Pretreatment of seawater for biodegradable organic content removal using membrane bioreactor

Desalination 153 (2002) 133–140

0011-9164/02/$– See front matter © 2002 Elsevier Science B.V. All rights reserved

Presented at the EuroMed 2002 conference on Desalination Strategies in South Mediterranean Countries:Cooperation between Mediterranean Countries of Europe and the Southern Rim of the Mediterranean.Sponsored by the European Desalination Society and Alexandria University Desalination Studies and TechnologyCenter, Sharm El Sheikh, Egypt, May 4–6, 2002.

*Corresponding author.

Pretreatment of seawater for biodegradable organic contentremoval using membrane bioreactor

Chettiyappan Visvanathana, Natapol Boonthanona, Arumugam Sathasivana,Veeriah Jegatheesanb*

aEnvironmental Engineering Program, Asian Institute of Technology, P.O. Box 4, Klong Luang,12120 Pathumthani, Thailand

Tel. +66 (2) 5245640; Fax +61 (2) 5245640; email: [email protected] of Engineering, James Cook University, Townsville, QLD 4811, Australia

Tel. +61 (7) 4781-4871; Fax +61 (7)4775-1184; email: [email protected]

Received 4 April 2002; accepted 20 April 2002

Abstract

Reverse osmosis (RO) is currently one of the most prevalent methods used for seawater desalination. During thepast four decades, the research and development has reduced the energy consumption from about 20 to 4 kWh/m3,while improvements in membrane science has led to a 20-fold increase in the specific membrane flux. Nevertheless,research is still underway to reduce the operation and maintenance problems and thus improve the performance ofRO systems. The most important maintenance problem associated with RO operation is the membrane fouling,especially biological fouling (biofouling). This work focuses on the aspects to eliminate biofouling in RO membranes,by adopting a proper pretreatment system. The experimental results revealed that fluidized bed biological granularactivated carbon, at 15 min empty bed contact time (with dissolved organic carbon, DOC concentration of 6–8 mg/L)can be utilized effectively to remove nearly 100% biodegradable DOC from seawater. Continuous experiments ofmembrane bioreactor (MBR) have been conducted concomitantly to gain insight into the long-term effects of MBRon biodegradable organic content removal and biofouling control. The results show that MBR system producedbetter effluent with 78% DOC removal and quasi-total biodegradable DOC removal. Dissolved oxygen was not alimiting factor for the DOC degradation. Short-term experimental runs were conducted with RO membrane usingboth pretreated and non-pretreated seawater. The results showed that filtrate from MBR yielded the highest permeateflux improvement, which was approximately 300% compared with non-pretreated seawater.

Keywords: Biodegradable organic matter; Biofouling; Membrane bioreactor; Microfiltration; Pretreatment; Reverseosmosis

134 C. Visvanathan et al. / Desalination 153 (2002) 133–140

1. Introduction

Currently, reverse osmosis (RO) process is oneof the most prevalent methods for desalinatingthe seawater. RO is gaining increased attentionover conventional distillation and ion exchangeprocesses by taking major shares of desalinationmarkets [1]. Recent technological advances in thedesign and manufacture of polyamide, celluloseacetate, and composite-polymer semi-permeablemembranes for a variety of commercial purposeshave made the RO the preferred method for theefficient recovery of high purity water directlyfrom raw, chemically contaminated, or brackishwater resources receiving a minimal pretreatment[2].

Many researches have been conducted so farto identify the optimum use of RO membraneincluding the selection of materials and optimumoperating conditions. These studies have achievedsuccesses in many fields of RO application. Duringthe past four decades the research and develop-ment have brought the energy consumption fromabout 20 down to 4 kWh/m3, while the improve-ments in membrane sciences have led to 20-foldincrease in specific membrane flux.

Nevertheless research is still underway to furtherreduce the operation and maintenance cost of theRO systems. The most important operation andmaintenance problem encountered is membranefouling, which is the undesirable formation ofdeposits on the surface of membrane. Depositstake place when the rejection is not transportedfrom membrane surface into bulk solution. Scaling,organic fouling, colloidal fouling and biofoulingare the four major types of fouling generally seemto occur on RO membrane surfaces.

While for the first three types of fouling, thereexist well-established pretreatment and chemicalcleaning methods, biofouling has been one of themost tenacious and least understood forms ofmembrane fouling. Biofouling is caused by thegrowth of microorganisms in modules [3] and isthe result of the complex interaction between the

membrane material, fluid parameters (such asdissolved substances, flow velocity, pressure. etc.)and microorganisms. The main source of microbialcontamination is the feed water. Feed water thatcontains high numbers of microorganisms tendsto cause microbial problems.

Since the RO systems generally have largemembrane surface areas the probability of bacterialadhesion onto the surfaces increases. Once thesebacterial cells established themselves onto themembrane surface over a period of time, theybecome self-supporting as the extra-polymericsubstances and bacterial matrix absorbs and con-centrates soluble organic and inorganic nutrientsnear the cell walls [4]. Thus, the originally solubleorganic and inorganic nutrients are now locallyimmobilized and converted from solution to asemisolid state establishing biofilm of variousmicroorganisms and inorganic nutrients. Thesebiofilms will be considered as biofouling, whichis not desirable in desalination processes and shouldbe removed to achieve better operation. In addition,RO membranes experience fluid transport towardthe rejecting surface and increase the probabilityof contact with suspended microorganisms.

The best way to prevent the formation of foulantsis to remove the materials responsible for foulingbefore they come into contact with the RO mem-brane. The various pretreatment methods that canbe applied are activated carbon, membrane filtration,and disinfection by chemical additions or ultraviolet(UV) irradiation or sanitization. All these processesare effective in removing these foulants, but arealso associated with some disadvantages. Someof these technologies are technically feasible butstill economically expensive.

In this experimental study the effectiveness ofmembrane bioreactor (MBR) as a pretreatmentprocess in removing organic content in seawater wasevaluated. MBR is the process in which con-ventional biological system is coupled with themembrane process (microfiltration, MF or ultra-filtration, UF). This process is found suitable forthe treatment of wastewater and removal of

C. Visvanathan et al. / Desalination 153 (2002) 133–140 135

organic compounds effectively [5]. The typical totalorganic carbon (TOC) concentration in seawateris reported to be in the range of 1.5–19.0 mg/L [6].Both suspended growth and attached growth MBRare found applicable in wastewater treatment. Forseawater, which normally has low organic load, theattached growth bioreactor could be effective. Themain objective of this study was to reduce theorganic content using a MBR in seawater as partof the RO pretreatment system and the effectiveoperating conditions for pretreatment system wereinvestigated.

2. Materials and methods

2.1. Feed water preparation

The raw seawater was taken from the Gulf ofThailand in the vicinity of Chonburi Province,Thailand. The initial analysis of the raw seawaterrevealed relatively low DOC concentration. How-ever, in order to conduct the experimental runswithin a DOC range of 3–15 mg/L, sodium acetate(CH3COONa) was added as an artificial sourceof DOC. Dissolved organic carbon was analyzedusing Standard Methods [8]. The biodegradableportion of the feed water was analyzed using themethod described by Hozalski et.al. [8].

2.2. Acclimatization phase

At the starting of acclimatization phase, DOCconcentration of the feed water was increased upto 150–200 mg/L to encourage the bacterial growthon the granular activated carbon, while main-taining DO concentration around 6–7 mg/L.

2.3. Experiments

The experimental runs were conducted in threestages. The first stage was conducted in batcheswith fixed and fluidized granular activated carbonto find the optimum empty bed contact time, inorder to determine the kinetic coefficients. In thesecond phase, continuous MBR experiments wereconducted as pretreatment for the RO desalination

processes. In the last stage, RO experiments wereconducted with different pretreated effluents toexamine the effectiveness of the pretreatmentsystem.

2.3.1. Effect of empty bed contact time

The experiments were conducted by varyingempty bed contact time (EBCT) between 15 and60 min, in two modes of operation: fixed andfluidized bed biological granular activated carbon(BGAC). The DOC of prefiltered raw seawater(filtered through 5 µm cartridge filter) was adjustedby adding the required amount of CH3COONa andpumped into the pretreatment reactors. Theeffluent sample was collected at the reactor outletin the fixed bed reactor and from the overflow offiltrate in the fluidized bed reactor. The initial DOCconcentration was varied between 3 and 15 mg/L.

2.3.2. Continuous experiments

Two operational modes are used to conductthe experiments (Fig. 1): (i) BGAC filter coupledwith the MF membrane, called MBR; (ii) MFmembrane. BGAC was taken from the first phaseof the operation along with a dead end hollow fiberMF membrane module (Sterapore manufacturedby Mitsubishi Rayon Co. Ltd.) with a pore size0.1 µm and surface area 0.42 m2 to construct a MBR.DOC concentration of the feed water, filtrate fromBGAC filter, filtrate from MBR and filtrate fromthe control membrane were analyzed. At the endof the experimental run, chemical cleaning of themembrane was carried out to remove the biologicalfouling from the membrane by filtering the alkalinecleaning agent at pH 10.5–11 through the membrane.

2.3.3. Desalination with reverse osmosis

A plate and frame type membrane module(dead end) made of stainless steel was used. Theinlet was connected at the top segment and theoutlet was connected at the bottom segment.Commercially available nitrogen (N2) cylinderequipped with a pressure regulator was used to

136 C. Visvanathan et al. / Desalination 153 (2002) 133–140

maintain high pressure in the system. First, thepretreated effluent from the second phase con-tinuous experiment was fed into the feed watertank at the top and both valves at the top andbottom of feed water tank were closed. The feedsystem pressure was maintained at 15 bar and thebottom valve was opened to collect and measurethe permeate mass continuously with an electronicbalance. The RO experimental runs were carriedout using raw seawater, the filtrate from the MBRand the filtrate from the control MF membranefor the runs that lasted for 3 days.

3. Results and discussion

3.1. Granular activated carbon isotherm

The adsorption capacity of GAC was determinedusing CH3COONa as a representative of DOC.From the experimental data the following Freundlichisotherm was obtained:

(x/m) = 0.0008 Ce1.2949 (1)

raw seawater Cartridge filter Feed seawater

Sodium acetate addition

Recycle Linefor making

Membrane Bioreactor

Filtrate

GAC column

Filtrate

fluidized condition

ControlMembranereactor

Fig. 1. Experimental setup for the membrane bioreactor.

where x is the amount of solute (CH3COONa)adsorbed, m is the weight of adsorbent (GAC) andCe is the concentration of solute in the solution atequilibrium. GAC isotherm will play a vital rolein adsorption of bacterial cells on GAC surface andpore despite the fact that the biological granularactivated carbon filtration does not use the GACadsorption as a principal mode of DOC removal.

3.2. Effect of empty bed contact time

For an initial DOC concentration of 13–15 mg/L, the batch experimental results showedan increase in the removal efficiency of DOC withthe increase in EBCT in the fluidized bed reactor(Fig. 2). The increase in the removal efficiency canbe attributed to the higher chances of organic carbonto be in contact with bacterial cells. Similar resultswere obtained with the initial DOC concentrationof 6–8 mg/L as shown in Fig. 3. When the initialDOC concentration was reduced to 6–8 mg/L, thefixed bed removal efficiency was augmented.

C. Visvanathan et al. / Desalination 153 (2002) 133–140 137

However, in both cases, the efficiency was stilllower than that obtained for the fluidized bed. Inorder to estimate the amount of the non-biode-gradable portion of the DOC used in the raw water,batch incubation experiments were conducted.Three different initial DOC concentration of sea-water with seeding were incubated at 37°C for30 days and DOC concentration was analyzed oncein every 4 days. It was found that the DOC concen-tration reduced down to 1.80–2.25 mg/L, irrespec-tive of the initial DOC concentrations. Thus theseawater sample tested contained around 2 mg/Lof non-biodegradable organic carbon. Therefore

0

2

4

6

8

10

12

14

16

10 15 20 25 30 35 40 45 50 55 60 65

Contact time (min)

DO

C (

mg/L

)

0

10

20

30

40

50

60

70

80

90

100

Rem

oval

effi

cien

cy (

%)

Seawater Fluidized bed effluentFixed bed effluent Fluidized bed efficiencyFixed bed efficiency

Fig. 2. Effect of EBCT on effluent DOC concentrationand removal efficiency (initial DOC concentration 13–15 mg/L).

0

2

4

6

8

10 15 20 25 30 35

Contact time (min)

DO

C (

mg

/L)

0

20

40

60

80

100

Rem

ov

al

effi

cien

cy (

%)

Seawater Fluidized bed effluentFixed bed effluent Fluidized bed efficiencyFixed bed efficiency

Fig. 3. Effect of EBCT on effluent DOC concentration andremoval efficiency (initial DOC concentration 6–8 mg/L).

in EBCT experimental runs, the biodegradableportions (BDOC) were found to degrade completelyin 15 min under fluidized bed conditions, withan initial DOC concentration of 6.8 mg/L.

The initial DOC concentration of 6–8 mg/Lyielded higher efficiencies as shown in Fig. 3.Although, the experimental runs with the initialDOC concentration of 6–8 mg/L at 30 min EBCTyielded lower final effluent concentrations thanat 15 min EBCT, it is not necessary that 30 minEBCT yielded significantly better efficiency than15 min EBCT because of the non-biodegradableportions of DOC in seawater. From these experi-mental observations, it was noted that when theseawater that has the initial DOC concentrationless than 8 mg/L, the fluidized bed BGAC with15 min EBCT, was effective in quasi-total removalof BDOC. As BDOC is the most important linkto the fouling of membrane for the next continuousphase experiments this optimum experimental valuewas used.

3.3. Kinetics of bacterial degradation in seawaterenvironment

The results of the fluidized bed EBCT experi-ments were used for the calculation of kinetic co-efficients. These coefficients were determined bythe Michaelis–Menton kinetic equation,

XkXSK

S

t

Xd

S

−+

= maxµ

d

d (2)

where X is the bacterial concentration, µmax is themaximum growth rate, S is the concentration ofBDOC, KS is the half-saturation constant, kd is theendogenous decay rate. The relationship betweenX and S is given by

t

SY

t

X

d

d

d

d = (3)

where Y is the specific yield. The values of thesekinetic parameters are summarized in Table 1.

138 C. Visvanathan et al. / Desalination 153 (2002) 133–140

3.4. Continuous experiments

Based on the optimum operating conditionsobtained from the earlier batch experiments(EBCT = 15 min, DOC = 6–8 mg/L), continuousexperiments on the MBR were conducted. Inparallel, control experiments were conducted wherefeed water was directly filtered with the biologicalGAC arrangement. Both reactors were operatedat a flow rate of 20 ml/min, which corresponds to15 min EBCT.

3.5. DOC removal

In order to analyze the DOC concentration,samples were taken from four different samplingpoints and the results are presented in Fig. 4. Itcan be noted as expected from the earlier set ofexperiments, the filtrate from the BGAC has a DOCof 2 mg/L, which corresponds to the non-biode-gradable portion of the DOC.

Table 1Kinetic coefficients of bacterial growth in the seawaterenvironment

Parameter Value

Endogeneous decay rate (kd), l/h 3.69

Half-saturation constant (Ks), mg C/L 4.87

Specific yield (Y) 0.356

Maximum specific growth rate (µmax), 1/h 9.81

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25Days

DO

C (

mg

/L)

Feed seawater

Filtrate from BGAC filterFiltrate from membrane bioreactor

Filtrate from control membrane

Fig. 4. DOC concentration in continuous experiments.

In terms of solids content, the feed seawatercan be categorized mainly into two parts, namelythe suspended solids and dissolved solids. In general,BGAC filtration will not be affected by the sus-pended solids concentration because the bacterialmass can consume mainly dissolved organic solids.However, these suspended solids could affect themicrofiltration membrane flux. Unlike suspendedsolids, dissolved solids in feed seawater can beseparated into two components. The first part isthe biodegradable portion, which is 5 mg/L andmostly composed of CH3COONa. The second partis the non-biodegradable portion around 2 mg/L,which can also be separated into two parts basedon the membrane pore size. Based on the experi-mental results presented in Fig. 4, the DOC of thefiltrate from BGAC filter in the membrane bio-reactor is considered as non-biodegradable. Sincethe membrane pore size of the microfiltrationmembrane used in this experiment is 0.1 µm, thenon-biodegradable part will be separated usingthis criteria, where the fraction less than 0.1 µmwill pass through the membrane. This fraction ofthe original non-biodegradable portion with lessthan 0.1 µm size was measured to be 1.75 mg/L.

From the microbiological population observa-tion, the filtrates from both membrane bioreactorand control membrane have no apparent bacterialcolony. The size of biomass generated was biggerthan 0.1 µm. For the control membrane, the experi-mental results showed that around 50% or3.5 mg/L DOC passed through the microfiltrationmembrane. These 3.5 mg/L DOC were comprisedof the non-biodegradable portion and biodegrad-able portion, which was CH3COONa. Since thefeed seawater for both reactors came from thesame source, the non-biodegradable part wasaccounted for 1.75 mg/L. Apart from the non-biodegradable portion, 1.75 mg/L of CH3COONapassed through the membrane while 3.25 mg/Lof them may be degraded because of bacterialaccumulation in the control membrane reactor andbacterial attachment onto the membrane surface.Here, in order to make a mass balance, the yield

C. Visvanathan et al. / Desalination 153 (2002) 133–140 139

coefficient (Y) from Table 1, was used to convertthe biodegradable portion into bacterial mass.

Fig. 5 presents the difference between solidsloading on the membrane in the membrane bio-reactor and the control membrane. Fig. 5 revealsthat, in terms of DOC removal efficiency, regardlessof membrane, BGAC filter is being utilizedeffectively to reduce the biodegradable portionsin seawater. The membrane plays an importantrole in removing bacterial cells and a small amountof the non-biodegradable portions (however, bio-fouling was prominent in the membrane bioreactor).At steady state, the DOC removal efficiencies forthe membrane bioreactor and control membraneare 78 and 50% respectively.

3.6. Transmembrane pressure

Transmembrane pressure required for bothreactors increased slightly as shown in Fig. 6,without any significant fouling. However, thecolor of the membrane fiber was found to changefrom white to yellow due to particle depositionand creation of bacterial cells at BGAC. Trans-membrane pressure of the membrane bioreactorwas found to be higher than that of the controlmembrane reactor. This could be due to higherbiomass formation in BGAC, which resulted inhigher biomass load in the MBR compared to thecontrol reactor.

3.7. Dissolved oxygen

DO was not a limiting factor for the DOC de-gradation. The DO was found to be 3–4 mg/L inthe samples collected after BGAC filter. DO con-centrations in the filtrate from both reactors wereat the level of 1–2 mg/L. This may be due to theoxygen demand of bacterial biomass that wasattached onto the surface of both microfiltrationmembranes.

3.8. Membrane cleaning

Due to higher biofouling formation of thepretreatment unit, the membrane used as the MBR

1.751.751.75

0.250.250.251.1571.780

1.7505

0

2

4

6

8

10

12

Feed seawater Membrane

bioreactor

Control

membrane

So

lid

co

nte

nt

(mg

/L) Na acetate

Newly generated biomassOriginal non-biodegradable 0.45-0.1 Original non-biodegradable < 0.1

Fig. 5. Solid composition of the feed seawater and seawaterbefore filtration.

0

1

2

3

4

5

6

7

8

9

10

0 2 4 6 8 10 12 14 16 18 20

Days

Tra

nsm

emb

ran

e p

ress

ure

(k

pa

)

0

1

2

3

4

5

6

7

8

9

10

Tra

nsm

emb

ran

e p

ress

ure

(k

pa

)

Membrane bioreactor

Control membrane reactor

Fig. 6. Variation of the transmembrane pressure with timefor the continuous experiment.

was found to be darker than the membrane usedin the control experiment. After cleaning themembrane with alkaline solution (pH=10.5–11.0)the membranes were found to have higher resist-ance (2.1 and 4.8% more than original resistance).However, the resistance was not so high indicatingthe possibility of restoring the membrane perform-ance during cleaning. The slight increase in themembrane resistance can be attributed to the internaland external absorption of micromolecules, whichcould not be removed by chemical cleaning.

3.9. Desalination with reverse osmosis

Fig. 7 shows the results of the desalinationexperiments with the three types of feed water.

140 C. Visvanathan et al. / Desalination 153 (2002) 133–140

Here the RO experiments were conducted to observethe effect of pretreatment in terms of biofoulingof the RO membrane. The three different typesof seawater were filtered through the dead endRO membrane modules under a pressure of 15 bar.Each experimental run lasted for 3 days. Thepermeate flux increased by 300% when the MBR-pretreated seawater was used as feed instead ofraw seawater. Since the filtration was conductedin dead end mode, the permeate flux was expectedto decrease continuously. However, in practicewhen RO modules are used in crossflow mode,almost a steady permeate flux will be obtainedthroughout a run.

4. Conclusions and recommendations

The experimental investigations conducted toremove organic content from seawater using themembrane bioreactor provide the following con-clusions:• Based on EBCT experiments, increasing EBCT

will increase DOC removal efficiency. Withinthe DOC concentration range studied (3–15 mg/L), the fluidized bed, BGAC filter systemyields higher removal efficiency comparedwith fixed bed.

• In continuous experiments, the membranebioreactor yields better DOC removal efficiencythan the control membrane reactor. The mem-brane bioreactor yields nearly 100% BDOC

0

1

2

3

4

5

6

7

8

9

10

11

12

13

0 500 1000 1500 2000 2500 3000 3500 4000 4500

Time (min)

Per

mea

te f

lux (

L/(

m2x

h))

Raw seawater

Control membrane

Membrane bioreactor

Fig. 7. Variation of permeate flux with time for threedifferent feeds.

removal. Moreover, dissolved oxygen does notplay an important role in both reactors becausenaturally seawater DO concentration alwayslies in the saturated level, which is sufficientfor bacterial degradation in the membranebioreactor. The results of membrane resistanceof the microfiltration membrane revealed thatperiodic simple chemical cleaning of the mem-brane using alkaline solution is effective torestore the flux to nearly the initial level.

• In reverse osmosis experiments, the filtratefrom the membrane bioreactor yields the highestpermeate flux compared with the filtrate fromthe control membrane reactor and raw seawaterdue to the lowest solid contents. The permeateflux of the filtrate from the membrane bioreactorwas improved up to 300% compared with rawseawater.

References[1] A.H. Khan, Desalination Processes and Multistage Flash

Distillation Practice, Elsevier, Amsterdam, TheNetherlands, 1986.

[2] H.F. Ridgway, A. Kelly, C. Justice and B.H. Olson,Microbial fouling of reverse-osmosis membranes usedin advanced wastewater treatment technology: chemical,bacteriological, and ultra structural analyses, Appl.Environ. Microbiol., 45(3) (1983) 1066–1084.

[3] J.J. Allard, J. Rovel and P. Treille, Importance of pre-treatment in RO plant design and its incidence onoperation and maintenance costs, Desalination, 19 (1976)169–174.

[4] A. Munro, Biological control on membrane surfaces,Asian Water, 15(8) (1999) 22–26.

[5] C. Visvanathan, R. Ben Aim and K. Parameshwaran,Membrane separation bioreactors for wastewatertreatment, Critical Rev. Environ. Sci. Technol., 30(1)(2000) 1–48.

[6] I.I. Frebrero, A Study of Seawater Quality in LaemChabang and the Evaluation of Impacts of Deep Sea Porton Marine Environment, AIT thesis no. EV-85-6,Thailand, 1984.

[7] A.E. Greenberg, R.R. Trussel and L.S. Clessceri, Eds.,APHA, AWWA, WEF, Standard Methods for theExamination of Water and Wastewater, 19th ed.,Washington, DC, USA, 1995.

[8] R.M. Hozalski, S. Goel and E.J. Bouwer, TOC removalin biological filters, J. AWWA, 87(12) (1995) 40–54.