- Email: [email protected]
Application of the IWA ADM1 model to simulate anaerobic co-digestion of organic waste with waste activated sludge in mesophilic condition
Bioresource Technology 100 (2009) 1539–1543
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Application of the IWA ADM1 model to simulate anaerobic co-digestion of organic waste with waste activated sludge in mesophilic condition K. Derbal a, M. Bencheikh-lehocine a,*, F. Cecchi b, A.-H. Meniai a, P. Pavan c a
Laboratoire de l’Ingénierie des Procédés de l’Environnement (LIPE), Département de Chimie Industrielle, Faculté des Sciences de l’Ingénieur, Université Mentouri, Constantine, Algeria Department of Science and Technology, University of Verona, Verona, Italy c Department of Environmental Science, University of Venice, Venice, Italy b
a r t i c l e
i n f o
Article history: Received 29 April 2008 Received in revised form 13 July 2008 Accepted 16 July 2008 Available online 26 October 2008 Keywords: Anaerobic co-digestion Simulation ADM1 Mesophilic temperature Solid wastes
a b s t r a c t Anaerobic digestion model no. 1 model of international water association was applied to a full scale anaerobic co-digestion process for the treatment of the organic fraction of municipal solid wastes along with activated sludge wastes originating from a municipal wastewater treatment plant. This operation was carried out in a digester of 2000 m3 in volume. It is operates at an average hydraulic retention time of 26.9 days with an average organic loading rate of 1.01 kg TVS/m3 day, at a temperature of 37 °C with an average gas production rate of 0.296 m3/m3 day. The aim of the present study is to compare the results obtained from the simulation with the experimental values. The simulated results showed a good fit for pH, methane and carbon dioxide percentages, biogas volume, chemical oxygen demand, total volatile fatty acids, inorganic nitrogen and inorganic carbon. Ó 2008 Published by Elsevier Ltd.
1. Introduction Anaerobic digestion is worldwide used, particularly in Europe for the treatment of numerous types of biodegradable wastes. This is confirmed by the important number of treatment plants using this process on the industrial scale, during the past few years (Mata-Alvares et al., 2000). In fact, anaerobic digestion of the organic fraction of the municipal solid wastes alone or combined with organic sludge can contribute efficiently in solid waste reduction and biogas production (Jewell, 1979; Kayhanian and Tchobanoglus, 1992; Vallini et al., 1992; Cout et al., 1994). This process can be used for the solid waste treatment under mesophilic or thermophilic conditions (Macé et al., 2003; Bolzonella et al., 2005a, 2003a), organic solid wastes with or without wastewater treatment plant sludge (Kayhanian and Rich, 1995; Bolzonella et al., 2005b), cheese whey (Erguder et al., 2001), agro-industrial wastewaters (Demirer et al., 2000), grey water from vacuum toilets (Feng et al., 2006), cow wastes (Igoud et al., 2002), olive mill waste (Fezzani and Ben Cheikh, 2007; Erguder et al., 2000), etc. Due to the importance of anaerobic digestion as a treatment process, different dynamic models exist, such as the AM2 which was developed jointly by researchers of the INRA of Narbonne and the INRIA of Sophia-Antipolis in 2001 (Olivier et al., 2001). It
* Corresponding author. Tel./fax: +213 31 81 88 80. E-mail address: [email protected] (M. Bencheikh-lehocine). 0960-8524/$ - see front matter Ó 2008 Published by Elsevier Ltd. doi:10.1016/j.biortech.2008.07.064
is based on experimental results obtained on the fixed bed reactor of the INRA of Narbonne. This model is made of two steps: acidogenesis and methanogenesis corresponding to acido-acetogens and methanogens bacteria populations, respectively. As a more recent and elaborate model, the anaerobic digestion model no. 1 (ADM1), was developed by an international water association (IWA) group (Batstone et al., 2002). Its main feature is the consideration of the principal steps of anaerobic digestion process that are, respectively, substrate disintegration (non biological step), hydrolysis, acidogenesis, acetogenesis and finally the methanogenesis with seven different bacteria groups. Since its development in 2002 and up to now, the ADM1 has been tested and used on different substrates where a great number of research works are reported in this literature. As examples, one can cite (Blumensaat and Keller, 2005) who modified the initial version of ADM1 for the simulation of a dynamic behaviour of a pilot scale digester using sludge, in order to ensure a faultless model implementation. They obtained accurate results for the cases of low to medium loading rates. However, the accuracy showed a decline with the increase of the loading rate. Parker (2005) considered the application of the ADM1 to a variety of anaerobic digestion configurations where the results showed, in most of the considered cases, that the model was able to reproduce the trends of the experimental results. Feng et al. (2006) found that the ADM1 is not sensitive to the distribution ratio of carbohydrates, proteins and lipids, whereas the fraction of short chain fatty acids (SCFA) in the influent is rather more important.
1540
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543
Consequently, the great capabilities of ADM1 in modelling different types of substrates and calculations have been the motivating factor to use it in the present work to evaluate the performances of a co-digestion process for the treatment of organic municipal solid waste and waste activated sludge in the above mentioned 2000 m3 reactor working at a temperature of 37 °C.
The effluent characteristics are shown in Table 2. Total and volatile solids were clearly reduced around 30% with respect to the influent parameters shown in Table 1, Values of alkalinity, pH and VFA were similar to typical results of literature (Bolzonella et al., 2003b) and stable enough, with low fluctuations. Finally the gas production characteristics are presented in Table 3.
2. Methods
3.2. Implementation of ADM1
2.1. The anaerobic digestion model (ADM1)
The substrate (solid waste + sludge) was characterized according to the original version of ADM1 (Batstone et al., 2002). Therefore, the input data were calculated on the basis of the influent substrate characteristics mentioned in Table 1. Thereafter, the ADM1 model was calibrated, using experimental data, through the optimisation of the different parameters, mainly the disintegration and hydrolysis ones, with a constraint on their values which should be within the physical range. The experimental and simulated results are discussed in the following paragraph.
As mentioned above the ADM1 was developed by the IWA group (Batstone et al., 2002) with the objective to build a full mathematical model based intimately on the phenomenological model in order to simulate, at best, anaerobic reactors. It includes, as a first step, the disintegration of solid complexes (non biological step) into carbohydrates, lipids, proteins and inert material (soluble and particulate inert). The second step is the hydrolysis process of the disintegration products under an enzymatic action to produce sugars, amino acids and long chain fatty acids (LCFA), successively. Then, amino acids and sugars are fermented to produce VFA, hydrogen and carbon dioxide (acidogenesis). Then LCFA, proprionic acid, butyric acid and valeric acid are anaerobically oxided into acetate, carbon dioxide and hydrogen (acetogenesis). Finally, methane can be produced through two paths: the first one is based on acetate whereas the second one is through the reduction of carbon dioxide by molecular hydrogen. The organic species and molecular hydrogen, in this model, are expressed as chemical oxygen demand (COD), whereas inorganic nitrogen and inorganic carbon species are expressed through their molecular concentrations. Extensions and modifications were brought to ADM1 to enlarge its prediction capabilities by, taking into account other factors such as, for instance, the sulfato-reductors or the degradation of certain substrates (Wolfsberger and Holubar, 2006; Batstone and Keller, 2003). Moreover, Usama (2004–2005) considered the toxic effects of cyanide as an inhibition process for acetate. 2.2. Reactor and monitoring As mentioned previously, experimental data were obtained from the monitoring of an anaerobic co-digestion of municipal solid waste mixed with wastewater treatment plant sludge, carried out in the industrial digester which was operated under mesophilic conditions (37 °C). The substrate is introduced daily with an influent mass loading of 1 kg TVS/m3/day and a hydraulic retention time (HRT) of around 27 days. Influent and effluent analysis were made daily for pH, total solids (TS), volatile solids (VS), volatile fatty acids (VFA), biogas volume produced and its composition, partial alkalinity (TA) and total alkalinity (TAC), ammoniacal nitrogen (NH3). Other analyses were made two or three times a week, like chemical oxygen demand (COD), Nitrogen djeldhal (TKN) and total phosphorous (Ptot). These parameters were determined according to the standard methods for water and wastewater examination. 3. Results and discussion 3.1. Experimental monitoring of the full scale reactor The full scale anaerobic digester was monitored for six months. Typical characteristics of influent feed are shown in Table 1. This is the stream resulting from the blend of biowaste and waste activated sludge. The typical solids concentration was some 3.9% while TVS were 65% of TS.
3.2.1. Estimation of kinetic parameter values Before starting the simulation step, the kinetics parameters concerning the disintegration and hydrolysis phases were first estimated. These were evaluated considering both experimental batch runs; i.e., for biowaste and sludge hydrolysis (batch tests in serum bottles) and the modelling of experimental data changing the values of the kinetic parameters. In particular, the ADM1 was first used to simulate experimental data obtained during the trials of anaerobic co-digestion runs. It was found that the constants for disintegration and hydrolysis suitable for this study were equal to typical values reported in literature for biowaste (Mata-Alvarez, 2003) as presented in Table 4. All the values of kinetics and stoichiometry constants were then maintained as in the ADM1 model. 3.3. Simulation of the process and comparison with experimental data After estimating the parameters of disintegration and hydrolysis of the substrate, the obtained results concerning the chemical oxygen demand, both total and dissolved (TCOD and SCOD) as well as the total volatile fatty acid (TVFA), are represented in Fig. 1. It can be noticed that the simulated results are in good agreement with the experimental values up to the 15th day. Thereafter, a deviation is shown between the two sets of values, i.e., experimental and simulated, for the total COD. However if the soluble results are considered this deviation is probably due to the particulate fraction of COD. Moreover, the substrate distribution between proteins, carbohydrates and lipids was not measured but default model values were adopted for this parameter. This can be justified by the fact that the used experimental rig is housed in the Treviso wastewater treatment plant and influent characteristics are daily varying. The reported values in the literature are specified to particular types of wastes, hence may not be safely used in this work. For the results of TVFA it can be seen that they show a kind of a good stability in the digester operations and they fit reasonably well the experimental values. Fig. 2 shows the variation of total gas volume produced with time, which clearly depends on the nature, composition and biodegradability of the waste. In fact, even though the mass loading in the digester is maintained almost constant (3% variation), the amount of sludge and biowaste can be different at any day, and hence leading to different biogas production. Moreover, the structural limitations of the ADM1 imply that the simulated gas production follows an average path; therefore simulated data are only partially confounded with experimental values.
1541
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 Table 1 Influent characteristics Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
pH NH3 (mg N/l) TKN (mg N/l) TCOD (mg COD/l) Ptot (mg P/g TS) TS (g/l) TVS (g/l) TVS (% TS) VFA (mg COD/l) TA at pH 6 (mg CaCO3/l) TA at pH 4 (mg CaCO3/l)
6.5 18. 47.9 691.9 24.0 39.1 25.8 65 225.8 201.6 590.5
5.9 8 40 591.1 669.2 29 23.2 57.1 22.6 39 380
6.9 46.5 52.5 822.1 1183 48.1 29.5 70 1358.7 400 1268.8
0.28 10 3.54 69.4 172.4 3.28 1.43 2.57 364.7 92.4 201.4
48 38 23 27 23 47 47 47 47 49 48
Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
pH NH3 (mg N/l) TKN (mg N/l) COD (mg COD/l) Ptot (mg P/g TS) TS (g/l) TVS (g/l) TVS (% TS) VFA (mg COD/l) TA at pH 6 (mg CaCO3/l) TA at pH 4 (mg CaCO3/l)
7.4 593.1 41.1 625.5 28.4 31.8 18 56.9 12.1 2342.3 1469.3
7.2 440 35.1 565.39 7 27.7 15.4 49.9 2.1 1100 2040
7.7 720 44.1 702.2 125 38.2 20.8 63.2 30.6 2163 2982
0.14 66 2.48 42.1 15 1.6 1 2.4 7.6 163.1 175.2
50 38 21 25 23 48 47 47 41 50 50
Table 2 Effluent characteristics
Table 3 Characteristics and biogas production Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
Biogas volume (m3/day) SGP (m3 biogas/kg TVS) Gas production rate (GPR) (m3 biogas/m3 day) % CH4 (%) % CO2 (%) V-CH4 (m3/day) V-CO2 (m3/day) H2S (ppm)
606.4 0.31 0.4 65.8 34.2 399.7 206.7 622.7
375 0.118 0.183 60.3 31.9 246 129 321
860 0.45 0.42 68.1 39.7 559.9 300 778
129.3 0.09 0.06 1.3 1.3 83.7 46.6 125.1
46 39 48 49 49 46 46 43
Table 4 Initial and estimated values of kinetic parameters Kinetic parameters
Names
Units
Kdis Khyd.Ch Khyd.Pr Khyd.Li
Disintegration constant Carbohydrate hydrolysis constant Proteins hydrolysis constant Lipids hydrolysis constant
Day Day Day Day
a b
1 1 1 1
Initial values used in ADM1
Initial values
Estimated values
0.5b 10b 10b 10b
0.7 1.25a 0.5a 0.4a
0.5 1.017 0.3842 0.999
Middle values obtained from Mata-Alvarez (2003). Values obtained from Batstone and Keller (2003).
In order to have an insight on the biogas production the input organic loading rate was calculated and plotted as shown in Fig. 2 where it can be seen that dynamic conditions are prevailing. Fig. 3 shows the experimental and simulated results of gas production which is composed of methane gas, carbon dioxide and hydrogen. However, since the hydrogen volume is not important it was not analysed in the total volume and it was assumed that the gas is only made of methane gas and carbon dioxide. Even though, this assumption could influence the errors, the obtained results are satisfactory. It can be noticed that at the beginning of the simulation, the results are overestimated for methane gas and underestimated for the carbon dioxide. However, these results
show that the reactor presents a good stability from the point of view of gas composition. To see better what is happening in the system, inorganic carbon (IC), inorganic nitrogen (IN) and pH were represented in the same Fig. 4 and hence the alkalinity is implicit. IC should be regarded as bicarbonate alkalinity (BA), since the variation of BA is due to the neutralisation of VFA in the solution. BA or IC are more sensitive to VFA accumulation than pH, and are correlated empirically to VFA accumulation (Bolzonella et al., 2003a). However, the variation of BA could not be converted into VFA units. From the simulation point of view ADM1 does not represent fluctuations but shows an average trend at the beginning of
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 0.10 (simulated TCOD) (simulated SCOD) (simulated TVFA) (measured TCOD) (measured SCOD) (measured TVFA)
40
30
20
10
14 simulated IC simulated IN measured IC measured IN simulated pH measured pH
0.08
12
10
0.06
8
6
0.04
Effluent pH
50
Effluent IC and IN (Kmol/m3)
Effluent TCOD, SCOD and TVFA (KgCOD/m3)
1542
4 0.02 2
0 0
0.00 0
-10 0
5
10
15
20
25
20
30
Time (days)
30
Time (days)
10
Fig. 4. Comparison between the simulation and the experimental IC, IN and pH.
2.0
1000 simulated gas flow measured gas flow calculated OLR
1.8 1.6
800
Gas flow (m3/d)
1.4 1.2
600
1.0 0.8
400
0.6 0.4
200
0.2 0
Organic loading rate (OLR) (KgTVS/m3*d)
Fig. 1. Comparison between the simulation and the experimental total and soluble COD and TVFA.
0.0 0
5
10
15
20
25
30
Time (days) Fig. 2. Comparison between the simulated and the experimental biogas production rate and the variation of the organic loading rate (OLR) with time.
100 simulated CH4 simulated CO2 measured CH4 measured CO2
C H 4 and C O 2 (%)
80
60
40
20
The simulated results of inorganic nitrogen do not present the average experimental trend and seem to be underestimated compared to the experimental values (Fig. 4) contrarily to the hydrolysis and ammonification of disintegrated particulate matter which was underestimated by the model. This also may be due to the level of the kinetic constant values. The results of pH are well simulated by ADM1 and are stable, in comparison to BA variations (Fig. 4). As a monitoring variable, BA is more sensitive than pH, and therefore it can be used as a control parameter for the operation of anaerobic digesters. 4. Conclusion The ADM1 model was tested to simulate the behaviour of a bioreactor for the anaerobic co-digestion of waste activated sludge and biowaste. ADM1 showed acceptable simulating results, regarding the number of parameters involved and processes considered. However, it is important to note, according to the findings of this study that the ADM1 model is relatively limited in simulating complex processes such as anaerobic digestion. In fact it cannot reproduce the intimate variations of the different parameters, but an average trend is exhibited. This can be explained, as mentioned previously, by the fact that not all the input kinetic parameters were obtained via analyses but extracted from this literature. For the present case, the obtained results can be tested for the prediction of the different operating parameters. ADM1 can, therefore, be used as a managing tool of anaerobic digestion. The simulated results show an acceptable fit. However, at the start of the experiment, where a transient state prevails, the simulated results do not show a good fit. It was confirmed that inorganic carbon or bicarbonate alkalinity is a very sensitive parameter to volatile fatty acids accumulation than pH or VFA variations and hence it can be used as a monitoring parameter for VFA accumulation. References
0 0
5
10
15
20
25
30
Time (days) Fig. 3. Comparison between the simulation and the experimental % of CO2 and CH4.
the experiment. The differences between experimental and simulated results are damped after the first 10 days, similarly to the other results (Fig. 4).
Batstone, D.J., Keller, J., 2003. Industrial applications of the IWA anaerobic digestion model No. 1 (ADM1). Water Sci. Technol. 47 (12), 199–206. Batstone, 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. International Water Association (IWA), Publishing, London, UK. Blumensaat, F., Keller, J., 2005. Modelling of two-stage anaerobic digestion using the IWA anaerobic digestion model no. 1 (ADM1). Water Res. 39, 171–183. Bolzonella, D., Innocenti, L., Pavan, P., Traverso, P., Cecchi, F., 2003a. Semi-dry thermophilic digestion of the organic fraction of municipal solid wastes: focusing on the start-up phase. Bioresour. Technol. 86, 123–129.
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 Bolzonella, D., Battistoni, P., Mata-Alvarez, J., Cecchi, F., 2003b. Anaerobic digestion of organic solid wastes: process behaviour in transient conditions. Water Sci. Technol. 48 (4), 1–8. Bolzonella, D., Francesco, Fatone, Pavan, P., Cecchi, F., 2005a. Anaerobic fermentation of organic municipal solid wastes for the production of soluble organic compounds. Ind. Eng. Chem. Res. 44, 3412–3418. Bolzonella, D., Pavan, P., Battistoni, P., Cecchi, F., 2005b. Mesophilic anaerobic digestion of waste activated sludge, influence of the solid retention time in the wastewater treatment process. Process Biochem. 40, 1453–1460. Cout, D., Genon, G., Ranzini, M., Romano, P., 1994. Anaerobic co-digestion of municipal sludge and industrial organic waste. In: Proceedings of the Seventh International Symposium on Anaerobic Digestion, Johannesburg, South Africa. Demirer, G.N., Duran, M., Erguder, T.H., Guven, E., Ugurlu, O., Tezel, U., 2000. Anaerobic treatability and biogas production potential studies of different agroindustrial wastewater in Turkey. Biodegradation 11, 401–405. Erguder, T.H., Guven, E., Demirer, G.N., 2000. Anaerobic digestion of olive mill wastes in batch reactors. Process Biochem. 36, 243–248. Erguder, T.H., Tezel, U., Guven, E., Demirer, G.N., 2001. Anaerobic biotransformation and methane generation potential of cheese whey in batch and UASB reactors. Waste Manage. 21, 643–650. Feng, Y., Behrendt, J., Wendland, C., Otterpohl, R., 2006. Parameters analysis and discussion of the anaerobic digestion model no. 1 (ADM1). Water Sci. Technol. 54 (4), 139–147. Fezzani, Boubaker, Ben Cheikh, Ridha, 2007. Anaerobic co-digestion of olive mill wastewater with olive mill solid waste in a tubular digester at mesophilic temperature. Bioresour. Technol. 98, 769–774. Igoud, S., Tou, I., Kehal, S., Mansouri, N., Touzi, A., 2002. Première approche de lacaractérisation du biogaz produit à partir des déjections bovines. Énergierenouvelable 5, 123–128. Jewell, W.J., 1979. Future trends in digester region. In: Proceedings of the First International Symposium on Anaerobic Digestion, Cardiff, Wales. Applied Science Publishers, London, pp. 17–21.
1543
Kayhanian, M., Rich, D., 1995. Pilot-scale high solids thermophilic anaerobicdigestion of municipal solid waste with an emphasis on nutriment requirements. Biomass Bioenergy 8 (6), 433–444. Kayhanian, M., Tchobanoglus, G., 1992. Pilot investigation of an innovative tow-stage anaerobic digestion and aerobic composting process for the recovery of energy and compost from the organic fraction of MSW. In: Proceedings of the First International Symposium on Anaerobic Digestion, Venice, Italy. Macé, S., Bolzonella, D., Cecchi, F., Mata-Alvarez, J., 2003. Comparison of the biodegradability of the grey fraction of municipal solid waste of Barcelona in mesophilic and thermophilic conditions. Water Sci. Technol. 48 (4), 21–28. Mata-Alvarez, J., 2003. Biomethanization of the Organic Fraction of Municipal Solid Wastes. IWA Publishing, London, UK, pp. 42. Mata-Alvares, J., Macé, S., Libres, P., 2000. Anaerobic digestion of organic solid wastes. An overview of research achievements and perspectives. Bioresour. Technol. 74, 3–16. Olivier, Bernard, Zakaria, Hadj-Sadok, Denis, Dochain, Antoine, Genovisi, Steyer, Jean-Philipe, 2001. Dynamical model development and parameter identification for an anaerobic wastewater treatment process. Biotechnol. Bioenergy 75 (4), 424–438. Parker, Wayne J., 2005. Application of the ADM1 model to advanced anaerobic digestion. Bioresour. Technol. 96, 1832–1842. Vallini, G., Cecchi, F., Pavan, P., Alvarez, M., Basseti, A., 1992. Recovery and disposal of the organic fraction of MSW by means of combined anaerobic and aerobic biotreatments. In: Proceedings of the Fifth International Symposium on Anaerobic Digestion of Solid Wastes, Venice, Italy. Wolfsberger, A., Holubar, P., 2006. WP7 Biokinetic Data. Modelling and Control Second Year CROPGEN Meeting, Vienne. Zaher Usama El-Sayed, 2004–2005. Modelling and monitoring the anaerobic digestion process in view of optimisation and smooth operation of WWTP’s. PhD Thesis in Appl. Biol. Sci.: Environ. Technol. Academic Year 2004–2005, Universiteit Gent, Faculty of Bioscience Engineering.
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Application of the IWA ADM1 model to simulate anaerobic co-digestion of organic waste with waste activated sludge in mesophilic condition K. Derbal a, M. Bencheikh-lehocine a,*, F. Cecchi b, A.-H. Meniai a, P. Pavan c a
Laboratoire de l’Ingénierie des Procédés de l’Environnement (LIPE), Département de Chimie Industrielle, Faculté des Sciences de l’Ingénieur, Université Mentouri, Constantine, Algeria Department of Science and Technology, University of Verona, Verona, Italy c Department of Environmental Science, University of Venice, Venice, Italy b
a r t i c l e
i n f o
Article history: Received 29 April 2008 Received in revised form 13 July 2008 Accepted 16 July 2008 Available online 26 October 2008 Keywords: Anaerobic co-digestion Simulation ADM1 Mesophilic temperature Solid wastes
a b s t r a c t Anaerobic digestion model no. 1 model of international water association was applied to a full scale anaerobic co-digestion process for the treatment of the organic fraction of municipal solid wastes along with activated sludge wastes originating from a municipal wastewater treatment plant. This operation was carried out in a digester of 2000 m3 in volume. It is operates at an average hydraulic retention time of 26.9 days with an average organic loading rate of 1.01 kg TVS/m3 day, at a temperature of 37 °C with an average gas production rate of 0.296 m3/m3 day. The aim of the present study is to compare the results obtained from the simulation with the experimental values. The simulated results showed a good fit for pH, methane and carbon dioxide percentages, biogas volume, chemical oxygen demand, total volatile fatty acids, inorganic nitrogen and inorganic carbon. Ó 2008 Published by Elsevier Ltd.
1. Introduction Anaerobic digestion is worldwide used, particularly in Europe for the treatment of numerous types of biodegradable wastes. This is confirmed by the important number of treatment plants using this process on the industrial scale, during the past few years (Mata-Alvares et al., 2000). In fact, anaerobic digestion of the organic fraction of the municipal solid wastes alone or combined with organic sludge can contribute efficiently in solid waste reduction and biogas production (Jewell, 1979; Kayhanian and Tchobanoglus, 1992; Vallini et al., 1992; Cout et al., 1994). This process can be used for the solid waste treatment under mesophilic or thermophilic conditions (Macé et al., 2003; Bolzonella et al., 2005a, 2003a), organic solid wastes with or without wastewater treatment plant sludge (Kayhanian and Rich, 1995; Bolzonella et al., 2005b), cheese whey (Erguder et al., 2001), agro-industrial wastewaters (Demirer et al., 2000), grey water from vacuum toilets (Feng et al., 2006), cow wastes (Igoud et al., 2002), olive mill waste (Fezzani and Ben Cheikh, 2007; Erguder et al., 2000), etc. Due to the importance of anaerobic digestion as a treatment process, different dynamic models exist, such as the AM2 which was developed jointly by researchers of the INRA of Narbonne and the INRIA of Sophia-Antipolis in 2001 (Olivier et al., 2001). It
* Corresponding author. Tel./fax: +213 31 81 88 80. E-mail address: [email protected] (M. Bencheikh-lehocine). 0960-8524/$ - see front matter Ó 2008 Published by Elsevier Ltd. doi:10.1016/j.biortech.2008.07.064
is based on experimental results obtained on the fixed bed reactor of the INRA of Narbonne. This model is made of two steps: acidogenesis and methanogenesis corresponding to acido-acetogens and methanogens bacteria populations, respectively. As a more recent and elaborate model, the anaerobic digestion model no. 1 (ADM1), was developed by an international water association (IWA) group (Batstone et al., 2002). Its main feature is the consideration of the principal steps of anaerobic digestion process that are, respectively, substrate disintegration (non biological step), hydrolysis, acidogenesis, acetogenesis and finally the methanogenesis with seven different bacteria groups. Since its development in 2002 and up to now, the ADM1 has been tested and used on different substrates where a great number of research works are reported in this literature. As examples, one can cite (Blumensaat and Keller, 2005) who modified the initial version of ADM1 for the simulation of a dynamic behaviour of a pilot scale digester using sludge, in order to ensure a faultless model implementation. They obtained accurate results for the cases of low to medium loading rates. However, the accuracy showed a decline with the increase of the loading rate. Parker (2005) considered the application of the ADM1 to a variety of anaerobic digestion configurations where the results showed, in most of the considered cases, that the model was able to reproduce the trends of the experimental results. Feng et al. (2006) found that the ADM1 is not sensitive to the distribution ratio of carbohydrates, proteins and lipids, whereas the fraction of short chain fatty acids (SCFA) in the influent is rather more important.
1540
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543
Consequently, the great capabilities of ADM1 in modelling different types of substrates and calculations have been the motivating factor to use it in the present work to evaluate the performances of a co-digestion process for the treatment of organic municipal solid waste and waste activated sludge in the above mentioned 2000 m3 reactor working at a temperature of 37 °C.
The effluent characteristics are shown in Table 2. Total and volatile solids were clearly reduced around 30% with respect to the influent parameters shown in Table 1, Values of alkalinity, pH and VFA were similar to typical results of literature (Bolzonella et al., 2003b) and stable enough, with low fluctuations. Finally the gas production characteristics are presented in Table 3.
2. Methods
3.2. Implementation of ADM1
2.1. The anaerobic digestion model (ADM1)
The substrate (solid waste + sludge) was characterized according to the original version of ADM1 (Batstone et al., 2002). Therefore, the input data were calculated on the basis of the influent substrate characteristics mentioned in Table 1. Thereafter, the ADM1 model was calibrated, using experimental data, through the optimisation of the different parameters, mainly the disintegration and hydrolysis ones, with a constraint on their values which should be within the physical range. The experimental and simulated results are discussed in the following paragraph.
As mentioned above the ADM1 was developed by the IWA group (Batstone et al., 2002) with the objective to build a full mathematical model based intimately on the phenomenological model in order to simulate, at best, anaerobic reactors. It includes, as a first step, the disintegration of solid complexes (non biological step) into carbohydrates, lipids, proteins and inert material (soluble and particulate inert). The second step is the hydrolysis process of the disintegration products under an enzymatic action to produce sugars, amino acids and long chain fatty acids (LCFA), successively. Then, amino acids and sugars are fermented to produce VFA, hydrogen and carbon dioxide (acidogenesis). Then LCFA, proprionic acid, butyric acid and valeric acid are anaerobically oxided into acetate, carbon dioxide and hydrogen (acetogenesis). Finally, methane can be produced through two paths: the first one is based on acetate whereas the second one is through the reduction of carbon dioxide by molecular hydrogen. The organic species and molecular hydrogen, in this model, are expressed as chemical oxygen demand (COD), whereas inorganic nitrogen and inorganic carbon species are expressed through their molecular concentrations. Extensions and modifications were brought to ADM1 to enlarge its prediction capabilities by, taking into account other factors such as, for instance, the sulfato-reductors or the degradation of certain substrates (Wolfsberger and Holubar, 2006; Batstone and Keller, 2003). Moreover, Usama (2004–2005) considered the toxic effects of cyanide as an inhibition process for acetate. 2.2. Reactor and monitoring As mentioned previously, experimental data were obtained from the monitoring of an anaerobic co-digestion of municipal solid waste mixed with wastewater treatment plant sludge, carried out in the industrial digester which was operated under mesophilic conditions (37 °C). The substrate is introduced daily with an influent mass loading of 1 kg TVS/m3/day and a hydraulic retention time (HRT) of around 27 days. Influent and effluent analysis were made daily for pH, total solids (TS), volatile solids (VS), volatile fatty acids (VFA), biogas volume produced and its composition, partial alkalinity (TA) and total alkalinity (TAC), ammoniacal nitrogen (NH3). Other analyses were made two or three times a week, like chemical oxygen demand (COD), Nitrogen djeldhal (TKN) and total phosphorous (Ptot). These parameters were determined according to the standard methods for water and wastewater examination. 3. Results and discussion 3.1. Experimental monitoring of the full scale reactor The full scale anaerobic digester was monitored for six months. Typical characteristics of influent feed are shown in Table 1. This is the stream resulting from the blend of biowaste and waste activated sludge. The typical solids concentration was some 3.9% while TVS were 65% of TS.
3.2.1. Estimation of kinetic parameter values Before starting the simulation step, the kinetics parameters concerning the disintegration and hydrolysis phases were first estimated. These were evaluated considering both experimental batch runs; i.e., for biowaste and sludge hydrolysis (batch tests in serum bottles) and the modelling of experimental data changing the values of the kinetic parameters. In particular, the ADM1 was first used to simulate experimental data obtained during the trials of anaerobic co-digestion runs. It was found that the constants for disintegration and hydrolysis suitable for this study were equal to typical values reported in literature for biowaste (Mata-Alvarez, 2003) as presented in Table 4. All the values of kinetics and stoichiometry constants were then maintained as in the ADM1 model. 3.3. Simulation of the process and comparison with experimental data After estimating the parameters of disintegration and hydrolysis of the substrate, the obtained results concerning the chemical oxygen demand, both total and dissolved (TCOD and SCOD) as well as the total volatile fatty acid (TVFA), are represented in Fig. 1. It can be noticed that the simulated results are in good agreement with the experimental values up to the 15th day. Thereafter, a deviation is shown between the two sets of values, i.e., experimental and simulated, for the total COD. However if the soluble results are considered this deviation is probably due to the particulate fraction of COD. Moreover, the substrate distribution between proteins, carbohydrates and lipids was not measured but default model values were adopted for this parameter. This can be justified by the fact that the used experimental rig is housed in the Treviso wastewater treatment plant and influent characteristics are daily varying. The reported values in the literature are specified to particular types of wastes, hence may not be safely used in this work. For the results of TVFA it can be seen that they show a kind of a good stability in the digester operations and they fit reasonably well the experimental values. Fig. 2 shows the variation of total gas volume produced with time, which clearly depends on the nature, composition and biodegradability of the waste. In fact, even though the mass loading in the digester is maintained almost constant (3% variation), the amount of sludge and biowaste can be different at any day, and hence leading to different biogas production. Moreover, the structural limitations of the ADM1 imply that the simulated gas production follows an average path; therefore simulated data are only partially confounded with experimental values.
1541
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 Table 1 Influent characteristics Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
pH NH3 (mg N/l) TKN (mg N/l) TCOD (mg COD/l) Ptot (mg P/g TS) TS (g/l) TVS (g/l) TVS (% TS) VFA (mg COD/l) TA at pH 6 (mg CaCO3/l) TA at pH 4 (mg CaCO3/l)
6.5 18. 47.9 691.9 24.0 39.1 25.8 65 225.8 201.6 590.5
5.9 8 40 591.1 669.2 29 23.2 57.1 22.6 39 380
6.9 46.5 52.5 822.1 1183 48.1 29.5 70 1358.7 400 1268.8
0.28 10 3.54 69.4 172.4 3.28 1.43 2.57 364.7 92.4 201.4
48 38 23 27 23 47 47 47 47 49 48
Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
pH NH3 (mg N/l) TKN (mg N/l) COD (mg COD/l) Ptot (mg P/g TS) TS (g/l) TVS (g/l) TVS (% TS) VFA (mg COD/l) TA at pH 6 (mg CaCO3/l) TA at pH 4 (mg CaCO3/l)
7.4 593.1 41.1 625.5 28.4 31.8 18 56.9 12.1 2342.3 1469.3
7.2 440 35.1 565.39 7 27.7 15.4 49.9 2.1 1100 2040
7.7 720 44.1 702.2 125 38.2 20.8 63.2 30.6 2163 2982
0.14 66 2.48 42.1 15 1.6 1 2.4 7.6 163.1 175.2
50 38 21 25 23 48 47 47 41 50 50
Table 2 Effluent characteristics
Table 3 Characteristics and biogas production Parameters
Middle
Minimum
Maximum
Standard deviation
Number of samples
Biogas volume (m3/day) SGP (m3 biogas/kg TVS) Gas production rate (GPR) (m3 biogas/m3 day) % CH4 (%) % CO2 (%) V-CH4 (m3/day) V-CO2 (m3/day) H2S (ppm)
606.4 0.31 0.4 65.8 34.2 399.7 206.7 622.7
375 0.118 0.183 60.3 31.9 246 129 321
860 0.45 0.42 68.1 39.7 559.9 300 778
129.3 0.09 0.06 1.3 1.3 83.7 46.6 125.1
46 39 48 49 49 46 46 43
Table 4 Initial and estimated values of kinetic parameters Kinetic parameters
Names
Units
Kdis Khyd.Ch Khyd.Pr Khyd.Li
Disintegration constant Carbohydrate hydrolysis constant Proteins hydrolysis constant Lipids hydrolysis constant
Day Day Day Day
a b
1 1 1 1
Initial values used in ADM1
Initial values
Estimated values
0.5b 10b 10b 10b
0.7 1.25a 0.5a 0.4a
0.5 1.017 0.3842 0.999
Middle values obtained from Mata-Alvarez (2003). Values obtained from Batstone and Keller (2003).
In order to have an insight on the biogas production the input organic loading rate was calculated and plotted as shown in Fig. 2 where it can be seen that dynamic conditions are prevailing. Fig. 3 shows the experimental and simulated results of gas production which is composed of methane gas, carbon dioxide and hydrogen. However, since the hydrogen volume is not important it was not analysed in the total volume and it was assumed that the gas is only made of methane gas and carbon dioxide. Even though, this assumption could influence the errors, the obtained results are satisfactory. It can be noticed that at the beginning of the simulation, the results are overestimated for methane gas and underestimated for the carbon dioxide. However, these results
show that the reactor presents a good stability from the point of view of gas composition. To see better what is happening in the system, inorganic carbon (IC), inorganic nitrogen (IN) and pH were represented in the same Fig. 4 and hence the alkalinity is implicit. IC should be regarded as bicarbonate alkalinity (BA), since the variation of BA is due to the neutralisation of VFA in the solution. BA or IC are more sensitive to VFA accumulation than pH, and are correlated empirically to VFA accumulation (Bolzonella et al., 2003a). However, the variation of BA could not be converted into VFA units. From the simulation point of view ADM1 does not represent fluctuations but shows an average trend at the beginning of
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 0.10 (simulated TCOD) (simulated SCOD) (simulated TVFA) (measured TCOD) (measured SCOD) (measured TVFA)
40
30
20
10
14 simulated IC simulated IN measured IC measured IN simulated pH measured pH
0.08
12
10
0.06
8
6
0.04
Effluent pH
50
Effluent IC and IN (Kmol/m3)
Effluent TCOD, SCOD and TVFA (KgCOD/m3)
1542
4 0.02 2
0 0
0.00 0
-10 0
5
10
15
20
25
20
30
Time (days)
30
Time (days)
10
Fig. 4. Comparison between the simulation and the experimental IC, IN and pH.
2.0
1000 simulated gas flow measured gas flow calculated OLR
1.8 1.6
800
Gas flow (m3/d)
1.4 1.2
600
1.0 0.8
400
0.6 0.4
200
0.2 0
Organic loading rate (OLR) (KgTVS/m3*d)
Fig. 1. Comparison between the simulation and the experimental total and soluble COD and TVFA.
0.0 0
5
10
15
20
25
30
Time (days) Fig. 2. Comparison between the simulated and the experimental biogas production rate and the variation of the organic loading rate (OLR) with time.
100 simulated CH4 simulated CO2 measured CH4 measured CO2
C H 4 and C O 2 (%)
80
60
40
20
The simulated results of inorganic nitrogen do not present the average experimental trend and seem to be underestimated compared to the experimental values (Fig. 4) contrarily to the hydrolysis and ammonification of disintegrated particulate matter which was underestimated by the model. This also may be due to the level of the kinetic constant values. The results of pH are well simulated by ADM1 and are stable, in comparison to BA variations (Fig. 4). As a monitoring variable, BA is more sensitive than pH, and therefore it can be used as a control parameter for the operation of anaerobic digesters. 4. Conclusion The ADM1 model was tested to simulate the behaviour of a bioreactor for the anaerobic co-digestion of waste activated sludge and biowaste. ADM1 showed acceptable simulating results, regarding the number of parameters involved and processes considered. However, it is important to note, according to the findings of this study that the ADM1 model is relatively limited in simulating complex processes such as anaerobic digestion. In fact it cannot reproduce the intimate variations of the different parameters, but an average trend is exhibited. This can be explained, as mentioned previously, by the fact that not all the input kinetic parameters were obtained via analyses but extracted from this literature. For the present case, the obtained results can be tested for the prediction of the different operating parameters. ADM1 can, therefore, be used as a managing tool of anaerobic digestion. The simulated results show an acceptable fit. However, at the start of the experiment, where a transient state prevails, the simulated results do not show a good fit. It was confirmed that inorganic carbon or bicarbonate alkalinity is a very sensitive parameter to volatile fatty acids accumulation than pH or VFA variations and hence it can be used as a monitoring parameter for VFA accumulation. References
0 0
5
10
15
20
25
30
Time (days) Fig. 3. Comparison between the simulation and the experimental % of CO2 and CH4.
the experiment. The differences between experimental and simulated results are damped after the first 10 days, similarly to the other results (Fig. 4).
Batstone, D.J., Keller, J., 2003. Industrial applications of the IWA anaerobic digestion model No. 1 (ADM1). Water Sci. Technol. 47 (12), 199–206. Batstone, 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. International Water Association (IWA), Publishing, London, UK. Blumensaat, F., Keller, J., 2005. Modelling of two-stage anaerobic digestion using the IWA anaerobic digestion model no. 1 (ADM1). Water Res. 39, 171–183. Bolzonella, D., Innocenti, L., Pavan, P., Traverso, P., Cecchi, F., 2003a. Semi-dry thermophilic digestion of the organic fraction of municipal solid wastes: focusing on the start-up phase. Bioresour. Technol. 86, 123–129.
K. Derbal et al. / Bioresource Technology 100 (2009) 1539–1543 Bolzonella, D., Battistoni, P., Mata-Alvarez, J., Cecchi, F., 2003b. Anaerobic digestion of organic solid wastes: process behaviour in transient conditions. Water Sci. Technol. 48 (4), 1–8. Bolzonella, D., Francesco, Fatone, Pavan, P., Cecchi, F., 2005a. Anaerobic fermentation of organic municipal solid wastes for the production of soluble organic compounds. Ind. Eng. Chem. Res. 44, 3412–3418. Bolzonella, D., Pavan, P., Battistoni, P., Cecchi, F., 2005b. Mesophilic anaerobic digestion of waste activated sludge, influence of the solid retention time in the wastewater treatment process. Process Biochem. 40, 1453–1460. Cout, D., Genon, G., Ranzini, M., Romano, P., 1994. Anaerobic co-digestion of municipal sludge and industrial organic waste. In: Proceedings of the Seventh International Symposium on Anaerobic Digestion, Johannesburg, South Africa. Demirer, G.N., Duran, M., Erguder, T.H., Guven, E., Ugurlu, O., Tezel, U., 2000. Anaerobic treatability and biogas production potential studies of different agroindustrial wastewater in Turkey. Biodegradation 11, 401–405. Erguder, T.H., Guven, E., Demirer, G.N., 2000. Anaerobic digestion of olive mill wastes in batch reactors. Process Biochem. 36, 243–248. Erguder, T.H., Tezel, U., Guven, E., Demirer, G.N., 2001. Anaerobic biotransformation and methane generation potential of cheese whey in batch and UASB reactors. Waste Manage. 21, 643–650. Feng, Y., Behrendt, J., Wendland, C., Otterpohl, R., 2006. Parameters analysis and discussion of the anaerobic digestion model no. 1 (ADM1). Water Sci. Technol. 54 (4), 139–147. Fezzani, Boubaker, Ben Cheikh, Ridha, 2007. Anaerobic co-digestion of olive mill wastewater with olive mill solid waste in a tubular digester at mesophilic temperature. Bioresour. Technol. 98, 769–774. Igoud, S., Tou, I., Kehal, S., Mansouri, N., Touzi, A., 2002. Première approche de lacaractérisation du biogaz produit à partir des déjections bovines. Énergierenouvelable 5, 123–128. Jewell, W.J., 1979. Future trends in digester region. In: Proceedings of the First International Symposium on Anaerobic Digestion, Cardiff, Wales. Applied Science Publishers, London, pp. 17–21.
1543
Kayhanian, M., Rich, D., 1995. Pilot-scale high solids thermophilic anaerobicdigestion of municipal solid waste with an emphasis on nutriment requirements. Biomass Bioenergy 8 (6), 433–444. Kayhanian, M., Tchobanoglus, G., 1992. Pilot investigation of an innovative tow-stage anaerobic digestion and aerobic composting process for the recovery of energy and compost from the organic fraction of MSW. In: Proceedings of the First International Symposium on Anaerobic Digestion, Venice, Italy. Macé, S., Bolzonella, D., Cecchi, F., Mata-Alvarez, J., 2003. Comparison of the biodegradability of the grey fraction of municipal solid waste of Barcelona in mesophilic and thermophilic conditions. Water Sci. Technol. 48 (4), 21–28. Mata-Alvarez, J., 2003. Biomethanization of the Organic Fraction of Municipal Solid Wastes. IWA Publishing, London, UK, pp. 42. Mata-Alvares, J., Macé, S., Libres, P., 2000. Anaerobic digestion of organic solid wastes. An overview of research achievements and perspectives. Bioresour. Technol. 74, 3–16. Olivier, Bernard, Zakaria, Hadj-Sadok, Denis, Dochain, Antoine, Genovisi, Steyer, Jean-Philipe, 2001. Dynamical model development and parameter identification for an anaerobic wastewater treatment process. Biotechnol. Bioenergy 75 (4), 424–438. Parker, Wayne J., 2005. Application of the ADM1 model to advanced anaerobic digestion. Bioresour. Technol. 96, 1832–1842. Vallini, G., Cecchi, F., Pavan, P., Alvarez, M., Basseti, A., 1992. Recovery and disposal of the organic fraction of MSW by means of combined anaerobic and aerobic biotreatments. In: Proceedings of the Fifth International Symposium on Anaerobic Digestion of Solid Wastes, Venice, Italy. Wolfsberger, A., Holubar, P., 2006. WP7 Biokinetic Data. Modelling and Control Second Year CROPGEN Meeting, Vienne. Zaher Usama El-Sayed, 2004–2005. Modelling and monitoring the anaerobic digestion process in view of optimisation and smooth operation of WWTP’s. PhD Thesis in Appl. Biol. Sci.: Environ. Technol. Academic Year 2004–2005, Universiteit Gent, Faculty of Bioscience Engineering.