You are on page 1of 92

START-UP O F A LABORATORY-SCALE

ANAEROBIC SEQUENCING BATCH REACTOR


TREATING GLUCOSE

Ioannis Sbizris

A tbesis submitted in conformity with the requirements


for the degree of Master of Applieà Science
Graduate Department of Civil Engineering
University of Toronto

@ Copyright by Ioannis Shizas, 2000


1+1 National Library
of Canada
Bibliithéque nationale
du Canada
Acquisitions and Acquisitions et
Bibliographie Services services bibliographiques
395 Wellington Street 395. rue Wdlingtm
O(Mwa0N K l A û N 4 OttawaûN K 1 A W
cana& canada

The author has granted a non- L'auteur a accordé une licence non
exclusive licence allowing the exclusive permettant à la
National Library of Canada to Bibliothèque nationale du Canada de
reproduce, loan, distribute or seil reproduire, prêter, distribuer ou
copies of this thesis in microform, vendre des copies de cette thèse sous
paper or electronic formats. la forme de microfiche/nlm, de
reproduction sur papier ou sur format
électronique.

The author retains ownership of the L'auteur conserve la propriété du


copyright in this thesis. Neither the droit d'auteur qui protège cette thèse.
thesis nor substantial extracts fiom it Ni la thèse ni des exnaits substantiels
may be p ~ t e or
d otherwise de celle-ci ne doivent être imprimés
reproduced without the author's ou autrement reproduits sans son
permission. autorisation.
Start-Up of a Laboratory-Scale Anaerobic
Sequcacing Batch Reactor Treating Glucose

Ioannis Shizris
Master of Applid Science, 2000
Graduate Department of C M Engineering
University of Toronto

A laboratory-scale anaerobic sequencing batch reactor (ANSBR) treating glucose

at 2 1O C was started-up and loaded to an organic loading rate (OLR)of 3 kg-m-3-d-'.

Higher loadings were not achieved due to acid accumulation and pH inhibition.

Operational parameters such as influent concentration, total cycle tirne, and fill-to-cycle

tirne (WC) ratios were identified that could be modined to improve the reactor's

pefiormance without the need for extemal pH control.

In addition, a straightfoward experimental methad was described that could

determine substrate-specific kinetic parameters and active biomass hctions necessary

for use in mathematical models of the ANSBR process. Low levels of aceticlastic

methanogens (0.48 %) and total active biomass (17.4 %) were measured in the

microcosm studies likely due to solids washout and temperature "shock". Mode1

predictions of the five sets of ANSBR oprational conditions shidied in the labonitory

generally agreed with experimental results for 6000 mg& glucose influent, but dBered

for 3000 mg/L.


ACKNOWLEDGMENTS

1thank my supervisor, Prof. David M. Bagley, for his comments and suggestions

which were key to developing and complethg this thesis and research work. Without bis

guidance this work would not have been possible. Also, 1thank my second reader, Prof.

Robert C. Andrews for his comments and suggestions regarding this thesis.

1 thank my feiiow students, Jerry, Mike, Yale, and Chi, for their help and

fiiendship which made working towards my Master's degree less troublesome thaa 1

initially thought. 1 must also thank Russell D'Souza and the other snidents in the lab for

helping me with analytical and laboratory procedures.

Finally, 1 wish to thank my family and friends who supported me throughout this

research work. 1would not have been able to make it without you.

This work was financiaily assisted by the Natural Sciences and Engineering

Research Council of Canada, the Ontario Ministry of Energy, Science, and Technology:

Singapore - Ontario Joint Research Programme, and the Centre for Research in Earth and

Space Technology.

iii
TABLE OF CONTENTS

Section Page

ABSTRACT

ACKNOWLEDGEMENTS iii

TABLE OF CONTENTS
LIST OF TABLES

LIST OF FIGURES vii

LIST OF SYMBOLS AND ABBREVIATIONS

1. INTRODUCTION

2. LITERATURE REVIEW
2.1 Anaerobic Bidegradation of Glucose
2.2 The Anaerobic Sequencing Batch Reactor (ANSBR)
2.3 AUcalinity and pH Issues
2.4 Modelling the Anaerobic Treatment Process
2.5 Substrate-Specifïc Kinetic Parameters and Active Biomass
2.6 Biomass Granulation

3. M A T E W S AND METHODS
3.1 Lab-Scale ANSBR Set-Up
3.2 Lab-Scale Reactor Operation
3.3 Lab-Scale Basal Medium
3.4 Measurement of Volatile Fatty Acids V A S )
3.5 Measurement of Hydrogen
3.6 Measurement of Glucose
3.7 Bottle Study Methods
3.8 Measurement of Aikalinity, pH, Soluble COD,and Suspended Solids

4.INITIAL OPERATION
4.1 Summary of Lab-Scale Operation
4.2 Phase 1 - Initial Set-Up
4.3 Phase 2 - First Addition of Granulated Sludge to the Reactor
4.4 Phase 3 - Final Configuration
TABLE OF CONTENTS, coat.
Section Page

5. MODIFMNG OPERATIONAL PARAMETERS


5.1 Experimental Plan
5.2 Soluble COD vs. Tirne
5.3 pH vs. Time
5.4 Glucose and VFAs vs. Time
5.5 Summary

6. SUBSTRATE-SPECIFIC KINETIC PARAMETERS AM) BIOMASS


6.1 Motivation for Mcasuring Suôstrate-Specific Parameters
6.2 Method for Measuring Substrate-Specinc Parameters
6.3 Experimental Protoc01
6.4 Resuits Obtained
6.5 Summary

7. COMPUTER MODEL SIMULATIONS


7.1 Simulation Methodology
7.2 Simulation Results for pH
7.3 Simulation Results for VFAs
7.4 Siunmaly

8. DISCUSSION AND CONCLUSIONS

10.REFERENCES
LIST OF TABLES

Table Page

2.1 Typical Acidogenic Pathways


2.2 Typical Acetogenic Pathways
2.3 Summary of ANSBR Experimental Work

3.1 Lab-Scale Measurements Taken and Average Frequency


3 -2 Initial Basal Medium Composition
3 -3 Modified Basal Medium Composition

4.1 Summary of Phases of Lab-Scale ANSBR Operation


4.2 Soiids Profile for the ANSBR on Day 28
4.3 pH and Effluent SCOD and VFAs for Days 206 and 207

5.1 Experimental Parameters


5.2 SCOD Removal Efficiencies

6.1 Substrate-Specific Khetic Parameters and Active Biomass


6.2 Cornparison of k Values for Anaerobic Bidegradation

7.1 Substrate-Specific Parameters Used in the Simulations


LIST OF FIGURES

Figure Page

3.1 Lab-Scale Reactor Set-Up


3.2 IC Eluent Profiles

4.1 Effluent SCOD Concentrations Measured for Days O Through 227


4.2 Initial SCOD Removal
4.3 Initial VFA Results
4.4 Effluent SCOD From Days 71 to 100
4.5 Effluent SCOD After First Addition of Granulated Sludge
4.6 Effluent SCOD and VFAs From Days 178 to 181
4.7 Glucose Concentration vs. Time for 5 hr and 10 hr Fil1 Times on Day 185
4.8 Effluent SCOD and Propionate From Days 185 to 188
4.9 influent Feed Titration Curve
4.10 Reactor Efnuent Titration Curve

5.1 SCOD and S u n of Individual Components vs. Time for Run 1


5.2 SCOD vs. Tirne for Runs 2 Through 5
5.3 Distribution o f Products vs. T h e for Run 1
5.4 pH vs. Time for Al1 Runs
5.5 Relationship Between pH and Total Acids for Run 1
5.6 Glucose Concentration vs. Time for AU Runs
5.7 Lactate Concentration vs. Time for All Runs
5.8 Propionate Concentration vs. Time for Al1 Runs
5 -9 Acetate Concentration vs. Time for Al1 Runs

6.1 Glucose Degradation Curves


6.2 Hydrogen Degradation Curves
6.3 Lactate Degradation Curves
6.4 Acetate Degradation Curves
6.5 Propionate Degradation Curves
6.6 Butyrate Degradation Curves

vii
LIST OF FIGURES, cont.

Figure Page
7.1 pH vs. Time for AU 5 Runs Using Parameters From Brodkorb (1998) 60
7.2 pH vs. Time for Ail 5 Runs Using Parameters From This Work 61
7.3 pH Cornparisons for Run 1 61
7.4 pH Cornparisons for Run 2 62
7.5 pH Cornparisons for Run 3 63
7.6 Lactate vs. Time for AI1 5 Runs Using Parameters From Brodkorb (1998) 64
7.7 Propionate vs. Time for Al1 5 Runs Using Parameters From Brodkorb (1998) 64
7.8 Acetate vs. T h e for Al1 5 Runs Using Parameters From This Work 65
7.9 ExpeNnental and Simulation Results for Run 1 Using Parameters
fiom Brodkorb (1 998) 65
7.10 Experimental and Simulation Results for Run 1 Using Parameters
fkom This Work 66
7.1 1 Experimental and Simulation Resdts for Run 3 Using Parameters
fiom Brodkorb (1 998) 66
7.12 Experimental and Simulation Results for Run 3 Using Parameters
fiom This Work 67
LIST OF SYMBOLS AND ABBREVIATIONS

SymboV
Ab breviation Definition
ADP adenosine diphosphate
Alk carbonate alkaiinity
ANSBR anaerobic sequencing batch reactor
ATP adenosine triphosphate
COD chemical oxygen demand
ECP extracelluiar polymer
F/C fill-to-cycle time ratio
FIM food-to-microorganism ratio
AG Gibb's k e energy change
GC gas chmatograph
HRT hydraulic retention t h e
WWQ International Association on Water Quality
IC ion chromatograph
ka maximum specific substrate utilization rate (based on X.)
kt maximum specific substrate utilization rate (based on X,)
Ks half-saturation constant
Kw
KI; K1;KH.
dissociation constant for water (10-l4at 25°C
carbonate system equilibrium constants (10 ,
23.
10-10.3.
, 1O-'.', respectively)
NAD nicotinamide adenine dinucleotide
NADH nicotinamide adenine dinucleotide (reduced form)
NFDM non-fat dry milk
OLR organic loading rate
PA partial pressure of gas A
S substrate concentration
SCOD soluble chemical oxygen demand
TCD thermal conductivity detector
TMS trace metal solution
TSS total suspended solids
UASB upflow anaerobic sludge blanket
VFA volatile fatty acid
VSS volatile suspended solids
Xa active biomass concentration
xt total biomass concentration
Y microbial growth yield
1. INTRODUCTION

Wastewater must be appropriately tnated befm it can be discharged to the

environment. Typically, wastewater treatment systems have included the use of

microorganisms which convert the biodegradable hction of the wastewater to end-

products and biomass which are then disposeci of in an appropriate manner.

Anaerobic treatment systems do not require oxygen as an electron accepter, thus

eliminating the cost associated with oxygen transfer. In addition, they produce

significantly less biomass than aerobic systems for the same level of treatment, greatly

reducing the cost of handlhg and disposing excess sludge. Also, anaerobic treatment

systems have the added advantage of producing methane gas which can be collected and

used as an energy source.

Many variations of anaerobic treatment systems have been developed including

upflow anaerobic sludge blankets (UASBs), upflow packed bed reactors, and anaerobic

digesters (Speece, 1983). A relatively new anaerobic process, called the anaerobic

sequencing batch reactor (ANSBR)has been developed by Dague and co-workers at

Iowa State University. The ANSBR has not been used as widely in indusûy as other

processes because the highest reported loading achievable by a lab-scale ANSBR

19 k g ~ ~ ~ - m ' 3(Angenent
-d-' and Dague, 1995), is significantiy lower than the loading of

100 k g ~ O ~ - r n "reported
d for lab-sale UASBs (Fang and Chui, 1993). However, the

ANSBR process has potential due to its relative ease of operation, flexibility (Fernandes

et al., 1993), and the fact that it uses the same vesse1 for both reacting and setthg the

wastewater (Schmit and Dague, 1993; Dague and Pidaparti, 1992).


Many mathematical models have been developed for anaerobic treatrnent, but few

have k e n developed for the ANSBR process. Previously developed ANSBR models

(e-g. Sung and Dague, 1992; Fernandes et al., 1993) have been overly simplistic, and they

have not provided the hdamental understanding necessary to improve the process. The

recent ANSBR d e l developed by Brodkorb (1998) and Bagley and Brodkorb (1 999)

more fully describes the physical, chernical, and microbiological processes that occur in

ANSBR reactors. As a result of the model's structure, substrate-specific kinetic

parameters and active biomass fiactions are required to accurately model the treatment

process, and these parameters have k e n difficult to measure in the laboratory. A simple,

inexpensive method to measure these parameters would be extremely beneficial to ensure

that the model can accurately predict experimental values.

Therefore, the objectives of this thesis are:

1. To maximize the organic loaàing treatable by a laboratory-scale ANSBR treating

glucose, and to determine barriers to reaching higher loadings.

2. To identie ANSBR operational parameters that may be modified to help improve the

reactor's biodegradation performance.

3. To directiy measure substrate-s~ifickinetic parameters and active biomass fiactions

in the laboratory using a simple and inexpensive method.

4. To sirnulate the performance of a laboratory-scale ANSBR using the model

developed by Brodkorb (1998) and Bagley and Brodkorb (1999), and to compare the

simulation results with experimental values obtained.


2. LITERATURE REVIEW

2.1 Anaerobic Biodegradation of GIucose

in aerobic treatment systems, glucose cau be converted completely to carbon

dioxide and water by individual microorganisms. However, in anaerobic systems, a

complex consortium of microorganisms is required to completely degrade glucose to the

end-products of methane (Ch)and carbon dioxide ( C a ) .

If the wastewater to be treated by au anaerobic system contains complex

carbohydrates, these are h t hydrofyzed by facultative, chemoheterotrophic

microorganisms to glucose molecules (Speece, 1983; McCarty and Smith, 1986). Next,

the glucose may be fermented iato volatile fatty acids (VFAs) other than acetate by

acidogenic, or acid-forming, microorganisms (McCarty and Smith, 1986). Some typical

acidogenic pathways and examples of microorganisms which mediate them are given in

Table 2.1.

Table 2.1 Typical Acidogenic Pathways

Substrate Product Example of Micmo~anism Reference


glucose butyrate Buîyribacter ium merhylotrophicunt Zeikus et al. (1980)
glucose propionate Propionibacterium acnes Wofin and Miller (1994)
glucose lactate Streptococcus spp. Brock and Maàigan
(199 1)
lactate propionate Propionibacterium spp. Brock and Madigan
(19911

Next, the VFAs (other than acetate) are converted by acetogenic microorganisms

into acetate. (For a thorough review of acetogenic processes, see Drake, 1994.) Some

typicd acetogenic pathways and examples of microorganisms which mediate them are

given in Table 2.2.


Table 2.2 Typical Acetogenic Pathways

Substrate Product Examplt of Microorganhm Refennce


giucose acetate Clostridium thetmoaceticum Fontaine et al. (1942)
propionate acetate Sjmtrophobacter wof inil Stams and Plugge ( 1994)
butyrate acetate Synh.ophomonas spp. Brock and Madigan (199 1)

The preferred reaction for acid-forming bacteria is the conversion of glucose to

acetate, since it provides the most energy for growth. The formation of other acids is

typically the bactena's response to shffik loadings and accumulations of bydrogen in the

reactor (Mosey, I 983).

Propionate and butyrate degradation to acetate are difficult due to thermodynamic

issues. For instance, the degradation of propionate to acetate at standard conditions

(temp = 25°C; pH = 7; al1 other components at unit activity) is thermodynamicdly

impossible, with a Gibb's free energy change, AG0'value of +7l.67 kJ/mol (McCarty and

Smith, 1986). Similady, the conversion of butyrate to acetate bas a AGo' value of +48.3O

kJ/mol (McCarty and Smith, 1986). For these reactions to become thermodynamically

favourabfe, the partial pressure of hydrogen in the reactor must be extremely low. For

propionate degradation, the partial pressure of hydrogen must f d l between approximately

1o4 to 104 atm (McCarty and Smith, 1986). To maintain these fow hydrogen partial

pressures, the H2must be consumed by another group of microorganisms referred to as

hydrogen-utilizing, or hydrogenotrophic methanogens. These microorganisms combine

hydrogen and carbon dioxide to produce methane and water. Stephenson and Strickland

(1933) were the first to isolate these methanogenic microorganisms, which include the

species Methanobacterium woljéi (Ferry, 1993). As a result of this process, these

bacteria regulate the formation and distribution of VFAs in anaerobic reactors. Thus,
these microorganisms have been d e d "the autopilot of the anaerobic digestion process"

(Mosey, 1983).

One h a i group of microorganisms, the aceticlastic methanogens (which includes

Methanosueta species Ferry, 1993]), degrade the acetate formed in the reactor to C&

and CO2. Sohngen (1910) was the nrst to show that 1 mol of C h and 1 mol of C a were

produced per mol of acetate consumed. Acetate is of particular importance in anaerobic

systems, as it has been shown to form more than 75 % of the methane produced in

anaerobic digesters (Aguilar et al., 1995). For a detailed review of methanogenesis, see

Ferry (1993).

The anaerobic microorganisms which mediate the biodegradation of glucose grow


at slow rates, even under ideal conditions of neutrai pH. If the rate of acid production in

an anaerobic reactor is not matched by the rate of methanogenesis, however, acids may

accumulate and lead to a &op in pH. Methanogenic bacteria, in parîicular, are severely

inhibited by pH levels below 6.5 (Speece, 1996). Thus, if the pH drops due to an

accumulation of acids in the reactor, the rate of methanogenesis will be reduced M e r

leading to even higher accumulations of acids and lower pH levels. As a result, the

operational protocol of anaerobic treatment systems is crucial to their success. In

addition, anaerobic sludge used in a reactor shodd be properly acclirnatized to select the

microorganisms that are best suited to degrade the VFAs produced by the acid formers

(Aguilar et al ., 1995).
2.2 The Anaerobic Scqucncing Batch Rcactor (ANSBR)

The anaerobic sequencing batch reactor (ANSBR)was developed by Dague and

CO-workersat Iowa State University during the 1980s and patented (U.S. Patent Number

5,185,079; Feb. 9, 1993).

An ANSBR consists of a vesse1 which acts as both a reactor and a settiing basin,

thus elirninating the need for separate vessels. The treatment process consists of four

stages: fill, react, settle, and decant. During the fill stage, the wastewater to be treated is

added to the reactor. The contents are typically mixed during the react phase (by either

liquid or gas flow) during which time most of the microbiological degradation of the

organic compounds occurs. h most of the early work perfomed with the ANSBR

(Dague et al., 1992; Sung and Dague, 1992), tilling the reactor witb the influent feed as

quickly as possible was assumed to be the most beneficial strategy because it ensured a

high initial food-to-microorganism ratio, and presumably high initial degradation

rates. This "fast feed" protocol is flawed, however, since the initial substrate

concentration would be so high that zeroth order kinetics with respect to substrate

concentration would quickly be reached. (From Monod kinetics, substrate levels above a

threshold concentration will no longer increase the degradation rate.) In addition, al1

substrates cannot be treated equally as simply chernical oxygen demand (COD) in

anaerobic treatment. Some complex, polymeric substrates may not form acids quickly,

and thus, fast feeding may not produce a pH problem. However, rapid feeding of readily

a c i d i m g substrates, such as glucose, can Iead to large accumulations of VFAs in the

reactor (Bagley and Brodkorb, 1999).


At the end of the react cycle, the concentratioa of biodegradable organics shouid

be at the level des- assuming no process upsets have occurred. Then, during the settle

stage, the biomass is allowed to senle under quiescent conditions to be retained withh the

reactor. During the decant stage, clarified supernatant is withdrawn fiom the reactor, and

the cycle is ready to be repeated again.

The ANSBR has been w d to treat a variety of wastewaters including swine

waste, municipal sludge, non-fat dry mik, sucrose, acetate, rice wastewater, and glucose

at varying organic loading rates (OLRs) (Table 2.3).

Table 2.3 Summary of ANSBR Experimental Work

Substrate OLR (IcgCo~rn~~-d-') Refcrtnce


sucroselacetate 2.5 - 9 Kennedy et al. (1991)
glucose 1.6 Suthaker et al. (199 1)
non-fat dry milk 0.5 5- Dague et al.(1992)
s w i n waste 1 - 5.5 Dague and Pidaparti (1992)
non-fat dry milk 2 - 12 Sung and Dague (1992)
swine waste 5 - 10 Schmit and Dague (1993)
municipal siudge 1-2 Chang et al. (1994)
sucrose 6 - 19 Angenent and Dague (1995)
non-fat dry mik 2 - 12 Sung and Dague ( 1995)
sucrose 3-12 Wirtz and Dague (1996)
non-fat dry milk 0.2 - 2 Ndon and Dague (1 997)
rice wastewater 1.6 - 3.3 Ng et al. (1998)
non-fat dry milk 0.6 - 2.4 Dague et al. (1998)

The treatment of swine wastes was performed at 20,25, and 35°C in 12-L reactors

with a 2-L fil1 volume (Dague and Pidaparti, 1992; Schmit and Dague, 1993). The solids

in the waste were effectively retained by the ANSBRs, although the COD removal rates

were relatively low (approx. 40 to 50 %). Municipal sludge was successfiilly treated at

35°C in 4-L reactors, with a fi11 volume of 1.2 to 1.6 L per cycle (Chang et al., 1994). A
relatively long hydraulic retention time (HRT)of 10 days was used to achieve OLRs of 1

to 2 k g ~ 0 ~ * r n - ' * d - ' .

Non-fat dry milk (NFDM) has also been successfully treated in ANSBRs. OLRs

- ' k e n treated at 3S°C with 12,24, and 48 hr HRTs and a 6 hr


up to 12 k g ~ 0 ~ - r n " - dhave

cycle tirne (Sung and Dague, 1992; Dague et al., 1992; Sung and Dague, 1995). Dilute

NFDM wastewatea (OLRs up to 2.4 k g ~ ~ ~ - m ' 3 -have


d - ' ) k e n treated at temperatures

ranging between 5 and 25°C (Ndon and Dague, 1997; Dague er al., 1998).

The highest organic loading reported to have been treated using an ANSBR was

- d - ' , sucrose as the COD source (Angenent and Dague, 1995).


19 k g ~ ~ ~ * m - 3utiliPiog

The temperature used in that study was 3S°C,and 4 L of feed was added to the 12-L

reactor each cycle, giving an HRT of 12 hrs. Wirtz and Dague (1996) also used sucrose

as the substrate (up to an OLR of 12 kg~O~-rn-'-d-')


during a granulation enhancernent

study. A soluble sucrose/acetate wastewater was used by Kennedy et al. (199 1) to

detemine the effect of varying the fill-to-react time ratio between 0.2 and 2. They found

that treatment efficiency at OLRs above 9 k g ~ 0 ~ = m 3 *was


d - 'reduced at the lower fill-

to-react ratios. It should be noted that the anaerobic biomass seeds used by Angenent and

Dague (1995) and Kennedy et al. (199 1) were granulated prior to beginning the studies.

N g er al. (1998) utilized a two-stage ANSBR (4.5 L each) to treat rice wastewater
and address acid fonnation. The first ANSBR was used to acidify the waste, whiie the

second ANSBR completed the microbiological degradation. HRT values of 1.5 and 3

days were used, with corresponding loadings of 3.3 and 1.6 k g ~ O ~ * r n J - drespectively.
-',

Suthaker et al. (199 1) had trouble using ANSBR technology to treat a hi&-strength

wastewater (35 gCOD/L as glucose) due to acid formation. T ' e u reactor achieved 73 %
soluble COD (SCOD)removal with a cycle t h e of 16 days and a .OLR o f

approximately 1.6 k g ~ 0 ~ - m 3 - d - ' .

2.3 Alkalinity and pH Issues

The pH and alkaliaity of an anaerobic treatment system are important operational

parameters. The alkalinity of an anaerobic treatment system is a rneasure of its ability to

neutralize the acids produced and thus control the pH of the system. In anaerobic

systems, the alkalinity consists primariiy of bicarbonate ion, HCa-. Wer carbonate

species are also present, but bicarbonate ion is predominant at pH levels around

neuîraiity. For a given CO2 partial pressure, the carbonate ailcaiinity and pH of a system

are related by the foilowing equation (Snoeyi.uk and Jenkins, 1980):

KI KHPco2 + 2K,K2KH,P + - -K,,


[H+]
Alk =
b+1 b+P [H+ 1
where:

Pco2= CO2 partial pressure, atm

Kw= 10-14;Ki = 1o".~; Kz= 1 0 - ' ~ ;- ~KH' = 1 0 - l . ~(equilibrium constants at 25OC)

Methanol may produce C a partial pressures of up to 25 %, while a substrate

such as glucose can produce CO2partial pressures of up to 50 % (Speece, 1996). In tbe

absence of significant acid concentrations, equation 2.1 predicts that because of the high

Pco2the alkalinity for a glucose-fed anaerobic system must be greater than 3500 mg/L as

CaC03 to maintain the pH at 6.8 or higher. Anaerobic microorganisms, and methanogens

in particular, are susceptible to inhibition by pH levels that dmp below 6.5 (Speece,

1996).
As mentioned previously, substrates that are consumed anaerobicaiîy produce

various orgauic acids which can lower the pH of the treatment system. Hence, pH levels

in anaerobic reactors must be monitored and controlled. If the system is working

properly, the acids produced should be readily coasumed to form acetate and eventually

C& and CO2. However, artificial pH control, in the form of strong acids or bases, may

need to be added to a reactor system particularly durhg start-up pends when the rate of

acid production may exceed the rate of acid consumption. Operationai parameters can

also be used to control the pH levels in some reactors, such as the ANSBR. For example,

the percentage of the reactor volume decanted each cycle can be varied to remove more

or less of the reactor's contents. As well, the feed c m be introduced more slowly into the

reactor to reduce the rate of acid formation.

To provide the best indication of anaerobic treatment process stability, and due to

their relative ease of measurement, the pH and alkalinity of a reactor should be monitored

closely, along with other parameters such as Hzpartial pressure (Inanc et al., 1996; Guwy

et al., 1997).

2.4 Modelling the Anaerobic Trelrtment Process

Mathematical models are required to predict the performance of complex systems,

and as such, many models for anaerobic treatment have been developed (e.g. Mosey,

1983; Costello et al., 1991 a&b; Wu and Hickey, 1995). The mode1by Mosey (1983)

specifically considered the anaerobic degradation of glucose. The mode1 developed by

Costello et al. (199 1 a&b) included most of the considerations required to properly detine

and mode1 the anaerobic treatment process. The "biological make-up" of the anaerobic

ecosystem, the physicaVchemical system and the reactor process were included, along
with mathematical terms to model product and pH inhibition of each group of

microorganisms. The giucose degradation pathway was altered as compared to Mosey

(1 983) in that lactic acid-consuming bacteria were also included. These bacteria would

either produce acetate or propionate, depending upon the hydrogen partial pressure in the

reactor. Overall, the d e l that was developed represented the anaerobic systems tested

accurately (Costello et al., 1991 b).

Early models for degradation in ANSBRs were developed by Sung and Dague

(1992) and Fernandes et al. (1993). The major drawback of these models was their use of

"buik" kinetic parameters, S and X. In other words, ali substrates were considered the

same and measured as COD (S), and ail biomass was considered the same and measured

as volatile suspended solids (VSS)(A). The subsequent model developed for the ANSBR

(Brodkorb, 1998; Bagley and Brodkorb, 1999) improved on these earlier models. The

model was based upon a systematic approach to modeiling microbiological treatment

processes developed by the International Association on Water Quaiity (TAWQ) (Henze

et al., 1987; Henze et al., 1995). Individual groups of micrmrganisms are modelled

separately, based upon the substrate they consume and the product they produce.

Individual yields and kinetic ami inhibition parameters can be assigned to each trophic

group. As well, because of the use of the iAWQ approach, any desired kinetic

expressions can be used. Thus, the model d o w s for a flexible and organized method of

modelling the ANSBR treatment process.


2.5 Substrate-Specific Kinetic Parameters and Active Biomass

As discussed previously, anaerobic biodegradation is a multi-step pmcess that

requires different groups of microorganisms working sequentiaily to completely degrade

the organics present. The interactions between these various groups are complex and

non-linear, and modelling anaerobic treatment systems can not be done sufnciently

through the use of bulk parameters such as COD and VSS. As weil, although it is

important to know the methanogenic activity of biomass (Ince et al., 1995; Soto et al.,

1993; James et al., 1990) it is also important to know the acid production rate and the

distribution of those acids, since propionic acid can be very slow to degrade (Iaanc et al.,

1996).

Biodegradation kinetics have been modelled in a number of ways, including using

first-order, Monod, and Contois models (Pavlostathis and Giralda-Gomez, 1991). For

whichever model is use4 obtaining diable kinetic parameters is essential to accurately

model and understand the anaerobic treatment process (MerkeI et al., 1996).

Nevertheless, large discrepancies exist between kinetic parameters reported by various

researchers for the same substrate (Pavlostathis and Giraldo-Gomez, 199l), and this may

occur due to differences in biomass sources, reactor operation, temperature, and

analytical methods used. Thus, kinetic parameters for specific substrates must be

measured with respect to the concentration of biomass actively consuming that substrate.

As a result, detennining concentrations of active biomass that degrade specific

compounds is important to accurately model the ANSBR process. Traditionally, bulk

quantities such as VSS have been used to characterize the o v e d l biomass present.

However, these methods do not give an insight into the population diversity and
dynamics existing in the biomass, nor do they allow for the differentiation between active

and dead rnicroorganisms. New methods have k e n developed which allow researchers

to identify specific species of microorganisrns through the use of targeted nucleic acid

probes (e.g. Merkel et al.,1999). These methods are expensive and time-consutning, and

require speciaiized equipment and knowledge of microbiology. As well, probes have not

yet k e n developed which will measure concentrations of active biomass based upon the

specific substrate they consume. These "'îrophic probes" will be much more usef'bl for

modelikg purposes (Merkel et al., 1999). Thus, a straightforward and accurate method is

required to measure both substrate-specific kinetic parameters and active biomass.

2.6 Biomass Granulation

To ensure that hydrogen produced in anaerobic reactors is consumed,

hydrogenotrophic methanogens must be in close physical proximity to hydrogen

producing organisms. For an anaerobic tteatment system, McCarty and Smith (1986)

calcdated that the microorganisms must be within about 10 bacterial widths. Thus,

treatment processes must "encourage" different microbial species to exist very close to

one another, to ensure efficient exchange and transfer of metabolites (McCarty and

Smith, 1986). This is particularly important for high organic loadings, which may lead to

high levels of acids in the reactor and possible system shutdown due to suppressed pH

levels. Granulation is a natural mechanism observed in anaerobic systems (such as the

ANSBR) that provides close contact between the microorganisms.

Initial work on grandar sludge focussed on UASB reactors. These reactors

selected for larger particles which settled more readily, since s d l e r particles would be

Iost through the top of the reactor due to the upfiow velocity of the wastewater. In
general, granular sludge consists of roughly spherical grains, approximately 1 to 3 mm in

diameter. Granules are fomed fiom sludge containing microorganisrns which wete

initial1y unattached. Typically, granules have densities ranging nom 1.O to 1.1 kg/m3

(HulshoE Pol et al., 1986). Tay and Yan (1997) and Yan and Tay (1 997) give a

description of the granulation process during start-up of a UASB reactor. Ia the first, or

acclimation, stage the granules are initiated. During this stage, extracellular polymers

(ECPs) or other attachment mechanisms cause the bacteria to attach to each other

(Wiegant, 1987). During the second stage, known as granulation, the granule mean

diameters increase at a maximum rate of approximately 0.1 mm-mm-'-6'. In the h a 1

stage, maturation, the granules grow at rates 20 % or below their maximum rate (Tay and

Yan, 1997; Yan and Tay, 1997). This declining rate of growth is likely due to limits

irnposed by mass transfer of metabolites into and out of the grande. Tay and Yan (1997)

and Yan and Tay (1997) suggest that the sludge loading rate to a reactor should be kept to

approximately 60 to 80 % of the specific methanogenic activity of the sludge to achieve

optimal granulation. Presumably, this allows for sufficient and balanced growth of the

microorganisms in the sludge, leading to the formation of strong and active granules.

Electron microscopy is typically used to ana1yze the microbiological structure of

granules and this can be very expensive and requin highiy trained staff to be performed

properly. In addition, diflerentiating between flocs and granules c m be difficult, and is

usually resolved by analyzing the particles with the naked eye and making an educated

guess. O h , the deterxnining factor has been mean diameter size, which is arbitrarily

set.
Granules have a layered structure, with different m i ~ ~ ~ ~ r g a n occupying
isrns the

various iayers. N o d y , granules contain signincant numbers of Methanosaeta-like

methanogenic bacteria (Quarmby and Forster, 1995; Forster and Quarmby, 1995;

MacLeod et al.,1990). Most of the micmrganisrns in granules have been detennined to

be Gram-negative, with areas containing carbohydrates decreasing towards the middle of

the granules (Quarmby and Forster, 1995; Forster and Quannby, 1995). The layered

structure of grandes is u s d l y modelIed as a core of methanogenic bacteria (normaliy

Methanosaeta) surrounded by an inner shell of acetogens and hydrogen-conswning

microorganisms, with a final outer shell of acidogenic bacteria (MacLeod et al-,1990).

T'us,glucose would first come in contact with the acidogenic bacteria on the outside of

the granule, fonning VFAs. These would then diffuse into the inner shell of acetogenic

bacteria which would convert lactate, propionate, and butyrate to acetate, CO2and

hydrogen. To allow these reactions to proceed, the hydrogen consumers in the imer shell

would need to continually consume the hydrogen produced. Because of the close

proximity of the bacteria in the granule, hydrogen transfer would be possible even at very

hi& loading rates. Finally, the acetate wouid diffiise to the core of the granule where it

would be converted to C a and C a . Subsequently, these biogases would diffuse back

through the grande to enter the bulk liquid. If the granules become too large, mass

transfer may become less efficient. As a result, insufficient food may reach the bacteria

within the grande, or the biogases accumulating in the granule may cause the granule to

disintegrate, leading to washouî of solids fiom the reactor (Jawed and Tare, 1998).

The substrate fed to a reactor system can influence the dominant species present
in granules. For example, grandes h m a UASB reactor treating potato-processing
wastewater contained predominantly Methanosueta-like species, while granules fiom a

reâztor treating brewery wastewater contained mostly Methanosaeta soehngenii-like

acetate utilizers (Wu et al-,1993 a&b). El-Mamouni, et al.(1 997) found that granulation

proceeded rapidly when Methanosaeta and Meîhanosarcina were used as the starîing

microbial nucleus type, while granulation was slowed signifïcantly when acidogenic flocs

were used as precursors. Increasing the influent substrate concentration has been found

to increase the median granule diameter (Grotenhuis et al., 1991). Work by Thaveersi et

al. (1995) showed that the surface tension of the environment surrounding granules c m

impact their structure. Low surface tensions (< 50 W m ) led to granules with

methanogenic cores surrounded by acetogens, with an outer shell consisting of

hydrophilic acidogens. High surface tensions (> 55 mN/m) led to granules with

methanogenic cores surrounded by acetogens, with an outer shell consisting of a

conglomerate of cells, with an overall hydrophobic surface. Intermediate surface

tensions led to difficult formation of granules. Work on granules fkom ANSBRs showed

similar structures to UASB granules, with predominantly Methanosueta-like bacteria

( B a d et al., 1997).

Mineral analyses on granules by Dubourguier et ai. (1987) showed that granules

contained a large amount of sulphide precipitates of iron and nickel. In addition, calcium

precipitates represented 30 to 40 % of the minerais in the granules. Mahoney et al-

( 1987) concluded that the addition of 100 mg/L of calcium to a reactor led to larger-shed

granules that settled 3 to 4 times faster than the control reactor's granules. Calcium has

been postulated to be an important factor in the formation and maintenance of strong

granular particles.
Granular sludge has also demonstrated other properties that make it s u p i o r to

non-granulated sludge. Kato et al. (1993) found that granular methanogenic sludge had a

high tolerance for oxygen in wastewater. Shen and Guiot (1996) found that grandar

sludge maintained its performance even after 3 months of exposure to oxygen. Wu et al.

(1995) reported that granules stored at 4°C for 18 months retained thei. metabolic

activities with respect to acetate, propionate, and butyrate degradation. It is possible that

this "resistance" to oxygen and low temperatures is due to *ion limitations that

protect the core of the granules.

Since granular sludge is essential to enswe that a treatment process can

successfüily treat high organic loadings, many researchers have done experiments dealing

with enhancement of granulation. Imai et al. (1997) reported that granulation was

enhanced in UASB reactors through the addition of water absorbing polymer. In ANSBR

reactors, Wirtz and Dague (1996) pedonned granulation enhancement studies utiiieog

powdered and granulated activated carbon, silica sanâ, gamet, femc chloride, and

cationic and anionic polymers. The cationic polymer was found to be the rnost effective

granulation enhancement aici, reducing the time required for granuiation by

approximately 75 % as compared to the control reactor (with no enhancement added).


3. MATERIALS AND METHODS

3.1 Lab-ScaJe ANSBR Sct-Up

The laboratory-scale ANSBR used was a cyiindrical plexiglass reactor

manufactured at the University of Toronto with an intemal diameter of approxirnately 15

cm and a height of 70 cm. The working liquid volume of the reactor was 12 L, with an

additional 1 L for headspace. There were 15 sampling ports (valves) evenly spread dong

the height of the reactor, aiiowing flexibility for s a m p h g and operating volume of the

reactor. The reactor was operated at an ambient lab temperature of 2 1 I1°C.

The reactor set-up aad ancillary components are shown in Figure 3.1.

1 Gas ) Ga"
Meter Effluent
I
I
Water
Sed

C
l

Recycle
Pump 2

ANSBR

Figure 3.1 Lab-Sale Reactor Set-Up


A pipermperistaitic pump was used as the i d u e n t purnp, ~ a s t e r f l e peristaltic
x~

pumps (LABCOR, Anjou, QC) were used as the effluent pump and recycle pump 1, and a

centrifbgal pump (LABCOR, Anjou, QC) was used as recycle pump 2


Littie Gad'

(Figure 3.1). The larger capacity Little Gantmcentrifbgal pump was added on &y 130 to

provide improved mixing. Liquid recycle was used throughout the experimental work.

For automated operation, plug-in timers were used which automatically tumed the pumps

on and off at prescnbed times (Canlrrlswide Scientific, Ottawa, ON). A 40-L feed tank

was used to store the innuent at 2 1°C for multiple automatic feedings. Typically, the

feed solution was prepared at least once every two days to minimize bacterial growth. To

ensure that the feed was weil-mixed, a small recirculating pump (normally used in

aquariums) was used to suspend any solids present in the tank.

For gas collection and volume measurement, a wet test gas meter was used in

combination with a water seal and gas bag. The gas bag (volume = 8 L) ensured that

there was gas available to fiîi the liquid volume decanted Erom the reactor (thus avoiding

creating a "vacuum in the reactor), while the water seal ensured that the gas bag was

refitled at the beginning of each cycle by providing extra tesistance to the gas flow. To

achieve this, only 3 to 5 cm of water was required for the water seal.

The reactor was initially seeded with anaerobic sludge obtained fiom the Toronto

Main Sewage Treatment Plant. The initial total suspended solids (TSS) in the reactor

was diluted to 6000 mg&, based upon settling tests performed on the sludge. The reactor

was re-seeded on day 101 with granulated sludge (Champlain Industries, Cornwall, ON)

fiom storage in the laboratory, and on &y 146 it was te-seeded again with a mix of

Toronto Main anaerobic sludge and granulated sludge.


3.2 Lab-Scale Reactor Operation

One objective of this research work was to maximize the organic ioading treatable
by a labscale ANSBR and to ident* obstacles to achieving higher loadings. Iaitiaiiy,

the glucose concentration of the infiuent feed was arbitrarily set to a value of 1 g/L. At

the end of each cycle, 6 L of the reactor's contents were decanteà, which gave an HRT of

simply twice the total cycle t h e . This decant strategy was used thtoughout the

experimental work In addition, no extemal pH control was used with the ANSBR

From &y 1 h u g h &y 60, the reactor was run with an 8 hr cycle, with a 5 hr

feed stage, 0.5 hr settle stage, and 0.5 hr decaut stage. From days 60 to 107, the cycle

time was increased to 2 or 3 days (varied throughout), with the settle and decant times

maintained at 0.5 hrs. The cycle t h e was increased to ailow for manual operation of the

pumps (since there were troubles with automated operation) and to increase the length of

the react phase (due to poor reactor performance). From day 108 to day 320 (end of the

experimental nui) the reactor was run on either a 1 or 2 day cycle (excluding holidays).

The influent glucose concentration was kept at 1 g/L fiom &y 1 through to &y

124, when it was increased to 2 g/L. On day 177 the feed concentration was increased to

4 g/L, which gave an OLR of 2 kg*rn-'-6' for 1 day cycles. On &y 200, the glucose

concentration of the feed was increased to 6 g/L,giving an OLR of 3 kg-m"-6' for 1 day

cycles. Nine days after the OLR = 3 gWd experiments were completed (on &y 273) the

influent concentration was dropped down to 2 g/L, simply to keep the organisms viable.

On day 290, the glucose concentration was increased to 4 g/L, and was maintained at that

fevel until the reactor was shut down on day 320. The teasons for these changes in

operation are explained in more detail in Chapter 4 (initial Operation).


Measurernents taken during the experimental nin and their average frequency are

given in Table 3.1, although the actual sampling fkquency at any given tirne varied.

Table 3.1 Lab-Scale Measurernents Taken and Average Frequency

Mtasunment Average Fnqutncy


PH twice per week
allcalinity once every two weeks
soluble COD (SCOD) once every three days
biogas produced (volume) once per day
VFA concentrations once per week

3.3 Lab-Scale Basal Medium

From the beginning of the experiments until day 25, the basai medium descnbed

by Brodkorb (1998) was used (Table 3.2).

Table 3.2 Initial Basal Medium Composition (Based on Brodkorb, 1998)


(Day O to Day 25)
Nutrient Overail Concentration Trace Meta1 Solution (TMS)
(mglg glucose) Concentration (rnR/L) 1
NaHC03 6000 mgA/ N/A~
&HP04 NIA
NHsCl N/A
EDTA NIA
Yeast extract N/A
MgCl2 - 4H20 NIA
CaC12 2H20 N/A
KCl NIA
Na2Se03 100
-
NiCl2 6H20 50
-
CoC12 6H2O 150
W)6M07@4 - 4WO 90
CuC12 . 2H20 30
-
MnC12 4H20 500
ZnC12 50
FeC12 - 4H20 2000
Na2S04 375
H3BO3 0.05 50
1
Volume of TMS required = 1 mL per gram of glucose per liter of feed
'Independent of glucose concentration
%/A = not applicable (i.e. compound not in TMS)
From &y 26 mtii the end of the experimental run (day 320), a modified fwd

solution was used (Table 3.3) based upon the recipe used by Fang and Chui (1993). (The

feed in Table 3.2 was used until this modified feed solution was fïnaliy developed.)

Their work with laboratory-scale UASBs involveci very high loadings (up to 100

- ' )it, was felt that a simiiar recipe would also help the ANSBR achieve
k g ~ ~ ~ - m ' 3 - dand

loadings greater than 19 k g ~ ~ ~ - m - 3reported


* d " by Angenent and Dague (1995). The

carbonate aikalinity added to the basal medium was selected to provide a pH of 6.8 for a

COz partial pressure of 50 % in the absence of significant acid accumulation (Section 2.3

and Equation 2.1).

Table 3.3 Modified Basal Medium Composition (Based on Fang and Chui, 1993)
(Day 26 to Day 320)
Nutrient Overail Concentration Trace Meta1 Solution (TMS)
(mdg glucose) Concentration ( m m )1 -
NaHC03 5600 m& N/A~
CaC03 250 mg@ NIA
MgClt . 6Hz0 83.66 NIA
NWcl 78.80 NIA
KCI 38.14 NIA
K2m04 28.15 NIA
(WhSO4 20.60 NIA
-
NiC12 6H20 12.50 12 500
Yeast extract 10.00 NIA
-
FeC12 4H20 8.90 8900
-
MKl2 4 H 2 0 1.79 1790
CoC12 - 6H20 1.25 1250
ZnC12 1.O4 1040
@-b)d%Ot4 4Hzo 0.71 710
CuClz 2H20 0.56 560
Na2Se03 0.1 1 110
H3B03 0.05 50
'volume of TMS required = 1 mL per gram of glucose per fiter of feed
%dependent of glucose concentration
%/A = not applicable (Le. compound not in TMS)
3.4 Measurement of Volatile Fatty Acids ( W h )

A DX 500 Ion Chromatograph (IC) (Dionex, Oakville, ON) equipped with a


CD20 conductivity detector and a GP40 gradient pump was used to measure volatile fatty

acids V A S ) based on a method developed by other researchers in the laboratory. To

provide adequate separation of the VFAs, au AS 1 1 analytical column and an AG1 1 guard

column were used in conjmction with an AMMS-II anion micromembrane suppressor

(Dionex, Oakville, ON). A 25 pL sample loop was used to deliver consistent sample

volumes to the K.

Three eluents were required for the VFA measurements. Eluent A, ~ i l l i - ~ @

water (Millipore, Nepean, ON), was distilled and deionized to 18.2 Mn. Eluent B was a

5 mM NaOH solution, and eluent C was a 50 m M NaOH solution. The regenerant used

was a 13.14 mM solution of H2S04. The regenerant and eluents were prepared with

~illi-Q@
water.

The eluent gradients used are shown in Figure 3.2. The flow rate was 2.0 W m i n

throughout the run, and injection of the sample ont0 the column occwred between O to

0.20 min. The regenerant flow was constant throughout the run at 10 W m i n .
O 3 6 9 12 15 18
tirne (min)
Figure 3-2 IC Eluent Profiles

Lactate, acetate, propionate, and n-butyrate were detected at approximately 2.2,

2.4,2.8, and 3.4 min,respectively. Fluoride ion eluted at approxhately the same time as

lactate, but this interference codd be removed by subtracting offthe peak area obtained

from a tap water blank, since the tap water used for the feed solution was the only source

of fluoride ion. The detection limit, based on the lowest standard concentration

measured, was 0.5 mg/L for each VFA.

Sample aliquots (40 pL for bottle tests and 1 mL for lab-scale reactor tests) were

diluted so that the concentration of individual VFA components lay between 0.5 to 5.0

mg/L (detection range for the instrument). A minimum of 3 mL of diluted sample was

filtered through an O n - ~ u a r dH-carüidge


~ (Dionex, Oakvüle, ON) to remove any

suspended solids. The first mL was discarded and the n e a 2 mL (minimum)was placed

into a 5 mL polypropylene IC viai. Prior to use, the H-cartridge was rinxd with 10 to 20

mL of Milli-~@
water.
3.5 Measunment of Hydrogen

A Hewlett Packard 5890 Series II Plus Gas Chromatograph (GC) (Mississauga,

ON) equipped with a thermal conductivity detector (TCD)was used for hydrogen

analyses, with direct injection of 20 to 25 pL gas samples into the GC (based on a

method developed by other researchers in the laboratory). The inlet and detector

temperatures were 200 and 250°C, respectively, and the oven temperature was set to a

constant 60°C. A fused silica capillaty wlumn (Carboxen Plot, 30 m x 0.53 mm,

Supelco, Oakville, ON) separated the gases present in the sample, and nitrogen was used

as the carrier gas at 5 W m i n to measure the hydrogen in the headspace. The detection

limit for this method, based on the lowest standard concentration measured was

6.2 x 10-~atm.

3.6 Measurement of Glucose

Glucose was measured spectrophotometrically using a kit fiom Sigma-Aldrich,

Ltd. (Oakville, ON) originally developed for glucose measurements of bodily fluids

(Kunst et al., 1983). The kit contained 0.75 mm01 of NAD, 0.5 mm01 of ATP, 500 uni&

of hexokhase enzyme, 500 units of glucose-6-phosphate dehydrogenase, and 1.05 mm01

~ ' per bottle. In addition, the kit aiso contained a bufEer, non-reactive
of M ~ ions

stabilizers, fillers, and 0.25 g of d u m azide as a preservative. The reagent was

prepared by adding 500 niL of distilled water to the dehydrated reagents.

Glucose reacts with ATP in the reagent to form glucose-6-phosphate and ADP in

the presence of hexokinase enzyme. Next, glucose-&phosphate is oxidized by NAD in

the presence of glucose-6-phosphate dehydrogenase to form 6-phosphogluconate and

NADH. The NADH produced can be detected specmphotometncally at 340 MI. For
lab-scale reactor samples, 5 mL aliquots were filtered through 0.45 pm ~crodisc' syrïnge

membrane filters (VWR Canlab, Toronto, ON) to remove suspended solids pnor to

analysis. Bottle study samples were not pre-fïitered due to the lack of s d c i e n t sample

volume for filtration. 50 pL samples were then added to 5 mL of reagent for glucose

analyses. The detection limit of the glucose test based on the lowest standard

concentration measured, as used, was 50 mgK. However, glucose levels as Iow as 1

mgL codd be detected by modifjing this procedure. First?250 mi,of distilled water is

added to the dehydrated reagents, giving a solution that is twice as concentrated as

normal. Then, 2.5 mL of sample (50 times more than normal) is added to 2.5 mL of the

concentrated reagent. This modification was not necessary for this work, since the

glucose concentrations of interest were > 100 mg& but could be usefûl for other

researchers.

3.7 Bottle Study Methods

To determine substrate-specific kinetic parameters and active biomass Eractions,

individual serum bottles were run that were fed a single substrate. Analytical methods

used to measure these substrates were described in Sections 3.4 to 3.6. For these

experiments, 20-mL bottles were used capped with butyl-rubber septa and alumiuum

crimp caps. The bottles were each filled with 15 mL of anaerobic biomass fiom a 4-L

reactor seeded with a mix of anaerobic sludge fiom the North Toronto Treatment Plant

and granulated sludge (Champlain Industries, Cornwall, ON) fiom storage in our

laboratory. The 20-mL bottles were then capped under a Hz/Nz/C@ atmosphere in an

anaerobic glove box. The s e m bottles were mixed by placing them on their sides in an

orbital shaker set to 200 rpm. Sampling times varied for each substrate, and were
detemillied firom prior expenence, fiom once every 15 minutes (e.g. glucose) to once

every day (e.g. propionate). The glucose and VFA bottles were centrifiiged for 1 minute

prior to sampling to separate-out the majonw of the suspended rolids, and 50 p i aliquots

were removed for glucose anaiyses, and 40 pL aîiquots for VFA analyses. The hydrogen

bonles were not centrifiiged, and 25 C<L headspace samples were removed for analyses.

3.8 Measurement of Alkdinity, pH, Soluble COD and Suspendcd Solids

Alkalinity, pH, soluble COD, and suspended solids were measured according to

rnethods outlined in Standard Methoak (APHA,AWWA, and WEF, 1998).


4. INITLAL OPERATION

4.1 Summary of Lab-Scale Operation

The labofatory-scale ANSBR was operated in thtee phases: Initial set-up; first

addition of granulated sludge; and final configuration(Table 4.1). A summary of the

influent and effluent SCOD values measured for &y O through &y 227 (before the

experiments descnbed in Chapter 5 were carried out) is given in Figure 4.1. In the

following sections, the three phases of operation will be described in greater detail, with

closer examination of effluent results.

Table 4.1 Sununary of Phases of Lab-Scale ANSBR Operation


--

Phase Days Opented Recycle Pump(s) Biomass Used


Used (Figure 3.1)
1. Initial Set-Up O to 100 Recycle Pump 1 Non-grandated

2. First Addition of 101 to 131 Recycle Pump 1 Granulated and


Granulated Sludge Non-granulated

3. Final Configuration 132 to 320 Recycle h p s Granulated and


1 and 2 Non-aranuiated
4.2 Phase 1- Initial Set-Up

The laboratory-scale ANSBR was operated without extemal pH control to

sirnplify its operation. In addition, a more robust reactor and microbial community

would be developed if extemal pH control was avoided. During the initial set-up phase,

fkom day 1 through &y 100, the laboratory-scde ANSBR was operated with the

following protocol to rapidly increase the loading rate. Half of the reactor contents (6 L)

were decanted at the end of each cycle, providing a hydrauiic retention tirne (HRT)equal

to twice the total cycle tirne. The influent concentration of the feed was initially

arbitrarily set to 1 g of giucose per Liter (1067 mg COD) to provide a low initial substrate

concentration. The reactor was run with an 8 hour cycie time and 5 hour feed stage to

coincide with an 8 hour work &y. The organic loading rate (OLR)to the reactor can

simply be calculated as the influent concentration divideci by the HRT. With an HRT of

16 hours, this led to an OLR of approximately 1.6 k g ~ ~ ~ - m - 3 -The


d - 'results
. fiom days

O to 35 for the ANSBR (with a total suspended solids (TSS) concentration of 6000 mg/L)

showed SCOD removais of < 40 % (Figure 4.2). The large variations in effluent SCOD

in Figure 4.2 are also indicative of the reactor's poor performance. However, the total

suspended solids in the efnuent of the reactor were consistently at, or below, 100 mg/L.
innn

200 - M u e n t SCOD :

O 5 10 15 20 25 30 35
time (days)

Figure 4.2 Initial SCOD Removal

Andysis of VFA concentrations in the reactor effluent (Figure 4.3) showed

relatively low concentrations of < 150 mg/L of lactate (as was expected due to tbe slow

feed) and fluctuating concentrations of acetate, propionate, and butyrate acid (between 50

to 500 m a ) .

15 20 25 30 35
time (days)

Figure 4.3 Initial VFA Results


It was also found chat very few solids were king recycled by the initial recycle

pump used (Recycle pump 1 in Figure 3.1) as is shown by the solids profile of the reactor

during the react phase on day 28 (Table 4.2).

Table 4.2 Solids Profile for the ANSBR on Day 28

Height (cm) TSS (mg&) VSS (mpr/L)


58.5 20 45

This analysis shows that almost 90 % of the biosolids existed at, or near, the

bottom of the reactor even during the recycle stage. This situation did not ailow d l of the

microorganisms to corne in contact with the glucose feed, thus partly causing the poor

treatment efficiency (< 40 % SCOD removal) of the reactor. In addition, it was

hypothesized that the anaerobic sludge obtained fiom the Toronto Main Treatment Plant

contained a significant amount of inert solids and dead or inactive microorganisms. This

may have occurred because the temperature difference between the treatment plant

digester (approx. 35OC) and the laboratory (2 I°C) '%hocked" the microorganisms. Also,

there may have been a community imbalance in the anaerobic biomass, since the digester

was not designed to run exclusively on glucose.

Due to the poor SCOD removal of the reactor, the feed solution was modified

(Table 3.3) on day 26 to provide more trace metals (especidly nickel and cobalt). The

trace metal concentrations were increased to prevent any deficiencies which may have

impaired the reactor's performance (Speece, 1996). In addition, the cycle tirne was
increased fiom 8 hrs to 2 - 3 &YS (varied) on day 70 to allow the microorganisms more

time to degrade the glucose and VFAs. The SCOD concentrations in the effluent fiom

day 71 to day 100 (Figure 4.4) were simiif?cantiy lower (> 80 % removal) due to the

increased cycle time and lowered OLR of 0.25 kg-rn"*d-l (effluent SCOD results for days

35 to 70 were not obtained).

i i n f l u e n t SCOD /

80 90
time (days)

Figure 4.4 Effluent SCOD From Days 7 1 to 100

4.3 Phase 2 - First Addition of Granulateci Sludge to the Reactor


On day 101, 1 L of granulated sludge fiom another ANSBR king ru.in our

laboratory was added to the ANSBR to attempt to "kick-start" the reactor, since

granulated sludge is requirrd to mat very high loadings (e.g. LOO k g C ~ ~ - r n *reported
~-d-~

by Fang and Chui, 1993). On day 109, the cycle tirne was reduced to 1 day to increase

the loading (since reactor performance was irnproving Figure 4.41). The length of the

feed stage was kept at 5 hours, and the influent glucose concentration was stiil 1 g/L.

Emuent SCOD concentrations fiom days 109 to 118 decreased h m 3 15 to 165 mglL (70

to 80 % removals) (Figure 4.9, and biogas production was 0.9 to 1.2 L each day.
E 600 - - ifluent SCOD I

Figure 4.5 Effluent SCOD Mer First Addition of Grandated Sludge

When the gas production increased to 1.5 L per day d e r &y 121, the glucose

concentration of the innuent was increased to 2 g/L with a cycle tirne of 1 &y. The

effluent SCOD concentration fiom day 124 to day 130 remained steady between 380 to

460 mg& resulting in an SCOD removal efficiency of 80 %. Biogas production

increased to 2.5 L per day.

On day 130, a new Little Gant@centrifuga1 recycle pump (LABCOR, Anjou,

QC) (Recycle pump 2 on Figure 3.1) was comected to provide improved mixing of the

reactor's biomass. The pump's flow rate was reduced to approximately 5 L h i n to

provide an upflow velocity of 15 mlhr which was 10 times greater than the rate provided

by the previous pump, and was considered more than adequate to provide efficient

mixing of the solids. M e r attachiiig the recycle pump and d g it ovemight, the

pump overheated and increased the temperature in the reactor to over 50°C, thus killing

the microorganisms present in the reactor.


4.4 Phase 3 - Final Configuration
Subsequently, the centrifugai recycle pump (Recycle pump 2 in Figure 3.1) was

placed in a water bath to prevent it fiom overheating. Six iiters of sludge were removed

£iom the reactor, and 1.5 L of Toronto Main Treatment Plant anaerobic sludge, 1.5 L of

granulated sludge fiom storage in our laboratory, and 3 L of basal medium (Table 3.3)

were added. From day 146 to &y 163 (over the Christmas holidays), the reactor was fed

once every 2 days with a 5 hr feed and 2 g/L influent glucose concentration. On Aay 171,

a new recycle strategy was developed to use the two recycle pumps efficiently. The

smaller peristaltic pump (Recycle pump 1 in Figure 3.1) was run throughout the cycle

(except for the settle phase) while the larger centrifugai pump was run intermittently for 2

minutes every 2 to 3 hours. Although the centrifùgal pump gave improved mixing of the

biosolids, and helped distribute them throughout the reactor, it did not allow for

granulation of the biomass. For instance, granules that were present in the seed sludge

were quickly crushed by the pump the f h t t h e it was used. Thus, any granules that

began to f o m in the reactor would aiso have beea crushed by the centrifuga1 p m p .

Nevertheless (since the OLRs used were still low enough to be treated with non-

granulated sludge), fiom day 178 to day 181 the glucose concentration of the infiuent was

increased to 4 gR,with a cycle time of 24 hours and a 5 hour feed (OLR = 2 kgmf3*d-').

The effluent SCOD and propionate concentrations accumulated over these 4 &YS (Figure

4.6).
179 180
time (days)

Figure 4.6 Effluent SCOD and VFAs Fmm Days 178 to 18 1

On &y 185, a test was conducted to determine the effect of increasing the length
of the fil1 time while maintainhg the 24 hr total cycle t h e and 4 g/Linfluent glucose

concentration (and hence, the same OLR). The results are given in Figure 4.7.

600 - Day 185

O 2 4 6 8
time (hrs)
Figure 4.7 Glucose Concentration vs. Time for 5 hr and 10 hr Fil1 Times on Day 185
The resuits in Figure 4.7 show that by increasing the fi11 t h e fiom 5 to 10 hrs, the

glucose concentration in the reactor was lowered by at l e s t 50 %. As a result, the rate of

acid production in the reactor was lower as weli, leading to improved reactor

performance. Effluent SCOD and propionate concentrations for days 185 to 188 with a

10 hr feed (Figure 4.8) showed improved results as compared to the 5 hr feed (Figure

O propionate .

185 186 187 188


time (days)
Figure 4.8. Effluent SCOD and Propionate fiom Days 185 to 188

On day 200, since the reactor was performing weU, the feed concentration and

OLR were increased to 6 g/L and 3 kgm*361,respectively . Effluent pH, SCOD and
VFA concentrations for days 206 and 207 (Table 4.3) show significant accumulations of

propionate to > LOO0 mgR. and acetate to > 400 m a . As well the effluent pH dropped

to nearly 6.4.

Table 4.3 pH and Effluent SCOD and VFAs for Days 206 and 207

Dav DH SCOD' Lactate' ce ta te' Pro~ionate' ~ u t v r a t e '

207 6.41 1345 63 443 1020 132


'AU concentrations are mgCOD/L
From &y 208 to &y 225, the reactor was maintained by feeding it 6 g& glucose

once every 3 days (OLR = 1 kgm-)dl). The pH of the reactor increased to between 6.8

and 6.9, and the effluent SCOD dropped to appmximately 400 mg/L. On day 226, 1 0 mL

of infiuent feed was titrated to determine if its buffering capacity was suffkient (Figure

4.9). The next day, 10 mL of effluent fiom the reactor was titrated to detennine its

bufTering characteristics (Figure 4.1 0).

Figure 4.9 Muent Feed Titration Curve

Figure 4.10 Reactor Effluent Titration Curve


The influent feed titration cuve (Figure 4.9) shows that of the 3800 mg&

aikalinity (as CaC03), 1600 mg/L was actually available for buffering above a pH of 6.5.

The other 2200 m g 4 of alkalinity was available h m pH 6.5 to 4.5, wbich is t w low to

be usefùl to anaerobic rnicroorganisms. Of the 3 150 mgL of alkalinity in the effluent,

only 650 mg/L was actuaily available in the reactor down to a pH of 6.5 (Figure 4. IO),

which means that a small additional accumulation of acids could have caused the reactor

pH to drop to inhibitory levels of below 6.5.

Thus, low pH levels a d accumulation of acids (particularly propionic and acetic

acids) prevented the ANSBR fiom treating OLRs of 3 ~ g m - ~ a dor- 'greater. As a result,

the objective of this research work changed fiom maximizing the organic loading

treatable by ANSBRs to i d e n m g ways to ameliorate the pH and acid accumulation

problems. From day 235 to day 273 experiments were conducted to determine if these

problems could be overcome by modifjing ANSBR operational parameters (Chapter 5).


5. MODIFYING OPERATIONAL PARAMETERS

5.1 Experimental Plan

There are at least two methods by which the performance of an ANSBR with

respect to pH and acid accumulation at low OLRs could be improved: external pH

controi and modifications of operational parameters such as the nIl-to-cycie time (FE)

ratio (Kennedy et al., 1991; Brodkorb, 1998). If external pH control was used (through

the addition of a strong base), the reactor performance at an OLR of 3 kg-m-3-d*'may be

irnproved. It was not pursued for this work, however, because at OLRs »3 kg.m-3.d-',

the base added would have caused large increases in the pH after the VFAs were

consumed. The high pH levels could have led to inhibition problems in their own right.

Muent glucose concentration and fill-to-cycle t h e (FK) ratios were chosen to

be the operational parameters that would be modified. it was hypothesized that

decreasing the Muent glucose concentration fiom 6 glL to 3 g/L (while also decreasing

the cycle time to 12 hours to maintain the OLR at 3 kgm-3d-1)would reduce acid

accumulations because a lower mass of glucose was added each cycle. It was also

hypothesized that increasing the F/C ratio for a given cycle time would reduce the rate of

acid formation and increase the pH of the reactor by "spreading out" the addition of

glucose (Figures 4.7 and 4.8).

To test these hypotheses, the laboratory-scale ANSBR was operated at a solids

concentration of approximately 16,000 mgVSS/L with different infiuent concentrations,

feeding times and total cycle tirnes while maintainhg a constant OLR of 3 kg-m-'b' .

Each of the five runs (Table 5.1) was conducted until the effluent pH was s 6.5

(signimg severe pH inhibition) or 5 days had elapsed (arbitrarily chosen), whichever


came first. For aU 5 nms, the settie and decant times were held constant at 90 min and 30

min, respectively. Each nia kgan d e r the reactor had k e n starved for 2 or 3 days in

order to reduce the levels of VFAs initially in the reactor. Runs 1 and 3 ended afler one

cycle each at which time their effluent pHs were less than 6.5. Run 2 lasted for 2 cycles,

Run 4 lasted for 3 cycles, and Run 5 lasted for 5 cycles before their effluent pHs

decreased below 6.5.

Table 5.1 Experimental Parameters

Run Cyck time (àrs) Feeà time (hm) F/C Glucose (mg&) OLR (kg-m-3d'1)
1 24 6 0.25 6000 3

5.2 Soluble COD vs. Time

Measured soluble COD (SCOD)values and the sum of individual components

(glucose plus VFAs) for Run 1 showed good agreement (Figure 5.1). The results of

SCOD vs. tirne for the other 4 nuis are shown in Figure 5.2.

React

10 15 20 25
time (hm)

Figure 5.1 SCOD and S w n of Individuai Components vs. Time for Run 1
O 10 20 30 40 50 60
time (hm)
Figure 5.2 SCOD vs. T h e for Runs 2 Through 5

Although al1 n m s showed an increase in SCOD values during the feeding stage,

some degradation was occwring, since SCOD concentrations would have reached > 3000

mg/L (Runs 1 and 2) or > 1500 mgK (Runs 3,4, and 5) otherwise. SCOD values

deciined approximately 10 to 20 % for the remainder of the cycle time after feeding. The

SCOD removal efficiencies (Table 5.2) showed little dserence between the m s after

one cycle of operation (Le. within 4 %). The SCOD removals at the end of one cycle for

al1 the runs were relatively high, e.g. > 79%. However, for those runs continuing for

more than one cycle, VFAs accurnulated, lowering SCOD removal efficiencies.

Table 5.2 SCOD Removal Efficiencies

SCOD removal SCOD removal at the


Run afterlcycle end of the rua
1 83 % 83 %
2 83 % 78 %
3 80 % 80 %
4 79 % 76 %
5 82 % 63 %
The distribution of products in the reactor for Run 1 (Figure 5.3) shows that the

glucose concentration increased to 1300 mg/L within 3 hours, but was completely

consumed by the end of the feed stage at hour 6. Lactate was detected during the f e d

stage (< 200 m&), but was consumed within 7 hows. Less than 100 mg& of butyrate

was measured towards the end of the cycle. Propionate and acetate increased throughout

the feed stage to 800 mg&, and 400 mg/L, respectively, but were only slightly degraded

(< 25 % removal) by the end of the cycle, with propionate forming the majority of the

SCOD present in the effluent. Similar results were obtained in al1 mm, and glucose,

lactate, propionate, and acetate levels for each case will be compared in section 5.4.

1400 -
+Glucose +Lactate 'Acetate
1200 - +Propimate +Butyrate
- 1000 -

O 5 10 15 20 25
tirne (hrs)

Figure 5.3. Distribution of Products vs. Time for Run 1

5.3 pH vs. Time

During the first hour of each test, the reactor pH increased sligbtly (0.1 to 0.2 pH

units) as a result of the alkalinity being intnxluced into the reactor (Figure 5.4). The pH

then decreased while VFAs were produced during the fill, and then increased d e r the fil1

was completed as VFAs were consumed. The discontinuities in subsequent figures for
Runs 4 and 5 were intervaIs during which time no experimental data was obtaiaed (due to

manpower limitations).

O 10 20 30 40 50 60
time (hm)
Figure 5.4 pH vs. Time for AU Runs

The depressed pH values encountered in al1 5 m s were due to the elevated levels

of acids (particularly propionate and acetate) in the reactor. Figure 5.5 shows the

relationship between pH and total acids for Run 1 (pH< 6.5 for acids > 1000 mg1L).

: +Total acids :

O 5 10 15 20 25
time (hrs)
Figure 5.5 Relationship Between pH and Total Acids for Run 1

44
The pH decreased to < 6.5 in 24 hours for Run 1, while the pH for Run 2 after 24
hours was > 6.7 (Figure 5.4) since the same mass of glucose was delivered faster to the

reactor for Run 1, aiiowing for faster acid production. The higher F/C ratios (0.42 for

Run 4 and 0.75 for Run 5) aiso showed improved pH results (> 6.5) after 12 hours as

compared to Run 3 (pH< 6.5 for an F/C ratio of 0.25). At an F/C ratio of 0.25, the lower

initial substrate concentration resulted in higher pH levels &er 12 hours (pH of 6.5 for

Run 3 compared to pH of 6.35 for Run 1). For an F/C ratio of 0.42, the pH levels for

both cases were fairly similar (Figure 5.4). The largest F/C ratio (0.75 for Run 5)

resulted in the best reactor performance in ternis of pH (Figure 5.4). Nevertheless, with

the feed solution used, none of the operating conditions examined could be sustained

without extemal pH control.

5.4 Glucose and VFAs vs. Time

For al1 5 runs, glucose concentrations in the reactor were highest during the feed

stage of the initial cycle (Figure 5.6). For runs in which more than one cycle took place,

the glucose concentration peak was significantly lower in subsequent cycles.

1400 - 1

i +Run1 -0-Run2 +Run3:

Figure 5.6 Glucose Concentration vs. T h e for Al1 Ruas


As expected, Runs 3,4, and 5 had lower concentration peaks than Run 1 which

had twice the influent glucose concentration (6000 mg/L vs. 3000 mgk). Similarly,

Runs 2 and 4 showed lower peaks than Runs 1 and 3 because glucose was added more
stowly (Table 5.1). Run 5, as expected, showed the lowest glucose peaks (Figure 5.6).

Because each nm was begun af€erstarving the reactor for 3 to 5 days, the glucose

fermenters may have been in a "dormant" state. When glucose was added to the reactor,

a finite amount of t h e was required for the bacteria to become active once more. In

subsequent cycles, the glucose fermenters would initiate fermentation more rapidly,

having been idle for a shorter length of time. Consequently, Run 5, with the largest F/C

ratio of 0.75 had almost no glucose measured in its third and nfth cycle due to the

reduced "idle tirne" for glucose fermenters. Because ANSBR reactors are fed

intennittently, the tirne between subsequent feedings may be important to their operation.

For al1 5 nins, lactate was produced early in the cycle at low concentrations and

was readily degraded (Figure 5.7). Propionate and acetate were the major constituents of

the SCOD remaining in the effluents (Figures 5.8 and 5.9, respectively).

O 10 20 30 40 50 60
time (hrs)
Figure 5.7 Lactate Concentration vs. Time for Al1 Runs
O 10 20 30 40 50 60
tim (hrs)
Figure 5.8 Propionate Concentration vs. Time for Ali Runs

O 10 20 30 40 50 60
tirne (hrs)

Figure 5.9 Acetate Concentration vs. Time for Al1 Runs

Both propionate and acetate were degraded slowly (< 20 m e per hour)

throughout early cycles, with m s consisting of multiple cycles showing accumulation of

these VFAs. This may have been caused by pH values < 6.5 (Figures 5.4 and 5.5) which

inhibited the microorganisms. In addition, there may have existcd a lack of active

propionate consumers and aceticlastic methanogens. The profiles measured also show
similarly shaped curves for each acid, mth propionate levels just more than double those

of acetate.

Run 1 produced higher propionate and acetate levels than Run 2. Runs 3-4, and 5

showed similar levels of acids, but ali three were lower after 24 hours than Runs 1 and 2,

indicating that for equal F/C ratios, lower acid production occurred with lower influent

glucose concentrations. Run 5, with the largest F/C ratio (0.75) and low influent glucose

concentration tan the longest before failure.

5.5 Summary

Al1 5 runs examined were unable to successfully treat an OLR of 3 kg-m'3-d-'.

Acids @articularly propionic and acetic) accumulated in the reactor lowering the pH, and

the depressed pH values M e r exacerbateci the acid removal problem by inhibithg the

microorganisms. As well, the aggressive feed strategy used (50 % of reactor volume)

may not have been the best choice. Decanting a large volume of effluent fkom the reactor

removes more of the total acids in the reactor and provides more alkalinity fiom the feed.

However, it also introduces a greater mass of glucose to the reactor per cycle which will

ultimately be fermenteci iato further acids. More studies are required to detennine the

effects of varying the fil1 volume on the performance of ANSBRs.


6. SUBSTRATE-SPECIFIC KINETIC PARAMETERS AND BIOMASS

6.1 Motivation for Measuriag Substrate-Specific Parameters

The 5 operational pmtocols çtudied in the experimental work discussed in Chapter

5 were simulated using the model developed by Brodkorb (1998) to determine whether or

not the model predictions agreed with the experimental results obtauied. Kinetic

parameters for individual populations in anaerobic microbial communities are required to

use the model. Historically, overall values for substrate concentration (S) and biomass

(X) have been measured in laboratory studies dealing with microbiological reactors.

However, as the results shown in Chaptet 5 indicate, anaerobic systerns produce and

consume numerous intermediates, even for a simple substrate such as glucose. individual

microbial populations can be measured using nucleic acid probe techniques (Merkel et

al..1999) which can be expensive and time consuming. An alternative method

developed by Brodkorb (unpublished) provides a simple and direct method to measure

both substrate-specific kinetic parameters and individual active biomass hctions.

6.2 Method for Measuring Substrate-Specific Parameters

The Monod equations for consumption of a specific substrate and microbial

growth in a batch systern, neglecting decay, are as follows:

where:

S = substrate concentration, mg COD/L

t = tirne, d
ka = maximum specific substrate utilization rate, mg substrate as COWmg active biomass
as COD/d

Ks= half-saturation constant, mg CODL


X, = active biomass concentration which consumes the substrate, m g COD/L
Y = microbial p w t h yield, mg active biomass as COD/mg substrate consumed as COD

ES >> Ks,then the substrate depletion equation (6.1) becomes zeroth-order with
respect to S, and it can be simplified to:

Substituting equation 6.3a into equation 6.2 gives:

integrating this expression gives:

h X a -hX, =k,Yt (6.6)

which is valid for batch systems where S >> Ks.SubstiMing equation 6.3b into equation

6.6 provides a relationship between the substrate removal rate and t (equation 6.7):

The absolute value signs are needed in equation 6.7 to provide a positive argument for the

natural logarithm since the slope of the substrate degradation curve will be negative. The
dope of a plot of ln
13 vs. tirne is kaY. The value of ka can be calculated using an

appropriate value for the yield, Y, X, can then be calculated as (Brodkorb, 1998):

where k, and Xt are based upon the total biomass rneasured (as VSS). ifthe total biomass

measured is approximately constant (resulting in a linear substrate depletion curve), k,

can be measured as the dope of the substrate depletion curve divided by XIwhich wouid

be measured as VSS:

6.3 Experimental Protocol

The inoccuium source used in experiments to determine the ka and X, values was

a 1:1 mix of anaerobic sludge fiom the North Toronto Treatment Plant and granuiated

sludge fiom storage in our laboratory. Unfortunately, this was not the exact same sludge

as was used for the laboratory-scale experiments (Chapter 5) because the method

descnbed in Section 6.2 was not developed until after the laboratory-scale experirnents

were completed. Nevertheless, the sludge sources were similar, and general trends and

cornparisons between experimental results and cornputer simulations could be made.

Three litres of this mix was placed in a 4-L glas bottle, and it was fed glucose at an OLR

of approximately 0.75 to 1 kgm-3*d-1.Every other day, 500 mL of sludge was removed

fiom the bottle, and replaced with 500 mL of basal medium along with 5 to 6 g of

gIucose. This resulted in an HRT and solids retention time (SRT) of 12 days. The pH of
the bottie remaineci at 7.5 * 0.3 and the alkalinity at 4000 * 200 mg/L as CaC03
throughout the experiment.

6.4 Results Obtaineà

Using the method described in Section 6.2, the ka and X, values were detennined

for 6 constituents of concern: glucose, hydrogen, lactate, acetate, propionate, and

butyrate. The glucose and VFA bottle experiments were run in triplicate, while the

hydrogen test was run in duplicate (due to experimental errors that occurred in the

triplicate bottle).

The bottles were fed twice for the VFA and hydrogen analyses and three times for

the glucose analysis to ensure that S remained »Ks. An overall ktvalue was caiculated

for each substrate using equation 6.9, by averaging both the slopes of the degradation

curves obtained (Figures 6.1 through 6.6) and the VSS values measured. For ail

substrates, the initial and final VSS values varied by less than 10 %, so X, was assuxned to

be approxirnately constant (which would lead to c 1O % error in k,and X, values). Note

that the hydrogen concentrations shown in Figure 6.2 are "nominal" values that are

calcdated as if the hydrogen present in the headspace at any sampling time is completely

dissolved in the liquid phase (Le. gas transfer effects are minimized). This was

considered reasonable, since the method and semm bottles used ailowed excellent mixing

of the bottle contents and good interaction between the gas and liquid phases (Guwy et

al., 1997).
. -

L R~ = 0.993 F? = 0.995
O
O 50 IO0 150 200
time (min)
Figure 6.1 Glucose Degradation Curves

O 20 40 60 80 1O0
time (min)
Figure 6.2 Hydrogen Degradation Curves
O 0.5 1 1.5 2 2.5 3
tirne (hrs)
Figure 6.3 Lactate Degradtion Curves

O 10 20 30 40 50
tirne (hrs)
Figure 6-4 Acetate Degradation Curves
O 40 80 120 160
tirne (hrs)
Figure 6.5 Propionate Degradation Curves

O IO 20 30 40 50
time (hrs)
Figure 6.6 Butyrate Degradation Curves

To determine ka values, la
Id (the natural logarithrn of the absolute value of the

dope of the regressed degradation curve) was plotted for each substnite depletion cuve

55
at its rnidpoint. A summary of the results obtained is given in Table 6.1. AN of the

substrates, except glucose, were fed twice and produceci only 2 points that could be used

to calcuiate ka and Xa. Future work should be conducted with multiple feedhgs (at least

3) for each substrate.

Table 6.1. Substrate-Specific Kinetic Parameters and Active Biomass

Substrate x,' d~ld* t a 3 14 (d-') k. (d-') x.' x,


(?4of Xt)
glucose 5635 12096 12518 2.22 0.071 19.3 650 11.5

hydrogen 878 490 512 0.58 0.069 39.1 13.1 1.5


533

lactate 5 126 2530 2686 0.52 0.064 24.9 108 2.1


2842

acetate 5522 159 201 0.036 0.054 7.56 26.5 0.48


243

propionate 4503 36.7 43.9 0.010 0.059 1.6 27.7 0.62


51.1

butyrate 5579 196 222 0.040 0.067 3.3 67 1.2


247
Sum = 17.4 %
~ einitiai and h a 1 total biomass measured as VSS and converted to COD using
' ~ v e r a of
the relationship 5.65 gVSS = 8 gCOD; units for X = (mg biomass as COD)-L"
2 ~ h slopes
e of the curves in Figures 6.1 - 6.6; units = (mg substrate as COD)-L" -a1
' ~ v e r a ~kx
e = average for the culture; units = r n g ~ O D * ~ - ' - b l
4~romBrodkorb (1998); uni& for Y = @COD biomass)*(gCODsubstrate cons~med)-~
using equation 6.8; units = (mg biomass as COD)-L-'
S~alculated
6.5 Summary

The results show that 11.5 % of the active biomass 0(,) is comprised of glucose

fermenters. Aceticlastic methanogens account for less than 0.5 % of the active biomass.

This low level may have resulted h m biomass washout in the inoccuiurn 4 L reactor, if

the SRT of 12 days was lower than the critical SRT for aceticlastic methanogens. in

general, the low total active biomass measured (17.4 % of VSS) could have been due to

the low temperature used for this work (21°C) as compared to 3 5 T used in the anaerobic

digester where part of the sludge was obtained. In addition, both sludges used (anaerobic

digester and granulated) were stored in a 4OC room pnor to use, and this temperature

shock may have inactivated some of the microorganisms.

Table 6.2 shows a cornparison between the ka values obtained for this work and

selected values fiom the literature.

Table 6.2 Comparison of k Values for Anaerobic Bidegradation

Substrate k a (d-9 k (d-9 Temperature Biomass Reference


(This Study) (Literature)
glucose 19.3 49.9 35 Total A
hydrogen 39.1 2.6 NA' Active B
lactate 24.9 34.6 25 Total C
acetate 7.56 1.8 20 Total A
2.5 NIA' Active B
propionate 1.6 5.5 25 Total A
2.5 NIAI Active B
butyrate 3.3 5.7 35 Total A
5.6 35 Total D
A. Pavlostathis and Giralda-Gomez (199 1)
B.Maillachem and Parkin (1996); '~ern~erature
not given
C. Costello et al. (1991 b)
D. Vavilin and Lokshina (1996)
The k value obtained for glucose is lower than one found in the literature
(Pavlostathis and Giraldo-Gomez, 1991). However, this study was conducted at ambient

lab temperatwe (2 1°C) as opposed to 3S°C,which would result in a lower kinetic

constant. The k values obtained for lactate, propionate and butyrate are similar but

slightly lower than those found in the Iiterature (Pavlostathis and Giraldo-Gomez, 1991;

Costello et al.,1991 b; Vavilin and Lokshina, 1W6), aiso due, in part, to the lower

temperatwe used in this study. The value obtained for acetate is higher than the lite-e

values, partly due to two main factors. First, the value reported by Pavlostathis and

Giraldo-Gomez (1991) was obtained at 20°C (the temperature was not given by

MaiIlachenivu and Parkin, 1996). Second, the k values obtained in this study were based

upon active biomass concentrations consuming the substrate, as opposed to total biomass

(as was used in Pavlostathis and Giralda-Gomez, 1991). Since the amount of active

biomass consuming a specific substrate will always be less than total biomass, k values

witl be higher if they are based on active biomass concentrations as opposed to total

biomass. The value obtained for hydrogen is 15 times greater than the value reported by

Maillacherwu and Parkin (1996). This is likely due to the fact that the bottles and

mixing strategy used in this study appeared to give good opportunity for gas transfer of

hydrogen into the liquid phase, allowing for faster degradation of hydrogen, which may

be similar to the situation that exists in granulated sludge.

Kinetic parameters and active biomass concentrations are not global parameters,

and as such, they should be measured for each treatment system of interest. The

technique described in this chapter is a simple method that can be used to measure

substrate-specific kinetic parameters and active biomass fractions.


7. COMPUTER MODEL SIMULATIONS

7.1 Simulation Methodology

Cornputer simulations of the 5 cases studied in the laboratory were run using the

computer mode1 developed by Brodkorb (1998) and Bagley and Brodkorb (1 999). One

simulation set was conducted using the ka values and X, concentrations given in

Brodkorb (1998) (Table 7.1) The second simulation set was run with the ka and active

biomass partitionhg values determineci for this research work (Chapter 6). The VSS

concentration for the simulation nai was set to 16,000 mg/L to approximate the solids

level in the lab-scale ANSBR during the research work. The ka and Xavalues used are

summarized in Table 7.1.

Table 7.1 Substrate-Specific Parameters Used in the Simulations

Brodkorb (1998) This work


Substmte k.(d") & (W&) k. (d-') x,
glucose 10.0 550 19.3 1840
hydrogen 2.6 175 39.1 240
lactate 5.O 2500 24.9 336
acetate 2.5 150 7.6 77
propionate 2.5 1200 1.6 99

The operational parameters simulated çorresponded to the parameters used in the

laboratory work (Chapter 5 and Table 5.1). As mentioned in Chapter 6, the studge used

to determine the ka and Xavalues was not the same as the sludge used in the laboratory-

scale experiments, but general cornparisons between experirnental resdts and computer

simulations could be made. Alkalinity in the simulation feed was set at 3400 mg& as

CaC03 for all cases examineci, and al1 inhibition and regdation parameters were kept at

the values given in Brodkorb (1998). The computer simulations were run for a total of 5
days each, which would have been the maximum lengh of the experimental nms if they

had not failed due to effluent pH levels below 6.5 (Chapter 5).

7.2 Simulation Results for pH

The simulation pH results ushg kinetic and biomass values fiom Brodkorb (1998)

and this work are given in Figures 7.1 and 7.2, respectively. Both simulation cases

predicted that Runs 1 and 2 would experience signifïcant drops in pH causing process

fdure (effluent pH c 6.5) atkr 1 cycle. A comparison of experimental and simulation

pH values for Run 1 (Figure 7.3) shows that the model predictions differ fiom each other

and the experimental results for the k t 12 hours (0.2 to 0.4 pH units). From hours 12 to

24, however, the simulation and experimental results are nearly identical. A comparison

of the pH results for Run 2 (Figure 7.4) shows that the simulation predictions are almost

the same for the first 33 hours, but then diverge fiom that point until hour 48. The

experimental results are similar to the model predictions for the k t 10 hours of Run 2,

but are signincantly higher fiom that point until hour 48 (> 1 pH unit).

4
O 20 40 60 80 1O0 120
time (hrs)

Figure 7.1 pH vs. T h e for AU 5 Runs Using Parameters h m Brodkorb (199%)


Figure 7.2 pH vs. Time for Ail 5 Runs Using Parameters From This Work

This work

O 5 10 15 20 25
tirne (hm)
Figure 7.3 pH Cornparisons for Run 1
Brodkorb (1998) 8
O 10 20 30 40 50
time (hrs)

Figure 7.4 pH Cornparisons for Run 2

Both simulation cases predicted that Runs 3,4, and 5 would not fail after 5 days

(Le. they would al1 resuit in effluent pH levels above 6.5) (Figures 7.1 and 7.2). Thus, the

mode1 results for pH do not agree with the results obtained experimentally for Runs 3,4,

and 5, because al1 of the cases examined failed in the laboratory. For example, a

cornparison of the pH results for Run 3 betwem the two modelling cases and the

experimental results is given in Figure 7.5. Although the modelling results differ slightiy

at the beginning of the cycle, the finai effluent pH is almost the sarne for both

simulations, and higher than the experimental results (Le. pH of 6.75 for modelling vs.

pH of 6.48 for the expriment).

Since pH is a function of acid concentrations, an examination of the simulation

results for acids is required.


O 2 4 6 8 10 12
time (hrs)

Figure 7.5 pH Cornparisons for Run 3

7.3 Simulation Results for VFAs

The computer sinidations using the parameters of Brodkorb (1 998) predicted that

the majority of the COD in the reactor would be present as lactate and propionate

(Figures 7.6 and 7.7, respectively). The computer simulations using the parameters

calculated in Chapter 6 predicted that acetate wouid be the only component present at any

significant concentration (Figure 7.8). The results obtained in the experimental work

showed that propionate forrned the majority (> 50 %) of the COD in the effluent of the

reactor, with acetate concentrations approximately 50 % of propionate concentrations.

Lactate appeared early in the cycles, and was readily consumed in the reactor. Overlays

of the experimentd results for Run 1 with simulation predictions ushg parameters fiom

Brodkorb (1998) and this work are given in Figures 7.9 and 7.10, respectively. Similar

overlays for Run 3 are given in Figures 7.1 1 and 7.1 2, respective1y.
O 20 40 60 80 1O0 120
tirne (hrs)

Figure 7.6 Lactate vs. Tirne for Al1 5 Runs Using Parameters From Brodkorb (1998)

s
O
2 900 -
Cu
&
P 600 -
Y

Q)
c.
rn
C
.O 300 -
CL
2
P 1
Runs 3 and 4

O 20 40 60 80 1O0 120
time (hrs)

Figure 7.7Propionate vs. T i e for Al1 5 Ruus Using Parameters Frorn Brodkorb (1 998)
3 and 4

Figure 7.8 Acetate vs. Time for Al1 5 Runs Using Parameters From This Work

/\ +lactate

O 5 10 15 20 25
time (hm)

Figure 7.9 Experimental and Simulation Results for Run 1 Using Parameten nom
Brodkorb (1 998) (Connecteci points are experimental, lines are simulation)
1500 -
1200 - acetate
P
0 900:
0,
E 600' +aceîate

O 5 10 15 20 25
time (hrs)

Figure 7.10 Experimental and Simulation Results for Run 1 Using Parameters fÎom
This Work (Comected points are experimental, lines are simulation)

, +lactate
+-onate

O 3 6 9 12
time (hm)
Figure 7.1 1 Experimental and Simulation Results for Run 3 Using Parameters fiom
Brodkorb (1998) (Connected points are experimental, lines are simulation)
! +acetate !

O 2 4 6 8 10 12
time (hrs)

Figure 7.12 Experimental and Simulation Results for Run 3 Using Parameters fiom
This Work (Comected points are experimental, lines are simulation)

7.4 Summary

The lactate accumulations predicted for Runs 1 and 2 using the parameters fiom

Brodkorb (1998) may have k e n due to pH inhibition. Glucose fermenters, which me not

as sensitive to pH drops (Speece, 1996), continue to ferment glucose throughout the feed

stage. By feeding the glucose at a higher initial concentration (6000 mg/L for Runs 1 and

2), lactic acid is produced too quickly initially, and the &op in pH causes inhibition of the

lactic acid consumers, thus leading to the predicted accumulation of lactate. Propionate

is modelled to accumulate at approximately 40 mg& per hour for the fïrst 20 hours of

Run 1 (Figure 7.9).

For the simulations run using the parameters calculateci in Chapter 6, only acetate

was predicted to occur at any significant concentration (Figure 7.8). This differed fiom

the predictions made using the parameters fiom Brodkorb (1998) due to the differences in
values of kinetic parameters used. The kinetic parameters in Brodkorb (1998), except for

propionate, are ali lower than the values calculated in Chapter 6 (Table 7.1). This is

particularly tme for the ka value for hydrogen which was 2.6 d-' in Brodkorb (1998) and

39.1 de' for this work. It is possible that the model predictions using parameters nom this

work predicted acetate would be formed almost exclusively fiom the fermentation of

glucose due to rapid consumption of hydrogen by the hydrogenotrophic methanogens.

By readily consuming the hydrogen produced, the product regulation fiinctions used in

the model (Brodkorb, 1998) may predict acetate would be formed much more readily

than the other VFAs. For Runs 1 and 2, the model may have predicted the acetate would

eventually have been consumed if the percentage of active aceticlastic methanogens was

higher than the value of 0.48 % obtained in Chapter 6, although the depressed pH values

may still have k e n sueEiciently inhibitory. It is possible that no lactate accumulated in

the experimental work because the lactate consumers were not as sensitive to pH drops as

was predicted by the model and the actual ka value may have differed.

Both sets of simulation results predicted that the lower innuent glucose

concentrations of Runs 3,4, and 5 wodd ensure that the reactor wouid perform

successfùlly. As was the case for Runs 1 and 2, the simulations run with the parameters

fiom Brodkorb (1998) predicted that lactate would accumulate, and propionate

production for Run 3 (Figure 7.1 1) was modeiied to be 40 mg/L per hou.. in addition,

the simulations performed with the parameters fiom Chapter 6 predicted that acetate

would be the only VFA produced to any significant extent for Runs 3,4, and 5 (Figtire

7.8). The possible reasons for these preâicted acid distributions are the same as those for

Ruas 1 and 2 discussed previously.


The experimental results did show improved results for the lower glucose

concentration, but none of the cases ran successfully for 5 days. It is possible that the

lower influent glucose concentration of 3000 mg& was just low enough to prevent

sufficient acid accumulation in the first few cycles to cause the simulated pH to drop

below 6.5. For example, at the end of the first cycle for Run 3, the computer simulations

predicted a total acid concentration of 650 to 700 mg/L (Figures 7.1 1 and 7.12) and a

corresponding pH value of approximatety 6.8 (Figure 7.5). In cornparison, at the end of

Run 1 the acid concentration was predicted by h t h sets of parameters to be

approximately 1200 mg& (Figures 7.9 and 7.1O), with a conesponding pH of 6.4 (Figure

7.3). From Figure 5.5, acid concentrations above 1000 mg/L caused the pH to drop to

intribitory levels of below 6.5 in the laboratory.

Thus, the specific parameters used in these simulations resulted in predictions of

severe inhibition for the higher glucose concentration of 6000 mg/L which agreed with

experimental results, and successful operation for an influent concentration of 3000 mg&

which did not agree with experimental results. It is possible that this occurred because

the pH inhibition parameters (Brodkorb, 1998) and kinetic parameters (Table 7.1 ) used in

the model were not representative of the values that existed in the reactor. Thus, some of

the microorganism groups may have k e n more or less susceptible to pH drops in reality

as compared to the model predictions. In addition, the actual ka and X, parameters that

existed in the laboratory rnay have contributed to the differences between model

predictions and experimental results. It will be necessary to measure these pH inhibition

parameters and ka and X, values experimentally for each system of interest to improve the

model's predictive capabiiity.


8. DISCUSSION AND CONCLUSIONS

A laboratory-sale anaerobic sequencing batch reactor (ANSBR) fed glucose was

initially run to maximize the organic loading rate (OLR) that it couid successfiilly treat.

However, the ANSBR was not able to treat an OLR of 3 kg-rn-'*d-l due to inhibitory

effluent pH levels of less than 6.5 caused by acids (padcularly propionate and acetate)

accumulating to concentrations of 1000 mg/L and greater. ANSBR operational

parameters, such as fiil-to-cycle tirne (WC)ratios, total cycle t h e , and influent

concentration, were modined to improve the reactor's performance at an OLR of

3 kg-m-3-d-'. Kennedy et al. (199 1) concluded that low nIl-to-react ratios of 0.2 - 0.5

significantly reduced ANSBR treatment efficiency for a soluble sucrose/acetate

wastewater at OLRs of 9 kgrn-'=d-' and above, which agrees with the results obtahed in

this research work. The work by Kennedy et al. (199 1) was able to achieve higher

ioadings because it was conducted at a higher temperature than this work (35°C vs. 21°C)

and granulated sludge was used and rnaintained throughout the experiments. In addition,

sucrose may be easier to biodegrade anaerobically than glucose, since the sucrose must

first be broken down into its constituent carbohydrate monomers (Le. glucose and

fructose), and this may slow down the overall production rate of acids such as propionate

which are difficult to degrade. Brodkorb (1998) also concluded that longer fil1 times for

a given total cycle time improve ANSBR treatment of a rapidly acidifjhg substrate such

as glucose. Experimental runs with a lower influent glucose concentration (3000 mg/L

vs. 6000 mg/L) for a given OLR and F/C ratio also showed improved ANSBR

performance. In practice, the influent concentration of a wastewater could possibly be


lowered by diluting the stock feed solution with treated wastewater prior to adding it to

the reactor.

Cornputer simulations for an OLR of 3 ~ ~ - r n - ~were


- d - 'run using the model of

Brodkorb (1998) and Bagley and Brodkorb (1999) using substrate-specific kinetic

parameters and active biomass fiactions based on values in Brodkorb (1998) and this

research work. Using both sets of parameters, the model predicted that the reactor would

fail due to pH inhibition for an influent glucose concentration of 6000 mg&, which

agreed with the experimental results. However, the model predicted that the ANSBR

would successfirlly treat dl t h e cases with an influent concentration of 3000 mg/L for

both sets of parameters, which differed h m the experimental results obtained. In

addition, for the three cases nui with the lower influent concentration, the model

predicted for both sets of parameters that varying the F/C ratio wodd not have a

significant impact on reactor pedonnance which difEered fiom the experimental resuits

obtained in this work. Thus, the model predicted that the reactor system was below its

maximum capacity since varying the F/C ratio had little effect. However, previous

modelling work by Brodkorb (1 998) did predict that reactor performance would be

iduenced by varying the F/C ratio. Thus, the substrate-specific kinetic parameters,

active biomass concentrations, and inhibition and regdation parameters used in the

simulations differed from the values that were present in the experimental reactor system.

For al1 of the experiments conducted for this research work, the fill volume-to-

total volume ratio was set to 0.5. Decanting such a large volume of effluent h m the

reactor did remove more of the total acids in the reactor and provided more alkalinity

fiom the feed. However, it also introduced a p a t e r mass of glucose per cycle which was
ultimately fermented into M e r acids. Some previous researchers (e.g. Sung and

Dague, 1995) have varied the fiii volume-to-total volume ratio, but this was done simply

to Vary the OLR by increasing or decreasing the hydraulic retention t h e (HRT). No

previous studies have closely looked at the effect varying the fill volume-to-total volume

ratio would have on reactor performance at a given OLR.

From this work, the following conclusions can be made:

Operational control alone is not sufncient for start-up of an ANSBR treating a rapidly

acidi-g substrate such as glucose; extemal pH control is also required.

Longer n1l times for a given cycle time (F/C ratios of 0.42 and 0.75 vs. 0.25) improve

the performance of an ANSBR treating a rapidly acidifying substrate such as glucose.

For identical fikcycle time ratios at the same loading rate, shorter cycle times with

lower initial substrate concentrations provide improved reactor performance.

Substrate-specific kinetic parameters and active biomass must be determined for the

anaerobic treatment systems of interest; the technique presented in this thesis is a

simple method that could be used for this detennination.

For two sets of kinetic parameters, the ANSBR cornputer model predicted that the

reactor would fail with the higher influent glucose concentration of 6000 mg/L and

successfully treat the Iower influent glucose concentration of 3000 mg& for al1 F/C

ratios used (at an OLR of 3 kg-m-3 ad'


1
).

For two sets of kinetic parameters, the ANSBR model predictions agreed with the

experimental results of reactor failure for 6000 mg/L infïuent and did not agree with

the experimental resuits of reactor failure for 3000 mg& influent due to differences

between parameter values existing in the laboratory and used in the model.
9. RECOMMENDATIONS

The foiiowing recommendations are made for possible friture work:


Study the effects of varying fili volume-to-reactor volume ratios on the performance

of a laboratory-scale ANSBR at a given organic Loading rate (OLR).

Run ANSBR bidegradation experiments using both pdysaccharides such as

carbohydrate dimers (e.g. sucrose; lactose; etc.) and glucose. Directly compare the

degradation resuits to detennine if more complex cabhydrates are easier to degrade

anaerobically than glucose.

Measure substrate-specific kinetic parameters, active biomass concentrations, and pH

and Hîregdation and inhibition coefficients used in the model in experimentai

studies. This measured &ta can then be used to M e r improve the model.

Directly compare the active biomass measured by the method presented in this thesis

to other methods, including nucleic acid probes. Simulations can then be rwi based

on the varying active biomass distributions obtained with the biomass then used in

lab-scale experiments and the results compared to those obtained via simulation. The

measurement method most representative of experimental reactor results can then be


10. REFERENCES

Aguilar, A., C . Casas, and J.M. Lema. 1995. Degradation of Volatile Fatty Aci& by

Differently Enricheci Methanogenic Cultures: Kinetics and Inhibition. Wat. Res. 29,2,

505-509.

Angenent, L.T. and RR. Dague. 1995. A Laboratory-Scale Cornparison of the UASB and

ASBR Processes. SO'~Purdue Industrial Wasîe Confrence Proceedings, A m A r b r

Press, Inc. Chelsea, MI. 365-377.

APHA, AWWA, and WEF. 1998. Standard Methods for the Examination of Water and

Wastewater,206 ed. Washington DC.

Bagley, D.M. and T.S. Brodkorb. 1999. Moàeling the Microbial Kinetics in an Anaerobic

Sequencing Batch Reactor - Model Development and Experimentai Validation. Water

Environ. Res. In Press.

Banik, G.C., T.G.Ellis, and R R . Dague. 1997. Structure and Methanogenic Activity of

Granules fiom an ASBR Treating Dilute Wastewater at Low Temperatures. Wat. Sci.

Tech. 36,6-7.149-156.

Brock, T.D. and M.T.Madigan. 1991. Biology of Microorganisms. dhEd. Prentice Hall,

New Jersey.

Brodkorb,T.S.1998. Dynamic Simulation of Microbial Kinetics in an Anaerobic

Sequencing Batch Reactor. MEng. Thesis, University of Toronto, Toronto, ON.

Chang, D., J.M. Hur, and T.H.Chung. 1994. Digestion of Municipal Sludge by

-
Anaerobic Sequencing Batch Reactor. Wat. Sci. Tech. 30, 12, 161 170.

Costello, D.J.,P.F. Greeafield, and P.L. Lee. 1991a. Dynamic Modelling of a Single-

Stage High-Rate Anaerobic Reactor - 1. Model Derivation. Wat. Res. 25,7,847-858.


Costello, D.J., P.F. Greenfield, and P.L.Lee. 1991b.Dynamic Modelhg of a Single-

Stage High-Rate Anaerobic Reactor - II. Mode1 Verification. Wut. Res. 25,7,859-

871.

Dague, R.R., C.E. Habben, and S.R. Pidaparti. 1992. Initial Studies on the Anaerobic

Sequencing Batch Reactor. Wat. Sci. Tech. 26,941,2429-2432.

Dague, R.R. and S.R. Pidaparti. 1992. Anaerobic Sequencing Batch Reactor Treatment of

Swine Wastes. 46hPurdue IndutriaI Wàste Conference Proceedings, Lewis

Publishers, Inc, Chelsea, MI. 75 1-760.

Dague, R.R, G.C. Banik, and T.G.Eilis. 1998. Anaerobic Sequencing Batch Reactor

Treatment of Dilute Wastewater at Psychrophilic Temperatures. Water Environ Res.

70,2, 155-160.

Drake, H.L. ed. 1994. Acetogenesis. Chapman & Hall,New York.

Dubourguier, H.C., M.N.Buisson, J.P.Tissier, G. Premier, and G. Albagnac. 1987.

Structural Characteristics and Metabolic Activities of Granular Methanogenic Sludge

on a Mixed Defined Substrate. ln Proceedings of the G A S M T Workshop.Lunteren,

The Netherlands. 78-86.

El-Mamouni, R., R. Leduc, and S.R. Guiot. 1997. Muence of the Starting Microbial

Nucleus Type on the Anaerobic Granulation Dynamics. Appl. Microbiol. Biotechnol.

47, 189-194.

Fang, H.H.P. and H.K. Chui. 1993. Maximum COD Lnading Capacity in UASB Reactoa

at 37°C. J Environ. Eng. 119, 1, 103-119.


Fernandes, L., K.J. Kennedy, and 2.Ning. 1993. Dynamic Modelling of Substrate

Degradation in Sequencing Batch Anaerobic Reactors (SBAR). Wat. Res. 27, 11,

1619-1628.

Ferry, J.G., ed. 1993. Methanogenesis: Ecology. Physiology. Biochemistry, and Genetics.

Chapman & Hd,New York.

Fontaine, F.E., W.H.Peterson, E. McCoy, M.J. Johnson, and GJ. Ritter. 1942. A New

Type of Glucose Fermentation by Clostridium thermoaceticum n. sp. J , Bucteriol. 43,

701-715.

Forster, C.F.and 1. Quarmby. 1995. The Physical Characteristics of Anaerobic Granular

Sludges in Relation to Their Intemal Architecture. Antonie van Leeuwenhoek. 67, 103-

110.

Cirotenhuis, J.T.C.,J.C. Kissel, C.M. Plugge, A.J.M. Stams, and A.J.B. Zehnder. 1991.

Role of Substrate Concentration in Particle Size Distribution of Methanogenic

Granular Sludge in UASB Reactors. Wat. Res. 25, 1,2 1-27.

Guwy, A.J., F.R. Hawkes, DL. Hawkes, and A.G. Rozzi. 1997. Hydtogen Production in

-
a High Rate Fluidised Bed Anaerobic Digester. Wat. Res. 31,6,1291 1298.

Henze, M., C. Grady, W.Gujer, G. Marais, and T. Matsuo. 1987. Activated Sludge

Model No. 1. IAWPRC Scientifïc and Technical Report No. 1, IAWPRC, London.

Henze, M., W. Gujer, T. Mino, T. Maîsuo, M.C. Wentzel, and G.v.R. Marais. 1995.

Activated Sludge Model No. 2. IAWQ Scientific and Technical Report No. 3, IAWQ,

London.
Huishoff Pol, L.W., J.J.M.van de Worp, G. Lettinga, and W.A. Beverlw. 1986. Physicai

Characterization of Anaerobic Gtanular Sludge. In Anaerobic Treatment. A Grown-Up

Technologv. Aquatech '86. Amsterdam, The Nethertands. 89- 10 1.

b a i , T., M. Ukita, J. Liu, M. Sekine, H. Nakanishi, and M. Fukagawa. 1997. Advanced

Start Up of UASB Reactors by Adding of Water Absorbing Polymer. Wat. Sci Tech.

36,6-7,399-406.

inanc, B., S. Matsui, and S. Ide, 1996. Propionic Acid Accumulation and Controllhg

Factors in Anaerobic Treatment of Carbohydrate: Effects of Hzand pH. Waf.Sci.

Tech. 34,s-6,3 17-325.

Ince, O., G.K. Anderson, and B. Kasapgil. 1995. Control of Organic Loading Rate Using

the Specific Methanogenic Activity Test During Start-Up of an Anaerobic Digestion

System. Wat. Res. 29, 1,349-355.

James, A., C.A.L. Chernicharo, and C.M.M. Campos. 1990. The Development of a New

Methodology for the Assessrnent of Specific Methanogenic Activity. Waf.Res. 24,7,

8 13-825.

Jawed, M. and V. Tare. 1998. Disintegration of Granular Sludge in UASB Reactor.

WEFTEC Asia 1998 Proceedings. 24 1-248.

Kato, M X , J.A. Field, and G. Lettinga. 1993. High Toletance of Methanogens in

Granular Sludge to Oxygen. Biotechnology und Bioengineering. 42, 11 , 1360-1366.

Kennedy, K.J., W.A. Sanchez, M.F. Hamoda, and R.L. Droste. 1991. Performance of

Anaerobic Sludge Blanket Sequencing Batch Reactors. Res. J. Water PofZuf.Confrol

Fed. 63, 1,75-83.


Kunst, A., B. Draeger, and J. Ziegenhom. 1983. UV-Methods with Hexokinase and

Glucose-6-Phosphate Dehydrogenase. In Methods of Emymatic Amlyszk, P E d .

Volume V7,Metabolites 1: Carbohydrates. H.U.Bergmeyer, e d in chief. Verlag

Chemie, Weinheim.

MacLeod, F.A., S.R. Guiot, and J.W.Costerton. 1990. Layered Structure of Bacterial

Aggregates Produced in an Upflow Anaerobic Sludge Bed and Filter Reactor. Appl.

Environ. Microbiol. 56,6, 1 598- 1607.

Mahoney, E.M., L.K.Varangu, W.L. Cairns, N. Kosaric, m d RG.E. Murray. 1987. The

Effect of Calcium on Microbiai Aggregation During UASB Reactor Start-Up. Wat.

Sci Tech. 19,2499260.

MaiIlacheruvu, KY.and G.F.Parkin. 1996. Kinetics of Growth, Substrate Utilization

and Sulfide Toxicity for Propionate, Acetate, and Hydrogen Utilizers in Anaerobic

Systems. Water Em~iron.Res. 68, 1099-1106.

McCarty, P.L. and D.P.Smith. 1986. Anaerobic Wastewater Treatment.Environ. Sci.

Technol. 20, 12, 1200- 1206.

Merkel, W., A. Schwarz, S. Fritz, M.Reuss, and K. Krauth. 1996. New Strategies for

Estimating Kinetic Parameters in Anaerobic Wastewater Treatment Plants. Watt Sci.

Tech. 34,s-6,393-401.

Merkel, W., W. Manz, U.Szewzyk, and K. Krauth. 1999. Population Dynamics in

Anaerobic Wastewater Reactors: Modelling and In Situ Characterization. Wat. Res.

33, 10,2392-2402.
Mosey, F.E. 1983Mathematicai M o d e h g of the Anaembc Digestion Process:

Regulatory Mechanisms for the Formation of Short-Chain Volatile Acids f b m

Glucose. Wat. Sci. Tech. 15,209-232.

Ndon, U.J. and RR Dague. 1997. Effects of Temperature and Hydraulic Retention Time

on Anaerobic Sequencing Batch Reactor Treatment of Low-Strength Wastewater.

Waf.Res. 31, 10,2455-2466.

Ng,WJ., W.J. Jiang, S.L.Ong, and MA. Aziz. 1998. Two-Stage Anaerobic SBR
Treatment of Rice Wastewater. WEFTEC Asia 1998 Proceedings. 563-566.

Pavlostathis, S.G., and E. Giraldo-Gomez. 1991. Kinetics of Anaerobic Treatment. Fat.

Sci. Tech. 24,8,35-59.

Quarmby, J. and C.F.Forster. 1995. A Comparative Study of the Intemal Architecture of

Anaerobic Granular Sludges. J. Chem. Tech. Biotechnol. 6 3 , 6 0 4 .

Schmit, C.G. and R.R Dague. 1993. Anaerobic Sequencing Batch Reactor Treatment of

Swine Wastes et 20°C, 2S°C, and 3S°C. 4ghPwdue Industrial Wmte Confience

Proceedings, Lewis Publishers, Inc. Chelsea, MI. 54 1-549.

Shen, C.F.and S.R.Guiot. 1996. Long-Tem Impact of Dissolved 02 on the Activity of

Anaerobic Granules. Biotechnolo~and Bioengineering. 49,6,6 11-620.

Snoeymk, V.L. and D.Jenkuiç. 1980. Water Chemistry. John Wiley & Sons, New York.

Sohngen, N.L.1910. Sur le Rôle du Méthane Dans la Vie Organique. Rec. Trav. Chirn.

29,238-244.

Soto, M., R. Méndez, and LM. Lema. 1993. Methanogenic and Non-Methanogenic

Activity Tests. Theoretical Basis and Experimental Set Up. Wat. Res. 27, 8, 1361-

1376.
Speece, RE. 1983. Anaerobic Biotechaology for Industrial Wastewater Treatment.

Environ. Sci. Technol. 17,9,4 16A-427A.

Speece, R.E. 1996. Anaerobic BiotechnoZogyfor Industrial Wartewaters. Archae Press.

Nashville, TN.

Stams, A.J.M. and C.M.Plugge. 1994. Occurrence and Function of the Acetyl-CoA

Cleavage Pathway in a Syntrophic Propionate-Oxidizing Bactenum. In Acetogenesis.

H.L. Drake, ed. Chapman & Hall, New York.

Stephenson, M. and L.H.Strickland. 1933. Hydrogenase. m.The Bacterial Formation of

Methane by the Reduction of One-Carbon Compounds by Molecular Hydrogen.

Biochem. J. 27, 15 17-1527.

Sung, S. and R.R. Dague. 1992. Fundamental Principles of the Anaerobic Sequencing

Batch Reactor Process. 4? Purdue Industrial Wmte Confrence Proceedings. Lewis

Publishers, Inc. Chelsea, MI. 393-408.

Sung, S. and R.R. Dague. 1995. Laboratory Studies on the Anaerobic Sequencing Batch

Reactor. Water Environ. Res. 67,294-30 1.

Suthaker, S., C. Polprasert, and RL. Droste. 1991. Sequencing Batch Anaerobic Reactors

for Treatment of a High-Strength Organic Wastewater. Wat.Sci. Tech. 23, 1249-1257.

Tay, J-H. and Y-G. Yan. 1997. Anaerobic Biogranulation as Microbial Response to

Substrate Adequacy. J. Environ Eng. 123, 10, 1002- 1010.

Thaveesn, J., D. DafTonchio, B. Liessens, P. Vandemeren, and W. Verstraete. 1995.

Granulation and Sludge Bed Stability in Upflow Anaerobic Sludge Bed Reactors in

Relation to Surface Thermodynamics. AppZ. Environ. Microbiol. 61, 10,368 1-3686.


Vavilin, V.A. and L. Ya. Lokshina. 1996.ModeLing of Volatile Fatty Acids Degradation

Kinetics and Evaluation of Microorganism Activity. Bioresource TechnoZow. 57,69-

80.

Wiegant, W.M. 1987. The "Spaghetti Theorf' on Anaerobic Granuiar Sludge Formation,

or the Inevitability of Granulation. in GranuIar Anaerobic Sludge; Microbiology and

Technology, Proceedings of the GASUAT Worhhop. Lunteren, The Netherlands. 146-

151.

Wirtz, R.A. and R-R Dague. 1996.Enhancement of Granulation and Start-Up in the

Anaerobic Sequencing Batch Reactor. Water Environ. Res. 68,5, 883-892.

Wolin, M.J.and T.L.Miller. 1994.Acetogenesis fiom C a in the Human Colonic

Ecosystem. In Acetogenesis, H.L.Drake, ed. Chapman & Hall, New York.

Wu, M.M. and R.F.Hickey. 1995.Dynamic Mode1 for UASB Reactor Including Reactor

Hydraulics, Reaction, and D i f i i o n . J. Environ. Eng. 123,3,244-252.

Wu, W-M., J.H. Thiele, M.K. Jain, H.S. Pankratz, RF. Hickey, and J.G. Zeikus. 1993a.

Cornparison of Rod- Versus Filament-Type Methanogenic Granules: Microbial

Population and Reactor Performance. Appl. Microbiol. Biotechnol. 39,795-803.

Wu, W-M., J.H. Thiele, M.K.Jain, and J.G. Zeikus. 1993b.Metabolic Properties and

Kinetics of Methanogenic Granules. Appl. Microbiol. Biotechnol. 39,804-811.

Wu, W-M., M.K.Jain, J.H. Thiele, and J.G. Zeikus. 1995.Effect of Storage on the

Performance of Methanogenic Granules. Wut. Res. 29,6, 1445- 1452.

Yan, Y-G.and J-H.Tay. 1997.Characterisation of the Granulation Process During

UASB Sm-Up. Wat. Res. 31,7, 1573-1580.


Zeikus, J.G., L.H. Lynd, TH.Thompson, JA. KrzycLi, P.J. Weimer, and P.W.Hegge.

1980. isolation and Cbaracterizatioa of a New MethyIotrophic, Acidogenic Anaerobe,

the Marburg Strain. C m Microbiol. 3,38 1-386.

You might also like