SIMULATION OF FLOW, MASS TRANSFER AND BIO-CHEMICAL REACTIONS IN

ANAEROBIC DIGESTION

By
LIANG YU

A dissertation submitted in partial fulfillment of

the requirements for the degree of
DOCTOR OF PHILOSOPHY
WASHINGTON STATE UNIVERSITY
Department of Biological Systems Engineering
DECEMBER 2012

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of LIANG YU
find it satisfactory and recommend that it be accepted.

JJJJjkkkk

Shulin Chen, Ph.D., Chair

JJJJj Robt Dillon Manuel Garcia-Perez, Ph.D.

JJJJj Robt Dillon Dillon Robert, Ph.D.

JJJJjkkkk

ii

Konstantin Matveev, Ph.D.

D

ACKNOWLEDGEMENTS
I would like to expess my sincere thanks to my advisor, Dr. Shulin Chen, for allowing me to
pursue this study at the Bioprocessing & Bioproducts Enigeering Laboratory (BBEL). I am
extremely thankful for his personal guidance, assistance and supervision over the course of this
study. My commetee members: Dr. Manuel Garcia-Perez, Dr. Dillon Robert and Dr. Konstantin
Matveev are also gratefully acknowledged for their valuable comments and suggestions during
the dissertation process.
I warmly thank Dr. Craig Frear for his valuable advice and friendly help on my dissertation
work. My warm thanks are also due to Jonathan Lomber, Cynthia Alwine for their kind help in
lab data analysis. I appreciate AD group members: Jingwei Ma, Quanbao Zhao, Baisuo Zhao,
and Nicholas Paul Kennedy, the former members and all BBEL members for their help and
cooperation. Thanks go to John Anderson, Patricia Huggins, Pat King, and Joan Hagedorn for
their administrative support. I extend my thanks to all those who have helped me with my work
in the Department of Biological Systems Engineering at Washington State University.
I also want to dedicate this dissertation to my parents, although they can not see it anymore.
Finally, I thank my wife Dongming Zhao for giving up her claim on my time for these past
four years. My gratitude also goes to my brother Jiang Yu for his support on my Professional
Engineer (PE) Exam.

iii

SIMULATION OF FLOW, MASS TRANSFER AND BIO-CHEMICAL REACTIONS IN

ANAEROBIC DIGESTION

ABSTRACT

By LIANG YU, Ph.D.

Washington State University
DECEMBER 2012
Chair: Dr. Shulin Chen
The aims of the dissertation were to understand and simulate flow, mass transfer and
bio-chemical reactions in anaerobic digesters. Computational fluid dynamics (CFD) and
anaerobic digestion model No. 1 (ADM1) were used to assist the development of a new high
solid anaerobic digestion system. The kinetic theory of granular flow (KTGF) was introduced
and a multi-fluid model was developed to describe the phenomena of settling and suspension in
the anaerobic digestor which is critical to increase biomass retention and improve digestion
performance.
To assess flow fields in a HSAD, a mechanical mixing model was constructed to predict
flow characteristics. The model results show impeller A-310 was not suitable for agitating a high
solid digester due to its impaired mixing performance and large shear rate. The helical ribbon
had a better potential to be more suitable for the mixing of high solids digesters than the Impeller
A-30. Its low shear environment is suitable for microbial flocs in anaerobic digestion. Through

iv

optimization, the power of the helical ribbon used in high solids digesters can be reduced
significantly.
A comprehensive model based on ADM1 was developed to describe a new two-stage HSAD
system. The predictions indicate that a high rate HSAD system could be achieved by adjusting
major parameters such as pH, recycled methanogenic bacteria, and UASB section area. Recycled
methanogenic bacteria increased methane concentration and decreased hydrogen concentrations
in the HSAD reactor, while pH increased in the batch mode, but decreased in the continuous
mode with an increase of recycling rate. The UASB height had little impact on the acetic acid
concentration and methane production.
The mechnisim of a high rate digestor was explored by a multi-fluid model with KTGF. The
simulation results suggest that the dairy manure particles tend to have soft and deformable fluid
properties due to the lower contribution of collisional and kinetic components. The evaluation for
biomass retention allowed the determination of the optimum SRT in anaerobic digestion. Aided
with CFD simulation, the scale-up effect of the hydrodynamic nature from the bottle reactor to
the column reactor was reduced.

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ........................................................................................................... iii
LIST OF TABLES ....................................................................................................................... viii
LIST OF FIGURES ........................................................................................................................ x
Chapter 1 : Introduction .................................................................................................................. 1
1.1. Introduction .......................................................................................................................... 1
1.2. Anaerobic Digestion Pathways and Reactions ..................................................................... 3
1.3. Bio-chemical kinetics models ............................................................................................ 17
1.4. Anaerobic Digestion Reactors ............................................................................................ 28
1.5. Black Box Models to Simulate Anaerobic Digestors......................................................... 37
1.6. Phenomenological Models for Simulating Anaerobic Digesters ....................................... 42
1.7. Conclusions ........................................................................................................................ 53
1.8. References .......................................................................................................................... 56
Chapter 2 : Numerical Simulation of Mechanical Mixing in High Solid Anaerobic Digester ....... 1
2.1. Abstract .............................................................................................................................. 74
2.2. Introduction ........................................................................................................................ 75
2.3. Methods .............................................................................................................................. 78
2.4. Results and discussion........................................................................................................ 84
2.5. Conclusions ........................................................................................................................ 95
2.6. References .......................................................................................................................... 96

vi

Chapter 3 : Experimental and Modeling Study of a Two-Stage Pilot Scale High Solid Anaerobic
Digester System .......................................................................................................................... 100
3.1. Abstract ............................................................................................................................ 100
3.2. Introduction ...................................................................................................................... 101
3.3. Methods ............................................................................................................................ 104
3.4. Results and discussion...................................................................................................... 113
3.5. Conclusions ...................................................................................................................... 132
3.6. References ........................................................................................................................ 134
Chapter 4 : Multiphase Modeling of Settling and Suspension in Anaerobic Digester ............... 139
4.1. Abstract ............................................................................................................................ 139
4.2. Introduction ...................................................................................................................... 140
4.3. Materials and methods ..................................................................................................... 143
4.4. Results and discussion...................................................................................................... 152
4.5. Conclusions ...................................................................................................................... 171
4.6. References ........................................................................................................................ 173
Chapter 5 : General Conclusions ................................................................................................ 178

vii

LIST OF TABLES
Table 1.1. Typical composition for the common substrates used in anaerobic digestion ............... 2
Table 1.2. Biochemical rate coefficients (vi,j) and kinetic rate equations (j) for soluble
components (i=1-12; j=1-19) ........................................................................................................ 15
Table 1.3. Biochemical rate coefficients (vi,j) and kinetic rate equations (j) for particulate
components (i=1-12; j=1-19) ........................................................................................................ 16
Table 1.4. Kinetic coefficients of the first order rate of hydrolysis .............................................. 23
Table 2.1. Geometrical details for digesters with impeller A-310 and helical ribbon (Hoffmann et
al., 2008; Karim et al., 2005) ........................................................................................................ 78
Table 2.2. Parameters of the power-law model for shear stress with temperature (T), flow
behavior index n and consistency coefficient K (El-Mashad et al., 2005) ................................... 85
Table 2.3. Rheological properties and densities of liquid manure for T = 35 C
(Achkari-Begdouri and Goodrich, 1992; Landry et al., 2004; Wu and Chen, 2008).................... 90
Table 2.4. Specifications and calculated results of ribbon mixers made in Aaron Process
Equipment ..................................................................................................................................... 93
Table 3.1. Conditions for the mathematic model for the high solids anaerobic digester system 106
Table 3.2. Model parameters in the HSAD system (Batstone et al., 2002; Gali et al., 2009; Rosen
and Jeppsson, 2006) .....................................................................................................................112
Table 3.3. Ranking of roots of mean squared sensitivities ......................................................... 121
Table 4.1. Particle size distribution in the bottle reactor............................................................. 144

viii

Table 4.2. Model parameters in the ADM1 model ...................................................................... 150

Table 4.3. Dairy manure initial concentration ............................................................................ 168
Table 4.4. Operational conditions and experimental results ....................................................... 168

ix

LIST OF FIGURES
Figure 1.1 Main pathways for anaerobic degradation of organic matter used in the model ........... 7
Figure 1.2. Main pathways for ADM1 model ............................................................................... 12
Figure 1.3. Plot shows the fit of the derived model to the experimental data points and the 95%
confidence interval (Karim et al., 2007) ....................................................................................... 19
Figure 1.4. Dependence of the solids removal reciprocal rate upon solids concentration according
to the firstorder (a) and Contois models with actual (b) and optimal design (c) Model parameters:
first-order (k1=0.047 day-1, k2=0.45 day-1); Contois (X0=0.005 g l-1, Y=0.13g g-1, m =4 day-1;
K*S=10g g-1) (Vavilin et al., 2001) ................................................................................................ 26
Figure 1.5. Batch reactor (Chynoweth and Isaacson, 1987) ......................................................... 30
Figure 1.6. Continuously stirred tank reactor (Chynoweth and Isaacson, 1987) .......................... 32
Figure 1.7. Plug flow reactor (Chynoweth and Isaacson, 1987) ................................................... 33
Figure 1.8. Up-flow anaerobic sludge blanket reactor (Chynoweth and Isaacson, 1987) ............ 35
Figure 1.9. Two-phase digesters (Chynoweth and Isaacson, 1987) .............................................. 37
Figure 1.10. Schematic diagram of (a) adaptive network-based fuzzy inference system (ANFIS)
models with all input variables and (b) inputoutput mapping structure of ANFIS model with
only on-line input variables (Pai et al., 2009) ............................................................................... 40
Figure 1.11. Prediction results of CODeff. (a) ANFIS2-1 and (b) ANN2-1 (Pai et al., 2009) ....... 41
Figure 1.12. Block diagram of Bolle et al.s (A) (Bolle et al., 1986) and Wu and Hickys (B) (Wu
and Hickey, 1997) ......................................................................................................................... 45

Figure 2.1. Comparisons of prediction and experimentation in radial average velocity (A) z=5cm
(B) z=13cm (Hoffmann et al., 2008) ............................................................................................ 83
Figure 2.2. Comparison of velocity contour in TS<5% and TS=10% .......................................... 85
Figure 2.3. Power number (Np),

flow number (Nq), maximum wall shear stress and shear rate

and average shear rate vs. impeller Re for TS<5% (A) and TS=10% (B) .................................... 86
Figure 2.4. Power number and flow number change with total solids ......................................... 89
Figure 2.5. Flow field of helical ribbon (1rpm) in HSAD (A) and higher agitator speeds in
20%TS (B) .................................................................................................................................... 91
Figure 3.1. Model structure of the two-stage HSAD system ...................................................... 103
Figure 3.2. (A) pH at the different ratio of effluent to waste, and (B) biogas production change
with time in the HSAD batch reactor without recirculation ........................................................115
Figure 3.3. Biogas production (A), VFA (B) and pH (C) change with time in the UASB seed
reactor ..........................................................................................................................................117
Figure 3.4. Acetic acid (A), propionic acid (B), butyric acid (C) and valeric acid (D) change with
time in the UASB seed reactor.................................................................................................... 120
Figure 3.5. Effect of recycled methanogenic seeds on methane concentration in the HSAD
reactor (A, B) the upper zone and the lower zone without recycled methanogenic bacteria; (C, D)
the upper zone and the lower zone with recycled methanogenic bacteria .................................. 122
Figure 3.6. Effect of recycling rate on pH: (A, B) batch mode at the recycling rate of 0.05 and
0.09 m3/day; (C, D) continuous mode at the recycling rate of 0.14 and 0.5 m3/day .................. 124

xi

Figure 3.7. Comparison of acetic acid production: (A) batch mode; (B) continuous mode; (C)
increase of UASB height; (D) increase of UASB cross section area; (E) increase of recycled
methanogenic seeds .................................................................................................................... 128
Figure 3.8. Comparison of bio-methane production: (A) batch mode; (B) continuous mode; (C)
increase of UASB height; (D) increase of UASB cross section area; (E) increase of recycled
methanogenic seeds .................................................................................................................... 132
Figure 4.1. Axial and radial distribution of multi-phase variables (A) gas volume fraction; (B)
solid volume fraction; (C) granular temperature; (D) granular collisional viscosity.................. 155
Figure 4.2. Wastewater residence time distributions at (A) the reactor exit height of 0.73 m, and
(B) the inlet liquid jet velocity of 0.12 m/s ................................................................................. 156
Figure 4.3. Particles settling process in bottle reactor (A) particle size = 2.33mm, ess=0.1,
ess=0.01; (B) particle size = 0.41mm, ess=0.01; (C) particle size = 0.06mm, ess=0.01 ............... 161
Figure 4.4. Particles suspending process in column reactor at the inlet velocity of 1.0 m/s (A)
solid volume fraction; (B) turbulent kinetic energy; and (C) liquid volume fraction ................. 164
Figure 4.5. Comparison of the different digester configurations (A, B: u = 2 m/s, t = 15 s) and
dimensional models (C, D: u = 1 m/s, t = 10 s) .......................................................................... 166
Figure 4.6. Effect of biomass retention time on biogas production (A) bottle reactor (B) column
reactor ......................................................................................................................................... 170

xii

Dedication

This dissertation is dedicated to my wife for her endless support and to my parents.

xiii

CHAPTER 1 : INTRODUCTION
1.1. Introduction
Anaerobic digestion (AD) has demonstrated growing importance in practice in recent years.
It is a green technology that can extract methane-rich biogas from the biological degradation of
regionally abound biomass such as municipal solid wastes and wastewaters. With the
development of high rate systems in the period from the 1950s to 1980s the technology was
applied to agro-industrial effluents. As the technology matured, increased knowledge on
toxicity and biodegradability enabled applications to include effluents containing toxic and
recalcitrant compounds from the chemical, petrochemical and pulp/paper industries
(Mata-Alvarez et al., 2000). The current designs of the AD systems reflect the need for shorter
hydraulic retention times, higher retention of biomass, smaller reactor volume and higher
loading rates, indicative of their application in urban locations. The users of the technology
benefit by using the biogas produced, reducing odor and the volume of sludge produced, as well
as sanitizing the wastes. Germany and Denmark, where environmental legislation concerning
waste disposal is stringent, lead the way. Since 2000, there has been almost a 25-fold increase
in the annual electricity generation from digester projects in USA, from about 14 million
kilowatt-hours (kWh) to an estimated 331 million kWh per year.
(Source: http://www.epa.gov/agstar/documents/digester_status_report2010.pdf).

Table 1.1. Typical composition for the common substrates used in anaerobic digestion

Composition

Cellulose
(mg/g DM)
Hemicellulose
(mg/g DM)
Lignin
(mg/g DM)
Crude protein
(mg/g DM)
Lipid
(mg/g DM)
Carbohydrate
(mg/g DM)

Dairy
manure
(Amon et
al., 2006;
Stafford et
al., 1980)

Sewage
sludge
(Inoue et
al., 1996;
Mikkelsen
and
Keiding,
2002)

Food
waste
(Buffiere
et al.,
2006)

Fruit
waste
(Buffiere
et al.,
2006)

310

39 126

120

122 190

Grass
(Buffiere
et al.,
2006)

Wheat
straw
(Kaparaju
et al.,
2009)

Sugar beet
(Khan and
Yaqoob,
1997;
Kryvoruchko
et al., 2009)

47 78

153

359.7

180

85 295

29 133

361

239.5

530

19 96

40 174

83

193.3

30

125 297

140 346

90 208

102 169

150

6.5

87 128

23.8 46.4

6.6 155

35 81

22 118

66

1.5

15 25

125

101 198

263 609

450 514

263

853.1

40 70

Note: DM: dry of matter

AD has long been used for the treatment of animal manure and sewage sludge from aerobic
wastewater treatment. Nowadays, most of the agricultural biogas plants digest manure from
pigs, cows, and chicken with the addition co-substrates to increase the content of organic
material for achieving a higher gas yield. Typical co-substrates are harvest residues, e.g., top
and leaves of sugar beets, organic wastes from agriculture-related industries, and food waste,
collected municipal bio-waste from households and energy crops (Appels et al., 2008; Cakir
and Stenstrom, 2005; Chynoweth and Isaacson, 1987). Table 1.1 shows the typical composition

for the common substrates used in anaerobic digestion. Factors such as substrate pH,
temperature, salinity, mineral composition, loading rate, hydraulic retention time (HRT),
carbon-to-nitrogen ratio and volatile fatty acid content influence the digestibility of the
substrate and the biogas production (Krzystek et al., 2001; Novak et al., 2003; Sanchez et al.,
2006).
1.2. Anaerobic Digestion Pathways and Reactions
There are very complicated processes involved in anaerobic digestion. The process is
accomplished with a consortium of microorganisms such as fermentative bacteria,
hydrogen-producing acetogenic bacteria, hydrogen-consuming acetogenic bacteria, carbon
dioxide-reducing methanogens, and aceticlastic methanogens (methane-forming bacteria) some
bacteria that secret related enzymes (Appels et al., 2008; Chynoweth and Isaacson, 1987).
Methanogenic bacteria grow more slowly than those in the nonmethanogenic group
(Chynoweth and Isaacson, 1987). The biodegradable materials in the organic matter such as
carbohydrates, proteins and lipids are broken down by extra-cellular enzymes such as cellulase,
protease and lipase to soluble products of a small enough to allow their transport across the cell
membrane. Before that, anaerobic degradation of viable, biological solids (e.g. waste activated
sludge, algae) also requires an additional conversion mechanism, namely death and lysis of
viable cells. These soluble products undergo a number of different metabolic processes being
converted to intermediate molecules including sugars, amino acids, fatty acids, hydrogen, and
acetic acid, finally being converted to biogas, mainly methane and carbon dioxide (Bagi et al.,

2007).
Generally, there are three temperature regimes operated for anaerobic digesters:
thermophilic (4565 C), mesophilic (2545 C) and psychrophilic (1025 C) (Chynoweth and
Isaacson, 1987). Most of the digestion processes take place at mesophilic and thermophilic
temperature conditions. It is important to keep a constant temperature during the digestion
process, as temperature changes or fluctuations will affect the biogas production negatively. In
most cases, methanogenic diversity is lower in plants operating at thermophilic temperatures
(Karakashev et al., 2005; Leven et al., 2007). Therefore, thermophilic processes are more
sensitive to temperature fluctuations and require longer time to adapt to a new temperature.
Mesophilic bacteria tolerate temperature fluctuations of 3 C without significant reductions in
methane production. The growth rate of methanogenic bacteria is higher at thermophilic process
temperatures making the process faster and more efficient. Therefore, a well-functioning
thermophilic digester can be loaded to a higher degree or operated at a lower hydraulic
retention time (HRT) than at mesophilic conditions. But the thermophilic process temperature
results in a larger degree of imbalance and a higher risk for ammonia inhibition. Ammonia
toxicity increases with increasing temperature, and washout of microbial population can occur
(Angelidaki et al., 1993).
Methane formation takes place within a relatively narrow pH interval, from about 6.5 to 8.5
with an optimum interval between 7.0 and 8.0. The process is severely inhibited if the pH
decreases below 6.0 or rises above 8.5. The pH value increases by ammonia accumulation

during degradation of proteins, while the accumulation of volatile fatty acid (VFA) decreases
the pH value (Weiland, 2010). The accumulation of VFA will often not always result in a pH
drop, due to the buffer capacity of the substrate. Animal manure has a surplus of alkalinity
which stabilizes the pH value at VFA accumulation. VFA are a key intermediate in the process
and are capable of inhibiting methanogenesis in high concentrations. Acetic acid is usually
present in higher concentration than other fatty acids, but propionic and butyric acids are more
inhibitory to methanogens (Mosche and Jordening, 1999; Wang et al., 1999). The ideal pH
range for hydrolysis (5.56.5) and methanation (6.87.2) is different (Parawira et al., 2008;
Vieitez and Ghosh, 1999).
Overall the bio-chemical reactions that happen during anaerobic digestion can be classified
into: heterogeneous reaction including hydrolysis and homogeneous reactions including
acidogenesis, acetogenesis and methanogenesis. These reactions lead to the formation of
intermediary products soluble in the aqueous phase, release of these products to the aqueous
phase, metabolic conversion of these intermediates into products. Acidogenic bacteria, through
the production of acids, reduce the pH of the tank. Methanogenic bacteria live in a strictly
defined pH range (Scheper and Ahring, 2003). In order to bring maximum control over the
bacterial communities living within the digesters, a two-stage or multi-stage digestion system
was presented. Typically hydrolysis, acetogenesis and acidogenesis occur within the first vessel.
The organic material is then heated to the required operational temperature (either mesophilic or
thermophilic) prior to being pumped into a methanogenic reactor. The initial hydrolysis or

acidogenesis tanks prior to the methanogenic reactor can provide a buffer to the rate at which
feedstock is added. Acidogenic bacteria produce organic acids and more quickly grow and
reproduce than methanogenic bacteria (Griffin et al., 1998).
With the recognition of various microbial groups and substrates in AD system, increasingly
complex kinetics models have been developed. However, it is difficult to identify each
composition in complex substrates. A few models have attempted simulation of co-digestion of
different wastes by defining the waste by its general composition. For example, waterwaste was
considered as composed of carbohydrates, proteins, lipids and others (Jeyaseelan, 1997). Based
on these aforementioned compositions, several researchers constructed different pathways for
anaerobic digestion (Tomei et al., 2009). The more advanced substrate pathways were presented
in the Angelidaki model and anaerobic digestion model No.1 (ADM1) model.
1.2.1 Angelidaki et al. model
Angelidaki et al. tried to describe a detailed pathway of intermediates and include the
interactions occurring during degradation of the intermediates (Angelidaki et al., 1999). They
presented a comprehensive model to simulate co-digestion of complex wastes with different
characteristics and composition. The main pathways are schematically shown in Figure 1.1. The
model involves two enzymatic processes: (a) hydrolysis of undissolved carbohydrates, and (b)
of undissolved proteins, and eight bacterial groups: (1) glucose-fermenting acidogens, (2)
lipolytic bacteria, (3) LCFA-degrading acetogens, (4) amino acid-degrading acidogens, (5)
propionate, (6) butyrate, (7) valerate-degrading acetogens, and finally (8) aceticlastic

methanogens.

Figure 1.1 Main pathways for anaerobic degradation of organic matter used in the model

The general composition definition has been linked as directly as possible to normal
analytical values commonly used to characterize organic waste to allow waste to be
characterized with a reasonable accuracy without experimental data.
Carbohydrates: Carbohydrates are found in the model as soluble, insoluble, and inert
carbohydrates. Insoluble carbohydrates (C6H10O5)is are enzymatically hydrolyzed to soluble
carbohydrates (C6H10O5)s and inert carbohydrates (C6H10O5)in.
(C6H10O5)is Yc (C6H10O5)s +

(1-Yc) (C6H10O5)in

(1.R1)

where Yc is the degradability of the carbohydrates.

Soluble carbohydrates are further degraded to acetate (C2H4O2), propionate (C3H6O2),

butyrate (C4H8O2) in acidogenic step (Angelidaki et al., 1993).

(C6H10O5)s + 0.1115NH3 0.1115C5H7NO2 +

0.744C2H4O2 +

C4H8O2 +

0.5C3H6O2 + 0.4409

0.6909CO2 + 0.0254H2O

(1.R2)

Propionate (C3H6O2) and butyrate (C4H8O2) are degraded to acetate (C2H4O2) and hydrogen
(H2) in acetogenic step.
C3H6O2 + 1.764H2O + 0.0458NH3

0.0458C5H7NO2 + 0.9345C2H4O2 + 2.804H2 +

0.902CO2

(1.R3)

C4H8O2 + 1.7818H2O + 0.0544NH3 + 0.0544CO2 0.0544C5H7NO2 + 1.8909C2H4O2 +

1.8909H2
In the methanogenic step, there are two pathways to generate methane (CH4).

(1.R4)
One is

hydrogen utlizing methanogenesis including Eq. (1.R3) and Eq.(1.R4) which derived from the
propinic step [Eq. (1.R3)] and the butyrate step [Eq. (1.R4)], respectively. Another is
aceticlastic methanogenesis. This is the major methanogenic step that acetate is cleaved to form
methane (CH4) and carbon dioxide (CO2).
2.804H2 + 0.01618NH3 + 0.7413CO2 0.001618C5H7NO2 + 0.6604CH4 + 1.45H2O
(1.R3)
1.8909H2 + 0.0109NH3 + 0.4999CO2 0.0109C5H7NO2 + 0.4452CH4 + 0.9780H2O
(1.R4)
C2H4O2 + 0.022NH3

0.022C5H7NO2 + 0.945CH4

+ 0.066H2O + 0.945CO2

(1.R5)
Lipid: Lipid uses glycerol trioleate (GTO, C57H104O6) as a model. Oleate is the most
abundant type of long-chain fatty acids (LCFA, C18H34O2) in many vegetable oils. GTO
degradation is derived by combining the GTO lipolysis to oleate and glycerol (C3H8O3) and the
glycerol degradation to biomass (C5H7NO2) and propionate (C3H6O2).
C57H104O6 +3 H2O C3H8O3 + 3 C18H34O2

(1.R6)

C3H8O3 + 0.04071 NH3 + 0.0291 CO2 0.04071 C5H7NO2 + 0.9418 C3H6O2 + 1.09305 H2O
(1.R7)
Degradation of the resulting oleate by LCFA-degrading acetogens is derived by combining
the LCFA step (1.R8), together with the hydrogen utilizing step (1.R9):
C18H34O2 + 15.2398 H2O + 0.1701 NH3 + 0.2500 CO2 0.1701 C5H7NO2 + 8.6998 C2H4O2 +
14.500 H2

(1.R8)

14.500 H2 + 0.0836 NH3 + 3.8334 CO2 0.0836 C5H7NO2 + 3.4139 C3H6O2 + 7.4997 H2O
(1.R9)
Protein: Protein uses gelatin (CH2.03O0.6N0.3 S0.001) as a model. Proteins are first
hydrolyzed to amino acids.
(Protein)is Yp (Amino acids) + (1 Yp) (Protein)in

(1.R10)

where Yp is the degradability of protein.

Further degradation of amino acids (CH2.03O0.6N0.3S0.001) to volatile fatty acids is performed
by acidogenic bacteria,

CH2.03O0.6N0.3S0.001 + 0.3006 H2O 0.017013 C5H7NO2 + 0.29742 C2H4O2 + 0.02904 C3H6O2 +

0.022826 C4H8O2+ 0.013202 C5H10O2 + 0.07527 CO2
+ 0.28298 NH3 + 0.001 H2S

(1.R11)

The main acids formed during gelatin degradation are acetate (C2H4O2), propionate
(C3H6O2), butyrate (C4H8O2), and valerate (C5H10O2). The degradation of valerate was
determined experimentally.
C5H10O2 + 0.0653 NH3 + 0.5543 CO2 + 0.8045 H2O 0.0653 C5H7NO2 + 0.8912 C2H4O2 +
0.02904 C3H6O2 + 0.4454 CH4

(1.R12)

When lipid and protein are converted to acetate, there are sequential reactions of
aceticlastic methanogenesis shown in Eq. (1.R5) to cleave it to methane (CH4) and carbon
dioxide (CO2).
For the hydrolytic step, first order reaction rates are applied to give appropriate description.
These reaction rates are assumed to be inhibited by the sum of VFA including acetate,
propionate, butyrate and valerate.

For the acidogenic, acetogenic, and methanogic steps,

Monod growth kinetics are used with respect to their primary substrates. The effect of pH,
ammonia and temperature on the growth rate are considered in all bateria steps.
1.2.2. Anaerobic digestion model No.1 (ADM1) model
A collaborative work of many international experts from various fields of anaerobic
process technology have been completed to develop the international water association (IWA)
ADM1 model

(Batstone et al., 2002a; Fedorovich et al., 2003). The main objective of the

10

work was to provide an instrument that could overcome the limits of the models developed over
the last few decades. The limits of the previous models were essentially due to their specificity
so that these models cannot be applied more widely. The main aim of the ADM1 model is the
possibility of simulating a broad category of processes. Some peculiar and specific aspects are
not included in order to obtain an instrument that is effective and at the same time easy to use.
Therefore, the ADM1 model can be regarded as a common platform from which applications to
specific processes and situations could be developed.
In developing the ADM1, the task group tried to establish common nomenclature, units and
model

structure,

consistent

with

existing

anaerobic

modeling

literature

and

the

popular-activated sludge models (Batstone et al., 2002a). Outputs from the model include
common process variables such as gas flow and composition, pH, separate organic acids, and
ammonium. The structure encourages specific extensions or modifications where required, but
still maintaining a common platform.

11

Figure 1.2. Main pathways for ADM1 model

ADM1 model is a structured model that reflects the major processes that are involved in the
conversion of complex organic substrates into methane and carbon dioxide and inert byproducts.
Figure 1.2 shows an overview of the substrates and conversion processes (Batstone et al., 2002a;
Parker, 2005). Table 1.2 and 1.3 shows biochemical rate coefficients (vi,j) and kinetic rate
equations (j) for soluble and particulate components (i=1-12; j=1-19), respectively. The model
includes disintegration of complex solids into inert substances, carbohydrates, proteins and fats.
The products of disintegration are hydrolyzed to sugars, amino acids and long chain fatty acids
(LCFA) respectively. Carbohydrates and proteins are fermented to produce volatile organic
acids (acidogenesis) and molecular hydrogen. LCFA are oxidized anaerobically to produce
acetate and molecular hydrogen. Propionate, butyrate and valerate are converted to acetate

12

(acetogenesis) and molecular hydrogen. Methane is produced by both cleavage of acetate to

methane (aceticlastic methanogenesis) and reduction of carbon dioxide by molecular hydrogen
to produce methane (hydrogenotrophic methanogenesis). The reaction kinetics in an anaerobic
digester is complex with a number of sequential and parallel steps. There are two main types of
reactions in the model which are biochemical reactions and physic-chemical reactions.
Biochemical reactions are normally catalyzed by extracellular enzymes and act on the pool of
biological available organic material. All biochemical extracellular steps were assumed to be
first order, which is a simplification based on empiricism, reflecting the cumulative effect of a
multi-step process. Substrate uptake Monod-type kinetics are used as the basis for all
intracellular biochemical reactions. As shown in Table 1.2, biomass growth is included in
substrate uptake equations via biomass concentration. A substrate uptake reaction rate is
proportional to the biomass growth.

Inhibition functions include pH (all groups), hydrogen

(acetogenic groups) and free ammonia (aceticlastic methanogens). pH inhibition is

implemented as one of two empirical equations, while hydrogen and free ammonia inhibition
are represented by non-competitive functions. The other uptake-regulating functions are
secondary Monod kinetics for inorganic nitrogen (ammonia and ammonium), to prevent growth
when nitrogen is limited, and competitive uptake of butyrate and valerate by the single group
that utilizes these two organic acids. Physic-chemical reactions are not biologically mediated by
micro-organisms and encompass ion association/dissociation and gas-liquid transfer.
Dissociation/association processes are very rapid compared to other reactions (especially

13

biochemical); they are often referred to as equilibrium processes, and can be described by
algebraic (rather than differential) equations. Liquid-gas transfer is described the dynamic gas
transfer equations which are based on two-film theory.

14

Table 1.2. Biochemical rate coefficients (vi,j) and kinetic rate equations (j) for soluble
components (i=1-12; j=1-19)
component

10

11

12

S su

Saa

Sfa

S va

Sbu

Spro

Sac

S h2

Sch4

S IC

SIN

SI

Rate (j, kg COD m-3 d-1)

j

process

Disintegration

Hydrolysis of
Carbohydrates

Hydrolysis of Proteins

Hydrolysis of Lipids

1-ffa, li

Uptake of Sugars

-1

Uptake of Amino Acids

Uptake of LCFA

Uptake of Valerate

Uptake of Butyrate

10

Uptake of Propionate

11

Uptake of Acetate

12

Uptake of Hydrogen

13

Decay of Xsu

kdec,Xsu X su

14

Decay of Xaa

kdec,Xaa X aa

15

Decay of Xfa

kdec,Xfa X fa

16

Decay of Xc4

kdec, Xc4 X c 4

17

Decay of Xpro

kdec,Xpro X pro

18

Decay of Xac

kdec,Xac X ac

Decay of Xh2

kdec,Xh2 X h2

k h yd,ch X ch

k hyd,pr X pr

k hyd,li X li

ffa, li

(1-Yaa)fva,aa

Ssu
X su I1
K S,su + Ssu

Naa-(Yaa)Nbac

k m,aa

Saa
X aa I1
K S,aa + S aa

(1-Yfa)0.3

-(Yfa)Nbac

km,fa

Sfa
X fa I1
K S,fa + Sfa

(1-Yc4)0.31

(1-Yc4)0.15

-(Yc4)N bac

km,c4

S va
1
X c4
I2
K S,va + S va
1 + Sbu S va

(1-Yc4)0.8

(1-Yc4)0.2

-(Yc4)N bac

km,c4

Sbu
1
X c4
I2
KS,bu + Sbu
1 + S va S bu

(1-Ypro)0.57

(1-Ypro)0.43

-(Y pro)N bac

k m,pro

(1-Y su)fac,su

(1-Ysu)fh2,su

(1-Y aa)fbu,aa

(1-Yaa)fpro,aa

(1-Y aa)fac,aa

(1-Yaa)fh2,aa

(1-Yfa)0.7

-1

(1-Yc4)0.54

-1

-1

-1

civi,6

i =19,11 24

ci vi,10

i =19,11 24

(1-Yh2)

ci vi,11

i =19,11 24

i =19,11 24

ci vi,12

S pro
K S,pro + Spro

X pro I 2

-(Yac)N bac

k m,ac

S ac
X ac I 3
K S,ac + S ac

-(Yh2)N bac

k m,h2

S h2
X h2 I 1
K S,h2 + S h2

15

Soluble Inerts (kgCOD m-3)

Inorganic Nitrogen (kgCOD m-3)

Inorganic Carbon (kgCOD m-3)

-3

Hydrogen Gas (kgCOD m-3)

-3

Total Acetate (kgCOD m )

Total Propionate (kgCOD m-3)

-3

Total Butyrate (kgCOD m )

Inhibition factors
Total Valerate (kgCOD m-3)

Long Chain Fatty Acid (kgCOD m-3)

-3

i =19,11 24

(1-Yac)

-1

Amino Acids (kgCOD m )

civi,5

k m,su

(1-Ysu)fpro,su

-1

-(Ysu)Nbac

(1-Y su)fbu,su

Methane Gas (kgCOD m )

-1

Monosacharides (kgCOD m-3)

19

k dis X c

fsI, xc

I1=IpHIIN,lim
I2=IpHIIN,limIh2
I3=IpHIIN,limINH3,Xac

Table 1.3. Biochemical rate coefficients (vi,j) and kinetic rate equations (j) for particulate
components (i=1-12; j=1-19)
component

13

14

15

16

17

18

19

20

21

22

23

24

Xsu

Xaa

Xfa

Xc4

Xpro

Xac

Xh2

XI

Rate (j, kg COD m-3 d -1)

j

process

Xc

Xch

Xpr

Xli

Disintegration

-1

fch, xc

fpr, xc

fli, xc

Hydrolysis
Carbohydrates

of

Hydrolysis
Proteins

of

Hydrolysis of Lipids

Uptake of Sugars

Uptake
Acids

Uptake of LCFA

Uptake of Valerate

Yc4

k m,c4

S va
1
X c4
I2
K S,va + S va
1 + Sbu S va

Uptake of Butyrate

Yc4

km,c4

Sbu
1
X c4
I2
1 + S va Sbu
K S,bu + Sbu

10

Uptake of Propionate

11

Uptake of Acetate

12

Uptake of Hydrogen

13

Decay of Xsu

14

Decay of Xaa

15

Decay of Xfa

16

Decay of Xc4

17

Decay of Xpro

18

Decay of Xac

19

Decay of Xh2

khyd,ch X ch

-1

khyd,pr X pr

-1

khyd,li X li

-1

Ysu

Amino

Yaa

Yfa

Ysu

Ysu

k m,fa

Sfa
X fa I1
K S,fa + Sfa

Spro
K S,pro + Spro

X pro I 2

k m,ac

S ac
X ac I 3
K S,ac + S ac

k m,h2

S h2
X h2 I1
K S,h2 + S h2

kdec, Xc4 Xc 4

-1

kdec,Xpro Xpro

-1

kdec,Xac Xac

16

kdec,Xh2 X h2

Inhibition factors

Particulate Inerts (kgCOD m-3)

-1

Hydrogen Degraders (kgCOD m-3)

Acetate Degraders (kgCOD m-3)

Propionate Degraders (kgCOD m-3)

Valerate and butyrate Degraders (kgCOD m-3)

LCFA Degraders (kgCOD m-3)

Amino Acid Degraders (kgCOD m-3)

-1

Sugar Degraders (kgCOD m-3)

Saa
X aa I1
K S,aa + S aa

kdec,Xfa Xfa

-1

Lipids (kgCOD m-3)

k m,aa

kdec,Xaa X aa

-1

Proteins (kgCOD m-3)

S su
X su I1
K S,su + S su

kdec,Xsu X su

-1

Carbohydrates (kgCOD m-3)

k m,su

k m,pro

Ysu

Composites (kgCOD m-3)

of

k dis X c

fsI, xc

I1=IpHIIN,lim
I2=IpH IIN,limIh2
I3=IpHIIN,limINH3,Xac

1.3. Bio-chemical kinetics models

Biological treatment processes have been mathematically described by the theory of
continuous cultivation of microorganisms (Kono and Asai, 1969; Metcalf & Eddy, 2003).
Biological growth kinetics was derived from two fundamental relationships: growth rate and
substrate utilization rate. The effect of the growth-limiting substrate concentration on the rate of
microbial growth has been described by various mathematical models. As usual, nutrients are
assumed to be excessive. If the growth limiting nutrients were studied, nutrients could be
assumed to be substrates in the kinetics models. The following equations are some common
kinetic correlations in anaerobic treatment.
First order kinetic model (Pavlostathis and Giraldogomez, 1991)

K S ,max S
b
S0 S

dS
= K S ,max S
dt

S=

S0
1 + K S ,max tSRT

(1.1)

Grau et al. kinetic model (Grau et al., 1975)

max S
S0

dS max XS
=
dt
YS0

S0 (1 + btSRT )
max tSRT

(1.2)

K S (1 + btSRT )
tSRT ( max b ) 1

(1.3)

S=

Monod kinetic model (Monod, 1949)

max S
KS + S

dS
max XS
=
dt
Y ( KS + S )

17

S=

Contois kinetic model (Contois, 1959)

max S
KX X + S

dS
max XS
=
dt
Y (KX X + S )

S=

K X YS0 (1 + btSRT )
K X YS0 (1 + btSRT ) + tSRT ( max b ) 1

(1.4)
Chen & Hashimoto kinetic model (Chen and Hashimoto, 1980)

max S
KS0 + (1 K ) S

dS
XS
= max
dt
KX + YS

S=

KS0 (1 + btSRT )
( K 1)(1 + btSRT ) + max tSRT

(1.5)

Haldane kinetic model (Lokshina et al., 2001)

dS

= max
dt
Y

SB

( K)

KS + S + S S

(1.6)

where is the specific growth rate; S0 and S are influent and effluent concentration of the
growth-limiting substrate, respectively; X is microorganism concentration; KS,max is

the

maximum specific substrate utilization rate; b is specific microorganism decay rate; tSRT is solid
retention time (Or mean cell residence time); Y is growth yield coefficient; max is the maximum
specific growth rate; KX is Contois kinetic constant; K is Chen and Hashimoto dimensionless
kinetic constant, B is biomass concentration, KS is the half saturation coefficient, KI is the
inhibition constant.
Based on the above equations, many modified kinetic models were applied into the different
AD kinetics studies and the good fit results were obtained (Karim et al., 2007; Vavilin et al.,
2001). The curve fit along with 95% confidence interval and the experimental data points have
been shown in Figure 1.3. Methane production rate (G) is calculated as (Karim et al., 2007)

18

G=

Yms ( S0 S )

(1.7)

where = V/Q = hydraulic retention time, V is the digester working volume, Q is the flow rate.

Figure 1.3. Plot shows the fit of the derived model to the experimental data points and the 95%
confidence interval (Karim et al., 2007)
Even in the complex mathematical models, these equations are also the fundamentals to
constitute the whole frame of AD process (Parker, 2005).

According to the method of

constructing a model, AD process kinetics models in the literatures are divided into two kinds by
the author. One is rate limiting models and the other is general composition models.

1.3.1. Rate limiting models

As described in the introduction, there are four key biological and chemical steps of

19

anaerobic digestion. Most of the early attempts to kinetically describe the anaerobic digestion
process were dependent on the so-called rate-limiting step approach. In general, when a process
is composed of a sequence of reactions, one step is usually very much slower that the other steps.
The last low step in a sequence of reactions has been called the rate controlling, rate limiting or
rate determining step. The first dynamic mathematical models of anaerobic digestion were based
on the assumption that a description of the rate-limiting step will also allow a sufficiently precise
description of the entire process. The first dynamic mathematical model was based on Andrews
(Andrews, 1969) who considered only the acetoclastic methanogenic process step. The model
describes the methane generation from acetic acid, assuming that hydrolysis and acidification are
not rate limiting. This model of one bacteria population includes dynamic mass balances of a
continuous stirred tank reactor (CSTR) concerning the concentration of acetic acid and biomass.
Main simplification of this model is the existence of a constant pH. Bicarbonate dissociation
equilibrium and ion charge balance is used to calculate the dissolved part of the produced carbon
dioxide. Only transport of carbon dioxide from the liquid to the gas phase is considered.
In anaerobic digestion, the rate limiting step is related to the nature of the substrate, process
configuration, temperature, and loading rate (Speece, 1983).

For example, methane formation

from organic matter contained in the samples of tundra soil was modeled in the wide range of
temperature conditions. It was concluded that hydrolysis is the rate-limiting step at 10 28 C,
but at 6 C the rate of acetoclastic methanogenesis becomes the rate-limiting stage in methane
production (Vavilin et al., 1997). Most of AD rate limiting model were focused on the hydrolysis

20

of particulate complex organic material such as sewage sludge (Vavilin et al., 2008).

1.3.2. Hydrolysis
Particulate organic material cannot be utilized by microorganisms unless it is broken down
to soluble compounds which can then pass the cell membrane. There are two mechanisms to
describe the process of hydrolysis (Batstone et al., 2002a).
(1) The organisms secrete enzymes to the bulk liquid where they are adsorbed onto a particle or
react with a soluble substrate (Jain et al., 1992).
(2) The organisms attach to a particle, produce enzymes in its vicinity and benefit from soluble
products released by the enzymatic reaction (Vavilin et al., 1996).
Vavilin et al. (Vavilin et al., 2008) reviewed the available kinetics of the hydrolysis process
in the previous literature. The following is a list of these kinetics models.
(1) The first order kinetics of carbohydrate, lipid and protein degradation: The most commonly
applied model for the description of hydrolysis rate is the first order with respect to the
concentration of degradable organic material.

dS
= khyd S
dt

dP
= kS
dt

(1.8)

where S is the volatile solids (VS) concentration, P is the product concentration, khyd is the first
order coefficient, and is the conversion coefficient of VS to product. Table 1.4 lists the kinetic
coefficients of the first order rate of hydrolysis for different substrates that can be found in the
literature. First order kinetics can only be applied when the rate-limiting is the surface of the

21

particulate substrate and biodegradability related phenomena do not interfere.

(2) Disintegration, solubilisation and enzymatic hydrolysis: The Michaelis-Menten kinetics was
applied for the hydrolysis of a soluble substrate, and is expressed as:

dS
S
S
= kE
= Vm
at
Km + S
Km + S

(1.9)

where S, E are the substrate and enzyme concentrations, Vm=kE is the maximum hydrolysis rate,

k is the maximum hydrolysis rate constant, and Km is the half saturation rate coefficient.
(3) Biodegradability of complex substrates: The first order kinetics was corrected by the
non-degradable fraction of a complex substrate:

dS
= k ( S S0 )
dt

(1.10)

where S0 is the initial substrate concentration, and is the non-degradable fraction of the
substrate.
Instead of Eq. (1.8), different types of kinetics including n-order reaction may be applied to
describe complex substrate hydrolysis.

dS
KS
=
at K S ( S S0 ) + S

(1.11)

where K is the maximum hydrolysis rate which in turn depends on hydrolytic biomass or enzyme
concentration, and K'S is an additional model coefficient.
Eq. (1.8) also can be changed to introduce a biomass concentration X:

dS
= k X n ( S S0 )
dt

22

(1.12)

where k' is a rate constant and n is a power index.

Table 1.4. Kinetic coefficients of the first order rate of hydrolysis

Substrate

khyd (day-1)

T
(C)

Reference

Carbohydrate

0.0250.2

55

(Christ
2000)

Carbohydrate

0.52.0

Proteins

0.0150.075

Proteins

0.250.8

Proteins
(gelatine)
Lipids

0.65
0.0050.01

Lipids

0.10.7

Lipids

0.76

Lipids

0.63

Cellulose

0.040.13

Cellulose

0.066

35

Cellulose

0.1

35

Cellulose

0.12

38

Kitchen waste

0.34

35

55

55

25

et

al.,

(Garcia-Heras,
2003)
(Christ et al.,
2000)
(Garcia-Heras,
2003)
(Flotats et al.,
2006)
(Christ et al.,
2000)
(Garcia-Heras,
2003)
(Shimizu et al.,
1993)
(Masse et al.,
2002)
(Gujer
and
Zehnder, 1983)
(Liebetrau et al.,
28 September-1
October 2004)
(Vavilin et al.,
1996)
(O'Sullivan
et
al., 2005)
(Liebetrau et al.,

23

Substrate

khyd
(day-1)

T
(C)

Reference

Biowaste

0.12

35

Cattle manure

0.13

55

Pig manure

0.1

28

Municipal
solid waste
Office paper

0.1

15

0.036

35

Cardboard

0.046

35

Newsprint

0.057

35

Food waste

0.55

37

Forest soil

0.54

30

Forest soil

0.090.31

20

Slaughthouse
house

0.35

35

(Liebetrau et
al.,
28
September-1
October 2004)
(Vavilin et al.,
2008)
(Vavilin et al.,
1997)
(Bolzonella et
al., 2005)
(Vavilin et al.,
2004)
(Vavilin et al.,
2004)
(Vavilin et al.,
2004)
(Vavilin et al.,
2004)
(Lokshina and
Vavilin, 1999)
(Lokshina and
Vavilin, 1999)
(Lokshina et
al., 2003)

Household
solid waste

0.1

37

Primary
sludge
Primary

0.41.2

35

0.99

35

(Vavilin
and
Angelidaki,
2005)
(O'Rourke,
1968)
(Ristow et al.,

Crops
and
crop residues
Grass

0.266

40

Napier grass

0.09

35

Paper
and
cardboard

0.012

35

28 September-1
October 2004)
(Lehtomaki
et
al., 2005)
(Veeken
and
Hamelers, 1999)
(Tong et al.,
1990)
(Qu et al., 2009)

Paper
and
cardboard

0.02

55

(Qu et al., 2009)

0.0090.094

sludge

2006)

Secondary
sludge
Wood grass

0.170.60

35

(Ghosh, 1981)

0.079

35

Wheat straw

0.087

35

Leaves

0.386

40

Straw

0.14

40

(Tong et
1990)
(Tong et
1990)
(Veeken
Hamelers,
1999)
(Veeken
Hamelers,
1999)

al.,
al.,
and

and

(4) Surface-related kinetics and two phase model of hydrolysis of particulate substrate: The
surface-related hydrolysis kinetics model that took into account the colonization of waste
particles by hydrolytic bacteria is expressed by

= m

X
S
1 + X KS + S

(1.13)

where and 'm are the current and maximum hydrolysis rates, respectively; S is the volatile
solid waste concentration, X is the concentration of hydrolytic (acidogenic) biomass, is the
equilibrium constant equal to the ratio between the adsorption and desorption rate constant in the
Langmuir function; and KS is the half saturation coefficient for the volatile solid waste
concentration S.
In addition, microorganisms attached to a particle produce enzymes in the vicinity of this
particle and benefit from soluble products released by the enzymatic reaction. The Contois model

24

that uses a single parameter to represent saturation of both substrate and biomass is expressed by

= m X

S
S
X
= m X
KX X + S
KX + S

(1.14)
X

where m is the specific maximum hydrolysis rates, and KX is the half saturation coefficient for
the ration S/X.
The differential equations for substrate, biomass and product are as following,

dS
= ( S, X )
dt

dX
= Y ( S, X )
dt

dP
= ( S , X )
dt

(1.15)

Assumed that hydrolysis, the first step of solid waste degradation, was the rate-limiting step
of the total anaerobic degradation process, Vavilin et al. (2001) also used Contois kinetics and
first order kinetics to describe the hydrolysis of biodegradable solids in a steady state two-stage
anaerobic digester system (see Figure 1.4). The results showed that Contois kinetics was
preferable to the traditional first-order kinetics when considering the optimal design.

25

Figure 1.4. Dependence of the solids removal reciprocal rate upon solids concentration according
to the firstorder (a) and Contois models with actual (b) and optimal design (c) Model parameters:
first-order (k1=0.047 day-1, k2=0.45 day-1); Contois (X0=0.005 g l-1, Y=0.13g g-1, m =4 day-1;

K*S=10g g-1) (Vavilin et al., 2001)

1.3.3. Acetogenesis and Methanogenesis
The formula of hydrolysis can also be applied into acetogenesis and methanogenesis when
considering these processes as rate limiting step. The stage of acidogenesis, following the
hydrolysis stage, is usually the quickest step during the anaerobic digestion of complex organic
material (Pavlostathis and Giraldogomez, 1991; Vavilin et al., 2008). The following steps,
acetogenesis and methanogenesis, can be the rate limiting steps in anaerobic digestion of a
complex substrate at a high organic loading (Vavilin et al., 2008). After treated at 165 oC in
electric mode, the results of AD showed that the acetate and propionate degradation steps can

26

also be considered as the limiting step for the anaerobic digestion of a complex substrate (waste
activated sludge) with an organic loading of 0.5 gCOD per gVS of inoculum (Mottet et al., 2009).
During overload with substrates, product inhibition of polymer hydrolysis/acidogenesis by a high
level of VFA might occur and methanogenesis rather than hydrolysis was the rate-limiting steps
of the overall anaerobic digestion process (Vavilin and Angelidaki, 2005). In the different
operational mode, the results showed that methanogenesis was rate limiting in the accumulation
system (AC) system while the hydrolysis was the rate-limiting step during batch digestion
(El-mashad et al., 2003). Carbohydrate was used to give pulses with an easily degradable
substrate, where the last steps in the degradation, acetogenesis and methanogenesis, would be
expected to be limiting (Bjornsson et al., 2001). Since methanogenic bacteria are the most
sensitive, with the lowest growth rates, methanogenesis is frequently considered to be the
rate-limiting step in modeling attempts (Azeiteiro et al., 2001). Andrews et al. (Andrews, 1969)
considered the acetoclastic methanogenic process as rate limiting step. Monod kinetics with
substrate inhibition are assumed,

max
K
I
1+ S +
S
KI

(1.16)

where is the specific growth rate, max is the maximum specific growth rate; KS is the half
saturation coefficient, KI is the inhibition constant, S is the concentration of growth-limiting
substrate, I is the inhibitor concentration.
Although no experimental validation of the Andrews model has been performed, the

27

proposed model structure has been the basis of numerous later models. Central issue in model
application was the difficulty in the determination of kinetic data describing the anaerobic
conversion of acetate to methane. Even using the same culture of Methanosarcina barkeri, strain
227, and acetate as substrate, different maximum growth rates were determined (Lubken et al.,
2010). This might be the reason why there were fewer rate limiting model studies in
methanogenesis than that in hydrolysis.
The rate limiting model leads to simple and readily usable models. It is possible to predict
the biogas production rate when the fermentation process is determined by one single process
step. However, main drawback of the rate limiting model is the fact that the limiting step
depends on the operating conditions and is, therefore, not always the same and can vary over
time. When only one bacterial species, e.g., acetoclastic methanogens, or only the hydrolysis step
is considered, intermediate fermentation products cannot be predicted. Intermediates like volatile
fatty acids (VFA) are important parameters to assess digester stability.

1.4. Anaerobic Digestion Reactors

Different reactor types are used in the AD system. The most common reactor configuration
employed for wet fermentation is the vertical continuously stirred tank fermenter which is
applied in nearly 90% of modern biogas plants in Germany (Gemmeke et al., 2009; Weiland,
2010). However, it is becoming apparent that in many applications the conventional stirred tank
is not the optimum configuration for AD. Limitation of this reactor include washout of unreacted
solids and active microorganisms at higher loadings, difficulty in achieving complete mixing,

28

energy requirements associated with mixing, and disruption of microbial consortia by mixing.
The other conventional digesters such as batch and plug flow reactors are also applied in the AD
system. Several innovative digester configurations are under development with the objectives of
increasing process stability and net energy output, and simplifying design and operation and
improving economics. These digesters include the upflow anaerobic sludge blanket (UASB), the
attached film, the leach bed, the membrane bioreactor (MBR) and the fluidized bed et al. (Busch
et al., 2009; Fuentes et al., 2009; Lopez et al., 2009; Sharrer et al., 2007). The commonly used
anaerobic digestion reactors are described as following.
1.4.1. Batch digesters
The batch digesters are the simplest and most common type of digestion used worldwide.
This type of operation has been used widely at the household or farm scale for hundreds years.
In batch-fed digesters shown in Figure 1.5, the total feed volume is added to the reactor where it
remains until digestion is completed. While these reactors may be mixed, larger-scale reactors,
especially those using high solids content feeds, are frequently nonmixed. The solids retention
time (SRT) and hydraulic retention time (HRT) are the same in the batch digesters. These
systems are relatively inexpensive to construct and operate compared to continuously stirred tank
reactors (CSTRs). This noncontinuous process is particularly suited for seasonally produced
biomass feeds and for feeds with very high solids content. A major disadvantage of batch
digestion is that it is relatively unstable and uncontrollable due to changes in the bacterial
population during the course of the fermentation. These changes can lead to population

29

imbalance, digestion failure, and variations in the quantity and composition of product gas.

Figure 1.5. Batch reactor (Chynoweth and Isaacson, 1987)

(Source: http://www.extension.org/pages/30307/types-of-anaerobic-digesters )

30

1.4.2. Continuously stirred tank reactors (CSTRs)

The CSTRs shown in Figure 1.6 are widely used and considered conventional in the
wastewater treatment plants. Feed is added and an equivalent volume of effluent is removed on a
continuous or intermittent basis from a continuously mixed reactor. The design incorporates
stirring or agitation of the contents to achieve good mixing using pumps, mechanical stirrers, gas
recycle, or other options. In this reactor, the SRT and HRT are equal. Some advantages reported
include: (1) Ability to process feeds with high levels of suspended solids; (2) Enhanced contact
of microbiological species with the substrate. (3) Uniform distribution of the substrate
throughout the tank; (4) prevention of scum layer formation when properly mixed; (5) Even
temperature distribution throughout the substrate; (6) Maintenance of minimum levels of
inhibiting substances at any one point in the digester; (7) Avoids plugging, gas entrapment, and
channeling; (8) Easily modeled. Disadvantages include: (1) Large reactor requirements due to the
inability of this digester to operate with a mean SRT great than the HRT; (2) Large power
requirements to operate adequate mixing mechanisms; (3) Difficult to achieve complete mixing
at commercial scale; (4) Incompletely digested substrate leaving the system; (5) No separation of
acid-forming bacteria and methane-producing bacteria, thereby limiting the optimum
environmental requirements of the different bacteria; (6) Loss of microorganisms with the
digester effluent.

31

Figure 1.6. Continuously stirred tank reactor (Chynoweth and Isaacson, 1987)
(Source: http://www.daviddarling.info/encyclopedia/A/AE_anaerobic_digestion.html)

1.4.3. Plug flow reactors (PFRs)

The PFRs shown in Figure 1.7 typically receive feed at one end and remove effluent from
the other. Such reactors are not mechanically mixed and ideally the liquid and solid portions of
the biomass enter and exit the reactor at approximately the same time. Some vertical mixing of
the digester contents occur during gas production. However, longitudinal mixing is minimal,
thereby promoting phase separation. The passive settling of solids, therefore, results in longer
SRT than HRT. Advantages of this digestion scheme are: (1) Simplicity and reduced energy input,
as mechanical mixing is not required; (2) Very synergistic with manure characteristics and
therefore, often most economical for farm applications; (3) High stability; (4) Best suited for

32

well-suspended, high-solids
solids wastes. Disadvantages are: (1) Solids may settle out, reducing the
effective reactor volume; (2) Need for solids/microorganism recovery; (3) Lower efficiencies for
cold feedstocks requiring substantial preheating to mesophilic or thermophilic
thermophilic temperatures
because of high surface-volume
volume ratios; (4) Difficult to maintain uniform temperatures and
substrate conditions; (5) Formation of thick crusts.

Figure 1.7.. Plug flow reactor (Chynoweth and Isaacson, 1987)

(Source: http://climatetechwiki.org/technology/jiqweb-anbt
http://climatetechwiki.org/technology/jiqweb anbt )

33

1.4.4. Upflow anaerobic sludge blanket (UASB) digesters

The UASB digester shown in Figure 1.8 features a bed of granular and flocculated sludge
suspended by an upflow of wastewater. Gas bubbles from the produced gas and upflow
wastewater provide mixing. The design should be careful that some of the influent may bypass
the sludge-bed zone via cracks and channels. As the liquid moves through the granular and
flocculated sludge, increased SRT is achieved relative to the HRT. Advantages of this system
include: (1) Simple construction with the exception of the gas/solids separator; (2) Higher
loading rates are obtainable; (3) Low concentrations of suspended solids possible in influent; (4)
No mechanical mixing or costly support media required; (5) Low effluent suspended solids; (6)
Improved environmental conditions for the microorganisms by separating acid and methane
bacteria. Disadvantages are: (1) Requires effective gas and liquid separator; (2) Needs efficient
means of distributing the feed over the bottom of the tank; (3) May lose solids and
microorganisms in the effluent at high solids loading rates; (4) May lose a portion of the sludge
bed during a hydraulic surge or toxic upset; (5) Requires effluent recycle for bed expansion.

34

Biogas

Effluent

Granular
sludge

Feed

Figure 1.8. Up-flow anaerobic sludge blanket reactor (Chynoweth and Isaacson, 1987)
(Source: http://www.shi.co.jp/ejsite/no167/04.html)

1.4.5. Two-phase digesters

The two-phase digesters shown in Figure 1.9 are carried out by two groups of bacteria
(methanogenic and nonmethanogenic) that differ significantly with respect to physiology,

35

nutritional requirements, growth and metabolic characteristics, environmental optima, and

sensitivity to environmental stress. It is also possible to optimize mixing conditions for different
groups of bacteria in two reactors. This is because nonmethanogenic bacteria prefer vigorous
mixing while methanogenic bacteria prefer minimal mixing (Vavilin and Angelidaki, 2005). The
fermentation products produced in the hydrolic/acidogenic digester can be added to a second
digester in which a large methanogenic bacterial population is promoted by operation at long
SRT. Besides the inherent advantages of individual digesters, two phase digestion has the
following additional advantages: (1) Ability to maintain appropriate densities of acid and
methane formers; (2) Maximization of conversion rates through independent control of
temperature, pH, oxidation-reduction potential, biomass recycle, retention time, and other
parameters for each phase; (3) Enrichment of the product gas with methane; (4) Greater stability
with respect to feedstock loading, pH, and toxic shock; (5) Higher solids reduction at lower
retention time and, therefore, reduced total volume per unit of influent processed; (6) Reduced
gas clean-up costs where required, due to the increased methane content of the product gas. The
main disadvantage is the cost of the more complex system. Capital expenditure for the additional
equipment, controls, and labor requirements may be higher than with conventional system.

36

Figure 1.9. Two-phase digesters (Chynoweth and Isaacson, 1987)

(Source: http://www.eoi.es/blogs/syafrinasharif/)

37

1.5. Black Box Models to Simulate Anaerobic Digesters

Despite several attempts to control the AD process, anaerobic digestion remains a kind of
black-box (Holubar et al., 2000). The main reason for this situation is that the mechanisms ruling
these processes are not adequately understood to formulate reliable nonlinear mathematical
models because AD is a complex biological process involving decomposition by several major
populations of microorganisms.
As an alternative, artificial neural networks are claimed to have a distinctive advantage over
some other nonlinear estimation methods used for bio-processes, because they do not require any
prior knowledge about the structure of the relationships that exist between important variables.
All that is needed is to specify the architecture of the net, and to feed it with sufficient and
consistent information. The model architecture and components given by Prendeci et al.
(Perendeci et al., 2008) is shown in Figure 1.10. It consists of five key components; inputs and
outputs, database and preprocessor, a fuzzy system generator, a fuzzy inference system, and an
adaptive neural network representing the fuzzy system. Input and output variables are selected or
generated from the variables commonly used for system description. Database containing system
performance information is a prerequisite for the model development. Generally, it is developed
by collecting the data of regularly monitored variables. For example, the effluent comes from the
designated sewers of communities or other residential area, only four effluent characteristics, i.e.,
suspended solids (SS), biochemical oxygen demand (BOD), chemical oxygen demand (COD)
and true color were regulated according to Effluent Standard. Meanwhile, in order to save cost,

38

effluent quality investigation from wastewater treatment plant were only carried out to meet
regulation standard, so their investigation data were few and incomplete compared with general
study cases. Under this situation, the effluent quality trend could not be predicted appropriately
using some numerical models, especially mechanism models. Some soft computation techniques,
such as artificial neural network, in which the mechanism reactions can be ignored are available
presently and applied in biological wastewater treatment process (Abu Qdais et al., 2010;
Chorukova and Simconov, 2008; Holubar et al., 2002; Horiuchi et al., 2001; Lee et al., 2005;
Sinha et al., 2002).
Pai et al. (Pai et al., 2009) applied two training algorithms to predict suspended solids (SSeff)
and chemical oxygen demand (CODeff) in the effluent from a hospital wastewater treatment plant.
One is artificial neural network (ANN) and another is adaptive neuro fuzzy inference system
(ANFIS). ANFIS's architecture consists of both ANN and fuzzy logic including linguistic
expression of membership functions (MFs) and ifthen rules. The mean absolute percentage
error (MAPE), correlation coefficient (R), mean square error (MSE), and root mean square error
(RMSE) values showed that the predicting performance of ANFIS prevailed. Figure 1.11(a) and
(b) shows the training and predicting results using ANFIS2-1 and ANN2-1, respectively.

39

Figure 1.10. Schematic diagram of (a) adaptive network-based fuzzy inference system (ANFIS)
models with all input variables and (b) inputoutput mapping structure of ANFIS model with
only on-line input variables (Pai et al., 2009)

40

Figure 1.11. Prediction results of CODeff. (a) ANFIS2-1 and (b) ANN2-1 (Pai et al., 2009)

The advantage of artificial neural network is that the reaction mechanisms are not required
and a lot of parameters are not determined experimentally. Therefore it is possible to monitor the
treatment performance of the unit as well as operating conditions for a reliable treatment process,
especially for the full scale plant. However, this is also the disadvantage of this method and
makes it impossible to design and scale up reactor. Furthermore, a common criticism of artificial
neural networks is that they require a large diversity of training for real-world operation.

41

1.6. Phenomenological Models for Simulating Anaerobic Digesters

1.6.1. Ideal Models (CSTR and PFR)

By use of fundamental kinetics relationships and applying mass balances for the biomass,
explicit equations have been developed for various reactor configurations. The most common
reactor models are constructed for continuous stirred tank reactor (CSTR) and plug flow reactor
(PFR). Angelidakis comprehensive model (Angelidaki et al., 1999) and ADM1 (Batstone et al.,
2002a) assumed their reactors are CSTRs, so it is feasible to take into account the more complex
kinetics of AD process due to low requirement of calculation. Vavilin et al. (Vavilin and
Angelidaki, 2005; Vavilin et al., 2007; Vavilin et al., 2003) developed a distributed model of
anaerobic digestion of solid waste to describe the balance between the rates of polymer
hydrolysis and methanogenesis during the anaerobic conversion of rich and lean wastes in batch
and continuous-flow reactors. This model was supposed to characterize mixing intensity in
CSTR.
Mathematical reactor models for anaerobic fluidized bed bioreactors (AFBR) have been
developed as a CSTR (Worden and Donaldson, 1987), a plug-flow reactor (Bonnet et al., 1997)
and a plug flow with dispersion reactor (Buffiere et al., 1998). Due to the high-recirculation rates
of typical AFBRs, many previously existing mathematical models of the AFBR were developed
as CSTRs with no spatial gradients of the substrates and products along the height of the reactor.
During continuous operation, however, the biofilm can be thickened and nonhomogeneously
developed, and the density of the colonized media can significantly decrease. Thus, a CSTR

42

model may no longer be valid as the recycle ratio in an AFBR is significantly reduced or bed
characteristics are significantly changed during operation. To consider the concentration
gradients along the height of AFBR, many investigators have proposed a model for a fluidized
bed reactor as a plug-flow regime with axial dispersion (Seok and Komisar, 2003a).
The other high rate digester models were also developed. A dynamic model for the anaerobic
digestion of glucose in the periodic anaerobic baffled reactor (PAFB) was developed (Skiadas et
al., 2000). The key assumption of the model is that the hydraulic behavior of a four compartment
PABR is equivalent with the behavior of four CSTR in series. Ramakant et al. (Ramakant et al.,
2002) made a material balance within the biofilm and applying Ficks law of molecular diffusion
and add flow rate into Monode equation to simulate the anaerobic moving bed reactor. The leach
bed process was modeled by coupling a moisture flow model with a biological reaction and a
physico-chemical equilibrium model (Nopharatana et al., 2003). The leach bed, made from
compacted municipal solid waste (MSW), consisted of both large and small pore spaces. The
flow of water through this heterogeneous bed was not uniform. Instead the flow will consist of
channeled flow, which is defined as the flow of leachate through narrow pore channels at high
velocities, and Darcian flow which is defined as the flow that percolates through the small pores
between small particles or within particles.
1.6.2. Non-ideal models

Some reactor models for high rate anaerobic digesters were constructed by combination of
CSTR and PFR. The different zones of a UASB reactor are modeled as CSTR or PFR having

43

dead volume with bypass flow between the zones. The flow behavior within a zone mainly
depends on the concentration and characteristics of the biomass. Bolle et al. (Bolle et al., 1986)
divided the reactor into three compartments (Figure 1.12(A)): sludge bed, sludge blanket and
settler. The liquid flow in the sludge bed and the sludge blanket were described by completely
stirred tank reactor systems. Liquid flow in the internal settler was described by a plug flow
model. Wu and Hickey (Wu and Hickey, 1997) have developed a flow model to describe the flow
pattern in a UASB reactor. They modeled the sludge bed and blanket as a non-ideal CSTR by
using a combination of an ideal CSTR along with a dead zone and a bypass flow. This CSTR is
in series with a dispersed plug flow reactor (PFR) (non-ideal PFR) that represents the
clarification zone above the sludge blanket (Figure 1.12(B)). Since many studies on the
fluid-flow pattern in UASB reactors was found that both the sludge bed and the sludge blanket
can be described as separate, well-mixed flow regions, Vlyssides et al. divided the UASB into
two areas with idea mixers and the recirculation tank was considered as the third ideal mixer
(Vlyssides et al., 2007).

44

Figure 1.12. Block diagram of Bolle et al.s (A) (Bolle et al., 1986) and Wu and Hickys (B) (Wu
and Hickey, 1997)
CSTR and PFR models have been extensively applied into the design, scale up and
optimization of commercial reactors. A lot of modifications such as tank in series or axial
dispersion models were also developed to account for non-idealities. However, they are not
sufficient to predict the reactor performance which is determined not only by the length of time
that a fluid element spends in the vessel, but also by its surrounding environment during its travel
from inlet to outlet (Leng et al., 2009). It is necessary to obtain the quantitative information on
the flow pattern in digesters so that the effective mixing strategies can be operated to avoid a
failed design such as recirculation (back-mixing), short circuits, and dead zones.
It is believed by some experiments that mixing also promotes improved digester
performance by enhancing contact between microorganisms and substrates. Hoffmann et al.
(Hoffmann et al., 2008) think that mixing plays several essential roles during anaerobic digestion
of sludge, including enhancing substrate contact with the microbial community, improving pH

45

and temperature uniformity, preventing stratification and scum accumulation, facilitating the
removal of biogas from the digesting, and aiding in particle size reduction. In Seok et al.s
opinion (Seok and Komisar, 2003b), good mixing promotes the efficient transfer of substrates
and heat to the microorganisms, maintains uniformity in other environmental factors and assures
effective use of the entire reactor volume by preventing stratification and formation of dead spots
and prevents pockets of the VFA from forming. Smith et al. (Smith et al., 2005) made a
sensitivity analysis of the hydraulic parameters showed that increasing dead zone volume and
bypass flow significantly reduced digester performance at pathogen removal, whereas increased
mixing improved pathogen destruction. Effects of mixing could depend on digester size and
configuration, feed composition and particle size, method of mixing, and a number of factors.
Mixing requirements are related to the viscosity of the culture and too many other factors, such
as density, particle size and shape, and water content. Dead zone occur unless the mixing
equipment, reactor configuration, and energy input can all be harmoniously utilized to achieve
the desired goal. Furthermore, over-frequent or excessive mixing can disrupt the anaerobic
microbes because microbial cells are susceptible to mechanical force (Morales-Barrera and
Cristiani-Urbina, 2006). Following convention, the many kinds of damaging forces on microbial
cells are collectively referred to here as shear force. Shear rate and shear stress are often used
to describe the cell damage. Typically, the shear rate values in airlift bioreactors and mechanical
stirred tanks are substantially lower than 105 s-1 that would be needed to damage cells if a 100
N/m2 shear stress value is taken as the threshold of mechanical damage (Chisti, 2001).

46

An understanding of process mechanisms and kinetics is essential for rational reactor design:
specifying operating conditions, and predicting methane production, system stability, and
effluent quality.

A number of mathematical models have been developed to provide more

detailed knowledge of the mechanisms governing the bio-chemical AD process. The first
dynamic mathematical models emerged in the late 1960s as an attempt to explain complex
behavior of anaerobic reactors (Andrews, 1969; Graef and Andrews, 1974). Since then, further
models of increasing complexity have been formulated along with the recognition of several
important inhibitions and interactions among various bacterial groups (Angelidaki et al., 1999;
Batstone et al., 2002b; Vavilin et al., 2007). Ideally, process models are suppose to describe the
qualitative and quantitative aspects of microbial reactions ranging from hydrodynamics and mass
transfer to population dynamics in different reactor configurations under different operational
and environmental conditions. However, the task of obtaining valid required kinetic constants is
complicated by the fact that anaerobic digestion is a complex multistage dynamics process
involving the concerted effort of several groups of bacteria, the composition of which varies in
an unknown manner with changes in retention time, feedstock, temperature, reactor type and
other operating conditions. It is further complicated by the lack of a valid and reliable method for
quantification of microbial cell mass in digesters containing insoluble substrates. The detail to
which modeling of anaerobic digestion is possible is limited by the lack of knowledge on the
specific bacteria involved, their metabolism and physiological limitations. Current modeling
efforts are, therefore, limited to construct a feasible model with a certain amount of

47

simplification, and certain aspects of reality will no longer captured by the model. The objective
of this review is to compile the information available in the scientific literature related to digester
design, scale-up, operation and optimization including kinetics, reactor models and mixing
models.
Since ADM1 model was published by IWA task group, it has been tested and proved their
success in simulating the anaerobic digestion of several organic wastes such as: industrial
wastewaters (Batstone and Keller, 2003); sludge from wastewater treatment plants (Blumensaat
and Keller, 2005; Parker, 2005); sewage sludge (Shang et al., 2005); black water from vacuum
toilet (Feng et al., 2006); olive mill solid wastes (Kalfas et al., 2006). There are two classes of
application of ADM1 model: one is applications of the standard model in a mixed tank, with the
recommended approach given in the STR, often using already implemented models, to assess
specific systems and; another is applications of the ADM1 in new distributed parameter
applications, often for theoretical analysis (Batstone et al., 2006).
However, ADM1 model was assumed to simulate a constant volume completely mixed
system (Batstone et al., 2002a). In real commercial scale, it is hard to find an idea mixing in any
reactors. The complex flow behavior in the anaerobic digesters would affect the accuracy of
ADM1 model in the prediction of large-scale digesters. Furthermore, ADM1 requires a large
number of input parameters due to the complex model structure leading to a high amount of both
kinetic and stoichiometric expressions. In most cases, only a limited amount of parameters is
significant for model performance, depending on the substrate chosen. The use of complex

48

models such as ADM1, although valuable for general process simulation, has severe
shortcomings if they are intended to be used for process control and optimization (Stamatelatou
et al., 2009). This is because there are difficulties in determining the numerous model parameters
(non-identifiability of parameters), while manipulating a large number of equations limits the
applicability for the dynamic analysis, process simulation and control. In addition, although the
model assumptions reflect quite well our current understanding of the physical processes
involved, many of the individual steps may actually be so fast so that they do not influence the
overall process dynamics.
1.6.3. Computational Fluid Dynamics (CFD) Models

Proper mixing of the AD is essential for providing an optimum performance. Mixing

provides intimate contact between the feed sludge and active biomass, yielding uniformity of
temperature, of substrate concentration, of other chemical, physical and biological aspects
throughout the digester, and preventing both the formation of surface scum layers and the
deposition of sludge on the bottom of the tank. Computational fluid dynamics (CFD) software
enables one to predict the effects that geometry, feed location, physical properties, and operating
conditions have on conditions in the vessel. Typical results predict velocity profiles, rates of
energy dissipation, concentrations, and flow streamlines as they would occur in the vessel. This
tool enables one to appreciate the good and bad features for each considered design. Therefore,
CFD has become effective method to assess the mixing effect on the AD process.
CFD is the numerical simulation of fluid motion, and the study of mixing is an obvious

49

application for CFD (Paul et al., 2004). The technique is very powerful and spans a wide range
of industrial and non-industrial application areas, such as aerodynamics of aircraft and vehicles,
power plants, chemical process engineering, meteorology and so forth (Versteeg and
Malalasekera, 1995). In the past decade, CFD has been used to simulate the flow behavior of
wastewater treatment units such as wastewater ponds, lagoons and tanks. Most of these
researches used single phase Euler approach to describe the flow behavior in digesters. The finite
volume method was a common numerical method to solve their problems.
CFD was applied to the design of wastewater ponds by Wood et al. (Wood et al., 1995;
Wood et al., 1998). They calculated the flow field in four different ponds typesrectangular
facultative, inlet baffle, outlet baffle and aeratedand concluded that current designs and
operating models pay little attention to the micro-scale effects within the treatment ponds. They
continued to simulate the effects of inlet formation, and basin inlet geometry and concluded that
the 2-D CFD models qualitatively show that inlet geometry has the biggest impact on flow
patterns in waste ponds. Shilton et al. (Shilton, 2000; Shilton and Mara, 2005) sought to
complement and extend on the work of Wood et al. (Wood et al., 1998) by presenting the results
of 3-D, turbulent model.

In order to evaluate the efficiency of an activated sludge reactor,

Karama et al. (Karama et al., 1999) used CFD code to simulate the anaerobic zone of a 5-stage
(anaerobic, anoxic, aerobic, secondary anoxic and secondary aerobic) Bardenpho reactor. The
dead zone in the wastewater pond was predicted so that they could increase mixing in this area.
Vega et al. (Vega et al., 2003) used a 2-D depth-integrated model MIKE 21 to simulate

50

hydrodynamic and advection-dispersion processes in a full-scale anaerobic pond located in

southwest Colombia.
CFD has also been applied to the study of lagoons, another common form of simple digester.
Salter et al. (Salter et al., 2000) focused on the hydraulic regime in facultative lagoons and
carried out two consecutive simulations with and without baffles; the first established steady
flow conditions, and the second used a chemical species transport model to obtain the Residence
Time Distribution (RTD). Baleo et al. (Baleo et al., 2001) employed two different numerical
methods to obtain theoretical predictions of residence time distributions in a lagoon. The first
consisted of solving a transport equation for the local mean age of the fluid, which is the average
time that a fluid particle takes to reach any point of the domain from a supply inlet. The second
consisted of injecting a virtual particle stream (i.e. a fluid particle having the same density as the
surrounding fluid and treated numerically as a tracer) and measuring the time elapsed between
the injection and the termination of the trajectory using a Lagrangian reference frame. Although
it is easy to construct and operate ponds and lagoons, the biggest drawback is the large surface
area (Wang et al., 2006) which presents hydrodynamic problems including short-circuiting,
stratification and solid settling that are serious barriers against high efficient digestion.
Tanks are relatively small compared with ponds and lagoons. Small size permits the use of a
more flexible mixing strategy to ensure effective mass and heat transfer in tank digesters.
Vesvikar et al. (Vesvikar and Al-Dahhan, 2005; Vesvikar et al., 2005) numerically simulated the
flow characteristics of a mimic flat bottom digester which was mixed by gas recirculation.

51

It

was shown that the gas flow rate has no effect on the dead volume, and that dead volume
decreases with increasing draft tube diameter. Wu and Chen (Wu and Chen, 2008) took slurry
circulation into account to obtain flow fields in lab-scale, scale up, and pilot-scale anaerobic
digesters and concluded that power input per unit digester volume increases logarithmically for
the scale-up digesters. Terashima et al. (Terashima et al., 2009) developed a new parameter, the
uniformity index (UI), to describe mixing performance in the full scale digester. If the high solids
digester requires more efficient mixing, mechanical mixing will be the best choice to keep
substrate close contact with microorganism (Karim et al., 2005). Hoffmann et al. (Hoffmann et
al., 2008) simulated mechanical mixing only in the low solid digester (TS<5%) treating animal
manure. The digester solution domain was reduced to one third of the vessel volume. Under
these conditions, the fluid property of animal manure is regarded as Newtonian and constant
viscosity was used in the CFD simulation of the A-310 impeller. For total solid greater than 5%
in manure slurries, many experiments have been done to identify rheological properties and
behavior in these media (Achkari-Begdouri and Goodrich, 1992; Chen, 1986; Landry et al., 2004;
Moeller and Torres, 1997; O'Neil, 1985). Since flow field and power consumption are greatly
affected by manures rheological properties, Wu and Chen (Wu and Chen, 2008) introduced a
mathematical model with non-Newtonian fluid theory to describe the mixing of slurry
recirculation.
However, bio-fluid has a very complex flow behavior. It is not sufficient to use single phase
model to obtain the flow characteristics in anaerobic digesters, even if the concept of

52

non-Newtonian fluid was applied into the relative research, because there are the phenomena of
segregation and aggregation which can be observed in the waste slurry. These phenomena
significantly affect the contact between the feed sludge and active biomass, and the
microorganism retention, subsequently the productivity of biogas. Furthermore non-Newtonian
multiphase flows differ greatly from flows involving smooth spherical particles in the Newtonian
fluid. The latter have been studied widely, while bio-particles non-Newtonian multiphase flows,
despite their widespread importance, have received limited attention. As a result, industrial
biomass processes are being limited and their potential is not being realized. Thus, further work
is needed on multiphase flow of biomass materials to understand and optimize industrial energy
and material conversion processes (Cui and Grace, 2007). The extreme properties and
heterogeneous nature of biomass particles create challenges in processes involving biomass
particles. Research in this area is difficult, but potentially rewarding in terms of aiding the ability
of these materials to be used effectively.
1.7. Conclusions

The microbiological process involved in AD is extremely complex and still requires

substantial research in order to be fully understood. Technical facilities for the production of
biogas are thus generally scaled in a purely empirical manner. The efficiency of the process,
therefore, corresponds to the optimum only in the rarest cases. An optimal production of biogas,
as well as a stable plant operation requires detailed knowledge of the biochemical processes in
the fermenter. The use of mathematical models can help to achieve the necessary deeper

53

understanding of the AD process.

This review shows that various models describing the anaerobic digestion process have been
developed. These models are related not only to improve anaerobic digestion but the whole chain
of operations of the wastes treatment plants. The first effort was focused on the rate limiting
models which were simple and readily usable. However, their large variety and explicit
developments for specific substrates was often an obstacle for a more widespread use. For
complex substrates, it is hard to identify the rate limiting at different conditions in the whole
process. Sometimes, the intermediate fermentation products are also very important to assess the
capability of anaerobic digesters. ADM1 represents the currently most comprehensive model of
the AD process which provides a common basis for the future development of the kinetics
models. However, there are still lot improvements for the further research due to the complex
model structure. For example, the molecular methods are expected to understand the interactions
and functions of anaerobic microorganisms in more detail. For a special anaerobic process, a
different pathway might be needed. It is just because of the complex reaction mechanisms in AD
which is hard to be completely understood, the black box models such as artificial neural
network were applied into the simulation of wastewater plants. It is also because lack of reaction
mechanisms, it would be better to use these model to control AD process instead of the design
and scale-up of digesters. CSTR and PFR models constructed the fundamental basis of the
different types of digesters with kinetics. Axial dispersion model was added to account for
non-idealities. In order to predict the reactor performance, fluid dynamics is needed to provide

54

local details in the digesters which can improve digester design. CFD was applied into the
simulation of anaerobic digester to study the mixing effects. From Newtonian fluid to
non-Newtonian fluid, the CFD application in AD process has gradually revealed the realistic
flow behavior in the digesters. However, it is not sufficient to use single phase model to describe
the complex bio-fluid system. In practical flows, it is difficult to isolate single phase from the
complex flow phenomena. Multiphase flow phenomena are frequently encountered in the design
of process and industrial plant in the anaerobic digestion such as dairy manure, municipal solid
waste, and wastewater. Therefore, multiphase flow models should be developed to understand
the phenomena of segregation and aggregation in digesters so that we can know more about the
mixing effect on the contact between microorganisms and substrates. Furthermore, CFD with
comprehensive kinetics should also be pursued for the AD processes, although such processes
are so complex, especially for multiphase systems, that progress is likely to be limited in the
short term. It is essential that modelers communicate effectively with experimentalists for the
mutual benefit of both groups in achieving progress in realistic model research for AD process.

55

1.8. References

Abu Qdais, H., Hani, K.B., Shatnawi, N., 2010. Modeling and optimization of biogas production
from a waste digester using artificial neural network and genetic algorithm. Resources
Conservation and Recycling, 54, 359-363.

Achkari-Begdouri, A., Goodrich, P., 1992. Rheological properties of Moroccan dairy cattle
manure. Bioresour Technol, 40, 149-156.
Amon, T., Amon, B., Kryvoruchko, V., Bodiroza, V., Potsch, E., Zollitsch, W., 2006. Optimising
methane yield from anaerobic digestion of manure: Effects of dairy systems and of
glycerine supplementation. International Congress Series, 1293, 217-220.
Andrews, J.F., 1969. A mathematical model for the continuous culture of microorganisms
utilizing inhibitory substrates. Biotechnology and Bioengineering, 10, 707723.
Angelidaki, I., Ellegaard, L., Ahring, B.K., 1993. A mathematical-model for dynamic simulation
of anaerobic-digestion of complex substrates - focusing on ammonia inhibition.
Biotechnology and Bioengineering, 42, 159-166.

Angelidaki, I., Ellegaard, L., Ahring, B.K., 1999. A comprehensive model of anaerobic
bioconversion of complex substrates to biogas. Biotechnology and Bioengineering, 63,
363-372.
Appels, L., Baeyens, J., Degreve, J., Dewil, R., 2008. Principles and potential of the anaerobic
digestion of waste-activated sludge. Progress in Energy and Combustion Science, 34,
755-781.

56

Azeiteiro, C., Capela, I.F., Duarte, A.C., 2001. Dynamic model simulations as a tool for
evaluating the stability of an anaerobic process. Water Sa, 27, 109-114.
Bagi, Z., Acs, N., Balint, B., Horvath, L., Dobo, K., Perei, K.R., Rakhely, G., Kovacs, K.L., 2007.
Biotechnological intensification of biogas production. Applied Microbiology and
Biotechnology, 76, 473-482.

Baleo, J.N., Humeau, P., Le Cloirec, P., 2001. Numerical and experimental hydrodynamic studies
of a lagoon pilot. Water Research, 35, 2268-2276.
Batstone, D., Keller, J., Angelidaki, I., Kalyuzhnyi, S., Pavlostathis, S.G., Rozzi, A., Sanders,
W.T.M., Siegrist, H., Vavilin, V.A., 2002a. Anaerobic digestion model No.1 (ADM1).
IWA Publishing, London, UK.
Batstone, D.J., Keller, J., 2003. Industrial applications of the IWA anaerobic digestion model No.
1 (ADM1), pp. 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., 2002b. The IWA Anaerobic Digestion Model No 1
(ADM1). Water Science and Technology, 45, 65-73.
Batstone, D.J., Keller, J., Steyer, J.P., 2006. A review of ADM1 extensions, applications, and
analysis: 2002-2005, pp. 1-10.
Bjornsson, L., Murto, M., Jantsch, T.G., Mattiasson, B., 2001. Evaluation of new methods for the
monitoring of alkalinity dissolved hydrogen and the microbial community in anaerobic
digestion. Water Research, 35, 2833-2840.

57

Blumensaat, F., Keller, J., 2005. Modelling of two-stage anaerobic digestion using the IWA
Anaerobic Digestion Model No. 1 (ADM1). Water Research, 39, 171-183.
Bolle, W.L., van Breugel, J., van Eybergen, G.C., Kossen, N.W.F., van Gils, W., 1986. An
integral dynamic model for the UASB reactor. Biotechnology and Bioengineering, 28,
1621-1636.
Bolzonella, D., Fatone, F., Pavan, P., Cecchi, F., 2005. Anaerobic fermentation of organic
municipal solid wastes for the production of soluble organic compounds. Industrial &
Engineering Chemistry Research, 44, 3412-3418.

Bonnet, B., Dochain, D., Steyer, J.P., 1997. Dynamical modelling of an anaerobic digestion
fluidized bed reactor. Water Science and Technology, 36, 285-292.
Buffiere, P., Fonade, C., Moletta, R., 1998. Mixing and phase hold-ups variations due to gas
production in anaerobic fluidized-bed digesters: Influence on reactor performance.
Biotechnology and Bioengineering, 60, 36-43.

Buffiere, P., Loisel, D., Bernet, N., Delgenes, J.P., 2006. Towards new indicators for the
prediction of solid waste anaerobic digestion properties. Water Science and Technology,
53, 233-241.
Busch, G., Gromann, J., Sieber, M., Burkhardt, M., 2009. A new and sound technology for
biogas from solid waste and biomass. Water, Air, &amp; Soil Pollution: Focus, 9, 89-97.
Cakir, F.Y., Stenstrom, M.K., 2005. Greenhouse gas production: A comparison between aerobic
and anaerobic wastewater treatment technology. Water Research, 39, 4197-4203.

58

Chen, Y.R., 1986. Rheological properties of sieved beef-cattle manure slurry - rheological model
and effects of temperature and solids concentration. Agricultural Wastes, 15, 17-33.
Chen, Y.R., Hashimoto, A.G., 1980. Substrate utilization kinetic-model for biological treatment
processes. Biotechnology and Bioengineering, 22, 2081-2095.
Chisti, Y., 2001. Hydrodynamic damage to animal cells. Critical Reviews in Biotechnology, 21,
67-110.
Chorukova, E., Simconov, I., 2008. Neural and Hybrid Modelling of Biotechnological Process.
Studies in Informatics and Control, 17, 305-314.

Christ, O., Wilderer, P.A., Angerhofer, R., Faulstich, M., 2000. Mathematical modeling of the
hydrolysis of anaerobic processes. Water Science and Technology, 41, 61-65.
Chynoweth, D.P., Isaacson, R., 1987. Anaerobic Digestion of Biomass. Elsevier Applied Science,
London and New York.
Contois, D.E., 1959. Kinetics of bacterial growth: relationship between population density and
specific growth rate of continuous cultures. Journal of General Microbiology 21, 40-50.
Cui, H.P., Grace, J.R., 2007. Flow of pulp fibre suspension and slurries: A review. International
Journal of Multiphase Flow, 33, 921-934.

El-mashad, H.M., Zeeman, G., van Loon, W.K.P., Bot, G.P.A., Lettinga, G., 2003. Anaerobic
digestion of solid animal waste in an accumulation system at mesophilic and thermophilic
conditions, start up. Water Science and Technology, 48, 217-220.
Fedorovich, V., Lens, P., Kalyuzhnyi, S., 2003. Extension of Anaerobic Digestion Model No. 1

59

with processes of sulfate reduction, pp. 33-45.

Feng, Y., Behrendt, J., Wendland, C., Otterpohl, R., 2006. Implementation of the IWA anaerobic
digestion model No.1 (ADM1) for simulating digestion of blackwater from vacuum
toilets, pp. 253-263.
Flotats, X., Palatsi, J., Ahring, B.K., Angelidaki, I., 2006. Identifiability study of the proteins
degradation model, based on ADM1, using simultaneous batch experiments. Water
Science and Technology, 54, 31-39.

Fuentes, M., Mussati, M.C., Aguirre, P.A., Scenna, N.J., 2009. Experimental and theoretical
investigation of anaerobic fluidized bed biofilm reactors. Brazilian Journal of Chemical
Engineering, 26, 457-468.

Garcia-Heras, J.L., 2003. Reactor sizing, process kinetics and modelling of anaerobic digestion
of complex wastes. in: J. Mata-Alvarez (Ed.) Biomethanization of the organic fraction of
municipal solid wastes. IWA Publishing, TJ International Ltd., Padstow, Cornwall, UK,
pp. 21-62.
Gemmeke, B., Rieger, C., Weiland, P., 2009. Biogas-Messprogramm II - 61 Biogasanlagen im
Vergleich. FNR, Glzow.
Ghosh, S., 1981. Kinetics of acid-phase fermentation in anaerobic digestion. Biotechnology and
Bioengineering, 11, 301313

Graef, S.P., Andrews, J.F., 1974. Mathematical modeling and control of anaerobic digestion.
AIChE Symp Ser, 136, 101131.

60

Grau, P., Dohanyos, M., Chudoba, J., 1975. Kinetics of multicomponent substrate removal by
activated sludge. Water Research, 9, 637-642.
Griffin, M.E., McMahon, K.D., Mackie, R.I., Raskin, L., 1998. Methanogenic population
dynamics during start-up of anaerobic digesters treating municipal solid waste and
biosolids. Biotechnology and Bioengineering, 57, 342-355.
Gujer, W., Zehnder, A.J.B., 1983. Conversion processes in anaerobic-digestion. Water Science
and Technology, 15, 127-167.

Hoffmann, R.A., Garcia, M.L., Veskivar, M., Karim, K., Al-Dahhan, M.H., Angenent, L.T., 2008.
Effect of shear on performance and microbial ecology of continuously stirred anaerobic
digesters treating animal manure. Biotechnology and Bioengineering, 100, 38-48.
Holubar, P., Zani, L., Hager, M., Froschl, W., Radak, Z., Braun, R., 2000. Modelling of anaerobic
digestion using self-organizing maps and artificial neural networks. Water Science and
Technology, 41, 149-156.

Holubar, P., Zani, L., Hager, M., Froschl, W., Radak, Z., Braun, R., 2002. Advanced controlling
of anaerobic digestion by means of hierarchical neural networks. Water Research, 36,
2582-2588.
Horiuchi, J., Kikuchi, S., Kobayashi, M., Kanno, T., Shimizu, T., 2001. Modeling of pH response
in continuous anaerobic acidogenesis by an artificial neural network. Biochemical
Engineering Journal, 9, 199-204.

Inoue, S., Sawayama, S., Ogi, T., Yokoyama, S.Y., 1996. Organic composition of liquidized

61

sewage sludge. Biomass & Bioenergy, 10, 37-40.

Jain, S., Lala, A.K., Bhatia, S.K., Kudchadker, A.P., 1992. Modeling of hydrolysis controlled
anaerobic-digestion. Journal of Chemical Technology and Biotechnology, 53, 337-344.
Jeyaseelan, S., 1997. A simple mathematical model for anaerobic digestion process. Water
Science and Technology, 35, 185-191.

Kalfas, H., Skiadas, I.V., Gavala, H.N., Stamatelatou, K., Lyberatos, G., 2006. Application of
ADM1 for the simulation of anaerobic digestion of olive pulp under mesophilic and
thermophilic conditions, pp. 149-156.
Kaparaju, P., Serrano, M., Thomsen, A.B., Kongjan, P., Angelidaki, I., 2009. Bioethanol,
biohydrogen and biogas production from wheat straw in a biorefinery concept.
Bioresource Technology, 100, 2562-2568.

Karakashev, D., Batstone, D.J., Angelidaki, I., 2005. Influence of environmental conditions on
methanogenic compositions in anaerobic biogas reactors. Applied and Environmental
Microbiology, 71, 331-338.

Karama, A.B., Onyejekwe, O.O., Brouckaert, C.J., Buckley, C.A., 1999. The use of
Computational Fluid Dynamics (CFD). Technique for evaluating the efficiency of an
activated sludge reactor, pp. 329-332.
Karim, K., Klasson, K.T., Drescher, S.R., Ridenour, W., Borole, A.P., Al-Dahhan, M.H., 2007.
Mesophilic digestion kinetics of manure slurry. Applied Biochemistry and Biotechnology,
142, 231-242.

62

Karim, K., Klasson, K.T., Hoffmann, R., Drescher, S.R., DePaoli, D.W., Al-Dahhan, M.H., 2005.
Anaerobic digestion of animal waste: Effect of mixing. Bioresource Technology, 96,
1607-1612.
Khan, M.R., Yaqoob, M., 1997. Sugar beet by product as a potential source of dietary fibers.
Journal of the Chemical Society of Pakistan, 19, 83-84.

Kono, T., Asai, T., 1969. Kinetics of continuous cultivation. Biotechnology and Bioengineering,
11, 19-36.
Kryvoruchko, V., Machmuller, A., Bodiroza, V., Amon, B., Amon, T., 2009. Anaerobic digestion
of by-products of sugar beet and starch potato processing. Biomass & Bioenergy, 33,
620-627.
Krzystek, L., Ledakowicz, S., Kahle, H.J., Kaczorek, K., 2001. Degradation of household
biowaste in reactors. Journal of Biotechnology, 92, 103-112.
Landry, H., Lague, C., Roberge, M., 2004. Physical and rheological properties of manure
products. Appl Eng Agric, 20, 277-288.
Lee, M.W., Joung, J.Y., Lee, D.S., Park, J.M., Woo, S.H., 2005. Application of a
moving-window-adaptive neural network to the modeling of a full-scale anaerobic filter
process. Industrial & Engineering Chemistry Research, 44, 3973-3982.
Lehtomaki, A., Vavilin, V., Rintala, J., 2005. Kinetic analysis of methane production from energy
crops. in: B.K. Ahring, H. Hartmann (Eds.), Proceedings of the Fourth International
Symposium on Anaerobic Digestion of Solid Waste, Copenhagen, Denmark.

63

Leng, D.E., Katti, S.S., Atiemo-Obeng, V., 2009. Industrial Mixing Technology. in: L.F. Albright
(Ed.) Albright's Chemical Engineering Handbook. CRC Press of Taylor & Francis Group,
New York, USA.
Leven, L., Eriksson, A.R.B., Schnurer, A., 2007. Effect of process temperature on bacterial and
archaeal communities in two methanogenic bioreactors treating organic household waste.
Fems Microbiology Ecology, 59, 683-693.

Liebetrau, J., Kraft, E., Bidlingmaier, W., 28 September-1 October 2004. The influence of the
hydrolysis rate of co-substrates on process behaviour. in: S.G. Guiot (Ed.) Anaerobic
digestion X. Selected Proceedings of the 10th IWA Congress on Anaerobic Digestion.
IWA Publishing, London, UK, Montral, Canada.
Lokshina, L.Y., Vavilin, V.A., 1999. Kinetic analysis of the key stages of low temperature
methanogenesis. Ecological Modelling, 117, 285-303.
Lokshina, L.Y., Vavilin, V.A., Kettunen, R.H., Rintala, J.A., Holliger, C., Nozhevnikova, A.N.,
2001. Evaluation of kinetic coefficients using integrated Monod and Haldane models for
low-temperature acetoclastic methanogenesis. Water Research, 35, 2913-2922.
Lokshina, L.Y., Vavilin, V.A., Salminen, E., Rintala, J., 2003. Modeling of anaerobic degradation
of solid slaughterhouse waste - Inhibition effects of long-chain fatty acids or ammonia.
Applied Biochemistry and Biotechnology, 109, 15-32.

Lopez, I., Passeggi, M., Pedezert, A., Borzacconi, L., 2009. Assessment on the performance of a
series of two UASB reactors compared against one of the same total volume using

64

Anaerobic Digestion Model No 1 (ADM1). Water Science and Technology, 59, 647-652.
Lubken, M., Gehring, T., Wichern, M., 2010. Microbiological fermentation of lignocellulosic
biomass: current state and prospects of mathematical modeling. Applied Microbiology
and Biotechnology, 85, 1643-1652.

Masse, L., Masse, D.I., Kennedy, K.J., Chou, S.P., 2002. Neutral fat hydrolysis and long-chain
fatty acid oxidation during anaerobic digestion of slaughterhouse wastewater.
Biotechnology and Bioengineering, 79, 43-52.

Mata-Alvarez, J., Mace, S., Llabres, P., 2000. Anaerobic digestion of organic solid wastes. An
overview of research achievements and perspectives. Bioresource Technology, 74, 3-16.
Metcalf & Eddy, I., 2003. Wastewater Engineering: Treatment and Reuse McGraw-Hill,
Newyork, US.
Mikkelsen, L.H., Keiding, K., 2002. Physico-chemical characteristics of full scale sewage
sludges with implications to dewatering. Water Research, 36, 2451-2462.
Moeller, G., Torres, L.G., 1997. Rheological characterization of primary and secondary sludges
treated by both aerobic and anaerobic digestion. Bioresource Technology, 61, 207-211.
Monod, J., 1949. The Growth of Bacterial Cultures. Annual Review of Microbiology, 3, 371-394.
Morales-Barrera, L., Cristiani-Urbina, E., 2006. Removal of hexavalent chromium by
Trichoderma viride in an airlift bioreactor. Enzyme and Microbial Technology, 40,
107-113.
Mosche, M., Jordening, H.J., 1999. Comparison of different models of substrate and product

65

inhibition in anaerobic digestion. Water Research, 33, 2545-2554.

Mottet, A., Steyer, J.P., Deleris, S., Vedrenne, F., Chauzy, J., Carrere, H., 2009. Kinetics of
thermophilic batch anaerobic digestion of thermal hydrolysed waste activated sludge.
Biochemical Engineering Journal, 46, 169-175.

Nopharatana, A., Pullammanappallil, P.C., Clarke, W.P., 2003. A dynamic mathematical model
for sequential leach bed anaerobic digestion of organic fraction of municipal solid waste.
Biochemical Engineering Journal, 13, 21-33.

Novak, J.T., Sadler, M.E., Murthy, S.N., 2003. Mechanisms of floc destruction during anaerobic
and aerobic digestion and the effect on conditioning and dewatering of biosolids. Water
Research, 37, 3136-3144.

O'Neil, D.J., 1985. Rheology and mass/heat transfer aspects of anaerobic reactor design. Biomass,
8, 205216.
O'Rourke, J.R., 1968. Kinetics of anaerobic treatment at reduced temperatures. Stanford
University, Stanford, CA, USA.
O'Sullivan, C.A., Burrell, P.C., Clarke, W.P., Blackall, L.L., 2005. Structure of a cellulose
degrading bacterial community during anaerobic digestion. Biotechnology and
Bioengineering, 92, 871-878.

Pai, T.Y., Wan, T.J., Hsu, S.T., Chang, T.C., Tsai, Y.P., Lin, C.Y., Su, H.C., Yu, L.F., 2009. Using
fuzzy inference system to improve neural network for predicting hospital wastewater
treatment plant effluent. Computers & Chemical Engineering, 33, 1272-1278.

66

Parawira, W., Read, J.S., Mattiasson, B., Bjornsson, L., 2008. Energy production from
agricultural residues: High methane yields in pilot-scale two-stage anaerobic digestion.
Biomass & Bioenergy, 32, 44-50.

Parker, W.J., 2005. Application of the ADM1 model to advanced anaerobic digestion.
Bioresource Technology, 96, 1832-1842.

Paul, E.L., Atiemo-Obeng, V.A., Kresta, S.M., 2004. Handbook of industrial mixing. John Wiley
& Sons, New Jersey.
Pavlostathis, S.G., Giraldogomez, E., 1991. Kinetics of anaerobic treatment Critical Reviews in
Environmental Control, 21, 411-490.

Perendeci, A., Arslan, S., Celebi, S.S., Tanyolac, A., 2008. Prediction of effluent quality of an
anaerobic treatment plant under unsteady state through ANFIS modeling with on-line
input variables. Chemical Engineering Journal, 145, 78-85.
Qu, X., Vavilin, V.A., Mazeas, L., Lemunier, M., Duquennoi, C., He, P.J., Bouchez, T., 2009.
Anaerobic biodegradation of cellulosic material: Batch experiments and modelling based
on isotopic data and focusing on aceticlastic and non-aceticlastic methanogenesis. Waste
Management, 29, 1828-1837.

Ramakant, Satyanarayan, S., Kaul, S.N., 2002. Kinetics of an anaerobic moving bed reactor
system treating synthetic milk wastewater. Journal of Environmental Science and Health
Part a-Toxic/Hazardous Substances & Environmental Engineering, 37, 1737-1755.

Ristow, N.E., Sotemann, S.W., Wentzel, M.C., Loewenthal, R.E., Ekama, G.A., 2006. The effects

67

of hydraulic retention time and feed COD concentration on the rate of hydrolysis of
primary sewage sludge under methanogenic conditions. Water Science and Technology,
54, 91-100.
Salter, H.E., Ta, C.T., Williams, S.C., 2000. Three-dimensional computational fluid dynamic
modelling of a facultative lagoon. I W a Publishing, pp. 335-342.
Sanchez, J.B., Alonso, J.M.Q., Oviedo, M.D.C., 2006. Use of microbial activity parameters for
determination of a biosolid stability index. Bioresource Technology, 97, 562-568.
Scheper, T., Ahring, B.K., 2003. Biomethanation II. Springer-Verlag Berlin Heidelberg New
York, Berlin, Germany.
Seok, J., Komisar, S.J., 2003a. Integrated modeling of anaerobic fluidized bed bioreactor for
deicing waste treatment. I: Model derivation. Journal of Environmental Engineering-Asce,
129, 100-109.
Seok, J., Komisar, S.J., 2003b. Integrated modeling of anaerobic fluidized bed bioreactor for
deicing waste treatment. II: Simulation and experimental studies. Journal of
Environmental Engineering-Asce, 129, 110-122.

Shang, Y., Johnson, B.R., Sieger, R., 2005. Application of the IWA Anaerobic Digestion Model
(ADM1) for simulating full-scale anaerobic sewage sludge digestion, pp. 487-492.
Sharrer, M.J., Tal, Y., Ferrier, D., Hankins, J.A., Summerfelt, S.T., 2007. Membrane biological
reactor treatment of a saline backwash flow from a recirculating aquaculture system.
Aquacultural Engineering, 36, 159-176.

68

Shilton, A., 2000. Potential application of computational fluid dynamics to pond design. I W a
Publishing, pp. 327-334.
Shilton, A., Mara, D.D., 2005. CFD (computational fluid dynamics) modelling of baffles for
optimizing tropical waste stabilization Cn pond systems, pp. 103-106.
Shimizu, T., Kudo, K., Nasu, Y., 1993. Anaerobic waste-activated sludge-digestion - a
bioconversion mechanism and kinetic-model. Biotechnology and Bioengineering, 41,
1082-1091.
Sinha, S., Bose, P., Jawed, M., John, S., Tare, V., 2002. Application of neural network for
simulation of upflow anaerobic sludge blanket (UASB) reactor performance.
Biotechnology and Bioengineering, 77, 806-814.

Skiadas, I.V., Gavala, H.N., Lyberatos, G., 2000. Modelling of the Periodic Anaerobic Baffled
Reactor (PABR) based on the retaining factor concept. Water Research, 34, 3725-3736.
Smith, S.R., Lang, N.L., Cheung, K.H.M., Spanoudaki, K., 2005. Factors controlling pathogen
destruction during anaerobic digestion of biowastes. Waste Management, 25, 417-425.
Speece, R.E., 1983. Anaerobic biotechnology for industrial wastewater treatment. Environmental
Science & Technology, 17, 416A-427A.

Stafford, D.A., Hawkes, D.L., Horton, R., 1980. Methane production from waste organic matter.
CRC Press.
Stamatelatou, K., Syrou, L., Kravaris, C., Lyberatos, G., 2009. An invariant manifold approach
for CSTR model reduction in the presence of multi-step biochemical reaction schemes.

69

Application to anaerobic digestion. Chemical Engineering Journal, 150, 462-475.

Switzenbaum, M.S., Giraldogomez, E., Hickey, R.F., 1990. Monitoring of the anaerobic methane
fermentation process. Enzyme and Microbial Technology, 12, 722-730.
Terashima, M., Goel, R., Komatsu, K., Yasui, H., Takahashi, H., Li, Y.Y., Noike, T., 2009. CFD
simulation of mixing in anaerobic digesters. Bioresource Technology, 100, 2228-2233.
Tomei, M.C., Braguglia, C.M., Cento, G., Mininni, G., 2009. Modeling of Anaerobic Digestion
of Sludge. Critical Reviews in Environmental Science and Technology, 39, 1003-1051.
Tong, X., Smith, L.H., McCarty, P.L., 1990. Methane fermentation of selected lignocellulosic
materials. Biomass, 21, 239-255.
Vavilin, V.A., Angelidaki, I., 2005. Anaerobic degradation of solid material: Importance of
initiation centers for methanogenesis, mixing intensity, and 2D distributed model.
Biotechnology and Bioengineering, 89, 113-122.

Vavilin, V.A., Fernandez, B., Palatsi, J., Flotats, X., 2008. Hydrolysis kinetics in anaerobic
degradation of particulate organic material: An overview. Waste Management, 28,
941-953.
Vavilin, V.A., Lokshina, L.Y., Flotats, X., Angelidaki, I., 2007. Anaerobic digestion of solid
material: Multidimensional modeling of continuous-flow reactor with non-uniform
influent concentration distributions. Biotechnology and Bioengineering, 97, 354-366.
Vavilin, V.A., Lokshina, L.Y., Jokela, J.P.Y., Rintala, J.A., 2004. Modeling solid waste
decomposition. Bioresource Technology, 94, 69-81.

70

Vavilin, V.A., Lokshina, L.Y., Rytov, S.V., Kotsyurbenko, O.R., Nozhevnikova, A.N., Parshina,
S.N., 1997. Modelling methanogenesis during anaerobic conversion of complex organic
matter at low temperatures. Water Science and Technology, 36, 531-538.
Vavilin, V.A., Rytov, S.V., Lokshina, L.Y., 1996. A description of hydrolysis kinetics in anaerobic
degradation of particulate organic matter. Bioresource Technology, 56, 229-237.
Vavilin, V.A., Rytov, S.V., Lokshina, L.Y., Rintala, J.A., Lyberatos, G., 2001. Simplified
hydrolysis models for the optimal design of two-stage anaerobic digestion. Water
Research, 35, 4247-4251.

Vavilin, V.A., Rytov, S.V., Pavlostathis, S.G., Jokela, J., Rintala, J., 2003. A distributed model of
solid waste anaerobic digestion: sensitivity analysis. Water Science and Technology, 48,
147-154.
Veeken, A., Hamelers, B., 1999. Effect of temperature on hydrolysis rates of selected biowaste
components. Bioresource Technology, 69, 249-254.
Vega, G.P., Pena, M.R., Ramirez, C., Mara, D.D., 2003. Application of CFD modelling to study
the hydrodynamics of various anaerobic pond configurations, pp. 163-171.
Versteeg, H.K., Malalasekera, W., 1995. An Introduction to Computational Fluid Dynamics
Addison Wesley Longman Limited, Harlow, England.
Vesvikar, M.S., Al-Dahhan, M., 2005. Flow pattern visualization in a mimic anaerobic digester
using CFD. Biotechnology and Bioengineering, 89, 719-732.
Vesvikar, M.S., Varma, R., Karim, K., Al-Dahhan, M., 2005. Flow pattern visualization in a

71

mimic anaerobic digester: experimental and computational studies, pp. 537-543.

Vieitez, E.R., Ghosh, S., 1999. Biogasification of solid wastes by two-phase anaerobic
fermentation. Biomass & Bioenergy, 16, 299-309.
Vlyssides, A., Barampouti, E.M., Mai, S., 2007. An alternative approach of UASB dynamic
modeling. Aiche Journal, 53, 3269-3276.
Wang, L.K., Hung, Y.-T., Lo, H.H., Yapijakis, C., 2006. Waste treatment in the food processing
industry CRC Press, London, the U.K.
Wang, Q.H., Kuninobu, M., Ogawa, H.I., Kato, Y., 1999. Degradation of volatile fatty acids in
highly efficient anaerobic digestion. Biomass & Bioenergy, 16, 407-416.
Weiland, P., 2010. Biogas production: current state and perspectives. Applied Microbiology and
Biotechnology, 85, 849-860.

Wood, M.G., Greenfield, P.F., Howes, T., Johns, M.R., Keller, J., 1995. Computational fluid
dynamic modeling of waste-water ponds to improve design. Pergamon-Elsevier Science
Ltd, pp. 111-118.
Wood, M.G., Howes, T., Keller, J., Johns, M.R., 1998. Two dimensional computational fluid
dynamic models for waste stabilisation ponds. Water Research, 32, 958-963.
Worden, R.M., Donaldson, T.L., 1987. Dynamics of a biological fixed film for phenol
degradation in a fluidized-bed bioreactor. Biotechnology and Bioengineering, 30,
398-412.
Wu, B.X., Chen, S.L., 2008. CFD simulation of non-Newtonian fluid flow in anaerobic digesters.

72

Biotechnology and Bioengineering, 99, 700-711.

Wu, M.M., Hickey, R.F., 1997. Dynamic model for UASB reactor including reactor hydraulics,
reaction, and diffusion. Journal of Environmental Engineering-Asce, 123, 244-252.

73

CHAPTER 2 : NUMERICAL SIMULATION OF MECHANICAL MIXING IN HIGH

SOLID ANAEROBIC DIGESTER
2.1. Abstract

Computational fluid dynamics (CFD) was employed to study mixing performance in high
solid anaerobic digester (HSAD) with A-310 impeller and helical ribbon. A mathematical model
was constructed to assess flow fields. Good agreement of the model results with experimental
data was obtained for the A-310 impeller. A systematic comparison for the interrelationship of
power number, flow number and Reynolds number was simulated in a digester with less than
5%TS and 10%TS (Total Solids). The simulation results suggested a great potential for using the
helical ribbon mixer in the mixing of high solids digester. The results also provided quantitative
confirmation for minimum power consumption in HSAD and the effect of share rate on
bio-structure.
Keywords: mixing; computational fluid dynamics (CFD); high solid anaerobic digester (HSAD)

74

2.2. Introduction

High Solid Anaerobic Digester (HSAD) has great potential for efficient conversion of
agricultural, municipal, and household wastes into green energy.

Increasing solids

concentration in the digester provides an option to substantially reduce the size of the reactor,
thus reducing the footprint of the system. If the reactor volume could be reduced significantly,
the economics of anaerobic digestion of these materials could also be improved.
However, high viscosity caused by high total solids concentration can severely reduce the
mass and heat transfer among enzymes, bacteria and substrates in the digester. Adequate mixing
is required to accelerate the diffusion processes thus improving the reaction kinetics in HSADs.
Mechanical agitation has been proved to be the most efficient approach in terms of energy input
and mixing performance (Karim et al., 2005). The study of flow fields caused by mechanical
mixing will help to improve the design of HSAD on a commercial scale.
Computational Fluid Dynamics (CFD) can assist in studying mixing through numerical
simulation of fluid motion (Paul et al., 2004). CFD can allow determination of flow
characteristics of a liquid/slurry system with a much lower cost than experimentation. Its
advantages are even greater in applications where it is difficult to experimentally detect the
mixing parameters. Consequently, CFD has been very effectively applied in a wide range of
industrial and non-industrial areas, such as aerodynamics of aircraft and vehicles, power plants,
chemical process engineering, meteorology and so forth (Ding et al., 2010; Norton et al., 2007;
Yin et al., 2010). In the field of environmental engineering, CFD has been used to simulate the

75

flow behavior of wastewater treatment units such as wastewater ponds(Shilton et al., 2008) ,
lagoons (Pougatch et al., 2007) and tanks (Terashima et al., 2009; Wu and Chen, 2008).
Nonetheless, the application of CFD in HSAD is still limited. Hoffmann et al. (Hoffmann et
al., 2008) simulated mechanical mixing in a low solid digester (TS < 5%) treating animal manure.
The digester solution domain was reduced to one third of the vessel volume. Under these
conditions, the fluid property of animal manure is regarded as Newtonian and constant viscosity
was used in the CFD simulation of an A-310 impeller. For total solid greater than 5% in manure
slurries, many experiments have been performed to identify the rheological properties and
behaviors in these media (Achkari-Begdouri and Goodrich, 1992; Landry et al., 2004; Moeller
and Torres, 1997). Since flow field and power consumption are greatly affected by manures
rheological properties and higher TS, a wide range of flow regimes from turbulence to laminar
flow should be studied to describe the mixing mechanism of HSAD.
To date, lab-, pilot- and commercial-scale high solid anaerobic digesters have been
constructed in which mixing is a key factor (Zhu et al., 2010). Research on mixing using CFD
will provide theoretical support for design, optimization and scale-up of HSAD systems. A
technical need in the study of mechanical mixing in HSAD is the consideration of A-310 and
helical ribbon designs. The A-310 is an axial flow impeller. It is regarded as a high-efficiency
hydrofoil impeller with minimal energy dissipation in the vicinity of the impeller (Kilander and
Rasmuson, 2005). It is often designed for turbulent flow and is well suited for flow-controlled
processes, such as liquid-solid mixing, miscible liquids and blending. The

76

helical ribbon

impeller (Paul et al., 2004), on the other hand, is typically used for industrial applications where
the viscosity is up to 25000Pas. Mixing in very viscous systems can be a formidable task, as
there are no turbulent eddies to help distribute components. Because of the high matrix viscosity,
diffusion coefficients for even very small molecules are exceedingly low (Paul et al., 2004).
Due to their ability to keep the entire vessel contents circulating, close-clearance impellers are
very well suited to mix highly viscous liquids. Among close-clearance impellers, helical ribbon
impellers are recognized to be the most efficient system applied in polymer industries. Since
HSAD is a high viscous and shear sensitive system, application of helical ribbon impellers might
improve the distribution of enzyme and microorganism into substrate.
This study applied non-Newtonian fluid theory to a 3-D mechanical mixing model to
describe flow characteristics in a HSAD. The theoretical model was validated against
experimental data from the literature (Hoffmann et al., 2008) in which the experiments were
conducted with lab-scale digesters made from clear PVC with a wet volume of 4.5 L. Computer
Automated Radioactive Particle Tracking (CARPT) was used to measure the flow fields.
The flow field, impeller pumping capacity, shear stress and power of impeller A-310 in a
HSAD were determined for TS = 10%, and compared with that of TS < 5%. Mixing performance
was assessed both qualitatively and quantitatively to determine the effects of rheological
properties and behavior. Predictions of mixing in higher viscous conditions (TS > 15%) by the
helical ribbon were compared with that of the A-310 impeller. Calculations of the unit energy
input of helical ribbon for HSAD and power optimization were performed based on Rivards

77

minimal mixing requirements (Rivard et al., 1990).

2.3. Methods
2.3.1. Mixing model

Simulating mechanical mixing of anaerobic digesters requires that the model be divided into
multiple fluid cell zones, and separating the zones with interface boundaries. The inner zone
includes the impellers which are responsible for the motion of fluid in digesters. The outer zone
contains no moving parts, and is passively driven by the impeller. Zones which contain moving
components are simulated with moving reference frame equations, whereas the stationary zone is
simulated using the stationary frame equations. A steady-state modeling method called the
Multiple Reference Frame (MRF) (Luo et al., 1994) was developed to solve this type of problem.
Table 2.1 presents actual dimensions of the digesters and mixers.

Table 2.1. Geometrical details for digesters with impeller A-310 and helical ribbon (Hoffmann et
al., 2008; Karim et al., 2005)
Digesters

Impeller A-310

Helical Ribbon

Reactor diameter(mm)

152

152

Reactor height (mm)

370

370

Liquid height (mm)

194

220

Cone height (mm)

26

Cone angle (deg)

25

Impeller diameter (mm)

62

120

Number of blades

78

The high level of total solids in liquid manure and waste slurries has a significant impact on
the fluid properties by significantly altering the flow characteristics in anaerobic digesters.
Turbulent flow is suppressed by high viscosity, and laminar flow prevails in HSAD systems. The
standard - turbulence model is used when Re > 1000; and the low-Re-- turbulence model is
used to describe transition flow when 10 < Re < 1000; when Re < 10, the laminar model is used
(Metzner and Otto, 1957). The non-Newtonian standard - turbulence model and the
low-Re-- turbulence model are activated by text command/turbulence-expert in FLUENT
v6.3.26, respectively. The CFD governing equations are provided in the FLUENT documentation
(FLUENT, 2006).
Liquid manure was described using the non-Newtonian Fluid Model:
ui

ij =

x j

u j

xi

(2.1)

where is non-Newtonian viscosity, which is only considered to be a function of the shear rate,
.

The viscosity in the non-Newtonian power law is expressed as:

= K n 1e

T0

(2.2)

where K is the consistency coefficient (Pa sn), is the shear rate (s-1), n is the power-law index
that determines the class of fluid, T0 is the reference temperature (K), and T is the slurry
temperature (K).
Conventionally, the Reynolds number (Re) is defined in Newtonian fluid, with the agitation

79

formula

Re =

NDm2

(2.3)

where Re is the Reynolds number for Newtonian fluid, N is agitator speed (rps) and Dm is
impeller diameter (m).
In this study, the flow characteristics are caused by agitators, and therefore the Reynolds
number can be accordingly described as (Metzner and Otto, 1957)
n
Dm2 N 2 n n
Re* =
8 6n + 2
K

(2.4)

where Re* is agitation Reynolds number for non-Newtonian fluid.

Impeller power can be directly obtained from impeller torque, and the power number is
defined as (Wu and Pullum, 2000)

P = 2 NM
Np =

P
N 3 Dm5

(2.5)
(2.6)

where P is impeller power (W), M is impeller torque ( Nm) and Np is power number.
To calculate the flow rate produced by the impeller, a surface needs to be created for the
discharge region. For an axial flow impeller, this surface is circular. By integrating the total
outflow through this surface, the flow rate and subsequently the flow number can be obtained
(Patwardhan, 2001; Paul et al., 2004).
Q = v z dA

80

(2.7)

Nq =

Q
NDm3

(2.8)

where Q is flow rate (m3/s), A is surface cell area (m2), and Nq is flow number.
2.3.2. Numerical considerations

The HSAD mixing model was solved in FLUENT (v.6.3.26). This CFD software is based on
the Finite Volume Method (FVM) which is the most well-established and thoroughly validated
CFD technique for general purpose. The unstructured grids were generated by GAMBIT
(v.2.3.16). The number of the grids for the digester with impeller A-310 was 179, 350 cells and
the number of the grids for the digester with helical ribbon was 418, 353. The minimum cell
volume was 4.3710-6 cm3 and the maximum cell volume was 0.607 cm3. The finest grids were
generated near the impeller wall, and the grids for the tank wall were also fine enough to offset
the effect of large gradient of variables. The mesh growth factor was set as 1.2 to create a
high-quality mesh. In this grid conditions, grid independency can be guaranteed by checking
with higher resolution grids (432, 561 cells). The pressure-based approach was used to simulate
incompressible manure slurry flow in anaerobic digesters. In order to obtain a converged
numerical solution, a segregated loop was carried out iteratively. The SIMPLE algorithm was
used to solve velocity-pressure coupled differential equations. The third-order convection
scheme was conceived from the original MUSCL (Monotone Upstream-Centered Schemes for
Conservation Laws) by blending scheme and second-order upwind scheme. The governing
equations were discretized into algebraic equations and then solved numerically. No-slip

81

boundary conditions were applied on the walls. The wall temperature was set as 35C. The
standard wall functions were used to calculate the variables at the near-wall. In this study, the
flow fields were calculated for a wide range of Reynolds number. Since the logarithmic law is
known to be valid for 30 < y* < 300, the criterion y* for higher agitator speeds (>2500 rpm in
5%TS slurry) was satisfied by gradient adaption which made grids refined automatically in
FLUENT. Therefore, the y* near the impeller wall would be less than 100 while the y* near the
tank wall was set less than 300. A steady-state calculation was performed. The iteration was
running until the variables such as velocity no longer changed. The criteria of the scaled
residuals used to stipulate an equilibrium flow were obtained when it decreased to 10-5 for
continuity, x-velocity, y-velocity, z-velocity, k and epsilon in FLUENT. The data from creating
surface variables were also checked to be stable.
2.3.3. Model validation

Figure 2.1 shows that the models predictions matched well with experimental results. The
model for mixing of Newtonian fluid was validated by comparing the computed velocities with
experimental data from Hoffmann et al (Hoffmann et al., 2008).

The small deviations between

predicted and experimental data may be attributed to differences in the blade width and chord
angle between the model and the actual structure of the A-310 impeller.

82

Figure 2.1. Comparisons of prediction and experimentation in radial average velocity (A) z=5cm
(B) z=13cm (Hoffmann et al., 2008)

83

2.4. Results and discussion

2.4.1. Flow characteristics

Figure 2.2 shows quantitative comparisons of flow patterns of the digester using A-310
impeller with 5%TS and 10%TS. Significantly different flow fields were formed by agitation of
the same impeller in these two slurries. The mixing intensity was set at 500 rpm so that medium
agitation could be achieved to give the digester a good ability to handle transient hydraulic and
organic shock load (Hoffmann et al., 2008). The cross section of the digester with 5%TS shows
that a large area of high velocity was formed close to the impeller.

Although fluid velocity

decayed with increased distance from the impeller, most flow fields of the digester could be over
0.03 m/s. Compared to the digester with 5%TS, the digester with 10%TS was confronted with
more challenges to achieve efficient mixing. The rheological parameters used in the digester with
10%TS is shown in Table 2.2. The cross section of the digester with 10%TS shows the
significant velocity zone only occurred at or near the impeller. Most flow fields of the digester
with 10%TS were less than 0.03 m/s. This indicates that high TS impeded the convection process
caused by the A-310 impeller.

84

Figure 2.2. Comparison of velocity contour in TS<5% and TS=10%

Table 2.2. Parameters of the power-law model for shear stress with temperature (T), flow
behavior index n and consistency coefficient K (El-Mashad et al., 2005)
T (C)

K (Pasn)

35

0.348

16.1

40

0.325

16.7

50

0.332

13.0

85

Figure 2.3. Power number (Np),

flow number (Nq), maximum wall shear stress and shear rate

and average shear rate vs. impeller Re for TS<5% (A) and TS=10% (B)

86

Power number, flow number and shear rate/stress are important parameters for the
assessment of mixing performance of impellers (Chudacek, 2002). These parameters were
computed for impeller A-310 in the digesters with 5%TS and with 10%, respectively. The
simulation results (Fig. 2.3) shown that the power number decreased with increasing Reynolds
number (Re). A curve drop occurred in the digester with 5% TS while a line drop occurred in the
digester with 10% TS. Once Re in digester with 5% TS was greater than 1000, the flow gradually
changed into turbulent flow and the power number stabilized. This prediction is in good
agreement with many researchers experiment results (Paul et al., 2004). The prediction of power
number in this study is 0.34 in the turbulent flow of a digester with 5%TS, and 0.47 in the
turbulent flow of a digester with 10% TS. This indicates that more energy consumption took
place in the higher solids digesters. The impeller discharge flow number, Nq, indicates the
pumping capacity of the impeller in a given geometry. Typically, the flow number increases as
Re increases up to 10,000, and becomes constant at higher Re (Paul et al., 2004). The prediction
of flow number was 0.76, while the experimental data were in the range of 0.55 ~ 0.73 under
turbulent conditions for impeller A-310. However, there was a different trend observed in the
higher solid digester (10% TS). When Re in digester with 10% TS was less than 200, the flow
number increased first then decreased. Once Re in digester with 10%TS was over 200, the flow
number increased quickly to approach a constant.

This is because the impeller pumping

direction was affected by the high viscosity of 10% TS fluid. The predicted fluid dynamics in the
digester with 10%TS showed that the upward flow dominated the impeller pumping direction at

87

low Reynolds numbers. With increases in Re, the downward flow gradually dominated the
impeller pumping direction. This flow behavior was different from that with 5% TS in which the
downward flow always dominated the impeller pumping direction.
Mixing characteristic also produce direct impact upon the microbial structure in the digester.
Microbial cells are susceptible to mechanical force (Morales-Barrera and Cristiani-Urbina, 2006),
which disrupts the cell membrane and eventually kills some microorganisms, animal cells and
plant cells. Continuous shear, in particular, negatively affects microbial flocs (Hoffmann et al.,
2008). CFD data on shear stress, shear rate and Re were obtained for impeller A-310 in the
digesters with 5% and with 10% TS, respectively (Fig. 2.3) while it is difficult to measure the
maximum of shear stress and shear rate on the impeller tip experimentally. The calculated results
shown that a small increase in shear stress and shear rate with increasing Re was observed at Re
of less than 10,000 in 5% TS slurry and 100 in 10% TS slurry, while with a sharp increase
following at this point. In addition, the maximal impeller wall shear stress and shear rate of the
10% TS digester were much higher than that of the 5%TS digester due to the high viscosity. If
the enzymes and bacteria stayed near the impeller tip, a large impact would be exerted by the
shear rate. However, the average shear rate was reduced by an order of magnitude compared with
the maximal shear rate. There was a great velocity gradient in the digester with 10% TS. Most of
enzymes and bacteria might find a low shear zone away from the impeller to live which needs
further validation.

88

2.4.2. Effect of TS on flow characteristics

The TS concentration of waste is an important parameter in anaerobic digestion. The fluid

dynamics in the digester can be affected by the TS related to the slurry rheological behavior
which subsequently influences the pH, temperature uniformity and effectiveness of the
microorganism in the decomposition process. The impact of TS on power number and flow
number were obtained for impeller A-310 in the digesters with 5% and with 10% TS,
respectively (Fig. 2.4). The response of power number to TS was nearly linear, at least in this
range, and indicates that higher power input was required to agitate digesters with higher solids.
A nearly exponential decay was observed in the relationship between flow number and TS,
indicating that the pumping capacity of impeller A-310 was impaired by higher total solids.

Figure 2.4. Power number and flow number change with total solids

89

2.4.3. Helical ribbon agitation in HSAD

The above results show that the application of the A-310 impeller in HSAD will face great
challenges. It is suggested that high solids digestion should be greater than 10% TS (Rivard et al.,
1989). According to our experimental observation, the manure waste slurry with 20%TS began
to lose flowability. In this study, rheological properties and density for 20% TS were obtained
from Landry et al. (Landry et al., 2004), shown in Table 2.3. The cross sectional velocity contour
of a helical ribbon mixer operated at the agitator speed of 1rpm is shown in Figure 2.5(A). It can
be seen that the flow pattern of a digester with 20% TS was similar to that of the digester with 10%
TS. The velocity gradient at the bottom of tank existed because the helical ribbon was not
designed with enough length to agitate the bottom. The reduction in intensity was slight in the
digester with 20% TS, indicating that the helical ribbon had good potential capacity for mixing in
HSAD.

Table 2.3. Rheological properties and densities of liquid manure for T = 35 C

(Achkari-Begdouri and Goodrich, 1992; Landry et al., 2004; Wu and Chen, 2008)
TS (%)

K(Pasn)

(s-1)

min(Pas)

max(Pas)

(kg/m3)

2.5

0.042

0.710

226702

0.006

0.008

1000.36

5.4

0.192

0.562

50702

0.01

0.03

1000.78

7.5

0.525

0.533

11399

0.03

0.17

1001.00

9.1

1.052

0.467

11156

0.07

0.29

1001.31

12.1

5.885

0.367

3149

0.25

2.93

1001.73

20

56.8

0.35

0.2423.9

34.47

172.5

1090

90

Figure 2.5. Flow field of helical ribbon (1rpm) in HSAD (A) and higher agitator speeds in
20%TS (B)

91

In order to reduce the cost of the system, Rivard et al. (1990) (Rivard et al., 1990) attempted
to establish mixing requirements for effective digestion rates (specific for each reactor
configuration). Experimental comparison of agitator speeds of 1 and 25 rpm in HSAD showed
no significant difference in fermentation performance. However, these experiments did not
directly show mixing situation.

Figure 2.5(B) illustrates the flow fields with mixing speeds of

10, 20, and 30 rpm. Compared to 1 rpm, higher agitator speeds only enhanced mixing intensity
without changing flow structure. It indicates that substrate and microorganism were transported
to everywhere in the digester even at 1 rpm. Vavilin et al. (Vavilin and Angelidaki, 2005)
suggested that intensive mixing resulted in acidification and failure of the process if organic
loading was high. This showed that low mixing intensity was crucial for successful digestion.
The maximal shear stress/rate calculated for helical ribbon in the digester with 20% TS
showed a linear rise in shear stress/rate in response to an increase in agitator speed. Judging from
previous studies of the impact of shear stress/rate on digestion, the helical ribbon should have
little negative effect on bio-structure such as microbial flocs and biofilm at the agitator speed of
1 rpm. Further comparing shear stress/rate of helical ribbon with that of impeller A-310, a
different mixing mechanism for helical ribbon and impeller A-310 can be revealed. Impeller
A-310 had relatively small wall shear stress but a large shear rate. It indicates that the slurry was
positively displaced by the fluid near the impeller. It would be better to operate impeller A-310 in
low viscosity and turbulent state. The helical ribbon had relatively large wall shear stress but a
small shear rate. It indicates that the slurry was positively displaced by the ribbons. Therefore,

92

the helical ribbon can also be used in high viscosity and laminar state.

Table 2.4. Specifications and calculated results of ribbon mixers made in Aaron Process
Equipment
Specifications
Model

Working capacity (ft )

Drive (HP)

Calculated results
Agitator speed (rpm)

Unit drive

Unit drive

at the required speed

at 1 rpm

(kW/m )

(kW/m3)

NR4

1.5

40

9.875

0.068

NR14

14

40

9.405

0.065

NR24

24

7.5

35

8.230

0.068

NR36

36

15

30

10.973

0.111

NR55

55

20

26

9.576

0.118

NR66

66

25

25

9.975

0.129

NR80

80

30

25

9.875

0.128

NR100

100

40

25

10.534

0.137

NR150

150

50

25

8.778

0.114

NR175

175

60

25

9.029

0.117

NR200

200

50

20

6.584

0.115

NR250

250

60

18

6.320

0.128

NR300

300

100

13

8.778

0.275

NR350

350

100

15

7.524

0.194

NR400

400

150

15

9.875

0.255

NR500

500

150

15

7.900

0.204

Using the blending results of Aaron Process Equipment (http://aaronprocess.com), the unit
drive power (power input per unit volume) can be estimated. Table 2.4 shows the specifications
of Ribbon mixers made in Aaron Process Equipment and the calculated results for the unit drive
power. If ribbon blending is running at the standard speed of 15 ~ 40 rpm, the unit drive power
volume will be very high (about 9 kW/m3). However, according to the research of Rivard et al.

93

(Rivard et al., 1990), no significant difference in fermentation performance is observed between

agitator speeds of 1 and 25 rpm. Therefore, 1 rpm can be used to optimize the power input per
unit volume of ribbon blends. It has been shown through experimental data that the power
consumption (P) of an impeller is proportional to the cube of the agitator speed of the impeller. It
is defined as (Rushton et al., 1950a; Rushton et al., 1950b):
Np =

Kp

Re*

(2.9)

where Kp is a geometrical parameter depending on the mixing system. Re* is defined in Equation
(4) for non-Newtonian fluids.

P2 =

P1 N12 n N 23
N13 N 22 n

(2.10)

where P1 and N1 are the power consumption (W) and agitator speed of helical ribbon impeller
(rpm) at the required agitator speeds (15 ~ 40 rpm), respectively. P2 and N2 are the power
consumption and agitator speed of the helical ribbon impeller at the agitator speed of 1 rpm,
respectively. n is the power law index.
The calculated results for the unit drive power are listed in Table 4. The unit drive power
was computed as drive power consumption divided by the reactor volume. The calculations were
carried out at the required agitator speeds and 1 rpm, respectively. The drive power consumption
at 1 rpm was derived from Equation (10). The digester content used in calculations was manure
slurry of 20% TS. The power law index n is 0.35.
As shown in Table 4, the unit drive power of a helical ribbon at the speed of 1 rpm had a

94

drop of about 50 fold compared to the required speeds. However, when the volume of the
digester increased, the higher unit power inputs were required.

Compared with the

conventional power input requirement, the predicted input for a HSAD system had a higher
value. Further optimization might be obtained following Rivards (1995) suggestion, using
intermittent mixing. In their experimental study, mixing power could be optimized as 1.48
HP/1000 ft3 (38.98 W/m3) at the sludge volume of 500 L.
2.5. Conclusions

Mechanical mixing in HSAD system was numerically simulated and relevant flow
characteristics were obtained, including flow patterns, pumping, shear and power. The numerical
results show that impeller A-310 mixing performance was impaired with increasing total solids
concentration. If higher agitator speed was used to maintain a good mixing in 10% TS, much
higher shear rate would be generated that may affect the growth of bio-structure after Re became
greater than 100.

The helical ribbon was suitable for mixing high solid digesters and mixing

power could be optimized. Its low shear environment would not significantly the growth of
bio-structure.
Acknowledgements

The authors thank Washington State Department of Ecology and the California Energy
Commission for sponsoring this study.

95

2.6. References

Achkari-Begdouri, A., Goodrich, P., 1992. Rheological properties of Moroccan dairy cattle
manure. Bioresource Technology, 40, 149156.
Chudacek, M.W., 2002. Impeller power numbers and impeller flow numbers in profiled bottom
tanks. Industrial & Engineering Chemistry Process Design and Development, 24,
858867.
Ding, J., Wang, X., Zhou, X.F., Ren, N.Q., Guo, W.Q., 2010. CFD optimization of continuous
stirred-tank (CSTR) reactor for biohydrogen production. Bioresource Technology, 101,
7005-7013.
FLUENT, 2006. FLUENT 6.3 User's Guide. FLUENT, Inc. , Lebanon, NH.
Hoffmann, R.A., Garcia, M.L., Veskivar, M., Karim, K., Al-Dahhan, M.H., Angenent, L.T., 2008.
Effect of shear on performance and microbial ecology of continuously stirred anaerobic
digesters treating animal manure. Biotechnology and Bioengineering, 100, 3848.
Karim, K., Klasson, K.T., Hoffmann, R., Drescher, S.R., DePaoli, D.W., Al-Dahhan, M.H., 2005.
Anaerobic digestion of animal waste: Effect of mixing. Bioresource Technology, 96,
16071612.
Kilander, J., Rasmuson, A., 2005. Energy dissipation and macro instabilities in a stirred square
tank investigated using an LE PIV approach and LDA measurements. Chemical

Engineering Science, 60, 68446856.

Landry, H., Lague, C., Roberge, M., 2004. Physical and rheological properties of manure

96

products. Applied Engineering in Agriculture, 20, 277288.

Luo, J.Y., Issa, R.I., Gosman, A.D., 1994. Prediction of impeller-induced flows in mixing vessels
using multiple frames of reference Institution of Chemical Engineers Symposium series,
pp. 549556.
Metzner, A.B., Otto, R.E., 1957. Agitation of non-newtonian fluids. AIChE Journal, 3, 310.
Moeller, G., Torres, L.G., 1997. Rheological characterization of primary and secondary sludges
treated by both aerobic and anaerobic digestion. Bioresource Technology, 61, 207211.
Morales-Barrera, L., Cristiani-Urbina, E., 2006. Removal of hexavalent chromium by
Trichoderma viride in an airlift bioreactor. Enzyme and Microbial Technology, 40,
107-113.
Norton, T., Sun, D.W., Grant, J., Fallon, R., Dodd, V., 2007. Applications of computational fluid
dynamics (CFD) in the modelling and design of ventilation systems in the agricultural
industry: A review. Bioresource Technology, 98, 2386-2414.
Patwardhan, A.W., 2001. Prediction of flow characteristics and energy balance for a variety of
downflow impellers. Industrial & Engineering Chemistry Research, 40, 38063816.
Paul, E.L., Atiemo-Obeng, V.A., Kresta, S.M., 2004. Handbook of industrial mixing. John Wiley
& Sons, New Jersey.
Pougatch, K., Salcudean, M., Gartshore, I., Pagoria, P., 2007. Computational modelling of large
aerated lagoon hydraulics. Water Research, 41, 2109-2116.
Rivard, C.J., Himmel, M.E., Vinzant, T.B., Adney, W.S., Wyman, C.E., Grohmann, K., 1989.

97

Development of a novel laboratory scale high solids reactor for anaerobic-digestion of

processed municipal solid-wastes for the production of methane. Applied Biochemistry

and Biotechnology, 20, 461478.

Rivard, C.J., Himmel, M.E., Vinzant, T.B., Adney, W.S., Wyman, C.E., Grohmann, K., 1990.
Anaerobic-digestion of processed municipal solid-waste using a novel high solids reactor
- maximum solids levels and mixing requirements. Biotechnology Letters, 12, 235240.
Rushton, J.H., Costich, E.W., Everett, H.J., 1950a. Power characteristics of mixing impellers Part
I. Chemical Engineering Progress 46, 395404.
Rushton, J.H., Costich, E.W., Everett, H.J., 1950b. Power characteristics of mixing impellers Part
II. Chemical Engineering Progress 46, 467479.
Shilton, A., Kreegher, S., Grigg, N., 2008. Comparison of computation fluid dynamics
simulation against tracer data from a scale model and full-sized waste stabilization pond.

Journal of Environmental Engineering-Asce, 134, 845-850.

Terashima, M., Goel, R., Komatsu, K., Yasui, H., Takahashi, H., Li, Y.Y., Noike, T., 2009. CFD
simulation of mixing in anaerobic digesters. Bioresource Technology, 100, 22282233.
Vavilin, V.A., Angelidaki, I., 2005. Anaerobic degradation of solid material: Importance of
initiation centers for methanogenesis, mixing intensity, and 2D distributed model.

Biotechnology and Bioengineering, 89, 113122.

Wu, B.X., Chen, S.L., 2008. CFD simulation of non-Newtonian fluid flow in anaerobic digesters.

Biotechnology and Bioengineering, 99, 700711.

98

Wu, J., Pullum, L., 2000. Performance analysis of axial-flow mixing impellers. AIChE Journal,
46, 489498.
Yin, C.E., Kaer, S.K., Rosendahl, L., Hvid, S.L., 2010. Co-firing straw with coal in a
swirl-stabilized dual-feed burner: Modelling and experimental validation. Bioresource

Technology, 101, 4169-4178.

Zhu, B.N., Zhang, R.H., Gikas, P., Rapport, J., Jenkins, B., Li, X.J., 2010. Biogas production
from municipal solid wastes using an integrated rotary drum and anaerobic-phased solids
digester system. Bioresource Technology, 101, 6374-6380.

99

CHAPTER 3 : EXPERIMENTAL AND MODELING STUDY OF A TWO-STAGE PILOT

SCALE HIGH SOLID ANAEROBIC DIGESTER SYSTEM
3.1. Abstract

This study established a comprehensive model to configure a new two-stage high solid
anaerobic digester (HSAD) system designed for highly degradable organic fraction of municipal
solid wastes (OFMSW). The HSAD reactor as the first stage was naturally separated into two
zones due to biogas floatation and low specific gravity of solid waste. The solid waste was
retained in the upper zone while only the liquid leachate resided in the lower zone of the HSAD
reactor. Continuous stirred-tank reactor (CSTR) and advective-diffusive reactor (ADR) models
were constructed in series to describe the whole system. Anaerobic digestion model No.1
(ADM1) was used as reaction kinetics and incorporated into each reactor module. Compared
with the experimental data, the simulation results indicated that the model was able to well
predict the pH, volatile fatty acid (VFA) and biogas production.
Keywords: organic fraction of municipal solid wastes (OFMSW); food waste; anaerobic

digestion model No.1 (ADM1); high solid anaerobic digester (HSAD)

100

3.2. Introduction

Anaerobic digestion (AD) for converting the organic fraction of municipal solid wastes
(OFMSW) to renewable energy and reducing environmental impact of late has focused on the
use of high solid anaerobic digester (HSAD) designs. This results primarily from their ability to
substantially reduce the size of the reactor, thus reducing the footprint and improving the
economics of the system (Rivard et al., 1989; Yu et al., 2011).

Given the large biodegradable

organic content of OFMSW (i.e. food waste, green waste), a major limitation of AD of these
wastes in a one stage system is rapid uncontrolled acidification and decrease in pH within the
reactor, which inhibits the activity of methanogenic bacteria (Arvanitoyannis, 2008; Schievano et
al., 2010). Furthermore, it is difficult to optimize the AD process in the one stage system because
methanogenic and non-methanogenic bacteria are significantly different with respect to
physiology, nutritional requirements, growth and metabolic characteristics, environmental
optima, and sensitivity to environmental stress (Jagadabhi et al., 2011).

Therefore, two-stage

systems appear to be more efficient technologies for AD of food waste (Demirel and Yenigun,
2002).
Various two-stage technologies have been used to overcome the above problems in
converting OFMSW into bio-methane. Leach bed reactors (LBR) have often been used as first
stage digesters in previous research (Lehtomaki et al., 2008; Nizami et al., 2011). However, LBR
are prone to mass transfer limitations because the solids tend to be agglomerated, resisting
enzyme and microorganism diffusion. Some researchers have employed continuous stirred tank

101

reactors (CSTR) as the first stage digesters to reduce this concern and enhance hydrolysis of the
solids (Cavinato et al., 2011; Dinsdale et al., 2000). However, total mixing may impair the
separation of liquid from solid, a step required for effective feed of liquid and soluble organics to
the second stage digester. Although a rotating drum mesh filter has been designed to mix and
separate solids and liquid simultaneously in a first stage digester (Walker et al., 2009), high
energy consumption and complicated manufacture remain a concern to project costs. In order to
design and optimize an efficient two-stage system, the development and use of models may be
favorable to comprehensively evaluate the impacts of all process variables on performance.
Many models had been developed for AD processes. Anaerobic digestion model No.1
(ADM1) was a collaborative work of many international experts from various fields of anaerobic
process technology to provide the possibility of simulating a broad category of processes
(Batstone et al., 2006). This model has been used as common basis for further development. The
extended applications of ADM1 model include anaerobic two-stage digestion of sewage sludge,
traditional Chinese medicine (TCM) wastewater, olive pulp, and grass silage (Blumensaat and
Keller, 2005; Chen et al., 2009; Koutrouli et al., 2009; Thamsiriroj and Murphy, 2011). ADM1
modeling has showed that two-stage anaerobic digestion is a stable, reliable and effective process
for energy recovery with ADM1 modeling proving to be a valuable tool for process design even
in the case of complex feedstock. Importantly, it is often necessary for purposes of time and cost
to develop and utilize model prediction for design, optimization and scale up of new anaerobic
digestion systems.

102

Figure 3.1. Model structure of the two-stage HSAD system

In this study, a new pilot-scale two-stage HSAD system was constructed to enhance system
stability and bio-methane productivity of a primarily food scrap solid waste. This system shown
in Figure 3.1 included a HSAD reactor and an upflow anaerobic sludge blanket (UASB) seed
reactor. According to experimental observation, the HSAD reactor was naturally separated into
two zones due to biogas floatation and low specific gravity of solid waste with respect to water.

103

In the HSAD reactor, the solid waste was agitated and retained in the upper zone while soluble
organics and liquid leachate comprised the lower zone without need for agitation. Liquid effluent
cycling was carried out in the two-stage HSAD system for seeding of bacteria, volatile fatty acid
(VFA) removal, and pH control. A comprehensive mathematical model with ADM1 kinetics was
established to understand this new system. The upper zone of the HSAD reactor was assumed to
be a continuous stirred-tank reactor (CSTR) while the lower zone of the HSAD reactor was
assumed to be advective-diffusive reactor (ADR). The UASB seed reactor was assumed to be an
ADR as opposed to a CSTR as the ADR model allowed for more effective analysis of reactor
cross section area and height for reactor design. The pH, volatile fatty acid (VFA) and biogas
production were experimentally verified in the two batch digesters including HSAD reactor and
UASB reactor, and then were predicted in the new pilot-scale two-stage HSAD system.
3.3. Methods

3.3.1. System description

Figure 3.1 shows the model structure used in the mesophilic (35oC) two-stage HSAD system.
The HSAD reactor had a working volume of 0.057 m3 with tank diameter of 0.3 m and height of
0.8 m. The volume ratio of the upper and lower zones for the HSAD reactor was 3.0. The UASB
reactor had a working volume of 0.14 m3 with tank diameter of 0.3 m and height of 2.0 m. Solids
and liquid retention as well as liquid recycle within both the HSAD reactor and UASB reactors
were modeled against varying solids retention, hydraulic retention, and recycle rates.
According to experimental observation, the HSAD reactor was naturally separated into two

104

zones due to biogas floatation and low specific gravity of solid waste with respect to water. The
solid substrates consist of complex composite particulate and particulate carbohydrates, proteins
and lipids (Batstone et al., 2002). About 99% of the solid substrates were suspended in the upper
zone of the HSAD reactor. The liquid leachate from the UASB effluent and the hydrolytic
solutes flowed into the lower zone of the HSAD reactor until ultimate feeding back to the UASB
reactor. Mechanical mixing, which consisted of a shaft with vertical impeller blades was carried
out in the upper zone of the HSAD reactor to enhance mass transfer of the solid waste. Soluble
substrates within the lower zone of the HSAD reactor flowed without need of active mixing,
allowing for an environment conducive to methanogenic bacteria growth and biogas flotation
(Vavilin and Angelidaki, 2005). The two-stage HSAD system was operated in both batch and
continuous mode. In batch mode, solid waste was fed into the HSAD reactor once and all
effluent was recycled in the system. The batch process was operated at a solid retention time
(SRT) of 9 days and a hydraulic retention time (HRT) of 1 day. In the continuous mode, solid
waste was continuously fed into the HSAD reactor and effluent was in part discharged out of the
UASB reactor.

The continuous process was operated at a SRT of 30 days and a HRT of 1 day.

After the SRT was achieved, the solid residuals were discharged out of the bottom of the HSAD
reactor for composting.
In the UASB seed reactor, bacteria biomass concentration present as retained granules was
assumed to be constant (Seghezzo et al., 1998) at an experimental determined average of 15 kg
COD/m3. Further data determined that approximately 4.87% of acetic acid and hydrogen

105

degrading organisms as well as all other degrading organism were recycled to the HSAD reactor
for seeding of bacteria, VFA removal and pH control. The UASB reactor was operated at a HRT
of 1 day.

Table 3.1. Conditions for the mathematic model for the high solids anaerobic digester system
Characteristic

Unit

Total solids (TS)

g/l

154

Volatile solids (VS)

g/l

118

Particulate COD (CODp)

gCOD/l

182

Soluble COD (CODs)

gCOD/l

18.8

Total organic carbon (TOC)

gC/l

53.3

Inorganic carbon (IC)

mol/l

10.7

Total Kjeldahl nitrogen (TKN)

gN/l

5.3

Total ammonium (TAN)

gN/l

1.3

Total phosphorus (TP)

gP/l

0.46

Total ortho-phosphorus (OP)

gP/l

0.9

Volatile fatty acids (VFA)

gCOD/l

2.3

pH

Solid food waste

5.47

g/l
g/l
g/l

Carbohydrate
Protein
Lipids

75.78 5.52
42.39 4.11
10.28 1.41

The food waste, obtained from the Washington State University (WSU) cafeteria, including
plastic and paper packaging (about 2.7%) was mechanically cut into shreds, producing a paste
with total solids (TS) of 30.6%, with other waste parameters summarized in Table 3.1. Upon
loading into the HSAD reactor, the initial TS was diluted to 15.4% while maintaining a pH of
5.47. Prior to system evaluation the HSAD and UASB systems were operated for 4 months to

106

allow for system acclimation. Bacteria inoculums and granular seeds were obtained from a
commercial starch water UASB reactor (Penford Foods, Richland, WA, USA).

The

experimental data were sampled from the two-stage HSAD system after four months of granule
inoculums because an inoculum adapted to the feed under study can be developed (Chynoweth
and Isaacson, 1987).

Since this system consisted of high rate digesters with short HRT of less

than 1 day (Seghezzo et al., 1998), one weeks experimental data were collected to show
dynamic change in pH, concentration and biogas production.
3.3.2. Model structure
The reaction kinetics in ADM1 model is complex with a number of sequential and parallel
steps (Batstone et al., 2002). There are two main types of reactions in the model. One is
biochemical reactions and another is physic-chemical reactions. The ADM1 model includes 19
biochemical kinetic processes. All biochemical extracellular steps are assumed to be first order
and all intracellular reactions are assumed to be Monod-type kinetics. The biochemical
extracellular kinetics including disintegration rate and hydrolysis rate of bio-particles, and decay
rate of microorganisms are expressed as:
rj = K dis , hyd , dec , j C X , j

(3.1)

where rj represents the disintegration rate (kgCODm-3d-1), hydrolysis rate and decay rate of
bio-particles and microorganisms, respectively. K is the coefficient of the first order reactions
(d-1). C is the concentration (kgCODm-3) including microorganisms of complex particles,
carbohydrates, proteins, lipids, sugars, amino acids, long chain fatty acids (LCFA), C4,

107

propionate, acetate and hydrogen.

The intracellular kinetics including uptake rate of substrates are expressed as:
rj = K m , j

CS , j
K S , j + CS , j

(3.2)

CX , j Ii

where Km is the specific Monod maximum uptake rate (kgCODm-3 _SkgCODm-3 _X d-1); KS
is the Monod half saturation constant (kgCODm-3); and Ii is inhibition function. Inhibitions
came from H2, NH3, pH and deficit of inorganic nitrogen.
Valerate and butyrate are assumed to be uptaken by the same microorganism responsible for
C4. Their kinetics differ from the other intracellular reactions and are expressed as:

rva = K m ,c 4

CS ,va
K S ,c 4 + CS ,va

rbu = K m ,c 4

C X ,c 4

CS ,bu
K S ,c 4 + CS ,bu

1
1 + CS ,bu CS ,va

C X ,c 4

I2

1
1 + CS ,va CS ,bu

I2

(3.3)

(3.4)

In this study, the ADM1 model was extended to describe a HSAD reactor and an UASB
reactor. The HSAD reactor includes two zones. The upper zone of the HSAD reactor is assumed
to be a CSTR. The solid food wastes were mechanically agitated to enhance mass transfer among
microorganisms, enzyme, and solid substrate. The temporal change of the concentration is given
as (Batstone et al., 2002)
dCliq ,i
dt

qin Cin ,i qout Cliq ,i

Vliq

rj vi , j

(3.5)

j =119

where C is concentration (kgCODm-3) including microorganisms and substrates; j=1-19rjvi,j is

the sum of the biochemical rate coefficients (vi,j) multiplied by the kinetic rate equations (rj) for

108

the process; V is the reactor volume (m3); and qin and qout are the flow rate of influent and
effluent, respectively.
The lower zone of the HSAD reactor was assumed to be an advective-diffusive reactor, with
mechanical agitation in the upper zone not affecting the lower zone of the HSAD reactor. Only
liquid leachate and soluble substrates reside in the low zone, with the solid waste having pushed
back to the upper zone by biogas flotation due to low specific gravity (SGfood waste=1.08). The
behavior of substrates transported in the compartment is given as (Reichert, 1998)
A

Cliq ,i
t

+q

Cliq ,i
x

= AD

2Cliq ,i
x 2

Arj vi , j

(3.6)

j =119

where A is the cross section area (m2) of the reactor, and D is the coefficient of dispersion which
was set as 20 m2/day.
Leachate from the lower zone is ultimately sent to the UASB reactor, a high rate
methanogenic digester. Most of methanogenic bacteria are immobilized in the granules (Molina
et al., 2008), allowing for a reach with short HRT and long SRT with highly active granules with
good settling abilities retained in the UASB reactor despite the upflow velocity of the wastewater
(Schmidt and Ahring, 1996). To simplify the flow model, the UASB seed reactor is assumed to
be an advective-diffusive reactor. Methanogenic bacteria including acetic acid and hydrogen
degrading organisms were assumed to be constant with an average of 15 kg COD/m3 based on
experimental measurement due to slow growth rate (Liu et al., 2002). The behavior of substrates
transported in the compartment is given as (Reichert, 1998)

109

Cliq ,i
t

+q

Cliq ,i
x

= AD

2Cliq ,i
x 2

Arj vi , j

(3.7)

j =119

Biogas hydrogen, methane and carbon dioxide were calculated from gas liquid transfer
kinetic rate.
3.3.3. Analytic method
The total chemical oxygen demand (COD), soluble COD, total phosphorus (TP), total
ortho-phosphorus (OP), total solids (TS) and volatile solids (VS) were determined according to
standard methods (Eaton et al., 2005). pH was measured daily using a Fisher Scientific AB15 pH
meter. Total organic carbon (TOC) and Inorganic carbon (IC) were analyzed using a
SHIMADZU TOC-5000. VFA was analyzed using a static headspace gas chromatograph (GC)
(SHIMADZU GC-2014). Gas composition was analyzed using a Varian CP-3800 GC. Total
Kjeldahl nitrogen (TKN) was measured using a 2300 Kjeltec Analyzer Unit (Tecator, Perstorp
Analytical). Biogas production was measured using Wet Tip Gas Meters (wettipgasmeter.com).
3.3.4. Model parameters estimation
The model was implemented and solved using AQUASIM 2.0 software tool which is a
program for the identification and simulation of aquatic systems (Reichert, 1998). The initial
parameters used in the ADM1 kinetics were selected from the scientific and technical work
reported by the IWA task group (Batstone et al., 2002; Gali et al., 2009; Rosen and Jeppsson,
2006). Kinetic parameters were estimated by minimizing the sum-of-the-squared-difference 2
between the measured (yexp) and calculated (ycal) values at the sampling ports over sampling

110

times (Ntime) as follows:

y y
= cal,i exp,i

i =1
exp,i

Ntime

(3.8)

where exp,I is the standard deviation of the measured values.

On the basis of the suggested biochemical parameter values, sensitivity and estimation in the
ADM1 report (Batstone et al., 2002), the estimation procedures were applied for the following
parameters: kinetic coefficients of Kdis, Km_ac, Ks_ac and stoichiometric coefficients of fch_xc, fpr_xc,
fsi_xc, fxi_xc. Initial and estimated kinetic coefficients for the model are shown in Table 3.2.

111

Table 3.2. Model parameters in the HSAD system (Batstone et al., 2002; Gali et al., 2009; Rosen
and Jeppsson, 2006)
Constant

Units

Disintegration and hydrolytic parameters

Kdis
d-1
Khyd_CH
d-1
d-1
Khyd_PR
Khyd_LI
d-1
Kinetics parameters of acidogenesis
Km_su
CODCOD-1d-1
Km_aa
CODCOD-1d-1
Ks_su
kgm-3
Ks_aa
kgm-3
Ysu
Yaa
Kdec_su
d-1
Kdec_aa
d-1
Kinetics parameters of acetogenecesis
Km_fa
CODCOD-1d-1
Km_c4
CODCOD-1d-1
Km_pro
CODCOD-1d-1
Ks_fa
kgm-3
Ks_c4
kgm-3
Ks_pro
kgm-3
Yfa
Yc4
Ypro
Kdec_fa
d-1
d-1
Kdec_c4
Kdec_pr
d-1
Kinetics parameters of methanogenesis
Km_ac
CODCOD-1d-1
Km_h2
CODCOD-1d-1
Ks_ac
kgm-3
Ks_h2
kgm-3
Yac
Yh2

112

Initial value

Estimated value

0.5
10
10
10

0.15

30
50
0.5
0.3
0.1
0.08
0.02
0.02
6
20
13
0.4
0.2
0.1
0.06
0.06
0.04
0.02
0.02
0.02
8
35
0.15
710-6
0.05
0.06

8.16
0.026

Kdec_ac
Kdec_h2
Stoichiometric parameters
fch_xc
fli_xc
fpr_xc
fsi_xc
fxi_xc

d-1
d-1

0.02
0.02
0.20
0.25
0.20
0.1
0.25

0.5
0.068
0.28
0.075
0.075

3.3.5 Sensitivity analysis

A sensitivity analysis was performed on the suggested sensitive parameters in the ADM1
report (Batstone et al., 2002) to understand their impact on the predicted acetic acid and CH4
production. The absolute-relative sensitivity function () was used to measure the absolute
change in acetic acid and CH4 production for a 100% change in kinetic coefficients of Kdis,
Khyd_CH, Khyd_PR, Km_ac, Ks_ac and stoichiometric coefficients of fch_xc, fpr_xc, fsi_xc, fxi_xc (Reichert,
1998). The behavior of the model was evaluated as a function of variations in input parameters
using a one-variable-at-a-time approach, in which only one parameter value was changed each
time, while the other parameters were kept constant.
3.4. Results and discussion

3.4.1. Model validation and Sensitivity analysis

Figure 3.2 shows comparison of experimental data and the simulation values in the HSAD
batch reactor. As shown in Figure 3.2(A), the HSAD reactor has a sharp pH drop at the first day,
and then the trend slows down with time. The effluent of the UASB seed reactor and the
feedstock were filled into the HSAD reactor in those different ratios. The volume ratios of 5.43,

113

2.19 and 1.1 were corresponding to 5%, 10% and 15%TS in the HSAD reactor, respectively. At
the beginning, the pH of the food waste fed in the HSAD reactor was 5.47 while the pH of
effluent was 7.9. The subsequent results show that HRT less than 1 day could maintain pH over
6.0 even at the low recirculation ratio of 1.1. This means that the effluent of the UASB seed
reactor provided a suitable environment for methanogenic bacteria at 15%TS at HRT as long as 1
day.

Figure 3.2(B) shows daily biogas production at pH of 7.5 and TS of 15% in the HSAD

reactor without mechanical mixing and recirculation. An obvious stratification of solid and liquid
was observed in the upper and lower zone of the HSAD reactor. The biogas production showed
stable operation on a daily and monthly basis. No trend towards acidification occurred because
only the solid waste interfaced while the liquid phase was slowly hydrolyzed into as if moved to
the lower zone for eventual biomethane production. The model shows good agreement (P > 0.05)
with the measurement of pH and biogas production in the pilot-scale HSAD reactor.

114

Figure 3.2. (A) pH at the different ratio of effluent to waste, and (B) biogas production change
with time in the HSAD batch reactor without recirculation

115

Figure 3.3 shows a comparison of daily biogas production, VFA and pH between predicted
and experimental data from the UASB seed reactor. The biogas production of the UASB seed
reactor shown in Figure 3.3(A) was measured when the feed rate of leachate from the high solids
digester was at 0.038 m3/week. In the first day, the highest biogas production was obtained due
to the feedstock with the highest VFA shown in Figure 3.3(B). Then an exponential decay of the
biogas production was observed in the following days. The pH was maintained at the range of
7.5 8.0. The model shows good agreement (P > 0.05) with the measurement of biogas
production, pH and VFA in the pilot-scale UASB seed reactor.

116

Figure 3.3. Biogas production (A), VFA (B) and pH (C) change with time in the UASB seed
reactor

117

Figure 3.4 shows a comparison of VFA compositions between predicted and experimental
data from the UASB seed reactor. The major composition of VFA includes acetic, propionic,
butyric and valeric acids. Compared with the other VFA compositions, the consumption rate of
propionic acids was relatively slow. The concentration of acetic and propionic acids are critical
factors that regulate anaerobic digestion processes, since oxidation of propionic acid to acetic
acid is the slowest among the VFA (Dohnyos et al., 1985; Inanc et al., 1999; Sonakya et al.,
2007). The model shows good agreement (P > 0.05) with measurement of acetic, propionic, and
butyric acid in the pilot-scale UASB seed reactor. However, the calculated results of valeric acids
do not fit well with the measurements. This is because butyric and valeric acids share the same
equations for degrading organism in the ADM1 kinetics shown in Eq. 3.3 and 3.4. The predicted
results of butyric and valeric acids only have proportional difference.

118

119

Figure 3.4. Acetic acid (A), propionic acid (B), butyric acid (C) and valeric acid (D) change with
time in the UASB seed reactor
The results of sensitivity analysis are shown in Table 3.3. Although there were different
sensitivity rankings in the HSAD and UASB reactors for acetic acid and CH4 production
respectively, Kdis was the higher sensitive parameter as a whole. This suggests that hydrolysis
was the rate-limiting step in this HSAD system (Arudchelvam et al., 2010; Zhao et al., 2009).
This conclusion is different from that in traditional food waste anaerobic digestion where
methanogenesis is the rate-limiting step (Velmurugan and Ramanujam, 2011), indicating the
design in this study could enhance the whole system production rate because sufficient
methanogen were retained to handle hydrolytic intermediate products in the HSAD system.

120

Table 3.3. Ranking of roots of mean squared sensitivities

1
2

HSAD upper zone

Acetic acid
2

HSAD lower zone

CH4

Acetic acid
2

UASB

CH4

Acetic acid
2

CH4

Rank

Rank

Rank

Rank

Rank

Rank

fch_xc

0.82

fch_xc

0.45

Kdis

0.66

Kdis

0.65

Km_ac

0.17

Kdis

1.70

fpr_xc

0.62

Kdis

0.21

fch_xc

0.65

Km_ac

0.63

Ks_ac

0.086

fch_xc

1.24

Kdis

0.53

fpr_xc

0.11

Km_ac

0.63

fch_xc

0.48

Kdis

0.072

fpr_xc

0.69

Km_ac

0.22

Km_ac

0.034

fpr_xc

0.35

fpr_xc

0.29

fch_xc

0.067

fsi_xc

0.57

Ks_ac

0.095

fsi_xc

0.019

fsi_xc

0.29

fsi_xc

0.16

fpr_xc

0.054

fxi_xc

0.38

fsi_xc

0.066

fxi_xc

0.013

fxi_xc

0.19

fxi_xc

0.11

fsi_xc

0.023

Km_ac

0.065

fxi_xc

0.044

Kh_CH

0.006

Ks_ac

0.10

Ks_ac

0.028

fxi_xc

0.015

Ks_ac

0.031

Kh_CH

0.012

Ks_ac

0.003

Kh_CH

0.011

Kh_CH

0.011

Kh_CH

0.001

Kh_CH

0.024

Kh_PR

0.008

Kh_PR

0.002

Kh_PR

0.007

Kh_PR

0.004

Kh_PR

Kh_PR

0.013

Note: 1 Compartments; 2 Outputs; 3 Comparison of sensitivity; 2 Roots of mean squared sensitivity

3.4.2. Effect of recycled methanogenic seeds on methane concentration in the HSAD reactor
The established model for the two-stage HSAD system was used to predict the effect of the
recycled methanogenic seeds from the UASB seed reactor to the HSAD reactor on methane
concentration in the HSAD reactor. It is difficult to separate the methanogenic bacteria from the
other microorganisms in experiments, especially large-scale experiments. This process can be
understood in terms of removing/un-removing the function of methanogenic bacteria in the
recycled effluent from the model. The recycling rate of effluent was set as 0.14 m3/day and
loading rate was set as 15 kg COD/m3/d. Figure 3.5(A) and (B) show the change of gas
composition with time in the HSAD reactor without methanogenic bacteria. High content (about
65%) of hydrogen was produced in the upper zone of the HSAD reactor while methane content
was close to zero due to high loading rate and short HRT. In the lower zone of the HSAD reactor,

121

methanogenic bacteria grew with time in a low TS environment. Therefore, the methane content
increases with time. Figure 3.5(C) and (D) show the biogas contents produced by a system with
methanogenic seeds (4.87% acetic acid degrading organisms and hydrogen degrading organisms)
in the HSAD reactor. No hydrogen was produced even in the upper zone of the HSAD reactor.
The biogas had high methane content (about 60%) in the HSAD system, which is in accordance
with our measurements (63.9% average).

Figure 3.5. Effect of recycled methanogenic seeds on methane concentration in the HSAD
reactor (A, B) the upper zone and the lower zone without recycled methanogenic bacteria; (C, D)
the upper zone and the lower zone with recycled methanogenic bacteria

122

3.4.3. Effect of recycling rate on pH in the HSAD system

The effect of recycling rate of effluent from the UASB seed reactor to the HSAD reactor on
pH in the HSAD system is shown in Figure 3.6. The results of batch and continuous mode
operation were compared. The difference of these two modes depends on feeding solid waste
once or continuously. The loading rate in continuous mode was set as 15 kg COD/m3/d. In the
batch mode operation, the increased recycling rate from 0.05 m3/d to 0.09 m3/d pushed the
HSAD reactor pH back to an optimum range of 7.0 8.0 (Chynoweth and Isaacson, 1987)
shown in Figure 3.6(A) and (B). However, in the continuous mode operation, the increased
recycling rate from 0.14 m3/d to 0.5 m3/d carried the pH of the HSAD reactor to deviate from the
optimum range as shown in Figure 3.6(C) and (D). This is because a high recycling rate leads to
a short retention time for fatty acids to be digested. This is in accordance with the prediction of
grass silage in a 2-stage CSTR anaerobic digester (Thamsiriroj and Murphy, 2011). In the
calculation, it was observed that the acetic acid consumption at the recycling rate of 0.14 m3/d
and the loading rate of 15 kg COD/m3/d was not balanced, although pH was maintained at the
range of 7.0 8.0. This is because ammonia ion released from solid waste offsets the
acidification trend caused by acetic acid. However, it also caused a high ammonia concentration,
which inhibited microbial activity.

123

Figure 3.6. Effect of recycling rate on pH: (A, B) batch mode at the recycling rate of 0.05 and
0.09 m3/day; (C, D) continuous mode at the recycling rate of 0.14 and 0.5 m3/day

3.4.4. Comparison of acetic acid in the HSAD system

Figure 3.7(A), (B), (C), (D) and (E) predict the change of acetic acid with time under the
conditions of batch mode, continuous mode, the increased UASB height, the increased UASB
cross section area, and the increased methanogenic seeds, respectively. The recycling rate of
effluent was set as 0.14 m3/day. In the batch mode shown in Figure 3.7(A), the concentration of
acetic acid first increased then decreased slowly in the HSAD reactor while acetic acid was

124

completely consumed in the UASB reactor. Most of complex particulates were decomposed in 8
days and the trend changed to smooth, although acetic acid still needed longer time to be
consumed in the HSAD reactor.
In the continuous mode at loading rate of 5 kg COD/m3/day shown in Figure 3.7(B), the
concentration of acetic acid first increased then maintained stable operation. If the UASB height
was increased 2-fold as shown in Figure 3.7(C), the acetic acid concentration had little change
compared with Figure 3.7(B). This indicates the UASB height had no impact on the change of
the acetic acid concentration. If the UASB cross section area was increased by 2 fold shown in
Figure 3.7(D), the acetic acid concentration decreased in the HSAD reactor compared with
Figure 3.7(B), especially in the upper zone of the HSAD reactor. This is because more acetic acid
in the HSAD reactor is transferred to the UASB reactor allowing for more complete consumption.
If the recycled methanogenic bacteria were increased 2-fold as shown in Figure 3.7(E), the acetic
acid concentration was reduced in the lower zone of the HSAD reactor compared with Figure
3.7(B). The concentration of acetic acid in the upper zone of HSAD reactor still remained at a
high level due to the high degradable solid concentration.

125

126

127

Figure 3.7. Comparison of acetic acid production: (A) batch mode; (B) continuous mode; (C)
increase of UASB height; (D) increase of UASB cross section area; (E) increase of recycled
methanogenic seeds
3.4.5. Comparison of bio-methane production in the HSAD system
Figure 3.8(A), (B), (C), (D) and (E) predict the change of bio-methane production with time
under the conditions of batch mode, continuous mode, the increased UASB height, the increased
UASB cross section area and the increased methanogenic seeds, respectively. Corresponding to
the results of acetic acid, an opposite tendency was shown in the methane production. The
highest methane productivity occurred in the UASB reactor while the highest concentration of
acetic acid occurred in the HSAD reactor. This indicates that more acetic acid was transported by
leachate from the HSAD reactor to the UASB seed reactor, and converted into methane. In the

128

batch mode shown in Figure 3.8(A), methane productivity first increased with time, reached a
peak, and then decreased to zero because acetic acid was depleted.
In the continuous mode at loading rate of 5 kg COD/m3/day shown in Figure 3.8(B), the
methane production was maintained continuously and steadily in the UASB reactor while there
was no methane production in the upper zone of the HSAD reactor. This indicates that the high
degradable solid concentration inhibited the methanogenic activity. If the UASB height was
increased 2-fold as shown in Figure 3.8(C), the methane production had little change compared
with Figure 3.8(B). This was in accordance with the acetic acid concentration change shown in
Figure 3.8(C). If the UASB cross-section area was increased by 2 fold shown in Figure 3.8(D),
the methane production increased in the UASB reactor. However, a longer time was needed to
reach the peak of the methane production. The contribution of the increased UASB cross section
area for reducing acetic acid might be attributed to lower velocity, i.e. lower mixing intensity.
This is in accordance with the research that a vigorous mixing is not suitable in a system where
methanogenesis is the rate limiting step, because only under imperfect mixing conditions do
methanogenic bacteria have the sites where they will be protected from rapid acidogenesis
(Padmasiri et al., 2007; Vavilin and Angelidaki, 2005). If the recycled methanogenic seeds were
increased 2-fold as shown in Figure 3.8(E), the methane production increased in the lower zone
of the HSAD reactor. This indicates methanogenic seeds would like to be isolated from the high
degradable solid concentration.

129

130

131

Figure 3.8. Comparison of bio-methane production: (A) batch mode; (B) continuous mode; (C)
increase of UASB height; (D) increase of UASB cross section area; (E) increase of recycled
methanogenic seeds
3.5. Conclusions

A comprehensive model for a new two-stage HSAD system was established to understand
the system interaction between a HSAD and UASB seed reactor. The predictions indicate that
recycled methanogenic bacteria increased methane concentration and decreased hydrogen
concentrations in the HSAD reactor, while pH increased in the batch mode, but decreased in the
continuous mode with an increase of recycling rate. An organic loading rate of 5 kg COD/m3/day
maintained continuous methane production enhanced by the increased UASB section area and
methanogenic seeds. The UASB height had little impact on the acetic acid concentration and

132

methane production.
Acknowledgements

The authors thank the Washington State Department of Ecology Waste to Fuels program and
their funding via contract # C1000172, and also thank Cynthia Alwine for assisting in
experimental analysis.

133

3.6. References

Arudchelvam, Y., Perinpanayagam, M., Nirmalakhandan, N., 2010. Predicting VFA formation by
dark fermentation of particulate substrates. Bioresource Technology, 101, 7492-7499.
Arvanitoyannis, I.S., 2008. Waste management for the food industries. Academic Press.
Batstone, D., Keller, J., Angelidaki, I., Kalyuzhnyi, S., Pavlostathis, S.G., Rozzi, A., Sanders,
W.T.M., Siegrist, H., Vavilin, V.A., 2002. Anaerobic digestion model No.1 (ADM1). IWA
Publishing, London, UK.
Batstone, D.J., Keller, J., Steyer, J.P., 2006. A review of ADM1 extensions, applications, and
analysis: 2002-2005. Water Science & Technology, 54, 1-10.
Blumensaat, F., Keller, J., 2005. Modelling of two-stage anaerobic digestion using the IWA
Anaerobic Digestion Model No. 1 (ADM1). Water Research, 39, 171-183.
Cavinato, C., Bolzonella, D., Fatone, F., Giuliano, A., Pavan, P., 2011. Two-phase thermophilic
anaerobic digestion process for biohythane production treating biowaste: preliminary
results. Water Science and Technology, 64, 715-721.
Chen, Z.B., Hu, D.X., Zhang, Z.P., Ren, N.Q., Zhu, H.B., 2009. Modeling of two-phase
anaerobic process treating traditional Chinese medicine wastewater with the IWA
Anaerobic Digestion Model No. 1. Bioresource Technology, 100, 4623-4631.
Chynoweth, D.P., Isaacson, R., 1987. Anaerobic Digestion of Biomass. Elsevier Applied Science,
London and New York.
Demirel, B., Yenigun, O., 2002. Two-phase anaerobic digestion processes: a review. Journal of

134

Chemical Technology and Biotechnology, 77, 743-755.

Dinsdale, R.M., Premier, G.C., Hawkes, F.R., Hawkes, D.L., 2000. Two-stage anaerobic
co-digestion of waste activated sludge and fruit/vegetable waste using inclined tubular
digesters. Bioresource Technology, 72, 159-168.
Dohnyos, M., Kosov, B., Zbransk, J., Grau, P., 1985. Production and Utilization of Volatile
Fatty Acids in Various Types of Anaerobic Reactors. Water Science & Technology 17,
191205.
Eaton, A.D., Franson, M.A.H., American Public Health, A., American Water Works, A., Water
Environment, F., 2005. Standard methods for the examination of water & wastewater.
American Public Health Association.
Gali, A., Benabdallah, T., Astals, S., Mata-Alvarez, J., 2009. Modified version of ADM1 model
for agro-waste application. Bioresource Technology, 100, 2783-2790.
Inanc, B., Matsui, S., Ide, S., 1999. Propionic acid accumulation in anaerobic digestion of
carbohydrates: An investigation on the role of hydrogen gas. Water Science and
Technology, 40, 93-100.
Jagadabhi, P.S., Kaparaju, P., Rintala, J., 2011. Two-stage anaerobic digestion of tomato,
cucumber, common reed and grass silage in leach-bed reactors and upflow anaerobic
sludge blanket reactors. Bioresource Technology, 102, 4726-4733.
Koutrouli, E.C., Kalfas, H., Gavala, H.N., Skiadas, I.V., Stamatelatou, K., Lyberatos, G., 2009.
Hydrogen and methane production through two-stage mesophilic anaerobic digestion of

135

olive pulp. Bioresource Technology, 100, 3718-3723.

Lehtomaki, A., Huttunen, S., Lehtinen, T., Rintala, J.A., 2008. Anaerobic digestion of grass
silage in batch leach bed processes for methane production. Bioresource Technology, 99,
3267-3278.
Liu, Y., Xu, H.-L., Show, K.-Y., Tay, J.-H., 2002. Anaerobic granulation technology for
wastewater treatment. World Journal of Microbiology and Biotechnology, 18, 99-113.
Molina, F., Garcia, C., Roca, E., Lema, J.M., 2008. Characterization of anaerobic granular sludge
developed in UASB reactors that treat ethanol, carbohydrates and hydrolyzed protein
based wastewaters. Water Science and Technology, 57, 837-842.
Nizami, A.S., Singh, A., Murphy, J.D., 2011. Design, Commissioning, and Start-Up of a
Sequentially Fed Leach Bed Reactor Complete with an Upflow Anaerobic Sludge
Blanket Digesting Grass Silage. Energy & Fuels, 25, 823-834.
Padmasiri, S.I., Zhang, J.Z., Fitch, M., Norddahl, B., Morgenroth, E., Raskin, L., 2007.
Methanogenic population dynamics and performance of an anaerobic membrane
bioreactor (AnMBR) treating swine manure under high shear conditions. Water Research,
41, 134-144.
Reichert, P., 1998. AQUASIM 2.0: Computer Program for the Identification and Simulation of
Aquatic Systems. Swiss Federal Institute for Environmental Science and Technology
(EAWAG), Switzerland.
Rivard, C.J., Himmel, M.E., Vinzant, T.B., Adney, W.S., Wyman, C.E., Grohmann, K., 1989.

136

Development of a novel laboratory scale high solids reactor for anaerobic-digestion of

processed municipal solid-wastes for the production of methane. Applied Biochemistry
and Biotechnology, 20-1, 461-478.
Rosen, C., Jeppsson, U., 2006. Aspects on ADM1 Implementation within the BSM2
framework,Technical Report. Department of Industrial Electrical Engineering and
Automation, Lund University, Lund, Sweden.
Schievano, A., D'Imporzano, G., Malagutti, L., Fragali, E., Ruboni, G., Adani, F., 2010.
Evaluating inhibition conditions in high-solids anaerobic digestion of organic fraction of
municipal solid waste. Bioresource Technology, 101, 5728-5732.
Schmidt, J.E., Ahring, B.K., 1996. Granular sludge formation in upflow anaerobic sledge blanket
(UASB) reactors. Biotechnology and Bioengineering, 49, 229-246.
Seghezzo, L., Zeeman, G., van Lier, J.B., Hamelers, H.V.M., Lettinga, G., 1998. A review: The
anaerobic treatment of sewage in UASB and EGSB reactors. Bioresource Technology, 65,
175-190.
Sonakya, V., Raizada, N., Hausner, M., Wilderer, P.A., 2007. Microbial populations associated
with fixed- and floating-bed reactors during a two-stage anaerobic process. International
Microbiology, 10, 245-251.
Thamsiriroj, T., Murphy, J.D., 2011. Modelling mono-digestion of grass silage in a 2-stage CSTR
anaerobic digester using ADM1. Bioresource Technology, 102, 948-959.
Vavilin, V.A., Angelidaki, I., 2005. Anaerobic degradation of solid material: Importance of

137

initiation centers for methanogenesis, mixing intensity, and 2D distributed model.

Biotechnology and Bioengineering, 89, 113-122.
Velmurugan, B., Ramanujam, R.A., 2011. Anaerobic digestion of vegetable wastes for biogas
production in a fed-batch reactor. International Journal of Emerging Sciences, 1,
478-486.
Walker, M., Banks, C.J., Heaven, S., 2009. Two-stage anaerobic digestion of biodegradable
municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter.
Bioresource Technology, 100, 4121-4126.
Yu, L., Ma, J., Chen, S., 2011. Numerical simulation of mechanical mixing in high solid
anaerobic digester. Bioresource Technology, 102, 1012-1018.
Zhao, B.-H., Yue, Z.-B., Ni, B.-J., Mu, Y., Yu, H.-Q., Harada, H., 2009. Modeling anaerobic
digestion of aquatic plants by rumen cultures: Cattail as an example. Water Research, 43,
2047-2055.

138

CHAPTER 4 : MULTIPHASE MODELING OF SETTLING AND SUSPENSION IN

ANAEROBIC DIGESTER
4.1. Abstract

Effective suspension and settling are critical for controlling biomass retention in a bioreactor.
In this paper, a multi-fluid model with kinetic theory of granular flow (KTGF) was established to
describe these phenomena in the biowaste particles flow in anaerobic digesters. Solid retention
time (SRT) was added as a parameter into anaerobic digestion No.1 (ADM1) model to evaluate
its effect on the biogas productivity. The model was experimentally validated in a
liquid-gas-solid column reactor with gas and solid volume fraction and granular temperature as
the major variables. The wastewater residence time distribution was also determined through
modeling and measurement to evaluate the mixing pattern in the pilot column reactor. The effect
of restitution coefficient on flow behavior of biowaste particles, including particle settling and
suspension were predicted. Settling and suspension processes of anaerobic digesters were
simulated for lab and pilot-scale reactors with comparisons made for reactor configuration and
geometry model, respectively. This study demonstrated that the multi-fluid model with KTGF
could provide better understanding of impact of suspension and settling upon retaining biomass
particles in the anaerobic digesters.
Keywords: computational fluid dynamics (CFD); anaerobic digester; Kinetic theory of granular

flow (KTGF); solid retention time (SRT); anaerobic digestion No.1 (ADM1) model

139

4.2. Introduction

Anaerobic digestion (AD) is a promising biochemical process for production of renewable

energy, stabilization of waste streams, and recycling nutrients from organic wastes. Retention
times of solid and liquid in an anaerobic digester are critical parameter for design and operation
of a high rate digester. For example, in wastewater treatment, high rate digesters are commonly
designed to retain anaerobic biomass particles including the particulate substrates and the
attached microorganisms, as these work with a very short hydraulic retention time (HRT) which
is less than the maximum growth rate for methanogens, and therefore mechanisms are required
to maintain biomass in the reactor (Climenhaga and Banks, 2008). Suspension and settling of the
biomass particles are essential mechanisms which significantly affect the operation of anaerobic
digesters. In particular, microorganism attachment on particle surfaces needs effective
suspension and settling to gain adequate contact and longer retention so that reactor performance
can be improved (Wang et al., 2010b). Suspension and settling of the biomass particles involves
solid-liquid-gas flow, in particular the nonlinear interactions among the particles and between the
particles and fluid (Cui and Grace, 2007). Although some models such as gravity settling theory
(Vesilind, 2003), selection pressure (Liu et al., 2005), and primary settling efficiency (Descoins
et al., 2012) etc. were developed to simulate the settling process in wastewater treatment
processes, the complicated phase interaction was not considered in these models. To better
understand these complex processes, it is necessary to apply advanced modeling methods.
With the advancement of computer performance in the last decades, many researchers have

140

turned to computational fluid dynamics (CFD) to predict flow fields and improve mixing
performance in anaerobic digesters (Hoffmann et al., 2008; Terashima et al., 2009). The earlier
relevant studies focused on change of velocity distribution in anaerobic digesters. However, this
variable did not completely describe flow fields within the reactors because the complexflow and
phase structure could not be completely revealed. Our earlier studies even with the introduction
of non-Newtonian fluid theory into CFD simulation of manure slurry (Wu and Chen, 2008; Yu et
al., 2011) did not either, explain the flow phenomena of suspension and settling in anaerobic
digester because wastewater and biomass particles were assumed sharing one phase. Some
researchers (Ren et al., 2009; Wang et al., 2010a) developed a three-fluid model (gas-liquid-solid)
to describe the hydrodynamics in an upflow anaerobic sludge blanket (UASB) reactor, the model
nonetheless did not describe biomass particle momentum exchange due to translation and
collision (FLUENT, 2006). To date, few studies of multiphase flow have focused on biomass
particles (Cui and Grace, 2007).
Kinetic theory of granular flow (KTGF) is a recognized approach for consideration of collision,
aggregation and breakup of particle clusters (Gidaspow et al., 2004). Introduced into a two-fluid
model, KTGF has been extensively applied into the simulation of gas-solid (Yu et al., 2007; Yu et
al., 2008), liquid-solid (Huang et al., 2009) and liquid-gas-solid fluidization (Matonis et al.,
2002). Gidaspow et al. (Gidaspow and Huang, 2009; Huang et al., 2009) further introduced
KTGF into non-Newtonian blood flow where plasma and red blood cells (RBCs) were
considered as two phase to describe Fahraeus-Lindqvist effect. Like glass beads in previous

141

studies, RCBs were also assumed to be partial elastic particles. However, unlike the rigid
particles (glass beads and sand particles) in gas-solid fluidization, most of the biowaste particles
(manure and activated sludge particles) are soft and deformable and trend towards inelastic
(Christensen et al., 2011). Parameters of KTGF must be adjusted to better account for such
unique biowaste particle flow phenomena.
In this study, CFD was used to evaluate the biomass retention physical process while
anaerobic digestion No.1 (ADM1) model was used to evaluate the biomass retention biochemical
process. Since the time scales of CFD and ADM1 are much different, they were calculated
separately. A multi-fluid model with KTGF was established to describe the settling and
suspension of dairy manure waste particles in the bottle reactors of lab scale and the pilot column
reactor, respectively. Since intermittent mixing was used in the anaerobic digesters, the settling
process was simulated at the laminar flow without agitation while the suspension process was
simulated at the turbulent flow with a jet agitation. Separate momentum balances equations were
solved for the continuous flowing liquid phase and for the discrete gas and solid phases.
Residence time distribution was measured and predicted to understand the dispersion feature of
the pilot column digester. Additional solid retention time (SRT) was added into anaerobic
digestion No.1 (ADM1) model to evaluate the effect on the biogas productivity in the bottle and
column reactors.

142

4.3. Materials and methods

4.3.1 Experimental section

Laboratory and pilot scale anaerobic digesters

Anaerobic digesters were constructed from 1L glass bottle reactor (lab-scale) and 70L steel
column reactor (pilot-scale). The glass bottle dimensions were 11.7 cm in height and 7.62 cm in
diameter. The bottle reactor was agitated on a 150 rpm shaker for 10 minutes per 2 hours. The
steel column reactor height was 100 cm, the diameter was 30 cm, the cone height was 7.6 cm,
and the cone bottom diameter was 3.8 cm. The wastewater was injected into the column reactor
from the bottom inlet as mixing mechanism. The wastewater flowed out from the side outlet
while the gas went out from the top outlet. The wastewater was recycled back to agitate the
column reactor for 10 minutes per 2 hours.
In the anaerobic digesters, dairy manure and anaerobic sludge were inoculated at a 1:1
volume ratio. The feeding concentration of dairy manure was 9.1 g/L total solids (TS). In the
anaerobic digesters, the solid contents was maintained at an average of 5%TS. When the dairy
manure and fiber particles were settled down in the digesters, there was 25% TS accumulated at
the bottom. The SRT value was adequately controlled by regular discharge of solids at 10 days
from the digesters while HRT was 6 days (Wang et al., 2010b). The source of flushed dairy
manure was the Washington State University Knotts Dairy Center in Pullman, WA, USA.
Anaerobic sludge was sampled from a 35 oC anaerobic digester operating within the Pullman
wastewater treatment plant. Dairy manure particle density was measured as 1700 kg/m3 with its

143

particle size distribution shown in Table 4.1. The samples were taken from top, middle and
bottom of the bottle reactor, respectively before digestion (Frear et al., 2011). The particle sizes
were calculated by weighted average. Table 4.1 shows that 70.06% particles at the range of 0.04
0.06 mm are not settled down from the top of the bottle reactor.

Table 4.1. Particle size distribution in the bottle reactor

Particle size distribution(mm)

Top of
the
bottle

Middle
of the
bottle

Bottom of
the bottle

Total bottle

Percentage (%)
0.04 0.06
0.06 0.07
0.07 0.09
0.09 0.1
0.1 0.13
0.13 0.18
0.18 0.25
0.25 0.30
0.3 0.35
0.35 0.42
0.42 0.59
0.59 1.00
1.0 2.0
2.0 3.33
3.33 5.00
Total percentage (%)
Average particle size (mm)

70.06
18.56
2.10
9.28

7.31
7.82
4.80
4.80
8.06
12.96
10.20
3.48
2.96
3.04
9.66
16.54
8.36

100.00
0.06

100.00
0.41

0.00
0.96
0.41
0.42
0.81
1.20
1.02
0.47
0.48
0.49
1.82
9.74
37.41
12.77
31.98
100.00
2.33

0.19
1.11
0.50
0.52
0.96
1.44
1.21
0.53
0.53
0.55
1.98
9.88
36.79
12.50
31.30
100.00
2.29

Analytic methods

A single phase (wastewater only) system was used to measure residence time distribution in
the column reactor for flow pattern validation and reactor design evaluation. There was no

144

inoculum in the wastewater system and the initial system pH was maintained at 7.0. Therefore,
acetic acid did not give perturbation in the measurement. 50 mL acetic acid was used as a pulse
tracer to measure residence time distribution in the column reactor. pH probe was used to receive
the signal at the exit of the column reactor. pH was monitored on-line with an OM-CP-PH101
pH and Temperature Data Logger connected to an OMEGAPHE-4200 pH probe.
Total solid (TS) was determined according to standard methods (Eaton et al., 2005).
Maximum packing limit was measured by the maximum TS packing at the reactor bottom
because of gravity.
In the digestion process, biogas production was measured using Wet Tip Gas Meters
(wettipgasmeter.com).
4.3.2 Models and Solution
4.3.2.1 Assumptions

The model was constructed to describe multi-phase flow characteristics including liquid
(wastewater), gas (biogas) and solid (dairy manure fiber and biomass particles) in anaerobic
digesters. An Eulerian treatment was used for each phase. The main assumptions were as
follows:
(a) Liquid (wastewater) had the same properties as water (35 oC), i.e. Newtonian fluid.
(b) Gas density and viscosity were 1.139 kg/m3 and 1.910-5 kg/ms, respectively. The gas
bubbles had a diameter of 0.1 mm (Ren et al., 2009).
(c) Solids were assumed to be spherical particles with the same diameter. The average particle

145

diameter was used in anaerobic digesters.

4.3.2.2 Kinetic Theory of Granular Flow (KTGF)

Kinetic theory of granular flow (KTGF) is a recognized approach for studying particle flow
(Gidaspow et al., 2004) based on the Chapman and Cowling kinetic theory of dense gasses
(Chapman and Cowling, 1970). Using a velocity distribution function f(r, Us, t) for a large
collection of granular particles that satisfies the Boltzmann integral-differential equation, and
taking the ensemble average of the single-particle quantity over Eq. 1, the transport equation for
mean quantity in Cartesian tensor notation (Ding and Gidaspow, 1990) is obtained:
f
f

f
+ U s,i
+
( Fi f ) =
t
xi U s,i
t Us

(4.1)

where f is the single-particle velocity distribution function; t is time (s); x is coordinate (m);

Us

is the instantaneous velocity of the particles (m/s); F is the external force per unit mass acting on
each particle (N).
To simulate particles flow, the conservation of mass and momentum equations derived from
an Eulerian-Eulerian approach and Eq.1 are mathematically solved for each phase and the
derivative details were provided in the literature (Ding and Gidaspow, 1990):

q q ) + q q uq = 0
(
t

(4.2)


n 

q q u q + q q u q u q = qp + q + q q g + R pq
t
p =1

( )

(4.3)

where , and u are the volume fraction, density (kg/m3), and velocity (m/s), respectively. p and

146

q represent the different phases including liquid, gas and solid, respectively. Rpq is the drag
coefficient between the pth phase and the qth phase, and g is gravity (m/s2). q is the stress tensor
of the qth phase (Pa). The details for the closure equations are provided in the FLUENT
documents (FLUENT, 2006).
Granular temperature s (m/s)2 is a pseudo-temperature which can be defined as the
random-particle kinetic energy, and is obtained from the frequency distribution of the
instantaneous velocities (Ding and Gidaspow, 1990).
3
1
s = < usus >
2
2

(4.4)

where us (m/s) is the fluctuating velocity of the particles which in turn can be derived from

us' = U s us

(4.5)

where Us (m/s) is the instantaneous velocity of the particles. The solid mean velocity us (m/s) is
defined as
us = U s,i =

1
U s,i fdU s
n

(4.6)

where f and n are the single-particle velocity distribution function and particle number density
respectively.
A transport equation which describes particle collision resulting in a random granular motion
is also derived from Eq.1 (Ding and Gidaspow, 1990; Gidaspow and Huang, 2009):


3

u
=

s : us + ( ks s ) 3R qss
(
)
(
)
s
s
s
s
s
s
s
s
s

2 t

The diffusion coefficient for granular energy ks (kg/ms) is given by

147

(4.7)

ks =

15 s d s s 12 2
12
1 + ( 4 3) s g 0 +
( 41 33 ) s g0

4 ( 41 33 )
5
15

(4.8)

1
(1 + ess )
2

(4.9)

The dissipation of fluctuating energy (W/m3) due to inelastic collision takes the form:
3

12 (1 ess2 ) s2 s g 0s 2

(4.10)

ds

The solids shear viscosity s (Pas) including the collisional, kinetics and frictional
components depends on the magnitude of the particle velocity fluctuations.

s = s,col + s,kin + s,fr

s,col
s,kin =

4

= s sd s g0 (1 + ess ) s
5

(4.11)
1

(4.12)

s sd s s 2

1 + g 0s (1 + ess )( 3ess 1)

(4.13)

ps sin
2 I2D

(4.14)

6 ( 3 ess )

s,fr =

where ess is the coefficient of restitution for particle collisions; g0 is the radial distribution
function; ds is the particle diameter (m); ps is the solids pressure (Pa); is the angle of internal
friction; and I2D is the second invariant of the deviatoric stress tensor.
Acetic acid was used as pulse tracer injected into the column reactor to simulate the
residence time distribution. The concentration of acetic acid was added into the species transport
equation:

l l l


Yac ) + ( l l u lYacl ) = l J acl + l Sacl
(
t

148

(4.15)

where Yac is the mass fraction of acetic acid; J is the diffusion flux (kg/m2s); and S is the species
mass sources (kg/m3s).

4.3.2.3 Solid retention time (SRT) model

The effect of solid retention on the biogas production was evaluated in the ADM1 model.
ADM1 is a highly comprehensive model for anaerobic digestion simulation. As a differential and
algebraic equation set, there are 26 dynamic state concentration variables, 19 biochemical kinetic
processes, 3 gas-liquid transfer kinetic processes and 8 implicit algebraic variables per liquid
vessel (Batstone et al., 2002). The retention time of solid components above hydraulic retention
time tret (day) was added into ADM1 model programmed in the Matlab/Simulink platform. In this
study, SRT was assumed to be constant. This is reasonable approach because SRT value is
adequately controlled by regular discharge of solids (Kleerebezem and van Loosdrecht, 2006).
dX liq ,i
dt

QX in ,i
Vliq

X liq ,i
V
tret, X + liq

rj i , j

(4.16)

j =119

where X is biomass concentration (kg COD/m3). i represents complex composite particulates,

particulates of carbohydrates, proteins, lipids and inerts, and degrading microorganisms of sugar,
amino acid, long chain fatty acid (LCFA), butyrate, propionate, acetate, and hydrogen. Vliq is the
liquid volume (m3). Q is the flow rate (m3/d). j=1-19rjvi,j is the sum of the biochemical rate
coefficients (vi,j) multiplying kinetic rate equations (rj) for process. The kinetic parameters and
their values used in the model are shown in Table 4.2 (Page et al., 2008; Schoen et al., 2009).

149

Table 4.2. Model parameters in the ADM1 model

Constant

Units

Value

Disintegration and hydrolytic parameters

Kdis
d-1
0.5
-1
d
10
Khyd_CH
-1
Khyd_PR
d
10
-1
Khyd_LI
d
10
Kinetics parameters of acidogenesis
Km_su
CODCOD-1d-1 30
Km_aa
CODCOD-1d-1 50
Ks_su
0.5
kgm-3
-3
Ks_aa
0.3
kgm
Ysu
0.1
Yaa
0.08
-1
Kdec_su
d
0.02
-1
d
0.02
Kdec_aa
Kinetics parameters of acetogenecesis
Km_fa
CODCOD-1d-1 6
Km_c4
CODCOD-1d-1 20
Km_pr
CODCOD-1d-1 13
0.4
Ks_fa
kgm-3
-3
Ks_c4
0.2
kgm
-3
Ks_pr
0.1
kgm
Yfa
0.06
Yc4
0.06
0.04
Ypr
Kdec_fa
d-1
0.02
-1
Kdec_c4
d
0.02
-1
Kdec_pr
d
0.02
Kinetics parameters of methanogenesis
Km_ac
(Page et al., 2008)
CODCOD-1d-1 15
-1 -1
Km_h2
35
CODCOD d
Ks_ac
0.05
(Page et al., 2008)
kgm-3
-6
-3
Ks_h2
710
kgm
Yac
0.07
(Page et al., 2008)
Yh2
0.06
Kdec_ac
d-1
0.04
(Page et al., 2008)
-1
Kdec_h2
d
0.02

150

Stoichiometric parameters
fch_xc
fli_xc
fpr_xc
fsi_xc
fxi_xc

0.165
0.277
0.144
0.016
0.101

(Schoen et al., 2009)

(Schoen et al., 2009)
(Schoen et al., 2009)
(Schoen et al., 2009)
(Schoen et al., 2009)

4.3.2.4 Numerical Solution

Two-dimensional (2D) axisymmetric and three-dimensional (3D) geometrical models were
generated by GAMBIT (v.2.3.16), respectively. Grid independence was checked using the
solution-adaptive mesh refinement. The multi-fluid model with KTGF was numerically solved in
FLUENT (v.6.3.26). The no-slip wall boundary condition was used for the liquid phase while the
solid phase is in Johnson and Jackson boundary condition (Johnson and R. Jackson, 1987),
which is partial-slip.
us,w =

6ss,max
3 s s g 0

k
w = s
w n

w =
The

Phase

Coupled

u
n

(4.17)

wall

2
3 ssus,w
g 0 w3 2
wall +
6s,max w

(4.18)

2
3 (1 ess,w
) s s g0 w3 2

SIMPLE

(4.19)

4 s,max

(PC-SIMPLE)

algorithm

was

used

to

solve

velocity-pressure coupled differential equations. The governing equations were discretized into
algebraic equations by the QUICK scheme. In order to decrease the impact of the strong
non-linear characteristic of the model and ensure the good convergence and acceptable

151

computational time, the selective algebraic multigrid (SAMG) solver was used to solve the
multiphase granular flow equations. The details on QUICK scheme and SAMG can be obtained
from FLUENT users Guide. The time step was 10-4 s. The convergence criteria require that the
scaled residuals decrease to 10-3 for continuity, momentum, turbulence and granular temperature
transport equations. The solution procedure includes the following steps:
(a) Define the solver as 2D axisymmetric or 3D, unsteady-state, implicit, and pressure-based.
(b) Activate the 3 phase Eulerian model, and laminar model for settling process while realizable
k- turbulence model for suspension process.
(c) Define material properties of wastewater, biogas and dairy manure fiber and biomass
particles.
(d) Define phase properties and interaction, and active KTGF model in solid phase.
(e) Define boundary conditions for liquid, gas, and solid phases, respectively.
(f) Set the under-relaxation factors and convergence criteria, iteratively solve the governing
equations.

4.4. Results and discussion

4.4.1 Model validation
The multiphase flow model was validated against experimental results from the literature
(Matonis et al., 2002) where an experiment was constructed to measure liquid-gas-solid flow in a
column reactor. Figure 4.1(A) and 4.1(B) show a comparison of gas and solid volume fraction
experimental curves and CFD results in bubbly-coalesced regime for ul = 2.04 cm/s and ug =

152

3.37 cm/s at 4 cm from the horizontal center of the bed. A good agreement (P > 0.05) is observed
between experimental data and CFD results. Further validation was made from a comparison of
granular temperature experimental curves and the CFD results shown in Figure 4.1(C) because
the granular temperature is a basic concept of KTGF (Gidaspow et al., 2004).

153

154

Figure 4.1. Axial and radial distribution of multi-phase variables (A) gas volume fraction; (B)
solid volume fraction; (C) granular temperature; (D) granular collisional viscosity

The model was also validated against the experimental results from wastewater residence
time distribution in the column reactor used in this study. Figure 4.2 shows good agreements (P >
0.05) between the computationally estimated distribution and those from the experiments.

155

Figure 4.2. Wastewater residence time distributions at (A) the reactor exit height of 0.73 m, and
(B) the inlet liquid jet velocity of 0.12 m/s

156

4.4.2 Evaluation of mixing pattern in the pilot column reactor

For lab-scale reactors, it is reasonable that the streams are assumed to be completely mixed.
However, mixing effects are always worse on scale up (Paul et al., 2004). The dimensionless
variance 2 was used to evaluate the mixing pattern in our column reactor. The concentration of
H+ was converted from pH value to calculate the distribution of residence times E(t):
E (t ) =

C (t )

C ( t ) dt

(4.20)

where C(t) (M) is the exit concentration of H+ at the time t.

Dimensionless method was used to characterize the mixing effect in the column reactor.
Dimensionless time , residence time distribution E() and variance 2 were expressed as (Paul
et al., 2004):

tavg = tE ( t )dt
0

(4.22)

tavg

E ( ) = tavg E ( t )

(4.21)

(4.23)

2 = ( 1) E ( ) d
0

(4.24)

where tavg is the mean residence time (day).

Figure 4.2(A) shows the residence time distributions of the column reactor at the inlet liquid
jet with velocity of 0.12 and 1.24 m/s, respectively. The reactor exit height was 0.73 m. The
dimensionless variance 2 is 0.22 and 0.52 at the inlet liquid jet velocity of 0.12 and 1.24 m/s,
respectively. In an ideal continuous flow stirred tank reactor, 2 is 1 while in a plug flow reactor,

157

2 is 0. Well designed reactors in turbulent flow have a 2 value between 0 and 1 (Paul et al.,
2004). This indicates that higher velocity at 1.24 m/s is favorable to approach completely mixing
in the column reactor.
Figure 4.2(B) shows the residence time distributions of the column reactor at the reactor exit
height of 0.25, 0.45 and 0.73 m, respectively. The inlet liquid jet velocity was 0.12 m/s. The
liquid level was the same as the reactor exit height. 2 is 0.22, 0.25, and 0.33 at the reactor exit
height of 0.73, 0.45, and 0.25 m, respectively. This indicates that lower reactor exit height of
0.25 m is favorable to approach completely mixing in the column reactor.
4.4.3 Effect of restitution coefficient ess
Restitution coefficient and maximum packing limit are two parameters that significantly
affect the estimation of the properties of particle flow (Du et al., 2006). Restitution coefficient is
a measure of the elasticity of the collision between two particles while maximum packing limit
specifies the volume fraction of the granular phase. Restitution coefficient and maximum
packing limit were reported at the range of 0.8 1.0 (Patil et al., 2005) and 0.55 0.63 (Du et al.,
2006), respectively.
Although manure wastes were reported as pseudo-plastic property (El-Mashad et al., 2005),
an obvious liquid-solid stratification can be observed in the dairy manure digesters at the range
of 1% 10% TS. The manure and activated sludge particles are soft and deformable
(Christensen et al., 2011). Therefore, the restitution coefficient of manure wastes needs to be
modified because a plastically inelastic collision has a restitution coefficient of zero. Furthermore,

158

the maximum packing limit of dairy manure waste was measured to be 20 30% total solid (TS)
in our experiments.
In this study, the effect of restitution coefficients of 0.01, 0.1 and 0.9999 on granular
temperature were compared. As shown in Figure 4.1(C), the granular temperature increases with
an increase of the restitution coefficient. This is in accordance with the trend of Du et al.s
simulation (Du et al., 2006). The explanation is shown in Figure 4.1(D) where the prediction of
the granular collisional viscosity increases with an increase of the restitution coefficient.
Referring to Eqs. (4.11), (4.12) and (4.13), the granular collisional viscosity and granular kinetic
viscosity increase with increasing restitution coefficient and granular temperature while granular
frictional viscosity is independent of the restitution coefficient and granular temperature. The
manure particle flow tends to have soft and deformable fluid properties due to the lower
contribution of collisional and kinetic components. This might be the reason that the overall
manure waste slurry in the previous research was assumed as a single-phase fluid in the
simulation.
Furthermore, the granular viscosity change calculated between the restitution coefficient of
0.01 and 0.1 is less than 8.4% compared with the granular viscosity change calculated between
the restitution coefficient of 0.9999 and 0.01. This indicates that any restitution coefficient
selected from the range of 0 and 0.1 gives an error small enough to satisfy the requirement of
engineering design for the nearly inelastic particles flow.

159

4.4.4 Manure fiber and biomass particles settling

The settling process of the particles in the bottle reactor was calculated in the laminar
conditions. There were noo external forces added in the reactor. Therefore, the large particles were
settled down to retain microorganism during agitating interval. Figure 4.3(A)
3(A) shows the effect of
the restitution coefficient on the particles settling process in the bottle reactor.
reacto The maximum
packing limits of the solid volume fraction was set as 0.15
15 or 25% total solids (TS). Results
show that there wass a higher particle concentration at the bottom of the bottle at the restitution
coefficient of 0.01. This is because it is easier
easier for particles to settle at lower granular viscosity.

160

Figure 4.3.. Particles settling process in bottle reactor (A) particle size = 2.33mm, ess=0.1,
ess=0.01; (B) particle size = 0.41mm, ess=0.01; (C) particle size = 0.06mm, ess=0.01

161

Figure 4.3(B) and (C) show that the solid volume fraction changes with time for particle
settling at the particle size of 0.41 and 0.06 mm. It can be observed that there was more
homogenous packing distribution for the small particles. This is in line with results from the
experiment of particles settling in the bottle reactor. With a decrease in particle size, it took more
time to settle to the bottom of the reactor. Actually, the particles with less than 0.06 mm were
never settled down to the bottom of the reactor as shown in Table 4.1. The promising feature of
the bioreactor is that with efficient biomass settling and long SRT; granular biomass can be
achieved, allowing for higher biomass to be maintained in the reactor. In a separate study, fibers
in the dairy manure were used as biomass carriers via particles settling (Wang et al., 2010b).

4.4.5 Manure fiber and biomass particles suspension

The suspension particles process in the column reactor was calculated in the turbulence
conditions. The manure waste and fiber suspension were described by the transient distributions
of the solid volume fraction in the column reactor as shown in Figure 4.4(A). The injected liquid
gradually suspended a stream of manure waste and fiber in a central core of the column reactor
to the maximum height, and tended to rain down onto the surrounding annular region between
the hollow core and the column wall. A systematic cyclic pattern of solids movement might be
thus established. The corresponding changes of the liquid volume fraction with time are shown
in Figure 4.4(C). The transient distributions of the turbulent kinetic energy in the liquid-gas-solid
column bed are displayed in Figure 4.4(B). With the inlet liquid jet velocity at 1.0 m/s and
maximum packing limits s,max at 0.15, the flow patterns of the column bed were simulated.

162

The results are at 0.2, 2, 10 and 20 s. These simulated flow patterns are similar to the results
simulated from gas-solid spouted beds in previous literature (Du et al., 2006). The difference is
that the spout zone is in the shape of a larger column in the liquid-gas-solid column reactor than
in the gas-solid spouted bed due to higher drag force between the liquid and solid. The simulated
results suggest that the solid biomass and substrates were well suspended and dispersed to the
global region in the column reactor at the inlet liquid jet velocity of 1.0 m/s.

163

Figure 4.4. Particles suspending process in column reactor at the inlet velocity of 1.0 m/s (A)
solid volume fraction; (B) turbulent kinetic energy; and (C) liquid volume fraction

164

Figure 4.4(B) shows that an egg-shaped zone for the turbulent kinetic energy is gradually
formed; indicating turbulence can enhance mixing effectiveness in this zone. This simulation can
provide a theoretical support for the design of the commonly used egg-shaped digesters which
are supposed to achieve higher mixing efficiency (Tchobanoglous et al., 2004). Since one of the
most important properties of turbulence is efficient mixing, which occurs as velocity fluctuations
break up macroscopic non-uniformities of momentum, temperature and compositions into
smaller scale, the results of the turbulent kinetic energy show that biomass particles in the spout
zone were effectively dispersed towards the mixed liquor from the central zone to the
surrounding zone. Therefore, biomass particles circulation inside the digester can be established
to realize effective contact between biomass and substrate.

4.4.6 Comparison of different digester configurations and dimensional models

Figure 4.5(A) and (B) show the effect of the liquid outlet height on the distribution of the
solid volume fraction in the column reactor. Reactor configuration has an impact on the flow
pattern of liquid and microbial aggregates in the reactor (Wang et al., 2009). A comparison of the
effects of different liquid outlet height shows the different hydrodynamic behaviors in terms of
interactive patterns between flow and microbial aggregation. The height of the liquid level was
0.45 m both at the liquid outlet height of 0.45 m and 0.25 m. The computed results at 15 s show
that more complete biomass suspension in the digester with the liquid outlet height at 0.45 m
than that at 0.25 m. This indicates that the configuration of the column reactor was optimized to
creat more biomass suspension in the operation of anaerobic digester.

165

Figure 4.5.. Comparison of the different digester configurations (A, B: u = 2 m/s, t = 15 s) and
dimensional models (C, D: u = 1 m/s, t = 10 s)

166

Figure 4.5(C) and (D) show a comparison of 2D axisymmetric and 3D computed results. The
inlet velocity was at 1 m/s and the results of 10 s were obtained. Similar flow patterns were
observed in the simulation. In the annulus zone the solids volume fraction simulated approached
the specified s,max while

the solid phase had the largest velocity at the spout channel. This

suggests that the 2D axisymmetric model can be used as a substitution for the 3D model to avoid
higher computational costs when predicting flow fields in the liquid-gas-solid column reactor.
The difference of numerical contour between the 2D and 3D is that the symmetrical shape can be
observed in 2D instead of 3D due to the effect of the liquid outlet. The effect of the liquid outlet
also makes this simulation different from traditional simulations for the gas-solid spouted bed
where the fountain went up vertically instead of deflecting to the sides (Du et al., 2006).

4.4.7 Biomass retention effect and digester performance

To understand the role of biomass retention caused by particles settling and suspension, the
effect of SRT on biogas production was tested and predicted in the bottle and column reactors.
The initial and operational conditions are shown in Table 4.3 and 4.4.

167

Table 4.3. Dairy manure initial concentration

Material properties

Unit

Dairy manure waste

Total solids (TS)

g/l

Volatile solids (VS)

g/l

7.6

Particulate COD (CODp)

gCOD/l

53.4

Soluble COD (CODs)

gCOD/l

9.6

Total Kjeldahl nitrogen (TKN)

gN/l

0.36

Total ammonium (TAN)

gN/l

0.13

Organic nitrogen

gP/l

0.23

Volatile fatty acids (VFA)

gCOD/l

1.69

Acetic acid
Propionic acid
Butyric acid
Valeric acid

gCOD/l

1.16
0.30
0.16
0.07

9.1

gCOD/l
gCOD/l
gCOD/l

Table 4.4. Operational conditions and experimental results

Operational conditions

Bottle reactor

Column reactor

0.17
6.0
9.58
5.33
0.89

10.0
6.0
9.58
5.33
0.89

0.25
0.28
0.18
0.20

0.22
0.25
0.18
0.16

Flow rate (L/d)

Hydraulic retention time (d)
Mean total solid content (g/L)
Mean volatile solid content (g/L)
Loading rate (g VS/L/d)
Experimental results
Biogas production rate (L/L/d)
Biogas yield (L/g VS)
Methane production rate (L/L/d)
Methane yield (L/g VS)

Figure 4.6(A) shows changes of biogas production with digestion time in the bottle reactor.
Figure 4.6(B) shows that the biogas production increased with digestion time in the column
reactor. The model has obtained good agreement with experiment for the biogas production at the

168

SRT of 10 days (P > 0.05). Furthermore, the model predicted that both biogas productions in the
bottle and column reactors increased faster with a first increase of SRT, and then gradually
reached a maximum value. This trend was in accordance with the obtained relationship between
biogas production and SRT in a semi-CSTR (continuous stirred tank reactor), suggesting that the
VFA production and consumption stabilize at SRT > 10 days and there is an optimum SRT for
the balance of biomass growth and lysis (Appels et al., 2008). The related experimental results
were also reported for AD of dewatered-sewage sludge using CSTR (Nges and Liu, 2010).

169

Figure 4.6. Effect of biomass retention time on biogas production (A) bottle reactor (B) column
reactor

170

To obtain the same digester performance in the column reactor as that in the bottle reactor,
the necessary mixing should be provided because it may result in severe manufacturing problems
on scale-up, ranging from costly corrections in the plant to complete failure of a process (Paul et
al., 2004). Based on the suggestion of the aforementioned CFD results, the inlet liquid jet
velocity at 1.24 m/s was used to mix microorganisms and substrates in the column reactor
intermittently. Via particles settling and suspension, HRT and SRT were uncoupled in the
anaerobic digestion allowing robust and stable operation over an extended period, as well as
resilience to disturbance (Climenhaga and Banks, 2008). Table 4.4 shows the experimental
results of anaerobic digestion compared between the bottle and column reactors. There is no
statistically significant difference (P > 0.05) in the experimental results of the biogas production
rate, biogas yied, methane production rate, and methane yeild between the bottle and column
reactors, suggesting that the scale-up effect of the hydrodynamic nature in the column reactor
was reduced by the suggested mixing strategy. Therefore, the multi-fluid model with KTGF can
be used to visually select the suitable flow patterns that are difficultly, expensively or impossibly
obtained for the digester design.

4.5. Conclusions
The mechnisim of a high rate digestor was explored by a multi-fluid model with KTGF. This
mechnisim was characterized by separating the SRT from HRT to retain more microoganisims in
the anaerobic digestors. The simulation results suggest that the dairy manure particles tend to

171

have soft and deformable fluid properties due to the lower contribution of collisional and kinetic
components. The mechanism of suspension and settling in bioreactors was clearly elucidated by
this model allowing visual selection of the suitable flow patterns. The evaluation for biomass
retention allowed the determination of the optimum SRT in anaerobic digestion. Aided with CFD
simulation, the scale-up effect of the hydrodynamic nature from the bottle reactor to the column
reactor was reduced. The approach established in this work would provide an effective
methodology to better explore the flow behavior in digester and the biowaste particles flow
dynamics.

Acknowledgement
The authors thank the Washington State Department of Ecology Waste to Fuels program and
their funding via contract # C1000172.

172

4.6. References
Appels, L., Baeyens, J., Degreve, J., Dewil, R., 2008. Principles and potential of the anaerobic
digestion of waste-activated sludge. Progress in Energy and Combustion Science, 34,
755-781.
Batstone, D., Keller, J., Angelidaki, I., Kalyuzhnyi, S., Pavlostathis, S.G., Rozzi, A., Sanders,
W.T.M., Siegrist, H., Vavilin, V.A., 2002. Anaerobic digestion model No.1 (ADM1). IWA
Publishing, London, UK.
Chapman, S., Cowling, T., 1970. The mathematical theory of non-uniform gases. Cambridge
University Press, London.
Christensen, M.L., Johansson, C., Sedin, M., Keiding, K., 2011. Nonlinear filtration behavior of
soft particles: Effect of dynamic cake compression. Powder Technology, 207, 428-436.
Climenhaga, M.A., Banks, C.J., 2008. Uncoupling of liquid and solid retention times in
anaerobic digestion of catering wastes. Water Science and Technology, 58, 1581-1587.
Cui, H.P., Grace, J.R., 2007. Flow of pulp fibre suspension and slurries: A review. International

Journal of Multiphase Flow, 33, 921-934.

Descoins, N., Deleris, S., Lestienne, R., Trouve, E., Marechal, F., 2012. Energy efficiency in
waste water treatments plants: Optimization of activated sludge process coupled with
anaerobic digestion. Energy, 41, 153-164.
Ding, J., Gidaspow, D., 1990. A bubbling fluidization model using kinetic theory of granular
flow. AIChE Journal, 36, 523-538.

173

Du, W., Bao, X.J., Xu, J., Wei, W.S., 2006. Computational fluid dynamics (CFD) modeling of
spouted bed: Influence of frictional stress, maximum packing limit and coefficient of
restitution of particles. Chemical Engineering Science, 61, 4558-4570.
Eaton, A.D., Franson, M.A.H., American Public Health, A., American Water Works, A., Water
Environment, F., 2005. Standard methods for the examination of water & wastewater.
American Public Health Association.
El-Mashad, H.M., van Loon, W.K.P., Zeeman, G., Bot, G.P.A., 2005. Rheological properties of
dairy cattle manure. Bioresource Technology, 96, 531535.
FLUENT, 2006. FLUENT 6.3 User's Guide. FLUENT, Inc. , Lebanon, NH.
Frear, C., Wang, Z.W., Li, C.L., Chen, S.L., 2011. Biogas potential and microbial population
distributions in flushed dairy manure and implications on anaerobic digestion technology.

Journal of Chemical Technology and Biotechnology, 86, 145-152.

Gidaspow, D., Huang, J., 2009. Kinetic theory based model for blood flow and its viscosity.

Annals of Biomedical Engineering, 37, 1534-1545.

Gidaspow, D., Jung, J., Singh, R.K., 2004. Hydrodynamics of fluidization using kinetic theory:
an emerging paradigm: 2002 Flour-Daniel lecture. Powder Technology, 148, 123-141.
Hoffmann, R.A., Garcia, M.L., Veskivar, M., Karim, K., Al-Dahhan, M.H., Angenent, L.T., 2008.
Effect of shear on performance and microbial ecology of continuously stirred anaerobic
digesters treating animal manure. Biotechnology and Bioengineering, 100, 3848.
Huang, J., Lyczkowski, R.W., Gidaspow, D., 2009. Pulsatile flow in a coronary artery using

174

multiphase kinetic theory. Journal of Biomechanics, 42, 743-754.

Johnson, P.C., R. Jackson, 1987. Frictionalcollisional constitutive relations for granular
materials, with application to plane shearing. Journal of Fluid Mechanics 176, 67-93.
Kleerebezem, R., van Loosdrecht, M.C.M., 2006. Critical analysis of some concepts proposed in
ADM1. Water Science and Technology, 54, 51-57.
Liu, Y., Wang, Z.W., Qin, L., Liu, Y.Q., Tay, J.H., 2005. Selection pressure-driven aerobic
granulation in a sequencing batch reactor. Applied Microbiology and Biotechnology, 67,
26-32.
Matonis, D., Gidaspow, D., Bahary, M., 2002. CFD simulation of flow and turbulence in a slurry
bubble column. AIChE Journal, 48, 1413-1429.
Nges, I.A., Liu, J., 2010. Effects of solid retention time on anaerobic digestion of
dewatered-sewage sludge in mesophilic and thermophilic conditions. Renewable Energy,
35, 2200-2206.
Page, D.I., Hickey, K.L., Narula, R., Main, A.L., Grimberg, S.J., 2008. Modeling anaerobic
digestion of dairy manure using the IWA Anaerobic Digestion Model no. 1 (ADM1).

Water Science and Technology, 58, 689-695.

Patil, D.J., Annaland, A.V., Kuipers, J.A.M., 2005. Critical comparison of hydrodynamic models
for gas-solid fluidized beds - Part II: freely bubbling gas-solid fluidized beds. Chemical

Engineering Science, 60, 73-84.

Paul, E.L., Atiemo-Obeng, V.A., Kresta, S.M., 2004. Handbook of industrial mixing: science and

175

practice. Wiley-Interscience.
Ren, T.T., Mu, Y., Ni, B.J., Yu, H.Q., 2009. Hydrodynamics of upflow anaerobic sludge blanket
reactors. AIChE Journal, 55, 516-528.
Schoen, M.A., 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, 5648-5655.

Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2004. Wastewater Engineering: Treatment and
Reuse. 4th ed. McGraw-Hill, New York, Metcalf & Eddy Inc.
Terashima, M., Goel, R., Komatsu, K., Yasui, H., Takahashi, H., Li, Y.Y., Noike, T., 2009. CFD
simulation of mixing in anaerobic digesters. Bioresource Technology, 100, 2228-2233.
Vesilind, P.A., 2003. Wastewater Treatment Plant Design. Water Environment Federation.
Wang, L.K., Shammas, N.K., Pereira, N.C., Hung, Y.T., 2009. Advanced Biological Treatment
Processes. Humana Press.
Wang, X., Ding, J., Guo, W.Q., Ren, N.Q., 2010a. A hydrodynamics-reaction kinetics coupled
model for evaluating bioreactors derived from CFD simulation. Bioresource Technology,
101, 9749-9757.
Wang, Z.W., Ma, J., Chen, S., 2010b. Bipolar effects of settling time on active biomass retention
in anaerobic sequencing batch reactors digesting flushed dairy manure. Bioresource

Technology, 102, 697-702.

Wu, B., Chen, S., 2008. CFD simulation of non-Newtonian fluid flow in anaerobic digesters.

176

Biotechnology and Bioengineering, 99, 700711.

Yu, L., Lu, J., Zhang, X.P., Zhang, S.J., 2007. Numerical simulation of the bubbling fluidized bed
coal gasification by the kinetic theory of granular flow (KTGF). Fuel, 86, 722-734.
Yu, L., Lu, J., Zhang, X.P., Zhang, S.J., Wang, X.L., 2008. Two fluid model using kinetic theory
for modeling of one-step hydrogen production gasifier. AIChE Journal, 54, 2833-2851.
Yu, L., Ma, J., Chen, S., 2011. Numerical simulation of mechanical mixing in high solid
anaerobic digester. Bioresource Technology, 102, 1012-1018.

177

CHAPTER 5 : GENERAL CONCLUSIONS

In this study, the flow, mass trassfer, and biochemical reactions in anaerobic digesters were
numerically studied. To develop economically attractive commercial technology for anaerobic
digestion of biomass feedstocks, it is necessary to take the physical and chemical properties of
the feedstock into design and performance considerations. CFD and ADM1 were used in this
study to assist in the development of a new high solid anaerobic digestion system for highly
degradable organic fraction of municipal solid wastes (OFMSW). A multi-fluid model with
KTGF was developed to help understand high rate anerobic digestion process. Major conclusions
obtained from the research are summarized below.
(1) Mechanical mixing in high solids digestion was numerically simulated, and flow
characteristics were obtained, including flow patterns, pumping, shear and power. The model
framework proposed in this study was used to predict mixing behavior of a HSAD reactor.
Impeller A-310 was not suitable for agitating a high solid digester, and its mixing performance
was impaired with increasing total solids, as the percentage of dead zone increases. Impeller
A-310 had relatively small shear stress but a large shear rate, while the helical ribbon had
relatively large shear stress but a small shear rate. The helical ribbon had a better potential to be
more suitable for the mixing of high solids digesters than the Impeller A-30. Its low shear
environment is suitable for microbial flocs in anaerobic digestion. Through optimization, the
power of the helical ribbon used in high solids digesters can be reduced significantly.
(2) A new two-stage HSAD system was developed as a high rate anaerobic digestion system

178

for highly degradable OFMSW (food waste). A comprehensive model based on ADM1 was used
to understand the system interaction between a HSAD and UASB reactor. Due to low
lignocellulose contents in food waste, it is highly suitable for rapid fermentation. Most of the
VFA and COD in food waste were consumed in about 10 days. The predictions indicate that a
high rate HSAD system could be achieved by adjusting major parameters such as pH, recycled
methanogenic bacteria, and UASB section area. Recycled methanogenic bacteria increased
methane concentration and decreased hydrogen concentrations in the HSAD reactor, while pH
increased in the batch mode, but decreased in the continuous mode with an increase of recycling
rate. An organic loading rate of 5 kg COD/m3/day maintained continuous methane production
and enhanced by the increased UASB section area and methanogenic seeds. The UASB height
had little impact on the acetic acid concentration and methane production.
(3) The mechnisim of a high rate digestor was explored by a multi-fluid model with KTGF.
This mechnisim was characterized by separating the SRT from HRT to retain more
microoganisims in the anaerobic digestors. The simulation results suggest that the dairy manure
particles tend to have soft and deformable fluid properties due to the lower contribution of
collisional and kinetic components. The mechanism of suspension and settling in bioreactors was
clearly elucidated by this model allowing visual selection of the suitable flow patterns. The
evaluation for biomass retention allowed the determination of the optimum SRT in anaerobic
digestion. Aided with CFD simulation, the scale-up effect of the hydrodynamic nature from the
bottle reactor to the column reactor was reduced. The approach established in this work would

179

provide an effective methodology to better explore the flow behavior in digester and the
biowaste particles flow dynamics.
There is a need for further research and improvement of the mathematical models to
establish the link between mixing and biochemical reactions. With additional knowledge, the
mass transfer at the cell level into the microorganisms within a flow fields might be predicted
and the effect of mixing on the performance of anaerobic digestion assessed. Integrated models
will promote their applications in full-scale plant design, operation and optimization.

180