application of anaerobic digestion model no. 1 for describing anaerobic digestion of grass, maize,...

7
Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine Piotr Biernacki , Sven Steinigeweg, Axel Borchert, Frank Uhlenhut EUTEC Institute, University of Applied Sciences Emden/Leer, Constantiaplatz 4, 26723 Emden, Germany highlights " Four substrates were characterized by Weender/van Soest analysis. " Anaerobic Digestion Model No. 1 was used to model biogas production. " As an optimization method downhill simplex methods algorithm was applied. " New kinetic constants indicated satisfactory correlation with experimental values. " Applicability of the model for optimization of biogas power plants is confirmed. article info Article history: Received 22 March 2012 Received in revised form 16 August 2012 Accepted 28 September 2012 Available online 13 October 2012 Keywords: Biogas technology Mesophilic anaerobic digestion Mathematical modeling ADM1 Hydrolysis kinetics abstract Anaerobic digestion of organic waste plays an important role for the development of sustainable energy supply based on renewable resources. For further process optimization of anaerobic digestion, biogas production with the commonly used substrates, grass, maize, and green weed silage, together with indus- trial glycerine, were analyzed by the Weender analysis/van Soest method, and a simulation study was performed, based on the International Water Association’s (IWA) Anaerobic Digestion Model No. 1 (ADM1). The simplex algorithm was applied to optimize kinetic constants for disintegration and hydro- lysis steps for all examined substrates. Consequently, new parameters were determined for each evalu- ated substrate, tested against experimental cumulative biogas production results, and assessed against ADM1 default values for disintegration and hydrolysis kinetic constants, where the ADM1 values for mes- ophilic high rate and ADM1 values for solids were used. Results of the optimization lead to a precise pre- diction of the kinetics of anaerobic degradation of complex substrates. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction 1.1. Biogas modeling through optimization As anaerobic digestion of organic waste is increasingly practiced to produce biogas as a renewable energy resource (Directive 2009/ 28/EC, 2009), it is necessary to carry out process optimization based on reliable simulation models. In order to ensure that a model is useful also for plant operators, a widely accepted model should be the basis of model develop- ment. The primary goal is to improve a model already applied in practice with respect to the reliable calculation of digester dynam- ics for a wide range of substrates. Therefore it was decided to make use of a common model and to analyze the agreement between experimental and calculated data. This analysis shows capabilities and limitations of an established model and gives information about necessary improvements. In the current study, a reliable model for anaerobic digestion of different substrates and their mix- tures was developed based on the Anaerobic Digestion Model No. 1 (ADM1). It was shown that ADM1 was capable of describing biogas production rate and composition without major changes to the model structure. Nevertheless, improvement of parameters was necessary since the initial biomass disintegration and hydrolysis phase was not reflected adequately for different substrates. 1.2. Anaerobic Digestion Model No. 1 The Anaerobic Digestion Model No. 1 (ADM1) was developed by the International Water Association’s (IWA) Task Group (Batstone et al., 2002). The strength of this model is in its consideration of separate biomass fractions and their decay, apart from incorporat- ing four main stages of anaerobic degradation, and dividing them into 31 processes and 33 groups of fractions. Moreover, the model includes a composite fraction (X C ), which represents a complex substrate. The composite fraction (X C ) is degraded into 0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2012.09.128 Corresponding author. Tel.: +49 4921 807 1876. E-mail address: [email protected] (P. Biernacki). Bioresource Technology 127 (2013) 188–194 Contents lists available at SciVerse ScienceDirect Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Upload: piotr-biernacki

Post on 24-Nov-2016

218 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

Bioresource Technology 127 (2013) 188–194

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestionof grass, maize, green weed silage, and industrial glycerine

Piotr Biernacki ⇑, Sven Steinigeweg, Axel Borchert, Frank UhlenhutEUTEC Institute, University of Applied Sciences Emden/Leer, Constantiaplatz 4, 26723 Emden, Germany

h i g h l i g h t s

" Four substrates were characterized by Weender/van Soest analysis." Anaerobic Digestion Model No. 1 was used to model biogas production." As an optimization method downhill simplex methods algorithm was applied." New kinetic constants indicated satisfactory correlation with experimental values." Applicability of the model for optimization of biogas power plants is confirmed.

a r t i c l e i n f o

Article history:Received 22 March 2012Received in revised form 16 August 2012Accepted 28 September 2012Available online 13 October 2012

Keywords:Biogas technologyMesophilic anaerobic digestionMathematical modelingADM1Hydrolysis kinetics

0960-8524/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.biortech.2012.09.128

⇑ Corresponding author. Tel.: +49 4921 807 1876.E-mail address: [email protected]

a b s t r a c t

Anaerobic digestion of organic waste plays an important role for the development of sustainable energysupply based on renewable resources. For further process optimization of anaerobic digestion, biogasproduction with the commonly used substrates, grass, maize, and green weed silage, together with indus-trial glycerine, were analyzed by the Weender analysis/van Soest method, and a simulation study wasperformed, based on the International Water Association’s (IWA) Anaerobic Digestion Model No. 1(ADM1). The simplex algorithm was applied to optimize kinetic constants for disintegration and hydro-lysis steps for all examined substrates. Consequently, new parameters were determined for each evalu-ated substrate, tested against experimental cumulative biogas production results, and assessed againstADM1 default values for disintegration and hydrolysis kinetic constants, where the ADM1 values for mes-ophilic high rate and ADM1 values for solids were used. Results of the optimization lead to a precise pre-diction of the kinetics of anaerobic degradation of complex substrates.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. Biogas modeling through optimization

As anaerobic digestion of organic waste is increasingly practicedto produce biogas as a renewable energy resource (Directive 2009/28/EC, 2009), it is necessary to carry out process optimizationbased on reliable simulation models.

In order to ensure that a model is useful also for plant operators,a widely accepted model should be the basis of model develop-ment. The primary goal is to improve a model already applied inpractice with respect to the reliable calculation of digester dynam-ics for a wide range of substrates. Therefore it was decided to makeuse of a common model and to analyze the agreement betweenexperimental and calculated data. This analysis shows capabilitiesand limitations of an established model and gives information

ll rights reserved.

(P. Biernacki).

about necessary improvements. In the current study, a reliablemodel for anaerobic digestion of different substrates and their mix-tures was developed based on the Anaerobic Digestion Model No. 1(ADM1). It was shown that ADM1 was capable of describing biogasproduction rate and composition without major changes to themodel structure. Nevertheless, improvement of parameters wasnecessary since the initial biomass disintegration and hydrolysisphase was not reflected adequately for different substrates.

1.2. Anaerobic Digestion Model No. 1

The Anaerobic Digestion Model No. 1 (ADM1) was developed bythe International Water Association’s (IWA) Task Group (Batstoneet al., 2002). The strength of this model is in its consideration ofseparate biomass fractions and their decay, apart from incorporat-ing four main stages of anaerobic degradation, and dividingthem into 31 processes and 33 groups of fractions. Moreover, themodel includes a composite fraction (XC), which represents acomplex substrate. The composite fraction (XC) is degraded into

Page 2: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194 189

carbohydrates (XCh), proteins (XPr), lipids (XLi) and inerts (XI) frac-tions during the disintegration step (Batstone et al., 2002; IFAK,2005). Additionally, due to its capability to describe the biogas pro-duction rate and composition, the ADM1 was commonly used as ananaerobic degradation model for different substances, like grass si-lage (Koch et al., 2010, 2009; Wichern et al., 2009), agro-waste(Gali et al., 2009), agricultural substrates (Lübken et al., 2010), cat-tle manure (Myint et al., 2007; Schoen et al., 2009; Wichern et al.,2008), cattle manure and maize (Amon et al., 2007), cattle manureand co-substrates (Lübken et al., 2007).

The ADM1 was updated by Wett et al. (2006) who added a newinert decay products fraction (XP) whose formation is described bya decayed biomass factor (fP). According to Koch et al. (2010), thisupdate ensures that nutrient mineralization is incorporated intothe model. Accordingly, the name of the model was changed toADM1xp by IFAK.

1.3. Disintegration and hydrolysis kinetic constants

ADM1 was originally developed to describe anaerobic digestionof sludge from waste water treatment plants. The degradation ofcomplex organic material is assumed to pass four stages startingfrom complex organic materials to monomers to gaseouscompounds.

The extracellular biological and non-biological breakdown ofcomplex organic substrates to soluble substrates is expressed asdisintegration and hydrolysis phase. The disintegration phase rep-resents degradation of composite fraction (XC) into carbohydrates(XCh), proteins (XPr), lipids (XLi) and inerts (XI) fractions. Furtherenzymatic degradation of the non-inert fractions into monosaccha-rides (SSU), amino acids (SAA) and long chain fatty acids (SFA) repre-sents the hydrolysis stage. This approach has been widely appliedby Gali et al. (2009), Wett et al. (2006) and Schoen et al. (2009).

Disintegration and hydrolysis are described in ADM1 using firstorder kinetics. The disintegration kinetic constant for compositedegradation is described as kdis, the hydrolysis constant for thehydrolysis of carbohydrates, lipids and proteins are khyd_ch, khyd_li

and khyd_pr, respectively (Batstone et al., 2002). The values forhydrolysis and disintegration phase kinetic constants as proposedby Batstone et al. (2002) are listed in Table 1. The hydrolysis wasreported to be the rate limiting step of the anaerobic degradation(Garcia-Heras, 2003; Flotats et al., 2006), and as noted by Fenget al. (2006), ADM1’s default values for solids are leading to theelimination of the influence of the hydrolysis step on the simula-tion. Consequently, as proposed by Garcia-Heras (2003), there isa need for further experimental validation of disintegration and

Table 1Kinetic constant values found in the literature for mesophilic digestion of different substr

Description References

UnitADM1 values for solids Batstone et al. (2002)ADM1 values for high rate Batstone et al. (2002)The most common values Garcia-Heras (2003)Different particular substances Christ et al. (2000)Grass silage Wichern et al. (2009)Grass silage Wichern et al. (2009)Grass silage Veeken and Hamelers (1999)Grass silage Koch et al. (2009)Grass silage Koch et al. (2010)Agro-residues Gali et al. (2009)Corn stover Hu and Yu (2005)Crops and crops residues Lehtomaki et al. (2005)

a kdis is an acronym of kinetic constant describing disintegration phase.b khyd_ch is an acronym of kinetic constant describing carbohydrates hydrolysis phasec khyd_pr is an acronym of kinetic constant describing proteins hydrolysis phase.d khyd_li is an acronym of kinetic constant describing lipids hydrolysis phase.

hydrolysis kinetic constants. Since the kinetic constants summa-rized by Vavilin et al. (2008) are much lower than the ADM1’s de-fault values, these values are presented in Table 1. Christ et al.(2000) also proposed much lower values for disintegration andhydrolysis constants. Furthermore, Wichern et al. (2008) per-formed a calibration of ADM1 values during their investigation ofcattle manure and lowered the disintegration constant from a de-fault value of 0.4 d�1 to 0.05 d�1; however, for monofermentationof grass silage, Wichern et al. (2009), increased the disintegrationconstant to 1.0 d�1. Therefore, there is a need for direct determina-tion of the kinetic constants for substrates commonly used for bio-gas production. Table 1 presents hydrolysis and disintegrationkinetic coefficients of the first-order rate for different substratesfound in the literature.

2. Methods

2.1. Batch reactors set-up and operation

Batch experiments were prepared in accordance with VDI 4630(VDI, 2006); however, because determination of the kinetic con-stants was of the main focus rather than the overall biogas pro-duced from a substrate, experiments were carried out for 15 daysto describe the beginning of biogas production as necessary forthe determination of the kinetic constants. Since the duration ofthe batch tests depends on the inoculum concentration and activ-ity of the inoculum (Angelidaki et al., 2007) the evaluated sub-strates accounted for only 1% of the reactor’s overall mass toensuring an authentic biogas power plant feeding scenario, thereduction in the duration of the experiments was consideredreasonable.

As reactors, bottles of 1100 ml volume were used. The contentswere manually stirred and incubated in a water bath at 38 �C. Bio-gas production was measured hourly with an ANKOM’s(N1v0,4RF2; RFS#194) gas production system and readings weretransmitted electronically to a computer.

Fresh inoculums were obtained from the EWE Wittmund biogaspower plant (Wittmund, Lower Saxony, Germany) prior to eachexperiment. The inoculums were characterized, and in additionto basic characteristic, also total volatile fatty acids/alkalinity ratio(FOS/TAC ratio) analysis was performed with the Biogas TitrationManager from HACH LANGE, and the ammonium content wasmeasured by use of the HACH LANGE cuvette test (LCK 303 andLCK 305).The main substrates used at the EWE Wittmund biogaspower plant are cattle manure and organic waste a mixture of foodresidues from kitchens, restaurants and a hospital. The batch reac-

ates.

kdisa khyd_ch

b khyd_prc khyd_li

d

[d�1] [d�1] [d�1] [d�1]0.5 10 10 100.4 0.25 0.2 0.1

0.5–2.0 0.25–0.8 0.1–0.70.025–0.200 0.015–0.075 0.005–0.010

10.26

0.266 0.266 0.2660.6 0.6 0.60.14/0.5 0.8 0.14/0.5

0.15 10 10 100.94 0.94 0.940.009–0.094 0.009–0.094 0.009–0.094

.

Page 3: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

190 P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194

tors (bottles) were filled with 495 g of inoculum before adding sub-strates to a level of 500 g. Three batch reactors were prepared inparallel for each substrate and the whole experiment was repeated.

Grass, maize, and green weed silages were collected from localfarmers. Industrial glycerine waste was provided by the EWE Witt-mund biogas power plant. Characteristics of the substrates areavailable in Table 2.

2.2. Characterization of complex substrates

In order to enhance ADM1’s capability of biogas plant optimiza-tion, commonly used substrates for biogas production were ana-lyzed and the kinetic constants for disintegration and hydrolysisphase were determined. The substrates were tested using thewell-established Weender analysis and van Soest method (vanSoest and Wine, 1967; Wichern et al., 2008,2009; Koch et al.,2010), described in Naumann and Bassler (1993). The outcome ofthe analysis indicates a fractionation of the organic matter be-tween raw lipids (RL), raw protein (RP), raw fiber (RF) and N-freeextract (NfE) (Naumann and Bassler, 1993). The sum of raw fiber(RF) and N-free extract (NfE) represents the carbohydrate contentof the substrate. The further split into starch, cellulose, hemicellu-lose and lignin can be reached with the use of the van Soest exten-sion, where three other fractions are introduced: Neutral Detergentfiber:fiber (NDF), Acid Detergent fiber:fiber (ADF) and AcidDetergent Lignin (ADL). This approach was also recommended byLübken et al. (2007) while testing inhomogeneous substrates,despite depending on COD measurements.Koch et al. (2010) alsoproposed a method to incorporate fodder analysis into ADM1 bycalculating theoretical oxygen demand (ThOD) for each fractionof the substance (proteins, lipids, carbohydrates), and then calcu-lating the composite material (XC) using the following equation:

XC¼qsubstrate �DM �ðRP �ThODPrÞþðRL�ThODLiÞþðADL �ThODLiÞþ

ðRFþNfE�ADLÞ�ThODCh

� �kgCOD

ms

� �;

ð1Þ

Table 2Characteristics of the examined substrates.

Parameter Unit Grass

Dry mass (DM) [%] 41.9Organic dry mass [% DM] 87.9Raw protein [% DM] 18.5Raw lipids [% DM] 3.5Raw fiber [% DM] 26.3Neutral Detergent fiber:fiber (NDF) [% DM] 57.2Acid Detergent fiber:fiber (ADF) [% DM] 15.0Acid Detergent Lignin (ADL) [% DM] 4.5Nitrogen free Extracts (NfE)* [% DM] 39.6Ash [% DM] 12.1Composite fraction (Xc)* [kgCOD/m3] 488.59Protein content stoichiometric factor(fPr_Xc)* [�] 0.21Lipids content stoichiometric factor(fLi_Xc)* [�] 0.04Carbohydrates content stoichiometric factor(fCh_Xc)* [�] 0.54Inerts content stoichiometric factor(fXi_Xc)* [�] 0.21

* Calculated value.

Table 3Degradability rate and methane content of analyzed substrates.

Substrate Degradability rate of orUnit [%]Source KTBL(2012)

Industrial glycerine 90.00Grass silage 79.00Maize silage 86.40Green weed silage 79.00

In the ADM1, XC is divided during the disintegration phase into car-bohydrates (XCH), proteins (XPR), lipids (XLI) and inert fractions (XI)(Batstone et al., 2002). The disintegration is described by the stoi-chiometric f-factors (e.g. fPr_Xc – protein content), which Kochet al. (2010) determined for instance by the equation:

fPr X�c ¼RP

oDM%

%

� �ð2Þ

However, Koch et al. (2010) included also the d factor, whichidentifies the degradable part of cellulose and hemicellulose, ob-tained from the degradation level (DVS). Since the kinetics of bio-gas production of each substrate were determined by usinginoculums as an initial reactor state, determination of the DVSparameter – by comparing the substances’ organic dry mass beforeand after the anaerobic digestion – was not applicable. Conse-quently, the degradability rate was taken from Association forTechnology and Structures in Agriculture e. V. (Kuratorium fürTechnik und Bauwesen in der Landwirtschaft e. V. (KTBL) 2012)(Table 3), and incorporated into calculations, obtaining values forf -factors. Table 2 presents characteristics of the four substratesand their calculated values. Koch et al. (2010) also examined grasssilage with fodder analysis, and their results are in good correlationwith those from the current study.

2.3. Parameters used in the modeling

For each substrate, a set of four parameters, kinetic constantsdescribing the phases of disintegration (kdis), hydrolysis of carbo-hydrates (khyd_ch), hydrolysis of lipids (khyd_li), and hydrolysis ofproteins (khyd_pr), were calibrated with experimental data, withuse of the optimization tool described in Section 2.4.

2.4. Optimization tool

In order to ensure precise determination of the optimal set ofthe kinetic constants, a numerical optimization algorithm was

silage Maize silage Industrial glycerine Green weed silage

31.1 64.3 26.193.2 47.8 88.910.3 0.5 12.0

5.1 22.9 4.515.5 1.1 30.871.0 2.2 65.033.4 2.5 47.211.6 0.7 3.262.4 23.4 41.6

6.8 52.2 11.1393.27 619.25 306.30

0 0.110 0.010 0.1350 0.055 0.478 0.0510 0.695 0.492 0.6040 0.140 0.020 0.210

ganic mass Methane content Methane content[%] [%]KTBL(2012) Simulation

50.00 49.3353.00 52.3652.00 52.0453.00 48.77

Page 4: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

Table 4Optimized disintegration and hydrolysis kinetic constants according to the downhillsimplex methods algorithm from Nelder and Mead (1965).

Substrate kdisa khyd_ch

b khyd_prc khyd_li

d

Industrial glycerine 1.3236 1.2516 0.0018 0.0086Grass silage 1.7433 0.7366 0.0104 0.0149Maize silage 0.7705 0.6865 0.2446 0.1216Green weed silage 0.8168 0.6659 0.0014 0.0513

a kdis is an acronym of kinetic constant describing disintegration phase.b khyd_ch is an acronym of kinetic constant describing carbohydrates hydrolysis

phase.c khyd_pr is an acronym of kinetic constant describing proteins hydrolysis phase.d khyd_li is an acronym of kinetic constant describing lipids hydrolysis phase.

0

200

400

600

800

1000

1200

1400

1600

0 2 4 6 8 10 12 14 16

Bio

gas

prod

uctio

n [m

l]

Duration [d]

Optimized

Experimental

Default 1

Fig. 1. Comparison of the experimental cumulative biogas production from grasssilage to simulation results, where the ADM1 default values for disintegration andhydrolysis kinetic constants were used (ADM1 values for mesophilic high rateidentified as ‘‘default 1’’ and ADM1 values for solids identified as ‘‘default 2’’), andassessed against simulation results, where new optimized parameters were used.

0

200

400

600

800

1000

1200

1400

0 2 4 6 8 10 12 14 16

Bio

gas

prod

uctio

n [m

l]

Duration [d]

OptimizedExperimentalDefault 1Default 2

Fig. 2. Comparison of the experimental cumulative biogas production from maizesilage to simulation results, where the ADM1 default values for disintegration andhydrolysis kinetic constants were used (ADM1 values for mesophilic high rateidentified as ‘‘default 1’’ and ADM1 values for solids identified as ‘‘default 2’’), andassessed against simulation results, where new optimized parameters were used.

0

100

200

300

400

500

600

700

800

900

0 2 4 6 8 10 12 14 16

Bio

gas

prod

uctio

n [m

l]

Duration [d]

OptimizedExperimentalDefault 1Default 2

Fig. 3. Comparison of the experimental cumulative biogas production from greenweed silage to simulation results, where the ADM1 default values for disintegrationand hydrolysis kinetic constants were used (ADM1 values for mesophilic high rateidentified as ‘‘default 1’’ and ADM1 values for solids identified as ‘‘default 2’’), andassessed against simulation results, where new optimized parameters were used.

0

500

1000

1500

2000

2500

0 2 4 6 8 10 12 14 16

Bio

gas

prod

uctio

n [m

l]

Duration [d]

OptimizedExperimentalDefault 1Default 2

Fig. 4. Comparison of the experimental cumulative biogas production fromindustrial glycerine to simulation results, where the ADM1 default values fordisintegration and hydrolysis kinetic constants were used (ADM1 values formesophilic high rate identified as ‘‘default 1’’ and ADM1 values for solids identifiedas ‘‘default 2’’), and assessed against simulation results, where new optimizedparameters were used.

P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194 191

used to simultaneously fit the four constants, describing the phasesof disintegration (kdis), hydrolysis of carbohydrates (khyd_ch), hydro-lysis of lipids (khyd_li), and hydrolysis of proteins (khyd_pr), for eachsubstrate to gas generation data determined from experiments.

As objective function which needs to minimize the absolute dif-ference between experimental and calculated data, the downhillsimplex methods algorithm from Nelder and Mead (1965) waschosen. This algorithm was already used and appreciated byBatstone et al. (2002) in the ADM1 parameters optimization. Thisalgorithm is implemented as function fminsearch in MATLAB(The MathWorks, Inc., 2011).

For each substrate, the four adjustable parameters (kdis, khyd_ch,khyd_li, khyd_pr) were simultaneously fitted. Data from Table 1 wereused as starting values for the optimization run. In order to ensurethat the total biogas yield was adequately represented by theresulting model parameter set, data from the Association for Tech-nology and Structures in Agriculture e. V. (Kuratorium für Technikund Bauwesen in der Landwirtschaft e. V. (KTBL) 2012) were in-cluded in the optimization run by calculating the data given bythe KTBL and introducing them as additional experimental datapoints with a time step of 100 days. The data points were not in-cluded in the graphs for scaling reasons.

Page 5: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

192 P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194

2.5. Optimization procedure

Experimental results obtained for substrates presented in Table 2were used for finding disintegration and hydrolysis kinetic con-stants. Initially, from the substrates’ experimental results, theexperimental result for blanks has been subtracted, this way allow-ing the determination of the kinetic constants for substrates. Subse-quently, composite fraction (XC) and f-factors stated in Table 2 wereused for modeling, together with the ammonium content of inocu-lum. Afterwards, the ADM1 default values for disintegration andhydrolysis kinetic constants were tested, where the ADM1 valuesfor high rate (Table 1) identified as ‘‘default 1’’and ADM1 valuesfor solids (Table 1) identified as ‘‘default 2’’ were used. Afterwardsthe optimization tool was used for determining optimal sets.

Fig. 5. Disintegration (kdis) and hydrolysis (khyd) kinetic constants sensitivity analysissimulated and experimental results.

3. Results and discussion

3.1. Characterization of the reactors initial state

The inoculums from the EWE Wittmund biogas power plantcontained 4.72% dry matter (DM) of which 69.8% was organic drymatter. The ammonium content was 3.069 g/L and the pH was7.8. The total volatile fatty acids/alkalinity ratio (FOS/TAC ratio)was 0.196 and therefore it was expected that inoculums from theEWE Wittmund biogas power plant would support efficient fer-mentation and stable operation since the FOS/TAC ratio was below0.3 (Rieger and Weiland. 2006). Additionally, the ammonium con-tent measured, was used as a value for the ammonium fraction(SNH4) in the ADM1xp model.

for maize silage, where the lowest error represents the best correlation between

Page 6: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194 193

3.2. Batch experiments and simulation

The new kinetic constants, describing the phases of disintegra-tion (kdis), hydrolysis of carbohydrates (khyd_ch), hydrolysis of lipids(khyd_li), and hydrolysis of proteins (khyd_pr), were determined byusing the optimization tool, and are presented in Table 4. Findingoptimal sets of disintegration and hydrolysis kinetic constants forcomplex substances characterized by the Weender analysis/vanSoest method (Figs. 1–4) illustrates that the goal of a more precisedescription of the anaerobic degradation kinetics was achieved.The ADM1 default values for disintegration and hydrolysis kineticconstants used for comparison, where the ADM1 values for highrate (Table 1) identified as ‘‘default 1’’ and ADM1 values for solids(Table 1) identified as ‘‘default 2’’. The results show a very goodcorrelation between experimental and simulation results, afteroptimization of the kinetic constants, and by keeping most of theparameters and fractions available in ADM1 unchanged. In addi-tion, the methane content results also indicate a very good correla-tion between the simulation and expected results (Table 3). Thisapplies especially for maize silage where the simulation fits theexperimental results perfectly. However, the result for glycerinedemonstrates that the model needs further improvement. Addi-tionally, the optimized values of disintegration and hydrolysis ki-netic constants are in accordance with those in the literature.Heukelekian (1958) has already stated that proteins are hydro-lyzed slower than carbohydrates, and those findings were con-firmed by Gavala et al. (2003). Moreover, Chirst et al. (2000) alsoproposed a range of kinetic constants (Table 1), and values for pro-teins and lipids are also lower than those for carbohydrates. Thefaster hydrolysis of lipids than of proteins was also confirmed byBischofsberger et al. (2005). Despite the fact that the value of kdis

is bigger than the values for hydrolysis for all analyzed substrates,they cannot be neglected. Since the values of kdis and khyd_ch are ofthe same order of magnitude, the relation between both values isimportant for the gas generation rate. Only if kdis is a lot faster thanthe value of khyd, does khyd not play a role, and this level has notbeen reached.

Based on the results, further research will focus on verificationof this approach for a large-scale biogas power plant (above3000 m3 digester volume).

3.3. Sensitivity analysis

Using a mathematical solver like the downhill simplex methodsalgorithm from Nelder and Mead (1965) does always mean thatthe final values are a ‘‘random output’’, and so there could be indef-inite pairs of kinetic constants giving a satisfying fitting. Therefore,a sensitivity analysis was performed, in order to verify correctnessof the determined parameters. As a result, the accuracy of the opti-mization’s output is confirmed by the three-dimensional graphs. InFig. 5 the graph for maize silage is presented, and the other threegraphs are included as Supplemental data. Additionally, two-pointcharts representing the proceedings of the optimization tool,where the starting points were chosen to be a boundary values,are included as a Fig. 5.

4. Conclusion

Commonly used substrates at biogas power plants were charac-terized and the results were transferred into the ADM1 simulationenvironment. New kinetic constants for disintegration and hydro-lysis phases were determined via the simplex algorithm from Nel-der and Mead, using other parameters and fraction’s default values.The obtained results indicate that the ADM1, with Wett et al.(2006) modification, is capable of simulating biogas production

from agricultural and industrial substrates, after precise character-ization of the substrates and adjustment of the kinetic constants.Further research will focus on broadening the database and testingthe transferability to industrial-sized biogas plants.

Acknowledgements

Funding for this study was provided by German Federal Minis-try for Education and Research (BMBF): project FKZ 17N1710. Mrs.M. Beyer from EWE Wittmund biogas power plant is acknowledgedfor her time, support, and providing us with substrates and data.We thank Michael Ogurek from IFAK for his kind assistance. Addi-tionally, discussions with Prof. Dr. M. Schlaak, Prof. Dr. E. Siefertand Dipl. Ing. Ingo Stein from EUTEC Institute, Hochschule Em-den/Leer, along with Dr. K. Koch from Institute of Water QualityControl, at Technische Universität München are highlyappreciated.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2012.09.128.

References

Amon, T., Amon, B., Kryvoruchko, V., Zollitsch, W., Mayer, K., Gruber, L., 2007. Biogasproduction from maize and dairy manure – Influence of biomass compositionon the methane yield. Agriculture, Ecosystems & Environment 118 (1–4), 173–182.

Angelidaki, I., Madalena, A., Bolzonella, D., Borzacconi, L., Campos, L., Guwy, A.,Jenicek, P., Kalyuzhnui, S., Van Lier, J., 2007. Anaerobic Biodegradation, Activityand Inhibition (ABAI). Task Group Meeting, 9th and 10th October 2006, inPrague. Institute of Environmental & Resources, Technical University ofDenmark, Kgs., Lyngby, ISBN 978-87-91855-44-3.

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 ModelNo.1. International Water Association, London.

Bischofsberger, W., Dichtl, N., Rosenwinkel, K.-H., Seyfried, C.-F., Böhnke, B. (Eds.),. Anaerobtechnik. Springer, ISBN 3-540-06850-3.

Christ, O., Wilderer, P.A., Angerhofer, R., Faulstich, M., 2000. Mathematical modelingof the hydrolysis of anaerobic processes. Water Science & Technology 41 (3),61–65.

Directive 2009/28/EC, 2009. Directive of the European Parliament and of the Councilof 23 April 2009 on the promotion of the use of energy from renewable sourcesand amending the subsequently repealing Directives 2001/77/EC and 2003/30/EC. Official Journal of The European Union.

Feng, Y., Behrendt, J., Wendland, C., Otterpohl, R., 2006. Parameter analysis of theIWA Anaerobic Digestion Model No.1 for the anaerobic digestion of blackwaterwith kitchen refuse. Water Science & Technology 54 (4), 139–147.

Flotats, X., Palatsi, J., Ahring, B.K., Angelidaki, I., 2006. Identifiability study of theproteins degradation model, based on ADM1, using simultaneous batchexperiments. Water Science & Technology 54 (4), 31–39.

Gali, A., Benabdallah, S., Astals, J., Mata-Alvarez, J., 2009. Modified version of ADM1model for agro-waste application. Bioresource Technology 100, 2783–2790.

Garcia-Heras, J.L., 2003. Reactor sizing, process kinetics and modelling of anaerobicdigestion of complex wastes. In: Mata-Alvarez, J. (Ed.), Biomethanization of theOrganic Fraction of Municipal Solid Wastes. International Water Association,London.

Gavala, H.N., Angelidaki, I., Ahring, B.K., 2003. Kinetics and modeling of anaerobicdigestion process. In: Scheper, T., Ahring, B.K. (Eds.), Biomethanization I.Springer-Verlag, Berlin, Heidelberg.

Heukelekian, H., 1958. Basic principles of sludge digestion. In: McCabe, J.,Eckenfelder, W.W. (Eds.), Biological Treatment of Sewage and IndustrialWastes. Reinhold, New York.

Hu, Z., Yu, H., 2005. Application of rumen microorganisms for enhanced anaerobicfermentation of corn stover. Process Biochemistry 40 (7), 2371–2377.

IFAK, 2005. SIMBA� 5.1 User’s Guide. ifak system GmbH, Magdeburg.Koch, K., Lübken, M., Gehring, T., Wichern, M., Horn, H., 2010. Biogas from grass

silage – measurements and modeling with ADM1. Bioresource Technology 101,8158–8165.

Koch, K., Wichern, M., Lübken, M., Horn, H., 2009. Mono fermentation of grass silageby means of loop reactors. Bioresource Technology 100, 5934–5940.

Lehtomaki, A., Vavilin, V.A., Rintala, J.A., 2005. Kinetic analysis of methaneproduction from energy crops, in: Ahring, B.K., Hartmann, H. (Eds.),Proceedings of the Fourth International Symposium on Anaerobic Digestion ofSolid Waste, vol. 2. Copenhagen, Denmark, pp. 67-72.

Page 7: Application of Anaerobic Digestion Model No. 1 for describing anaerobic digestion of grass, maize, green weed silage, and industrial glycerine

194 P. Biernacki et al. / Bioresource Technology 127 (2013) 188–194

Lübken, M., Gehring, T., Wichern, M., 2010. Microbiological fermentation oflignocellulosic biomass: current state and prospects of mathematicalmodeling. Applied Microbiology and Biotechnology 85 (6), 1643–1652.

Lübken, M., Wichern, M., Schlattmann, M., Gronauer, A., 2007. Modelling the energybalance of an anaerobic digester fed with cattle manure and renewable energycrops. Water Research 41, 4085–4096.

The MathWorks, Inc., 2011. MATLAB� – Optimization ToolboxTM 6, User’s Guide.The MathWorks, Inc., Natick, MA, USA.

Myint, M., Nirmalakhandan, N., Speece, R.E., 2007. Anaerobic fermentation of cattlemanure: modelling of hydrolysis and acidogenesis. Water Research 41 (2), 323–332.

Naumann, C., Bassler, R., 1993. Die Chemische Untersuchung von Futtermitteln.VDLUFA-Verlag, Darmstadt.

Nelder, J.A., Mead, R., 1965. A simple Method for Function Minimization. ComputerJournal 7, 308–313.

Rieger, C., Weiland, P., 2006. Prozessstörungen frühzeitig erkennen. Biogas Journal4, 18–20.

Schoen, M., Sperl, D., Gadermaier, M., Goberna, M., Franke-Whittle, I., Insam, H.,Ablinger, J., Wett, B., 2009. Population dynamics at digester overload conditions.Bioresource Technology 100 (23), 5648–5655.

van Soest, P.J., Wine, R.H., 1967. Use of detergents in the analysis of fibrous feeds IV.Determination of plant cell-wall constituents.. Journal of Association of OfficialAnalytical Chemists 28, 50–59.

Verein Deutscher Ingenieure (VDI), 2006. Fermentation of Organic Material.Characterization of the Substrate, Sampling, Collection of Material Data,Fermentation Tests. Verein Deutscher Ingenieure, Düsseldorf.

Vavilin, V.A., Fernandez, B., Palatsi, J., Flotats, X., 2008. Hydrolysis kinetics inanaerobic degradation of particulate organic material: an overview. WasteManagement 28, 939–951.

Wett, B., Eladawy, A., Ogurek, M., 2006. Description of nitrogen incorporation andrelease in ADM1. Water Science & Technology 54 (4), 67–76.

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 modelling with ADM1. BioresourceTechnology 100 (4), 1675–1681.

Wichern, M., Lübken, M., Schlattmann, M., Gronauer, A., Horn, H., 2008.Investigations and mathematical simulation on decentralized anaerobictreatment of agricultural substrates from livestock farming. Water Science &Technology 58 (1), 67–72.