We investigated the ability of the default ADM1 parameter set to simulate mesophilic anaerobic CSTRs fed with WAS and PS. For acetotrophic methanogenesis, ADM1 default parameters for biomass growth and ammonia inhibition enabled accurate simulation of biomass activity of WAS acclimated inoculum. On PS acclimated inoculum, parameter estimation was necessary due the high concentration of free ammonia, which induced biomass shifts and consequently, invalidated the default ammonia inhibition incorporated into the ADM1. The propionate degradation stage also required a calibration step and this paper highlights inaccurate modelling hypotheses for propionate acetogenic biomass. Concerning acidogenesis, anaerobic activity related to the degradation of pure monomers was shown to be non-representative of the overall activity of these biomass groups. Indeed, this kind of test cannot be used for acidogenesis calibration, thus highlighting the difficulties involved in determining the initial state for batch tests and in estimating ADM1 parameters using specific substrate degradation tests. Nevertheless, except for LCFA acidogenesis, results showed that this step is not rate-limiting. Hence, in addition to specific disintegration constants, the kinetics of methane production in continuous reactor modelling were shown to be mainly driven by cellulose hydrolysis, LCFA acidogenesis (in PS acclimated inoculum) and acetotrophic methanogenesis.
CemOA : archive ouverte d'Irstea / Cemagref
6. References
Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J.L., Guwy, A.J., Kalyuzhnyi, S., Jenicek P. Van Lier, J.B. 2009. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: A proposed protocol for batch assays.
Water Science and Technology 59 (5), 927-934.
Batstone D.J., Keller J., Angelidaki I., Kaluzhnyi S.V., Pavlostatis S.G. Rozzi A., Sanders W.T.M., Siegrist H., Vavilin. 2002. Anaerobic Digestion Model No.1, Scientific and Technical Report No.13, IWA Publishing, ISBN: 1 900222 78 7
Batstone, D.J., Pind, P.F., Angelidaki, I. 2003. Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate. Biotechnology and Bioengineering 84 (2), 195-204
Batstone, D.J., Torrijos, M., Ruiz, C., Schmidt, J.E. (2004). Use of an anaerobic sequencing batch reactor for parameter estimation in modelling of anaerobic digestion. . Wat. Sci.
and tech., 50(10), 295-303.
Batstone, D.J., Keller, J., Steyer, J.P. 2006. A review of ADM1 extensions, applications, and analysis: 2002-2005. Water Science and Technology, 54 (4), 1-10.
Batstone, D.J., Tait, S., Starrenburg, D. 2009. Estimation of hydrolysis parameters in full-scale anerobic digesters. Biotechnology and Bioengineering 102 (5), pp. 1513-1520 Bernard, O., Chachuat, B., Hélias, A., Rodriguez, J. 2006. Can we assess the model
complexity for a bioprocess: Theory and example of the anaerobic digestion process.
Water Science and Technology 53 (1), pp. 85-92.
Blumensaat, F., Keller J.. 2005. Modelling of two-stage anaerobic digestion using the IWA Anaerobic Digestion Model No. 1 (ADM1). Water Research, 39(1) 171-183.
Boubaker, F., Ridha, B.C. 2006. Modelling of the mesophilic anaerobic co-digestion of olive mill wastewater with olive mill solid waste using anaerobic digestion model No. 1 (ADM1). Bioresource Technology 99 (14). 6565-6577.
Conklin, A., Stensel, H.D., Ferguson, J. 2006. Growth kinetics and competition between Methanosarcina and Methanosaeta in mesophilic anaerobic digestion. Water Environment Research, 78 (5), 486-496.
Feng, Y., Behrendt, J., Wendland, C., Otterpohl, R. 2006. Parameter analysis of the IWA anaerobic digestion model No. 1 for the anaerobic digestion of blackwater with kitchen refuse. Water Science and Technology 54 (4), 139-147.
Galí, A., Benabdallah, T., Astals, S., Mata-Alvarez, J. 2009. Modified version of ADM1 model for agro-waste application. Bioresource Technology 100 (11), 2783-2790.
Ganesh Kumar, A., Kamatchi, P., Umashankari, J., Vidhya, S., Sriyutha Murthy, P., &
Sekaran, G. (2008). Acidogenic fermentation of proteinaceous solid waste and characterization of different bioconversion stages and extracellular products. Biodegradation, 19(4), 535-543.
Girault R., Rousseau P., Steyer J.P., Bernet N., Béline F. 2011. Combination of batch experiments with continuous reactor data for ADM1 calibration: application to anaerobic digestion of pig slurry. Water Scienbce and Technology. 63 (11), pp. 2575-2582.
CemOA : archive ouverte d'Irstea / Cemagref
Kalfas, H., Skiadas, I.V., Gavala, H.N., Stamatelatou, K., Lyberatos, G. 2006. Application of ADM1 for the simulation of anaerobic digestion of olive pulp under mesophilic and thermophilic conditions. Water Science and Technology 54 (4), 149-156.
Kalyuzhnyi, S.V. 1997. Batch anaerobic digestion of glucose and its mathematical modeling.
II. Description verification and application of model. Bioresource Technology, 59(2-3), 249-258.
Karakashev, D., Batstone, D.J., and Angelidaki, I. (2005) “Influence of the environmental conditions on the methanogenic composition of anaerobic biogas reactors” Applied Environmental Microbiology 71(1):331-338.
Koch, K., Lübken, M., Gehring, T., Wichern, M., Horn, H. 2010. Biogas from grass silage - Measurements and modeling with ADM1. Bioresource Technology 101 (21), pp. 8158-8165
Lübken, M., Wichern, M., Schlattmann, M., Gronauer, A., Horn, H. 2007. Modelling the energy balance of an anaerobic digester fed with cattle manure and renewable energy crops. Water Research 41 (18), 4085-4096.
Lucas T., Le Ray D., Peu P., Wagner M., Picard S. 2007. A new method for continuous assessment of CO2 released from dough baked in ventilated ovens Journal of Food Engineering, 81( 1), 1-11
Nelder J. A.and Mead R. 1965. A Simplex Method for Function Minimization. The Computer Journal. 7 (4) 308-313.
Pavlostathis, S.G., Giraldo-Gomez, E. 1991. Kinetics of anaerobic treatment: A critical review. Critical Reviews in Environmental Control 21 (5-6), 411-490.
Pind, P.F., Angelidaki, I., Ahring, B.K. 2003. Dynamics of the anaerobic process: Effects of volatile fatty acids. Biotechnology and Bioengineering 82 (7), 791-801.
Ramirez, I., Mottet, A., Carrère, H., Déléris, S., Vedrenne, F., Steyer, J.-P. 2009. Modified ADM1 disintegration/hydrolysis structures for modeling batch hermophilic anaerobic digestion of thermally pretreated waste activated sludge. Water research 43 (14), 3479-3492.
Raposo, F., Fernández-Cegrí, V., De la Rubia, M., Borja, R., Béline, F., Cavinato, C., Demirer, G., (...), de Wilde, V. 2011. Biochemical methane potential (BMP) of solid organic substrates: Evaluation of anaerobic biodegradability using data from an international interlaboratory study. Journal of Chemical Technology and Biotechnology.
In press.
Rodríguez, J., Lema, J.M., van Loosdrecht, M.C.M., Kleerebezem, R. 2006. Variable stoichiometry with thermodynamic control in ADM1. Water Science and Technology 54 (4), 101-110.
Schnürer, A., Nordberg, Å.. 2008. Ammonia, a selective agent for methane production by syntrophic acetate oxidation at mesophilic temperature. Water Science and Technology 57 (5), 735-740
Siegrist, H., Vogt, D., Garcia-Heras, J.L., Gujer, W. 2002. Mathematical model for meso- and thermophilic anaerobic sewage sludge digestion. Environmental Science and Technology 36 (5), 1113-1123.
Thamsiriroj, T., Murphy, J.D. 2011. Modelling mono-digestion of grass silage in a 2-stage CSTR anaerobic digester using ADM1. Bioresource Technology 102 (2), pp. 948-959
CemOA : archive ouverte d'Irstea / Cemagref
Vavilin, V.A., Fernandez, B., Palatsi, J., Flotats, X. 2008. Hydrolysis kinetics in anaerobic degradation of particulate organic material: An overview. Waste Management, 28 (6), 939-951
Wett, B., Schoen, M., Phothilangka, P., Wackerle, F., Insam, H. 2007. Model-based design of an agricultural biogas plant: Application of Anaerobic digestion model No. 1 for an improved four chamber scheme. Water Science and Technology 55 (10), 21-28
Wichern, M., Gehring, T., Fischer, K., Andrade, D., Lübken, M., Koch, K., Gronauer, A., Horn, H. 2009. Monofermentation of grass silage under mesophilic conditions:
Measurements and mathematical modeling with ADM 1. Bioresource Technology 100 (4), 1675-1681.
Yasui, H., Goel, R., Li, Y.Y., Noike, T. 2008. Modified ADM1 structure for modelling municipal primary sludge hydrolysis. Water Research 42 (1-2), 249-259.
CemOA : archive ouverte d'Irstea / Cemagref
Sous-chapitre 4 B:
A waste characterisation procedure for ADM1 implementation based on degradation kinetics
Girault R.a,b, Bridoux G.c, Nauleau F.c, Poullain C.c, Buffet J.a,b, Sadowski A.G.d, Steyer J.P.e, Béline F.a,b
a Cemagref, UR GERE, 17 av. de Cucillé, CS 64427, F-35044 Rennes, France.
(Tel. (+33)2 23 48 21 42 – Fax: (+33)2 23 48 21 15 – e-mail: romain.girault@cemagref.fr)
b Université Européenne de Bretagne, F-35044 Rennes, France
c SAUR, Recherche et développement, Atlantis, 1, av. Eugène Freyssinet, F-78280 Guyancourt, France.
d IMFS de Strasbourg (CNRS-UdS-ENGEES-INSA), France
e3 INRA, UR50, Laboratoire de Biotechnologie de l'Environnement, Avenue des Etangs,Narbonne, F-11100, France.
Abstract
In this study, a procedure accounting for degradation kinetics was developed to split the total COD of a substrate into each input state variable required for Anaerobic Digestion Model n°1. The procedure is based on the combination of batch experimental degradation tests (“anaerobic respirometry”) and numerical interpretation of the results obtained (optimisation of the ADM1 input state variable set). The effects of the main operating parameters, such as the substrate to inoculum ratio in batch experiments and the origin of the inoculum, were investigated. Combined with biochemical fractionation of the total COD of substrates, this method enabled determination of an ADM1-consistent input state variable for each substrate with affordable identifiability. The substrate to inoculum ratio in the batch experiments and the origin of the inoculum influenced input state variables. However, based on results modelled for a CSTR fed with the substrate concerned, these effects were not significant. Indeed, if the optimal ranges of these operational parameters are respected, uncertainty in COD fractionation is mainly limited to temporal variability of the properties of the substrates. As the method is based on kinetics and is easy to implement for a wide range of substrates, it is a very promising way to numerically predict the effect of design parameters on the efficiency of an anaerobic CSTR. This method thus promotes the use of modelling for the design and optimisation of anaerobic processes.
Keywords: Anaerobic digestion, modelling, ADM1, fractionation, calibration, waste activated sludge
CemOA : archive ouverte d'Irstea / Cemagref
1. Introduction
Modelling of anaerobic digestion is widely used for process optimisation, benchmarking and the investigation of biological process at full or laboratory scale. The most commonly used model is the “Anaerobic Digestion Model 1” (ADM1, Batstone et al., 2002).
But before modelling begins, two key tasks need to be completed: (i) fractionation and characterisation of the influent (definition of the composition of the influent according to model input variables) and (ii) calibration (estimation of the most sensitive parameters of the model). Some reference data are available for well-known anaerobic digestion processes such as sewage sludge digestion. Today however, anaerobic digestion processes are more and more often used for a wide range of organic wastes, mainly aimed at energy recovery (Alatriste-Mondragón et al., 2006; Mata-Alvarez et al., 2011). Hence, before numerical process optimisation, a crucial step is characterising the substrate according to ADM1 hypotheses.
Since ADM1 was published, many methods have been developed to this end:
(i) Physical-chemical analysis (Lübken et al., 2007; Wichern et al., 2008). In this method, biochemical fractionation (protein, lipid and carbohydrate analysis), VFA analysis, and fibre extraction are used to split the total COD into each ADM1 input state variable. This kind of method uses quite simple concepts but the conversion of analytic fractions into COD units is tricky. Moreover, additional tests or fractionation calibration of reactor data are needed to estimate inert fractions and hydrolysis kinetics.
(ii) Elemental analysis (Kleerebezem et al., 2006; Zaher et al., 2009a). This is an improvement of the previous method. Here, the elemental composition of the substrates (C, H, O, N and P) is used to improve the biochemical fractionation of the substrate according to ADM1 hypotheses. Nevertheless, the above comments concerning biodegradability and kinetics also apply in this case.
(iii) Physical-chemical analysis combined with online calibration on the simulated reactor (Batstone et al., 2009). This method can be used for most substrates and reactor configurations and the results are suitable for full-scale applications. However, the combined determination of biodegradability and degradation kinetics can be difficult for most full-scale reactors due to identifiability issues, which may reduce the validity domain of the results of the simulation (Batstone et al., 2009). In addition, this kind of method
CemOA : archive ouverte d'Irstea / Cemagref
requires analytical monitoring of the simulated reactor. Hence, this method cannot be used for prediction as the reactor concerned does not actually exist.
(iv) The conversion of another model output into ADM1 input state variables (Copp et al., 2003; Vanrolleghem et al., 2005; Zaher et al., 2007; Nopens et al., 2009). This kind of approach is almost always reserved for simulation of sewage sludge digestion in a plant-wide modelling approach. In this case, the characterisation data come from the output of ASM-type models. Model interfaces have been developed to convert ASM-type outputs into ADM1 inputs. The same approach was developed for piggery wastewater by Rousseau et al. (2008). However, depending on the circumstances, a discrepancy between aerobic and anaerobic biodegradability may lead to model inconsistencies (Buendia et al., 2008).
(v) “Anaerobic respirometry”. The principle of “anaerobic respirometry” is the identification of COD fractions and the kinetic parameters associated with their degradation based on interpretation of the methane production rate (MPR) curves obtained during the anaerobic degradation of the substrate in batch experiments. This method has already been applied by Yasui et al. (2008) and Zaher et al. (2009b) to investigate, using reduced-order anaerobic digestion models other than ADM1, batch degradation of specific substrates (primary sludge and dairy manure). The procedure was developed as an analogy to aerobic respirometry used for the determination of influent COD fractions for activated sludge system models (Ekama et al., 1986). It enables simultaneous determination of ADM1 input state variables (COD fractionation) and related hydrolysis kinetics.
Regarding its ability to provide ADM1 compatible substrate fractionation and related hydrolysis kinetics, “anaerobic respirometry” appears to be in accordance with numerical predictive studies for reactor design and optimisation for many anaerobic digestion processes and applicable to a wide range of substrates. However, very few studies have been published on this topic to date (Yasui et al. (2008), Zaher et al. (2009b) and methodological advances are required to investigate the influence of the operational parameters (especially the substrate to inoculum ratio and the origin of the inoculum in batch experiments) on the fractionation results and to allow identification of a complete set of ADM1-compatible input state variables. In ADM1, the influent COD is split into 13 input state variables including 11 biodegradable COD fractions: composite substrate (Xc), polymers: carbohydrates (Xch), proteins (Xpr) and lipids (Xli) monomers: sugars (Ssu) amino acids (Saa) and long chain fatty
CemOA : archive ouverte d'Irstea / Cemagref
acids (Sfa) and VFAs: butyrate (Sbu), valerate (Sva), propionate (Spro) and acetate (Sac) Inert COD of the influent is split into a soluble fraction (Si) and a particular fraction (Xi).
This paper describes the complete procedure based on “anaerobic respirometry” to provide ADM1 input state variables and related hydrolysis kinetics. The resulting parameters could be used to predict, with ADM1, the methane production linked to a given substrate in an anaerobic CSTR configuration. In this study, the influence of crucial operating conditions for
“anaerobic respirometry” (i.e. the substrate to inoculum ratio for batch experiments and origin of the inoculum) on the estimated ADM1 input state variables and the results of the simulations were investigated. Finally, the temporal variability of the results obtained using waste activated sludge sampled in the same WWTP was compared to the uncertainty related to operational parameters.
2. Material and methods
2.1. SubstratesTwo substrates were used. The first was pig slurry (PS) produced by a livestock fattening unit, which was sampled in the storage tank at a commercial farm in Brittany (France) at the end of the livestock cycle. The second substrate was waste activated sludge (WAS) sampled in a French activated sludge wastewater treatment plant (Mordelles, France) with a capacity of about 10,000 p.e. After settling in a secondary clarifier, the WAS was thickened with a thickening table to reach a dry content of about 5-6%. For anaerobic digestion batch tests and analysis, samples were stored at 4 °C for less than one week before experiments.
2.2. Chemical Analysis
Classical parameters were measured using standard methods (APHA, AWWA, WEF, 1998): total solids (TS), volatile solids (VS), total Kjeldahl nitrogen (TKN), total ammonia nitrogen (TAN) and total chemical oxygen demand (COD).
Total lipid content was determined by Soxhlet extraction on substrate dry matter. Each substrate was first dried at 105 °C and ground to powder (<1 mm). Soxhlet extraction was
CemOA : archive ouverte d'Irstea / Cemagref
carried out for 5 h using hexan/isopropanol (60/40) solvent. After evaporation of the solvent, the percentage of hexane extractable materials (HEM) was determined by gravimetry.
For the biochemical fractionation of the total COD of each substrate, the following
Equation 4: Equation to assess a biochemical fractionation of total COD.
TKN in gN.kg-1 ; TAN in gN.kg-1 ; COD in gO2.kg-1 ; HEM in g.kg-1
*: Dintziz et al., 1988
In addition, volatile fatty acids (VFAs) were analysed on a high performance liquid chromatographer (HPLC, Varian, U3000) equipped with an evaporative light scattering detector. For VFAs, raw samples were first centrifuged and only the supernatant was used for analysis.
2.3. Batch experiments for “anaerobic respirometry”
2.3.1. Inoculums
Depending on the tests, two different inoculums were used for the batch experiments:
a waste activated sludge acclimated inoculum (WAS-INO) and a pig slurry acclimated temperature was maintained at 36 °C.
PS-INO was sampled from a CSTR fed with a mixture of pig slurry (44% of the total COD) and horse feed (56% of the total COD) mainly consisting of a mixture of wheat gluten feed, oats, straw and sunflower cake. HRT was 27 days and OLR was 3.9 kgCOD.m-3.d-1 (feed COD concentration of 106.3gO2.kg-1). Based on biodegradability, the OLR was 2.0
CemOA : archive ouverte d'Irstea / Cemagref
kgCODbiodegradable.m-3.d-1 (pH=7.8 ; [N-NH4+
]=3.2gN.L-1). The temperature of this reactor was 38 °C and its working volume was 87L. Both reactors were kept at steady state conditions throughout the study and were operated for at least for three times their HRT before the first samples were taken.
2.3.2. Batch reactors
For the batch experiments, eight similar reactors with a 1.2 L working volume were used. The contents were continuously mixed by a magnetic stirrer (1200 rpm) and maintained at 38 °C using a specific chamber (Aqualytic, ET637-6, Germany). Each reactor was equipped with a manometer (Vegabar 14, Vega, Germany) and a solenoid valve to allow continuous monitoring of biogas production. Gas production was calculated taking the temperature, the headspace volume of the reactor and pressure into account. Over-pressure was automatically released by opening the solenoid valve when an overpressure of 50 mbar was measured. The released gas was collected in a 1 L Tedlar® Bag (232-01, SKC, USA) which enabled regular determination of CH4 and CO2 concentrations in the biogas. CH4 and CO2 concentrations were determined using a gas chromatographer equipped with an electron capture detector (Agilent Technologies, 6890N, USA) according to the method described in Lucas et al. (2007).
2.3.3. Experimental procedure
Batch reactors were filled with 1 L of inoculum and incubated to allow the stabilization of biogas and methane production. After one day of incubation (corresponding to time = 0 day in the figures), substrate was added and the methane production rate (MPR) was monitored for 10 days. The mass of substrate added was calculated on the basis of a defined biodegradable substrate to inoculum ratio (gCODbiodegradable.gCODbiomass-1
). The concentration of biomass was calculated as the sum of the biomass in the initial state of experiments for ADM1 (see section 2.4). However, in this study, the total COD of the inoculum cannot be considered as a representative indicator of biomass concentration because it was mainly comprised of inert materials. The concentration of each substrate in biodegradable COD was
CemOA : archive ouverte d'Irstea / Cemagref
first estimated according to the literature. Then, the real substrate to inoculum ratio was calculated based on the sum of the biodegradable fractions of the substrate (see section 2.5.)
Immediately after setting up the inoculum and the substrate pulse, the headspace of each reactor was purged with a gas mixture of N2 and CO2 (70/30). All tests were compared with a control without the substrate pulse.
As the viscosity of the WAS-INO was relatively high which caused problems with stirring, this inoculum was diluted (1/1) with its centrifuge supernatant. Dilution of PS-INO was not necessary. The dilution factor of the WAS-INO was taken into account in the calculation of the amount of substrate added.
Repeatability had been previously tested in parallel in the eight batch reactors with an acetate pulse (results not shown). The relative standard deviation for MPR was below 5%
except in the first and the last hours of the pulse, where a slight time lag was observed.
2.4. Modelling of batch experiments
The ADM1model (Batstone et al., 2002) was used in the simulations to represent the major biochemical and physicochemical processes that occurred during anaerobic digestion.
The ADM1 was implemented in Scilab® and solved with the ordinary differential equation solver “ode” (package ODEPACK, solver Isoda). The stoichiometric parameters from Batstone et al. (2002) were used along with kinetic parameter sets from Girault et al.
(submitted) except for hydrolysis. Parameter values are listed in Table 15.
Parameter Units WAS-INO PS-INO
INH3×km_ac* kgCOD.kgCOD-1.d-1 2.73 2.51
Ks_ac kgCOD.m-3 0.088 0.30
km_pro kgCOD.kgCOD-1.d-1 13 7.8
Ks_pro kgCOD.m-3 0.3 0.077
Table 15: Kinetic parameters for biomass growth from Girault et al. (submitted) used to simulate batch experiments.
Other parameters used were default parameter from Batstone et al. (2002). *the inhibition factor by free ammonia was considered to be constant in all simulations with each inoculum.
Due to the low substrate to inoculum ratios, the substrate added had no significant impact on the concentration of free ammonia.
CemOA : archive ouverte d'Irstea / Cemagref
Initial biomass concentrations for batch experiments were taken from the results of the steady state ADM1 simulation of each CSTR from which the inoculums were sampled. The inert fraction (Xi) was estimated on each CSTR result to accurately simulate degradation of COD. Biodegradable COD was split into proteins (Xpr), carbohydrates (Xch) and lipids (Xli) according to the biochemical fractionation of the biodegraded COD into each CSTR (application of Equation 4 on the influent and the effluent of the CSTR). All other input state variables for the influent were considered as equal to zero. Next, the CSTR was simulated for four times the HRT to provide steady state data. Simulated concentrations of each specific biomass were used as initial conditions for the inoculum used in the batch experiments.
Before modelling the batch experiments with added substrate, an accurate simulation
Before modelling the batch experiments with added substrate, an accurate simulation