Waste Management 32 (2012) 382–388
Contents lists available at SciVerse ScienceDirect
Waste Management
journal homepage: www.elsevier .com/ locate/wasman
Dry-thermophilic anaerobic digestion of organic fraction of municipal solidwaste: Methane production modeling
L.A. Fdez-Güelfo a,⇑, C. Álvarez-Gallego a, D. Sales b, L.I. Romero García a
a Department of Chemical Engineering and Food Technology, Faculty of Science, University of Cadiz, 11510 Puerto Real, Cádiz, Spainb Department of Environmental Technologies, Faculty of Marine and Environmental Sciences, University of Cadiz, 11510 Puerto Real, Cádiz, Spain
a r t i c l e i n f o a b s t r a c t
Article history:Received 6 July 2011Accepted 3 November 2011Available online 30 November 2011
Keywords:Bioprocess engineeringDegradationEnvironmental engineeringFermentationReactor analysis
0956-053X/$ - see front matter � 2011 Elsevier Ltd.doi:10.1016/j.wasman.2011.11.002
⇑ Corresponding author.E-mail address: [email protected] (L.A. Fd
The influence of particle size and organic matter content of organic fraction of municipal solid waste(OFMSW) in the overall kinetics of dry (30% total solids) thermophilic (55 �C) anaerobic digestion havebeen studied in a semi-continuous stirred tank reactor (SSTR). Two types of wastes were used: syntheticOFMSW (average particle size of 1 mm; 0.71 g Volatile Solids/g waste), and OFMSW coming from a com-posting full scale plant (average particle size of 30 mm; 0.16 g Volatile Solids/g waste).
A modification of a widely-validated product-generation kinetic model has been proposed. Resultsobtained from the modified-model parameterization at steady-state (that include new kinetic parame-ters as K, YpMAX and hMIN) indicate that the features of the feedstock strongly influence the kinetics ofthe process. The overall specific growth rate of microorganisms (lmax) with synthetic OFMSW is 43%higher compared to OFMSW coming from a composting full scale plant: 0.238 d�1 (K = 1.391 d�1;YpMAX = 1.167 L CH4/gDOCc; hMIN = 7.924 days) vs. 0.135 d�1 (K = 1.282 d�1; YpMAX = 1.150 L CH4/gDOCc;hMIN = 9.997 days) respectively.
Finally, it could be emphasized that the validation of proposed modified-model has been performedsuccessfully by means of the simulation of non-steady state data for the different SRTs tested with eachwaste.
� 2011 Elsevier Ltd. All rights reserved.
1. Introduction
Anaerobic digestion (AD) of organic fraction of municipal solidwaste (OFMSW) has been a growing concern over the last yearsdue to the potential of the technology for fast production of meth-ane and/or hydrogen rich biogas and to save fossil energy (Jewellet al., 1993 and Mata-Álvarez et al., 2000). Methane productionrates must be increased to maximize revenues from power gener-ation and hence, to make digestion facilities more profitable. Opti-mization can be achieved by means of mathematical models forAD. Such models can be useful to predict both substrate utilizationand methane production rates for different operating conditions,and to determine the optimum and critical values for operationvariables.
Therefore, kinetic modeling is increasingly needed for abetter understanding of the performance of such systems, thedesign and operation of biological waste-treatment systems topredict system stability, effluent quality and waste stabilization(Angelidaki et al., 1993).
Currently, there are four main approaches to anaerobic modelingwhich help predict the behavior of the reactors (Garcia-Ocha et al.,
All rights reserved.
ez-Güelfo).
1999): based on kinetic equations (Chen and Hashimoto, 1980;Chen, 1983; Romero García, 1991), unstructured and non-segre-gated, unstructured and segregated (Hill, 1983) and, finally, struc-tured kinetic models. The last one type of models for dynamicsimulation of AD includes a complex matrix with kinetic constantsfor the different degradation steps of organic matter. Several authors(Bagley and Brodkorb, 1999; Batstone et al., 2002) have used theaforementioned models to predict the process response to specificoperating conditions in Continuous Stirred Tank Reactors (CSTR).Moreover, the Organic Loading Rate (OLR) and the solids retentiontime (SRT) are the main variables in everyday practice to carry outboth a proper design of reactor and an analysis of the CSTR perfor-mance (Halalsheh et al., 2001; Kiyohara et al., 2000).
In general, when using a specific waste (with roughly the samecharacteristics) as substrate, a decrease in SRT produces an in-crease in the consumed organic loading rate (OLRC), thus, in thespecific methane production (De la Rubia et al., 2006a,b). However,such a behavior presents a maximum from which the effect is theopposite. Depending on the feeding characteristics (particle sizeand organic matter content mainly), every system presents an opti-mum SRT below which the destabilization of the reactor and a dropin biogas generation and organic matter removal take place.
The effect SRT has on the effluent organic matter concentrationand methane production and its relation with the fed OLR (both
L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388 383
measured as Dissolved Organic Carbon –DOC- or Volatile Solids –VS-) can be used to determine the parameters of an adequate ki-netic model, such as the one proposed by Romero García (1991),for the prediction of process behavior.
2. Kinetic model development
The kinetic modeling helps predict the effect of the main pro-cess variables on system performance. In this sense, the develop-ment of adequate models and their parameterization, by fittingmodel equations to the experimental results obtained in specificassays, is a critical step.
What is more, since industrial plants for the treatment ofOFMSW by AD are characterized by the use of a wide range ofwaste composition to feed reactors, kinetic modeling plays a keyrole in determining optimum operating conditions.
In this work, the biogas generation model proposed by RomeroGarcía (1991) has been used. Such a model has successfully provedto fit experimental results of AD of different organic wastes: winevinasses (Romero García, 1991; Pérez et al., 2001a, 2001b), sludgesfrom Wastewater Treatment Plant (WWTP) (De la Rubia et al.,2006a, 2006b), and OFMSW (Fernández et al., 2010).
The model assumes that the substrate consumption rate (�rS) isobtained through the following equation:
ð�rsÞ ¼ lmaxðh� SÞ � ðS� SNBÞðh� SNBÞ
ð1Þ
where, SNB represents the Non-Biodegradable substrate concentra-tion (ML�3), S represents the substrate concentration (ML�3), h isthe maximum substrate concentration that can be utilized in thebiomass formation including cell material from microorganisms(ML�3) and lmax represents the maximum specific growth rate ofthe microorganisms (T�1).By including Eq. (1) in the mass balanceon a completely mixed reactor and assuming steady state condi-tions, the model equation used in this work can be obtained:
S ¼ SNB þS0 � SNB
lmaxhð2Þ
where S0 represents the initial substrate concentration (ML�3) and hrepresents the solids retention time (T).
In Eq. (2), as posited by Romero García (1991), it was assumedthat parameter h is equivalent to the initial substrate concentrationS0 for continuous operation. Besides, the above equation is basedon the overall growth rate (lmax) of the main microbial groups in-volved on the process and its value is representative of the rate-limiting step.
Eq. (2) can be modified by assuming that the quantity of non-biodegradable matter is proportional to the total substrate concen-tration in the feeding. Thus, SNB can be expressed as the fraction ofnon-biodegradable substrate (0 6 a < 1) with respect to the initialsubstrate concentration (S0), as in Eq. (3):
S ¼ aS0 þS0ð1� aÞlmaxh
ð3Þ
Evidently, the aforementioned equation could be easily linearizedas follows:
SS0¼ aþ 1� a
lmax
� �1h
� �ð4Þ
Based on Eq. (4) and assuming that substrate is converted nearlystoichiometrically into methane, Eq. (5) can be obtained:
cCH4¼ YP
S0 � Sh
� �¼ YP
S0 � ðaS0 þ S0ð1�aÞlmaxh
h
!
¼ YPS0ð1� aÞ
h1� 1
lmaxh
� �ð5Þ
where, cCH4 represents the volumetric methane production rate (ex-pressed as Litres of CH4/Litresreactor/d) and YP the methane yieldcoefficient (expressed as Litres of CH4/g consumed organic matter).
By linearizing Eq. (5), the following can be obtained:
cCH4h
S0¼ YPð1� aÞ � YPð1� aÞ
lmax
� �1h
� �ð6Þ
However, according to Leatherbarrow (1990), linearization ofequations implies a change in the fitted variables because it mod-ifies the random error distribution associated to the aforemen-tioned values, and thus, the hypotheses of the linear- regressiontechniques are invalidated. Nevertheless, non-linear regressiondoes not modify the error distribution since it is not necessary totransform any variables. In this case, the adjustment equation isas follows:
cCH4¼ YPð1� aÞS0
1h� 1
lmaxh2
!ð7Þ
Eq. (7) represents product generation model of Romero García(1991) and the coefficient YP is assumed to be constant. However,in this work it has been checked that YP is dependent on the SRTused. Such a behavior has also been reported by Bernd Linke(2006), who worked with solid waste from the potato processingindustry in thermophilic range. The author reported a decrease inboth the biogas and methane yield of 23.5% and 14%, respectively,when the feeding OLR was increased from 0.8 to 3.4 gVS/L/d.
In this sense, a mathematical empirical relationship is proposedby the authors in this work to explain the relationship between YP
and the SRT. The observed dependence of both parameters suggesta Monod-type curve, as expressed in Eq. (8).
Yp ¼YpMAX � ðh� hMINÞ
K þ ðh� hMINÞð8Þ
where YpMAX is the maximum methane production (expressed as LCH4/gDOCc), hMIN is the minimum SRT (expressed as days) belowwhich the system would become unstable and K is a Monod-typeconstant whose value is the difference (h�hMIN), expressed as days,for which Yp reaches half of its maximum value.
By analogy with the Monod equation, the value of K may beassociated with the ability of microorganisms to produce methanefrom substrate. Thus, the high values of the aforementionedparameter indicate that microorganisms present greater difficultyin the synthesizing methane from degraded organic matter.
The mathematical relation between YP and SRT depends on thetype of substrate. Thus, a mathematical relation must be obtainedfor each case. Last but not least, from the introduction of such anexpression into the equation of product generation model ofRomero García (1991) the ‘Modified Product Generation Model(MPGM)’ can be obtained.
For continuous processes that have not reached steady-state ofoperation and taking into account the material balance for an idealmixed reactor, the equations resulting from the extension of themodel of Romero García (1991) are:
SðS0; h; tÞ ¼S0 � ðSt¼0 � SSSÞ þ SSS � ðS0 � St¼0Þ � eðlMAX�1
h�t
ðSt¼0 � SSSÞ þ ðS0 � St¼0Þ � eðlMAX�1hÞ�t
ð9Þ
SSS ¼ a � S0 þS0 � ð1� aÞlmax � h
� �ð10Þ
where, St = 0 represents the substrate concentration when new con-ditions are imposed on the system (ML�3), SSS represents the sub-strate concentration of the system under Steady-State conditions(ML�3) and t represents the operating time elapsed since the newconditions were imposed on the system (t).
Fig. 1. Semi-continuous stirred tank reactor (SSTR).
Table 1Physicochemical characterization of wastes.
Analytical parameter OFMSWSYN OFMSWFSP
pH 7.78 7.98Density (kg/m3) 750 650Alkalinity (gCaCO3/L) 4.29 18.13Ammonia (gNH3-N/L) 1.68 0.79Total nitrogen (g/kg) 23.0 29.0Total solids (g/g sample) 0.90 0.71Volatile Solids (g/g sample) 0.71 0.16Fixed solids (g/g sample) 0.19 0.55Dissolved organic carbon (mg/g) 112.6 39.8Total volatile fatty acids (mgAcH/ L) 1440 32.65
384 L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388
Knowing the value of the substrate concentration in the efflu-ent, the methane generation may be estimated since the modelof Romero García (1991) considers that the methane productivityis directly proportional to the consumed organic loading rate.
cCH4¼ YP �
S0 � Sh
� �ð11Þ
Combining the expressions 8, 9 and 11, the following expressioncan be obtained for non steady-state process:
cCH4¼ 1
hYpMAXðh� hMINÞK þ ðh� hMINÞ
� �½S0ðS0 � St¼0 � SSSÞ þ SSSSt¼0�e lMAX�1
hð ÞtSt¼0 � SSS þ ðS0 � St¼0Þe lMAX�1
hð Þt
" #
ð12Þ
On the basis of the previously developed model for steady-stateand non steady-state operations, the two main aims of the paperare:
1. To obtain the kinetic parameters of the dry-thermophilic anaer-obic digestion in order to determine the influence of the type ofwaste in the overall kinetic process. To reach this objective, theequations of the MPGM for steady-state operation will beapplied to the experimental data of methane productivity.
2. Next, the results obtained from the parameterization of thekinetic equations in steady-state will be used to simulate andvalidate the MPGM, at non steady-state conditions, throughthe methane productivity data for the different SRTs tested witheach waste.
3. Materials and methods
3.1. Software
To fit kinetic equations to the experimental data and determinethe kinetic parameters of the model, linear and non-linear regres-sions have been applied using a statistical software program (Stat-graphics Plus v. 5.0 Professional Edition). The non-linear regressionis based on minimization-square-residues algorithm of Marquardt(1963).
3.2. Experimental equipment
A 5 L reactor without biomass recycling with a working volumeof 4.5 L was used (Fig. 1). The thermophilic conditions (55 �C) weremaintained by circulating water through the jacket from a thermo-static water bath (7 L). The reactor was equipped with a dischargeball valve and several input/output ports located at the top: a stir-ring paddle (stirring rate of 13 rpm), pH probe, biogas outlet, feedinlet and two extra pH controllers. In this type of reactor, theHydraulic Retention Time (HRT) and the Solid Retention Time(SRT) are equal.
The pH was adjusted using an on/off controller and correctionswere carried out by means of 5 N NaOH and 1 N H3PO4 solutions.The pH was maintained stable in the 6.5–8 range, which is suitablefor methanogenic microorganisms.
About the feeding regime, the SSTR reactor was fed once a day.The TS concentration of the feeding was adjusted to 30% (which ischaracteristic of dry anaerobic digestion) by adding tap water.
3.3. Feedstock
Two types of OFMSW were used: synthetic waste (OFMSWSYN),with an average particle size of 1 mm and prepared following sug-gestions of Martin et al. (1997) to provide the nutritional require-ments of the main microorganisms involved in the process, and theindustrial one, with an average particle size of 30 mm. The indus-
trial OFMSW coming from the trommel of the full scale treatmentplant ‘‘Las Calandrias’’ located at Jerez de la Frontera, Cádiz- Spain.This OFMSW will be called as OFMSWFSP, where FSP means ‘‘fullscale plant’’. The physicochemical characterization of both wastesis displayed in Table 1.
The synthetic waste was employed in order to avoid the distor-tion in the process caused by the changes in feed compositionwhen industrial feeding is used and, therefore, evaluate as theanaerobic digestion is influenced by this aspect. About the compo-sition of the OFMSWSYN, the main constituents were potato (6.2%w/w), cabbage (5.3% w/w), orange (4.9% w/w), apple (4.9% w/w),bread (3.5% w/w) and paper (55.8% w/w).
It is important to emphasize that, as shown in Table 1, com-pared to the OFMSWSYN, the OFMSWFSP had a low content of organ-ic matter (measured as Volatile Solids and dissolved organiccarbon). According to Pavan et al. (2000), OFMSWFSP can be classi-fied as non-easily biodegradable, since the VS/TS ratio is lower than0.7. In this study, the VS/TS ratio is 0.22, which, based on theassumptions of the authors, implies the following: first, the systemcan work with a high organic loading rate without the occurrenceof acidification problems, and, second, the biogas and methaneproduction is considerably smaller compared with the productionachieved from an easily biodegradable substrate.
Table 3Variables in the methane production model with OFMSWSYN.
h(days) DOCINFLUENT
(g/L)DOCEFFLUENT
(g/L)cCH4 (L CH4/LREACTOR d)
Yp (L CH4/gDOCC)
20 28.075 10.303 0.93 1.0515 28.075 10.806 1.15 0.9710 28.075 14.803 0.93 0.708 28.075 10.475 0.13 0.06
L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388 385
3.4. Methodology
As regards to OFMSWSYN, an initial sequence of four SRTs (20,15, 10 and 8 days) was tested once reactor stabilization at 25-day SRT was obtained. The start-up and stabilization stages arefully described by Fdez-Güelfo et al. (2010). The imposed organicloading rate during each SRT, expressed as mgDOC/L day andmgVS/L day, is shown in Table 2.
Once destabilized the reactor at 8-days SRT, the OFMSWFSP be-gan to be fed to the digester. In this case, the next sequence of fourSRTs was 15, 13, 12 and 10 days (see Table 2).
For both wastes, each SRT was maintained for at least threeperiods in order to reach stable operation. The operation time ofeach SRT is detailed in Table 2.
It is worth pointing out that the organic loading rate exceededlimit value suggested by Angelidaki et al. (2006) for a successfulstart-up. In fact, along stages 3 and 4, when the OFMSWSYN wasfed, the organic loading rate expressed as Volatile Solids (VS) washigher than 15 gVS/L/d.
3.5. Sampling and analysis
The following analytical determinations were used for wastecharacterizations and process monitoring and control: Total Solids(TS), Volatile Solids (VS), alkalinity, pH, Dissolved Organic Carbon(DOC), ammonium and Volatile Fatty Acids (VFA). To verify the sys-tem’s performance, all the parameters were analyzed once a dayand determinations were performed according to Standard Meth-ods (APHA-AWWA-WEF, 1995). TS, VS, pH, alkalinity and ammo-nium were determined directly from digested OFMSWSYN
samples. The DOC concentration was analyzed from the filtratesupernatant obtained by means of a lixiviation (10 g of digestedwaste in 100 ml of Milli-Q water during 20 min) of the effluentsamples. Samples for the DOC analysis were further filteredthrough a 0.47 lm glass fiber filter.
The volume of gas produced in the reactor was measured di-rectly by using a high precision flow gas meter – WET DRUM TG0.1 (mbar) – Ritter – through the Tedlar bag. The gas composition(hydrogen, methane and carbon dioxide) was determined by gaschromatography (SHIMADZU GC-14 B) with a stainless steel col-umn packed with Carbosive SII (diameter of 3.2 mm and lengthof 2 m) and a thermal conductivity detector (TCD). The injectedsample volume was 1 mL and the operational conditions were asfollows: 7 min at 55 �C; ramped at 27 �C min–1 until 150 �C; detec-tor temperature: 255 �C; injector temperature: 100 �C. The carrierwas helium and the flow rate used was 30 ml min–1. A standardgas (by Carburos Metálicos, S.A; composition: 4.65% H2; 5.33%N2; 69.92% CH4 and 20.10% CO2) was used for the systemcalibration.
Individual VFA (acetic, propionic, iso-butyric, butyric, iso-vale-ric, valeric, iso-caproic, caproic and heptanoic) levels were
Table 2Initial organic loading rate (OLR0) used in the different experimental stages.
Type ofwaste
Stage OLR0
gDOC/Ld gVS/Ld SRT(days)
Operation time(days)
OFMSWSYN 1 1.404 8.862 20 602 1.872 11.817 15 453 2.808 17.725 10 484 3.509 22.156 8 24
OFMSWFSP 5 0.701 2.929 15 456 0.809 3.380 13 397 0.877 3.661 12 368 1.052 4.393 10 30
analyzed, as in the case of the DOC, from the filtrate supernatantobtained by means of a lixiviation (10 g of digested waste in100 ml of Milli-Q water during 20 min) of the effluent samples.The samples for the VFA analysis were further filtered through a0.22 lm Teflon filter. VFA levels were determined by gas chroma-tography (SHIMADZU GC-17 A) with a flame ionization detectorand a capillary column filled with Nukol (polyethylene glycol mod-ified by nitro-terephthalic acid). The temperatures of the injectionport and detector were 200 �C and 250 �C, respectively. Heliumwas the carrier gas at 50 ml min�1. In addition, nitrogen gas wasused as make up at 30 ml min�1 flow rate. The total VFA was calcu-lated as the sum of individual VFA levels.
4. Results and discussion
The evolution of methane production vs. the SRT for each waste,expressed as LCH4/LREACTOR�d and LCH4/gDOCc, is displayed in Table3. Each SRT was maintained at least for 3 times to ensure the sys-tem stabilization.
4.1. Proposed ratio YP vs. SRT: parametrization of the new function
4.1.1. Synthetic OFMSW (OFMSWSYN)By using non-linear regression of Eq. (8) to fit experimental data
(Fig. 2), the following values were obtained:
YpMAX ¼ 1:167LCH4=gDOCc hMIN ¼ 7:924days
K ¼ 1:391days r2 ¼ 0:999
According to Fdez-Güelfo et al. (2011a), the critical solid reten-tion time (SRTCR) has been defined as the tested SRT that causes asignificant decrease in the daily methane production. In the case athand, the value obtained for hMIN (7.92 days) was very close to theSRTCR obtained experimentally (8 days).
Consequently, the destabilization of the system (Table 3) and aclear drop in the methane production and organic matter removaltake place.
Fig. 2. Average YP (expressed as litres of CH4 per gram of consumed DOC) versusSRT (expressed as days) for OFMSWSYN.
Table 4Variables in the methane production model with OFMSWFSP.
h(days)
DOCINFLUENT
(g/L)DOCEFFLUENT
(g/L)cCH4 (L CH4/LREACTOR�d)
Yp (L CH4/gDOCC)
15 10.525 7.425 0.19 0.9013 10.525 8.024 0.16 0.8412 10.525 8.160 0.13 0.6810 10.525 7.192 N.D. N.D.
N.D. not-detected.
Fig. 3. Average YP (expressed as litres of CH4 per gram of consumed DOC) versusSRT (expressed as days) for OFMSWFSP.
Table 5Results from the MPGM of Romero García (1991).
aAdjustment
equation:cCH4¼ YpMAX ðh�hMIN Þ
Kþðh�hMIN Þ
� �ð1� aÞS0
1h � 1
lmaxh2
� � Parameter r2
OFMSWSYN lmax = 0.238 d�1
a = 0.1640.999
OFMSWFSP lmax = 0.135 d�1
a = 0.4250.996
a The adjustment equation has been obtained combining Eqs. (7) and (8).
386 L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388
4.1.2. OFMSW coming from a composting full scale plant (OFMSWFSP)On closer inspection of the experimental values of average YP
for OFMSWFSP (Table 4) vs. the SRT (Fig. 3), a very similar relation-ship to the one obtained for OFMSWSYN was observed. Similarly, bynon-linear regression of YP vs. SRT (based on Eq. (8)), the parame-ters YpMAX, hMIN and K can be estimated:
YpMAX ¼ 1:150LCH4=gDOCc hMIN ¼ 9:997days
K ¼ 1:282days r2 ¼ 0:996
Again, the value of hMIN (9.997 days) obtained was very similarto the experimental SRTCR (Table 4): 10 days for this type of waste.For synthetic and industrial OFMSW, the SRTCR values determinedby non-linear regression were 7.924 days and 9.997 days, respec-tively. Such values indicate that hMIN is influenced by the type ofwaste. Considering that in the kinetic of the AD process the rate-limiting step is generally the hydrolysis stage (Chulhwan et al.,2005), this fact may be associated with transfer-matter limitationand microbial colonization of the particles of the wastes. Thus,higher hMIN values may be due to higher average particle size ofthe waste, as occurs with the OFMSWFSP.
By contrast, the YpMAX values obtained for both wastes were al-most equal (1.167 for OFMSWSYN and 1.150 L CH4/gDOCc forOFMSWFSP), which may suggests that the methane generationyield, in term of organic matter removal expressed as DOC, is inde-pendently of waste source (synthetic or industrial) and its biode-gradability and, therefore, this value could be considered like astoichiometric ratio.
In addition, the observed difference between the K values forboth wastes was lower than 8%, i.e. the values are not significantlydifferent. Taking into account that K parameter is representative ofthe microbial affinity to methane production from organic matterdegradation, this finding indicates that, although both wastes pres-ent different particle size and organic matter content, the nature ofthe organic matter is the same and, therefore, the affinity of themicroorganisms versus waste are analogous.
4.2. Application of the modified product generation model (MPGM)
Table 5 shows the values obtained for lmax and a from the non-linear regression of the Modified Product Generation Model(MPGM), which fits the Yp dependence on the SRT, to experimentaldata of methane production for both wastes: synthetic and indus-trial OFMSW.
The specific growth rate of the microorganisms (lmax) washigher when OFMSWSYN was fed to the system: 0.238 d�1 forOFMSWSYN vs. 0.135 d�1 for OFMSWFSP. In batch processes, AD ofMSW shows a spontaneous sequential phase separation and thespecific growth rate of every microbial group may be characterizedin the same run (Álvarez Gallego, 2005). However, it is very impor-tant to emphasize that the lmax value obtained in this work repre-sents the global maximum specific growth rate for the overallmicrobial populations involved in the process since the systemwas operated under semi-continuous regime. Thus, such a valueis representative of the microbial group rate-limiting of theprocess.
On the other hand, the value of a was lower when the OFMSW-SYN was fed: a reasonable finding taking into account the composi-tional characterization of the wastes (Table 1). The OFMSWSYN
presented higher organic matter concentration compared to theOFMSWFSP: 0.71 gVS/g for the synthetic waste and 0.16 gVS/g forthe industrial one. The value of parameter a obtained for OFMSW-SYN and OFMSWFSP was 0.164 and 0.452 respectively. The afore-mentioned results indicate that the biodegradable fraction ofsuch a waste is 83.6%. Therefore, it is a substrate with high biode-gradability compared with the industrial OFMSW which presents abiodegradable fraction of 57.4%.
The values of lmax obtained in this study using synthetic andindustrial OFMSW match up the literature values for the differentmicrobial populations (Fdez-Güelfo et al., 2011b). As can be seen,the values of lmax in this study are very similar to those obtainedby Álvarez Gallego (2005) in batch dry-thermophilic anaerobicdigestion of OFMSW. The author reported that the maximum spe-cific growth rate for the hydrolytic and acidogenic populations ran-ged between 0.08 and 0.18 d�1, and for methanogenic acetate-utilizing Archaeas between 0.23–0.28 d�1.
According to the Álvarez Gallego (2005) and the values of lmax
obtained, when OFMSWSYN is fed to the system, methanogenesisfrom acetate can be considered the rate-limiting step of the processrather than the hydrolytic stage. Such a fact can be associated withthe small particle size of the aforementioned waste (1 mm), whichencourages hydrolysis and solubility of the organic matter. Basedon the aforementioned assumptions, when the OFMSWFSP is fed,hydrolysis and acidogenesis of the waste may be considered therate-limiting step of the process; a reasonable finding taking intoaccount that the average particle size of such OFMSW is 30 timeshigher than that of the synthetic waste.
4.3. Modified product generation model (MPGM) validation
The application of MPGM in steady-state to the experimentaldata has allowed the parameterization of the equations and,
Fig. 4. Comparison between the accumulated methane values predicted by Eq. (12),based on MPGM, and experimental results when OFMSWSYN is used.
Fig. 5. Comparison between the accumulated methane values predicted by Eq. (12),based on MPGM, and experimental results when OFMSWFSP is used.
L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388 387
therefore, it has been possible to obtain the values of the kineticcoefficients. These coefficients allow predicting the behavior thatshould make the system when operating conditions are modified,i.e. the system has not reached steady-state conditions.
Therefore, the values of these coefficients allow checking thevalidity of the model since they have a particular physical andmicrobiological significance and, therefore, they cannot take ran-dom values.
In this way, the simulation of the process at non steady-statewas developed in order to determine the behavior of the systemwhen synthetic and industrial OFMSW are fed to reactor at differ-ent SRTs.
In the Figs. 4 and 5 are shown the comparison between the accu-mulated methane values predicted by Eq. (12), based on MPGM,and experimental results when OFMSWSYN and OFMSWFSP are fedto the system and the SRT is modified.
As can be seen, the Eq. (12) fits to experimental data and, hence,it can be concluded that MPGM is valid to describe successfully thebiomethanization process of both OFMSWs, independently of itsorganic matter content and particle size.
5. Conclusions
The application of the MPGM has allowed the characterizationof the overall specific growth rate of the microorganisms (lmax)for dry-thermophilic biomethanization of synthetic and industrialOFMSW. The validation of the MPGM has been performed success-fully by means of the simulation of non-steady state data for thedifferent SRTs tested with each waste.
From the kinetic analysis may be concluded that the type ofwaste (particle size and organic matter content) significantly influ-ences the hMIN parameter and, consequently, the lmax. However,the obtained values of YpMAX and K have been analogous for
synthetic and industrial feeding indicating that these parametersare independent of the type of waste.
The obtained results of hMIN and lmax indicate that wastes withhigher biodegradability and lower particle size enhance the kinet-ics of the biomethanization process. Furthermore, the findingspoint to methanogenesis from acetate as the rate-limiting step ofdry-thermophilic biomethanization when wastes of small particlesize are degraded. Otherwise, hydrolysis and acidogenesis are therate-limiting steps of the process.
Acknowledgements
This work was supported by the ‘‘Ministerio de Ciencia e Innova-ción’’ of Spain (Projects CTM2007-62164/TECNO and CTM2010-17654), the ‘‘Consejería de Innovación, Ciencia y Empresa’’ of the‘‘Junta de Andalucía, Spain’’ (Project P07-TEP-02472), the EuropeanRegional Development Fund (ERDF) and the ‘‘Ministerio de Educa-ción y Ciencia’’ of Spain (Project NovEDAR_Consolider CSD2007-00055).
References
Álvarez Gallego, C. J., 2005. Testing different procedures for the start up of a dryanaerobic co-digestion process of OFMSW and sewage sludge at thermophilicrange. University of Cádiz. Doctoral thesis.
Angelidaki, I., Ellegaarg, L., Ahring, B., 1993. A mathematical model for dynamicsimulation of anaerobic digestion of complex substrates: focusing on ammoniainhibition. Biotech. Bioeng. 42, 159–166.
Angelidaki, I., Chen, X., Cui, J., Kaparaju, P., Ellegaard, L., 2006. Thermophilicanaerobic digestion of source-sorted organic fraction of household municipalsolid waste: start-up procedure for continuously stirred tank reactor. WaterRes. 40, 2621–2628.
Bagley, D.M., Brodkorb, T.S., 1999. Modeling microbial kinetics in an anaerobicsequencing batch reactor: Model development and experimental validation.Water Environ. Res. 71, 1320–1332.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A.,2002. The IWA Anaerobic digestion Model no 1 (ADM1). Water Sci. Technol. 45,65–73.
Chen, Y.R., 1983. Kinetic analysis of anaerobic digestion of pig manure and itsdesign implications. Agric. Wastes 8, 65–81.
Chen, Y.R., Hashimoto, A.G., 1980. Substrate utilization kinetic-model for biologicaltreatment process. Biotechnol. Bioeng. 22, 2081–2095.
Chulhwan, P., Chunyeon, L., Sangyong, K., Yu, Ch., Howard, C.H., 2005. Upgrading ofanaerobic digestion by incorporating two different hydrolysis processes. J.Biosci. Bioeng. 100 (2), 164–167.
De la Rubia, M.A., Pérez, M., Romero, L.I., Sales, D., 2006a. Effect of solids retentiontime (SRT) on pilot anaerobic thermophilic sludge digestión. Process Biochem.41, 79–86.
De la Rubia, M.A., Pérez, M., Sales, D., Romero, L.I., 2006b. Municipal sludgedegradation kinetic in thermophilic CSTR pilot-scale reactors. AIChE J. 52 (12),4200–4206.
EF, A.P.H.A.-A.W.W.A.-W., 1995. Standard Methods for the examination of Waterand Wastewater. 19th ed., Washington DC. New York, USA.
Fdez-Güelfo, L.A., Álvarez-Gallego, C.J., Sales, D., Romero, L.I., 2010. Start-up ofthermophilic–dry anaerobic digestion of OFMSW using adapted modifiedSEBAC inoculum. Bioresour. Technol. 101, 9031–9039.
Fdez-Güelfo, L.A., Álvarez-Gallego, C.J., Sales, D., Romero, L.I., 2011a. Determinationof critical and optimum conditions for biomethanization of OFMSW in a Semi-continuous Stirred Tank Reactor. Chem. Eng. J. 171, 418–424.
Fdez-Güelfo, L.A., Álvarez-Gallego, C.J., Sales, D., Romero, L.I., 2011b. Dry-thermophilic anaerobic digestion of simulated organic fraction of municipalsolid waste: process modeling. Bioresour. Technol. 102, 606–611.
Fernández, J., Pérez, M., Romero, L.I., 2010. Kinetics of mesophilic anaerobicdigestion of the organic fraction of municipal solid waste: Influence of initialtotal solid concentration. Bioresour. Technol. 101 (16), 6322–6328.
Garcia-Ocha, F., Santos, V.E., Naval, L., Guardiola, E., Lopez, B., 1999. Kinetic modelfor anaerobic digestion of livestock manure. Enzyme Microb. Technol. 25,55–60.
Halalsheh, M., Copes, J., Den Elzen, J., Zeeman, G., Fyyad, M., Lettinga, G., 2001. Effectof SRT and temperature on biological conversion and the related scum formingpotential. In: Proc. 9th World Congress on Anaerobic Digestion, 505–510.Antwerpen, Belgium.
Hill, D.T., 1983. Simplified Monod kinetics of methane fermentation of animalwastes. Agric. Wastes 5, 1–16.
Jewell, W.J., Cummings, R.J., Richards, B.K., 1993. Methane fermentation of energycrops maximum conversion kinetics and in-situ biogas purification. Biomassand Bioenergy 5, 261–278.
388 L.A. Fdez-Güelfo et al. / Waste Management 32 (2012) 382–388
Kiyohara, Y., Miyahara, T., Mizuno, O., Naike, T., Ono, K., 2000. A comparative studyof thermophilic and mesophilic sludge digestion. J. Water Environ. Manag. 14(2), 150–154.
Leatherbarrow, R.B., 1990. Using linear and non-linear regression to fit biochemicaldata. Trends Biochem. Sci. 15 (12), 455–458.
Linke, Bernd, 2006. Kinetic study of therophilic anaerobic digestion of solid wastesfrom potato processing. Biomass Bioenergy 30, 892–896.
Marquardt, D.W., 1963. An algorithm for least-squares estimation of non-linearparameters. J. Soc. Ind. Appl. Math. 11, 431–441.
Martin, D.J., Potts, L.G.A., Reeves, A., 1997. Small-Scale simulation of wastedegradation in landfills. Biotechnol. Lett. 19, 683–686.
Mata-Álvarez, J., Mace, S., Llabres, P., 2000. Anaerobic digestion of organic solidwastes an overview of research achievements and perspectives. Bioresour.Technol. 74, 3–16.
Pavan, P., Battistoni, P., Mata-Álvarez, J., Cecchi, F., 2000. Performance ofthermophilic semi-dry anaerobic digestion process changing the feedbiodegradability. Water Sci. Technol. 41 (3), 75–82.
Pérez, M., Romero, L.I., Sales, D., 2001a. Kinetics of thermophilic anaerobes in fixed-bed reactors. Chemosphere 44 (5), 1201–1211.
Pérez, M., Romero, L.I., Sales, D., 2001b. 2001b. Organic Matter Degradation Kineticsin an Anaerobic Thermophilic Fluidised Bed Bioreactor. Anaerobe 07, 25–35.
Romero García, L.I., 1991. Development of a general mathematical model forfermentation processes: kinetics of anaerobic digestion. Doctoral Thesis.University of Cádiz. ISBN, 84-7786-109-9.
