R

Eco

BB

a

ARRAA

KMSAAPOOM

1

wooOdorgafApaci(a

T

0d

Journal of Hazardous Materials 172 (2009) 1430–1438

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journa l homepage: www.e lsev ier .com/ locate / jhazmat

esearch article

xtension of the anaerobic digestion model No. 1 (ADM1) to include phenolompounds biodegradation processes for simulating the anaerobic co-digestionf olive mill wastes at mesophilic temperature

oubaker Fezzani ∗, Ridha Ben Cheikhiogas Laboratory, URSAM, Industrial Engineering Department, Ecole Nationale d’Ingénieurs de Tunis, Université Tunis EL-Manar, BP. 37 Le Belvédère 1002 Tunis, Tunisia

r t i c l e i n f o

rticle history:eceived 9 February 2009eceived in revised form 31 July 2009ccepted 4 August 2009vailable online 11 August 2009

eywords:

a b s t r a c t

The anaerobic digestion model No. 1 (ADM1) was extended and enhanced to describe the anaerobic degra-dation processes of phenol compounds and homologues in olive mill wastewater (OMW) and olive millsolid waste (OMSW) at mesophilic temperature (37 ◦C). The original ADM1 basic structure was extendedby to the inclusion of phenolic compounds degradation processes into benzoate and then into acetate.The inhibitory effect of phenolic compounds on the fermenting process was accounted for by the use ofnon-competitive inhibition functions. New sensitive phenolic and benzoate parameters were calibratedand validated using updated experimental data from our previous study dealing with the mesophilic

athematical modellingimulationDM1naerobic co-digestionhenol compoundslive mill wastewater

anaerobic co-digestion of OMW with OMSW in semi-continuous tubular digesters. The simulating resultsrevealed that the extended ADM1 could predict with adequate accuracy the steady-state results of gasflow rate, effluent pH and soluble phenol concentrations of various influent concentrations at differenthydraulic retention times (HRTs).

© 2009 Elsevier B.V. All rights reserved.

live mill solid wasteesophilic temperature

. Introduction

Modelling and simulation of anaerobic digestion of olive millastes at mesophilic temperature could provide a guide line for

peration and optimisation of anaerobic reactors and to improveur understanding of the difficulties observed when co-digestingMW with olive mill solid wastes. Applying the IWA anaerobicigestion model No. 1 (ADM1) to simulate the mesophilic anaer-bic digestion of olive wastes is done in our previous work [1]. Theesults indicated that the ADM1 is capable to predict accuratelyas flow rates, methane and carbon dioxide percentages, pH, VFA,lkalinity and ammonium nitrogen levels in effluents for differenteed concentrations and under different HRTs. Nevertheless, theDM1 model lacks the ability in taking into account phenol com-ounds nor it could predict and monitor phenol levels in effluentfter anaerobic digestion. In fact, olive mill wastewater (OMW) is

haracterized by high concentrations of several organic compoundsncluding carbohydrates, proteins, lipids and phenolic substancesup to 30 g COD/l) [2]. The latter are responsible of phyto-toxicitynd antibacterial activity of OMW at high levels. Previous reports

∗ Corresponding author at: 28 Rue Larbi ZARROUK, 9000 Beja, Tunisia.el.: +216 97 37 69 69.

E-mail addresses: bfezzani2007@yahoo.fr, b.fezzani@laposte.net (B. Fezzani).

304-3894/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.jhazmat.2009.08.017

have concluded that the presence of either phenolic compoundsor volatile fatty acids (VFA) in OMW at high level (20–30 g COD/l)is the major cause of methanogenic inhibition of the anaerobicdigestion process of OMW. Anaerobic degradation of phenol com-pounds is a complex process and requires a consortium of variousmicroorganisms. Two possible pathways for mineralisation of phe-nol have been reported; either via benzoate into the benzoyl-CoApathway (mesophilic temperature) or via caproate (thermophilictemperature). At thermophilic temperature phenol is assumed tobe degraded through caproate pathway which is further convertedby acetogens to acetate but caproate is not confirmed experi-mentally as an intermediate, nor the bacteria responsible for thecaproate production has been identified [3].

However, at ambient and mesophilic temperatures someauthors have suggested that phenol is reduced in the presenceof nitrate to cyclohexanone and then n-caproate, which is subse-quently undergone beta-oxidation to form lower VFAs [4]. Later,others authors have assumed that during anaerobic degradation,phenols are first converted to benzoate [4–7]. Benzoate is furtherdearomatized to form cyclohexane carboxylic acid which is then

supposed to be converted to heptanoate [6]. The latter is eitherdegraded through beta-oxidation to form valerate, propionate andacetate [8] or is degraded directly to form propionate and butyrateboth of which can be further oxidized to acetate [9]. More recently,it has confirmed by many authors [10–13] that during mesophilic

B. Fezzani, R. Ben Cheikh / Journal of Hazard

Nomenclature

COD chemical oxygen demandDE differential equationsHRT hydraulic retention time (days)Ka,ph phenolic acid equilibrium constant (mol/l)kB,ph kinetic rate constant of phenol acid–base reaction

(mol/l/d)NH+

4 -N total ammonium nitrogen (mg/l) or (mg/kg TS)OMW olive mill wastewaterOMSW olive mill solid wasteQ influent and effluent flow rate (m3/d)SCOD soluble chemical oxygen demand (g COD/l)Sph soluble phenol concentration (g COD/l)Sbnz− benzoate ion concentration (mol/l)Sbnz total benzoate concentration (g COD/l)Sph− phenol ion concentration (mol/l)TS total solids (g/l)TCOD total chemical oxygen demand (g COD/l)TKN total Kjeldahl Nitrogen (g/l) or (g/kg TS)VS volatile solids (g/l)Vliq liquid reactor volume (m3)Xph particulate phenol concentration (g COD/l)XBph concentration of phenol biomass (g COD/l)XB,bnz concentration of benzoate biomass (g COD/l)

Greek letters�j kinetic rate equation for process j of ADM1 model

3 −1

azwvgSpeapet

benzoate degrading organisms) and six phenols conversion pro-

(kg COD m d )�i,j stoichiometric coefficients of ADM1 model

naerobic degradation, phenols are first degraded through ben-oate pathway or more precisely via benzoate 4-hydroxybenzoatehich is further transformed through the benzoyl-CoA pathway

ia cyclohexane carboxylate which is further converted by aceto-ens to acetate and hydrogen (see Fig. 1) using benzoate degradingyntrophus-like bacteria [14–17] accounted for about 49% of theopulation [16]. Furthermore, Li et al. [18] have deduced fromxperimental results that there is no intermediate VFA except

cetate from benzoate degradation process. In fact, neither pro-ionate, butyrate, valerate nor heptanoate are found in effluentven when the reactor is operated at high OLR. Taking into accounthese deductions about phenol degradation pathway at mesophilic

Fig. 1. Phenol compounds d

ous Materials 172 (2009) 1430–1438 1431

temperature and continuing the research of ADM1 enhancementmodelling, the objective of the present work is to include thephenolic compounds and phenol degradation processes into theADM1 with emphasize placed on simulating phenol contents ineffluent generated from digesters treating in co-digestion OMWand OMSW at mesophilic temperature (37 ◦C). The results of theextended ADM1 model were compared to phenol experimentalresults obtained from the updated study of the mesophilic anaero-bic co-digestion of OMW with OMSW in semi-continuous tubulardigesters fed with various initial substrate concentrations at differ-ent HRTs.

2. ADM1 model enhancement

2.1. Suggested modifications of the ADM1 basic structure

Fig. 2 illustrates the new ADM1 basic structure modified totake into account phenol compounds as composite, particulate andsoluble substrates. As can be seen, original ADM1 structure is mod-ified to include the following steps: disintegration of compositesolids (like OMSW) into particulate phenol compounds; hydroly-sis of particulate phenols to produce soluble phenols. Furthermore,we assume that all phenolic compounds (soluble and particulate)are being considered as a “lumped compound” without differencesbetween their own rates of degradation and expressed in terms ofphenolic acid equivalent. Finally and according to phenol degrada-tion benzoate pathway as presented above we assume that solublephenols and homologues are first converted to benzoate which isfurther converted by acetogens to acetate, hydrogen and to carbondioxide according to the reactions outlined in Table 1. Based onthese modifications the ADM1 model will become more realistic inphenol process monitoring, data analysis and to be a good startingpoint to model phenol degradation in olive mill wastes submittedto anaerobic digestion at ambient and mesophilic temperatures.

2.1.1. Additional growth kineticsThe inclusion of phenol degradation processes via benzoate

pathway into the ADM1 model requires the addition of five extrastate variables described in terms of COD (two for soluble and par-ticulate phenols, one for total benzoate and two for phenol and

cesses via benzoate pathway, one for phenol disintegration, onefor phenol hydrolysis, one for uptake of phenol, one for uptakeof benzoate and two for decay of phenol and benzoate degrad-ing organisms. Phenolic compounds conversion processes were

egradation pathway.

1432 B. Fezzani, R. Ben Cheikh / Journal of Hazardous Materials 172 (2009) 1430–1438

1 mo

dccwsgepsieft

22usi

TP

Fig. 2. Biochemical conversion processes according to IWA ADM

escribed by a number of kinetic expressions that describe theonversion rates in terms of substrate concentrations and rateonstants. The hydrolysis of particulate phenol compounds (Xph)as described by first order rate expression. The conversion of

oluble phenol compounds to benzoate was expressed by Monodrowth kinetic equation. The conversion of benzoate to acetate wasxpressed by Haldane growth kinetic equation. Endogenous decayrocess of phenol and benzoate degrading biomass were repre-ented by first order kinetics, and dead biomass were maintainedn the system as composite particulate material. The new kineticquations, new yield coefficients and new derived stoichiometryor all steps included in the extended ADM1 model are shown inhe updated ADM1 Petersen Matrix in Table 2.

.1.2. Additional basic equations

.1.2.1. Phenol liquid phase equations. The mass balance equationssed by the extended ADM1 to describe the dynamic behaviour ofoluble and particulate phenol, benzoate and biomass componentsn the liquid phase are as follows:

dXph

dt= Q

Vliq(Xph,in − Xph) + fXph,Xc �1 − �4a (1)

dSph

dt= Q

Vliq(Sph,in − Sph) + fSph,Xph

�4a − �7a (2)

dSbnz

dt= Q

Vliq(Sbnz,in − Sbnz) + fbnz,Sph

(1 − Yph)�7a − �7b (3)

dXBph

dt= Q

Vliq(XBph,in − XBph) + Yph�7a − �20 (4)

able 1henol compounds reactions of degradations as implemented in the updated ADM1.

No. Compound names Reactio

1 Phenol (mesophilic temperature) C6H6O2 Benzoic acid (mesophilic temperature) C7H6O

del [19] extended for phenol compounds degradation pathway.

dXB,bnz

dt= Q

Vliq(XB,bnz,in − XB,bnz) + Ybnz�7b − �21 (5)

where Xph,in and Xph are the input and output concentrations of par-ticulate phenol compounds, Sph,in and Sph are the input and outputconcentrations of soluble phenol compounds, Sbnz,in and Sbnz arethe input and output concentrations of total benzoate, XBph,in andXBph are the input and output concentrations of phenol biomass,XB,bnz,in and XB,bnz are the input and output concentrations of ben-zoate biomass, Vliq is the liquid reactor volume, Q is the flow intoand out of the reactor and the term

∑�j�ph are the sums of the

specific kinetic rates for process j multiplied by the stoichiometriccoefficient �ph (see Table 2).

2.1.2.2. Modified gas liquid phase equations. To involve solublephenol compounds and benzoate contributions according to thesuggested biochemical reactions presented in Table 1, acetate,hydrogen, inorganic carbon and inorganic nitrogen liquid phaseconcentration (Sac, Sh2

, SIC and SIN) equations were modified asfollowing:

dSac

dt= Q

Vliq(Sac,in − Sac) +

∑

j=1−19

�ac,j�j + (1 − Ybnz)fac,bnz�7b (6)

dSh2

dt= Q

Vliq(Sh2,in − Sh2

) +∑

j=1−19

�h2,j�j + (1 − Yph)fh2,ph�7a

+(1 − Ybnz)fh2,bnz�7b − �T,h2(7)

ns

+ 0.0139NH3 + 0.1596H2O + 0.3073CO2 → 0.0139C5H7NO2 + 0.87C7H6O2 + 0.51H2

2 + 0.02NH3 + 10.84H2O → 0.02C5H7NO2 + 0.51C2H4O2 + 5.9CO2 + 12.82H2

B. Fezzani, R. Ben Cheikh / Journal of Hazard

Tab

le2

Bio

chem

ical

rate

coef

fici

ents

(�i,j

)an

dki

net

icra

teeq

uat

ion

s(�

j)fo

rp

arti

cula

teco

mp

onen

ts(o

nly

add

itio

nal

pro

cess

esan

dco

mp

onen

tsto

AD

M1

are

show

n).

Com

pon

ent

jPr

oces

sC

omp

onen

ti

Kin

etic

rate

equ

atio

ns

(�j,

kgC

OD

m3

d−1

)7

87a

7b10

1116

a23

a23

bTo

tala

ceta

te(k

gC

OD

/m3

)H

ydro

gen

(kg

CO

D/m

3)

Solu

ble

ph

enol

(kg

CO

D/m

3)

Tota

lBen

zoat

e(k

gC

OD

/m3

)In

orga

nic

carb

on(k

mol

C/m

3)

Inor

gan

icn

itro

gen

(km

olN

/m3

)

Part

icu

late

ph

enol

(kg

CO

D/m

3)

Phen

olbi

omas

sd

egra

der

s(k

gC

OD

/m3

)

Ben

zoat

ebi

omas

sd

egra

der

s(k

gC

OD

/m3

)S a

cS h

2S p

hS b

nz

S IC

S IN

Xp

hX

Bp

hX

B,b

nz

1D

isin

tegr

atio

nf X

ph

,Xc

CX

ph

f Xp

h,X

ck d

isX

C

4aH

ydro

lysi

sof

par

ticu

late

ph

enol

f Sp

h,X

ph

CX

ph

−C

ph

−1k h

yd,p

h,X

ph

7aU

pta

keof

ph

enol

f ac,

bnz(1

−Y b

nz)

f h2,

bnz(1

−Y b

nz)

−1f b

nz,

ph

(1−

Y ph

)−[

C ph

−(1

−Y p

h)C

bnz

−Y p

hC b

iom

]−Y

ph

Nbi

omY p

hk m

,ph

(Sp

h/(

KS,

ph

+

((S b

nz)2

/KI,p

h))

)XB

ph

7bU

pta

keof

ben

zoat

e−1

C bn

z−

(1−

Y bn

z)C

ac−

Y bn

zC b

iom

−Ybn

zN

biom

Y bn

zk m

,bn

z(S

bnz/K

S,bn

z+

S bn

z)X

B,b

nzI 2

20D

ecay

ofp

hen

olbi

omas

sC

biom

−C

Xc

Nbi

om−

NX

c−1

k dec

,Bp

hX

Bp

h

21D

ecay

ofbe

nzo

ate

biom

ass

Cbi

om−

CX

cN

biom

−N

Xc

−1k d

ec,B

bnzX

B,b

nz

ous Materials 172 (2009) 1430–1438 1433

dSIC

dt= Q

Vliq(SIC,in − SIC)+

∑

j=1−19

�IC,j�j−fph,Xc CXph�1+(CXph

−Cph)�4a

− (Cph−(1−Yph)Cbnz − YphCbiom)�7a + (Cbnz − (1 − Ybnz)Cac

− YbnzCbiom)�7b+(Cbiom−CXc )�20+(Cbiom−CXc )�21−�T,CO2

(8)

dSIN

dt= Q

Vliq(SIN,in − SIN) +

∑

j=1−19

�IN,j�j − YphNbiom�7a

+(Nbiom − Nxc)�20 − YbnzNbiom�7b + (Nbiom − Nxc)�21 (9)

where∑

i=1−19�i�j (i = ac, h2, IC and IN) are the sums of the stoi-chiometric coefficients (�i) multiplied by the specific kinetic rates(�j) for process j of the original ADM1 model [19].

Besides, composite substrate concentration (Xc) equation waschanged, to involve composite substrate issued from phenolbiomass decay, in this manner:

dXC

dt= Q

Vliq(XC,in − XC) +

∑

j=13−19

�j + �20 + �21 − �1 (10)

where∑

i=13−19�j is the sum of the specific kinetic rates (�j) of theoriginal ADM1 model.

2.1.3. Modified pH simulation equationsThe charge balance equation of the original ADM1 was extended

to include the contribution of soluble phenolic compounds andbenzoate into acido-base reactions as follows:

SH+ − SOH− = SHCO−3

+ Sac−

64+ Spro−

112+ Sbu−

160+ Sva−

208+

Sph−

220+ Sbnz−

210

+ SAn− − SCat+ − SNH+4

(11)

where Sph− and Sbnz− are the phenol and benzoate ion concentra-tions implemented in the extended ADM1 model as kinetic rateequation as following:

dSph−

dt= −�A,ph

dSbnz−

dt= −�A,bnz (12)

where:

�a,ph = kB,ph(Sph− (Ka,ph + SH+ ) − Ka,phSph) (13)

�a,bnz = kB,bnz(Sbnz− (Ka,bnz + SH+ ) − Ka,bnzSbnz) (14)

where Ka,ph (mol/l) is the phenolic acid equilibrium constant andkB,ph (mol/l/d) is the kinetic rate constant of phenol acid–base reac-tion.

Ka,bnz (mol/l) is the benzoic acid equilibrium constant and kB,bnz(mol/l/d) is the kinetic rate constant of benzoic acid–base reaction.

2.1.4. Modification of the acetogenic inhibitory factorThe inhibition factor I5 applied to the rate of acetate uptake of

the original ADM1 was altered as follows:

I5 = IpH,ac · IIN,lim · INH3 · ITVFA · Iphenol (15)

Instead of:

I5 = IpH,ac · IIN,lim · INH (16)

3

in the original ADM1.Where I,TVFA and I,phenol are non-competitive functions added

to take into account the inhibition of methanogenic steps by highTVFA and high phenol levels [20], respectively.

1434 B. Fezzani, R. Ben Cheikh / Journal of Hazardous Materials 172 (2009) 1430–1438

Table 3Characteristics of the OMW and the sludge used in ADM1 as input main influent and initial conditions respectively.

Parameters Units OMW* Sludge*

pH – 7.5 ± 0.3 7.5 ± 0.1TCOD g COD/l 80 ± 2.5 20.5 ± 1.5SCOD g COD/l 60 ± 2.5 0.72 ± 0.1Total carbohydrates g COD/l 21 ± 1.5 0.5Total proteins g COD/l 14 ± 1.5 0.45Total Lipid g COD/l 17 ± 1.5 0.3Total Phenol g COD/l 24 ± 1.2 0.00Total Inert g COD/l 4 ± 1.5 19Sugars (monosaccharide) g COD/l 13 ± 1.5 0.005Amino acids g COD/l 4.5 ± 1.5 0.02LCFA g COD/l 7 ± 1.5 0.004Soluble phenols g COD/l 17 0.00Soluble inert g COD/l 7.5 ± 1.5 0.65Acetic acid g COD/l 5.5 ± 0.5 0.01Propionic acid g COD/l 1.75 ± 0.5 0.01Butyric acid g COD/l 2.85 ± 0.5 0.01Valeric acid g COD/l 0.85 ± 0.5 0.01Alkalinity g CaCO3/l 3.67 ± 0.25 2.5 ± 0.5NH+

4 -N mg N/l 470 ± 50 1300 ± 70Anions mol/l 0.053 ± 0.002 0.091 ± 0.002

( among

I

I

TC

(a

TI

Cations mol/l

*) Each value is an average of three replicates. Symbol (±) shows standards errors

Their expressions are as follows:

1

TVFA =

1 + STVFA/KI, TVFA(17)

phenol = 1

1 + (Sphenol)2/KI,phenol

(18)

able 4haracteristics of OMSW used as co-substrate.

Parameters Units OMSW*

TCOD g/kg TS 1100 ± 25Carbohydrates g/kg TS 345 ± 5Total lignin g/kg TS 440 ± 5Total proteins g/kg TS 72 ± 5Total lipids g/kg TS 100 ± 5Total polyphenols g/kg TS 18 ± 5TKN g N/kg TS 12 ± 1.5

*) Each value is an average of three replicates. Symbol (±) shows standard errorsmong replicates.

able 5nitial and estimated values of stoichiometric parameters.

Stoichiometricparameters

Names

fSI,Xc Soluble inert fraction in OMSWfXI,Xc Particulate inert fraction in OMSWfch,Xc Carbohydrates fraction in OMSWfpr,Xc Proteins fraction in OMSWfli,Xc Lipids fraction in OMSWfXph,Xc Particulate phenol fraction in OMSWfSph,Xph

Soluble phenol fraction issued from particulate phenol degradatifbnz,ph Benzoate fraction issued from soluble phenol degradationfac,bnz Acetate fraction issued from soluble phenol degradationfh2,Xc Hydrogen fraction issued from soluble phenol degradationfh2,bnz Hydrogen fraction issued from benzoate degradationNXc Nitrogen content in OMSWNI Nitrogen content in inert substrates of OMSWCXph Carbon content in particulate phenol compoundsCph Carbon content in soluble phenol compoundsKa,ph Phenolic acid equilibrium constantkB,ph Kinetic rate constant of phenol acid–base reactionCbnz Carbon content in benzoateKa,bnz Benzoate acid equilibrium constantkB,bnz Kinetic rate constant of benzoate acid–base reaction

0.343 ± 0.02 0.027 ± 0.004

replicates.

3. Lab-scale experimental data

Experimental results, against which the extended ADM1 simu-lations were compared, were obtained from our previous updatework dealing with the mesophilic anaerobic co-digestion ofOMW with OMSW in semi-continuous tubular digesters of 22 lvolume [21]. The OMW concentrations used were 24, 56 and80 g COD/l. The amount of the dry OMSW was 56 g TS perlitre of OMW. Alkalinity in the form of Ca(OH)2 was addedto all OMW concentrations (5–25 g/l of OMW) to provide aneutral medium (pH: 7.0–7.4) for the methanogenic archaeagrowth. In the run R1 to run R3 each digester was fed with aninfluent substrate concentration of 24 g COD/l at a HRT of 36, 24and 12 days respectively. In the run R4 to run R6, each digester was

fed with an influent substrate concentration of 56 g COD/l at a HRTof 36, 24 and 12 days respectively. Finally, in the run R7 to run R9each digester was fed with an influent substrate concentration of80 g COD/l at a HRT of 36, 24 and 12 days respectively.

Units Original suggested values Estimated values

– 0.1 0.013– 0.2 0.45– 0.2 0.35– 0.2 0.074– 0.3 0.1– – 0.026

on – – 1– – 0.87– – 0.51– – 0.51K mol N/g COD – 0.51K mol N/kg COD 0.0376/14 0.0107/14K mol N/kg COD 0.06/14 0.00K mol C/kg COD – 0.033K mol C/kg COD – 0.0319mol/l – 3.16e-10mol/l/d – 1e+10K mol C/kg COD – 0.034mol/l – 3.16e−10mol/l/d – 1e+10

B. Fezzani, R. Ben Cheikh / Journal of Hazardous Materials 172 (2009) 1430–1438 1435

Table 6Initial and estimated kinetic parameters of the implemented extended ADM1.

Kinetic parameters Name Unit Initial value Estimated value

kdis Disintegration constant d−1 0.5 0.001khyd,ph Phenol hydrolysis constant d−1 0.007 0.0015km,ph Monod maximum specific rate for phenol uptake d−1 3.43 15kS,ph Half saturation constant for phenol uptake kg COD m−3 0.175 30.5KI,ph Inhibition constant for phenol uptake kg COD m−3 0.214 50kdec,Bph Phenol biomass decline constant d−1 0.02 0.02Yph Phenol yield coefficient kg CODX/kg CODs 0.025 0.010km,bnz Monod maximum specific rate for benzoate uptake d−1 3.43 8kS,bnz Half saturation constant for benzoate uptake kg COD m−3 0.175 15.5kdec,Bbnz Benzoate biomass decline constant d−1 0.02 0.02Ybnz Benzoate yield coefficient kg CODX/kg CODs 0.025 0.0135kI,TVFA TVFA inhibition constant for acetate utilisers kg COD m−3 – 47.5kI,NH3 Ammonia inhibition constant for acetate utilisers K mol m−3 – 2.4e−3

Fig. 3. Effluent soluble phenol concentration simulation in comparison with exper-imental data for an influent TCOD concentration of 56 g COD/l at a HRT of 36 daysafter phenol parameters calibration.

Fig. 4. Validation results of simulations with experimental data for an influent TCODconcentration of 56 g COD/l at a HRT of 24 days.

4. Results and discussion

4.1. Substrates characteristics

Tables 3 and 4 show the characteristics of OMW, OMSW and thesludge used to determine the values of input steady-state variablesand the initials values of steady-state variables for the extended

ADM1 simulations of OMW mixed with OMSW.

1436 B. Fezzani, R. Ben Cheikh / Journal of Hazard

Fc

4

cwm

4

4

otptcetzt

4

ptip

was validated with the results of an influent TCOD concentration of56 g COD/l digested at a HRT of 24 days. In the second scenario themodel was validated with results of an influent TCOD concentrationof 24 g COD/l digested at HRTs of 24 and 12 days respectively. Finally

ig. 5. Validation results of simulations with experimental data for an influent TCODoncentration of 24 g COD/l at a HRT of 24 days.

.2. Model implementation

The ordinary differential equations of the ADM1 model wereoded and implemented using Matlab 7.0 software and integratedith the ODE15s solvers which solves stiff ODE systems as recom-ended by Rosen et al. [22].

.3. Model calibration

.3.1. Initial conditionsExperimental results of the mesophilic anaerobic co-digestion

f OMW (TCOD = 56 g/l) and OMSW at a HRT of 36 days were usedo assist the model calibration. Initial values of the original ADM1arameters were those suggested by Rosen and Jeppsson [23]. Ini-ial values of the model state variables were obtained from averageomposition of the sludge. Initial values of the extended param-ters (related to phenol degradation via benzoate pathway) werehose extracted from literature. Initial values of phenol and ben-oate yield parameters (Yph and Ybnz) were determined accordingo the reactions presented in Table 1.

.3.2. New parameters estimation

The estimation method used to identify the new sensitive

arameters (of the extended ADM1) to fit phenol model outputso effluents phenol experimental results was as follows: first, allnitial values were set to the model parameters. Then an heuristicrocedure was applied in adjusting the new sensitive parameters

ous Materials 172 (2009) 1430–1438

until fitting the simulation ADM1 outputs to the experimental efflu-ent phenol concentration results. The other parameters with lowsensitivity on phenol model outputs were applied without anymodification. Modified Stoichiometric coefficients implied in theextended ADM1 are presented in Table 5. Estimated parameter val-ues that fit better the experimental effluent phenol concentrationresults are given in Table 6.

Fig. 3 shows measured and simulated results of phenol concen-tration, gas flow rate, methane and carbon dioxide percentages andeffluent pH after model calibration. As can be seen effluent phe-nol concentration was predicted quite well at both transient andsteady-state periods.

4.4. Model validation

The calibrated model was validated on three quite different sce-narios by comparison with different experimental effluent phenolconcentration results of our previous work obtained at differentHRT and feed concentrations [21]. In the first scenario the model

Fig. 6. Validation results of simulations with experimental data for an influent TCODconcentration of 24 g COD/l at a HRT of 12 days.

B. Fezzani, R. Ben Cheikh / Journal of Hazardous Materials 172 (2009) 1430–1438 1437

Fc

iia

4

eedet0ais

4

ra2bmppw

ig. 7. Validation results of simulations with experimental data for an influent TCODoncentration of 80 g COD/l at a HRT of 36 days.

n the third scenario the model was validated with the results of annfluent TCOD concentration of 80 g COD/l digested at HRTs of 36nd 24 days respectively.

.4.1. Scenario one: medium feed concentrationFig. 4 shows measured and simulated results of gas flow rate,

ffluent pH and soluble phenol concentration issued from an influ-nt TCOD concentration of 56 g COD/l digested at a HRT of 24ays and applying the same pervious calibrated parameters. Efflu-nt pH and soluble phenol concentration were well predicted byhe extended model with minor deviations within 0.2–0.5 and.3–0.5 g COD/l respectively. Whereas, gas flow rate was predictedccurately in steady-state period by the extended ADM1. However,t revealed some deviations within 5–7 and 1–2 l/d at transient andteady-state periods respectively.

.4.2. Scenario two: low feed concentrationFigs. 5 and 6 show measured and simulated results of gas flow

ate, pH and soluble phenol concentrations in effluents issued fromn influent TCOD concentration of 24 g COD/l digested at HRTs of4 and 12 days using the previous calibrated parameters. As can

e seen, effluent pH values were well predicted by the extendedodel at HRTs of 24 and 12 days. Whereas, gas flow rates were

redicted with high accuracy at both transient and steady-stateeriods of the HRT of 24 days but they revealed some discrepanciesithin 5–7 l/d at transient period of the HRT of 12 days. Also some

Fig. 8. Validation results of simulations with experimental data for an influent TCODconcentration of 80 g COD/l at a HRT of 24 days.

deviations, of about 0.5–2 g COD/l, were noted between simulationsand measurements for effluent soluble phenol levels at steady-stateperiods of both HRTs.

4.4.3. Scenario three: high influent phenol concentrationFigs. 7 and 8 show measured and simulated results of gas flow

rate, pH and soluble phenol concentration in effluents issued froman influent TCOD concentration of 80 COD/l digested at HRTs of36 and 24 days applying the previous calibrated parameters. Gasflow rates and effluent pH values were predicted with some devia-tions within 1–3 l/d and 0.2–0.5 respectively at steady-state periodsof both HRTs. Whereas, effluent soluble phenol concentrationswere well predicted in spite of the minor discrepancies within0.5–1 g COD/l noted at both HRTs.

5. Conclusions

This work has proved that the extended ADM1 model was able topredict the steady state of gas flow rate, pH and phenol concentra-tion in effluents rejected from anaerobic semi-continuous tubulardigesters treating in co-digestion OMW with OMSW at mesophilic

◦

temperature (37 C) and under different lab-scale operating con-ditions and could tolerate the change in both feed concentrationsand HRTs with the same calibrated parameters. Effluents pH val-ues and phenol concentrations were predicted quite well in mostcases. Also gas flow rates were well predicted for different feed

1 azard

cbfi3

dldpoaea

R

[

[

[

[

[

[

[

[

[

[

[

[

[

438 B. Fezzani, R. Ben Cheikh / Journal of H

oncentrations digested at different HRTs. But some deviationsetween measurements and model simulations were observedrstly for feed concentration of 80 g COD/l digested at a HRT of6 days.

Secondly at the transient periods of all feed concentrationsigested at different HRTs. These inconsistencies between simu-

ations and experimental results may due to the fact, that ADM1ifferential equations were non-linear equations and it was com-licated to optimize all the sensitive parameters by adjusting ADM1utputs with all main experimental results simultaneously withny parameters identification method. Finally, the essentials of thextended model outputs are generally valid, although more studiesre still needed to improve this model.

eferences

[1] B. Fezzani, R. Ben Cheikh, Modelling of the mesophilic anaerobic co-digestionof olive mill wastewater with olive mill solid waste using anaerobic digestionmodel No. 1 (ADM1), Bioresour. Technol. 99 (2008) 6565–6577.

[2] G. Lyberatos, M.S. Fountoulakis, S.N. Dokanakis, M.E. Kornaros, G.G. Aggelis,Removal of phenolics in olive mills wastewaters using the white-rot fungusPleurotus ostreatus, Water Res. 36 (2002) 4735–4744.

[3] H.H.P. Fang, D.W. Liang, T. Zhang, Y. Liu, Anaerobic treatment of phenol inwastewater under thermophilic condition, Water Res. 40 (2006) 427–434.

[4] G. Knoll, J. Winter, Anaerobic degradation of phenol in sewage-sludge—benzoate formation from phenol and CO2 in the presence ofhydrogen, Appl. Microbiol. Biotechnol. 25 (1987) 384–391.

[5] G. Knoll, J. Winter, Degradation of phenol via carboxylation to benzoate by adefined, obligate syntrophic consortium of anaerobic bacteria, Appl. Environ.Microbiol. 30 (1989) 318–324.

[6] T. Kobayashi, T. Hashinaga, E. Mikami, T. Suzuki, Methanogenic degradationof phenol and benzoate in acclimated sludge, Water Sci. Technol. 21 (1989)55–65.

[7] K.L. Londry, P.M. Fedorak, Benzoic-acid intermediates in the anaerobicbiodegradation of phenols, Can. J. Microbiol. 38 (1992) 1–11.

[8] C.L. Keith, R.L. bridges, L.R. Fina, K.L. Iverson, J.A. Cloran, The anaerobic decom-position of benzoic acid during methane fermentation. I: Dearomatization ofthe ring and volatile fatty acids formed on ring rupture, Arch. Microbiol. 118(1978) 173–176.

[

ous Materials 172 (2009) 1430–1438

[9] L.R. Fina, R.L. Bridges, T.H. Coblentz, F.F. Roberts, The anaerobic decompositionof benzoic acid during methane fermentation. III: The fate of carbon four andthe identification of propionic acid, Arch. Microbiol. 118 (1978) 169–172.

10] B.R. Sharak Genthner, G.T. Townsend, P.J. Chapman, Para-hydroxybenzoate asan intermediate in the anaerobic transformation of phenol to benzoate, FEMSMicrobiol. Lett. 78 (1991) 265–270.

11] A. Karlsson, J. Ejlertsson, D. Nezirevic, B.H. Svensson, Degradation of phenolunder meso- and thermophilic, anaerobic conditions, Anaerobe 5 (1999) 25–35.

12] B. Schink, B. Philipp, J. Müller, Anaerobic degradation of phenol compounds,Naturwissenschaften 87 (2000) 12–23.

13] H.H.P. Fang, Y. Liu, S.Z. Ke, T. Zhang, Anaerobic degradation of phenol in wastew-ater at ambient temperature, Water Sci. Technol. 49 (2004) 95–102.

14] T. Li, J.G. Bisaillon, R. Villemur, L. Letourneau, K. Bernard, F. Lépine, R. Beaudet,Isolation and characterization of a new bacterium carboxylating phenol tobenzoic acid under anaerobic conditions, J. Bacteriol. 178 (1996) 2551–2558.

15] M.S. Elshahed, V.K. Bhupathiraju, N.Q. Wofford, M.A. Nanny, M.J. McIner-ney, Metabolism of benzoate, cyclohex-1-ene carboxylate, and cyclohexanecarboxylate by “Syntrophus aciditrophicus” strain SB in syntrophic associ-ation with H-2-using microorganisms, Appl. Environ. Microbiol. 67 (2001)1728–1738.

16] M.S. Elshahed, M.J. McInerney, Benzoate fermentation by the anaerobicbacterium Syntrophus aciditrophicus in the absence of hydrogen-usingmicroorganisms, Appl. Environ. Microbiol. 67 (2001) 5520–5525.

17] T. Zhang, S.Z. Ke, Y. Liu, H.P. Fang, Microbial characteristics of a methanogenicphenol-degrading sludge, Water Sci. Technol. 52 (2005) 73–78.

18] Y.Y. Li, H.H.P. Fang, T. Chen, H. Chui, USAB treatment of wastewater containingconcentrated benzoate, J. Environ. Eng. 121 (1995) 748–751.

19] D.J. Batstone, J. Keller, I. Angelidaki, S.V. Kalyuzhnyi, S.G. Pavlostathis, A. Rozzi,W.T.M. Sanders, H. Siegrist, V.A. Vavilin, Anaerobic digestion model No.1(ADM1), in: Scientific and Technical Report No. 13., International Water Asso-ciation (IWA) Publishing, London, UK, 2002.

20] P. Olguin-Lora, L. Puig-Grajales, E. Razo-Flores, Inhibition of the acetoclasticmethanogenic activity by phenol and alkyl phenols, Environ. Technol. 24 (2003)99–100.

21] B. Fezzani, R. Ben Cheikh, Anaerobic co-digestion of olive mill wastewaterwith olive mill solid waste in a tubular digester at mesophelic temperature,Bioresour. Technol. 98 (2007) 169–179.

22] C. Rosen, D. Vrecko, K.V. Gernaey, U. Jeppsson, Implementation ADM1 for

benchmark simulations in matlab/simulink, in: The First International Work-shop on the IWA Ananerobic Digestion Model No. 1 (ADM1), IWA publishing,UK, 2005, pp. 11–18.

23] C. Rosen, U. Jeppsson, 2002, Anaerobic COST benchmark model description,version 2.0, Technical report, Department of Industrial Electrical Engineeringand Automation (IEA), Lund University, Lund, Sweden.