modelling

eo

ering, P

diment

ctivated sludge in an

ss models, reasonable

gen, chemical oxygen

inematics of sludge

2014 Elsevier Ltd. All rights reserved.

maintenance costs, high and flexible capacity, and low sludge

production (Hong et al., 2003). More than 10,000 oxidation

ditches are to be found in China and the USA alone. However,

land, consume sub-

osits of sludge (Yang

eing undertaken tomand and Carlsson,

reduce energy con-

sumption (see e.g. Zhou et al., 2012).

Mathematical models offer an effective means of simu-

lating the physical, chemical and biological processes in ODs

* Corresponding author. Department of Environmental Engineering, Peking University, Beijing 100871, China. Tel.: 86 10 62751185; fax:86 10 62756526.

E-mail address: [email protected] (J. Ni).

Available online at www.sciencedirect.com

ScienceDirect

.e ls

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 41. Introduction

Oxidation ditches (ODs) are widely used in wastewater treat-

ment due to their simple construction, low capital and

oxidation ditches occupy large areas of

stantial energy, and produce uneven dep

et al., 2011). Much work is presently b

optimize the treatment process (see e.g. A

2012), to improve sludge deposition andPseudo-solid phase

Sedimentation

Mass transfer

Biochemical kinetics

Oxidation ditch

satisfactory agreement with laboratory data on the behavior of a

oxidation ditch. By coupling species transport and biological proce

predictions are made of: (1) the biochemical kinetics of dissolved oxy

demand (COD) and nitrogen variation, and (2) the physical k

sedimentation.Three-dimensional three-phase

model

sludge viscosity, sludge density, oxygen mass transfer rate, and carbon substrate uptake

due to adsorption onto the activated sludge. The validation test results were in veryLi Lei a,b, Jinren Ni a,b,*aDepartment of Environmental EnginebThe Key Laboratory of Water and Se

China

a r t i c l e i n f o

Article history:

Received 7 June 2013

Received in revised form

6 October 2013

Accepted 15 January 2014

Available online 23 January 2014

Keywords:0043-1354/$ e see front matter 2014 Elsevhttp://dx.doi.org/10.1016/j.watres.2014.01.021eking University, Beijing 100871, China

Sciences, Ministry of Education, Peking University, Beijing 100871,

a b s t r a c t

A three-dimensional three-phase fluid model, supplemented by laboratory data, was

developed to simulate the hydrodynamics, oxygen mass transfer, carbon oxidation, nitri-

fication and denitrification processes in an oxidation ditch. The model provided detailed

phase information on the liquid flow field, gas hold-up distribution and sludge sedimen-

tation. The three-phase model described water-gas, water-sludge and gasesludge in-

teractions. Activated sludge was taken to be in a pseudo-solid phase, comprising an

initially separated solid phase that was transported and later underwent biological re-

actions with the surrounding liquidmedia. Floc parameters were modified to improve thedenitrification in oxidation ditchan

transfer, carbon oxidation, nitrification andThree-dimensional three-phassimulation of hydrodynamics,

journal homepage: wwwier Ltd. All rights reservemodel forxygen mass

evier .com/locate /watresd.

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 201Nomenclature

aj,i stoichiometric number of species i in the j-th

process, dimensionless1which experience complicated alternating aerobic and anoxic

conditions for nitrification and denitrification. For example,

the Activated Sludge Model (ASM) predicts the effluent water

quality and biomass production of wastewater treatment

plants (Henze et al., 2000). In ODs, the wastewater treatment

efficiency is influenced not only by the bio-reaction of acti-

vated sludge, but also by the dynamics of liquid-bubble flows

bA decay coefficient for XB,A, d

bH decay coefficient for XB,H, d1

Cs wall roughness constant, dimensionless

dG bubble diameter, m

DL diffusivity of oxygen in the liquid phase, m2 s1

dO diameter of aeration orifice, m

F!

lift; q lift force, kg m2 s2

fP fraction of biomass leading to particulate

products, dimensionless

F!

q external body force, kg m2 s2

F!

vm; q virtual mass force, kg m2 s2

g! gravitational acceleration, m s2hKs wall roughness height, m

iXB mass of nitrogen per mass of COD in biomass,

g g1

iXP mass of nitrogen per mass of COD in products

from biomass, g g1

J!

q; i diffusion flux of species i in phase q, kg m2 s1

ka ammonification rate, m3 g1 d1

kh maximum specific hydrolysis rate, d1

KLaL mass transfer coefficient, s1

KNH ammonia half-saturation coefficient for XB,A,

g m3

KNO nitrate half-saturation coefficient for XB,H, g m3

KO,A oxygen half-saturation coefficient for XB,A, g m3

KO,H oxygen half-saturation coefficient for XB,H, g m3

KS half-saturation coefficient for XB,H, g m3

KX half-saturation coefficient for hydrolysis of slowly

biodegradable substrate, g g1

mpq mass transfer from phase p to q, kg m3 s1

O2 oxygen concentration in air, g m3

OF normalized standard error, dimensionless

p pressure, N m2

Q flow rate, m3 s1

R!

pq interaction force between phase p and q,

kg m2 s2

Rq, i source term representing mass transfer of species

i from other phases to phase q, and the

production/consumption rate of the species i for

biochemical reactions, kg m3 s1

S soluble constituent concentration, g m3

SI soluble inert pollution concentration, g m3

SND soluble organic nitrogen concentration, g m3

SNH ammonium concentration, g m3

SNO nitrate and nitrite concentration, g m3

SO oxygen concentration in liquid, g m3SO(S) saturated dissolved oxygen, g m3

Sq source term of phase q, kg m3 s1

SRT sludge age, d

SS soluble biodegradable pollution concentration,(Insel et al., 2005). Carbon oxidation process was firstly

coupled in one-dimensional (1D) convectionedispersion

equation by Stamou (1994), followed by coupling more pro-

cesses such as nitrification and denitrification (Stamou, 1997;

Stamou et al., 1999; Mantziaras et al., 2011) to reveal the ef-

fects of local hydrodynamics on water quality in ODs. For

more detailed understanding of the effects of hydrodynamics,

g m3

Uslip slip velocity between a gas bubble and water,

m s1

v!pq interphase velocity from phase p to q phase, m s1v!q velocity vector of phase q, m s1X particulate component concentration, g m3

XB,A autotrophic biomass, g m3

XB,H heterotrophic biomass, g m3

xci calculated result of the i-th parameter

XI particulate inert pollution concentration, g m3

xmi measured result of the i-th parameter

XND particulate organic nitrogen concentration, g m3

XP inert biomass, g m3

XS particulate biodegradable pollution

concentration, g m3

YA yield for XB,A, g g1

YH yield for XB,H, g g1

Yq, i mass fraction of species i in phase q,

dimensionless

Greek letters

a modification coefficient for SO(S), dimensionless

adG modification coefficient for dG, dimensionless

aq volume fraction of phase q, dimensionless

b modification coefficient for KLaL, dimensionless

g modification coefficient for KLaL, dimensionless

3 dissipation rate of turbulent kinetic energy, m2 s3

hg correction factor for mH under anoxic conditions,

dimensionless

hh correction factor for hydrolysis under anoxic

conditions, dimensionless

mA maximum specific growth rate for XB,A, d1

mH maximum specific growth rate for XB,H, d1

mq shear viscosity of phase q, kg m1 s1

rj process rate of the j-th bioreaction, kg m3 s1

rq density of phase q, kg m3

s surface tension, kg s2

sq stress-strain tensor of phase q, kg m1 s2

Subscripts

G gas phase

in inflow

L liquid phase

out outflow

rec recirculation flow

S pseudo-solid phase

two-dimensional (2D) or three-dimensional (3D) models

would be preferred (Yang et al., 2011). Littleton et al. (2007)

introduced the Activated Sludge Model No. 2 (ASM2) to a 3D

fluid dynamicsmodel for elucidating the role of the bioreactor

macro-environment in simultaneous biological nutrient

removal. Other investigators also attempted to describe the

complex phenomena in OD (Zhang et al., 2010). On the other

hand, effects of hydrodynamics would be reflected bymatters

in different phases. However, most previous models have

been limited to one or two phases, although multiple-phase

phenomena and processes have been observed (e.g. Pipes,

1969; Schmid et al., 2003; Fayolle et al., 2007) in ODs.

Understanding of coupled physicalechemicalebiological

processes relies on accurate assessment of the transport

processes and phase interactions. There are three basic pha-

ses of the primary medias needed to be fully considered in

ODs. First, oxygen must be supplied in order to maintain the

level of dissolved oxygen (DO) during the aerobic process and

so directly affects the effluent water quality (Fayolle et al.,

2007), which implies the model that does not take gas phase

od

T d

cs;

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4202Fig. 1 e Framework of the three-dimensional three-phase m

heterotrophs; AGA denotes aerobic growth of autotrophs; OM

of heterotrophs; HEO denotes hydrolysis of entrapped organiHEON denotes hydrolysis of entrapped organics nitrogen; DH d

autotrophs).el for an oxidation ditch (AGH denotes aerobic growth of

enotes oxygenmass transfer; ANGH denotes anoxic growth

ASON denotes ammonification of soluble organic nitrogen;enotes decay of heterotrophs; DA denotes decay of

into consideration cannot reasonably simulate the oxygen

mass transfer between gas and sewage water. Second, acti-

vated sludge comprises a strongly hydrated solid phase, and

has different physical properties to those of pure water

(Schmid et al., 2003). Furthermore, sludge settling can lead to

septic sludge formation at the dead angle in OD, with associ-

ated odor (Pipes, 1969). The complicated phase interactions

and transformations in an OD system include transfer of

dissolved oxygen from the gas phase, and carbon oxidation,

nitrification and denitrification in the liquid and solid phases.

In this paper, a 3D three-phase model was developed by

between the phases is outlined. Coupled gas transport and

modified oxygen mass transfer models simulate the DO dis-

liquid, and is termed a pseudo-phase. To approximate the

pseudo-phase behavior in the model, the activated sludge is

first represented as in a separated solid phase regarding

transport and sedimentation processes, and later as in a sol-

ideliquid phase after biological changes have taken place (see

Fig. 2). The 3D three-phase model quantifies the phase-

dependent behavior of the sewage, activated sludge, and

gas, and their complicated interactions in an OD (Fig. 1). To

permit analogy between the behavior of activated sludge and

that of granular particles, the model parameters require

adjustment to account for differences between floc sludgeand

granular sludge. Floc sludge has much higher water content,

mass and momentum conservation laws. The model as-

sumptions are as follows:

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 203tribution. Coupled species-transport and modified biological

kinetic models simulate the sludge distribution and pollutant

degradation, and include the biochemical kinetics of chemical

oxygen demand (COD) and nitrogen removal in an OD. More

specifically, the 3D and three-phase model is able to simul-

taneously describe the 3D transport of sewage water (e.g.

secondary flow, see Yang et al., 2011), bubbles and activated

sludge (settling process), in addition to the phase interactions

and the sedimentation processes related to the activated

sludge. During calibration, the input parameters are deter-

mined by iteration for target values of sludge viscosity,

settling capacity, oxygen mass transfer rate, and carbon sub-

strate uptake due to adsorption onto the pseudo-solid phase.

The sludge transport process and the effect of inclusion of the

pseudo-solid phase on mass transfer and pollutant trans-

formation are investigated through simulations of variations

in concentration of activated sludge, DO, COD, ammonia ni-

trogen and nitrate in a pilot-scale OD.

2. Methodology

2.1. Model development

A hydrated activated sludge floc is essentially a gelatinous,

coagulated material whose phase lies between solid andtaking the sewage water, air bubbles and activated sludge to

be in liquid, gas and pseudo-solid phases, respectively. The

proposed model could not only vividly describe local hydro-

dynamic structures with 3D fluid velocities but also reason-

ably simulate the interactions of sewage water, air bubbles

and activated sludge treated as pseudo-solid phase. Fig. 1

shows the proposed framework in which the relationshipFig. 2 e Dual roles of the(1) sewage water, activated sludge, and air are in liquid,

pseudo-solid, and gas phase, respectively;

(2) pollutants are divided into soluble and particulate

components regarded as species of liquid and pseudo-

solid phases, respectively;

(3) heterotrophic and autotrophic biomass are species in

the pseudo-solid phase;

(4) DO is a species in the liquid phase;

(5) oxygen mass transfer is a biological rather than a

physical process;

(6) accumulation of biomass with ammonia may be

neglected;

(7) alkalinity is not a limiting parameter;

(8) no biological reaction occurs in the secondary settling

tank.

2.1.1. Multiphase hydrodynamics modelA standard 3D steady-statemulti-phase flowmodel, described

in detail by Fluent Corporation (2006), is used to describe the

complicated hydrodynamic behavior of the sewage water,

activated sludge, and gas transportation behavior in an OD.

The steady-state equilibrium mass conservation equation is

given by:contains extracellular polymeric substances, and has a

negatively-charged surface, whereas granular sludge is more

permeable (Wang et al., 2012). To describe the transport and

evolution of the pollutants and biomass, an advection-

diffusion species transport model is coupled with the modi-

fied oxygen mass transfer model and modified biological ki-

netic process models. The governing equations are based onpseudo-solid phase.

ical

ore

iXBS

1YA

iX

fPX

fPX

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4204transport and growth/decay of activated biomass and con-

taminants in the ditch:V$aqrq v

!q

X3p1

mpq mqp

Sq (1)where aq is the volume fraction, rq is density, v

!q is velocity

vector, Sq is the source term, p and q are the different phases,

and mpq characterizes the mass transfer from phase p to q.

Subscripts q W, G and S denote water, gas and active sludge,respectively. The steady-state equilibrium momentum equa-

tion is:

V$aqrq v

!q v!

q

aqVp V$sq aqrq g!

Xnp1

R!

pqmpq v!

pq

mqp v!qp

Fq! F!lift; q F!vm; q (2)

where p is the pressure shared by all phases, g! is gravitationalacceleration, R

!pq is the interaction force between phases, v

!pq

is the inter-phase velocity from phase p to q, mpq v!

pq mqp v!qpdenotes themomentum change due tomass transfer between

the p-th and the q-th phases, F!

q is an external body force,

F!

lift;q is a lift force, F!

vm;q is a virtual mass force and sq is the

stress-strain tensor for the q-th phase. The multi-phase flow

model is closed using a steady-state k- 3turbulencemodel (see

e.g. Fluent Corporation, 2006).

2.1.2. Species transport modelThe following 3D steady-state advection-dispersion species

transport equation with source term is used to describe the

Table 1 e Quantifications of the major chemical and biolog

No. Process Bi

1 Aerobic growth of heterotrophs 1YHSS 1YHYH SO

2 Aerobic growth of autotrophs 4:57YAYA

SO iXB

3 Oxygen mass transfer O2/SO4 Anoxic growth of heterotrophs 1

YHSS 1YH2:86YH SNO

5 Hydrolysis of particulate organics XS/SS6 Ammonification SND/SNH7 Hydrolysis of particulate organics nitrogen XND/SND8 Decay of heterotrophs XB, H/(1fP)XS9 Decay of autotrophs XB, A/(1fP)XSV$aqrq v

!qYq; i

V$aq J!q; i Rq; i (3)

where Yq,i is the mass fraction of species i in phase q, Rq, i is a

source term representingmass transfer of species i from other

phases to phase q, and the production/consumption rate of

the species i for biochemical reactions, J!

q; i is the diffusion

flux of species i in phase q.

2.1.3. Biochemical kinetics modelThe source term Rq, i in Equation (3) is given by the modified

Activated Sludge Model No.1 (ASM1) (Henze et al., 2000).

Biochemical processes, such as carbon oxidation, nitrification

and denitrification, are described by considering thirteen

components (Fig. 1). Autotrophic, heterotrophic and inert

biomass is regarded as particulate matter. Fig. 1 and Table 1

summarize the detailed biological interactions among thethree phases and corresponding kinetic processes. The source

term Rq, i is expressed by

Rq; i X9j1

aj; irj

(4)

where j is the number of the process in Table 1, aj,i is the

stoichiometric number of species i in the j-th process, and rj is

the process rate. In Equation (4), if species i is produced, istaken positive; whereas if species i is consumed, is takennegative.

Carbon substrate removal from storage always occurs in an

activated sludge system, because activated sludge is in the

separated phase (Carucci et al., 2001). COD then accumulates

in sludge (Beccari et al., 2002), leading to the heterotrophs

having a competitive growth advantage (Cggn et al., 2011).

Hence, certain stoichiometric and kinetic parameters (such as

mH, YH, KS and bH) are modified herein, given that activated

sludge is treated as in the separated pseudo-solid phase. Table

2 lists the proposed parameters used by ASM1 and the modi-

fied input parameters used in the present work.

2.1.4. Oxygen mass transfer modelThe oxygen mass transfer rate r3 (see Table 1) from the gas to

the liquid phase is determined by Kulkarni (2007)

r3 KLaL wastewateraSOS SO

(5)

in which KLaL wasterwater is the mass transfer coefficient, SO(S) is

saturated DO concentration in clean water, SO is oxygen con-

centration in the liquid phase, and a expresses the propor-

processes.

action Process rate

NH/XB; H r1 mH SSKSSSSO

KO; HSO XB; HSNH/XB; A 1YA SNO r2 mA

SNHKNHSNH

SOKO; ASO XB; A

r3(KLaL)wastewater(aSO(S)SO)BSNH/XB; H r4 mH SSKSSS

KO; HKO; HSO

SNOKNOSNO hgXB; H

r5 kh XS=XB; HKXXS=XB; Hh

SOKO; HSO hh

KO; HKO; HSO

SNOKNOSNO

iXB; H

r6kaSNDXB, Hr7 r5XND=XS

P(iXBfPi)XPXND r8 bHXB,HP(iXBfPi)XPXND r9 bAXB,Ationality between saturated DO in wastewater and its clean

water value. For bottom aeration, the oxygen mass transfer

coefficient for sewage water is expressed as:

KLaL wastewater gb12aGdG

DLUslippdG

s(6)

where g is introduced to consider the pseudo solid-effect, b is

the ratio of wastewater to clean water mass transfer co-

efficients, aG is the volume fraction occupied by the gas phase,

dG is the Sauter mean diameter of the bubbles, DL is diffusivity

of oxygen in the liquid phase, and Uslip is the slip velocity

between a gas bubble and water, and can be estimated by

evaluating jvL-vGj. For surface aeration, the oxygen masstransfer correlates directly with the dissipation rate of energy

(Kumar and Rao, 2009) such that

Table 2 e Summary of parameters used in modeling of biological processes.

Parameter Unit ASM1 Su and Yu (2006) Present

Stoichiometric YA g(COD) g(COD)1 0.24 0.24 0.24

YH g(COD) g(COD)1 0.67 0.58 0.63e0.67

fP Dimensionless 0.08 0.08 0.08

iXB g(N) g(COD)1 0.086 0.086 0.086

iXP g(N) g(COD)1 0.06 0.06 0.06

Kinetic mH d1 6.00 4.98 5.50e6.00

KS g(COD) m3 20.0 26.1 20.0e23.0

KO,H g(O2) m3 0.20 0.20 0.20

KNO g(N) m3 0.50 0.50 0.50

bH d1 0.62 0.92 0.62e0.77

hg Dimensionless 0.8 0.8 0.8

hh Dimensionless 0.40 0.44 0.44

kh d1 3.0 3.0 3.0

KX g(COD) g(COD)1 0.03 0.03 0.03

mA d1 0.80 0.80 0.80

KNH g(N) m3 1.0 1.0 1.0

KO,A g(O2) m3 0.4 0.4 0.4

ka m3 g(COD)1 d1 0.08 0.08 0.08

bA d1 0.15 0.15 0.15

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 205equations governing the interaction force R!

pq are listed byKLaL3

p 0:64 exp0:29 30:98 1:6 106 exph0:72 3 3:912:0i(7)

where 3is the dissipation rate of the turbulent kinetic energy.

Hence, the oxygenmass transfers due to the surface impellers

and bottom aeration system can be evaluated by using Equa-

tions (5)e(7).

2.2. Phase interaction

2.2.1. Liquidegas interactionAir bubbles introduced into OD experience a drag force due to

the different velocities in the liquid and gas phases. The

Fluent Corporation (2006). Liquid-gas interaction directly in-fluences both physical and mass transfer processes, with the

Fig. 3 e Plan view sketch of the pilot-scale oxidation ditch (aeffect of the latter incorporated via the source term in Equa-

tion (3).

2.2.2. Liquid-solid interactionThe density of activated sludge density ranges from 1010 to

1060 kg/m3 (Dammel and Schroeder, 1991), and so sludge

settling always occurs. Flow turbulence influences sludge

settling, resulting in intensive interaction between sewage

water and activated sludge floc. The interaction force R!

pq

between pseudo-solid and liquid phases is also described

using interaction equations listed by Fluent Corporation

(2006).

In activated sludge systems, intensive interaction occurs

between different species in the sewage and activated sludge.

To sustain growth, heterotrophs and autotrophs biomassutilize ammonium and nitrate species in sewage (Table 1).

) and monitoring points of liquid velocity (b) (Unit: mm).

Meanwhile, heterotrophs and autotrophs decompose to par-

ticulate biodegradable organic nitrogen and slowly biode-

gradable substrate, which are further hydrolyzed respectively

to soluble biodegradable organic nitrogen and readily biode-

particulate components Xi in Table 1) is initially also pre-

scribed. At steady state, when the internal condition of the OD

is balanced after sludge return, Xin is assumed to be the same

as the sludge concentration at the outlet (Xout) provided the

average sludge concentration (Xaverage) in OD is less than a

pre-set value (Xset). For Xaverage > Xset, the surplus sludge

(Xsurplus) is removed from the OD by pumping in order to

maintain the sludge concentration at a desired level. Full de-

tails about the surplus sludge pumping model and the inlet

sludge concentration condition are given by Stamou (1997).

The recirculation flow recycles water to the inlet. If Sin and Srecdenote the soluble constituents (e.g. ammonium and nitrate)

of wastewater and recirculation flow, the inlet value of the

soluble constituent variable is:

S QinSin QrecSrec=Qin Qrec (8)

where Qout is the outflow rate, Qin is the inflow rate, and Qrec is

Table 3 e Operation modes of impellers and stirrers.

Movingpart

Case I Case II

Speed (rpm) Direction Speed (rpm) Direction

Impeller 1 80 a 40 Impeller 2 80 40 Stirrer 1 90 b 70 eStirrer 2 90 70 Stirrer 3 90 70 Stirrer 4 70 50 a rotation in clockwise direction.b - rotation in anticlockwise direction.

ria

O

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4206activated sludge floc has properties different from pure water,

it is better to treat the floc as in a separated pseudo-solid

phase, with oxygen mass transfer between gas and liquid

phases modified accordingly in terms of solid concentration

(Mena et al., 2011). For this purpose, the parameter g is

introduced in Equation (6).

2.3. Boundary conditions

For the OD (Fig. 3), open boundary conditions are applied at

the influent inlet and effluent outlet. The walls are treated as

fixed, impermeable boundaries. At the inlet, the velocity is

prescribed (according to the inflow rate), and the inlet sludge

concentration (Xin, the sum of the concentrations of all

Table 4 e Experimental conditions and range of primary va

Variablesgradable substrate. The present model describes the in-

teractions between species in liquid and pseudo-solid phases,

unlike traditional models that neglect the pseudo-solid

phase.

2.2.3. Gasesolid interactionMany models, such as ASM-series models, consider activated

sludge to be perfectly soluble in the liquid phase. Given that anRotation mod

Model Calibration Liquid velocity Case I

MLSS Case I

DO Case I

COD Case I

Ammonium Case I

Nitrate Case I

Model Verification Liquid velocity Case II

MLSS Case II

DO Case II

COD Case II

Ammonium Case II

Nitrate Case IIthe recirculation flow rate. The wastewater constituents are

determined from experiments, following Henze et al. (2000).

At the OD outlet, the boundary pressure is atmospheric.

For bottom aeration, the introduction of gas can be treated

as source term in the Equations (1)e(3), and the oxygen con-

centration in the pumped air is evaluated using an equation

provided by Fayolle et al. (2007). Surface aeration refers to

rotation of the impellers which aerated the water in their vi-

cinity. The oxygen mass transfer rate due to surface aeration

is obtained by solving Equations (5) and (7).

A rigid-lid, slip wall boundary condition (see e.g. Yang

et al., 2011) is applied to the liquid and pseudo-solid phases

at the water surface. Injected air from bottom aeration es-

capes the OD at the gaseliquid surface, and so a degasification

condition (Le Moullec et al., 2011) is applied to the gas phase at

the water surface.

No-slip boundary conditions are assigned for all other

walls, including the bottom surface, the side and central walls

of the ditch. The roughness constant and the roughness

height at the no-slip boundaries are calibrated to the

measured data, using the Fluent values of 1 and 0.02 m,

respectively following Yang et al. (2011).

bles for model calibration and verification.

peration conditions Range of variables

e Aeration rate (m3/h)

2.2 ux: 0.000e0.155 m/s

uy: 0.000e0.044 m/s

uz: 0.000e0.116 m/s

2.2 3.80e4.16 g/L

2.2 0.32e1.66 mg/L

2.2 12.8e14.0 mg/L

2.2 0.33e0.90 mg/L

2.2 16.6e17.5 mg/L

2.2 ux: 0.000e0.132 m/s

uy: 0.000e0.031 m/s

uz: 0.000e0.101 m/s

2.2 3.42e4.55 g/L

2.2 0.28e1.07 mg/L

2.2 14.7e15.8 mg/L2.2 3.04e3.91 mg/L

2.2 11.78e14.01 mg/L

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 207Table 5 e Description of the parameters used in themodel.

Parameter Notation Unit Value

Sewage water density rL kg/m3 1000

Sewage water viscosity mL kg/m/s 0.0010

Dissolved oxygen of

synthetic wastewater

SO mg/L 0.20

Soluble inert organic SI mg/L 10.0The multiphase flow in OD is agitated by the moving parts

such as rotating blades, disc aerators and submerged impellers.

Here, the rotation of the rotating blades and disc aerators is

simulated by a moving wall model and the submerged impeller

describedbya fanmodel.For furtherdetails seeYangetal. (2011).

2.4. Parameter estimation

Activated sludge flocs are considered to be in a pseudo-solid

phase, whose density ranges from 1010 to 1060 g/mL

maximum specific storage rate and size of floc-like sludge

particles are both less than for granular sludge, and their

substrate of synthetic

wastewater

Readily biodegradable

substrate

synthetic wastewater

SS mg/L 80.0

Particulate inert organic of

synthetic wastewater

XI mg/L 0.0

Slowly biodegradable

substrate of synthetic

wastewater

XS mg/L 160.0

Nitrate and nitrite

of synthetic wastewater

SNO mg/L 0.0

Ammonium of synthetic

wastewater

SNH mg/L 50.0

Soluble biodegradable

organic nitrogen of

synthetic

wastewater

SND mg/L 0.0

Particulate biodegradable

organic nitrogen of

synthetic wastewater

XND mg/L 0.0

Sludge floc density rs kg/m3 1010

Sludge floc viscosity ms kg/m/s 0.0046

Sludge floc diameter ds mm 0.40

Pre-set average sludge

concentration

Xset g/L 3.8

Sludge age SRT d 25

Air density rG kg/m3 1.225

Air viscosity mG kg/m/s 1.8 105Air bubble diameter dG mm 2.60

Air bubble diameter

modification coefficient

adG Dimensionless 0.58

Wall roughness constant Cs Dimensionless 1

Wall roughness height hKs m 0.02

Modification coefficient a a Dimensionless 0.92

Modification coefficient b b Dimensionless 0.44

Modification coefficient g g Dimensionless 0.75

Yield for heterotrophic

biomass

YH g/g 0.64

Maximum specific growth

rate for heterotrophic

biomass

mH 1/d 5.80

Half-saturation coefficient

for heterotrophic

biomass

KS g/m3 22.0

Decay coefficient for

heterotrophic biomass

bH 1/d 0.70physical properties (e.g. moisture content and negative sur-

face charge) are also different (Grijspeerdt and Verstraete,

1997; Liu et al., 2005; Wang et al., 2012). In general, the pa-

rameters in the present model lie between those of ASM1 and

the granular sludge system (Su and Yu, 2006).

3. Experimental measurements in pilot-scale oxidation ditch

Fig. 3 depicts a plan view of the pilot-scale carrousel-type

oxidation ditch. The ditch was fabricated from plexiglass, and

comprised four straight 1.15 m lengths of channel each with

semi-circular end channels, the smaller semi-circles having

radius 0.35 m, the larger semi-circles having radius 0.7 m. The

total working volume was 1.4 m3. The channels had rectan-

gular cross-section of width 0.35 m and still depth 0.5 m. Two

surface impellers (Impeller 1 and Impeller 2) and four sub-

merged stirrers (Stirrer 1, Stirrer 2, Stirrer 3 and Stirrer 4)

located in the curved channels drove the recirculating flow in

the oxidation ditch. Each spindle-like impeller consisted of 18

steel strips; each stirrer comprised an S-shape blade of

diameter 0.2 m Table 3 lists the operating rotational speeds

and angular directions of the impellers and stirrers for the

calibration Case I and validation Case II. Air was introduced

from a series of 1 m long gas distributors located at the base of

the second and third channels, and also from the surface

entrainment effect of the impellers. Each orifice of the gas

distributorswas of diameter 0.1mm. The overall aeration rate,

controlled by a rotameter, was 2.2 m3/h. Synthetic waste-

water, originally stored in a tank of volume 1.8 m3, was(Dammel and Schroeder, 1991) and dynamic viscosity ranges

from 3.8 to 11.0 mPa S (Jin et al., 2004), andmean floc diameter

from 0.05 to 0.5 mm (Grijspeerdt and Verstraete, 1997).

The interaction force between the gas and liquid phases is

closely related to the bubble diameter, and is given by:

dG 2:9adGsdogrL

1=3(9)

where do is diameter of the aeration orifice, s is surface tension

of liquid phase, g is the acceleration due to gravity, rL is den-

sity of the liquid phase, and adG is a coefficient allowing for the

aeration orifice configuration and the presence of pseudo-

solid phase.

The mass transfer rate corresponding to the pseudo solid-

phase is greatly dependent on the coefficients a[0.92,1.0](Stenstrom and Gilbert, 1981), b[0.44,0.98] (Gillot and Heduit,2008), and g (to be calibrated experimentally).

Table 2 lists the major stoichiometric and kinetic param-

eters (after eliminating the pseudo-solid phase effect). Pa-

rameters related to heterotrophs growth and decay processes

(such as mH, YH, KS and bH) are modified in the present model

compared to ASM1 reflecting an improved understanding of

biodegradable material storage (see e.g. Gujer et al., 1999;

Krishna and Van Loosdrecht, 1999; Su and Yu, 2006). Thepumped into the ditch at a flow rate of 0.1 m3/h, and this flow

rate then maintained for a hydraulic residence time of 14 h.

The composition of synthetic wastewater was as follows:

1000 L of tap water, 250.0 g of sugar (approximately 250 mg/L

COD), 107.2 g of ammonium chloride (50 mg/L total nitrogen

concentration), 61.3 g of Na3PO4,12H2O, 500.0 g of NaHCO3,

3.0 g of FeSO4,7H2O, 10.0 g of CaCl2, 12.0 g of MgSO4, and 50mL

of trace element solution. The trace element solution con-

tained (per liter of tap water): 3.5 g of ethylene diamine tet-

raacetic acid, 2.0 g of ZnSO4,7H2O, 1.0 g of CuSO4,5H2O, 1.0 g of

MnSO4,7H2O, 1.0 g of Na2MoO4,2H2O, 1.0 g of H3BO3 and 0.2 g

of CoCl2,6H2O. A settling tank of volume 0.15m3 separated the

sludge which is recycled at the ditch inlet. The sludge recycle

ratio was 100%, and the average activated sludge concentra-

tion in the ditch was maintained at about 3.8 g/L. The sludge

retention time was 25 d. All experiments were performed at

atmospheric pressure and room temperature.

A High-resolution Acoustic Doppler Velocimeter for the

Laboratory (Vectrino II, developed by Nortek AS, Vangkro-

ken, Norway) was used to measure 3D liquid velocities at the

Sections 1-1, 2-2 and 3-3 as shown in Fig. 3. Twice a week,

activated sludge was sampled 0.1 m from the bed at M1, M2,

M3, M4, M5 and M6 (Fig. 3). The mixed liquor suspended

solid (MLSS) of each sample was determined using Method

2540 D (APHA, 1998). Twice daily, water samples for were

obtained 25 cm above the bed at W1, W2, W3, outlet, W4 and

W5 (Fig. 3). Each water sample was immediately filtered

using a 0.45 mm filter, and then stored at 4 C in refrigerator.Soluble COD was determined by the closed reflux titrimetric

method 5220C (APHA, 1998). Ammonia nitrogen and nitrate

concentrations were measured by Nesslers reagent spec-

trophotometry HJ 535-2009 (Environmental Protection

Agency of China, 2009) and ultraviolet spectrophotometric

screening 4500-NO3 B (APHA, 1998), respectively. DO con-

centrations were also monitored twice daily by an oxygen

probe (YSI-550A, YSI, Yellow Springs, Ohio, USA) located

0.25 m above the bed at W1, W2, W3, outlet, W4 and W5.

Ranges of primary variables experimental conditions are

listed in Table 4.

4. Results and discussion

4.1. Numerical model calibration

A 3D unstructured tetrahedral mesh was created for the OD

system using GAMBIT (Fluent Corporation, 2005) with mesh

refinement in the rotating and aerator zones. The governing

equations at steady state were solved using the finite volume

computer code, FLUENT. The grid independent analysis was

done, in which the liquid velocity was selected to conduct the

mesh test. Three different sets of meshes with cells of 55946,

161987 and 367881 respectively, were chosen to simulate the

liquid velocity field in the OD. As a result, the second set was

selected for all computations, considering the low difference

(less than 5%) of liquid velocities simulated with the optimal

mesh (161987 cells) and the refined mesh (36788 cells). The

segregated solver of Fluent 6.3 was used, with default

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4208Fig. 4 e Comparison between measured and simulated concent

nitrogen, and (e) nitrate, at the sampling locations for Case I.rations of: (a) MLSS, (b) DO, (c) soluble COD, (d) ammonia

Fig. 5 e Comparisons of simulated and measured liquid velocity components at different layers over cross-section 1-1:

surface (a), top (b), middle (c) and bottom (d) for Case II.

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 209

parameter settings applied. All simulationswere computed on

a workstation equipped with two Intel Xeon 2.93 GHz pro-

cessors and 24 GB RAM. Each run took about 24 h of CPU time

to reach steady state condition.

The inlet boundary conditions in the numerical model of

the pilot-scale OD utilized measured parameters relating to

the liquid and pseudo-solid phases. The inlet had a rectan-

gular cross-section of breadth 0.15 m, depth 0.05 m, and area

0.0075 m2. At the inlet, the flow speeds of the combined liquid

and pseudo-solid phases were kept at 0.0074 m/s. The syn-

theticwastewaterwas composed of SNO 0mg/L, SNH 50mg/L, SND 0 mg/L and XND 0 mg/L. The quantity of slowlybiodegradable substrate was about twice that of readily

biodegradable substrate (Henze et al., 2000; Orhon et al., 1997).

Sugar provided the carbon source for the synthetic waste-

water. The particulate inert organic concentration was

assumed zero, following Le Moullec et al. (2011). Hence, the

COD compositionwas: SI 10mg/L, SS 80mg/L,XS 160mg/L and XI 0 mg/L.

In order to calibrate the numerical model, the input pa-

rameters were adjusted iteratively in order to minimize the

following objective function (Squires, 2001):

OF

1nn1

Pni1 xci xmi2

q1n

Pni1 xci

(10)

where OF was the normalized standard error; xmi and xci were

themeasured and the calculated results of the i-th parameter,

and nwas the number of monitored samples. Table 5 lists the

input parameters determined by minimizing the objective

function for MLSS, DO, COD, ammonia nitrogen and nitrate

concentration, where the corresponding errors were 2.7%,

12.7%, 2.8%, 6.7% and 6.6%, respectively. Fig. 4 shows the close

agreement between the simulated and measured variables

after calibration, Case I.

4.2. Activated sludge distribution

The model was calibrated using one group of data at Case I

and further verified using another group of data obtained

under Cases II. In Fig. 5, comparison of the simulated and

measured 3D liquid velocities at Section 1-1 demonstrated

reasonable agreements between them with normalized stan-

dard error less than 9.6% at Case II.

Fig. 6 shows the numerically predicted horizontal velocity

component field 0.1 m from the bed and the simulated

streamwise-vertical velocity component field at longitudinal

sections taken along the and third channel in the carrousel

oxidation ditch for Case II. In Fig. 6a, vectors of the horizontal

velocity components are superimposed on contours of

magnitude of the horizontal velocity components. Strong

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4210Fig. 6 e Case II: (a) predicted horizontal velocity component distr

velocity component distributions along the third channel.ibution, 0.1 m above bed; (b) predicted stream-wise-vertical

clockwise-rotating forced vortices are evident at Stirrers 2, 3,

and 4, and a weaker anti-clockwise flow deflection can be seen

at Stirrer 1. The surface impellers help direct flow from thewall

to the interior of the OD. The gas distributors cause an up-

welling effect driving a ring vortex (which appears like two

counter-rotating vortices in the two-dimensional plane view)

throughwhich a vertical flow passes, later radiating away from

a stagnation point close to the free surface, as shown in Fig. 6b.

From the inlet onwards, the overall horizontal plane flow

conditions in Fig. 6a are as follows: the flow spreads out to

move along Channel 1; it then runs into the opposing flow from

Impeller 1 and enters a strong mixing zone towards the end of

the straight portion of Channel 1. It appears that somematerial

can remain trapped for considerable time in Channel 1. On

entering Channel 2, the flow is essentially uniform across the

breadth of the ditch, but runs into the opposing flow generated

by Stirrer 1 and is driven upwards by the air bubbles at the

bottom aerator, causing a persistent horizontal eddy-like

feature to form about halfway along Channel 2, perhaps

linked to the ring vortex generated by the bottom aeration

system. Stirrer 1 rotates in the anti-clockwise direction, and

this causes a weakly deflected flow into Channel 3. This flow

meets the rotating flow from Impeller 2 and the vertical upflow

from the bottom aerators (Fig. 6b), and again an eddy-like

feature forms in the mixing region towards the middle of

Channel 3. The flow in Channel 4 is partly driven by Impeller 2

and is initially fairly uniform, but appears to separate at a

stagnation point about 0.25 m along the internal wall, after

which a recirculation zone develops. The flow then meets the

strongly rotating vortex associated with Stirrer 2, and stag-

nates. A free anti-clockwise rotating eddy occupies the final

quarter of Channel 4, and is divided from the forced vortex of

Stirrer 2 by a transverse flow towards another stagnation point

at the external wall. Strong forced clockwise rotating vortices

can be seen at Stirrers 1, 2, and 3.Mixing regions and stagnation

points are evident between the stirrers.

Figs. 7 and 8 present the results obtained for the water

quality parameters for Case II. Fig. 7a shows the predicted

MLSS distribution, where the liquid sludge was taken to be in

the pseudo-solid phase closely coupled with liquid and gas

phases. Superimposed on Fig. 7a are six experimental mea-

surements of MLSS taken from the pilot-scale laboratory test

samples. It can be seen that the 3D three-phase model suc-

cessfully predicted the observed processes, i.e. activated

sludge tended to settle out where the liquid velocity was

small, especially in the nearly stagnant zones and slowly

rotating free eddies. Fig. 8a presents a comparison of the

modeled andmeasuredMLSS concentrations. The normalized

standard error between the numerically predicted and labo-

ratory measured MLSS concentration is less than 2.8%,

demonstrating that the proposed model usefully described

the interaction between liquid and pseudo-solid phases in the

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 211Fig. 7 e Case II: (a) MLSS concentration distribution, 0.1 m above

nitrate distribution, 0.25 above the bed (where D indicates meathe bed; (b), (c), (d) and (e) DO, COD, ammonia nitrogen and

sured data values at the sampling points in Fig. 3).

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4212OD in reproducing the transport and sedimentation of acti-

vated sludge.

4.3. Modified oxygen mass transfer

Fig. 7b shows the predicted DO distribution, on which are

superimposed five experimental measurements of DO taken

from the pilot-scale laboratory test samples. The DO content

rises in a fairly uniform fashion from aminimumat Impeller 1

to a maximum in Channel 3, after which DO content reduces

along Channel 4. Fig. 8b also shows the predicted DO con-

centrations for the validation Case II without and with the

pseudo-solid phase (based on the mass transfer model pre-

and post-modification) for Case II. It is obvious that the

modified model (with pseudo-solid phase) provides a much

closer fit to the measurements. This is further verified by

calculating the normalized standard errors obtained by

comparing the two sets of predicted values against the

experimental data. Here, the normalized standard error is

24.6% using the unmodified model for a 0.92 and b 0.44.The normalized standard error is reduced to 9.9% by modi-

fying the pseudo-solid phase effect through the parameter g

in Equation (6) (Mena et al., 2011; Su and Yu, 2006). The very

satisfactory agreement between the simulated and measured

DO concentrations confirms that the proposed model is

capable of providing an accurate description of the liquid-

egasesolid interaction in terms of oxygen mass transfer.

Fig. 8 e Comparison between measured and simulated concent

nitrogen, and (e) nitrate, at the sampling locations for Case II.4.4. Water quality

In predicting the water quality, the stoichiometric and kinetic

parameters in ASM1 were modified along with the oxygen

mass transfer model. Fig. 7cee show the predicted soluble

COD, ammonia nitrogen, and nitrate distributions. In each

plot, the five experimental measurements are superimposed.

For the COD and ammonia nitrogen, both contour plots tell a

similar story: a contaminant hot spot can be seen immediately

downstream of the inlet, and a further high concentration

zone at the end of Channel 1. The water quality then consis-

tently improves throughout Channels 2, 3, and 4. Fig. 7e shows

that the nitrate distribution follows an almost inverse trend to

the ammonia nitrogen, as would be expected. Fig. 8cee show

the close agreement between the numerically predicted and

measured values of soluble COD, ammonia nitrogen, and ni-

trate respectively at different locations in the OD for Case II. In

general, both soluble COD (Figs. 7c and 8c) and ammonia ni-

trogen (Figs. 7d and 8d) decline in the OD from a peak at the

inlet to a low values of about 16 mg/L and 4 mg/L respectively

at W2, after which there is little further change. The nitrate

concentration (Figs. 7e and 8e) increases substantially from

the inlet to W2, and seems to saturate at about 13 mg/L

beyond. The normalized standard errors between the simu-

lated and measured soluble COD, ammonia nitrogen and ni-

trate concentrations are 1.5%, 1.5% and 3.3%, respectively.

Hence, it can be concluded that the present model provides a

rations of: (a) MLSS, (b) DO, (c) soluble COD, (d) ammonia

dynamics was developed that satisfactorily represents phase

Amand, L., Carlsson, B., 2012. Optimal aeration control in anitrifying activated sludge process. Water Res. 46 (7),

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4 2132101e2110.APHA, 1998. Standard Methods for the Examination of Water and

Wastewater, 20th ed. American Public Health Association,Washington, DC.

Beccari, M., Dionisi, D., Giuliani, A., Majone, M., Ramadori, R.,motion and phase interactions in an oxidation ditch. Acti-

vated sludge flocs were interpreted as having pseudo-solid

phase, and the related parameters modified by calibrating

the sludge viscosity, settling capacity, oxygen mass transfer

rate, and carbon substrate uptake due to adsorption on acti-

vated sludge. The assumption that activated sludge was in a

pseudo-solid phase made it possible to describe its transport

and sedimentation. By modifying the oxygen mass transfer

model, and the stoichiometric and kinetic parameters, the

numerical model was able to represent biochemical trans-

formation of sludge. Experimental data on mixed liquor sus-

pended solid (MLSS), DO concentration, soluble COD,

ammonia nitrogen, and nitrate were obtained from a pilot-

scale carrousel oxidation ditch. The numerical predictions of

flow field in the OD showed the great importance of the im-

pellers and stirrers in promoting mixing. The excellent

agreement obtained between the numerical simulations and

sampled data measurements for water quality parameters

indicated that the numerical model accurately simulated the

kinematics of the multi-phase flow and the carbon oxidation,

nitrification and denitrification processes in the OD. The pre-

sent paper has focused on model calibration and verification,

and used the results to provide some insights into the

behavior of an oxidation ditch. In future, it is recommended

that the model be used for parameter studies as a design tool

in helping to select optimal arrangements of impellers, stir-

rers, and aeration zones in planned oxidation ditches.

Acknowledgments

Financial support from National Natural Science Foundation

of China (Grant No. 21261140336/B070302) is very much

appreciated. Sincere thanks are also to Professor Alistair G.L.

Borthwick at Department of Engineering Science, Oxford

University for his careful editing on the manuscript.

r e f e r e n c e sreasonable description of the interactions between species in

liquid and pseudo-solid phases. The modified stoichiometric

and kinetic parameters in presence of solid phase lie between

those of ASM and granular sludge system (Su and Yu, 2006).

5. Conclusions

A 3D three-phase fluid model based on computational fluid2002. Effect of different carbon sources on aerobic storage byactivated sludge. Water Sci. Technol. 45 (6), 157e168.Carucci, A., Dionisi, D., Majone, M., Rolle, E., Smurra, P., 2001.Aerobic storage by activated sludge on real wastewater. WaterRes. 35 (16), 3833e3844.

Cggn, A.S., Orhon, D., Rossetti, S., Majone, M., 2011. Short-termand long-term effects on carbon storage of pulse feeding onacclimated or unacclimated activated sludge. Water Res. 45(10), 3119e3128.

Dammel, E.E., Schroeder, E.D., 1991. Density of activated sludgesolids. Water Res. 25 (7), 841e846.

Environmental Protection Agency of China, 2009. Water quality-Determination of ammonia nitrogen-Nesslers reagentspectrophotometry (HJ 535-2009). China EnvironmentalScience Press, Beijing.

Fayolle, Y., Cockx, A., Gillot, S., Roustan, M., Heduit, A., 2007.Oxygen transfer prediction in aeration tanks using CFD.Chem. Eng. Sci. 62 (24), 7163e7171.

Fluent Corporation, 2005. Gambit 2.2 Users Guide, Lebanon, NH,USA.

FluentCorporation, 2006. Fluent6.3UsersGuide, Lebanon,NH,USA.Gillot, S., Heduit, A., 2008. Prediction of alpha factor values for

fine pore aeration systems. Water Sci. Technol. 57 (8),1265e1269.

Grijspeerdt, K., Verstraete, W., 1997. Image analysis to estimatethe settleability and concentration of activated sludge. WaterRes. 31 (5), 1126e1134.

Gujer, W., Henze, M., Mino, T., van Loosdrecht, M., 1999. Activatedsludge model No. 3. Water Sci. Technol. 39 (1), 183e193.

Henze, M., Gujer, W., Mino, T., van Loosdrecht, M., 2000. ActivatedSludge Models ASM1, ASM2, ASM2d and ASM3. Scientific andTechnical Report 9, IWA Task Group on MathematicalModelling for Design and Operation of Biological WastewaterTreatment. IWA Publishing, London.

Hong, K.H., Chang, D., Hur, J.M., Han, S.B., 2003. Novel phasedisolation ditch system for enhanced nutrient removal and itsoptimal operating strategy. J. Environ. Sci. Health Part A-Toxic/Hazard. Subst. Environ. Eng. 38 (10), 2179e2189.

Insel, G., Artan, N., Orhon, D., 2005. Effect of aeration on nutrientremoval performance of oxidation ditch systems. Environ.Eng. Sci. 22 (6), 802e815.

Jin, B., Wilen, B., Lant, P., 2004. Impacts of morphological, physicaland chemical properties of sludge flocs on dewaterability ofactivated sludge. Chem. Eng. J. 98 (1e2), 115e126.

Krishna, C., Van Loosdrecht, M.C.M., 1999. Substrate flux intostorage and growth in relation to activated sludge modeling.Water Res. 33 (14), 3149e3161.

Kulkarni, A.A., 2007. Mass transfer in bubble column reactors:effect of bubble size distribution. Industrial Eng. Chem. Res. 46(7), 2205e2211.

Kumar, B., Rao, A.R., 2009. Oxygen transfer and energy dissipationrate in surface aerator. Bioresour. Technol. 100 (11), 2886e2888.

Le Moullec, Y., Potier, O., Gentric, C., Leclerc, J.P., 2011. Activatedsludge pilot plant: comparison between experimental andpredicted concentration profiles using three differentmodelling approaches. Water Res. 45 (10), 3085e3097.

Littleton, H.X., Daigger, G.T., Strom, P.F., 2007. Application ofcomputational fluid dynamics to closed-loop bioreactors: II.Simulation of biological phosphorus removal usingcomputational fluid dynamics. Water Environ. Res. 79 (6),613e624.

Liu, Y., Liu, Y., Wang, Z., Yang, S., Tay, J., 2005. Influence ofsubstrate surface loading on the kinetic behaviour of aerobicgranules. Appl. Microbiol. Biotechnol. 67 (4), 484e488.

Mantziaras, I.D., Stamou, A., Katsiri, A., 2011. Effect of operationalcycletime length on nitrogen removal in an alternatingoxidation ditch system. Bioprocess Biosyst. Eng. 34 (5), 597e606.

Mena, P., Ferreira, A., Teixeira, J.A., Rocha, F., 2011. Effect of somesolid properties on gas-liquid mass transfer in a bubble

column. Chem. Eng. Process. 50 (2), 181e188.

Orhon, D., Ates , E., Sozen, S., Cokgor, E.U., 1997. Characterizationand COD fractionation of domestic wastewaters. Environ.Pollut. 95 (2), 191e204.

Pipes, W.O., 1969. Types of activated sludge which separatepoorly. J. Water Pollut. Control Fed. 41 (5), 714e724.

Schmid, M., Thill, A., Purkhold, U., Walcher, M., Bottero, J.Y.,Ginestet, P., Nielsen, P.H., Wuertz, S., Wagner, M., 2003.Characterization of activated sludge flocs by confocal laserscanning microscopy and image analysis. Water Res. 37 (9),2043e2052.

Squires, G.L., 2001. Practical Physics. Cambridge University Press,Cambridge, UK.

Stamou, A.I., 1994. Modelling oxidation ditches using the IAWPRCactivated sludge model with hydrodynamic effects. Water Sci.Technol. 30 (2), 185e192.

Stamou, A.I., 1997. Modelling of oxidation ditches using anopen channel flow 1-D advection-dispersion equation andASM1 process description. Water Sci. Technol. 36 (5),269e276.

Stamou, A.I., Katsiri, A., Mantziaras, I., Boshnakov, K.,Koumanova, B., Stoyanov, S., 1999. Modelling of analternating oxidation ditch system. Water Sci. Technol. 39(4), 169e176.

Stenstrom, M.K., Gilbert, R.G., 1981. Effects of alpha-factor, beta-factor and theta-factor upon the design, specification andoperation of aeration systems. Water Res. 15 (6), 643e654.

Su, K.Z., Yu, H.Q., 2006. A generalized model for aerobic granule-based sequencing batch reactor. 1. Model development.Environ. Sci. Technol. 40 (15), 4703e4708.

Wang, J., Qiu, Z., Chen, Z., Li, J., Zhang, Y., Wang, X., Zhang, B.,2012. Comparison and analysis of membrane fouling betweenflocculent sludge membrane bioreactor and granular sludgemembrane bioreactor. PLoS One 7 (7).

Yang, Y., Yang, J., Zuo, J., Li, Y., He, S., Yang, X., Zhang, K., 2011.Study on two operating conditions of a full-scale oxidationditch for optimization of energy consumption and effluentquality by using CFD model. Water Res. 45 (11), 3439e3452.

Zhang, D., Guo, L., Xu, D., Chen, Y., 2010. Simulation ofcomponent distributions in a full-scale carrousel oxidationditch: a model coupling sludge-wastewater two-phaseturbulent hydrodynamics with bioreaction kinetics. Environ.Eng. Sci. 27 (2), 159e168.

Zhou, X., Guo, X., Han, Y., 2012. Enhancing nitrogen removal in anOrbal oxidation ditch by optimization of oxygen supply:practice in a full-scale municipal wastewater treatment plant.Bioprocess Biosyst. Eng. 35 (7), 1097e1105.

wat e r r e s e a r c h 5 3 ( 2 0 1 4 ) 2 0 0e2 1 4214

Three-dimensional three-phase model for simulation of hydrodynamics, oxygen mass transfer, carbon oxidation, nitrification ...1 Introduction2 Methodology2.1 Model development2.1.1 Multiphase hydrodynamics model2.1.2 Species transport model2.1.3 Biochemical kinetics model2.1.4 Oxygen mass transfer model

2.2 Phase interaction2.2.1 Liquidgas interaction2.2.2 Liquid-solid interaction2.2.3 Gassolid interaction

2.3 Boundary conditions2.4 Parameter estimation

3 Experimental measurements in pilot-scale oxidation ditch4 Results and discussion4.1 Numerical model calibration4.2 Activated sludge distribution4.3 Modified oxygen mass transfer4.4 Water quality

5 ConclusionsAcknowledgmentsReferences