Bioresource Technology 100 (2009) 2783–2790
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Modified version of ADM1 model for agro-waste application A. Galí, T. Benabdallah, S. Astals, J. Mata-Alvarez * Department of Chemical Engineering, University of Barcelona, Martí i Franquès, No. 1, 6th Floor, 08028 Barcelona, Spain
a r t i c l e
i n f o
Article history: Received 23 October 2008 Received in revised form 23 December 2008 Accepted 24 December 2008 Available online 7 February 2009 Keywords: Anaerobic digestion Agro-wastes Biodegradability Modelling Particulate fraction
a b s t r a c t Agro-residues account for a large proportion of the wastes generated around the world. There is thus a need for a model to simulate the anaerobic digestion processes used in their treatment. We have developed model based on ADM1, to be applied to agro-wastes. We examined and tested the biodegradability of apple, pear, orange, rape, sunflower, pig manure and glycerol wastes to be used as the basis for feeding the model. Moreover, the fractions of particulate COD (Xc) were calculated, and the disintegration constant was obtained from biodegradability profiles, considering disintegration to be the limiting process. The other kinetic and stoichiometric parameters were taken from the ADM1 model. The model operating under mono-substrate and co-substrate conditions was then validated with batch tests. At the same time the model was validated on a continuous anaerobic reactor operating with pig manure at lab scale. In both cases the correlation between the model and the experimental results was satisfactory. We conclude that the anaerobic digestion model is a reliable tool for the design and operation of plants in which agro-wastes are treated. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Nowadays alternative sources of energy like bio-fuels and biogas are becoming necessary. In this context, anaerobic digestion (AD) to produce biogas is a clean technology that can significantly contribute to the supply of energy, especially at the farm level (Mata-Álvarez et al., 2000). Focussing on biogas production, there are different types of waste that have a strong biogas potential, like the residues from agro-activities, sludge from wastewater treatment plants or municipal solid wastes (MSWs). Agro-wastes comprise an extensive family of residues that can be classified into two types: fruit or plant residues and animal manures. Energy crops and their sub-products (including glycerine) are very common in the world of bio-fuels due to their high energy capacity, but plant wastes and manures can also produce a significant amount of biogas if they are correctly treated. However, AD of some substrates can present limitations and low efficiency. In these cases co-digestion could be an interesting option to improve the yields of AD. Inhibition is another type of limitations that can occur during anaerobic digestion, which can also be avoided using co-digestion. The development and use of models may save time and money. The power of models lies in their capacity to reproduce empirical behaviour in a computer. Accordingly, in the last few years various models have been developed to simulate AD processes.
* Corresponding author. E-mail address:
[email protected] (J. Mata-Alvarez). 0960-8524/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2008.12.052
In 1987, ASM1 (Activated Sludge Model 1) was published by Henze et al. (1987). In subsequent years ASM1 and its derivations (ASM2 and ASM3,) became a standard in wastewater treatment (Henze et al., 1999; Henze et al., 2000). Based on the same theory, in 2002, Anaerobic Digestion Model 1 (ADM1) was published (Bastone et al., 2002). ADM1 unified the concepts in anaerobic digestion of wastes, focussing basically on sewage sludge (Fig. 1). Other residues like agro-wastes and MSW were not studied in the same way as sewage sludge. Although some work has been done (Angelidaki et al., 1998; Christ et al., 2000; Kalfas et al., 2006; Lübken et al., 2007; Derbal et al., 2009) further research is required to improve the model, and to render it equally applicable to sewage sludge. In ADM1, disintegration and hydrolysis convert the particulate compounds into soluble compounds. Suitable substrate characterisation will yield appropriate values for the following fractions: carbohydrates (fchxc1), proteins (fprxc1), lipids (flixc1) that are to be hydrolysed and the inert fraction (fsixc1 for soluble and fxixc1 for particulate). Hydrolysis is assumed to be the first limiting step of the anaerobic digestion process when a high solid content is present and so the kinetic parameters must be calculated accurately (Miron et al., 2000). Normally hydrolysis (Bastone et al., 2002) is assumed to follow first-order kinetics (r = khXi, where kh is the hydrolysis constant and Xi is the particulate substrate). The soluble compounds generated during hydrolysis (sugars, aminoacids and long-chain fatty acids) are then transformed in one step (acidogenesis) to valeric, butyric, and propionic acids; in the second step (acetogenesis) to acetic acid and hydrogen; and
2784
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
Fig. 1. General scheme reaction for ADMI (Bastone et al., 2002).
finally (methanogenesis) into CH4 and CO2 following Monod kinetics (r = kSi/(Si + Ki)Xi, where k is the maximum growth rate and Ki is the half-saturation constant). Given the importance of agro-wastes, a model that could predict biogas production and biodegradability for each type of waste would be of interest. Moreover, it would be useful to be able to predict the inhibition caused in the process and the possibility of co-digesting more than one substrate in order to compensate for the occasional lack of nutrients or enlarge the organic loading rate (Gavala et al., 1996). The aim of the present paper is to characterise residues from the family of agro-wastes for use as a basis for the development of a modified version of ADM1. The new ADM1 will then be checked and validated by means of mono-substrate and co-substrate batch and continuous digestion at lab scale. 2. Model structure and development The ADM1 modified by CEIT (de Gracia et al., 2006), using differential equations instead of algebraic equations, has been taken as a reference. It has been modified and adapted to a series of agro-residues under study in the present paper. The final modified ADM1 is composed by 32 processes (19 biological processes, 10 equilibrium processes and 4 gas transfer processes). It is also composed by 41 components divided into 24 soluble (Si), 13 particulate (Xi) and 4 gas (Gi) compounds. Table 1 presents the Petersen matrix where all the processes and compounds are detailed whereas, in Table 2 the 41 states variables are briefly described. From Tables 1 and 2 it can be seen that the model is structured with a differential equation for each acid/base pair of soluble compounds. In this way, there are 10 equilibrium or compensation terms considering the elemental compounds charge, N, C, P, and S. The implementation of H2S in soluble and gas states, involved in processes 28 and 32, respectively, is a new modification with respect to ADM1 and the model proposed by CEIT for sewage sludge (de Gracia et al., 2006). In the bottom of the Petersen matrix, there is the composition matrix, where, the quantity of carbon, nitrogen, phosphorus and the charge component for each compound are expressed. This is
an important tool in order to apply the continuity equation for each of the 32 processes for C, N, P and charge. In this way, it is posþ sible to obtain the stoichiometry for the components Sco2, Snh4 , 2 + Shpo4 and Sh . These terms correspond to the concentrations of carbon dioxide, ammonia, hydrogen phosphate and protons, respectively and they are the result of the mass balance for C, N and P and charge (Huete et al., 2006). When proceeding like this, it is possible to control and follow in a continuous way the alkalinity, the phosphates concentration, the total Kjelhdal nitrogen (TKN), and finally the pH with the proton concentration. Most of stoichiometric and kinetic parameter values used in the present model comes from the default values of ADM1 model. Tables 3 and 4 show the parameters used in the present model. The particulate material (Xc) fractions (fsixc1, fxixc1, fchxc1, fprxc1, flixc1) and the disintegration constants (Kdis) are the parameters that have been characterised for each of the studied substrates and their numeric values are not included in these latter tables. The values from Table 4 are for mesophilic conditions but the model could be operated also at thermophilic conditions by applying the values proposed in ADM1 model (Bastone et al., 2002). Moreover, the inhibitions considered by the model (bottom Table 1) are the ones proposed by the ADM1 including the inhibition of acetoclassic methanogens by hydrogen sulphide. The general expressions for inhibition terms can be found in Petersen matrix (Table 1) but the individual inhibitions are shown in Table 5. The current pH inhibition used in the present model is the same considered by the ADM1 model when the lower pH limit is achieved (Bastone et al., 2002).
3. Methods 3.1. Model implementation The model is developed in MATLAB/SIMULINK (Rosen et al., 2006) with the code written in C language. The data from the substrate characterisation, stoichiometry and kinetics is written in Excel and then it is exported to Matlab. Matlab, using different vectors with the initial conditions, influent concentration, stoichiometry, kinetic data and other relevant information, take the information from the Excel files. Simulink acts as flow sheet diagram software where the different units (reactors) are connected with the influent flow-rates. In each unit (S function) it is necessary to specify (a) which vectors from Matlab interact and (b) which is the file containing the code for the mathematical resolution of the model. The different differential equations are set in the code written in C. They are structured in order to be solved for each compound or variable. At the same time, the variables of the model and the important engineering aspects (COD, biogas production etc.) are defined and connected with Matlab through Simulink. In order to perform the connection, the C files must be compiled and converted to MEX files and then can be used by Matlab/Simulink. 3.2. Analytical methods Analysis of total chemical oxygen demand (COD), total solids (TSs) and volatile solids (VSs), pH, bicarbonate alkalinity, total ammonia nitrogen (TAN), total Kjeldahl nitrogen (TKN) and phosphorus were performed according to the standard methods (APHA, 1998). Individual volatile fatty acids and gas composition were analysed by gas chromatography (HP 5800) equipped with flame ionization detector (FID) and thermal conductivity detector (TCD), respectively (APHA, 1998). Proteins amount was estimated by
Table 1 Peterson matrix.
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
2785
2786
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
Table 2 States variables considered in the model.
Table 3 Stoichiometric parameters (Bastone et al., 2002). Fractions (CODproduct COD1sustrate)
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
*
3
Sh (kmole m ): Proton Soh (kmole m3): Hydroxyl 2 Shpo4 (kmole m3): Hydrogen phosphate 3 Sh2po 4 (kmole m ): Dihydrogen phosphate þ Snh4 (kmole m3): Ammonium Snh3 (kmole m3): Free ammonia Sco2 (kmole m3): Carbon dioxide Shco3 (kmole m3): Bicarbonate Sh2s (kmole m3): Sulphide acid Shs (kmole m3): Hydrogen sulphide Ssu (kg COD m3): Soluble sugar Saa (kg COD m3): Soluble aminoacids Sfa (kg COD m3): Soluble LCFA Shva (kg COD m3): Valeric acid Sva (kg COD m3): Valerate Shbu (kg COD m3): Butyric acid Sbu (kg COD m3): Butyrate Shpro (kg COD m3): Propionic acid Spro (kg COD m3): Propionate Shac (kg COD m3): Acetic acid Sac (kg COD m3): Acetate Sh2 (kg COD m3): Hydrogen acid Sch4 (kg COD m3): Methane Si (kg COD m3): Soluble inert Xc1 (kg COD m3): Composite Xc2 (kg COD m3): Death biomass composite Xch (kg COD m3): Carbohydrate Xpr (kg COD m3): Protein Xli (kg COD m3): Lipid Xsu (kg COD m3): Sugar biomass Xaa (kg COD m3): Aminoacid biomass Xfa (kg COD m3): LCFA d biomass Xc4 (kg COD m3): Valeric/butyric biomass Xpro (kg COD m3): Propionic biomass Xac (kg COD m3): Acetic acid biomass Xh2 (kg COD m3): Hydrogen biomass Xi (kg COD m3): Particulate inert Gco2 (kg mole m3): Carbone dioxyde gas Gh2 (kg COD m3): Hydrogen gas Gch4 (kg COD m3): Methane gas Gh2s (kg COD m3): Sulphide gas
multiplying organic nitrogen (TKN minus NHþ 4 -N) by 6.25 and lipids were quantified using Soxhlet extraction procedure (APHA, 1998). Carbohydrates were estimated by subtracting the amount of proteins and lipids from volatile solids. The main fractions of particulate carbohydrates, namely: cellulose, hemicellulose and lignin, were determined according to Goering and Van Soest (1970). 3.3. Biodegradability test Anaerobic biodegradability tests were carried out under mesophilic temperature conditions (35 °C) to quantify the potential of methane production (ml CH4 g1VSadded) for each substrate. Wastes under study were previously homogenised and sieved to achieve a particle size lower than 2 mm before the addition to the lab-reactor. These reactors with a total volume of 250 ml, were filled with substrate and inoculum at 0.5 of VSsubstrate/VSinoculum ratio, and with an excess of alkalinity. Equal amounts of substrate were added to all digesters and the effective reactor volume was set to 230 ml by deionised water. The methane production during a running test was measured using displacement liquid device equipped with a biogas wash vessel (1 N NaOH solution to remove CO2). The test was performed in quadruplicate for each substrate. The inoculum previously filtered through 1 mm was obtained from an industrial mesophilic pig manure digester situated in Juneda (Lleida, Spain) operating with a hydraulic retention time (HRT) of 18 days and 260 ml CH4 g1TS as an average value of specific methane production (SMP).
Soluble inert from initial residue Particulate inert from initial residue Carbohydrates from initial residue Proteins inert from initial residue Lipids inert from initial residue Soluble inert from death biomass Particulate inert from death biomass Carbohydrates from death biomass Proteins inert from death biomass Lipids inert from death biomass LCFA from lipids H2 from sugars Butyric from sugars Propionic from sugars Acetic from sugars H2 from aminoacids Valeric from aminoacids Butyric from aminoacids Propionic from aminoacids Acetic from aminoacids
fsixc1 fxixc1 fchxc1 fprxc1 flixc1 fsixc2 fxixc2 fchxc2 fprxc2 flixc2 ffali fh2su fbusu fprsu facsu fh2aa fvaaa fbuaa fpraa facaa
variable variable variable variable variable 0.1 0.1 0.1 0.67 0.033 0.95 0.19 0.13 0.27 0.41 0.06 0.23 0.26 0.05 0.40
Yields (COD COD1) Yield in sugar degradation Yield in aminoacid degradation Yield in LCFA degradation Yield in valeric and propionic degradation Yield in propionic degradation Yield in acetic degradation Yield in h2 degradation
Ysu Yaa Yfa Yc4 Ypro Yac Yh2
0.10 0.08 0.06 0.06 0.04 0.05 0.06
3.4. Laboratory scale semi-continuous reactor (S-CSTR) Two completely stirred and jacketed anaerobic digesters with a 4 l working volume were used in this study. They were maintained at 35 °C by external heating and the HRT was fixed at 20 days. Both digesters were equipped with pH probe and were connected to a cooled feed tank (4 °C) and to a biogas collection-measuring system. Feed addition and digestate removal was carried out three times a day with two peristaltic pumps.
4. Experimental results 4.1. Substrate characterisation The substrates under study and their characteristics are detailed in Table 6 (soluble fractions for each sample were also analysed but is not shown in Table 6). It should be noted that, due to the longevity of some tests and the consequent large quantities of waste used during the semi-continuous digester, substrates and inoculum presented some variability in their characterisation. 4.2. Fractions of particulate COD (Xc) Particulate COD from each substrate (Xc) is composed by carbohydrates (fchxc1), proteins (fprxc1), lipids (fprxc1) and inerts (fsixc1, fxixc1). With the extended characterisation of each substrate presented in Table 6 it is possible to calculate the percentages in which Xc is structured. The latter is very important because the mechanism of degradation of carbohydrates, proteins and lipids is not the same. It is important to state that in this partition, the soluble biodegradable COD is not included because it is already specified in the characterisation. Table 7 presents these fractions of the non-soluble COD and the particulate partition of each waste.
2787
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790 Table 4 Kinetic parameters at mesophilic conditions (Bastone et al., 2002).
Table 5 Individual inhibition terms (Bastone et al., 2002).
Constant
Units
Value
Hydrolisis rates Kdis1 Kdis2 Kh_xch Kh_xpr Kh_xli
d1 d1 d1 d1 d1
Variable 0.15 10 10 10
Degradation rates Km_su Km_aa Km_fa Km_c4 Km_pr Km_ac Km_h2
COD COD1 d1 COD COD1 d1 COD COD1 d1 COD COD1 d1 COD COD1 d1 COD COD1 d1 COD COD1 d1
30 50 6 20 13 8 35
Half saturation constants Ks_su Ks_aa Ks_fa Ks_c4 Ks_pr Ks_ac Ks_h2
kg m3 kg m3 kg m3 kg m3 kg m3 kg m3 kg m3
0.5 0.3 0.4 0.2 0.1 0.15 7.00E-06
Decay rates Kdec_xsu Kdec_xaa Kdec_xfa Kdec_xc4 Kdec_xpr d-1 Kdec_xh2
d1 d1 d1 d1 d1 d1 d1
0.02 0.02 0.02 0.02 0.02 0.02 0.02
Inhibition constants KI_h2s KI_h2pro KI_h2fa KI_h2c4 KI_in KI_nh3
kg m3 kg m3 kg m3 kg m3 kg mole m3 kg mole m3
0.64 3.50E-06 5.00E-06 1.00E-05 1.00E-04 0.0018
pH inhibition limits pHLLaa pHLLaa pHLLxh2 pHLLxh2 pHLLxac pHLLxac
– – – – – –
4 5.5 5 6 6 7
Equilibrium acid/base parameters KaacidIC k mole H m3 KaacidIN k mole H m3 KaacidAC k mole H m3 KaacidPRO k mole H m3 KaacidBU k mole H m3 KaacidVA k mole H m3 KaacidIP k mole H m3 KaacidH2S k mole H m3
4.94E-07 1.09E-09 1.74E-05 1.32E-05 1.50E-05 1.38E-05 6.60E-08 8.90E-08
Mass rate coefficients Kla_h2 Kla_ch4 Kla_co2 Kla_h2s
195 184 163 150
Kh_h2 Kh_ch4 Kh_co2 Kh_h2s Acid/base dissociation rate Kab
d1 d1 d1 d1 kg m3 k mole COD m3 bar1 k mole COD m3 bar1 k mole C m3 bar1 k mole S m3 bar1 3
(k mole H m
1
d)
0.000738 0.00116 0.0271 0.0017 1215752192
Term
Expression
Comments
In_iNlim
Lack of inorganic nitrogen
In_h2fa In_h2c4
1/(1+KSin/ SNH4 + SNH3) 1/(1+SH2/KI_h2fa) 1/(1+SH2/KI_h2c4)
In_h2pro In_nh3xac In_h2sxac
1/(1+SH2/KI_h2pro) 1/(1+SNH3/KI_nh3) 1/(1+SH2S/KI_h2s)
Inhibition of Inhibition of degradation Inhibition of Inhibition of Inhibition of
H2 in fatties degradation H2 in butyric/valeric H2 in propionic degradation NH3 in acetate degradation H2S in acetate degradation
no accumulation of intermediary products. Assuming first-order kinetics for the hydrolysis of particulate organic matter, the cumulative methane production can be described by means of Eq. (1) (Hill, 1982; Veeken and Hamelers, 1999)
SMPðtÞ ¼ SMPo ð1 eðkdis
tÞ
Þ
SMP (t) represents the cumulative specific methane production at time t at standard temperature of 0 °C and standard pressure of 1 atm (STP conditions), SMPo the maximum methane yield of the substrate (in STP: L kg VS1), kdis the first-order total hydrolysis (disintegration) rate constant (in day1) and t time (in days). Parameter values for SMPo and kdis were estimated using non-linear least squares curve fitting of the net cumulative methane production. SMPo is directly linked with biodegradability and can be used as a first approach. For each of the analysed substrates the disintegration constant is estimated by fitting Eq. (1) with the so-called least square method. After doing that, a value of the disintegration constant is found for each substrate (Table 8). 5. Model structure and performance Once the characterisation of substrates is finished (Table 6), the fractions of Xc are calculated (Table 7) and the disintegration constants are obtained (Table 8), the model can be built up. From Table 6 and Table 7 it is possible to establish the initial concentration for each of the 41 model variables by knowing the quantity of substrate added, the reactor volume and the hydraulic retention time to be applied. Then, the model will work in a steady state or dynamic way by knowing the latter information daily. The model is thought to be operated for one or two indistinct stirred reactors in series filled separately in a continuous or semicontinuous way. The number of substrates included can be chosen indistinctly for each of the reactors. Every reactor has its own operating screen that can be modified separately. As explained above the required general data to be introduced in the model to start, is the hydraulic retention time (HRT) of the selected reactor, the operating temperature (T), the volume of the liquid portion of the reactor (Vliq) and the number of times that the reactor is filled per day. Moreover, the quantities of the selected daily substrate in kg and the operation temperature conditions (mesophilic or thermophilic) have also to be specified. 6. Model validation The validation of the model has been carried out in different steps. Thus, several batch tests have been simulated using monosubstrate digestion and a single one with co-substrate digestion. Finally, the model has been tested with lab-scale experiments.
4.3. Disintegration constants
6.1. Substrates batch test
Methane production is mainly limited by the total hydrolysis rate (disintegration) of particulate organic matter when there is
The first step consists of validating the model in the simplest situation like mono-substrate digestion. At this point five
2788
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
Table 6 Characteristics of pig manure and pulp fruits. Parameter
Units
Apple pulp
Orange pulp
Pear pulp
Pig manure
Rape
Sunflower
Glycerol
Inoculum
Total solids Volatile solids COD VFA TAN TKN Alkalinity pH
1
g TS kg g VS kg1 g O2 kg1 TS mM HAc kg1 1 g NHþ 4 -N kg g N kg1 TS mg CaCO3 kg1 –
137.7 130.2 1700 18.1 n.d 4.8 0 3.9
144.5 139.9 1692 41.0 n.d 13.7 0 2.45
118.2 113.6 2074 117.2 n.d 6.2 0 4.1
33.1 21.7 882 0.4 1.0 29.5 11800 8.0
152 135 1558 n.d n.d 22 0 4.44
116 101 1529 n.d n.d 32.7 0 5.0
850 850 1500 n.d n.d n.d n.d 5
11.6 7.2 1357 0.5 470.7 513.0 8.235 8.0
Proteins Lipids Carbohydrates Cellulose Hemicellulose Lignin
g kg1 TS g kg1 TS g kg1 TS g kg1 TS g kg1 TS g kg1 TS
30.2 33.1 882.0 198.3 210.4 131.6
85.4 6.35 876.6 78.0 260.3 25.1
38.4 28.4 894.3 147.4 190.4 80.3
191.6 39.0 424.0 90.5 65.3 115.2
137.4 48.5 697 232 41.3 71.8
204.5 52.7 607 407 85.5 115.4
n.d n.d 1272 n.d n.d n.d
86.7 77.3 874.4 284 80.7 503.3
Elemental C Elemental N Elemental P
% % %
39.0 6.7 0.2
42.8 1.2 0.2
41.9 0.6 0.5
32.6 2.7 2.2
40.2 1.84 3.56
39.3 2.1 5.2
n.d n.d n.d
29.6 3.35 0.97
Table 7 COD Particulate substrate fraction.
6.2. Lab-scale experiments
Substrate
fchxc1
fprxc1
flixc1
fxixc1
fsixc1
Pig manure Rape Sunflower Orange Pear Apple Glycerol
0.461 0.556 0.506 0.477 0.399 0.256 1
0.202 0.126 0.198 0.020 0.016 0.011 0
0.161 0.122 0.034 0.0140 0.084 0.055 0
0.033 0.166 0.078 0.337 0.1340 0.255 0
0.143 0.122 0.184 0.153 0.367 0.422 0
Table 8 Disintegration kinetic constant. Substrate
SMPo
Kdis
Units Pig manure Rape Sunflower Glycerol Orange Pear Apple
(L CH4 kg VS1) 0.150 0.250 0.200 0.300 0.250 0.150 0.180
(days1) 0.17 ± 0.01 0.24 ± 0.01 0.23 ± 0.01 0.5 ± 0.01 0.29 ± 0.01 0.18 ± 0.01 0.15 ± 0.01
biodegradability tests have been performed following the methodology explained in Section 3 (Methods). After 30 days of operation the biodegradability profiles are obtained (Fig. 2). The continuous line in this figure corresponds to the simulation results obtained with the same operating conditions than in the experiments. As can be seen in Fig. 2, the correlation between experimental and simulation data using orange, apple and pig manure wastes is very good. In the case of rape the simulation data predicted more methane production than the experimental data. This could be due to some kind of inhibition occurring inside the batch reactor that the model cannot totally predict, like slowly degradation of lignin. Leaving this fact aside, the correlation between model and experimental values is satisfactory considering the slopes of the profiles in all the cases and the prediction of the final biogas production in the case of orange, apple and pig manure wastes. Moreover, a batch test using pig manure (60%, total weight) and glycerine (40%, total weight) has been done in order to provide nutrients to the glycerine. When doing this simulation the model also shows a very good correlation with the experimental results.
Once shown that the model predicts suitable results for batch tests, some simulations have been run to reproduce the performance of a semi-continuous laboratory reactor working with pig manure (see full characterisation in Table 6) under steady state conditions. This semi-continuous reactor has been operated under the working conditions presented in Table 9. After four months of operation working under steady state conditions, the results presented in Table 10 have been obtained. Then, the model has been loaded with initial data from Table 9 and run within a simulation period of 70 days. The outputs from the experimental reactor and the simulation process are also shown in Table 10. From Table 10 it can be seen that the final pH (steady state) fits perfectly around 8.5 and the methane composition of biogas was 80% as it was obtained in the experimental results. Moreover, the biogas production obtained in the model or (0.38 Nm3/kg VS) presents a good correlation respect to the one found in the experimental results. The difference could be explained by some kind of inhibitions and substrate degradation (i.e. lignin) that the model cannot predict, resulting in better conversions. Other parameters like ammonia, soluble COD, alkalinity and volatile solids were also analysed and fitted correctly with the model (Table 10). 7. Conclusions
A modified version of the ADM1 based on the model developed by CEIT (de Gracia et al., 2006) has been built in order to be used for the simulation of agro-wastes. After the characterisation of the substrates, the partition of the particulate COD (Xc) and the calculation of the disintegration constants have been done. The validation of the model has been done for mono-substrate and co-substrate cases in batch and continuous reactors. In all the cases there were good correlation between experimental and simulation results which lead to the conclusion that the model predicts correctly the degradation of agro-wastes. As a general conclusion it could be stated that the modification of the model in order to be used for agro-wastes is satisfactory and constitute an excellent tool to design and monitory digester operation.
2789
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
Orange
180
Apple
120 100
140 120
CH4 (NmL)
CH4 (NmL)
160
100 80 60 40
80 60 40 20
20 0
0 0
10
20
0
30
10
Pig Manure
180
30
20
30
Rape
250
150
200
CH4 (NmL)
CH4 (NmL)
20
Time (days)
Time (days)
120 90 60
150 100 50
30 0
0 0
10
20
30
0
10
Time (days)
Time (days) PM (60%) + Glyverine (40%)
CH4 (NmL)
400 300 200 100 0
0
10
20
30
40
Time (days) Fig. 2. Mono-substrate and co-substrate biodegradability profiles (h experimental; – simulation).
References
Table 9 Lab-anaerobic reactor operational conditions. Fermentation volume
4
L
Feed (FM mix)a HRT VS feed Organic load Fermentation temperature
200 20 4.3 1.09 35
g/d d g/d g/(L d) °C
a
Filled 3 time per day.
Table 10 Outputs from the reactor: experimental and simulated.
Biogas production (Nm3/kg VS) Methane fraction (%) pH VS final (g/l) NHþ 4 -N (g/l) COD soluble (g/l) Alkalinity (g CaCO3/l)
Experimental
Simulation
Relative error (%)
0.350 77 8.0 11 2.7 18.0 11.4
0.381 77 7.9 12 3.0 16.0 12.0
9 0 1 9 7 8 5
Acknowledgements This research has been supported by European Commission (Project No. 030348-AGROBIOGAS).
Angelidaki, I., Ellegaard, L., Ahring, B.K., 1998. A comprehensive model of anaerobic bioconversion of complex substrates to biogas. Biotechnology and Bioengineering 63 (3), 363–372. APHA, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed. Washington D.C. Bastone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M., Siegrist, H., Vavilin, V.A., 2002. Anaerobic Digestion Model No. 1 (ADM1). IWA Scientific and Technical Report No. 13. IWA Publishing, London. Christ, O., Wilderer, P.A., Angerhöfer, R., Faulstich, M., 2000. Mathematical modelling of the hydrolisi of anaerobic processes. Water Science and Technology 41 (3), 61–65. De Gracia, M., Sancho, L., García-Heras, J.L., Vanrolleghem, P., Ayesa, E., 2006. Mass and charge conservation check in dynamic models: application to the new ADM1 model. Water Science and Technology 53 (1), 225–234. Derbal, K., Bencheikh-lehocine, M., Cecchi, F., Meniai, A.-H., Pavan, P., 2009. Application of the IWA ADM1 model to simulate anaerobic co-digestion of organic waste with waste activated sludge in mesophilic condition. Bioresource Technology 100 (4), 1539–1543. Gavala, H.N., Skiades, I.V., Bozinis, N.A., Lyberatos, G., 1996. Anaerobic co-digestion of agricultural industries wastewaters. Water Science and Technology 34 (11), 67–75. Goering, H.K.,Van Soest, P.J., (1970). Forage Fibre Analyses (Apparatus, Reagents, Procedures and some Applications). Agricultural Handbook No. 379, ARS-USDA, Washington D.C., United States of America. Henze, M., Grady, C.P.L., Gujer, W., Marais, G.V.R., Matsuo, T., 1987. Activated Sludge Model No. 1. Scientific and Technical Report No. 1. IWA Publishing, London. Henze, M., Gujer, W., Mino, T., Matsuo, T., Wentzel, M.C., Marais, G.R., 1999. Activated Sludge Model No. 2d. Scientific and Technical Report No. 3. IWA Publishing, London. Henze, M., Gujer, W., Mino, T., van Loosdrecht, M.C.M., 2000. Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. Scientific and Technical Report No. 9. IWA Publishing, London.
2790
A. Galí et al. / Bioresource Technology 100 (2009) 2783–2790
Hill, D.T., (1982). A Comprehensive Dynamic Model for Animal Waste Methanogenesis. Transaction of the ASAE. Huete, E., de Gracia, M., Ayesa, E., Garcia-Heras, J.L., 2006. ADM1-based methodology for the characterisation of the influent sludge in anaerobic reactors. Water Science and Technology 54 (4), 157–166. 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. Lübken, M., Wichern, M., Schlattmann, M., Gronauer, A., 2007. Modelling the energy balance of an anaerobic digester fed with cattle manure and renewable energy crops. Water Research 41, 4085–4096.
Mata-Álvarez, J., Macé, S., Llabrés, P., 2000. Anaerobic digestión of organic solid waste. An overview of research achievements and perspectives. Bioresource Technology 74, 3–16. Miron, Y., Zeeman, G., van Lier, J.B., Lettinga, G., 2000. The role of sludge retention time in the hydrolysis and acidification of lipids, carbohydrates and proteins Turing digestion of primary sludge in CSTR systems. Water Research 34 (5), 1705–1713. Rosen, C., Vrecko, D., Gernaey, K.V., Pons, M.N., Jeppsson, U., 2006. Implementing ADM1 for plant-wide benchmark simulations in Matlab/Simulink. Water Science and Technology 54 (4), 11–19. Veeken, A., Hamelers, B., 1999. Effect of temperature on hydrolysis rates of selected biowaste components. Bioresource Technology 29, 249–254.