-03 1.08E-03 0.00351 4.20E-02 75.22% 84.20%

5 1000 10.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0009 1.10E-02 90.55% 95.00%

1000 14.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0009 1.10E-02 96.85% 96.50%

1000 20.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0009 1.10E-02 100.00% 98.00%

1000 28.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0009 1.10E-02 100.00% 99.00%

1000 40.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0009 1.10E-02 100.00% 100.00%

1000 10.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00136 1.10E-02 84.81% 91.00%

1000 14.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00136 1.10E-02 90.71% 94.00%

1000 20.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00136 1.10E-02 97.42% 97.00%

1000 28.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00136 1.10E-02 100.00% 99.00%

1000 40.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00136 1.10E-02 100.00% 100.00%

1000 10.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 75.99% 75.00%

549




Droplet diameter

Bed media

diameter

Bed height



Density difference

Dynamic viscosity

of dispersed

phase


phase

Empty bed

velcity

Interfacial tension

Calculated removal

efficiency from

DAMAK's model

Observed efficiency




(N.s)/m2 (=0.1000cP)


% %

1000 14.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 81.28% 86.00%

1000 20.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 87.29% 92.00%

1000 28.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 93.37% 97.00%

1000 40.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 99.00%

1000 10.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00406 1.10E-02 71.19% 67.00%

1000 14.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00406 1.10E-02 76.15% 79.00%

1000 20.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00406 1.10E-02 81.78% 84.00%

1000 28.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00406 1.10E-02 87.47% 92.00%

1000 40.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00406 1.10E-02 93.94% 94.00%

1000 10.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00541 1.10E-02 68.00% 61.00%

1000 14.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00541 1.10E-02 72.73% 65.00%

1000 20.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00541 1.10E-02 78.11% 71.50%

1000 28.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00541 1.10E-02 83.54% 85.50%

1000 40.00 0.36 0.02 914.5 998 83.5 2.10E-03 1.07E-03 0.00541 1.10E-02 89.72% 92.00%

1000 10.00 0.36 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 79.78% 85.00%

1000 14.00 0.36 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 85.34% 90.00%

1000 20.00 0.36 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 91.65% 94.50%

1000 28.00 0.36 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 98.03% 98.00%

1000 40.00 0.36 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 100.00%

1000 10.00 0.36 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 84.83% 89.00%

1000 14.00 0.36 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 90.73% 92.00%

1000 20.00 0.36 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 97.44% 96.00%

1000 28.00 0.36 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 99.00%

1000 40.00 0.36 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 100.00%

1000 10.00 0.36 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 92.18% 93.50%

1000 14.00 0.36 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 98.60% 95.00%

1000 20.00 0.36 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 99.00%

1000 28.00 0.36 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 100.00%

1000 40.00 0.36 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 100.00%

1000 10.00 0.56 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 66.86% 71.00%

1000 14.00 0.56 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 71.51% 76.00%

1000 20.00 0.56 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 76.80% 80.00%

1000 28.00 0.56 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 82.15% 85.00%

1000 40.00 0.56 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 88.22% 94.00%

1000 10.00 0.56 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 71.08% 78.00%

1000 14.00 0.56 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 76.03% 81.00%

1000 20.00 0.56 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 81.66% 85.00%

1000 28.00 0.56 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 87.34% 92.00%

1000 40.00 0.56 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 93.80% 98.00%

1000 10.00 0.56 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 77.25% 86.00%

1000 14.00 0.56 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 82.63% 90.00%

1000 20.00 0.56 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 88.74% 98.00%

1000 28.00 0.56 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 94.91% 98.00%

1000 40.00 0.56 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 100.00% 99.00%

1000 10.00 0.73 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 60.13% 60.00%

1000 14.00 0.73 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 64.32% 65.00%

1000 20.00 0.73 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 69.07% 70.00%

1000 28.00 0.73 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 73.88% 76.00%

550




Droplet diameter

Bed media

diameter

Bed height



Density difference

Dynamic viscosity

of dispersed

phase


phase

Empty bed

velcity

Interfacial tension

Calculated removal

efficiency from

DAMAK's model

Observed efficiency




(N.s)/m2 (=0.1000cP)


% %

1000 40.00 0.73 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 79.34% 91.00%

1000 10.00 0.73 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 63.93% 63.00%

1000 14.00 0.73 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 68.38% 69.00%

1000 20.00 0.73 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 73.44% 75.00%

1000 28.00 0.73 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 78.55% 81.00%

1000 40.00 0.73 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 84.36% 95.00%

1000 10.00 0.73 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 69.48% 71.00%

1000 14.00 0.73 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 74.31% 76.00%

1000 20.00 0.73 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 79.81% 84.00%

1000 28.00 0.73 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 85.36% 90.00%

1000 40.00 0.73 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 91.68% 98.00%

1000 10.00 0.94 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 54.35% 53.00%

1000 14.00 0.94 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 58.13% 58.00%

1000 20.00 0.94 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 62.43% 61.00%

1000 28.00 0.94 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 66.77% 66.00%

1000 40.00 0.94 0.03 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 71.71% 75.00%

1000 10.00 0.94 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 57.78% 57.00%

1000 14.00 0.94 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 61.81% 61.00%

1000 20.00 0.94 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 66.38% 65.00%

1000 28.00 0.94 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 71.00% 71.00%

1000 40.00 0.94 0.05 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 76.25% 82.00%

1000 10.00 0.94 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 62.79% 63.00%

1000 14.00 0.94 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 67.17% 67.00%

1000 20.00 0.94 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 72.13% 72.00%

1000 28.00 0.94 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 77.15% 78.00%

1000 40.00 0.94 0.1 914.5 998 83.5 2.10E-03 1.07E-03 0.0027 1.10E-02 82.86% 92.00%

1000 10.00 0.73 0.03 795 998 203 1.35E-03 1.07E-03 0.0027 4.20E-02 69.68% 68.00%

1000 14.00 0.73 0.03 795 998 203 1.35E-03 1.07E-03 0.0027 4.20E-02 74.53% 77.00%

1000 20.00 0.73 0.03 795 998 203 1.35E-03 1.07E-03 0.0027 4.20E-02 80.04% 84.50%

1000 28.00 0.73 0.03 795 998 203 1.35E-03 1.07E-03 0.0027 4.20E-02 85.61% 90.00%

1000 40.00 0.73 0.03 795 998 203 1.35E-03 1.07E-03 0.0027 4.20E-02 91.94% 92.00%

1000 10.00 0.73 0.03 994.1 998 3.9 1.09E-03 1.07E-03 0.0027 1.60E-02 44.33% 40.50%

1000 14.00 0.73 0.03 994.1 998 3.9 1.09E-03 1.07E-03 0.0027 1.60E-02 47.42% 45.50%

1000 20.00 0.73 0.03 994.1 998 3.9 1.09E-03 1.07E-03 0.0027 1.60E-02 50.93% 51.00%

1000 28.00 0.73 0.03 994.1 998 3.9 1.09E-03 1.07E-03 0.0027 1.60E-02 54.47% 56.00%

1000 40.00 0.73 0.03 994.1 998 3.9 1.09E-03 1.07E-03 0.0027 1.60E-02 58.50% 73.00%

1000 10.00 0.73 0.03 684 998 314 4.20E-04 1.07E-03 0.0027 3.62E-02 64.47% 62.00%

1000 14.00 0.73 0.03 684 998 314 4.20E-04 1.07E-03 0.0027 3.62E-02 68.96% 66.00%

1000 20.00 0.73 0.03 684 998 314 4.20E-04 1.07E-03 0.0027 3.62E-02 74.06% 72.10%

1000 28.00 0.73 0.03 684 998 314 4.20E-04 1.07E-03 0.0027 3.62E-02 79.21% 89.00%

1000 40.00 0.73 0.03 684 998 314 4.20E-04 1.07E-03 0.0027 3.62E-02 85.07% 90.00%

1000 10.00 0.73 0.03 860 998 138 7.16E-04 1.07E-03 0.0027 3.03E-02 61.93% 57.50%

1000 14.00 0.73 0.03 860 998 138 7.16E-04 1.07E-03 0.0027 3.03E-02 66.24% 64.50%

1000 20.00 0.73 0.03 860 998 138 7.16E-04 1.07E-03 0.0027 3.03E-02 71.14% 75.00%

1000 28.00 0.73 0.03 860 998 138 7.16E-04 1.07E-03 0.0027 3.03E-02 76.09% 83.00%

1000 40.00 0.73 0.03 860 998 138 7.16E-04 1.07E-03 0.0027 3.03E-02 81.71% 89.00%

551

Annex A2.2 Corresponding Kozeny-Carman's Porosity (ε) for lower (critical) and upper part of bed and Porosity for single step caluculationRef. [3], pp.126,and [26], pp.49Case Bed media

diameterBed

heightDispersed

phase density


Dynamic viscosity of dispersed

phase

Dynamic viscosity

of continuous

phase

Empty bed

velcity

Interfacial tension Observed headloss

Predicted Hc Corresponding porosity for

H<Hc

Correcponding porosity for H>Hc

corresponding porosity for one-step calculation

dp H ρd ρc μd μc V γo/wm m kg/cu.m kg/cu.m (N.s)/m2

(=1000cP)(N.s)/m2 (=1000cP)

m/s N/m (=1000dyn/cm)

m m ε for H<Hc ε for H>Hc

1 6.50E-04 0.025 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 0.422 0.101 0.151 0.000 0.1506.50E-04 0.05 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 0.654 0.101 0.162 0.000 0.1626.50E-04 0.075 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 1.144 0.101 0.155 0.000 0.1556.50E-04 0.1 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 1.539 0.101 0.155 0.000 0.1556.50E-04 0.2 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 1.879 0.101 0.154 0.250 0.179




5 6.50E-04 0.1 829 998 2.95E-03 1.08E-03 0.0017 2.20E-02 1.906 0.1456 2.20E-04 0.02 829 998 2.95E-03 1.08E-03 0.0019 2.20E-02 3.541 0.1477 3.00E-04 0.045 829 998 2.95E-03 1.08E-03 0.002 2.20E-02 1.226 0.2118 3.00E-04 0.045 829 998 2.95E-03 1.08E-03 0.002 2.20E-02 0.286 0.3189 3.00E-04 0.045 829 998 2.95E-03 1.08E-03 0.005 2.20E-02 8.171 0.16210 3.00E-04 0.045 829 998 2.95E-03 1.08E-03 0.005 2.20E-02 5.447 0.18311 0.00022 0.02 829 998 2.95E-03 1.08E-03 0.0025 2.20E-02 3.813 0.15612 2.20E-04 0.02 829 998 2.95E-03 1.08E-03 0.02 2.20E-02 11.575 0.20713 0.00065 0.02 829 998 2.95E-03 1.08E-03 0.0025 2.20E-02 0.817 0.13014 6.50E-04 0.02 829 998 2.95E-03 1.08E-03 0.02 2.20E-02 2.043 0.183

552

Annex A2.3 Comparison between calculated efficiency and experimental result for brush-type bed coalescer (Based on Newly proposed SRIJAROONRAT's and TAPANEEYANGKUL's model)Proposed SRIJAROONRAT's model TAPANEEYANGKUL's modelEff(%) = 104.5(ρc Vp D/μc)-0.77 (dE/D)0.18 (dF/D)-0.18(H/D)0.694(1-ε)0.35 Eff(%) = 1.76 (ρc Vp D/μc)-0.21 (dE/D)0.58 (dF/D)-0.58 (H/D)0.35 (1-ε)0.35 (DN/Vp)0.53

Case Ref. page Total hydrocar

bon concentr

ation

Droplet diameter

Fiber diameter

Bed diameter

Porosity Bed height Dispersed phase density

Continuous

phase density

Density difference

Dynamic viscosity

of dispersed

phase


phase

Empty bed

velcity

Interfacial

tension

Calculated removal

efficiency from Srijaroonrat's

model

Observed efficiency

from experiment(by

droplet size)

Assumed rotating speed

for Tapaneeyangkul

model

Calculated removal

efficiency from Tapaneeyungkul

's modelCo dE dF D ε H ρd ρc Δρ μd μc Vp γo/w Re Eff Eff exp N Eff tapanee

mg/l m m m - m kg/cu.m kg/cu.m kg/cu.m (N.s)/m2 (=1000cP)

(N.s)/m2 (=0.1000cP)

m/s N/m (=1000dyn/cm)

% % RPM

1 10 198 1000 1.8E-06 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 39.06% 39.66% 450 15.45%2 10 198 1000 2.9E-06 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 42.51% 35.59% 450 20.29%3 10 198 1000 1.6E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 58.01% 57.97% 450 55.26%4 10 198 1000 2.4E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 62.49% 67.12% 450 70.21%5 10 198 1000 4.8E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 70.60% 100.00% 450 100.00%6 10 198 1000 6.4E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.02 3.30E-02 932.71 74.45% 100.00% 450 100.00%7 10 198 1000 1.429E-06 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 63.99% 69.15% 450 22.67%8 10 198 1000 1.786E-06 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 66.61% 68.14% 450 25.81%9 10 198 1000 2.857E-06 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 72.49% 61.02% 450 33.89%

10 10 198 1000 1.214E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 94.06% 75.25% 450 78.44%11 10 198 1000 1.607E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 98.93% 85.42% 450 92.29%12 10 198 1000 2.429E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 100.00% 85.42% 450 100.00%13 10 198 1000 3.214E-05 1.00E-04 5.00E-02 0.91 0.3 795 998 203 1.35E-03 1.07E-03 0.01 3.30E-02 466.36 100.00% 100.00% 450 100.00%14 16 67 520 5.00E-06 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 71.56% 19.46% 720 34.85%15 16 67 520 6.00E-06 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 73.95% 57.30% 720 38.73%16 16 67 520 8.00E-06 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 77.88% 71.35% 720 45.77%17 16 67 520 9.00E-06 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 79.54% 81.08% 720 49.00%18 16 67 520 1.19E-05 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 83.62% 90.81% 720 57.56%19 16 67 520 1.28E-05 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 84.75% 94.05% 720 60.11%20 16 67 520 1.86E-05 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 90.65% 97.30% 720 74.66%21 16 67 520 2.34E-05 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 94.47% 99.46% 720 85.29%22 16 67 520 2.94E-05 2.00E-04 5.00E-02 0.845 0.5 900 998 98 1.10E-03 1.07E-03 0.02 N/A 932.71 98.43% 100.00% 720 97.36%23 8 84 1000 8.00E-06 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 1.89% 450 100.00%24 8 84 1000 9.00E-06 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 20.38% 450 100.00%25 8 84 1000 1.20E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 24.91% 450 100.00%26 8 84 1000 1.50E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 40.75% 450 100.00%27 8 84 1000 1.90E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 46.79% 450 100.00%28 8 84 1000 2.40E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 64.15% 450 100.00%29 8 84 1000 3.00E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 77.74% 450 100.00%30 8 84 1000 3.80E-05 1.00E-04 5.20E-02 0.96 0.2 795 998 203 1.35E-03 1.07E-03 0.001 0.042 48.50 100.00% 90.57% 450 100.00%31 11 229 1% V/V 1.00E-06 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 16.42% 52.00% 450 7.51%

32 11 229 1% V/V 2.00E-06 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 18.60% 48.00% 450 11.23%

33 11 229 1% V/V 3.00E-06 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 20.01% 55.00% 450 14.21%

34 11 229 1% V/V 8.00E-06 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 23.87% 43.00% 450 25.10%

35 11 229 1% V/V 1.20E-05 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 25.68% 34.00% 450 31.76%

36 11 229 1% V/V 1.60E-05 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 27.04% 44.00% 450 37.52%

37 11 229 1% V/V 2.40E-05 1.00E-04 5.00E-02 0.91 0.1 795 998 203 1.35E-03 1.07E-03 0.02 4.20E-02 932.71 29.09% 44.00% 450 47.47%

553

Annex A.3 DAF tested by SIEM[12] pp.130

ValueWastewater flowrate

Qin cu.m/s 3.9E-06

Pressurized water flowrate

Qwater cu.m/s 4.2E-06

Qin/Qwater 0.92Total flow Qtotal cu.m/s 8.14E-06

l/hr 29.31Column cross section area

Ao m2 0.0177

Velocity based on total flow

Vo m/s 4.60E-04

Air flowrate degassed in DAF column

Φ cu.m/s 4.20E-07

Dissolved air quantity in pressurized water

kg/cu.m 0.119

Effective column height

H m 0.7

Bubble diameter dB m 7.00E-05Dispersed phase density

ρd kg/cu.m 850


ρc kg/cu.m 998

Air density kg/cu.m 1.201Dispersed phase viscosity

μd kg/m.s 0.0011

Continuous phase viscosity

μc kg/m.s 0.00108

Bolzman constant K 1.38E-23Temperature T Kelvin 293

Description

554

Annex A.3 Comparison between observed efficiency and predicted efficiency for DAF[12] pp.130

Item Bubble diameter

Droplet diameter

Rising velocity of droplet

Rising velocity of

bubble

Relative velocity

Sedimentation

efficiency

Interception efficiency

Diffusion

efficienc

Theoritical totla

efficiency

Predicted

efficiency

predicted (Co-Cs)/Co

Observed (Co-Cs)/Co

Observed total

efficiency dB dE u U U-u ηS ηI ηD ηt theo αηexp αηexp m m m/s m/s m/s predicted observed

1 7.00E-05 2.10E-06 3.29E-07 2.46E-03 2.46E-03 1.34E-04 1.35E-03 4.3E-04 1.91E-03 2.21E-04 15.75% 0.00%2 7.00E-05 2.60E-06 5.05E-07 2.46E-03 2.46E-03 2.05E-04 2.07E-03 3.7E-04 2.64E-03 2.68E-04 18.75% 16.01% 2.25E-043 7.00E-05 3.30E-06 8.13E-07 2.46E-03 2.46E-03 3.30E-04 3.33E-03 3.2E-04 3.98E-03 3.42E-04 23.24% 20.66% 2.99E-044 7.00E-05 4.10E-06 1.26E-06 2.46E-03 2.46E-03 5.10E-04 5.15E-03 2.7E-04 5.93E-03 4.33E-04 28.46% 26.79% 4.03E-045 7.00E-05 5.10E-06 1.94E-06 2.46E-03 2.46E-03 7.89E-04 7.96E-03 2.4E-04 8.99E-03 5.54E-04 34.84% 37.42% 6.06E-046 7.00E-05 6.50E-06 3.16E-06 2.46E-03 2.46E-03 1.28E-03 1.29E-02 2.0E-04 1.44E-02 7.32E-04 43.26% 50.42% 9.07E-047 7.00E-05 8.20E-06 5.02E-06 2.46E-03 2.46E-03 2.04E-03 2.06E-02 1.7E-04 2.28E-02 9.61E-04 52.44% 58.48% 1.14E-038 7.00E-05 1.03E-05 7.92E-06 2.46E-03 2.46E-03 3.23E-03 3.25E-02 1.5E-04 3.58E-02 1.26E-03 62.15% 67.49% 1.45E-039 7.00E-05 1.30E-05 1.26E-05 2.46E-03 2.45E-03 5.15E-03 5.17E-02 1.3E-04 5.70E-02 1.65E-03 72.16% 74.05% 1.74E-0310 7.00E-05 1.64E-05 2.01E-05 2.46E-03 2.44E-03 8.22E-03 8.23E-02 1.1E-04 9.07E-02 2.17E-03 81.41% 79.93% 2.08E-0311 7.00E-05 2.06E-05 3.17E-05 2.46E-03 2.43E-03 1.30E-02 1.30E-01 9.4E-05 1.43E-01 2.85E-03 88.96% 86.61% 2.60E-0312 7.00E-05 2.59E-05 5.01E-05 2.46E-03 2.41E-03 2.07E-02 2.05E-01 8.1E-05 2.26E-01 3.74E-03 94.45% 93.94% 3.62E-0313 7.00E-05 3.27E-05 7.99E-05 2.46E-03 2.38E-03 3.35E-02 3.27E-01 7.0E-05 3.61E-01 4.93E-03 97.79% 96.58% 4.36E-03

555

Annex A3.1 Two-Phases hydrocyclone tested by MARef. [26] pp.105

Description Value

Flowrate Q cu.m/s 5.40E-04Diameter of upper cylindical part

D m 4.00E-02

Nominal diameter of hydrocyclone

Dn m 2.00E-02

Inlet diameter (2 ports) Di m 7.00E-03Overflow diameter Dover m 2.00E-03Underflow diameter Dunder m 1.00E-02Angle of conical part β Rad 2.62E-02Dispersed phase density ρd kg/cu.m 9.00E+02Continuous phase ρc kg/cu.m 9.98E+02Viscosity μc Ns/m2 1.00E-03

Comparison between Observed efficiency and predicted efficiency from Ma's and COLMAN's model for Two-Phases hydrocycloneRef. [26] pp.105, [16] pp.49d75% for COLMAN model is based on [29] pp.15.

Efficiency

Observed data from

[16], p105

Droplet diameter

predicted from COLMAN ([16]

Droplet from Ma's

Trajectory method

eff% dE dE dE% m m m

5% 8.4910% 9.6615% 10.8920% 12.19 13.0825% 13.5830% 22 15.10 17.9135% 16.7040% 24 18.43 22.5250% 25.6 22.36 26.6360% 28.12 27.18 30.5870% 30.75 33.39 34.1275% 37.3280% 34.7 42.12 37.2690% 40 57.13 40.0995% 72.0599% 55.78 105.49 42.33

100% 156.93 42.54

556

Annex A3.2 Three-Phases hydrocyclone tested by MARef. [16] pp.151

Description Value

Flowrate Q cu.m/h 1.40E+00Diameter of upper cylindical

D m 4.00E-02

Nominal diameter of inlet of oil

Do m 1.40E-02

Inlet diameter (2

Di m 5.60E-03

Underflow diameter for SS

Du m 6.00E-03

Underflow diameter for oil

Dp m 1.00E-03

Angle of conical oil

1.5o

Angle of conical SS part

12o

Split ratio 2%

Comparison between Observed efficiency and predicted efficiency forThree-Phases hydrocycloneRef. [16] pp.51

Efficiency %

Droplet from predicted efficiency

curve (micron)

Droplet from observed efficiency

curve (micron)10 5.07 8.520 9.00 930 12.4640 15.81 10.550 18.82 1260 21.73 14.2580 26.83 20.5

100 31.02 42100 50

557

Annex A5.1 Comparison between calculated flux and experimental result from resistance model of UF (reference temparature 20oC)

Item Transmembrane pressure

Velocity Concentration (by volume

of oil, not the

Observed flux Membrane resistance

Gel resistance coefficient (Rg

= Φ Vα Pt)

Alpha (exponent of

V)

Gel concentration

Predicted flux (J = Pt/ (Rm + Rg))

Ref.

ΔP or Pt V Co Jobserved Rm Φ α Cg Jpredictedbar m/s % l/(h.m2) % l/(h.m2)

1 0 2.8 4 0 0.430 0.022 -1.500 16.200 0.00 see note12 0.1 2.8 4 14 0.430 0.022 -1.500 16.200 20.97 see note13 0.5 2.8 4 70.8 0.430 0.022 -1.500 16.200 75.21 see note14 1 2.8 4 114 0.430 0.022 -1.500 16.200 111.17 see note15 1.5 2.8 4 137.29 0.430 0.022 -1.500 16.200 132.24 see note16 2 2.8 4 150.8 0.430 0.022 -1.500 16.200 146.08 see note17 2.5 2.8 4 162.16 0.430 0.022 -1.500 16.200 155.87 see note18 3 2.8 4 165.4 0.430 0.022 -1.500 16.200 163.16 see note19 3.5 2.8 4 168.6 0.430 0.022 -1.500 16.200 168.80 see note1

10 0 2.1 4 0 0.430 0.022 -1.500 16.200 0.00 see note111 0.24 2.1 4 33.5 0.430 0.022 -1.500 16.200 39.77 see note112 0.5 2.1 4 60 0.430 0.022 -1.500 16.200 63.17 see note113 1 2.1 4 84.32 0.430 0.022 -1.500 16.200 86.74 see note114 1.5 2.1 4 97.3 0.430 0.022 -1.500 16.200 99.05 see note115 2 2.1 4 107.6 0.430 0.022 -1.500 16.200 106.62 see note116 2.5 2.1 4 110.8 0.430 0.022 -1.500 16.200 111.74 see note117 3 2.1 4 114 0.430 0.022 -1.500 16.200 115.44 see note118 0 1.4 4 0 0.430 0.022 -1.500 16.200 0.00 see note119 0.24 1.4 4 33.5 0.430 0.022 -1.500 16.200 32.05 see note120 0.5 1.4 4 45.4 0.430 0.022 -1.500 16.200 45.70 see note121 1 1.4 4 54 0.430 0.022 -1.500 16.200 56.88 see note122 1.5 1.4 4 59 0.430 0.022 -1.500 16.200 61.93 see note123 2 1.4 4 60 0.430 0.022 -1.500 16.200 64.80 see note124 2.5 1.4 4 62 0.430 0.022 -1.500 16.200 66.66 see note125 3 1.4 4 62.7 0.430 0.022 -1.500 16.200 67.96 see note126 3.5 1.4 4 63.24 0.430 0.022 -1.500 16.200 68.92 see note127 0 0.7 4 0 0.430 0.022 -1.500 16.200 0.00 see note128 0.1 0.7 4 14 0.430 0.022 -1.500 16.200 12.41 see note129 0.5 0.7 4 31.9 0.430 0.022 -1.500 16.200 21.66 see note130 1 0.7 4 33.5 0.430 0.022 -1.500 16.200 23.89 see note131 1.5 0.7 4 33.5 0.430 0.022 -1.500 16.200 24.73 see note132 2 0.7 4 33.5 0.430 0.022 -1.500 16.200 25.18 see note133 2.5 0.7 4 33.5 0.430 0.022 -1.500 16.200 25.46 see note134 3 0.7 4 33.5 0.430 0.022 -1.500 16.200 25.64 see note135 1.0 1.0 3.2 36.51 0.640 0.019 -1.401 16.200 40.07 see note236 1.5 1.0 3.2 48.68 0.640 0.019 -1.401 16.200 43.82 see note237 2.0 1.0 3.2 48.68 0.640 0.019 -1.401 16.200 45.96 see note238 2.5 1.0 3.2 48.68 0.640 0.019 -1.401 16.200 47.36 see note239 3.0 1.0 3.2 48.68 0.640 0.019 -1.401 16.200 48.33 see note240 1.0 2.0 3.2 60.85 0.640 0.019 -1.401 16.200 74.47 see note241 1.5 2.0 3.2 93.31 0.640 0.019 -1.401 16.200 88.53 see note242 2.0 2.0 3.2 101.42 0.640 0.019 -1.401 16.200 97.76 see note243 2.5 2.0 3.2 109.54 0.640 0.019 -1.401 16.200 104.29 see note244 3.0 2.0 3.2 117.65 0.640 0.019 -1.401 16.200 109.14 see note245 1.0 2.8 3.2 97.37 0.640 0.019 -1.401 16.200 92.70 see note246 1.5 2.8 3.2 109.54 0.640 0.019 -1.401 16.200 115.55 see note247 2.0 2.8 3.2 133.88 0.640 0.019 -1.401 16.200 131.79 see note248 2.5 2.8 3.2 154.16 0.640 0.019 -1.401 16.200 143.94 see note249 3.0 2.8 3.2 170.39 0.640 0.019 -1.401 16.200 153.35 see note2

Note 1 Ref. [18] pp.86&88, Macroemulsion: Elf Seraft A, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)2 Ref. [11] , Microemulsion: Elf G3EAB, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)

558

Annex A5.2 Comparison between calculated flux and experimental result from film model of UF (reference temperature 20oC)

Item Velocity Transmembrane pressure

Concentration (by volume of

oil, not the concentrate)

Observed flux Constant Constant Constant Predicted flux J = k Vα ln (Cg/Co)

Ref.

V ΔP or Pt Co Jobserved k β Cg Jpredictedm/s bar % l/(h.m2) % l/(h.m2)

1 2.8 3.5 4 168.6 34.953 1.165 15.60 157.86 see note12 2.1 3.5 4 114 34.953 1.165 15.60 112.91 see note13 1.4 3.5 4 64 34.953 1.165 15.60 70.40 see note14 0.7 3.5 4 33.5 34.953 1.165 15.60 31.40 see note15 1.4 3.5 2 105.6 34.953 1.165 15.60 106.25 see note16 1.4 3.5 8 36 34.953 1.165 15.60 34.55 see note17 2.8 2 4 150.8 33.554 1.092 16.20 144.39 see note18 2.1 2 4 107.6 33.554 1.092 16.20 105.48 see note19 1.4 2 4 61.6 33.554 1.092 16.20 67.76 see note110 0.7 2 4 33.5 33.554 1.092 16.20 31.80 see note111 1.4 2 2 101.6 33.554 1.092 16.20 101.34 see note112 1.4 2 8 36 33.554 1.092 16.20 34.18 see note113 1.4 2 20 28.57 12.293 1.092 100.00 28.57 see note114 1.4 2 30 21.78 12.293 1.092 100.00 21.37 see note115 1.4 2 40 17.14 12.293 1.092 100.00 16.27 see note116 1.4 2 50 11.78 12.293 1.092 100.00 12.30 see note117 1 3 0.64 42.56 15.448 1.146 14.505 48.21 see note218 1.5 3 0.64 99.72 15.448 1.146 14.505 76.72 see note219 2 3 0.64 113.59 15.448 1.146 14.505 106.67 see note220 2.3 3 0.64 130.72 15.448 1.146 14.505 125.19 see note221 2.8 3 0.64 142.60 15.448 1.146 14.505 156.84 see note222 2.8 3 1.28 110.91 15.448 1.146 14.505 122.00 see note2

Note 1 Ref. [18] pp.86&88, Macroemulsion: Elf Seraft A, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)2 Ref. [11] , Microemulsion: Elf G3EAB, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)3 For item 13 to 16, Cg changes from 16.2% (of item 7-12) to 100%. So there is an inflection point in flux vs. ln(Co) curve. In this case, the concentration at the inflection point is around 8.0%.4 Data in tiem 1-6 and item 7 to 12 are observed from the same emulsion but at different pressure.

559

Annex A5.3 Comparison between calculated flux and experimental result from combination film and resistance model of UF (reference temperature 20oC)

Item Transmembrane pressure

Velocity Concentration

Observed flux

Membrane resistance

Ref Gel resistance coefficient

(Rg = Φ Vα Pt)

Ref Alpha

(exponent of V)

Oil concentration at inflection

point

Gel concentrat

ion (P=3.5

bar)

Refernce k

Reference Beta

J at C,Vref

J at C,V J at Cinf,V

Alpha (exponent of V) at C

Ref Gel resistance coefficient

(Rg = Φ Vα Pt)

Predicted flux (J = Pt/ (Rm + Rg))

Ref.

ΔP or Pt V Co J Rm Φ ref α ref Cinf Cg ref k ref β ref α at C Φ at C Jbar m/s % l/(h.m2) % % l/(h.m2) l/(h.m2) l/(h.m2) l/(h.m2)

1 0.2 1.400 2.0 27.2 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 34.22 see note12 0.5 1.400 2.0 68.8 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 61.27 see note13 1 1.400 2.0 86.4 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 83.19 see note14 1.5 1.400 2.0 96 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 94.45 see note15 2 1.400 2.0 101.6 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 101.31 see note16 2.5 1.400 2.0 104.8 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 101.314 101.313 34.172 -1.396 0.012 105.93 see note17 0.2 1.400 8.0 27.2 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 20.57 see note18 0.5 1.400 8.0 34.8 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 28.00 see note19 1 1.400 8.0 36 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 31.84 see note1

10 1.5 1.400 8.0 36 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 33.36 see note111 2 1.400 8.0 36 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 34.18 see note112 2.5 1.400 8.0 36 0.430 0.022 -1.500 8.000 16.200 33.540 1.092 34.178 34.178 34.172 -1.179 0.040 34.69 see note113 1 2.800 1.28 63.38 0.64 0.015 -1.130 14.505 15.448 1.146 82.975 122.002 -1.463 0.027 80.24 see note 214 1.5 2.800 1.28 87.15 0.64 0.015 -1.130 14.505 15.448 1.146 82.975 122.002 -1.463 0.027 96.81 see note 215 2 2.800 1.28 99.03 0.64 0.015 -1.130 14.505 15.448 1.146 82.975 122.002 -1.463 0.027 107.95 see note 216 2.5 2.800 1.28 106.95 0.64 0.015 -1.130 14.505 15.448 1.146 82.975 122.002 -1.463 0.027 115.97 see note 217 3 2.800 1.28 110.91 0.64 0.015 -1.130 14.505 15.448 1.146 82.975 122.002 -1.463 0.027 122.00 see note 2

Note 1 Ref. [18] pp.86&88, Macroemulsion: Elf Seraft A, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)2 Ref. [11] , Microemulsion: Elf G3EAB, UF membrane: IRIS3042 (polyacrylic membrane), MWCO: 50 Kdalton (15 nm.)3 For data in item 1 to 12, The velocity at required condition is the same as reference velocity (1.4 m/s). In this case, dummy condition at inflection point (C =8%) is introduced to show that the model can be used at any cond (see Part2, section 6.1.3)

560

Annex A5.4 Comparison between calculated flux and experimental result of macro.microemulsion mixture(reference temperature 20oC)

Item Velocity Transmembrane pressure

%V of macroemulsi

on

%V of microemulsi

on

% oil in macroemulsi

on

% oil in microemulsion concentrate

% oil of mixture

Observed flux

Flux of macroemulsion at Coil,mix (2)

Flux of microemulsion at Coil,mix (2)

Predicted flux of the

mixtureV ΔP or Pt Cmac Cmic Roil,mac Roil,mic Coil,mix J Jmac:Coil,mix Jmic:Coil,mix Jmix

m/s bar % % % % l/(h.m2) l/(h.m2) l/(h.m2) l/(h.m2)1 1.00 0.0 4.0 1.0 80.0 32.0 3.52 0 0.00 0.00 0.002 1.00 1.0 4.0 1.0 80.0 32.0 3.52 27.73 46.12 20.01 43.753 1.00 1.5 4.0 1.0 80.0 32.0 3.52 27.73 49.39 20.90 46.804 1.00 2.0 4.0 1.0 80.0 32.0 3.52 35.65 51.20 21.38 48.495 1.00 2.5 4.0 1.0 80.0 32.0 3.52 43.57 52.35 21.67 49.566 1.00 3.0 4.0 1.0 80.0 32.0 3.52 43.57 53.15 21.88 50.317 2.00 0.0 4.0 1.0 80.0 32.0 3.52 0.00 0.00 0.00 0.008 2.00 1.0 4.0 1.0 80.0 32.0 3.52 63.38 88.40 40.11 84.019 2.00 1.5 4.0 1.0 80.0 32.0 3.52 71.30 101.23 43.87 96.01

10 2.00 2.0 4.0 1.0 80.0 32.0 3.52 87.15 109.14 46.02 103.4111 2.00 2.5 4.0 1.0 80.0 32.0 3.52 95.07 114.52 47.42 108.4212 2.00 3.0 4.0 1.0 80.0 32.0 3.52 99.03 118.41 48.40 112.0413 2.80 0.0 4.0 1.0 80.0 32.0 3.52 0.00 0.00 0.00 0.0014 2.80 1.0 4.0 1.0 80.0 32.0 3.52 91.11 117.71 54.59 111.9715 2.80 1.5 4.0 1.0 80.0 32.0 3.52 110.91 141.61 61.78 134.3516 2.80 2.0 4.0 1.0 80.0 32.0 3.52 126.76 157.60 66.14 149.2817 2.80 2.5 4.0 1.0 80.0 32.0 3.52 142.60 169.06 69.06 159.9718 2.80 3.0 4.0 1.0 80.0 32.0 3.52 146.56 177.67 71.16 167.9819 1.00 0.0 4.0 2.0 80.0 32.0 3.84 0.00 0.00 0.00 0.0020 1.00 1.0 4.0 2.0 80.0 32.0 3.84 29.09 43.74 18.88 39.6021 1.00 1.5 4.0 2.0 80.0 32.0 3.84 29.09 46.67 19.67 42.1722 1.00 2.0 4.0 2.0 80.0 32.0 3.84 37.40 48.28 20.09 43.5823 1.00 2.5 4.0 2.0 80.0 32.0 3.84 41.55 49.31 20.35 44.4824 1.00 3.0 4.0 2.0 80.0 32.0 3.84 45.71 50.01 20.53 45.1025 2.00 0.0 4.0 2.0 80.0 32.0 3.84 0.00 0.00 0.00 0.0026 2.00 1.0 4.0 2.0 80.0 32.0 3.84 45.71 84.27 38.05 76.5727 2.00 1.5 4.0 2.0 80.0 32.0 3.84 62.33 95.85 41.41 86.7828 2.00 2.0 4.0 2.0 80.0 32.0 3.84 70.64 102.92 43.32 92.9929 2.00 2.5 4.0 2.0 80.0 32.0 3.84 74.79 107.69 44.56 97.1730 2.00 3.0 4.0 2.0 80.0 32.0 3.84 74.79 111.12 45.42 100.1731 2.80 0.0 4.0 2.0 80.0 32.0 3.84 0.00 0.00 0.00 0.0032 2.80 1.0 4.0 2.0 80.0 32.0 3.84 78.95 112.63 51.98 102.5233 2.80 1.5 4.0 2.0 80.0 32.0 3.84 91.41 134.31 58.46 121.6734 2.80 2.0 4.0 2.0 80.0 32.0 3.84 103.88 148.62 62.35 134.2435 2.80 2.5 4.0 2.0 80.0 32.0 3.84 103.88 158.76 64.94 143.1236 2.80 3.0 4.0 2.0 80.0 32.0 3.84 108.03 166.33 66.79 149.7437 1.00 0.0 4.0 4.0 80.0 32.0 4.48 0.00 0.00 0.00 0.0038 1.00 1.0 4.0 4.0 80.0 32.0 4.48 23.77 37.51 16.85 31.6039 1.00 1.5 4.0 4.0 80.0 32.0 4.48 27.73 39.64 17.47 33.3140 1.00 2.0 4.0 4.0 80.0 32.0 4.48 31.69 40.80 17.80 34.2341 1.00 2.5 4.0 4.0 80.0 32.0 4.48 31.69 41.53 18.01 34.8142 1.00 3.0 4.0 4.0 80.0 32.0 4.48 35.65 42.03 18.15 35.2043 2.00 0.0 4.0 4.0 80.0 32.0 4.48 0.00 0.00 0.00 0.0044 2.00 1.0 4.0 4.0 80.0 32.0 4.48 39.61 73.27 34.28 62.1345 2.00 1.5 4.0 4.0 80.0 32.0 4.48 55.46 81.87 36.99 69.0446 2.00 2.0 4.0 4.0 80.0 32.0 4.48 67.34 86.97 38.50 73.1247 2.00 2.5 4.0 4.0 80.0 32.0 4.48 67.34 90.35 39.48 75.8148 2.00 3.0 4.0 4.0 80.0 32.0 4.48 71.30 92.75 40.15 77.7249 2.80 0.0 4.0 4.0 80.0 32.0 4.48 0.00 0.00 0.00 0.0050 2.80 1.0 4.0 4.0 80.0 32.0 4.48 63.38 98.88 47.16 84.1051 2.80 1.5 4.0 4.0 80.0 32.0 4.48 83.18 115.21 52.44 97.2852 2.80 2.0 4.0 4.0 80.0 32.0 4.48 91.11 125.58 55.54 105.5753 2.80 2.5 4.0 4.0 80.0 32.0 4.48 99.03 132.75 57.59 111.2754 2.80 3.0 4.0 4.0 80.0 32.0 4.48 99.03 138.00 59.04 115.44

Note 1 Use reference data of flux of microemulsion and macroemulsion from Annex A5.1 and 5.2 and culculate flux at any concentration by combination model (as shown in Annex 5_3).

561

Point d'Inflection

Documents

tambour et

tambour ou

disque dans

design et

eaux rsiduaires huileuses

genie des procedes et

et 291x10

units procd et