A review of ADM1 extensions, applications, and

analysis: 20022005
D.J. Batstone*, J. Keller** and J.P. Steyer*
*Environment and Resources, Building 113, Technical University of Denmark Lyngby 2800, Denmark
(E-mail: djb@er.dtu.dk)
**Advanced Wastewater Management Centre, The University of Queensland, St Lucia, 4058, Australia
Abstract Since publication of the Scientic and Technical Report (STR) describing the ADM1, the model
has been extensively used, and analysed in both academic and practical applications. Adoption of the ADM1
in popular systems analysis tools such as the new wastewater benchmark (BSM2), and its use as a virtual
industrial system can stimulate modelling of anaerobic processes by researchers and practitioners outside
the core expertise of anaerobic processes. It has been used as a default structural element that allows
researchers to concentrate on new extensions such as sulfate reduction, and new applications such as
distributed parameter modelling of biolms. The key limitations for anaerobic modelling originally identied in
the STR were: (i) regulation of products from glucose fermentation, (ii) parameter values, and variability, and
(iii) specic extensions. Parameter analysis has been widespread, and some detailed extensions have been
developed (e.g., sulfate reduction). A veried extension that describes regulation of products from glucose
fermentation is still limited, though there are promising fundamental approaches. This is a critical issue, given
the current interest in renewable hydrogen production from carbohydrate-type waste. Critical analysis of the
model has mainly focused on model structure reduction, hydrogen inhibition functions, and the default
parameter set recommended in the STR. This default parameter set has largely been veried as a
reasonable compromise, especially for wastewater sludge digestion. One criticism of note is that the ADM1
stoichiometry focuses on catabolism rather than anabolism. This means that inorganic carbon can be used
unrealistically as a carbon source during some anabolic reactions. Advances and novel applications have
also been made in the present issue, which focuses on the ADM1. These papers also explore a number of
novel areas not originally envisaged in this review.
Keywords ADM1; anaerobic; model; review
Introduction
It is undeniable that interest, and activity in academic and applied anaerobic digestion
simulation is rapidly developing. There have been some 750 publications regarding
anaerobic digestion modelling in the last 30 years, of which half have been in the last 5
years (Figure 1). Much of this work has been a driver for development of a standardised
anaerobic model.
The ADM1 was rst presented at the 9th IWA Conference on Anaerobic Digestion in
2001 (AD2001) in Antwerp (Batstone et al., 2002a), and the related Scientic and Tech-
nical Report (STR) was published by IWA in early 2002 (Batstone et al., 2002b).
Between then, and mid-2005 there have been 30 ISI citations of the report. Additionally,
the model itself is available in popular aquatic modelling packages such as WEST, GPS-
X, Matlab Simulink, SIMBA, and Aquasim with cross-verication between the
implementations. The two most widely used implementations are the Aquasim 2.1
implementation, created by the ADM1 Taskgroup, and the Matlab implementation, in
Matlab C, and Matlab script (Rosen and Jeppsson, 2002; Rosen et al., 2006). These will
not be discussed widely here, but each implementation needed specic numerical tech-
niques to overcome limitations, and were cross-veried against each other. The
W
a
t
e
r
S
c
i
e
n
c
e
&
T
e
c
h
n
o
l
o
g
y
V
o
l
5
4
N
o
4
p
p
1

1
0
Q
I
W
A
P
u
b
l
i
s
h
i
n
g
2
0
0
6
1 doi: 10.2166/wst.2006.520
availability of the code has helped considerably in distribution and use of the model, as
implementation of a complex bioprocess model such as the ADM1 is a daunting task and
can lead easily to errors that are difcult to identify and correct.
The original goals of publication of the ADM1, as stated in the STR were:
increased model application for full-scale plant design, operation and optimisation;
further development work on process optimisation and control, aimed at direct
implementation in full-scale plants;
common basis for further model development and validation studies to make outcomes
more comparable and compatible;
assisting technology transfer from research to industry.
The key objective of the ADM1 STR was to enable a shift in focus from development
of model fundamentals (e.g. basis, kinetics etc.) to investigation of specic aspects and
model application in industry situations. The Taskgroup identied, in the STR, a number
of key limitations in anaerobic digestion modeling (as of 2002). The main limitations
were: (i) models for regulation of VFA products from glucose are incomplete, and have
not been validated: (ii) uncertainty in parameter values, and parameter variation; and (iii)
lack of sufcient understanding of a number of related processes, such as sulfate
reduction, precipitation, etc.
The objective of the present paper is to review extensions, applications, and critical
analysis of the ADM1, especially in relation to whether the publication of a standardised
model has achieved these goals. In addition, we attempt to assess future requirements for
standardised anaerobic process modelling. This review is part of a special issue on the
ADM1, and is mainly intended to establish current knowledge. Therefore the other papers
in this issue are not specically addressed, except where they address critical limitations.
Applications
Published applications can be divided into two classes: applications of the standard
model in a mixed tank, with the recommended approach given in the STR, often using
already implemented models, to assess specic systems and; applications of the ADM1
in new distributed parameter applications, often for theoretical analysis.
Mixed tank applications
Many of the papers cited in other sections of this review include mixed tank implemen-
tations in systems that include mixed sludge digesters, manure digesters, and UASB
Figure 1 ISI publications regarding anaerobic digestion modelling per year since 1972. Note that some
papers from 2004/2005 have not yet been indexed
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
2
systems. However, there are two applications of the ADM1 in particular that are likely to
receive widespread use. The COST wastewater activated sludge benchmark (BSM1) was
produced by COST Action 624 and COST Action 682 for standardised comparison of
control and operating strategies. It describes a standardised wastewater treatment plant
with 2 anoxic tanks, 3 aerated tanks, and a secondary settler. The description included
the full model with all parameter values, and detailed input data. Reactions were
described using the IWA Activated Sludge Model Number 1 (ASM1; Henze et al.,
1987). The BSM1 was a success, resulting in 100 publications worldwide (Jeppsson et al.,
2006), with use not only in the core area of plant-wide control and optimisation, but also
as a teaching, training, and theoretical analysis tool.
A new benchmark (i.e. BSM2) has been proposed (Jeppsson et al., 2006) that also
considers a primary settler, thickener, and anaerobic digester, modelled using the ADM1
(Figure 2). This has been one of the drivers for cross-verication of the different versions
of the ADM1, as well as implementation of the reference versions in Matlab, WEST, For-
tran, GPS-X, and SIMBA. It has also resulted in detailed critical analysis of some con-
cepts proposed in the STR. These include discovery of an ammonia leak during the
disintegration process, and much improved interfacing of the ADM1, and ASM series
models (Copp et al., 2003; Vanrolleghem et al., 2005).
The ADM1 was also used within the TELEMAC project, a 5th Frame programme EU
project for improved remote monitoring, control, and operation of high-rate anaerobic
wastewater treatment plants. One interesting subproject is production of a model-based
operator training tool (Bernard et al., 2005). The ADM1 is sufciently complex such that
it can predict complex overload behaviour, which is a necessary requirement for operator
training. Additionally, the ADM1 was used as a virtual plant to evaluate different control
strategies and software sensors for advanced monitoring of industrial anaerobic digesters.
The ADM1 has also been used to simulate and analyse the response and behaviour of an
anaerobic respirometer (Mosche and Meyer, 2003).
Distributed parameter modelling. The three key distributed parameter systems of interest
are biolm systems, plug ow reactors (including UASB reactors), and municipal solid
waste reactors. Biolm systems have been modelled using the ADM1 in one dimension (1D)
Figure 2 Plant layout for the proposed BSM2 from Jeppsson et al. (2006), used with permission
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
3
(Batstone et al., 2004a), and 2/3D (Picioreanu et al., 2005). The second application
demonstrates the power of current computing, as the full ADM1, with pH solution can be
simulated in 10
5
discrete elements over several days, with single processor computers. An
image of the simulated granule from the 2D model is shown in Figure 3.
A plug ow ADM1 implementation was also used to simulate a laboratory-scale
UASB (Batstone et al., 2005), against experimental data. The model could effective
simulate acetate conversion rates against measured data, and demonstrated the higher
conversion efciencies of a plug-ow system when compared to mixed reactors.
The ADM1 is not widely used for distributed parameter modelling of solid-phase
anaerobic digestion of solid waste. However, Vavilin and Angelidaki (2005) used a sim-
plied model to investigate the impact of separation in space of methanogenic and acido-
genic groups in solid phase batch digesters (VFA was assumed to inhibit methanogens)
and the same implementation can be readily applied to the ADM1.
Extensions
The ADM1 was designed to be readily extendible, and additional functions work very
well, and are easy to document. The most commonly requested extensions are sulfate
reduction, description of the behaviour of phosphorous, and mineral precipitation. A key
limitation in the existing model is regulation of products from glucose fermentation, as
recognised in the STR.
Sulfate reduction is the most complex of these, as sulfate acts as electron acceptor for
oxidation of VFAs such as propionate, butyrate and acetate, but reacts preferably with
hydrogen (H
2
) formed in the fermentation. Additionally, the product, sulde, is inhibi-
tory, is ionic (and hence affects pH), and is gaseous. Therefore to have a comprehensive
model of sulfate reduction, every part of the ADM1 needs modication. A very extensive
sulfate reduction extension was published by Fedorovich et al. (2003). This involves four
Figure 3 Granule simulated using the ADM1 in a 2D model after 31.25 days using 10
5
discrete elements.
From Picioreanu et al. (2005), used with permission. Subscribers to the online version of Water Science
and Technology can access the colour version of this gure from http://www.iwaponline.com/wst
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
4
additional groups of microbes that can oxidise butyrate/valerate, propionate, acetate and
hydrogen respectively, with sulfate as electron acceptor to produce hydrogen sulde.
Inhibition, acid-base chemistry, and gas stripping were also included. The new microbes
compete with those currently in the ADM1 for these substrates. The model was effective
in predicting the behaviour of a UASB fed with sulfate up to 6 gSL
21
. It is also possible
to use a reduced model, with a single group of sulfate reducers competing with
hydrogen utilisers, but this is effective only with inuent S:COD ratios up to
0.1 gS gCOD
21
(Batstone, 2006). Above this level, the sulfate reducers will oxidise
VFAs and the model will incorrectly predict sulfate in the efuent. Sulde is an import-
ant contributor to odour, and a number of other anaerobic byproducts contribute to odour.
This has been further explored in this issue.
Phosphorous has been requested in order to close phosphorous balances in integrated
wastewater treatment plant modelling and particularly to account for the signicant phos-
phorus retention and release in the biomass during digestion of activated sludge from
domestic wastewater treatment plants. Papers that have described an extension for this
have not been found, but it is a relatively simple extension. Soluble phosphorous can be
described by a single state, and organic (or inorganic) bound phosphorous can be
described by a separate state, as with nitrogen in the ASM1, or bound to complex particu-
lates (X
c
), as with nitrogen in the ADM1. If required, acid-base chemistry for phosphor-
ous can be implemented and using this, precipitation of relevant complexes may also be
included (CaHPO
4
, struvite {MgNH
4
PO
4
} or others). The main difculty in this regard is
not inclusion of individual processes, but the large number of processes. There are a huge
number of potential, (and maybe unknown) precipitants (Wild et al., 1996; Wild et al.,
1997), and release mechanisms are complex depending on upstream phosphorous removal
methods. Currently, there is likely insufcient fundamental knowledge available to enable
an ab initio prediction, but often data from existing, similar processes can be used to
approximate the expected solubilisation.
Single component precipitation extensions are relatively simple. The fully dissociated
anion needs to be modelled as a separate state (normally CO
3
2
), either by algebraic, or
differential equations, and the cation and precipitate need to be modelled as differential
states. Precipitation can be based either on equilibrium (using a K
SO
), or a rst-order (or
more complicated) reaction. An extension for CaCO
3
precipitation has been published
(Batstone and Keller, 2003), and can be used as a template for other precipitation
reactions.
Other extensions have focused on specic issues. For example, ethanol oxidation was
included to simulate the degradation of winery wastewater in an Anaerobic Sequencing
Batch Reactor (ASBR), and was found to have similar kinetic rates to propionate and
butyrate oxidation (Figure 4, Batstone et al., 2004b). Isovalerate was found to degrade to
acetate only, in comparison with normal-valerate, which degrades to propionate and acet-
ate, and an extension was written to address this (Batstone et al., 2003).
Despite strong interest, acidogenesis of glucose is still a key limitation in the existing
model. It is known that environmental factors, including pH and hydrogen concentration
regulate the production of products, including butyrate, propionate, acetate, ethanol, and
hydrogen (Mosey, 1983; Costello et al., 1991). However, there is still not a good model
available of this regulation as is becoming clearly evident from the results of many pro-
cesses aiming to achieve hydrogen production. Currently, the ADM1 has xed stoichio-
metric parameters for glucose fermentation products, which is clearly not representative
of the actual acidogenic processes in most systems. This is especially critical, given the
rapidly raising interest in biohydrogen production where (mono)saccharides are the most
common source material, and the hydrogen production optimisation is the main focus of
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
5
these processes. An interesting metabolic model has been proposed recently (Rodriguez
et al., 2004), but needs experimental validation. Further evaluation of this is presented in
Rodriguez et al. (2006).
Critical analysis and modications
Model structure and basis. Most of the criticisms of the ADM1 have been that it is too
simple, or too complex. Simplicity can be addressed by adding relevant extensions to the
model (where feasible), and on the other hand researchers have also used the ADM1 as a
basis for more simplied models. Model reduction usually aims to decrease simulation
time (especially in integrated systems modelling), decrease parameter estimation
requirements, or decrease implementation workload. The Siegrist model (Siegrist et al.,
2002) is a slight simplication of the ADM1, orientated towards primary sludge
digestion. It removes the butyrate and valerate components but has a good basis for both
thermophilic and mesophilic sludge digestion. Elmitwalli et al. (2003) used the ADM1 as
a basis for a two-step model to assess the impact of temperature on anaerobic treatment
of domestic sewage. Batstone (2006) has recommended a two-step model similar to that
of Elmitwalli et al. (2003) for simulation of hydraulics in UASB reactors.
Copp et al. (2004) reduced the ADM1 substantially for an alternative implementation
in GPS-X. The changes included lumping all soluble components more complex than
acetate as soluble COD, and separating nitrogen states from COD states, in a similar way
to the ASM1. The changes improved simulation speed by a factor of 5. However, while
model reduction is valid, and often recommended, it should be recognised that it
decreases specic capabilities. For example, the reduced GPS-X model will not effec-
tively simulate hydrogen inhibition of propionate or butyrate oxidation, and will incor-
rectly predict pH during propionate or butyrate accumulation.
Another important limitation was recognised by Kleerebezem and van Loosdrecht
(2006), in that stoichiometry is based only on catabolic reactions. Anabolic reactions
source COD from the substrate, but in cases of carbon limitation, they source excess car-
bon from carbon dioxide. This is unrealistic for most organisms (except hydrogen utili-
sers), especially those that do not produce bicarbonate (e.g. butyrate oxidisers). Because
biomass yields in general are relatively small, and because most reactions produce carbon
dioxide during catabolism, the overall impact is small. However, when the ADM1 is used
Figure 4 Measured values for propionate ( ), acetate (S) and residual COD (soluble COD-approximate
inert COD-organic acid COD) (o), and simulated data (solid lines as marked?) for ethanol, propionate, and
acetate over a single cycle fed with winery wastewater in an ASBR (Batstone et al., 2004b), used with
permission
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
6
as a basis for assessing metabolic products, the updated basis proposed by Kleerebezem
and van Loosdrecht (2006) should be used.
Kinetics. Critical analysis of kinetics has largely focused on hydrogen inhibition. The
STR recognised that thermodynamics played a critical role in inhibition, but that
diffusion limitations within biolms may allow reactions to occur, that seem
thermodynamically impossible from concentrations in the bulk liquid. Therefore a simple
non-competitive function was chosen. Zheng and Bagley (2006) recommended
replacement of the non-competitive function with a function dependent on the actual free
energy (rate is negative if free energy is positive or zero). The results were veried using
data from an anaerobic sequencing batch reactor (ASBR). Kleerebezem and van
Loosdrecht (2006) recommended also using the hydrogen ion as a driver for the reaction.
Both papers are essentially based on the work of Hoh and CordRuwisch (1996). While
the inhibition function is certainly valid when the model observes in-situ hydrogen
concentrations, and elegant (no inhibition constant is required), the original limitations
identied in the STR relating to diffusion in biolm or occular systems remain.
Additionally, the thermodynamic functions are discontinuous at DG
0
0, which can
cause considerable problems when using implicit differential equation (stiff) solvers.
Continuous approximations of this function also cause integration problems.
Parameter values. The recommended parameter set in the ADM1 is based on low
decay rates (0.010.02 d
21
). This reects the consensus of previously published decay
rates. However, decay, and uptake (and hence growth) rates are heavily correlated
(Batstone et al. 2003), and it is possible to increase the decay rate while simultaneously
increasing the uptake rate to have zero net impact on outputs. In parallel to the
publication of the ADM1, the Siegrist model was also presented (Siegrist et al., 2002).
Siegrist et al. (2002) based parameter values on batch experiments, with decay rates on
the order of 0.05 d
-1
. An example of the two models is shown in Figure 5, with k
m
(kgCODS kgCODX
21
d
21
) and K
S
(kgCODm
23
) of the ADM1 set to 6.6, 0.030 (acetate
utilisers), and 9.0, 0.029 (propionate utilisers), respectively. In this case, the higher decay
rates used by Siegrist et al. (2002) offset the higher K
S
values. Indeed, there is
independent indication from biolm modelling (Batstone et al., 2004a) that higher decay
rates may be more valid.
There has been extended analysis of the applicability of the recommended parameter
set for digestion of sludge from domestic sewage treatment plants. Blumensaat and Keller
(2005) applied the ADM1 to simulation of mixed primary and secondary sludge in a two-
stage thermophilic/mesophilic digester. Parameters were optimised, with small (2050%)
increases in uptake and half-saturation coefcients for acetate and propionate compared
Figure 5 A comparison of the ADM1 (solid line), with optimised k
m
, K
S
for acetate and propionate, with the
Siegrist model (Siegrist et al., 2002), with published parameters
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
7
to the recommended parameter set. They also found that the hydrolysis rate under meso-
philic conditions was double that under thermophilic conditions, the opposite to that rec-
ommended in the STR. Parker (2005) applied steady-state predictions to a range of
sewage sludge digesters operating at different retention time. It was found that at short
retention times, the model overpredicted acetate, and underpredicted butyrate and propio-
nate. This could be corrected by two modications: (i) decrease in acetate K
S
, and
increase in propionate/butyrate K
S
, (ii) increasing ammonia inhibition constant, and
decreasing hydrogen inhibition constant. Shang et al. (2005) simulated the steady state of
mixed primary and secondary sludge digesters, and found very good agreement between
model and measured data, but with a 10% overprediction of biogas production. This was
related to overprediction of solids destruction, and is probably related to either the disin-
tegration parameter, or inert fraction in the sludge. In summary, out of the four studies,
two studies (Siegrist et al., 2002; Parker, 2005) found K
S
values effectively lower than
the ADM1 set, one study (Blumensaat and Keller, 2005) found K
S
values effectively
higher than the ADM1 set, and the other did not assess organic acids, but found good
overall prediction. This indicates that the recommended parameter set in the STR is a
reasonable compromise but there may be case-specic adaptations necessary in some
situations.
Parameters for the ADM1 have also been applied in a range of other systems, includ-
ing an ASBR treating winery wastewater (Batstone et al., 2004b), and a thermophilic
manure digester (Batstone et al., 2003). Generally modications to kinetic parameters
have been on the order of 2050%, which is relatively small compared to the large vari-
ations found in some parameters used for activated sludge systems.
Conclusions
Publication of the ADM1 has largely addressed its primary objective of reducing dupli-
cate published model structures. Indeed, no references to new, complex structural models
were found. There has been a considerable amount of effort in producing extensions to
the ADM1, and extensive critical analysis. The most serious criticism is that of Kleerebe-
zem and van Loosdrecht (2006), relating to inorganic carbon use during anabolism. How-
ever, its impact is relatively small in environmental systems, and probably does not
warrant an overhaul of the current model structure and stoichiometry.
Among the goals, most relate to increased model application in full scale, and by tech-
nology developers and users. This has also been achieved, and several of the papers cited
in this review are from these users and we are aware of several other applications in prac-
tice that have not been published or presented in conferences. Application is primarily in
sewage sludge digesters, and hopefully, future use will also involve analysis and optimis-
ation of industrial high-rate digesters, as in this context, modelling can provide real
gains. Publication of the ADM1, and its adoption into the BSM2 has also introduced the
eld of anaerobic modelling to a variety of researchers working in control and optimis-
ation of activated sludge treatment plants, and in the future, the end-users of this
benchmark.
Of the three core limitations regulation of glucose fermentation products; parameter
values; and key extensions the last two are being well addressed by subsequent publi-
cations. Additional work is needed on parameter variability across industries, and specic
extensions. However, the key limitation of glucose fermentation products needs model
development and verication, given the high level of interest in carbohydrate fermenta-
tion technologies for hydrogen production. This should have a fundamental, metabolic
basis, as has been proposed. Such research would be invaluable for interpretation and
optimisation of renewable hydrogen production processes.
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
8
Acknowledgements
We thank the ADM1WS1 scientic committee for reading and commenting on this
manuscript.
References
Batstone, D. (2006). Mathematical modeling of anaerobic reactors treating domestic wastewater: rational
criteria for reactor projects. Reviews in Environmental Science and Biotechnology, 5(1), 5771.
Batstone, D.J. and Keller, J. (2003). Industrial applications of the IWA anaerobic digestion model No. 1
(ADM1). Water Science and Technology, 47(12), 199206.
Batstone, D.J., Hernandez, J.L.A. and Schmidt, J.E. (2005). Hydraulics of laboratory and full-scale UASB
reactors. Biotechnology and Bioengineering, 91(3), 387391.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S., Pavlostathis, S., Rozzi, A., Sanders, W., Siegrist, H.
and Vavilin, V. (2002a). The IWA Anaerobic Digestion Model No1. Water Science and Technology.,
45(10), 6573.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M.,
Siegrist, H. and Vavilin, V.A. (2002b). Anaerobic Digestion Model No. 1 (ADM1), IWA Task Group for
Mathematical Modelling of Anaerobic Digestion Processes, IWA Publishing, London, UK.
Batstone, D.J., Keller, J. and Blackall, L.L. (2004a). The inuence of substrate kinetics on the microbial
community structure in granular anaerobic biomass. Water Research, 38(6), 13901404.
Batstone, D.J., Pind, P.F. and Angelidaki, I. (2003). Kinetics of thermophilic, anaerobic oxidation of straight
and branched chain butyrate and valerate. Biotechnology and Bioengineering, 84(2), 195204.
Batstone, D.J., Torrijos, M.J., Ruiz, C. and Schmidt, J.E. (2004b). Use of an anaerobic sequencing batch reactor
for parameter estimation in modelling of anaerobic digestion. Water Science and Technology, 50(10),
295303.
Bernard, O., Chachnat, B., Helias, A., Le Danteg, B., Sialve, B., et al. (2005). TELEMAC: An integrated
system to remote monitor and control anaerobic watewater treatment plants through the internet. Water
Science and Technology, 52(12), 457464.
Blumensaat, F. and Keller, J. (2005). Modelling of two-stage anaerobic digestion using the IWA Anaerobic
Digestion Model No. 1 (ADM1). Water Research, 39(1), 171183.
Copp, J.B., Jeppsson, U. and Rosen, C. (2003). Towards an ASM1-ADM1 state variable interface for plant-wide
wastewater treatment modelling. In 76th Annual WEF Conference and Exposition, October 1115, Los
Angeles, CA, USA.
Copp, J.B., Peerbolte, A., Snowling, S., Schraa, O., Froelich, D., and Belia, E. (2004) Integrating anaerobic
digestion into plant-wide wastewater treatment modelling experience with data from a large treatment
plant. In Anaerobic Digestion 2004, 10th World Congress on Anaerobic Digestion 2004, Guiot, S. and
Pavlostathis, S.G. (eds), Montreal, Canada, pp. 13621365.
Costello, D.J., Greeneld, P.F. and Lee, P.L. (1991). Dynamic modelling of a single-stage high-rate
anaerobic reactor -I. Model development. Wat. Res., 25(7), 859871.
Elmitwalli, T.A., Sayed, S., Groendijk, L., van Lier, J., Zeeman, G. and Lettinga, G. (2003). Decentralised
treatment of concentrated sewage at low temperature in a two-step anaerobic system: two upow-hybrid
septic tanks. Water Science and Technology, 48(6), 219226.
Fedorovich, V., Lens, P. and Kalyuzhnyi, S. (2003). Extension of Anaerobic Digestion Model No. 1 with
processes of sulfate reduction. Applied Biochemistry and Biotechnology, 109(13), 3345.
Henze, M., Grady, C.P.L., Gujer, W., Marais, G.v.R. and Matsuo, T. (1987). Activated Sludge Model No. 1.
IAWQPRC Scientic and Technical Report. IAWQPRC, London, UK.
Hoh, C. and CordRuwisch, R. (1996). A practical kinetic model that considers endproduct inhibition in
anaerobic digestion processes by including the equilibrium constant. Biotechnology and Bioengineering,
51(5), 597604.
Jeppsson, U., Rosen, C., Alex, J., Copp, J., Gernaey, K.V., Pons, M.N. and Vanrolleghem, P. (2006).
Towards a benchmark simulation model for plant-wide control strategy performance evaluation of
WWTPs. Water Science and Technology, 53(1), 287295.
Kleerebezem, R. and van Loosdrecht, M.C.M. (2006). Critical analysis of some concepts proposed in ADM1.
Water Science and Technology, 54(4), 5157.
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
9
Mosche, M. and Meyer, U. (2003). Factors affecting constancy of acetate concentration and correct
determination of methanogenic activity in pH-stat experiments. Water Science and Technology, 48(6),
111118.
Mosey, F.E. (1983). Mathematical modelling of the anaerobic digestion process: Regulatory mechanisms for
the formation of short-chain volatile acids from glucose. Water Science and Technology, 15, 209232.
Parker, W.J. (2005). Application of the ADM1 model to advanced anaerobic digestion. Bioresource
Technology (in Press).
Picioreanu, C., Batstone, D.J. and van Loosdrecht, M.C.M. (2005). Multidimensional modelling of anaerobic
granules. Water Science and Technology, 52(12), 501507.
Rodriguez, J., Kleerebezem, R., Lema, J.M. and van Loosdrecht, M.C.M. (2004). A promising approach for
modelling product formation in mixed culture fermentations. In Anaerobic Digestion 2004, 10th World
Congress on Anaerobic Digestion 2004, Guiot, S. and Pavlostathis, S.G. (eds), Montreal, Canada, pp.
14001405.
Rodriguez, J., Lema, J.M., van Loosdrecht, M.C.M. and Kleerebezem, R. (2006). Variable stoichiometry with
thermodynamic control in ADM 1. Water Science and Technology, 54(4), 101110.
Rosen, C. and Jeppsson, U. (2002). Anaerobic COST Benchmark Model Description: Version 1.2.
Department of Industrial Electrical Engineering and Automation, University of Lund, Lund, Sweden.
Rosen, C., Vrecko, D., Gernacy, K.V., Pons, M.N. and Jeppsson, U. (2006). Implementing ADM1 for plant-
wide benchmark simulation in Matlab/Simulink. Water Science and Technology, 54(4), 1119.
Shang, Y., Johnson, B.R. and Sieger, R. (2005). Application of the IWA Anaerobic Digestion Model
(ADM1) for simulatiing full-scale anaerobic sewage sludge digestion. Water Science and Technology,
52(12), 487492.
Siegrist, H., Vogt, D., Garcia-Heras, J. and Gujer, W. (2002). Mathematical model for meso and thermophilic
anaerobic sewage sludge digestion. Environmental Science and Technology, 36, 11131123.
Vanrolleghem, P., Rosen, C., Zaher, U., Copp, J., Benedetti, L., Ayesa, E. and Jeppsson, U. (2005).
Continuity-based interfacing of models for wastewater systems described by Peterson matrices. Water
Science and Technology, 52(12), 493500.
Vavilin, V.A. and Angelidaki, I. (2005). Anaerobic degradation of solid material: Importance of initiation
centers for methanogenesis, mixing intensity, and 2D distributed model. Biotechnology and
Bioengineering, 89(1), 113122.
Wild, D., Kisliakova, A. and Siegrist, H. (1996). P-xation by Mg, Ca and zeolite A during stabilization of
excess sludge from enhanced biological P-removal. Water Science and Technology, 34(12), 391398.
Wild, D., Kisliakova, A. and Siegrist, H. (1997). Prediction of recycle phosphorus loads from anaerobic
digestion. Water Research, 31(9), 23002308.
Zheng, Y. and Bagley, D.M. (2004). Improved thermodynamic inhibition and regulatory functions for
modeling anaerobic processes. In Anaerobic Digestion 2004, 10th World Congress on Anaerobic
Digestion 2004, Guiot, S. and Pavlostathis, S.G. (eds), Montreal, Canada, pp. 174179.
D
.
J
.
B
a
t
s
t
o
n
e
e
t
a
l
.
10