 

 

'.
1%..
”n. u;

w,
t w

11¢
3...

 

 

 

LIBRARY
Michigan State
L.— UHIWTSI‘IY

 

 

This is to certify that the
dissertation entitled

DESIGN IMPLICATIONS FOR ANAEROBIC MEMBRANE

BIOREACTORS AND THE METABOLIC INFLUENCE OF

CYCLE TIME FOR THE TREATMENT OF LIQUID DAIRY
MANURE

presented by

JAMES M. WALLACE

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Biosystems and Agricultural
Engineering

 

 

Jam. J2. AM“...~

Major Professor’s Signature

4‘44} ‘1"; Z00?
Date

MS U is an Afﬁnnative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:lProj/Aoc&Pres/ClRC/DateDue.indd

 

DESIGN IMPLICATIONS FOR ANAEROBIC MEMBRANE BIOREACTORS AND
THE METABOLIC INFLUENCE OF CYCLE TIME FOR THE TREATMENT OF
LIQUID DAIRY MANURE

By

James M. Wallace

A DISSERTATION

Submitted to
Michigan State University

in partial fulfillr'nent of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Biosystems and Agricultural Engineering

2009

ABSTRACT

DESIGN IMPLICATIONS FOR ANAEROBIC MEMBRANE BIOREACTORS AND
THE METABOLIC INFLUENCE OF CYCLE TIME FOR THE TREATMENT OF
LIQUID DAIRY MANURE
By
James M. Wallace
This research developed a design approach for an anaerobic membrane bioreactor

(AnMBR) treating liquid dairy manure with consideration of the cycle time impact on
microbial activity. The research builds from initial comparison experiments with an
AnMBR and a complete mix digester (CMD) and concludes with testing of various cycle
time conditions necessary to develop a qualitative understanding of the associated
microbiology. The results from this research and those of previous researchers were
integrated into speciﬁc design considerations for an AnMBR treating liquid dairy
manure.

A pilot-scale AnMBR and an identically sized CMD were designed and constructed to
treat a sand-separated dairy manure. The CMD produced 54% more methane than the
AnMBR operating at a cycle time of 84. Despite the apparent negative impact on
microbial activity, the AnMBR produced an efﬂuent permeate devoid of suspended solids
with a COD reduction of 89%. There was also a strong correlation between membrane
ﬂux rate and the total solids (TS) concentration of the digester system that indicated
declining ﬂux rate with increasing digester TS concentration.

Based on the initial results, a combined CMD/AnMBR digester conﬁguration was

studied where the CMD efﬂuent was used as the AnMBR inﬂuent. Metabolic evaluation

of the biomass from the CMD and the AnMBR using a respirometer setup indicated a

reduction in the interaction between fatty acid oxidizing bacteria and hydrogen
consuming methanogens (syntrophic relationship); however, some activity remained.

A ﬁnal set of experiments evaluated the impact of cycle time, digester volatile solid
concentration and cross-ﬂow velocity on the rate of methane production for two AnMBR
systems and a control CMD. All digesters received the same sand and solid-liquid
separated manure feedstock.

Cycle times as high as 27/day and cross ﬂow velocities up to 4.5 m/s did not produce
a negative effect on methane production compared to a CMD control while total VFA
concentration for the AnMBR digesters was lower than that of the CMD. Metabolic
evaluation illustrated a reduction in syntrophic activity compared to the CMD; however,
even at a cycle time of 27/day, the AnMBR biomass retained approximately 25% of the
syntrophic activity of the CMD biomass.

Operation at the higher VS concentration of the AnMBR did not confer a methane
production advantage compared to the CMD for the operating conditions tested.
Considering low VF A concentrations in all of the systems, it was theorized that once
steady-state operation was attained, hydrolysis mass transfer limitations controlled
available substrate for anaerobic degradation.

Based on the ﬁndings of this research, the AnMBR process, when operated at cycle
times of 27/day or less, provided equal gas production to a CMD while reducing the
COD, phosphorus and pathogen/virus loading by approximately 90%, 95% and 99.96%

respectively.

Dedicated to my wife Amy and daughters Abigail, Margaret and Lucille

iv

ACKNOWLEDGEMENTS

I would like to express my gratitude to my major professor and research advisor, Dr.
Steve Safferman, for his guidance and support during my Ph.D. study. Special thanks are
also extended to my other research advisor, Dr. Bill Bickert, for his support throughout

the entire process.

TABLE OF CONTENTS
LIST OF TABLES ............................................................................ x

KEY TO ABBREVIATIONS xiv

CHAPTER 1

INTRODUCTION ........................................................................... 1
CHAPTER 2
BACKGROUND ............................................................................ 4
2.1 Anaerobic Digestion Process ............................................. 4
2.2 Hydrolysis .................................................................. 5
2.3 Acidogenesis ................................................................ 6
2.4 Acetogenesis ................................................................. 6
2.5 Methanogenesis ............................................................ 8
2.6 High Rate Anaerobic Digestion ......................................... 9
2.7 Anaerobic Reactors Coupled with External UF Membranes
(AnMBR) ................................................................... 13
2.8 Advantages of AnMBR for Nutrient and Pathogen Management... 17
2.9 Objective .................................................................... 19
2.10 Research Outline ........................................................... 20
CHAPTER 3
UTILITYOF THE ANAEROBIC MEMBRANE BIOREACTOR ................... 26
3.1 Introduction ................................................................. 26
3.2 Materials and Methods .................................................... 26
3.2.1 Analytical Methods .......................................... 26
3.2.2 Substrate 27
3.2.3 CMD System .................................................. 28
3.2.4 AnMBR System ............................................. 29
3.2.5 Combined CMD/AnMBR .................................... 29
3.3 Results and Discussion ................................................... 33
3.3.1 Comparison of an AnMBR and a CMD .................. 33
3.3.2 Combined CMD/AnMBR ................................... 42
3.3.3 Nutrients ......................................................... 53
3.3.4 Removal of Virus and Pathogen Indicators ............... 54
CHAPTER 4
CYCLE TIME COMPARISON ............................................................ 56
4.1 Introduction ................................................................. 56
4.2 Materials and Methods .................................................... 60
4.2.1 General AnMBR Conﬁguration ............................ 60
4.2.2 Phases 1 ........................................................ 61
4.2.3 Phase 2 ......................................................... 64

vi

4.2.4 Phase 3 .......................................................... 67

4.2.5 Substrate. ...................................................................... 68
4.3 Results and Discussion ................................................... 69
4.3.1 Phase 1 ......................................................... 69
4.3.2 Phase 2 ......................................................... 78
4.3.3 Phase 3 ......................................................... 92
4.4 Summary .................................................................... 101
CHAPTER 5
METABOLIC EVALUATION OF CYCLE TIME .................................... 102
5.1 Introduction ................................................................. 102
5.2 Materials and Methods .................................................... 105
5.2.1 Experimental Setup .......................................... 105
5.2.2 Methanogenic Activity Setup .............................. 105
5.2.3 Acidogenic Activity Setup .................................. 106
5.2.4 Dilution Media composition ................................ 106
5.2.5 Operational Procedure ....................................... 107
5.3 Analytical Methods ........................................................ 109
5.3.1 General ......................................................... 109
5.3.2 Microscopic Observations .................................. 109
5.4 Results and Discussion ................................................... 110
5.4.1 CMD/AnMBR Respirometer ............................... 111
5.4.2 Phase 1 Respirometer ................................................... 117
5.4.3 Phase 2 Respirometer. . . . . . .. .......................................... 123
5.4.4 Microscopy ................................................... 129
5.4.5 Most Probable Number ...................................... 130
5.5 Summary .................................................................... 131
CHAPTER 6
AnMBR DESIGN CONSIDERATIONS ................................................ 135
6.1 Introduction ................................................................. 135
6.2 Cycle time .................................................................. 135
6.3 Cross Flow Velocity and Membrane Conﬁguration .................. 136
6.4 Operating Pressure ......................................................... 140
6.5 Total Solids Concentration and Flux Rate .............................. 141
6.6 HRT and SRT .............................................................. 143
6.7 Pump Selection ......................................... . .................. 145
6.8 Membrane Pore Size ...................................................... 146
6.9 Cleaning Protocol ......................................................... 147
6.10 Summary .................................................................... 149
CHAPTER 7
ENGINEERING SIGNIFICANCE AND FUTURE WORK .......................... 150
7.1 Summary of Research Findings .......................................... 150
7.2 Future work ................................................................. 151
7.2.1 Increased OLR ................................................ 151

vii

7.2.2 Temperature impact on ﬂux rate ........................... 153

7.2.3 Flux recovery with cleaning........................ ........ 154
APPENDICES
A. Reprint of “Removal of Viruses and Indicators by Anaerobic Membrane
Bioreactor Treating Animal Waste ................................................. 155
B. Volatile Fatty Acid Titration Procedure ........................................... 174
C. Most Probable Number Methodology ............................................. 175
D. Example AnMBR Analysis ........................................................... 177
E. GCMS Procedure ..................................................................... 180
F. Photographs of Pilot Digesters ...................................................... 181
LIST OF REFERENCES ................................................................... 183

viii

TABLE
2.1

2.2
3.1
3.2
3.3
3.4
3.5
3.6

3.7

3.8
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.1 1

4.12

LIST OF TABLES

High rate anaerobic digester conﬁgurations ....................................
Summary of anaerobic membrane bioreactor systems ........................
Substrate characteristics for initial CMD and AnMBR comparison. . . . .
Substrate characteristics for combined CMD/AnMBR .......................
Summary of operating data for AnMBR and CMD ...........................
Summary of water quality data for AnMBR and CMD ......................
Summary of operating data for combined CMD/AnMBR ...................
Summary of water quality data for combined CMD/AnMBR days 1-69..
Summary of water quality data for combined CMD/AnMBR days 70-
Water quality at each sampling point ............................................
Characteristics of substrate for Phase 1 .........................................
Characteristics of substrate for Phase 2 .........................................
Characteristics of substrate for Phase 3 .........................................
Summary of operating data for Phase 1 .........................................
Summary of water quality data for Phase 1 ....................................
Summary of operating data for Phase 2 .........................................
Summary of water quality data for Phase 2 — SM AnMBR ..................
Summary of efﬂuent quality data for Phase 2 — MM AnMBR ...............
Summary of efﬂuent water quality data for Phase 2 — CMD ................
Volatile fatty acid data ............................................................
Summary of operating data for Phase 3 .........................................

Summary of water quality data for Phase 3 ....................................

ix

21
22
27
27
33
33
43

43

43
'52
67
67
67
69
69
79
80
81
81
82
9O

9O

4.13

4.14

4.15

5.1

5.2

5.3

5.4

5.5

5.6

5.7

5.8

5.9

6.1

6.2

6.3

6.4

Volatile fatty acid data, SMD .................................................... 96
Volatile fatty acid data, MMED .................................................. 96
Volatile fatty acid data, CMD .................................................... 96
Substrates used for metabolic testing ............................................ 101
Dilution media composition ...................................................... 104
CMD/AnMBR respirometer feed ratios, g substrate/ g VSS .................. 108
Summary of CMD/AnMBR respirometer results, mL CH4/ g VSS/hr ...... 109
Phase 1 respirometer feed ratios, g substrate/g VSS ........................... 114
Summary of Phase 1 respirometer results, mL CH4/ g VSS/hr ............... 114
Phase 2 respirometer feed ratios, g substrate/ g VSS .......................... 119
Summary of Phase 2 respirometer results, mL CH4/ g VSS/hr ............... 120
MPN results after 144 hours of incubation, estimated using 5-tube

dilution .............................................................................. 125
Flux summary ...................................................................... 132
Phase 3 CFV comparison ......................................................... 133
Phase 2 comparison of SM AnMBR and MM AnMBR ...................... 133
Design Consideration for AnMBR System ..................................... 144

LIST OF FIGURES

FIGURES
2.1 Stages of methanogenesis .........................................................
2.2 AnMBR schematic ................................................................
3.1 CMD schematic ....................................................................
3.2 AnMBR schematic ................................................................
3.3 CMD/AnMBR schematic .........................................................
3.4 CMD and AnMBR, organic loading rate for COD and VS ..................
3.5 CMD and AnMBR, COD and VS
3.6 AnMBR (from CMD and AnMBR comparison study), permeate rate
and digester TS concentration ...................................................
3.7 CMD and AnMBR, pH ...........................................................
3.8 CMD and AnMBR, methane production .......................................
3.9 AnMBR without solids wasting .................................................
3.10 Combined CMD/AnMBR, organic loading rates for COD and VS. . . . .
3.11 Combined CMD/AnMBR, COD and VS .......................................
3.12 Combined CMD/AnMBR, gas production and pH ...........................
3.13 Combined CMD/AnMBR, volatile acid concentration .......................
3.14 Combined CMD/AnMBR, ﬂux rate and digester TS concentration. . . . .
4.1 Comparison of membrane elements connected in a serial versus parallel
conﬁguration .......................................................................
4.2 General AnMBR layout from which the SM AnMBR and MM AnMBR
are derived .........................................................................
4.3 SMD operated with a single element in a complete mix conﬁguration. . ..
4.4 MED with 13.9 mm pipe surrogate for membrane ...........................

xi

5

13

29

30

31

35

36

37

38

39

41

46

47

48

49

51

57

60

61

62

4.5 CMD system ........................................................................ 63

4.6 MM AnMBR with 7 elements connected in series ........................... 64
4.7 MM AnMBR module illustrating manifold .................................... 65
4.8 MMED with four (4) 3000 mm x 13.9 mm diameter PVC pipes ........... 66
4.9 Phase 1, organic loading rate, VS and COD ................................... 70
4.10 Phase 1, COD and VS concentration ........................................... 71
4.11 Phase 1, pH ......................................................................... 72
4.12 Phase 1, volatile acid concentration ............................................. 73
4.13 Phase 1, methane production .................................................... 74
4.14 Phase 1, SMD ﬂux rate versus digester TS concentration ................... 75
4.15 Phase 2 pH ......................................................................... 80
4.16 Phase 2, volatile acid concentration, MM AnMBR .......................... 81
4.17 Phase 2, organic loading rate, COD and VS .................................. 82
4.18 Phase 2, VS concentration ....................................................... 83
4.19 Phase 2, COD concentration ..................................................... 84
4.20 Phase 2, SM AnMBR ﬂux rate and digester TS concentration ............. 85
4.21 Phase 2, MM AnMBR ﬂux rate and digester TS concentration ............ 86
4.22 Phase 2, methane production .................................................... 87
4.23 Phase 3, organic loading rates for VS and COD .............................. 92
4.24 Phase 3, COD and VS concentration ........................................... 93
4.25 Phase 3, pH ......................................................................... 94
4.26 Phase 3, methane production ..................................................... 95
5.1 Respirometer setup ................................................................ 106

xii

5.2 CMD/AnMBR acetate ............................................................ 109

5.3 CMD/AnMBR, propionate ....................................................... 110
5.4 CMD/AnMBR, formate .......................................................... 1 10
5.5 CMD/AnMBR, acetate + formate ............................................... 111
5.6 CMD/AnMBR glucose consumption per mass VSS .......................... 111
5.7 Phase 1, acetate .................................................................... 115
5.8 Phase 1, propionate ............................................................... 115
5.9 Phase 1, formate ................................................................... 116
5.10 Phase 1, acetate + formate ........................................................ 116
5.11 Phase 1, glucose consumption per mass VSS ................................. 117
5.12 Phase 2, acetate .................................................................... 121
5.13 Phase 2, propionate ............................................................... 121
5.14 Phase 2, formate ................................................................... 122
5.15 Phase 2, acetate + formate ........................................................ 1222
5.16 Phase 2, glucose consumption per mass VSS ................................. 123
5.17 SEM Images, (A) CMD, (B) SM AnMBR and (C) MM AnMBR .......... 124
6.1 TSS concentration versus TS concentration .................................... 137

xiii

AnMBR

CFV

CMD

COD

IMMS

MLTSS

MLVSS

OLR

PFD

SRT

TS

TSS

VS

VSS

VFA

KEY TO ABBREVIATIONS

Anaerobic membrane bioreactor
Cross-ﬂow velocity

Complete mix digester

Chemical oxygen demand

Hydraulic retention time

Integrate manure management system
Mixed liquor total suspended solids
Mixed liquor volatile suspended solids
Organic loading rate

Plug ﬂow digester

Solids retention time

Total solids

Total suspended solids

Volatile solids

Volatile suspended solids

Volatile fatty acids

xiv

Chapter 1
INTRODUCTION

During the last four decades, there has been a steep decline in the number of dairy
farms in the United States. Thirty-ﬁve years ago, there were 124,000 dairy farms. By the
mid-19805, the number decreased to 42,000 and from 1986 to 2006, to 15,500.
Projections suggest that only 8,000 dairies will remain by the year 2018 (Hoard’s
Dairyrnan, June 2008). This decline is due to increased milk production per cow coupled
with increasing herd sizes resulting from industry consolidation.

Land application is the traditional method of animal manure management; however,
increasing animal density, growing regulatory oversight and negative public perceptions
are acting to shift the manure management paradigm. Three speciﬁc characteristics of
animal waste management present signiﬁcant technical challenges. Animal manure has a
high organic strength and can exert a chemical oxygen demand in excess of 75,000 mg/L
(Pain, West et a1. 1984; Demirer and Chen 2005) . Animal manure has a high nutrient
concentration that presents serious water quality concerns because it promotes excessive
algal growth in receiving waters. Excessive algal growth leads to a depression of
dissolved oxygen which can have a deleterious impact on aquatic biota (Vesilind and
Peirce 1983). Dairy manure derived wastewater phosphorus concentrations are typically
in the range of 300—600 mg/L as P (Voge12003; Demirer and Chen 2005) and total

nitrogen concentrations are often in excess 2,000 mg/L (Lo, Bulley et a1. 1983; Lo and
Liao 1 985; Ghaly and Echiegu 1992) . Finally, animal manure contains pathogens and

viruses (Wong, Xagoraraki et al. 2009).

Manure management is a critical component of every animal agriculture operation and
inﬂuences the farm’s management structure. A farmer’s ability to remain economically
competitive is often predicated on effective manure management. Manure is spread
consistent with required crop nutrient uptake. However, in some cases, ﬁelds are nutrient
saturated and the farmer must increase the distance that manure is transported, adding
cost and complexity. As a result, livestock farmers are seeking new methods to
effectively and efﬁciently manage the manure generated by their operations (Bickert
2006)

Many farmers are turning to anaerobic digestion to enhance their manure
management performance (Knight 2003). Anaerobic digestion is a renewable energy
technology that has gained signiﬁcant popular appeal related to the beneﬁts of energy
production, manure treatment cost savings, nutrient conversion, odor and pathogen
control and co-product recovery (Moser, Mattocks et a1. 1998).

The long-term viability of animal agriculture in the United States is largely
dependent on integrated manure management systems that incorporate new technologies
to provide effective treatment of livestock manure (Bickert 2006). One such technology
is the adaptation of the conventional anaerobic digester with a membrane system to form
a process commonly known as an anaerobic membrane bioreactor or AnMBR.

Many studies have evaluated the use of AnMBRs for the treatment of a variety of
waste streams with the vast majority focused on the operational and water quality

outcomes. A few have investigated the impact of the membrane system on microbial
activity and the potential for nutrient management or pathogen and virus removal. None

has Presented a framework for the design of a farm-based AnMBR system. The concept

of coupling an anaerobic process with a cross-ﬂow ultraﬁltration membrane process
holds promise for enhanced organic treatment, nutrient management and pathogen and
virus removal. The objective of this work was to develop a design approach for an
anaerobic membrane bioreactor for the treatment of dairy manure with consideration for

the metabolic impacts associated with the pump/membrane system.

Chapter 2
BACKGROUND

The basic anaerobic processes to convert organic substrate to its most reduced state,
carbon dioxide and methane is ﬁrst presented. A review of digestion technology, with a
speciﬁc emphasis on the difference between the complete mix digester and high rate
digester systems follows. High rate digester systems are distinguished by equipment
conﬁgurations that enable the separation of hydraulic retention time (HRT) from solids
retention time (SRT). The last portion of this chapter presents speciﬁc detail of a high
rate digester system that couples a complete mix digester with an ultraﬁltration (UF)
membrane to produce what is commonly referred to as an anaerobic membrane bioreactor

(AnMBR). Chapters 4-6 provide further literature speciﬁc to the content.

2.1 Anaerobic digestion process
Anaerobic digestion is a multi-faceted and complex process. No one organism is capable
of completely reducing carbonaceous matter to methane. A four step process is required
to complete this transformation (Bryant 1979; Speece 1996). Complex organic matter
such as proteins, carbohydrates and lipids are ﬁrst hydrolyzed into less complex
compounds such as sugars, amino acids and peptides and these are further fermented to
fatty acids by acidogenesis. Long-chained fatty acids (> C2) are converted to acetate, H2
and CO; by acetogenesis. Lastly, acetate and H2 are converted to methane and carbon
dioxide by methanogenesis (McInemey 1979; McCarty and Smith 1986; Samsoon,
Loewenthal et al. 1987; Oremland 1988; Speece 1996). Figure 2.1 presents a graphical
representation of the process ﬂow and the following subsections discuss hydrolysis,

acldogenesis, acetogenesis and methanogenesis in more detail.

4

 

20°/ , 50/
° Complex Organics - 0

(Proteins, lipids, carbohydrates)

 

 

 

 

 

Hydrolysis

 

 

350/ . . 100/
° Simple Organrc Compounds °

(sugars, amino acids, peptides)

 

 

 

 

 

Acidogenesis

 

 

17% Long-Chain Fatty Acids 13%
(propionate, butyrate, etc)

 

 

 

 

 

Acetogenesis

 

 

 

 

 

 

 

 

 

 

Acetate H2, CO2

 

 

 

72% 28%

 

 

 

CH4, co2

 

 

Methanogenesis

Figure 2.1 Stages of methanogenesis

McCarty and Smith reprinted with permission from Environ. Sci. Technol. Copyright 1986, American
Chemical Society

2.2 Hydrolysis

Cellulose and hemicellulose compose a signiﬁcant portion of the digestable fraction of
dairy manure (Amon 2007). Hydrolysis is catalyzed by a variety of different bacteria
secreted enzymes such as proteases, lipases and cellulases. In anaerobic environments,

the initial enzymatic attack of cellulose is dependent on the activity of a relatively select
group of microorganisms (Chayovan, Gerrish et a1. 1988). Noike et al. (1985) found that
the percentage removal of cellulose fed to a reactor apparatus increased as the solids

retention time (SRT) increased. Only 2% of the cellulose fed was removed at an SRT of

 

 

1.94 days, while 54% was removed at an SRT of 13.7 days. Based on these results, it
was concluded that cellulose is slowly broken down in the hydrolysis phase of anaerobic
digestion.

The hydrolysis step is typically rate controlling when the substrate contains a high
concentration of particulate matter (Eastman and Ferguson 1981; Vavilin, Rytov et al.
1996; Miron, Zeeman et al. 2000; Rittmann and McCarty 2001; Mahmoud, Zeeman et al.
2004; Zhang, He et al. 2007). Veeken et al. (2000) illustrated that hydrolysis proceeds at
pH values between 5.0-7.0; however, they illustrated that lowering the pH below neutral

did not provide a hydrolysis rate advantage.

2.3 Acidogenesis

Acidogens ferment the less complex compounds to acetate, formate or to other volatile
fatty acids (VF A) and H2 (Kaspar and Wuhrmann 1978; Boone and Bryant 1980;
McCarty and Smith 1986). The optimum pH for acidogenic bacteria is 5.2 — 6.5 and they

exhibit doubling times of approximately 2 days (Demirer and Chen 2004).

2.4 Acetogenesis
Acetogenic bacteria represent a complex of species involved in B-oxidation of fatty
acids of even numbered carbons to acetate and H2, conversion of fatty acids of odd-
numbered carbons to acetate, propionate and H2 and decarboxylation of propionate to
acetate, CO2 and H2 (Boone and Bryant 1980). As an example, according to Boone and

Bryan (1980), propionate is fermented per Equation 2.1.

 

Propionate + 3H2O —» Acetate + Hco3‘ + H+ +3H2, AGO = +76.1 kJ /reaction (2.1)

Propionate conversion is endergonic under standard conditions and only proceeds under

low concentrations of H2 below 10'4 atmospheres, while H2 conversion to methane is
only thermodynamically possible at concentrations above 10'6 atmospheres (Speece

1996). The H2 concentration is typically kept low by hydro gentrophic methanogens

working in partnership with acetogenic propionate degrading fermenters (Kaspar and
Wuhrmann 1978; Boone and Bryant 1980; McCarty and Smith 1986). This relationship
is referred to as a syntrophic interaction. The term syntrophic was coined to describe the
close cooperation of fatty acid-oxidizing fermenting bacteria with hydro gentrophic
methanogens (McInemey 1979; Boone and Bryant 1980). This process is also known as
“interspecies hydrogen Transfer (Ianotti 1973) and, in the absence of this syntrophic
relationship, fatty acids accumulate. Kasper and Wuhrman (1978) reported that
propionate-degrading systems were saturated to only 10-15% of their capacity. This
suggests that in a well operating digester system, there should not be a build-up of
propionate.

Ideal conditions for acetogenic bacteria are quite different than those favored
by hydrolysis and acidogenesis and more closely mirror the conditions under which
methanogens thrive. The optimum pH for acetogenic bacteria is 6.6 — 7.6 and they

exhibit a minimum doubling time of 3.6 days (Speece 1996).

2.5 Methanogenesis
Methanogens form a unique group of Archae capable of metabolizing a limited

number of simple organic compounds, primarily acetate, H2 and CO2 to methane.

Acetate and H2 are the two immediate precursors of CH4 (Yao and Conrad 2001). There

are two primary methane forming paths that are relevant to a manure-based anaerobic
digestion process, methanogenic respiration and acetate fermentation (McCarty and
Smith 1986), and each is discussed below.

For methanogenic respiration, hydrogen acts as the electron donor and CO2 acts as the

electron acceptor as illustrated in Equation 2.2.

2H2 (g) + 1/2 CO2 (g) + H+ (aq) -—» 1/2 CH4 (g) + H20 (1) AG° = -65.37 (2.2)

Kaspar and Wuhrmann (1978) reported that hydrogen removal by hydrogen consuming
methanogens (or hydrogentrophic methano gens) was less than 1% of the maximum
possible rate, suggesting a large unused capacity able to buffer the partial pressure of
dissolved hydrogen in the system. Approximately 30% of methane produced in the
anaerobic digestion process results from methanogenic respiration of H2 and CO2 (Smith

and Mah 1966).

Methanogenesis by acetate fermentation forms CH4 and CO2 per Equation 2.3.

CH3C00'(aq) +H+ (aq) -—» CH4 (g) + €02 (g) AG° = -35.83 (2.3)

There are only two known genera of methanogens capable of degrading acetate including
the species Methanosarcina, which is also capable of utilizing H2/CO2 and
Methanosaeta, which is only able to convert acetate to methane (Harper 1985). Despite
a limited number of known organisms capable of degrading it, acetate fermentation
accounts for 70% of the total methane produced (Smith and Mah 1966).

Speece (1996) indicated that the generally accepted pH range for methanogenic
bacteria is 6.5-8.2. Rittmann and McCarty (2001) suggest a similar range of 6.6 to 7.6.

Doubling times for hydrogen consuming methanogens have been reported between 6-
24 hours (Archer and Powell 1985; Rittmann and McCarty 2001), a rate that is
considerably greater than that for the acetate consuming methanogens which exhibit
reported doubling times ranging between 2 and 9 days, as summarized by Harper and
Pohland (1985). Due to the very slow kinetics associated with the methanogenic process,
adequate digester retention of the methanogenic consortia is critical to successful

anaerobic treatment.

2.6 High Rate Anaerobic Digestion
A signiﬁcant advantage of the complete mix digester is the simplicity of design and
operation. There are minimal internal components required. Submersible or external
mounted mixers provide satisfactory agitation with minimal power consumption. Typical
organic loading rates for complete mix systems range from 1-4 g COD/L/day (Rittmarm
and McCarty 2001). Further, complete mix systems are capable of handling total solids
loadings consistent with those found in most animal agricultural operations with an

inﬂuent total solid concentration of 4-10% (Hills and Roberts 1981; Lo, Liao et a1. 1984;

 

Pain, West et al. 1984; Oliver, Pain et a1. 1986; Chapman, Phillips et a1. 1990; Moller,
Sommer et al. 2004). ,

Hydraulic retention times in standard complete mix digesters are commonly in the
range of 10 to 20 days. This is considerably greater than the minimum detention time of
4 days required for acetate using methanogens (Rittmann and McCarty 2001). Dague et
al. (1970) reported that the critical solids retention time for anaerobic waste treatment
systems was 10 days and that virtually no waste stabilization occurred at solids retention
times of 3 days or less. Due to the slow growing nature of methanogens, long retention
times are necessary, without which, the anaerobic digestion process will come to a halt.

A negative attribute of the complete mix system is the biomass retention time (or
SRT) is equal to the HRT. As a result, the active biomass concentration available to
convert substrate entering the digester is limited by its growth rate within the operating
HRT. Theoretically, if the SRT is de-coupled from the HRT, a higher concentration of
active biomass is available for treatment and a greater degree of substrate conversion
achievable.

The development of the anaerobic contact process (Schroepfer, Fullen et al. 1955)
resulted from an effort to enhance digester performance by segregating SRT from HRT.
By adding a settling tank and recycling the biomass back to the digester tank, separation
of HRT from SRT resulted. This process is analogous to the aerobic activated sludge
process. Typical organic loading rates associated with the contact process range between

2 — 8 g COD/L/day (Schroepfer, Fullen et al. 1955; Hamdi and Garcia 1991; Hickey

7.007). However, entrained biogas in the anaerobic efﬂuent leads to poor settling

10

characteristics and washout of biomass, degrading efﬂuent quality (Hawkes, Donnelly et
al. 1995).

In the late 19603, the anaerobic ﬁlter was developed by Young and McCarty (1969).
This process originally used a rock medium for attaching the biosolids, which was
eventually replaced with plastic media. Design loadings are often in the 6 to 16 g
COD/L/day range (Hawkes, Donnelly et al. 1995; Powers, Wilkie et al. 1997; Rittmann
and McCarty 2001). Anaerobic ﬁlter systems, also known as ﬁxed-ﬁlm systems, are
particularly well suited for the treatment of soluble organic waste streams. Powers et al.
(1997) employed a ﬁxed ﬁlm process for treating dilute dairy manure resulting from a
ﬂush manure collection system that operated at HRTs of 1.5 and 2.3 days and had
approximate VS and TS reductions equivalent to a CSTR operated at a HRT of 10 days.

Lettinga et al. (1980) introduced a novel mechanism for segregating HRT from
SRT through the use of a process known as the upﬂow anaerobic sludge blanket reactor
(UASB). In a UASB “granules” naturally form after several weeks of digester operation.
These compact spherical particles are about 0.5 mm in diameter and consist primarily of
a dense mixed population of microorganisms necessary to carry out anaerobic digestion
(Rittmann and McCarty 2001). The UASB process is capable of managing organic
loading rates as high as 16 g COD/L/day (Lettinga, Vanvelse‘n et al. 1980). Like the
anaerobic ﬁlter process, the UASB is particularly well suited for waste streams with high
concentrations of soluble COD, but have little tolerance for suspended solids (Hickey
2007). However, Castrillon et a1. (2002) used a lab-scale UASB to treat cattle manure
and operated this system continuously for approximately one year. During this period,

the UASB operated at organic loading rates were between 1.67 and 5.06 g/L/day and

11

inﬂuent TS concentrations between 22.38 and 39.94%. The total sludge accumulation in
the UASB was controlled through wasting.

The anaerobic sequential batch reactor (ASBR) was developed at Iowa State
University in the late 19903. Operation is similar to the contact process with the
exception that the solids are separated directly in the reactor rather than in an external
clariﬁer. The operation of an ASBR involves four distinct stages: feed, reaction, settling
and decanting. The purpose of the settling and decanting stage is to allow the biomass to
settle and remain in the tank while removing the digested efﬂuent such that HRT is
decoupled from SRT. Biomass granulation has been reported to occur with this process
producing a highly active granular mass with good settling properties (Zhang, Yin et al.
1997). This design has been demonstrated for treating swine waste at organic loading
rates of 1.6 to 4.5 g VS/L/day with VS reduction ranging from 55 to 61% and BOD
reduction of 81 to 86% (Zhang, Yin et al. 1997). With dairy manure, VS reductions of
26.1 to 44.2% have been reported at organic loading rates of 2 to 6 g VS/L/day (Dugba
and Zhang 1999). Considering the typical characteristics of swine and dairy manure, this
suggests the ASBR is capable of treating waste streams containing relatively high
concentrations of suspended solids. Table 2.1 presents a summary of the common high
rate digester systems that use various mechanisms to segregate SRT from HRT.
Irrespective of conﬁguration, digesters are typically operated in the mesophilic range
with an optimum temperature around 35°C or the thermophilic range with an optimum

temperature of 55-60°C.

12

2.7 Anaerobic Reactors Coupled with External UF Membranes (AnMBR)

All of the high rate digester systems described above rely on settling of biomass or
adhesion of biomass to media in the digester tank. As a result, each of these systems, to
varying degrees, allow biomass to exit the system with the treated efﬂuent. Coupling an
anaerobic reactor with an external UF membrane to create an AnMBR is another
adaption of traditional digestion technology that seeks to decouple SRT ﬁom HRT to
improve reactor substrate conversion efﬁciency. Figure 2.2 is a schematic of the

AnMBR and illustrates the placement of the membrane external to the digester tank.

‘ l

i Digester ~ Permeate

1 Tank 1

, l l

E l l

' ‘ I * UF

.__ . ‘.. _______ . Membrane
_ _ _ _ Conssmrate

Figure 2.2 AnMBR schematic

A full-scale system will be comprised of multiple membranes, placed in series, parallel or
a combination with the placement of the membranes referenced as the membrane
conﬁguration. Biomass from the anaerobic reactor is pumped through the UP membrane

WhiCh provides a physical barrier to prevent wash-out of biomass. Clariﬁed efﬂuent, or

13

permeate, which is devoid of solids, is removed from the system and concentrated
biomass is returned to the digester tank. The biomass concentration in the digester is
described by the mixed liquor volatile suspended solids (MLVSS) concentration and is
deﬁned based on design conditions. Other parameters commonly used to describe the
biomass concentration in the digester include mixed liquor total solids (MLTSS), total
solids (TS) and volatile solids (VS). The rate that biomass is pumped through the
membrane is known as the cross-ﬂow velocity (CF V) and is determined based on system
design.

Much work has been conducted related to the advantages of AnMBRs. The ﬁrst
known research took place in the United States in the mid to late 1970s (Grethlein 1978;
Sutton, Berube et a1. 2004) and employed a membrane ﬁlter coupled to a domestic septic
tank system. In the early 19803, Epstein and Korchin et al. (1981) and Choate,
Houldsworth et al. (1983) conducted research with a combined anaerobic reactor and UP
membrane in an industrial wastewater treatment capacity. This was followed by the
development of the anaerobic digestion ultraﬁltration (ADUF) process (Ross, Barnard et
al. 1992; Strohwald and Ross 1992; Ross 1994) which utilized an unsupported tubular UF
membrane and organic loading rates of 10 g COD/L/day and greater were reported, up to
four times that of conventional processes at reduced volume and capital requirements.
The ADUF work was followed by the development of the cross-ﬂow ultraﬁltration
membrane anaerobic reactor (CUMAR) system (Anderson, Kasapgil et al. 1994; Ince,

Anderson et al. 1995; Anderson, Kasapgil et al. 1996; Ince 1998; Ince, Ince et al. 2000;
Ince, Ince et a1. 2001). In a series of publications, a CUMAR system was evaluated for

treating brewery wastewater with COD removal efﬁciencies no lower than 97% while

14

operating at organic loading rates as high as 28.5 kg COD/m3/day (Anderson, Kasapgil et
al. 1996; Ince, Ince et al. 2000; Ince, Ince et al. 2001). Fakhru’l-Razi (1994) reported
operating an AnMBR to treat high strength industrial wastewater at an organic loading
rate as high as 19.7 g COD/L/day and achieving COD removal of greater than 96%.
Cadi, Huyard et al.(1994) achieved an organic loading rate of 24 g/L/day with a COD
removal yield of 87% using starch as the sole carbon source for the study. Fuchs, Binder
et al. (2003) achieved COD removal rates of 90% for an artiﬁcial wastewater (loading
rate of 20 g COD/L/d), sauerkraut brine (8 g COD/L/d) and an animal slaughterhouse

wastewater (6-8 g COD/L/d).

Much research supports the AnMBR as an effective process capable of producing
excellent efﬂuent quality while providing a very high level of organic conversion.
However, other research suggests microbial inhibition due to the shearing impacts
associated with turbulent transport of biomass through the membrane system or other
high shear applications (Brockmann 1995; Brockmann and Seyﬁied 1996; Choo and Lee
1996; Brockmann and Seyfried 1997; Ghyoot and Verstraete 1997; He, Xu et al. 2005;
Padmasiri, Zhang et al. 2007). Brockman and Seyfried conducted methane potential
testing on the biomass from an AnMBR and demonstrated a 50% reduction of microbial
activity when the entire contents of the reactor were pumped through the membrane 20
times per day. They theorized that this reduction was due to an interruption in syntrophic
activity resulting in an accumulation of VFA. Ghyoot and Verstraete (1997) subjected

biomass to displacement through the membrane system (treated biomass) of an AnMBR
and compared its activity to that of a control sample (untreated). The treated biomass

exhibited a lower biogas production potential and it was concluded that the mechanical

15

stress of the AnMBR damaged the interaction between the different species in the
anaerobic consortia. Padmasiri et al. (2007) also reported a reduction in microbial
activity with an anaerobic membrane bioreactor used for the treatment of swine waste.
The deterioration in reactor performance was manifested by increased VF A, in excess of
the metabolic capacity of the methanogens, and thought to be a direct result of an
increase in the hydrolysis rate due to the high shear environment of the AnMBR. Choo
and Lee (1996) reported a dramatic reduction in the reactor biomass concentration while
operating an AnMBR (3,000 mg/L to 300 mg/L as MLVSS) . A signiﬁcant amount of
biomass was observed attached to the membrane surface during the experimental run and
it was theorized that the microbial cells moved from the reactor to the membrane surface

to avoid the shear stress of the pump.

Evaluating a similar phenomenon, Stroot, McMahon et al. (2001) evaluated the impact
of various mixing conditions on the digestion of municipal solid waste and found that
vigorous and continuous mixing had a detrimental impact on microbial activity and
caused a disruption in the syntrophic interaction or an increase in hydrolysis leading to an
excess of fermentation intermediates (in excess of the methanogens capacity to process).
However, in a similar manner, Hoffman, Garcia et al. (2008) evaluated the effect of
mixing shear on performance and microbial ecology of continuously stirred anaerobic
digesters treating dairy manure and concluded that at four different mixing intensities
(50,250, 500 and 1,500 RPM), with the exception of at startup, there was no effect on the

biogas production rates and yields at steady-state conditions.

Table 2.2 provides a detailed summary of the AnMBRs described above as well as

other AnMBR work of interest to this research. The heading “cycle time”, a concept

16

introduced by Seyfried and Brockmann (1995; 1996; 1997), is used as a metric to
compare various levels of AnMBR pump circulation rates. Speciﬁcally, cycle time is
deﬁned as the period of time required for a discreet particle to travel from the digester
tank, through the pump/membrane and return to its initial starting location in the tank.
Cycle time is typically presented as number of cycles completed in a 24 hour period. For
example, a cycle time of 10 indicates the biomass has, on average, been completely
circulated through the pump, membrane and digester tank 10 times in a 24 hour period.
In a number of cases, cycle time (or necessary data to calculate cycle time) was not
provided and often, the research indicated that excess permeate was returned to the
digester tank. This is a typical situation for a laboratory setup because the membranes
used are often industrial size units; therefore, the biomass pumping rate is high in relation
to the digester tank size and results in excess permeate production. Where permeate is

returned to the digester tank, it is likely that the system is operating at a high cycle time.

2.8 Advantages of AnMBR for Nutrient and Pathogen Management
Recent surveys suggest that for many Midwestern dairy farms, phosphorus inputs are

greater than phosphorus outputs. This leads to a buildup of phosphorus in the soil and the

potential for phosphorus runoff to surface water exists (Beede 2003). Understanding

both the fate and chemical composition of nutrients existing in an anaerobic digester is of

great interest and importance, particularly to dairy farmers in the Midwest. Converse
and Karthikeyan (2002) conducted a series of settling tests to evaluate both ﬂushed dairy

manure and efﬂuent from a screw press. After long-term settling (49 days),

approximately 75-80% of the total phosphorus was concentrated in the bottom 25% of

17

the test vessel. Inglis et al. (Inglis 2007) evaluated the phosphorus content of a plug-ﬂow
digester during a cleanout operation. The phosphorus concentration of the supernatant
was 465 mg/kg, the crust phosphorus level was 686 mg/kg and the bottom phosphorus
concentration was 874 mg/kg. Qureshi Lo et al. (2006) evaluated the nutrient recovery
balance for a sequencing batch reactor (SBR) treating dairy manure. The phosphorus
remaining in the settled fraction of the SBR ranged between 45% - 59%. Masse and
Droste (2000) evaluated the phosphorus fate for a psychrophilic anaerobic sequencing
batch reactor and, after two cycles, the bioreactors retained on average, 25.5% of the total
phosphorus. These ﬁndings suggest that phosphorus tends to partition with the solid
fraction in an animal manure digester.

The literature contains a little detail regarding the impact of the AnMBR system on
the removal of nutrients. Ghyoot and Verstraete (1997) reported 82% removal of total
phosphorus, 56% ortho—phosphorus, 66% organic nitrogen and 32% ammonia nitrogen in
the AnMBR permeate. Vogel (2003), using a thermophilic AnMBR to treat dairy
manure, reported an inﬂuent total phosphorus concentration of 478 mg/L and a permeate
concentration of 17 mg/L. Wong et a1. (Wong, Xagoraraki et a1. 2009), using an
AnMBR for the treatment of dairy manure found a phosphorus reduction of 96%, a TKN
reduction of 31% and no ammonia reduction. Ammonia is soluble and therefore would
be expected to pass through the membrane, whereas, organic nitrogen could partition
with the solid fraction and would be excluded. The membrane in the AnMBR acts as a

very efﬁcient ﬁlter precluding solid particles larger than 0.03 pm and therefore, high
removal efﬁciency should be anticipated for constituents that tend to partition with the

solid fraction such as phosphorus and organic nitrogen.

18

Following the same line of reasoning, pathogens and viruses should be excluded based
on the membrane pore size. Cicek, Franco et al. (1998) reported operation of an aerobic
MBR treating simulated municipal wastewater that completely excluded viruses from the
MBR permeate. Grethlein (1978) coupled a membrane with a septic tank system and

reported treated efﬂuent that contained no E. coli. Total coliforrns were removed with an
efﬁciency greater than 99% from liquid pig manure (Fugere, Mameri et al. 2005). Vogel
et a1. (2003) operated a thermophilic AnMBR for the treatment of liquid dairy manure
and reported a 5 log removal of fecal coliforrn for both ﬁltered and settled (no
membrane) efﬂuent. Work conducted by Wong et al. (2008) evaluated the removal of

pathogen and virus indicators in the efﬂuent of the AnMBR used for the present research

and results are presented in Chapter 3, Section 3.4.

2.9 Objective

Much of the published research related to coupling anaerobic digesters with
membranes has focused on water quality outcomes. Little is presented related to speciﬁc
criteria needed to design an AnMBR. Further, there is also uncertainty in the literature

n‘v‘géll‘ding the impact of the AnMBR system on methane productivity.

The purpose of this research work is to develop a design approach for an AnMBR
treating liquid dairy manure. Cycle time is thought to exert signiﬁcant inﬂuence over the
methane productivity of an AnMBR system and, understanding this impact, is central in
the effort to deﬁne its role in the AnMBR design. Much effort is dedicated in this

research to evaluating the effect of cycle time on biogas production and exploring the

19

potential mechanism(s) inﬂuencing the microbial biota under various digester

conﬁgurations.

2. l 0 Research Outline

The research builds ﬁ'om initial comparison experiments with an AnMBR and CMD to
testing of speciﬁc cycle time conditions that incorporate a qualitative understanding of
the associated microbiology, followed by an integration of these ﬁndings into design

considerations for an AnMBR treating liquid dairy manure. A brief summary of the

content of each chapter is presented below.

Chapter 3 — Acquire a general applied knowledge of AnMBR performance and operating

characteristics necessary to provide basis for future work.

Chapter 4 — Evaluate methane production at cycle times consistent with anticipated full-

scale design and incorporate important design parameters fundamental to the deﬁnition of

CyCle time including digester VS concentration, CF V and membrane conﬁguration.

Chapter 5 — Use activity measurements to characterize and evaluate the microbial

Pathways associated with the digester conﬁgurations presented in Chapters 3 and 4 and

explain the affect of cycle time at a metabolic level.

20

Chapter 6 — Formulate AnMBR design considerations based on the ﬁndings of Chapters

3, 4 and 5.

Chapter 7 - Conclusions and Future work

21

 

 

 

 

 

 

 

  

88m .3 a 8H .mﬁﬁv e§m> w 4 l N caaooaaeomez Sea: :32 ”£95..an
poem .3 a
cam .80 meme .3 Ho ﬂﬁmaom 6326 «3:33
mowﬁ .3 Ho :ow_o>§> .mmﬁtoqv mm - 2. 8:6me omen—m oEomocdﬁoE macaw 95952:“ 32.3:
zoom €822 Ea gang 2. 2 secs BEE 23.... 95235.
A88 385 :SN .3
Ho 85 .85 Jog «6ch Ba €235 w - m @538 Lmﬁoaxm 88on
.5500 u
895:0qu :00 .33— uﬁega 55.33% mums—2m no “5502
2.5th Egan.

 

L

 

 

Sana...— a

 

 

22

 

 

#83 Be 85.5 m Co 293 mﬁeoom co>o can 2vo oMEo><
8% Banana Sow wean—820

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N
_

 

 

 

£33822 awe—2 82.522082 3 232082 .8295 9 .530 5 no u H.
“census—man Euﬁooﬂmma 32:829me
9E8; 328m 358m <7 395550 web 955
ceméi oomég of <2 ocvéomu «.5 6.530.. 53.5 0
<2 w.wm <2 <2 <2 .55
m.o-m.o wad mmd <2 on awn—v AH:
NUEEEV mm
2-2 <2 4 <2 cases 838 :58: .2: SE
we
w
2 he 3 8 S bag—Eco u «.5
3.3 ca 5 ea 3-3 .x. £285.38“. 900
80.3
om 10m
<7 33% 8de 813 Nd v ARE .QOU too..—
<2 <2 <2 <2 <2 .mh
<2 <2 25.: <2 <2 .m>
coo. m <2 <2 ooode coo m .mwﬂ .mmAE
<2 cm H .nm <2 <2 <2 éwm ”mmxﬁmz
sexua—
ooodv coca: <2 <2 GEE—03 Geode £83 3325 3.3232
<2 <2 No Nd ﬂown—«bow _.o . nEo—z
om o2 <2 5 es A 63m Sunni
oqtm Sabﬁosam owes:
omnoﬁoﬁwsﬂm DEE @88me
8mm? boaocm 833 28305 nogm Eouﬁ bioﬁocoua< 8.2.3.6
o.~-n._ <2 m.~ <2 «.méd a): . 322, Be: 820
mmm <2 Home E 8 mmooxm Sumo E 2 mmooxm 378— «Emu—u o .oEE. 29 U
as: as: daz $2... a: 232 0322.5»
2: 23:22 _~&._.E§«e .1 e 1.1.5.238 a e :25 .225 2: .829 .2255

“So: a 3.3933 235.55 9323:: a: «Ea—5 Na 03:

23

“.83 .2. won... m 20 0.0.3 wig»... 5.5 on... £23 owio><

 

 

N

8% @858... Sou 285.5030.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

4% 0.5.3803. £2.90on ﬂamoE a .8530 o .88 .6 .3 u H.
3.5.22. <2 Eoﬁooa.%€ 03:8... <2 <2 2.3 9......
<2 8. <2 09. OS. «A... .958». .530 0
<2 <2 <2 <2 <2 .55
o <2 N... Om N6 awn: mum:
<2 <2 <2 2m 3-. .w 25:5 .3... .3...
Q. o.oA n.m~ .. ad anew—ECU w 240
8 <2 8 ma 8 .x. 52.25.... now
<2 <2 ooodo - 80.0.. 08.5. 806 .
<7. <2 <7 <2 <2
<7 <2 <2 <2 <2
<7 <2 <2 25% So...
<2 <2 08.3 I coo... <2 <2
coo. .N <2 ooodo~ <2 <2 :31... 83.5 E .03 h.w...o\...e—>.
<2 ..o <2 <2 .. .o .36 P... 2.55.55.
c coo... om. ooQN ooo.ooo.~ A an? .88“:—
895..."...
25.5... 033m .895 9.2.085 52.3883 2.9265 8.33 28:5... 053 923882-383. oaabmnnm
... o.~-m.. «him QN o. a): . .922. 3c... 99.0
.83.. cm . s G <2 .83»... o. wmooxm. m. 2:38.93 6...: g.umU
9%. BE»;
.2... gun—v.99;—
33. 325m
99“." .1. .o E... 53.5.3.5 923 8.8. Aug—
n u a .
Erasnga .mao. 5.25.8.8 .a .o 8... 8.... .u .9 9.2:; .23: ._.. .9 2.5.5 £3: 52:25.

B====QUN.~ ﬁsh

24

 

.283 Hum 883 m .20 £93 92¢qu .55 0:3 293 uwm~o><

~

8% 08:0on Sou vogue—a0.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

23.332 £3882 0:59.82 0229::qu uzﬁmonuoﬁ :23: o 238 =. .3 o a.
<2 <2 <2 <2 <2 0&5 mash
n: 00m <2 <2 <2 am: .9539. 3:20 0
<2 <2 Wok. 1 N6 0m <2 Elm
<2 0.2 Q: I 5; MN <2 what Arm”
<2 mm - w m: 8 - 9. mm - om ENE: .32 SE
0.5 0.m <2 Om <2 NaEdA—OU M “AC
<2 <2 om av <2 .x. icﬂuzbmue m>
mo 00 <2 <2 <2 o\.. 5235.580 GOD
<2 89% <2 08. H <2 Alum. doc :82
<2 <2 000.02 <2 000.0w I 000.0v me .ma.
<2 <2 <2 <2 <7 mum m>
<2 <2 <2 <2 <7 mum .ww‘us
<2 <2 <2 <2 <7 mum .mm>x=2
ﬂog—«u
000.000.N <2 <2 <2 000.0.v 53:0 EH03 23:00—32
<2 <2 _.0 N0 <2 Jim 9.3 2:22:52
<2 000.m <2 03m 000.m A .eﬁm .53qu
Eon—bu
533233 8:05 wimmoooaéﬁﬂz mom—cm 236252 8:58 Eun— oBaaE ma Satan—am
<2 3 <2 mm 2 I 2 as . €22 2.: 280
<2 <2 <2 NNv We I QN antic—aha .oEi. 2920
as: ._a 3 @8— ~22 “5E3 @122 .1 .0 was ._a E
a . a .
«35:3. 2:.»an 13 1.35.5 835 .5: 62.550 55:25 350:5 326.52 .30: :3 28052.5

35.5 2 2...:

25

CHAPTER 3
UTILITY OF THE ANAEROBIC MEMBRANE BIOREACTOR
3.1 Introduction
An initial plan was developed to assess the AnMBR operating characteristics and

conduct a side-by—side comparison with a CMD (CMD and AnMBR comparison study),
including gas production. Based on the results of the CMD and AnMBR comparison
study, a second experiment was conducted in which the AnMBR was coupled with the
CMD such that the efﬂuent of the CMD was the inﬂuent to the AnMBR (CMD/AnMBR
study). This CMD/AnMBR study emphasized understanding the impact of cycle time on
biogas production, enhancing the flux rate of the AnMBR system (including long-term

fouling/cleaning impacts) and evaluating the fate of nitrogen and phosphorus for the

CMD/AnMBR system.

3.2 Materials and Methods
3.2.1 Analytical Methods
Total solids (TS) and volatile solids (VS) were measured according to AWWA

Standard Methods 2540 B and 2540 B respectively. Chemical oxygen demand (COD)
was evaluated using Hach (Loveland, Colorado) high range COD test kits. Total
Kl eldab] nitrogen (TKN), ammonium nitrogen and total phosphate (TP) were conducted
according to “Recommended Methods of Manure Analysis”, Bulletin A3769, University
of Wisconsin Extension (2003). Methane was measured by gas chromatography using a

SR1 3 1 0C equipped with a high temperature TCD and an AllTech Porapak Q 80/100

26

column (6’ x 1/8” x 0.85 stainless steel) column (SR1, Torrance, CA). Volatile acid, total
alkalinity and bicarbonate alkalinity were measured using a titration method adopted
ﬁ'om O’Brien and Donlan (1977), procedure is detailed in Appendix B.

3.2.2 Substrate

The substrate used for the studies was collected on a weekly basis from an operating

3 200 cow dairy that uses sand to bed their cows. At this dairy, the manure is scraped

from the alleys into reception pits and then processed through sand-manure separators
(McLanahan Corporation, Hollidaysburg, PA) for primary sand and grit removal.
Samples were collected at the discharge of the sand-manure separator. The manure was
collected one time per week and stored in 5 gallon carboys at room temperature. During
warm weather, the 5 gallon carboys were stored in a freezer and defrosted as needed.
Typical manure characteristics at collection for the CMD and AnMBR comparison study
are presented in Table 3.1. Table 3.2 presents the typical substrate characteristics for the

CMD/AnMBR study.

27

Table 3.1 Substrate characteristics for initial CMD and AnMBR comparison

 

 

 

 

 

Parameter Value Standard Deviation
COD, mg/L 53,700 11,900
TS, % 4.7 1.4
VS, % 3.4 0.6
pH 7.30 0.07

 

 

 

 

 

Table 3.2 Substrate characteristics for Combined CMD/AnMBR

 

 

 

 

 

 

 

 

 

Parameter Value Standard Deviation
COD, mg/L 35,700 p 10,500
Total solids, % 3.7 0.8
Volatile solids, % 2.4 0.6
pH 7.04 0.22
3.2.3 CMD System

Figure 3.1 provides a schematic of the CMD system. The CMD system consisted of a
175 cm tall x 30.5 cm diameter section of schedule 40 PVC with ﬂanged ends, a working
Volume of 105 liters and approximately 30 cm of headspace for gas collection. Mixing
was achieved with a 1”x1.5” centrifugal pump (AMT, Inc. Mansﬁeld, OH), operated 6
times per day for 5 minutes. The circulation rate was approximately 45 LPM. A 0.64 cm
diameter tube directed the biogas in the digester headspace to a wet tip meter (Wet Tip
Meter Company, Nashville, TN). The digester was heated using an external heat blanket
and thermostat (BriskHeat, Columbus, OH). Gas samples for GC analysis were collected
between the digester and the wet tip meter via a luer-style, 3—way valve. Digested
manure was removed from the system once per day based on mass and the same amount

0f fresh manure was added to the system once per day from a 100-L mix tank. The mix

28

 

tank was agitated with a submersible pump for approximately 5 minutes prior to feeding

the digester systems.

3 -2.4 AnMBR System

A schematic of the AnMBR is presented in Figure 3.2. The digester portion of the
AnMBR was constructed identical to the CMD. The membrane was a 0.03 micron, 14.4
mm diameter, 0.079 m2 PVDF tubular product ( X-F low, Inc., Netherlands) and was
operated in a cross—ﬂow conﬁguration using a centrifugal pump (AMT, 1.5” self-priming

centrifugal) to generate a circulation rate of approximately 33 L/min (cross ﬂow velocity

= 3.4 m/s).

3.2.5 Combined CMD/AnMBR

A schematic of the combined system is presented in Figure 3.3 and consisted of the
AnMBR and CMD described in Sections 3.2.3 and 3.2.4 respectively. The AnMBR
circulation pump was energized by a timer to achieve a speciﬁc cycle time and permeate
was removed from the digester during these periods of pump operation. The difference
between the mass of substrate fed to the digester (based on design HRT) and the mass of

permeate removed from the digester was wasted directly from the digester (See Figure
3 .3).

29

2.2.5.2 920 1“.

3352.8 :20 ﬁn charm

.1 1111 1.J
E T
Sumowa _
_, ..1..-...-. 36 .
$.23 -. 1 1| 1
@282 .
532
.5
so?

9 95

Q20 8 28:05 . 2 . 8:62 wEEEam

2282
ESQ
3388-28
as. 8%

30

era—55¢ mmzn< «d Baum-.—

Bmoccom mm2=<ﬂ ; warms? mow—om . Mm22< 9 25:05 _ ”35oz wEn—Emm

28m.
1-1 128m
an . - . - .
53: >65 ouocwﬂz. . . . * How—mowwa H 3.1V 1.1....9 . .
I.-- 111-11me184 2.1..- 1.9:. . -1. Togas .. 2
.1... 1.2.1.2...“ .12.
5530.20 830% 1. + woman—“Mum
88:50:00 ” ”BE
.1 1 .- 1 1-1 ; E.
.11 1.1 1 W . 23m1 83
. . 380
.. 0:89:82 .. 7
1.5
3:53.
58:2

 

mod

31

2352.3. $255.20 3. 2%,...

maﬁa; £50m U 1-.. EozEm NEE—E - 63:5 020 - QED BEBE: . . ”mHEOQ mam—macaw
- _ -1 ; ‘ 111-11 ., ._ 1.20
a 1
mm: < 1 1 1. 1 _ 28m
U 1...~1 1.1 . -- 11
1?. 1H11tH1> 1.. u _ .1 1 M _ . W-1llf.1/ .
1212 1 . 1f -11 1 _ “EMF 1 .1 - 1.1wmw1z-
30: omuwcwwz . 1.1-. :21. How—amow_a jg . . HUMWAMWQ _ human—
11 11---. 111111.11. .- mmeac< m . :1m.11.1. 1.1- . . ,1 1_ ﬁBEaQom
..._ - 1 95E . 93m
1. 1 . was: as:
530% 1 852 . 9
v -1 -. - oamb1coo=o101 QC. 1 3%
I1 1 1-0—on u «03 1 Cu mac
.1 . 11.1. 9 30 .

 

32

3.3 Results and Discussion
3.3.1 Comparison of an AnMBR and a CMD
The AnMBR was Operated for 108 days in parallel with a CMD. During this period,

both systems had a HRT of 19 days. The AnMBR cross-ﬂow pump was operated on a
timer energized approximately 1/2 of the time until day 45. This was necessary because a
digester heating system had not yet been installed and the pump was used to both
circulate biomass through the membrane and provide heat to the system (thermal energy
imparted from the pump to the circulating biomass). When the circulation pump was
energized, the permeate was continuously returned to the digester tank with the exception
of one time per day when it was discharged from the AnMBR for approximately 1.5
hours (to match the quantity of fresh manure added). Beginning on day 41, an external
heating blanket was added to the digester to maintain system temperature. At this time,
the pump circulation rate was reduced to 15 minutes every two hours to provide periodic
mixing plus an additional 1.5 hours per day for permeate removal.

A summary of the operating data is shown in Table 3.3. Table 3.4 presents the water
quality data. Figures 3.4 and 3.5 present the organic loading rate for COD and VS and

irlfluent/effluent COD and VS concentrations for the digesters.

33

Til?“ 3.3 Summary of ogerating data for AnMBR and CMD

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter AnMBR Days AnMBR Days CMD Days CMD Days
1-40 41-108 1-40 41- 108
UF Rate, /min 36 31 NA NA
Perm. Rate, mL/min 78 48 NA NA
Flux, L/lehr‘ 59 36 NA NA
Avg. TS in reactor, 8.0 10.2 4.0 3.6
%
L CH4/kg VS 118 177 245 272
fed/day'
VS Destruction, % NA NA 15 29
Cycle Time 256 84 NA NA
HRT, days 19 19 19 19
*Actual conditions
Tible 3.4 Summary of water quality data for AnMBR and CMD
Parameter Influent APMBR AnMBR CMD
_ Digester Permeate
COD, 53,700 :1: 18,200 94,900 is 12,400 6,600 :1: 2,400 32,600 :t 8,100
mg/L
Ts,% 4711.4 9.31 1.6 1.11 10.135 3.6i0.8
173.1%. 3.2 :l: 0.9 6.7 :l: 1.1 0.5 :h 0.09 2.4 :t 0.6
?H 7.32 :l: 0.27 7.67 i 0.11 7.81 :1: 0.09 7.75 i 0.10

 

Figure 3.6 illustrates the impact of AnMBR total solids concentration on the ﬂux rate.

This is consistent with the ﬁndings of other researchers (Anderson 1986; Beaubien, Baty

et al. 1996; Brockmann and Seyﬁ‘ied 1996) who also reported a degradation in ﬂux with

cOrresponding to increasing digester TS. There was a steady increase in AnMBR digester

tOtal solids concentration from day 1 to approximately day 46 and a steady decay in the

34

 

 

permeate ﬂux to approximately day 25 at which time ﬂux held stable to day 51. The TS
content of the digester increased to above 10% at day 53, and the ﬂux rate experienced a
stepwise reduction from approximately 60 L/mz/hr to 40-50 L/mz/hr with a steady _
decline beginning day 99 to the end of the experiment at day 108. The membrane was
not cleaned during this study and likely lead to the slow decay in ﬂux rate beginning after

day 51. The step reduction in the ﬂux rate at day 51 appears to be related to attainment of

a critical TS concentration in the digester.

35

 

g COD/L/day
4b
_.....__ .. “I-.. .. .'
I
I
'-

 

Days

 

OLR, g VS/L/day

 

 

 

0.00 i . . . « 1
0 10 20 30 40 SO 60 7O 8O 90 100 110

Days

Figure 3.4 CMD and AnMBR, organic loading rate for COD and VS

36

6 Feed Cl AnMBR Digester

AAnMBR Permeate X CMD

 

 

 

140,000 g
[3
120,000
El
El E1 1:] E]
100,000 1:]
DD 1:1 1:1 '3
E] C] C] DC] 1:] D
S 80000 E1 13 D D 9
I . .
E , . .
d E] O .
8 60,000 ,6 9 ”.0 °
; Q .0 O 0
>2 0
40,000 :xx x x X X
1 o X x>3< x
1 X
20,000 if XX XXXXX )$<
1
019A - - . W4 AMA/AMA?
o 10 20 30 40 so 60 70 80 90 100110
Days
+Feed -B-AnMBR Digester -A—AnMBRPermeate -)(-CMD
9.0 .

 

 

Figure 3.5 CMD and AnMBR, COD and VS

37

Days

 

-B-F|ux +TS Conc.

 

 

 

80 , . 12.0
;n
70 111' n 1
13:1 ‘ .", , 100
l '1‘: l
50 1::: .1 1111 1"1r.‘1:!1”1::‘!.1'n
u :1 ' \ 111 u 1.1 ,
1"1 .1 8.0 BE
‘50 _ 1n
é K L" Fri-41¢?! -151 E
e V224“ -' 11 tau 8
> U '31 .--'.1 Tu
0‘40 " H "' 6.0 ‘6
g 1; ‘ a
£30 . u — E
1 L“ 4.0 E’
20 3 i
g 5' 2.0
10 1 ‘
0 1 , . . 0.0
0 1o 20 30 40 so 60 70 80 90 100
Days

Figure 3.6 AnMBR (from CMD and AnMBR comparison study), permeate rate and
digester TS concentration

Figures 3.7 and 3.8 present a comparison of pH and methane production. During the
period of days 60 through 72, there was an unknown condition at the dairy that resulted
in collected manure with a VS content considerably lower than the previous average.
Once the VS content recovered, the AnMBR gas production returned to a value
consistent with previous readings. During this period of low VS organic loading (Figure
3.4), the AnMBR produced more methane (per mass VS fed) than the CMD. As it was
operated at a VS concentration that was approximately 3 times higher than the CMD, the
probable explanation is that the residual VS of the AnMBR was converted to methane

during this period, accounting for the apparent increase in methane production.

38

+AnMBR -12-.-C1v10
7.8 ,

 

 

 

 

Figure 3.7 CMD and AnMBR, pH

Despite the greater biomass concentration of the AnMBR, the average methane
production for the CMD over the period of the experiment was 262 L CH4/kg VS fed/day
compared to 155 L CH4/kg VS fed/day for the AnMBR. These results suggest a
negative impact on methane productivity related to the cross-ﬂow membrane system.
However, the impact is not as extreme as the ﬁndings of Brockmann and Seyfried (1995;
1996; 1997) who reported a 50% reduction in microbial activity at a cycle time of 20 with

only 10 to 15% of the activity remaining after 120 tolSO cycles/day.

39

 

 

 

600
500 -
4‘ 41
\ . z . o
a A 5'. 4 s ‘
i 300 - A AA' 5AA A ,3
4 hasf‘lxiga 46‘, . 9,.
5 % .2 1 . . .' . 1 g.
_. 200 4,535 :: tiger: 215:: i .:
1A I 1: ’ ‘ . D ' I .
i ' 7 AA ‘ ‘V‘ i ‘
100 ‘ ‘v _, -
l A
o 1 — .— . ~ .
0 1o 30 41 52 63 75
Days

Figure 3.8 CMD and AnMBR, methane production

.‘.'.'.’lIo-'B>

Ghyoot and Verstraete (1997) compared the methane generating potential for biosolids

subjected to the shearing impact of an AnMBR (“treated”) operated at a cycle time of

245/day, with biosolids that had not been impacted by an AnMBR system (“untreated”)

and found a 18% increase in biogas production for the untreated biosolids. The results

for the present research fall between the ﬁndings of Brockmann and Seyfried and those of

Ghyoot and Verstraete.

VS destruction during the period of study for the CMD was 24%. The calculated

VS destruction for the AnMBR was skewed upwards by apparent solids settling in the

digester tank, likely caused by operating for an extended period of time without wasting.

As a result, the data are not valid. Vogel (2003), who conducted the only known

40

AnMBR work on dairy manure, reported a VS reduction of 49% and a COD reduction of
5 0% for the operation of a thermophilic AnMBR treating dairy manure at an average
cycle time of 42 (system was fed three times per week). Gas production estimates were
simulated in Vogel’s work using serum bottles and the basis for predicting gas production
for the pilot anaerobic digester was not clear.

The average methane production for the AnMBR during the ﬁrst phase of the

experiment (cycle time of 256/day, days 1-40) was 118 L CH4/kg VS/day compared to

the CMD during the same period which produced 245 L CH4/kg VS/day. When the

AnMBR cycle time was reduced to 84/day (days 41-108), the biogas production for the
AnMBR improved (Figure 3.8). As previously discussed, the manure fed to the digesters
for days 60-72 was very low in VS (Figure 3.5) and was inconsistent with previous and
future data and the methane production per kg VS fed to the AnMBR during this period
was unusually high, likely due to endogenous decay of existing VS retained in the
digester tank. Comparing days 41-108 skews the AnMBR methane production upwards;
therefore, days 75-108 were used for the methane production comparison between a cycle
time of 256/day and 84/day. The CMD produced 319 L CH4/kg VS fed/day during days
75-108, a 30% increase from the days 1-40. The AnMBR produced 171 L CH4/kg VS
fed/day during days 75-108, a 45% increase. Therefore, the AnMBR experienced a
Signiﬁcant increase in methane production when the cycle time was reduced from
245/day to 84/day indicating an apparent positive impact with cycle time reduction.
The AnMBR COD removal efﬁciency, as measured by comparing the feed to the
Permeate, during the course of the 108 day experimental period equaled 88% (average

feed COD = 53,700 mg/L and average permeate COD = 6,570 mg/L and when evaluated

41

independently for the two cycle times, also resulted in the same removal efﬁciency). The
CMD removal efﬁciency during this same period equaled 39%. The AnMBR ﬁndings
are consistent with the ﬁndings of previous researchers (Cadi, Huyard et al. 1994;
Fakhrulrazi 1994; Anderson, Kasapgil et al. 1996; Ince, Ince et a1. 2000; Ince, Ince et al.
2001; Fuchs, Binder et a1. 2003). However, it is important to note that this does not
speak to the overall COD removal for the process as biomass was not wasted during this

period of study and, as a result, COD accumulated in the digester tank (per Figure 4.9).

Biogas
l

1 Digger
Inﬂuent _,,_. -3... T0113

I
i 1

i ‘—/*l if: "I -1
: Accumulated f * 1

l , ,__SglidsA___ UF Membrane ‘

1
Permeate

Figure 3.9 AnMBR without solids wasting

3.3.2 Combined CMD/AnMBR .

Following the comparison experiment, the CMD and AnMBR were placed in series so
that the effluent from the CMD was acting as the inﬂuent to the AnMBR. This was
conducted to determine if an operational advantage could be leveraged by reducing the
total solids loading to the AnMBR with the expectation of increasing the ﬂux rate while
diminishing the cost of operation. The ﬂux rate is directly related to the solids loading

applied to the membrane (Beaubien, Baty et a1. 1996; Madaeni 1997), and, as a result,

42

reducing the solids loading will improve the ﬂux rate and reduce the energy required per
unit of permeate produced. A second potential advantage is that the readily degradable
substrate will be available for conversion by the CMD and the AnMBR, due to its longer
SRT, will be more effective at converting the more recalcitrant organic matter.

The CMD/AnMBR was operated at a cycle time of 34 for days 1 through 69 and a
cycle time of 56 for days 70 through 282. Table 3.5 outlines the general operating
conditions and Tables 3.6 and 3.7 provide a summary of the water quality data.

Figure 3.10 provides a summary of the organic loading rate in terms of COD and VS
applied to the system. The VS concentration in the feed to the CMD began to decline
around day 215, most likely caused by a problem with operation of the sand-manure
separators at the dairy farm where the manure was collected, the manure feed tank
provided some equalization and, as a result, the decline in the feed VS concentration (and
corresponding organic loading rates) slowed until day 247 before beginning to slowly

increase to a value consistent with the balance of the data.

43

Table 3.5 Summary of operating data for combined CMD/AnMBR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter AnMBR AnMBR CMD Days $1;qu
Days 1-69 Days 70-282 1-69 70_ 282
UP Rate, LPM 36.3 34.8 NA NA
Perm. Rate, mL/min 73 41 NA NA
Flux, L/MZ/hr' 55 31 NA NA
Avg. TS in reactor, % 5.5 5.7 3.4 2.8
VS destruction, % 20 27 22 26
COD destruction, % NA NA 29 39
L CH4/Kg vs fed/day. 82 83 133 190
Cycle Time 34 56 NA NA
HRT, days 10.7 9.7 9.5 9.4
SRT, days 23 25 9.5 9.4
'Actual conditions
Table 3.6 Summary of water quality data for combined CMD/AnMBR Days 1 - 69
CMD
Parameter CMD Influent Effluent/AnMBR AnMBR Permeate
Contents
Inﬂuent
COD, mg/L 42,900 i 9,200 29,500 :t 5,800 53,000 a: 9,800 3,054 i 730
TS, % 4.3 :t 0.5 3.4 :1: 0.5 5.5 i 1.2 0.9 i 0.1
VS, % 2.9 :t 0.3 2.2 i 0.4 3.6 i 0.8 0.3 i 0.1
pH 7.02 d: 0.15 7.73 i 0.08 7.66 :t 0.10 7.77 i 0.12

 

 

 

 

 

 

Table 3.7 Summary of water quality data for combined CMD/AnMBR Days 70-282

 

 

 

 

 

 

 

 

 

 

CMD AnMBR
Parameter CMD Inﬂuent Effluent/AnMBR Permeate
Contents
Inﬂuent
COD, mg/L 33,100 i 9,400 22,600 :1: 5,700 53,800 i 11,500 2,450 i 800
TS, % 3.5 i 0.8 2.8 :1: 0.7 5.7 3:12 0.8 i 0.10
VS, % 2.3 :t 0.6 1.7i 0.4 3.6 i 0.8 0.3 i 0.3
pH 6.98 a: 0.27 7.62 i 0.1 7.59 :t 0.1 7.68 i 0.10

 

44

 

 

 

Figure 3.11 presents inﬂuent and efﬂuent COD and VS data for the digester systems
and Figure 3.12 shows pH and methane production and, though the pH declined during
the period of reduced organic loading to the digester systems, the gas production
remained reasonably consistent over the course of the experiment.

Figure 3.13 shows the titrated volatile acid concentration for samples removed from
the AnMBR. The volatile acid concentration was between 200-400 mg/L as HAc during
the experiment. Volatile acid analysis for samples from the CMD, with few exceptions,
were non-detect (data is not shown). A common metric used to assess the relative health
of a digester system is the ratio of volatile acids to total alkalinity (Speece 1996) .
Manure contains a high concentration of ammonia and organic nitrogen from protein.
During the anaerobic digestion process, the organic nitrogen is degraded and forms
ammonium bicarbonate alkalinity (Speece 1996). Consequently, the volatile acids/total
alkalinity ratio never exceeded 0.06 at any time for the present systems. Vogel (Vogel
2003) operated a thermophilic AnMBR with dairy manure as the substrate at similar
HRTs and SRTs to the present work but at approximately double the organic loading rate
and reported efﬂuent VFA concentrations between 1,000 and 2,000 mg/L during the ﬁrst
100 days of operation with a steady reduction to values less than 250 mg/L. Padmasiri,
Zhang et a1. (2007) reported volatile acid excursions for swine manure greater than 3,000
mg/L as HAc at an organic loading rate of 2.0 g COD/L/day when cross-ﬂow velocity
was increased from 0.9-1.9 m/s but reported VF A concentrations of less than 500 mg/L
as HAc during periods of stable operation. Brockmann and Seyfried (1995; 1996; 1997),

when treating potato waste with an AnMBR, experienced a volatile acid concentration of

45

4,000 mg/L when the organic loading rate was increased from the range of 1.5-3.0 g
COD/L/day to 4.0 g COD/L/day and with a reduction in the organic loading rate,
stabilized at approximately 3,500 mg/L as HAc. Though the AnMBR gas production
lagged the CMD, the system was stable, as measured by VFA production, and consistent

with other reported AnMBR research.

46

IAnMBR ACMD

 

 

 

7.0 ,
gA
‘ A
6.0
AAA1Q A AA A
5.0:; 69 E EA
f A A
. A
6 1| 9&0 A 9A A A [A AA A
B . I AAA ”21 A
8 30 i 'I g I I E Adi A
,, - ‘I': I f f A. A
' ' I {a ":h E3 A
20 ,1 — I
' . I
1 I
1.0
O 25 51 77 104 139 163 189 212 238 261
Days
IAnMBR ACMD
so . . _._ ..,.
4.5 A

s VS/L/dav

 

 

Days

Figure 3.10 Combined CMD/AnMBR, organic loading rates for COD and VS

47

I Feed X CMD Effluent/AnMBR Feed AAnMBR O Permeate

90,000

 

80,000 A
70,000 Aﬁ

. A
60,000 1. % A A
50,000 II? M E A I ﬁ§
- - -'.- 1w 1..
30,000 WXX§S< ‘ W %

COD, mg/ L

20,000

 

  

10,000 4

 
 

. g_ 4 .‘
Kmafﬂut'ljﬂ'ﬂ"!

.;.

. . o . - . , , . . . . . - .
5.1.2.1! [tum u! u! '1 "m“ it." nitluoygmju"11.111311: 1mm 1'11 1‘.
, ,._-...;.-._...._,......V...._...I . - ,‘ V....I....... .7; r. HT. - -_ ;.

    

0 25 51 79 106 141 165 191 215 240 263

Days

I Feed XCMD Effluent/AnMBR Feed AAnMBR O Permeate

 

 

 

Figure 3.11Combined CMD/AnMBR, COD and VS

48

I AnMBR A CMD ——Linear(AnMBR)

 

 

L CH4/kg VS fed/day

 

'1"’!; ""T "‘Y"" I V'"“"'" "‘7'" v 7 r . r y l , . 1 . r
20 40 65 87 111 145 166 186 210 230 250 270

 

days

UAnM BR ACMD

 

9.0 ;

 

6.5

 

 

0 20 40 63 85 107 140 164 184 205 225 245 265

Days

Figure 3.12 Combined CMD/AnMBR, gas production and pH

49

700

 

600 ,

500 v

400 ~

300 \

 

Volatile Acids, mg/L as Hac

200 ‘

 

100 ,

 

 

 

Figure 3.13 Combined CMD/AnMBR, volatile acid concentration

The methane production rate of the AnMBR, during days 1-69 and operating at a
cycle time of 34/day, equaled 82 L CH4/kg VS/day. Following the increase in cycle time
to 56, the methane production was nearly identical at 83 L CH4/kg VS/day for days 70-
282. During the period between days 1-69, the CMD methane production rate equaled
133 L CH4/kg VS/day but increased to 190 L CH4/kg/day between days 70-282.

AanfBR VS destruction for days 1-69 was 20% and the CMD VS destruction for this
same period was 22%. VS destruction for days 70-282 for the AnMBR was 27% while
the CMD VS destruction for this same period was 26%. The VS destruction results are at

odds with the methane production rate. One potential explanation relates to the type of

50

VS being converted to methane. Methane is produced consistent with the chemical
makeup of the substrate (Bryant 1979). It is possible that the VS degraded by the CMD
contained greater methane potential per mass than the residual VS fed to the AnMBR.
Therefore, because the AnMBR was fed with the CMD efﬂuent, VS destruction may not
provide a good metric for comparing the systems. Also, because the AnMBR is operated
at a higher biomass concentration, there is a greater propensity for solids to settle in the
digester tank. The increase in AnMBR VS destruction from days 1- 69 compared to days
70- 282 could be partially the result of increased solids settling and an overstatement of
the VS destruction.

There was also a difference in the VS content of the manure for days 70-282 which
averaged 2.3% compared to 2.9% for days 1-69. The resulting gas production per mass
VS fed to the CMD was signiﬁcantly greater for days 70-282. Though there was less
total mass of VS during the second period, it is possible that a higher percentage of this
VS was readily degradable and this may account for the increased gas production of the
CMD. As a result, it cannot be concluded that there was a difference in gas production
between a cycle time of 34 and 56.

The COD reduction based on the difference between the COD concentration of the
CMD feed and the AnMBR permeate equals 93% (average permeate COD equals 2,450
mg/L and average CMD feed equals 33,127 mg/L). It is important to note that this does
not represent the total system COD reduction as it does not account for COD wasted from
the AnMBR. COD reduction data is presented in this fashion for consistency with

previous AnMBR research.

51

+ Flux, L/MZ/hr m-A Digester TS

 

 
   
 

120 , , 9.00
8.00
100 «
.1; "- ,. 7.00
80 . I 5 1' 5.00 BE
1 5 8
5 5.00 §
. _ n
60 'r A H; ; "2‘
4.00 c
2
40 '3' " 3.00 '5
2.00
20
l 1.00
7
0 1r 1 . 1 . . . . . ,; . . 1 I . . . . 1 1 0.00

 

0 20 40 63 85 107 138 159 179 200 220 240 260 280

Figure 3.14 Combined CMD/AnMBR, flux rate and digester TS concentration

AnMBR TS concentration (in the AnMBR digester tank, reference Figure 3.3) versus
flux rate is shown in Figure 3.14. The digester TS concentration started at 2% and
increased to approximately 6% by day 24 with a corresponding decrease in the ﬂux rate
from approximately 100 L/mZ/day to 50 L/mz/day. The digester TS and the ﬂux rate
were stable between days 24 to 60. Aﬂer day 60, there was a spike in digester TS
concentration and a corresponding decrease in ﬂux rate followed by a decline in digester
TS and an increase in ﬂux rate between days 77 and 87. Following day 87 until day 179,

there was a steady degradation in flux, despite a decline in the TS content of the digester,

52

mostly likely caused by membrane fouling. At day 179, the membrane was cleaned.

There was an immediate increase in ﬂux from the range of 10-20 L/mz/day to a little

more than 60 L/mz/day. For the next 100 days, the digester TS increased to

approximately 7% and then steadily decreased from day 218 to 282 to approximately 4%

while the ﬂux rate was stable. It is possible that the decreasing TS content was balanced

by a corresponding increase in membrane fouling resulting in a stable ﬂux during this

period.

3.3.3 Nutrients

Table 4.8 provides data for various nutrients of interest. The AnMBR permeate

phosphorus concentration equaled 16 mg/L (approximately 95% removal efﬁciency).

The permeate TKN concentration equaled 1,454 mg/L (approximately 30% TKN

removal efﬁciency). The AnMBR membrane did not signiﬁcantly impact the ammonium

N or potassium concentration because both were present in the soluble state and thus

passed through the membrane.

Table 3.8 Water quality at each samplingpoint

 

 

 

 

 

 

 

 

 

 

 

 

Parameter CMD CMD AnMBR AnMBR

n Influent Effluent Permeate Digester

Total P, mg/L 7 339i27 322i68 16:1:5 791i156
TKN, mg/L 7 2,070i205 2,130:t131 1,450188 3,090i550
Organic N, mg/L 7 8711325 848i69 121i23 1,810:t339
Ammonium N, mg/L 7 1,200:t416 1,280i91 1,330i84 1,5702t76
Potassium, mg/L 7 1,840:I:189 1,7901205 1,780i148 1,900i113

 

During the period of study, the average input feed to the combined CMD/AnMBR was

11.0 kg/day of raw manure, the average permeate discharge rate was 6.8 kg/day and the

53

 

average wasting rate was 4.2 kg/day. A consistent HRT was maintained throughout the
experiment. Based on this mass balance accounting, nearly 97% of the total phosphorus
can be accounted for in the system. This is consistent with other researchers who

reported near perfect mass balances for total phosphorus for plug ﬂow digestion systems

(Wright 2004; Martin 2007).

3.3.4 Removal of Virus and Pathogen Indicators

The membrane provides a barrier that excludes suspended solids that are larger than
the pore size of the membrane. Limited work has been conducted with respect to
quantifying the ability of AnMBR systems to exclude viruses and pathogens. A study
was conducted by Wong et al. (2009) to evaluate the removal of bacterial and virus
indicators in the efﬂuent of the AnMBR used for the present research. E. coli,
Enterococci and C. Perfringens were the bacterial indicators and somatic coliphage was
the viral indicator monitored in this study. The paper published from this work is
attached, with the permission of the Journal of Environmental Quality, as Appendix A.
Referencing Table 2 from Appendix A, the inﬂuent tested positive in all 8 sampling
events for E. coli, Enterococci, C. perfringens and coliphage. The CMD efﬂuent tested
positive for E. coli (8/8), Enterococci (7/8), C. perfringens (8/8) and coliphage (8/8). The
AnMBR efﬂuent tested positive for E. coli (2 of 8 samples), Enterococci (3 of 8
samples), C. perfringens was not detected in any of the AnMBR samples and Coliphage
(5 of 8 samples). The average values for E. coli, Enterococci, C. perfringens and
Coliphage in the combined CMD/AnMBR efﬂuent were 0.31, 0.51, ND and 2.47

loglocfu/L respectively with loglO removals of 6.7, 7.3, 6.5 and 4.2, respectively. The

54

coliphage exhibited the highest occurrence frequency and concentration in the AnMBR
efﬂuent which is to be expected because viruses are generally smaller than bacteria, with
diameters as small as 0.01 pm. The average pore size for the AnMBR membrane was
0.03 pm. The loglO removals of the indicators are illustrated in Figure 2 of Appendix A
and the results illustrate that most of the removal was due to the AnMBR.

The high rate of removal attributed to the AnMBR was a direct result of the
membrane. However, it is not clear if the membrane pore size is solely responsible for
the rejection of the virus and pathogen indicators or if the removal efﬁciency is due to the
ﬁltering impact of the gel layer due to concentration polarization. He et al. (2005)
reported bacteria removal ranging from 5.65 loglo removal to 5.14 loglo removal
(>99.9%) with membrane pore sizes ranging between 20,000 and 70,000 Da. 20,000 Da
is approximately equal to 0.01 um and 70,000 Da is approximately equal to 0.06 mm.
Concentration polarization is described as the formation of a gel layer at the membrane
surface due to retained solutes. This gel layer forms a secondary barrier to ﬂow through
the membrane (Baker 2000). It seems likely that the gel layer provides additional
ﬁltration capability and may explain the relatively small difference between the bacterial
removal of 20,000 Da membrane versus the 70,000 Da membrane. Nevertheless, the

bacteria removal rate is consistent with that reported by Wong et al. (2009).

55

CHAPTER 4
CYCLE TIME COMPARISON
4.1 INTRODUCTION

AnMBRs have been reported to provide robust treatment at high organic loading
rates. However, other researchers have found reductions in microbial activity, reportedly
due to the shearing impact of the pump/membrane system. A determining factor appears
to relate cycle time, a measurement of pumping frequency, to microbial activity. Higher
cycle times have been reported to reduce microbial activity. This relationship and

advantages and disadvantages of AnMBRs are discussed in detail in Chapter 2, Section 7.

A typical HRT for a complete mix, manure-based digester system is in the range of 10
to 30 days with dairy manure as the substrate (Hills 1979; Oliver, Pain et al. 1986;
Summers, Hobson et al. 1987; Pain, Phillips et al. 1988; Ghaly and Echiegu 1992; Vogel
2003). Considering average ﬂux rates identiﬁed from this research (Chapter 3, Tables
3.3 and 3.5) and typical HRTs discussed, the estimated cycle time for a manure-based
AnMBR is in the range of 4 to 30 cycles per 24 hour period (Appendix D). The cycle
time depends on the total system volume and the conﬁguration of the membranes. By
way of example, if the cross ﬂow circulation rate through an AnMBR equals 7,500 LPM
(10,800,000 LPD) and the digester tank volume equals 400,000 liters, then the cycle time
equals 27 cycles/day (10,800,000 L/day + 400,000 liters). If all other parameters remain
equal, one way to decrease the cycle time is to increase the size of the digester tank. For
example, if the digester tank volume were increased to 800,000 liters, the cycle time for

the above example becomes 14 cycles/day. The obvious disadvantage is the increased

56

capital cost. In an optimum conﬁguration, the tank size will be maintained at the least

possible volume. However, as volume decreases, cycle time also increases.

An alternative is to change the membrane conﬁguration. Figure 4.1 illustrates the
cycle time differences that result if the membranes are placed in a serial conﬁguration.
Placing 4 membranes in series, rather than in parallel, results in a cycle time of
approximately ‘A of a parallel conﬁguration. A serial conﬁguration, however, results in a
signiﬁcant increase in pressure to maintain the desired cross-ﬂow velocity through the
membrane and creating a more complex design. System pressure limitations also dictate

the number of membrane modules that can be placed in series.

57

5:23:59 3:23 a 23...? 3.8... a E $885.8 35.5.0 2.29:2: he Eaten—~80 ﬁv v.5»:—

 

 

 

 

     

EESQ E 85
mo an 68: 293 a E 3.38% “802
cougmmcoo n50 :ozﬁzwwcoo
Ear—mm ooov . 3Com Q90
1 l 41 i 1 l . llll. :1; . OSM
OED ‘1 . .
88 _ L L - iL l
.A M
. L : ,. 96
r ., .. ,. 82 ..
. 0:5 . . . , e
, , . _ 838.5 _ . l l _w- A. .i- .1, 0:5.
_ 82 _ M 8?: z “ dogmowﬁw
885.3 + r - -- ll 1 888.5%
‘ . 1,1 -- it

58

The UF membranes used in this research were full-size and designed for industrial
applications. Timers started and stopped pumps in order to achieve target operating
conditions. A series of tests were performed at various cycle times, digester VS
concentrations (biomass concentrations), cross-ﬂow velocities and membrane
conﬁgurations to determine the implications of these parameters on system design with
the goal of maximizing methane production. Three distinct phases of experimentation

were undertaken as discussed in the following subsection.

Phase 1 compared the methane production of three digester systems identiﬁed as
single membrane digester (SMD), membrane equivalent digester (MED) and complete
mix digester (CMD). The objective of this experiment was to determine if a PVC pipe
could be used as a surrogate membrane for future conﬁguration evaluation and to
establish a baseline for methane production at a cycle time of 6 and cross-ﬂow velocity of
4.5 m/s. The SMD was operated with 100% of the membrane permeate returned to the
digester tank so that it resembled a complete mix digester with HRT equal to SRT. The
MED employed a 1750 mm x 13.9 mm diameter PVC pipe to simulate the turbulence and
pressure drop of the membrane used for the SMD. All three systems were operated at an

approximate cycle time of 6 (the SM and MED = 7/day and the CMD = 5/day).

Phase 2 compared the methane production of three digester systems identiﬁed as single
membrane AnMBR (SM AnMBR), multi-membrane AnMBR (MM AnMBR) and CMD.
The objective of Phase 2 was to determine if there was a methane production advantage

between a cycle time of 6 (MM AnMBR) and 27 (SM AnMBR). A second objective of

59

Phase 2 was to evaluate whether the operational time of the CMD mixing pump affected
methane production. The mixing pump was operated 6 times per day x 3 minutes for
days 1 through 54 (equivalent cycle time of 5/day) and 4 times per day x 1 minute for

days 55 through 92 (equivalent cycle time of 1/day).

Phase 3 compared the methane production of three digester systems identiﬁed as
SMD, multi-membrane equivalent digester (MMED) and CMD. The SMD was operated
with 100% permeate recycle at a cycle time of 27 which enabled comparison with the SM
AnMBR of Phase 2. The MMED consisted of four 3,000 mm x 13.9 mm diameter PVC
pipes and was operated at a cycle time of 6 to evaluate the pressure and turbulence impact
of placing four membrane modules in series and was selected because it represented a
probable full-scale conﬁguration. The MMED and the SMD were compared with a
control CMD operating with an equivalent cycle time of 1. The objective of Phase 3 was
to evaluate the impact of biomass concentration on methane productivity and to assess
the difference in methane production between the torturous path of the MM AnMBR

compared with the more realistic design of the MMED.

4.2 MATERIALS AND METHODS
4.2.1 General AnMBR Configuration
Two AnMBR systems were operated in the previously described conﬁgurations. A

schematic of the general AnMBR layout is presented in Figure 4.2.

60

0.03

 
 
 
  

Micron ;
Tubular
; UF
l ; Membrane;
Gas to _: , - '1
Wet Scale ._ I 1'
Ti . r a *
Sand and Meir 1Concentrate
Solid/Separated j Return . .
Dairy ; _ _ l Circulation
Manure if 1 pump
1 ,1 . Digester ___ _ﬂ 4;"pr 11315;» -~—»-
..Vg,’ . 4 1“" " Tank I Magnetic Flow
_ 7 l ' Meter
7 Scale . ' - a T
l
Scale
Sampling Points ‘ " Inﬂuent to AnMBR ’ Solids Wasting AnMBR Permeate

Figure 4.2 General AnMBR layout from which the SM AnMBR and MM AnMBR

are derived

4.2.2 Phase 1

Three digester systems operated in this phase: SMD, MED and CMD and each is

described in detail below.

SMD - A schematic of the SMD is presented in Figure 4.3. Permeate is returned to the
digester. The digester was a 175 cm tall x 30 cm diameter section of schedule 40 PVC

pipe with ﬂanged ends. The working volume was 115 liters with approximately 10 cm of

headspace for gas collection. A 0.64 cm diameter tube directed the biogas in the

headspace to a wet tip meter (Wet Tip Meter Company, Nashville, TN).

61

Biogas to Wet
Tip Meter

i Permeate

1"" ____

 

Biomass

Feed * - --—-‘ Tank '

 

Membrane

l
Wasted Efﬂuent

Figure 4.3 SMD operated with single element in a complete mix configuration

Gas samples for GC analysis were collected between the digester and the wet tip meter
via a luer-style, 3—way valve. The digester was heated using an external heat blanket and
thermostat (BriskHeat, Columbus, OH). Digested manure was removed from the system
once per day based on mass and fresh manure was added to the system once per day
based on mass. The manure was collected from the Car-Min-Vu Dairy, Williamston, MI
one time per week and stored in 5 gallon carboys at room temperature. Carboys were
added to a IOO—L mixing tank (mixed with submersible pump prior to feeding for
approximately 5 minutes). The SMD was operated in a cross-ﬂow conﬁguration and
used a 0.03 micron, 14.4 mm diameter, 0.079 m2 PVDF tubular ultraﬁltration product (X-
Flow, Inc., Netherlands). A 1.5” self-priming centrifugal pump (AMT Inc., Mansﬁeld,
OH) was used to generate a circulation rate of approximately 43 L/min (cross flow

velocity = 4.5 m/s).

62

MED — A diagram of the MED is presented in Figure 4.4. With the exception of a 13.9
mm diameter PVC pipe of 1750 mm in length used as a surrogate for the UF membrane,

all other aspects of the digester tank, manure addition and manure source were as

described for the SMD of this section.

Biogas to Wet
Tip Meter

l

1 l
l

l Digester

l ._ 7f? —- ---*C:‘:_:ZT"”y——~J
‘ I 1500 mm x 13.9 mm
* —‘—"‘ “*4 Diameter" PVC Pipe

Wasted
Effluent

Figure 4.4 MED with 13.9 mm pipe surrogate for membrane

CMD - A CMD was operated as a control to compare performance with the AnMBR
systems. Figure 4.5 provides a schematic of the CMD system. The CMD consisted of a
122 cm tall x 55 cm diameter HPDE vessel with a working volume of 166 liters. Mixing
was achieved with an AMT 1 x 1.5 centriﬁigal pump (AMT, Inc. Mansﬁeld, OH),
operated 6 times per day for 3 minutes. The circulation rate of the pump was
approximately 45 LPM. Digester heating, gas collection, manure addition and manure

source as described in for the SMD of this section.

63

Gas to

Wet
Tip
Sand Meter
. 311d i Mixing
Solld-Separated ‘ . .A 1| Pgmp 7-
Dairy i «4:. ..;.. r
Manure l Digester ' -- I ,, ,i .
. ‘ ﬂ ' Tank l . l
f ‘4... 7 I 2
. I v . .
Scale
Scale
Sampling Points: : i Inﬂuent to CMD CMD Efﬂuent
Figure 4.5 CMD system
4.2.3 Phase 2

Three digester systems were operated for this phase of experimentation: SM AnMBR,

MM AnMBR and CMD and each is described in detail below.

SM AnMBR — Referencing Figure 4.3, the SM AnMBR was the same as the Phase 1
SMD with the exception that permeate was removed from the system (rather than
returned to the digester). The feed rate was set based on a design HRT of 12 days. The
quantity wasted equaled the difference between the feed rate and the permeate removal

rate.

64

MM AnMBR - The MED from Phase 1 was replaced with a module containing seven,
14.4 mm diameter x 1750 mm x 0.03 pm pore size PVDF ultraﬁltration membranes with
a total area of 0.55 m2, manufactured by X-Flow, Inc. (Netherlands). A schematic is
shown in Figure 4.6. The working volume was 119 liters with approximately 10 cm of
free board for gas collection (slightly higher than the SM AnMBR due to the additional
membranes used in the module). The MM AnMBR was operated in a cross—ﬂow
conﬁguration using a Summit 2196LF, 1x1.5x8 centrifugal pump (Summit Pump, Inc.,
Green Bay, WI) to generate a circulation rate of approximately 28.5 L/min (cross-ﬂow
velocity = 2.9 m/s). A manifold was constructed to allow the elements to be operated in
series (Figure 4.7) such that the system contained enough membrane surface area to allow
for operation at a cycle time of 6. All other aspects of the digester tank, manure addition

and manure source as described in Section 4.2.2 for the SMD.

CMD —CMD conﬁguration was identical to that described in Section 4.2.2.

65

Biomass

Biogas to Wet
Tip Meter

4

Permeate

Digester
Feed __ Tank

l
Wasted
Efﬂuent

Figure 4.6 MM AnMBR with 7 elements connected in series

 

Figure 4.7 MM AnMBR module illustrating manifold

66

4.2.4 Phase 3

Three digester systems were operated for this phase of experimentation: SMD, MMED

and CMD and each is described in detail below.

SMD — The SMD was identical to that described in Section 4.2.2

MMED - The seven element membrane module was replaced with four 3000 mm x 13.9
mm diameter sections of PVC pipe (Figure 4.8). The same Summit centrifugal pump was
used to generate a circulation rate of approximately 41.5 L/min (cross-ﬂow velocity = 4.5
m/s) to provide a cycle time of approximately 6. All other aspects of the digester tank,

gas collection, manure feeding and manure source are as described in Section 4.2.2.

 

 

Biogas to Wet
Tip Meter
i W? ._L._, . _[ i
, *4—————t_’__ _ _ .14-
l . l Tl
, Digester , . _ be?” “3-! 4
Feed — >~: Tank l I
‘ - L——*l _ n T 416
l
i ' l
, 1’ ” 13.9mm x 3000 m
t PVC Pipe (TYP)
Wasted
Efﬂuent

Figure 4.8 MMED operated with four 3000 mm x 13.9 mm diameter PVC pipe

CMD — The CMD was identical to that described in Section 4.2.2

67

4.2.5 Substrate

The substrate fed to all systems was collected from the Cal-Min-Vu Dairy

(Williamston, MI) on a weekly basis. Cal-Min-Vu Dairy employs a McLanahan sand-

manure separator followed by a McClanahan UL T RA cyclone (McLanahan Corp.,

Hollidaysburg, PA) for recovery of ﬁne sand particles. The sand-separated manure is

further processed through a press screw separator, PSS (Fan Separator, Carol Stream, IL).

Liquid manure from the F AN unit was used as the substrate for the pilot-scale

experiments outlined in this chapter. Manure was collected on a weekly basis. As with

any “real-world” operation, the quality and character exhibited variability depending

conditions at the dairy. Table 4.1 summarizes the average feed manure characteristics for

Phases 1, 2 and 3.

Table 4.] Characteristics of substrate for Phase 1

 

 

 

 

 

 

 

 

 

 

 

Parameter Value Standard Deviation
COD, mg/L 41,800 3,500

TS, % 3.3 0.2

VS, % 2.2 0.1

pH 7.11 0.14

Total alkalinity, mg/L as CaCO3 10,400 1,060
Volatile acids, mg/L as HAc 2,110 417
Bicarbonate alkalinity, mg/L as CaC03 8,610 1,020

 

68

 

Table 4.2 Characteristics of substrate for Phase 2

 

 

 

 

 

 

 

 

Parameter Value Standard Deviation
COD, mg/L 52,300 14,800

TS, % 4.0 1.1

VS, % 2.6 0.6

pH 7.02 0.14

Total alkalinity, mg/L as CaCO3 10,600 1,420
Volatile acids, mg/L as HAc 1,810 873
Bicarbonate alkalinity, mg/L as CaCO3 9,100 1,260

 

 

 

 

 

Table 4.3 Characteristics of substrate for Phase 3*

 

 

 

 

 

Parameter Value Standard Deviation
COD, mg/L 42,200 17,500

TS, % 3.9 1.3

VS, % 2.4 0.9

pH 7.03 0.33

 

 

 

 

 

.Total alkalinity, volatile acids and bicarbonate alkalinity not analyzed for this phase because systems were
stable and test was consistently non-detect

4.2.8 Analytical Methods

General analytical methods used in this chapter were described in Chapter 3.2.2. A

GCMS analysis procedure used in this chapter is presented in Appendix E.

4.3 Results and Discussion
4.3.1 Phase 1

This conﬁguration compared the methane production and VS destruction of a single
1750 mm long x 14.4 mm diameter membrane operated under complete permeate recycle

conditions (SMD, Figure 4.3) with a 1750 mm x 13.9 mm. diameter pipe (MED, Figure

69

4.4) and a control (CMD, Figure 4.5). Table 4.4 provides a summary of the average
performance parameters and Table 4.5 details the water quality results with plus/minus
values indicating the standard deviation for each result.

The SMD and MED were operated at a CF V of 4.5 m/s. There is general agreement
in the literature that permeate ﬂux increases with increasing cross-ﬂow velocity (see
discussion, Chapter 6, Section 3). Most AnMBR work has been conducted at CFVs of
3.0 m/s or less (see Chapter 2, Table 2.2). A higher CFV was selected for this
experiment to enable ﬂux rate comparison with a lower CFV used in Phase 2 and to
evaluate its impact on methane production. Based on comparing methane production of
the SMD and MED with the CMD, there does not appear to be a negative impact on
methane production at a CFV as high as 4.5 m/s. In an effort to mirror the operating
conditions for the SMD and the MED, the CMD was also operated at a cycle time of
approximately 6 and this may have had a negative impact on methane production for the
CMD. The mixing rate for the CMD and its impact on gas production is explored further
in Phase 2.

Figure 4.9 presents the organic loading rate for VS and COD and Figure 4.10 the feed
and efﬂuent VS and COD for all three digesters. Data collection began on the 12th day of
operation for VS and on the 16th for COD. There were a limited number of COD

sampling events for Phase 1 and, as a result, COD destruction was not presented in Table

4.4.

70

Table 4.4 Summary of operating data for Phase 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter SMD MED CMD
Cross-ﬂow velocity, m/s 4.5 4.5 NA
Flux, L/mZ/hr 118 Na NA
Circulation rate, LPM 43 43 45
Transmembrane pressure, 85 87 NA
kPa
Membrane entry pressure, 124 124 NA
kPa
Avg. TS in Reactor, % 2.6 2.6 2.5
V8 destruction, % 27 30 33
L CH4/kg vs fed/day. 264 268 280
CH4 concentration, % 72 72 71
Cycle time, day'1 7 7 5
HRT, days 15.3 15.8 14.8
SRT, days 15.3 15.8 14.8
.actual conditions
Table 4.5 Summary of water quality data for Phase 1
Parameter Inﬂuent SMD MED CMD
COD, 41,8002t3,500 30,800i2,940 29,70014,270 26,800i2,990
mg/L
TS, % 3.33:0.2 2.6i0.1 2.6i0.2 2510.1
VS, % 2.2:t0.l 1.63:0.1 1.5:t0.1 1.5:120.l
pH 7.11i0.14 7.71i0.23 7.7li0.23 7.83:0.17

 

 

 

 

 

 

71

 

2.5

1.5

OLR, g VS/L/day

0.5

4.5

3.5

2.5

1.5

OLR, 3 COO] l./ day

0.5 -

+SMD +MED "-XHCMD

 

 

 

 

 

 

 

 

 

 

70

l

i

i

l

l

0 S 1015202530354045505560657075
Days

mnElSMD --o--MED —&—CMD

l

0 10 20 30 40 SO 60
Days

Figure 4.9 Phase 1, organic loading rate, V8 and COD

72

+CMD —0—Feed "-AHSMD --)(--MED

 

2.60 5

2.40

2.00

VS, 96

1.80 5,
1.60

1.40

 

1.20 ~

 

1_oo . . . a,
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75

Days

—0—Feed -€--SMD +MED —)(—CMD

 

 

55,000

50,000

45,000

40,000

COD, mg/ L

35,000
30,000

25,000

 

20,000

Days

Figure 4.10 Phase 1, COD and VS concentration

73

 

The desired operating HRT (11-12 days) was reached around day 57. Figure 4.11 and
4.12 show pH and volatile acid concentration during startup of the three systems in Phase

1.

+Feed +CMD "'AHSMD --)(--MED

 

8.20
8.00 -
7.80 J . ‘
l . - A ‘ _ 1‘ ‘ ‘
. a «wax MM
, A\/‘\v A It ‘6‘-
7.60 - , “ t :
10:. “A: 1:
. A n

740 >5 I A

ll 1. " A ‘

" o ‘ A A v

' '\ A (s
7.20 ‘ ' 4 , /’\

x x l I \~\ ‘ I, V v’ r 0 x ‘ (,5
\ Q
7.00 - V v A w ,H,
\‘v ’4 v v v
V v

6.80 """""‘ T"'”""‘ . . T“ l r l r l v v

 

 

 

Days

Figure 4.11, Phase 1, pH

Ghaly and Echiegu (1993) reported a total volatile acid concentration of approximately
2,000 mg/L for raw dairy manure with a TS concentration of 3.3% and a volatile acid
content of approximately 26 mg/L using a GC method for the digested manure in a
continuous mix anaerobic reactor. Hoffrnann, Garcia et al. (2008) reported volatile acid
concentrations for stable dairy manure digesters in the range of 250 mg/L using a titration
procedure for quantiﬁcation. The systems in the present research were considered stable

74

when the volatile acid content, as measured via a titration procedure (see Chapter 3,
Section 3.2.2 Analytical Methods and Appendix B), was less than 250 mg/L and the pH
was consistent and stable. The CMD and the MED were largely stable by day number
40. The SMD was slower to reach stable operation and did not reach a volatile acid

concentration of less than 250 mg/L until day 57 (Figure 4.12).

+SMD "-AHMED ->(- CMD
4500

 

4000 -

3500 ‘

3000 .

2500 ,.

2000

1500

Volatile Acid, mg/L as Hac

1000 -

500 -

t A. .A

 

 

 

 

. .aﬁ-i-x ' . «pk-x-rxm .
11 16 19 22 25 30 32 36 38 4o 43 44 49 55 57 59

Days

Figure 4.12 Phase 1, volatile acid concentration

75

+SMD -13'MED "-A"CMD

 

500 1“” .-.
450 -
400 .
350 «
300 -

250 -

L methane per kg VS Fed

150 ‘

 

100 -i

501'

 

0 10 20 30 4O 50 60 7O 80

 

Days

Figure 4.13 Phase 1, methane production

The methane production rate (Figure 4.13) for all three systems stabilized at
approximately the same value from day 55 through the end of the experiment with very
low volatile acid production and stable pH.

Though the SM system was operated with 100% permeate recycle, its ﬂux rate was
still measured on a daily basis and this is presented in Figure 5.14. The ﬂux rate for the
SMD was relatively high based on previous work conducted in this research (Chapter 3,
Section 3.1 and 3.2, Tables 3.3 and 3.5 and Figures 3.6 and 3.14); however, this test was
performed at a digester total solids concentration of 2.6%, much lower than previous
work (Figure 4.14). The average ﬂux rate of 118 L/mz/hr is higher than most other
AnMBR work found in the literature. Zitomer, Bachman et al. (2005) also working with

76

dairy manure, reported ﬂux rates between 40 to 80 L/mz/hr at an operating CFV of 3.3
m/s and a digester TS concentration of approximately 3%. Pierkeil and Lanting (2005)
reported an operating ﬂux of 145 L/mZ/hr with municipal solids as the substrate;
however, they were operating at a total solid content of 1% without a reported CFV.
Strohwald and Ross (1992) found a linear relationship between CFV and membrane ﬂux
for a membrane bioreactor treating brewery efﬂuent. The reported ﬂux for Phase 1 of
this research was conducted at a CFV of 4.5 m/s, substantially higher than most reported
rates and this likely contributes to the higher ﬂux rate observed in this research compared
to ﬂux rates reported in the literature.

....B- Flux + Digester TS
160

 

2.9

140

120

100 -*

80

Flux, L/mzlhr

60~

 

Digester Total Solids, %

40-

20 7 ‘ 2.3

 

 

 

0 1 x I X l l l l l l l ' l ' 2-2
0 5 10 15 20 2530354045 5055 6065 70

Days

Figure 4.14 Phase 1, SMD flux rate versus digester TS concentration

77

The following conclusions for Phase 1 follow:
1. Cross-ﬂow velocity as high as 4.5 m/s does not inﬂuence methane production
when compared with the CMD control mixed 6 times per day x 3 minutes.
2. The 13.9 mm Pipe can be used as a surrogate for a UF membrane under the tested

conditions

4.3.2 Phase 2

Testing in Phase 2 evaluated the biogas production differences between two AnMBR
systems, one operating at a cycle time of 6 (Figure 4.6) and the other operating at a cycle
time of 27 (Figure 4.3, permeate not returned to digester tank) with a control CMD
(Figure 4.5). The operating data and water quality data are presented in Tables 4.6-4.9,
respectively with plus/minus values indicating the standard deviation for each water
quality result.

During days 1-54, the CMD mixing pump was operated 6 times per day x 3 minutes,
consistent with its operation for Phase 1. The mixing rate for the CMD was reduced to 4
times per day x 1 minute for days 55-92 in an effort to determine if the higher mixing rate

was negatively impacting methane production.

78

Table 4.6 Summary of operating data for Phase 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter SM AnMBR MM AnMBR CMD
UF circulation rate, LPM 43 29 NA
Cross-ﬂow Velocity, m/s 4.5 2.9 NA
Flux, L/mZ/hr 53 24 NA
Transmembrane Pressure, kPa 86 240 NA
Membrane entry pressure, kPa 124 450 NA
Avg. TS in Reactor, % 4.3 5.0 2.8
VS destruction, % 38 33 33
COD destruction, % 36 34 42
LCH4/kg vs fed/day" 252 246 247
L CH4/kg VS fed/day (days 1- 252 NA 245
54)‘

L CH4/kg VS fed/day (days 252 NA 251
55-92)’

CH4 concentration, % 67 67 65
Cycle time, day'1 27 6 1
HRT, days 12 12 12
SRT, days 24 27 11.6

 

 

 

 

 

 

.
Actual conditions

COD analysis was conducted approximately 3 times per week while VS analysis was

conducted every day. The COD (or VS) destruction was calculated based on the

79

difference between COD (or VS) added to the digester, COD (or VS) discharged from the
digester plus (or minus) any accumulation of COD (or VS) in the digester. The VS
destruction for all three systems was similar; however, the COD destruction for the CMD
was considerably higher than for the AnMBR digesters. The reported COD destruction
for the CMD (42%) appears too high based on comparison to the AnMBR results and
methane production and may be a result of the timing of COD analysis. Because VS data
was collected every day, it is a better metric for comparison.

The methane production for the CMD, shown in Table 4.6, was very similar for both
mixing regiments, indicating the higher rate of mixing during days 1 through 54 did not
negatively inﬂuence methane production. As a check to estimate if data was skewed by
changes in manure fed to the CMD, it was conﬁrmed that the methane production for the
SM AnMBR was the same for days 1-54 and 55-92, suggesting consistency in the
methane production potential of manure for both periods. The MM AnMBR methane
production data was not shown during the two periods of interest because it was still in

the start-up phase during days 1 through 15.

Table 4.7 Summary of water quality data for Phase 2 - SM AnMBR

 

 

 

 

 

 

Parameter Inﬂuent Wasted Efﬂuent Membrane

Permeate
COD, mg/L 52,300:H4,800 50,20017,000 5,260il,550
TS, % 4011.1 4.3106 0810.1
VS, % 2610.6 2810.4 0.310.]
pH 7.021014 7.701006 7.771006
Average 10.0 4.9 5.1
fed/removed, kg/day

 

 

 

 

 

80

 

Table 4.8 Summary of effluent water quality data for Phase 2 - MM AnMBR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Inﬂuent Wasted Efﬂuent Membrane

Parameter
Permeate
COD, mg/L 52,300114,800 56,50017,620 5,59011,950
TS, % 4.011.] 5.0107 0.9101
VS, % 2.6106 3.2105 0310.1
pH 7.021014 7.671006 7.751006
Average removed, 10.1 4.4 5.7
kg/day
Table 4.9 Summary of effluent water quality data for Phase 2 - CMD

Parameter Inﬂuent Effluent
COD, mg/L 52,300114,800 29,90016,010
TS, % 4.011.] 2.8105
VS, % 2.6106 1.7103
pH 7.021014 7.761006
Average removed, kg/day 14.3 14.3

 

 

 

 

 

The volatile acid concentration for the SM AnMBR and the CMD remained below the
detection limit of the titration procedure used for analysis, consistent with the steady-state
condition of Phase 1 (data not shown). The MM AnMBR exhibited a slightly depressed
pH and increased volatile acid concentration during the ﬁrst 14 days of operation (Figure
4.15 and 4.16). GCMS analysis (Appendix E) was conducted for the ﬁnal day of Phase 2

operation and is presented in Table 4.10.

81

Table 4.10 Volatile fatty acid data

 

 

 

 

Acetic Propronrc Butyrrc Isobutyric Total,
Acid mg/L Ac'd’ ' Ac'd’ Acid m IL mg/L
9 mg/IJ M14 9 g
SM AnMBR 192 85 15 1 299
MM AnMBR 182 75 14 0.4 271
CMD 199 91 16 38 344

 

 

 

 

 

 

 

Acetate represented the majority of the VFA present in all three digesters. As detailed
in Chapter 2, Section 2.3, hydrogen consuming methanogens must work in syntrophic
cooperation with fatty acid oxidizing fermenting bacteria to maintain low H2
concentrations. When H2 concentrations increase, a shift towards more reduced products
such as propionate, as opposed to acetate, occurs (Ianotti 1973; Bryant 1979). The
syntrophic relationship requires spatial proximity between the fatty acid oxidizing
fermenting bacteria and the hydrogen consuming methanogens (McCarty and Smith
1986). Because a build—up of propionate was not observed in either of the AnMBRs
(concentrations are consistent with the CMD control), there is not an apparent break-

down in syntrophic activity due to the shearing impact of the pump/membrane system.

82

 

pH

+ Feed HQ—SM AnMBR

'"X'- MM AnMBR

-x—

CMD

 

0 510152025303540

Figure 4.15 Phase 2, pH

300

250

200

150

100

Volatile Acid, mg/ L as Hac

50 ‘3

1.1“-.. .. H-

 

 

45 50 55 60 65 70 75 80 85 90 95 100

Days

 

Days

Figure 4.16 Phase 2, volatile acid concentration, MM AnMBR

83

 

 

30

Figures 4.18 and 4.19 illustrate a steady increase in the VS and COD concentrations
respectively for the MM AnMBR beginning at day 1 through 17. At the start-up of the
MM AnMBR, the timer operation was set at 2 minutes x 24 times per day. The
frequency of this initial setting resulted in a permeate generation rate that approximated
the feed rate and, as a result, there was very little biomass wasted from the digester,
consequently the digester VS/TS/COD concentration increased at a faster rate than the
SM AnMBR. At day 16, the timer operation was adjusted to 1 minute x 24 times per day
to more closely approximate the TS concentration of the SM AnMBR.

The volatile acid titration procedure (Appendix B) identiﬁes volatile acids present in a
sample; however, at stable operation, the titration procedure was non-detect for the
digester systems of this research. GCMS indicated there were volatile acids present, even
during stable operation (Table 4.10) suggesting that GCMS is a more accurate method for

quantifying volatile acids at low concentration.

84

- El -SM AnMBR

+MM AnMBR

00.x... CMD

 

6.0

5.0 e

01R, g VS/l/day

 

 

 

 

 

 

0.0 .1..-........,._-_..._..-...._._..,,.....__.-_.r..____ r "l’""" -..
0 10 20 30 40 50 60 70 100
Days
-El -SM AnMBR +MM AnMBR ---X--- CMD
8.0 ,
l
7.0 *4, “’51
r a. '8 _' o

6.0 ' 1 \p
> 1.1 ’5 i i
3 s 0 "4 ‘v
E .
a - . 9" ... ’ ‘ 3',
8 4.0 “ . '1“ ’ a. W 6.,-
” ° y
1‘ 3.0 ' In T 7'. l
-l i
O \ -

2.0 -< - I

1.0 — d]

0_0 . - . 1.--,” .. 1..-, ..-. . - . _ ,. . _. 11.----. .-....-1.-.....--_-._...-_--.1.1-1-.. -T- --.--- 111,1...

 

3 8 13182328

 

 

33 38 43 48 53 58 63 68 73 78 83

Days

Figure 4.17 Phase 2, organic loading rate, COD and VS

85

- A- SM AnMBR Permeate + MM AnMBR Permeate m)K--- CMD
—O—SM AnMBR --El--MM AnMBR +Feed
4.50

 

4.00 -

3.50 -

3.00 ~

2.50 ~

VS, 96

2.00 *~

1.50 -

 

 

 

 

 

Figure 4.18 Phase 2, VS concentration

Figure 4.17 presents the organic loading rates for VS and COD for all three digesters and
shows a signiﬁcant increase in loading rate at day 50 which resulted in similar increases
in efﬂuent VS and COD (Figures 4.18 and 4.19) as well as methane production (Figure
4.22). Volatile acids were not tested during this period; however, pH remained stable
(Figure 4.15) indicating the digester systems did not have any problem processing the

increased loading rate.

86

- El - SM AnMBR

+ MM AnMBR ...x... CMD

 

OLR, g COD/leay
I
J

3.0 w i A
2.0 ~ I

1.0 1

."
‘1

.“

 

 

0.0 T l I W '

T T T T

3 8 13 18 23 28 33 38 43 48 53 S8 63 68 73 78 83

Days

Figure 4.19 Phase 2, COD concentration

The ﬂux data for the SM AnMBR and MM AnMBR are presented in Figures 4.20 and

4.21 respectively. The SM system averaged 53 L/mZ/hr and the MM Membrane

averaged 24 L/mZ/hr. Consistent with the previous work presented in Chapter 3, Section

3.1, a declining ﬂux rate follows an increasing digester total solids concentration

(Anderson, Saw et a1. 1986; Beaubien, Baty et al. 1996; Brockmann and Seyfried 1996).

The SM AnMBR system was cleaned on day 16 and there was an immediate increase in

ﬂux rate from approximately 33 L/mZ/hr to 104 L/mz/hr followed by a slow decay in ﬂux

rate between days 18-68 when the ﬂux rate was 13 L/mz/hr and the membrane was

87

 

-T. .-.. . ......r___~_ 1-

cleaned again. Flux stability and cleaning protocol are discussed in greater detail in
Chapter 6, Section 10. A leaking valve on the day 16 cleaning resulted in low pH
cleaning solution (approximately pH = 4.5) entering the digester tank. Approximately
25% of the digester volume was displaced with the cleaning solution. This resulted in a
reduction in the digester total solids concentration (Figure 4.20) and a reduction in gas
production (Figure 4.22). The SM system recovered quickly and within 8 days was
producing gas consistent with the CMD. The low pH cleaning solution was acetic acid-
based (speciﬁc detail is presented in Chapter 6, Section 9).

ME! SM AnMBR +Digester TS
120 , 6

100

 

 

 

. s
l

l as
80 'J ' 4 v?
c. I!
i E
To
} 60 ~ - 3 g
2 ':
2 o
“ 11
40 2 an
o

20 - 1

0 5101520253035404550556065707580859095

Days

Figure 4.20 Phase 2, SM AnMBR ﬂux rate and digester TS concentration

88

“El MM AnMBR + Digester TS

 

 

, 7
l
16
l
l
l
l~-5
g as
l u?
E l 3
NE .' -4 ,2
E . .3
D 8
§ ' Cl 3 3
1 15 [a @410 . :4,
' M
5&1 ' 2 '5
10 4 l
‘ n
l1
5"! l
.
z
o 1 . . , 0

 

0 5101520253035404550556065707580859095

Days

Figure 4.21 Phase 2, MM AnMBR ﬂux rate and digester TS concentration

Figure 4.21 shows an initial increase in ﬂux for the MM AnMBR from day l to 22.
As detailed above, the initial timer operation was set too high. To counter the very rapid
increase in digester TS concentration, the permeate was returned to the digester between
days 18 and 22 and the digester contents were wasted consistent with the feed rate
causing a decline in the digester TS concentration. At day 23, normal operation was
resumed and the general trend of decreasing ﬂux against increasing digester TS also
resumed.

The MM AnMBR required approximately 12-15 days of operation to reach a gas

production rate consistent with the other two digesters (Figure 4.22). This lag was most

89

likely due to the change in operating conditions from Phase 1 to Phase 2 and is consistent

with the period of volatile acid production identiﬁed in Figure 4.16.

"'El"SMAnMBR +MMAnMBR -)(- CMD

 

Methane Production, L CH‘lkg VS/day

 

 

 

0.ir l 1 I i l I v I . r . .’ l I t I I l
1 6 11 16 21 263136414651 566166 7176818691

Days

Figure 4.22 Phase 2, methane production

Initially, there appeared to be a signiﬁcant difference in gas production for the MM
AnMBR (data not shown); however, on day 59 it was recognized that the wet tip meter
measuring biogas production for the MM AnMBR was double counting gas production.
A review of the data indicated the double counting started around Day No. 14. The data
was corrected by dividing the meter reading by 2. The MM AnMBR membrane was
cleaned at day 74. Following this cleaning, its gas production lagged the other digesters
for the balance of the experiment. The MM AnMBR’s module that houses its 7

90

membranes is considerably larger than the SM AnMBR module. When cleaning the
membranes, the ﬁnal step was to ﬂush the membrane with clean water for 15 to 30
minutes (Chapter 6, Section 9 provides membrane cleaning detail). However, due to the
greater volume of the MM AnMBR module, there was probably still residual cleaning
solution in the module. This may have led to a decline in performance of the MM
AnMBR digester.

It was anticipated that at low cycle time, the MM AnMBR (cycle time = 6/day) would
outperform the CMD and the SM AnMBR (cycle time = 27/day). However, based on the
fact all systems produced equal amounts of methane, the following conclusions were

made.

1. The potential advantage of operating at a higher VS concentration for the SM

AnMBR is off-set by the impact of a high cycle time (cycle time = 27).

2. The potential advantage gained by operating at a low cycle time and higher VS
concentration for the MM AnMBR (cycle time =6) is off-set by the high degree of
turbulence in the system due to the torturous path the biomass must negotiate.

The torturous path is a design issue unique to the nature of the pilot-scale setup.
Figure 5.7 presents a picture of the manifold that was constructed to direct ﬂow in

a series fashion through the membrane elements. This issue is further explored in

Phase 3.

91

3. Methane production is equal for both SM AnMBR and MM AnMBR because

methane production is independent of the operating conditions of these systems.

4. CMD methane production was independent of the two mixing conditions tested
and, referencing the conclusions of Phase 1, methane production is independent of

cross-ﬂow velocity at 4.5 m/s or less.

4.3.3 Phase 3

For the final phase of testing, the SM AnMBR was converted back to a SMD
(resembling a complete mix digester with the permeate returned to the digester tank) and
operated at a cycle time of 27 (consistent with Phase 2 operation of the SM AnMBR).
The purpose of this experiment was to evaluate whether the higher biomass concentration
of the SM AnMBR provided a methane production advantage. Phase 2 results indicated
that that the MM AnMBR, operating at a cycle time of 6/day, did not have a methane
production advantage over the CMD or the SM AnMBR (cycle time equaled 27/day). It
was theorized that the MM AnMBR may have been negatively impacted by the system
design. Therefore, in Phase 3, the MM AnMBR was removed and replaced with four
13.9 mm diameter x 3000 mm PVC pipes (MMED) to mimic a design that might be used
in a full-scale application. This design provided a less torturous path for the biomass to
negotiate compared to that of the MM AnMBR which, in order to modify a full-scale
membrane/module for pilot-scale work, necessitated the use of a manifold to route ﬂow

between membranes (Figure 4.7). Table 4.11 presents a summary of the operating data

92

and Table 4.12 a summary of the water quality data for each of the digesters with

plus/minus values indicating the standard deviation for each water quality result.

Table 4.11 Summary of operating data for Phase 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter‘ SMD MMED CMD
UF circulation rate, LPM 42 44 NA
Cross—ﬂow Velocity, m/s 4.3 3.7 NA
Flux, L/mZ/hr 48 NA NA
Avg. TS in Reactor, % 3.0 3.2 2.5
VS destruction, % 35 38 37
COD destructionl NA NA 41
L CH4/kg vs fed/dayz 275 250 280
CH4 concentration, % 70 69 67
Cycle time, day'1 27 6 1
HRT, days 12 12 12
SRT, days 1 2 1 2 12

 

 

|COD destruction not reported for SM AnMBR with 100% recycle or Four 3000 mm Pipes due to limited
number of COD data points.

2Actual conditions

Table 4.12 Summary of water quality data for Phase 3

 

 

 

 

 

 

 

 

 

 

Parameter Inﬂuent SMD MMED CMD
COD, mg/L 42,200117,500 32,80018,850 34200110870 26,20015,l80
TS, % 3.9113 3,010.8 3211.0 2.5105
VS, % 2.4109 1.9105 2.0107 1.5103
pH 7.031033 7.671005 7.621007 7.721005

 

93

 

 

Figure 4.23 presents the COD and VS organic loading rate data and Figure 4.24 presents
the feed and efﬂuent COD and VS data for Phase 3. The SMD and MMED were
previously operated as AnMBRs. As a result, the initial digester VS and COD
concentrations were much higher for these two systems at the start of Phase 3 and
required approximately 25 days of operation to reach a point where the VS and COD
concentrations for these digesters were consistent with the CMD. Despite the initial VS
differences between the systems, the gas production (Figure 4.26) for all three systems
was very similar for the duration of the Phase 3 testing. At day 31, there was a spike in
the VS and COD concentration of the manure due to farm operations (Figures 4.23 and

4.24) resulting in a decrease in methane production and pH (Figures 4.25and 4.26).

94

mEluSMD +MMED -x- CMD

 

 

 

 

 

 

 

4.0
3.5
>
II
'0
:l‘
3
>
u
a?
d
O
0.0 "LT—u“ ! T" ' r T‘ T ‘ .—-—--—.._T ..-..__-_...- ....-.-.- 7-..... ..- _
0 5 10 15 20 25 30 35 40

Days

+SMD cacao. MMED -X -CMD

 

7.0 ~
6.0
5.0 l

4.0 --

OLR, g COD/Liday

 

2.0 ~

1.0 .4

 

I I

Days

Figure 4.23 Phase 3, organic loading rates for VS and COD

95

 

+Feed *SMD -)(-MMED "-IHCMD

80,000

 

70,000

60,000 -

”.41.“... ... . ,. _ _ . .... .

50,000 4 ,
l ‘ ‘X.
40,000 ‘ .

......l.......,..... I“; I.
....... \J‘

COD, mg/l.

30,000 4

 

20,000 4 --..... “

10,000

30

Days

35

 

40

 

4.5 -!—Feed —A-SM0 mX-oMMED -)K-CMD

4.0 -

3.5 -

3.0 4

2.5 *

 

VS, 96

2.0 -r

 

1.5

1.0

0.5

......L... -..... -..-_. .

 

0.0

.,-

*1
l

,

 

 

0 5 10 15 20 25 30

Days

Figure 4.24 Phase 3, COD and VS concentration

96

35

40

 

45

—S—|nﬂuent mop-SMD -)(- MMED —)l(—CMD

 

 

 

8.00 ,
7.80 3 r n.
. A “ - ‘
an L Axﬁ A 3. .‘ﬁ . ) .1 n L
7.60 l I
n
7.40 l
:I:
O.
7.20 1 r1
7.00 1 ll
c _ r:
E n 2'5: u
6.80 “ u___‘ 5:1
‘ V ‘ . L'~
, u L .4
0 5 10 15 20 25 30 35 40

Figure 4.25 Phase 3, pH

97

 

+SMD "'AuMMED ->(- CMD

 

600 z
500
400
300 -

200 .

Methane Production, L CHJkg Vs/day

100 -

 

 

 

Days

Figure 4.26, Phase 3, methane production

However, both pH and methane production quickly recovered as the systems responded
to the change in organic loading. The pH impact on the MMED was more dramatic than
for the other two digesters, suggesting this system was not as stable and more prone to an
increase in VFA when perturbed, though there is not a clear explanation for this
condition.

Tables 4.13-4.15 present acetic, propionic, butyric and isobutyric acid concentrations
for the three digester systems as determined via gas chromatography mass spectroscopy
(GCMS, protocol outlined in Appendix E). Total acetic, propionic, butyric and isobutyric
acid concentrations were similar during the periods tested. The results are very similar to
those in Phase 2 and indicate stable operation.

98

Table 4.13 Volatile fatty acid data, SMD

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Acetic 1’ng :o‘nic Bxgsic Isobutyric Total,
Acid, mg/L mg/I: mg/I: Acid, mg/L mg/L
Day 2 187 82 13 0 282
Day 7 132 47 23 6 208
Day 12 153 53 19 0 225
Average Value 168 67 18 2 254
Table 4.14 Volatile fatty acid data, MMED
Acetic P12: :Znic Biggie Isobutyric Total,
Acid, mg/L m g/I: m g/I: Acid, mg/L mg/L
Day 2 208 89 14 0 311
Day 7 169 77 18 36 300
Day 12 200 97 16 43 356
Average Value 190 85 16 20 310
Table 4.15 Volatile fatty acid data, CMD
Acetic Prxpignic Bxgsic Isobutyric Total,
Acid, mg/L m g/I: m g/L, Acid, mg/L mg/L
Day 2 296 89 14 27 427
Day 7 184 70 14 0 268
Day 12 193 71 13 20 297
Average Value 218 80 14 22 334

 

 

 

 

 

 

 

The SMD experiment was designed to examine the impact of operating at a biomass

concentration consistent with that of a complete mix digester compared to the elevated

biomass concentration and extended SRT of the SM AnMBR of Phase 2. A cycle time of

99

 

 

 

27 was maintained (as in Phase 2) and the SMD produced 275 L CH4/kg VS/day. The
SM AnMBR of Phase 2 produced 251 L CH4/kg VS/day. To account for differences in
manure gas production potential between Phases 2 and 3, the CMD acted as a control and
produced 247 L CH4/kg VS/day during Phase 2 and 280 L CH4/kg VS/during phase 3.
Based on these methane production rates, there was virtually no difference between the
SMD and SM AnMBR.

The MMED lagged the SMD and the CMD by about 10% during this period, a
consistent trend beginning on day 74 of Phase 2 (same biomass for MMED and MM
AnMBR), when the MM AnMBR membrane was cleaned. A potential reason for the
lower than anticipated methane production rate may be a residual effect of this cleaning
operation as previously discussed in Section 3.2 of this chapter. However, the operation

of all three digesters resulted in similar VS destructions of 35%, 38% and 37% for the

SMD, MMED and CMD respectively.

Based on the results of Phase 3, the following conclusions are made:
1. At a HRT of 12 days and a cycle time of 27, the AnMBR conﬁguration that
provides for an extended SRT compared to a complete mix conﬁguration (HRT =

SRT), did not provide a gas production advantage.

2. There does not appear to be a pronounced advantage or disadvantage to operating
with the less turbulent condition of Phase 3 (MMED) compared to the MM
AnMBR of Phase 2, suggesting its membrane/manifold conﬁguration did not

have a negative impact on gas production.

100

4.4 Summary

There was not an increase in methane production associated with operating at the
higher biomass concentration of the AnMBR system. Nor was there an apparent biogas
production difference between an operating cycle time of 27/day and 6/day. All three
systems operated at volatile acid concentrations that were not detectable via the titration
procedure. A GCMS technique was used to measure the concentrations of acetic,
propionic, butyric and Isobutyric acid for Phase 2 and Phase 3 showed all three digesters
operated in the range of 200 mg/L to 450 mg/L. Acetic acid was the predominate VFA in
all three systems suggesting the systems were stable and that syntrophic activity was not
disrupted. Chapter 5, “AnMBR Metabolic Evaluation of Cycle Time”, explores the
metabolic level interactions in an effort to explain the observed pilot-scale results

presented in this chapter.

101

Chapter 5
AnMBR METABOLIC EVALUATION OF CYCLE TIME
5.1 Introduction

The goal of this phase of the research was to develop a better understanding of the
impact of cycle time on microbial activity. Much of this effort focused on three sets of
respirometer experiments. These experiments were developed to allow for comparison of
biomass from the AnMBRs and the CMD with the objective of using activity
measurements to characterize and compare the microbial pathways associated with
digester conﬁgurations described in Chapter 3 and 4.

Acetate was used as a substrate to evaluate the activity of acetate consuming
methanogens by measuring methane production of a known quantity of digester biomass
provided with a known quantity of substrate. Referencing the ﬂow of electrons in Figure
2.1, acetate is converted directly to methane via acetate consuming methanogens. James
et al. (1990) outlined a methodology for evaluating speciﬁc methanogenic activity (SMA)
using a Warburg respirometer and sodium acetate as the substrate. The SMA test
provides a basis for evaluating the methane generating potential for active biomass. The
speciﬁc methanogenic activity is estimated from the methane production rate or the
substrate depletion rate and the amount of sludge present (V andenbe.L, Lentz et a1. 1974;
Owen, Stuckey et al. 1979; Valcke and Verstraete 1983; Dolﬁng and Bloemen 1985;
James, Chernicharo et al. 1990; Soto, Mendez et a1. 1993). As previously discussed in in
Chapter 2, Section 5, acetate is fermented directly to methane and accounts for
approximately 70% of total methane production, therefore, as a test substrate, acetate

provides a very good indication of the maximum methane generating potential of a given

102

biomass. The results of this test are sometimes used to optimize organic loading rate for
a faster and more reliable start-up (James, Chernicharo et al. 1990; Soto, Mendez et al.
1993). The SMA test is also used to evaluate ongoing process performance (Soto,
Mendez et al. 1993). It provides a maximum gas generation potential against which
actual performance can be compared.

Formate was used as a surrogate for hydrogen to assess the metabolic activity of the
hydrogentrophic methanogens. Dolﬁng and Bloemen (1985) illustrated that
methanogenic activity on H2-CO2 was comparable with the activity on formate. Dolﬁng
and Bloemen (1985) also indicated that the relative contribution of mixed function
methanogenic biomass (for example, methanosarcina spp) can be estimated by
comparing the activities associated with a mixture of formate (as a hydrogen surrogate)
and acetate and comparing with formate and acetate individually. The presence of
Methanosarcina will result in a lower activity on hydrogen plus acetate as compared to
the sum of the activities on acetate and formate individually because it has been shown to
preferentially degrade hydrogen over acetate at high substrate concentrations (Dolﬁng
and Bloemen 1985).

Propionate was used as a substrate to evaluate the syntrophic activity of the biomass to
degrade propionate to acetate, H2 and CO2. Referencing Figure 2.1, propionate must ﬁrst
be degraded to acetate, H2, and CO2 prior to the occurrence of methanogenesis. As
described in Chapter 2, Sections 2.4 and 2.5, propionate degradation is endergonic under
standard conditions and requires syntrophic interaction between fatty acid oxidizing
bacteria and hydrogen consuming methanogens. An evaluation of the biomass’ ability to

degrade propionate provides insight relative to the existence of syntrophic cooperation.

103

It is also of interest to compare the acidogenic activity of the biomass from the CMD
and from the AnMBR. Similar approaches are presented in the literature using glucose as
the substrate to measure acidogenic activity (Soto, Mendez et al. 1993; GarciaMorales,
Nebot et al. 1996). Padmasiri et a1. (2007) reported a decrease in methanogenic activity
resulting from a build-up of volatile fatty acids and theorized that this was a direct result
of an increase in the rate of hydrolysis caused by the high shear environment of the
AnMBR, occurring at a rate that exceeded the metabolic capacity of the methanogens.
Based on this theory, comparison of the acidogenic activity was undertaken to explore the
potential of the AnMBR to select a more robust community of primary fermenters
compared to the CMD.

Table 5.1 presents a summary of the substrates with their utility for the respirometer

experiments.

Table 5.1 Substrates used for metabolic testing

 

 

 

Substrate Objective
Acetate Used to measure activity of acetate consuming methanogens
Forrnate Used as a hydrogen surrogate to assess hydrogentrophic activity

 

Formate/Acetate Methanosarcina will preferentially consume H2 prior to acetate
when both present in high concentrations.

Propionate Used to compare syntrophic activity

Glucose Used to compare acidogenic activity

 

 

 

 

 

In addition to the respirometer work, most probable number enumeration was
conducted to allow for comparison of the viable organisms present in the SM AnMBR,

the MM AnMBR and the CMD for Phase 2. Microscopic evaluation was also performed

104

 

n 9
D
:0er

Lie .1".

MS

pro

011}

Z.)-

3111

for these samples to gain a general sense of the spatial relationship of the organisms of

the AnMBRs compared with the CMD.

5.2 Materials and Methods

5.2.] Experimental Setup

The respirometer setup (Challenge Technology AR-200, Springdale, AZ) consisted of
sixteen (16) reaction vessels of 675 ml, each with a working volume of 600 ml. Biogas
ﬂow measuring cells were dedicated to each respirometer reaction vessel. Gas ﬂow data
was logged automatically using a computer software interface. The reaction vessels were
provided with gas-tight screw membrane caps allowing insertion of a needle in the
headspace for removal of both gas and liquid from the reaction vessel without impacting

ongoing experiments.

5.2.2 Methanogenic Activity Setup

To ensure a reliable estimation of a speciﬁc biochemical activity, the biomass should
be present in the test bottles such that it is the limiting factor in the reaction to be studied
(Dolﬁng and Bloemen 1985; Chynoweth, Turick et al. 1993). Typical values used in
previous work ranged between 0.3 to 2.5 g HAc/ g VSS (Valcke and Verstraete 1983;
James, Chernicharo et al. 1990; Soto, Mendez et a1. 1993), although values ranged as
high as 30 g HAc/ g VSS (Dolﬁng and Bloemen 1985). Soto et al. (1993) proposed a
VFA concentration of 2.0 and 0.5 for HAc and HPr respectively.

In this research, approximately 1.2 g acetate, 1.2 g formate, 1.2 g acetate plus formate

and 0.3 g propionate were selected for methanogenic testing with the objective of

105

335161

0. 322'

1 “1.
min

1.12
N

resp

Ac -

:JI
! J

it

an
iti

la:

achieving substrate to VSS ratios of 0.5 to 1.0 for acetate , formate and the combination
of acetate and format and, for propionate, between 0.20 to 0.50. For a given respirometer
bottle, the biomass was diluted to a VSS concentration that resulted in approximately 1.2-
1.8 g VSS/bottle. Using Table 5.3 as an example, 1.2 g of acetate was added to a
respirometer bottle that contained 1.38 g VSS, such that the Ac/V SS ratio became 1.2 g

Ac +1.38 g VSS = 0.87 for the CMD.

5.2.3 Acidogenic Activity Setup

Glucose was used for acidogenic activity measurement because it is considered as the
main intermediate pathway of anaerobic digestion of carbohydrate complex organics
(Soto, Mendez et a1. 1993). This is also appropriate for manure as cellulose and
hemicellulose comprise a signiﬁcant fraction of dairy manure (Amon 2007) and degrade
to glucose. The half-saturation constant for acidogenic bacteria (K5) is about 0.2 g COD/L
(Henze and Harremoes 1983). A glucose concentration of 1.2 g/L was used in this
research to ensure the initial substrate concentration was signiﬁcantly greater than the
half-saturation constant as recommended by Soto et al. (1993) and was close to the value

of 1.5 g/L used by Soto et al. (1993).

5.2.4 Dilution Media Composition
The composition of the nutrient media solution used in this work is shown in Table 5.2
and was adapted from Garcia-Morales et al. (1996). The media composition

recommended by Valcke and Verstraete (1983) was very similar and was also used by

James et al. (1990).

106

Table 5.2 Dilution media composition

 

 

 

 

 

 

 

 

Chemical Acidogenic Activity, g/L Methanogenic Activity, g/L
Yeast extract 0.2 0.2

NaHC03 1 NA

K2HPO4 NA 1

KH2PO4 NA 2.5

NH4C1 NA 1

MgCl2 NA 0.1

 

 

 

The acidogenic activity dilution media uses a sodium bicarbonate buffer to maintain an
alkaline pH, which is favored by acidogenic bacteria. The Methanogenic activity dilution
media uses a phosphate buffer to maintain the pH close to neutrality, which is favored by
methanogens. Macro nutrients were provided for methanogenic grth but were not

considered necessary for the acidogenic growth.

5.2.5 Operational Procedure

A schematic of the respirometer setup is shown in Figure 5.1. The biogas generated in
the respirometer bottle was bubbled through a 1 M solution of potassium hydroxide
(KOH) to scrub the CO2 from the gas prior to measuring the generated volume for the
CMD/AnMBR and Phase 2 testing. KOH scrubbing was not used for the Phase 1 test.
Instead, biogas was measured directly and gas chromatograph analysis used to determine
methane generation rate. A control was run for each digester and the gas generated from

the control bottle was subtracted from the gas generated by each of the bottles testing the

107

 

various substrates. The Operational procedure was adapted from James et al. (1990) and

presented below.

10.

ll.

12.

. Determine the volatile suspended solids concentration of the sludge to be

analyzed prior to the start of the respirometer study.

Prepare the dilution media solution per Table 5.2

. Dilute inoculum with media solution to desired VSS concentration and introduce

into respirometer bottles.

Flush respirometer vessel headspace with nitrogen.

Seal respirometer vessels and connect to gas measuring cell.

Initiate water circulation in water bath and set temperature to 35° C.

Activate stir mechanism at a rate of 60 RPM.

Add substrate to bottles following an acclimation period of approximately 12
hours.

Continuously measure gas production for approximately 100 hours.

Measure glucose concentration every 1 to 2 hours for (acidogenic test bottles).
The samples for glucose analysis to be removed using a needle/syringe setup via
the reaction vessel’s membrane cap.

Record methane production every 10 minutes.

Make periodic measurements of gas using a gas chromatograph to ensure all CO2

is being removed by the potassium hydroxide.

108

Biogas Sampling Port

3 I

7 1 i ;, . l Flow Measuring .
‘ 7 7 7 7 i I “17 1 Cell - -
' ll '7”— ——“ ’ 7" Computer for Data
‘ } Logging
1 1 l l

l i g
l 2 l L
l l l Potassium
l L . 1 l Hydroxide

l ‘~~ espirometer Bottle

Figure 5.1 Respirometer Setup

5.3 Analytical Methods

5.3.1 General

General analytical methods used in this chapter were described in Chapter 3, Section 2.1.

Glucose concentration was determined using a glucose assay kit (Sigma Aldrich, GAGO-
20, St. Louis, MS). Most probable number enumeration was used to assess the estimated
number of viable cells in the digester biomass and speciﬁc detail can be found in

Appendix C.

5.3.2 Microscopic Observations
A scanning electron microscope was used to view the biomass from Phase 2 of the

research which included the SM AnMBR operated at a cycle time of 27, the MM

109

AnMBR operated at a cycle time of 6 and the CMD control. Samples were taken from
the systems on Day 92 and ﬁxed at 4°C for ‘/2 hour in 4% glutaraldehyde buffered with
0.1 m sodium phosphate at pH 7.4. One drop of 1% poly-L—lysine (Sigma P1399) was
placed on a plastic petri dish and a 12 mm round glass coverslip was placed on top of the
drop and allowed to stand for 5 minutes. The coverslip was removed and gently washed
with several drops of water and drained but not allowed to dry. One drop of the cells
ﬁxed in suspension was placed on the side of the coverslip which previously faced down.
The suspension was allowed to settle ten minutes before it was gently washed with
several drops of distilled water. Next, the coverslip was placed in a graded ethanol series
(25%, 50%, 75%, 95%) for ﬁve minutes in each with three ﬁve minute changes in 100%
ethanol (Klomparens, Flegler et al. 1986). The coverslips were mounted with epoxy on
aluminum stubs and coated with osmium. The preparatory work described above was
conducted by personnel in the Center for Microscopy at Michigan State University. A

J EOL 6400 scanning electron microscope in the Center for Microscopy at Michigan State

University was used for viewing samples.

5.4 Results and Discussion

Three sets of respirometer experiments were performed on the biomass from various
digester conﬁgurations of Chapter 3 and 4. The ﬁrst set was conducted on the biomass
from the combined CMD/AnMBR described in Chapter 3, Section 2.5 and referenced in
this chapter by the same heading. The second and third set of respirometer experiments
were conducted on the biomass from the digester systems detailed in Chapter 4, Section

2.2 and 2.3 and described as Phase 1 and Phase 2.

110

.2 \\§

 

5.4.1 CMD/AnMBR Respirometer

The CMD/AnMBR respirometer work compared the biomass from the AnMBR to the
CMD (outlined in Chapter 3, “Utility of an Anaerobic Membrane Bioreactor”). For this
conﬁguration, the pilot CMD efﬂuent was the AnMBR inﬂuent (See Chapter 3, Figure
3.3). There was a signiﬁcant difference in the gas production for the two systems. The
CMD averaged 176 L CH4/kg VS fed/day and the AnMBR averaged 82 L CH4/Kg VS '
fed/day (Chapter 3, Table 3.5). This difference is not surprising considering the CMD
was converting the readily degradable substrate and its efﬂuent was the AnMBR inﬂuent.
The respirometer tests provided a basis for evaluation by pairing equal concentrations of
biomass from each digester with equal substrate concentrations in the reaction vessels.
Table 5.3 outlines the substrate to biomass ratios used for the respirometer experiments
described in Section 5.2.2 with results presented in Table 5.4. Figures 5.2-5.5 present
the respirometer methanogenic results in graphical format. The CMD biomass produced
methane at a higher rate than the AnMBR biomass for all methanogenic substrate. Figure
5.6 presents the glucose consumption data for the AnMBR/CMD system. The
acidogenic testing illustrated that the biomass from the AnMBR degraded glucose at a

higher rate than the CMD.

111

1 \\§

 

Table 5.3 CMD/AnMBR respirometer feed ratios, g substrate/g VSS

 

 

 

 

 

Substrate1 CMD AnMBR
Acetate 0.87 0.78
Propionate 0.58 0.52
Formate 0.87 0.78
Acetate and Formate 0.87 0.78

 

 

 

 

 

lSubstrate was sand-separated dairy manure

Table 5.4 Summary of CMD/AnMBR respirometer results, mL CH4/g VSS/hr

 

 

 

 

 

 

 

Substratel CMD AnMBR
Acetate 2.22 1.81
Formate 4.48 2.59
Acetate + Formate 6.74 2.28
Propionate 0.98 0.24
Cycle time, day'1 NA 56
L CH4/kg VS/day2 176 82

 

 

 

 

 

Substrate was sand-separated dairy manure
2 . .
Actual conditions

112

' \xi

'1
j
...;g.

 

The acetate activity of the CMD was similar to that for the AnMBR.

"'13" AnMBR +CMD

 

250 r

200 -«

150 ~

100

ml. Methane Produced

50-

 

 

 

 

0 S101520253035404550556065707580859095

Hours

Figure 5.2 CMD/AnMBR, acetate

113

°°°°Ei AnMBR +CMD

60

 

 

ml. Methane Produced

 

 

 

Figure 5.3 CMD/AnMBR, propionate

+AnMBR oovoo CMD

 

Methane Produced, ml.

 

 

0 10 20 30 40 50 60 7O 80 90 100

Hours

Figure 5.4 CMD/AnMBR, formate

114

 

"'El" AnMBR +CMD
140

 

120
100 --
80

60-

Methane Produced, ml

40

20

 

 

 

Hours

Figure 5.5 CMD/AnMBR, acetate + formate

+ CMD onA" AnMBR
2,000

 

1,800 ~

1,600 1

 

1,400

1,200 1

.w

.A-" _.

1,000
800

600

Glucose concentration, rag/L

400 -‘

 

200 -'

4.1....“ .. ._... .._...

O . ' ° ’ A33. 3'... 3.3. {A333 ° ° ' ' °

C

0 2 5.7 7.5 8.5 9.25 10 13.25

Hours

Figure 5.6 CMD/AnMBR glucose consumption per mass VSS

115

 

_ .. .. .1..- . . .. .. .. _ .. . _. __ W‘- .. .. . .._..... . .11....” --_.. -... .. - . .1... - .... ._ .._.- -..1 1....-......-.... _ . 1' “TNT 1....-.

' kx'x

 

However, of particular note, there was a signiﬁcant difference between the rate of
methane production for the CMD on propionate compared to the AnMBR (Figure 5.3).
Propionate requires a syntrophic interaction between acetogenic bacteria and
hydrogentrophic methanogens (Kaspar and Wuhrmann 1978; Boone and Bryant 1980;
McCarty and Smith 1986). The syntrophic interaction appears to be signiﬁcantly greater
for the CMD. This suggests the high shear environment of the AnMBR negatively

impacted the juxtaposition between the acetogenic bacteria and the hydrogentrophic

AI.
....

methanogens. The AnMBR also exhibited a lower activity for formate, consistent with a

breakdown in syntrophic interaction (Figure 5.4). Lastly, referencing Table 5.4, the
activity of formate and hydrogen individually are approximately equal to the activity for
formate and hydrogen combined, suggesting that Methanosarcina like organisms did not
make a signiﬁcant contribution to the overall acetate consuming methanogenic
population present for the CMD (Dolﬁng and Bloemen 1985). Following this same logic,
it appears that Methanosarcina like organisms are present in the AnMBR.
Methanosarcina cells grow as cocci whereas Methanosaeta cells grow as long ﬁlaments
in anaerobic biomass (Hoffmann, Garcia et al. 2008). Due to their morphology, the
Methanosarcina cells are likely to experience a competitive advantage in a high shear
environment. Referencing Chapter 3, Figure 3.13, the volatile acid concentration of the
AnMBR (CMD efﬂuent was AnMBR inﬂuent) efﬂuent averaged 232 mg/L as HAc as
determined via a titration procedure (Appendix B). It is generally accepted that under
conditions of high acetate concentration, Methanosarcina spp. will outcompete
Methanosaeta (McMahon, Stroot et al. 2001; Hoffrnann, Garcia et al. 2008). During this

same period, the CMD exhibited a volatile acid concentration of zero (VFA method

116

4 (xi

 

located in Appendix B), suggesting that the AnMBR environment was more selective
towards Methanosarcina spp.

Figure 5.6 illustrates the acidogenic activity for the two digester systems as
determined by the rate of glucose consumption. This rate was markedly higher for the
AnMBR compared to the CMD. This ﬁnding was also observed at the time of the
methanogenic testing (data not shown). To conﬁrm, fresh biomass samples were taken
from each of the digesters and re-tested. The results (shown in Figure 5.6) were
consistent with the initial testing. It is not clear why the AnMBR biomass exhibited a
higher acidogenic activity than the CMD. Padmasiri et a1. (2007) suggested the high
shear environment of an AnMBR treating swine waste increased the rate of hydrolysis,
thus increasing the rate of fermentation. Considering this theory, the AnMBR may have
selected for a more robust acidogenic population due to an increased rate of hydrolysis

compared to that of the CMD.

5.4.2 Phase 1 Respirometer

Respirometer experimentation was completed for the biomass for the Phase 1 digester
systems. In place of CO2 scrubbing, methane content was measured via gas
chromatograph. As described in detail in Chapter 4, Section 2.2, the digesters are
referenced as SMD, MED and CMD. All three systems were operated at approximately
the same cycle time of 6. The CMD was mixed with a centriﬁrgal pump that was
energized on the same schedule as the circulation pumps for the SMD and MED. Figure

4.13 illustrates that the methane production for the SMD and MED lagged the CMD at

117

 

startup; however, all three systems produced methane at nearly the same rate once the
SMD and MED reached stable operation.

Tables 5.5 and 5.6 present the respirometer feed ratios and results for Phase 1
respectively. Figures 5.7—5. 10 present the respirometer methanogenic results in graphical
format for the SMD, MED and the control CMD and Figure 5.11 presents the glucose

consumption data for this period.

Table 5.5 Phase 1 respirometer feed ratios, g substrate/g VSS

 

 

 

 

 

Substrate] CMD SMD MED
Acetate 0.70 0.73 0.83
Propionate 0.18 0.18 0.21
Formate 0.70 0.73 0.83
Acetate and Formate 0.70 0.73 0.83

 

 

 

 

 

 

lSubstrate was sand and solid-liquid separated dairy manure

Table 5.6 Summary of Phase 1 respirometer results, mL CH4/g VSS/hr

 

 

 

 

 

 

 

 

 

 

 

Substrate‘ CMD SMD MED
Acetate 1.53 1.49 1.50
Formate 3.57 3.10 3.70
Acetate + Formate 2.94 2.37 2.80
Propionate 0.25 0.36 0.34
Cycle time, day"! 5 7 7
L CH4/kg VS/dayz 280 264 268

 

 

lSubstrate was sand and solid-liquid separated dairy manure

Actual conditions
The results for the Phase 1 respirometer show nearly equal activity for the biomass from
each of the digester systems and approximate the ﬁndings for the CMD/AnMBR

118

respirometer work with the exception of the activity for each of the digesters on

propionate.

Methane Production, ml

~~<>~°SMD -3. MED +CMD

 

140
120
100

80

60

 

40

20

Hours

Figure 5.7, Phase 1, acetate

30

25

20

15-3

10

.....o SMD + MED --A-- CMD

 

 

 

 

I 1 I ‘ .1, 1! "T l I ...—w . .... .. T" .. -.- . T""'"""" ..

0 5101520253035404550556065707580859095

5.8 Phase 1, propionate

119

 

 

 

Figure

(

 

—0—SMD "-BuMED +CMD
80

 

4O

30

20

 

 

0 5 10 15 20 25 30 35 40 45 50 55 60 65 7O 75 80 85 90 95 100

Figure 5.9 Phase 1, formate

--~0-°SMD +MED —Al'—CMD
100

 

90
80
70
60
50
40

30

 

 

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Figure 5.10 Phase 1, acetate + formate

120

+SMD "'23" MED -)(- CMD

2,000 7

1,800 4

L

1,600
1,400 \-
1,200 «
1,000 J
800 «

6004

 

Glucose conentration, mall.

400 1

 

200

 

 

0 -1..____._-1 -, ,. . --.l--. .....1. 1.----.-.-- --. __..,---._...-. ._-r-.-..__.--_-._---,- ---...

0 2 3 5.8 6.7 8.7 9.7 10.5 11

Hours

Figure 5.11 Phase 1, glucose consumption per mass VSS

The resulting activity on propionate was similar for the SMD and MED; however, the
CMD curve for propionate was very different in appearance and not consistent with
earlier (or later data) (Figure 5.8). GC analysis resulted in a methane concentration of
38% for the SMD and the MED control bottles; however, the methane concentration for
the CMD control bottle was 25% and this likely skewed the propionate curve. If the
methane content for the CMD control was assumed to approximate the other respirometer
bottles, the CMD would have looked similar to the SMD and MED curves (data not
shown). Further, none of the Phase 1 systems performed as well on propionate as did the
CMD in the CMD/AnMBR work. This suggests a disruption of the syntrophic

interaction occurred for all three digesters of Phase 1.

121

 

K

_J

 

The measured activity on fonnate+acetate (Figure 5.10) was less than on formate
alone for all three digesters, indicating that Methanosarcina like organisms made a
signiﬁcant contribution to the methanogenic population (Dolﬁng and Bloemen 1985).
All three digesters were in a startup mode during much of Phase 1 and exhibited
signiﬁcant volatile acid concentrations through day 36 (Figure 4.12). The SMD and the
MED had low but measureable volatile acid until Day 55 (MED) and day 59 (SMD).
Methanosarcina has been shown to exhibit high growth rates at elevated acetate

concentrations, while Methanosaeta, with its higher afﬁnity for acetate, results in a

 

competitive advantage at low acetate concentrations (McMahon, Stroot et al. 2001;
Hoffrnann, Garcia et al. 2008). One explanation for the apparent contribution of
Methanosarcina like organisms to the methanogenic structure could be driven by the
initial elevated concentrations of volatile acid of which acetate was likely a signiﬁcant
contributor.

The rate of glucose consumption (Figure 5.11) was very similar for all three
digesters suggesting there was not a discemable difference with respect to the acidogenic

activity of the systems.

122

 

\ xi

 

5.4.3 Phase 2 Respirometer

The biomass from the SM AnMBR, the MM AnMBR and CMD were compared using
the substrate feed ratios shown in Table 5.7. CO2 scrubbing was used for the Phase 2
respirometer work. The SM AnMBR was operated at a cycle time of 27, the MM
AnMBR was operated at a cycle time 6 and the CMD was operated at. a low level of
mixing equivalent to a cycle time of 1/day (the mixing pump was operated for 1 minute x
4 times per day).

Table 5.8 provides a summary of the methanogenic activity for each system. Figures
5.12 - 5.15 present the respirometer methanogenic results in graphical formate for the SM
AnMBR, the MM Membrane AnMBR and the control CMD and Figure 5.16 presents the
glucose consumption data for this period. The methane production for the pilot systems
was nearly equal over the course of Phase 2. Despite this, there were signiﬁcant

differences in metabolic activity for these systems.

Table 5.7 Phase 2 respirometer feed ratios, g substrate/gVSS

 

 

 

 

 

 

Substrate] CMD SM AnMBR MM AnMBR
Acetate 0.72 0.93 0.76
Propionate 0.21 0.23 0.19
Formate - 0.71 1.0 0.80
Acetate and Formate 0.80 0.95 0.68

 

 

 

 

 

 

Substrate was sand and solid-liquid separated dairy manure

123

‘3 , \\.'\

 

Table 5.8 Summary of Phase 2 respirometer results, mL CH4/LVSS/hr

 

 

 

 

 

 

 

Substrate‘ CMD SM AnMBR MM AnMBR
Acetate 1.54 2.18 2.98
Formate 4.95 3.58 1.81
Acetate + Formate 3.61 2.63 1.74
Propionate 0.91 0.24 0.0
Cycle time, day"1 " 27 6

L CH,/kg VS/day2 247 251 246

 

 

 

 

 

 

‘Substrate was sand-separated solid-liquid separated dairy manure
2Actual conditions

The CMD results were similar with those of previous experiments and the CMD
activity on propionate returned to a level similar to that of the CMD in the CMD/AnMBR
respirometer experiment. In Phase 1, the CMD contents were circulated through a
mixing pump at a rate consistent with the two AnMBR systems. For Phase 2, the CMD
pumping rate was reduced to a low rate of mixing. The gentle mixing of the CMD in
Phase 2 may explain the increase in the propionate activity compared to Phase 1.

Both the SM AnMBR and the MM AnMBR exhibited increased acetate activity
compared to the Phase 1 results and the CMD (Figure 5.12 and Tables 5.4 and 5.7). One
possible explanation may be that the SM AnMBR was operated at a higher cycle time
than in Phase 1. In a similar fashion, the MM AnMBR was operated at a cycle time of 6;
however, due to the design of the system, the biomass was pumped through a series of 7
elements connected by common manifolds (Figure 4.7) causing the biomass to travel a
much more torturous path than in Phase 1. As previously discussed, Methanosarcina,
due to its morphology, has an advantage in higher shear environments and, as also

previously discussed, Methanosarcina is particularly effective at higher acetate

124

i‘ 1 s1

 

concentrations, such as in the respirometer bottles. This may explain the increase in
acetate activity for the SM AnMBR and the MM AnMBR compared to the CMD and the
Phase 1 results.

As described above, the MM AnMBR design caused the biomass to travel a torturous

path and may have negated the beneﬁt of operating at a lower cycle time.

"08-- SM AnMBR +MM AnMBR +CMD
250

 

200
150 ED

1004

Methane Production, mL
El

50 - ..

 

 

 

" r r r r 1 I r: I r 1 f

0 51015202530354045505560657075808590

Days

Figure 5.12 Phase 2, acetate

125

+SMAnMBR -‘Ei--MMAnMBR —$—CMD
80

 

70 r i.
60 .. ‘

50 -- ‘

 

40 -‘l A A

30 % -

Methane Production, ml.

204 .1
10 ._4‘

0 '.‘.'“.':‘:':‘:::::::uv.:':f::::::z.Ei£hB-&BBW=::
0 5 101520253035404550556065707580859095

 

 

Days

Figure 5.13 Phase 2, propionate

+SMAnMBR *MMAnMBR '-°)<°°°CMD

 

 

 

 

l
l
!
t
100 *3
t
I
l
E ’
|
l
c‘ 80 ’- ,
l ......................
t .0 p ' ’00 ........ 3
02 ' ...... t O“:':' ............ O
u . ..... "c‘O ' t .........
Q o~ .....'. ................ r'.~. O
a 00000000000
‘3 60 ..
E
5 40 ~
0
20 e
0 . . .. ...._,_. : : : :4: : : : : : : :_-_:;
’ ' ' . . . A A A n A‘a’a'n' 'A'A'A'l' 'A'I'A‘A' 'n“- T I j 1 I I I - I .. I _~

 

0 51015202530354045505560657075808590

Hours

Figure 5.14 Phase 2, formate

126

+SMAnMBR +MMAnMBR "-X-"CMD
120

 

100
80 -'

60 -~

Methane Production, ml.

20-

 

 

 

 

Hours

Figure 5.15 Phase 2, acetate + formate

127

mEi-oSMAnMBR +MMAnMBR --)<--CMD

2,500 4

2,000 4

 

1,500 -~

1,000

Glucose concentration, mg/l.

500

 

 

...L.___--___.__ _- .

 

0 2.25 4.5 5.5 6.5 6.8 9.25 10 10.8

Hours

Figure 5.16 Phase 2, glucose consumption per mass VSS

This is likely reﬂected in the fact that the MM AnMBR exhibited no activity on
propionate (Figure 5.13) and little activity on formate (Figure 5.14). The SM AnMBR
activity on propionate was less than Phase 1, indicating the increase in cycle time from 6
to 27 had a slight impact on syntrophic activity.

All of three of the digester systems exhibited lower activity for the combination of
acetate+forrnate compared to formate alone (Figure 5.15), suggesting that
Methanosarcina like organisms comprised an important component of the methanogenic
community (Dolﬁng and Bloemen 1984).

The glucose consumption test used to measure the acidogenic conversion rate

(conducted at approximately the same VSS concentration for each digester system)

128

produced similar results for all three digesters (Figure 5.16). In practice, the pilot
AnMBRs were operated at a higher VSS concentration than the CMD. Considering this,
if provided with a greater concentration of readily fermentable substrate, they should
produce more methane than the CMD. Since this did not occur, it appears that hydrolysis

could be limiting the rate of substrate conversion.

5.4.4 Microscopy
Biomass from each of the three digesters at day 92 of Phase 2 (see Chapter 4, Section
3.2 for speciﬁc operating detail) was viewed using a scanning electron microscope. A

representative image from each of the digester systems is presented in Figure 5.17.

 

(A) (C)

 

(B)

Figure 5.17 SEM Images, (A) CMD, (B) SM AnMBR, (C) M AnMBR

129

Signiﬁcant groupings of organisms were observed for all three digester systems,
suggesting the potential for spatial proximity of hydrogen consuming methanogens and

fatty acid oxidizing bacteria remained intact.

5.4.5 Most Probable Number

Samples were collected from the Phase 2 digester experiment at day 92 (see Chapter
4, Section 3.2 speciﬁc operating detail) for most probable number evaluation of cell
viability. Because hydraulic retention time is decoupled from solids retention time,
theoretically, the AnMBR systems are expected to have higher concentrations of viable
cells than the CMD. The results are presented in Table 6.8. Surprisingly, there was very
little difference between the three systems with respect to the predicted number of viable
cells.

Table 5.9 MPN results after 144 hours of incubation, estimated using 5-tube serial
dilution

 

 

 

 

 

 

 

 

 

 

CMD SM AnMBR MM AnMBR

MPN x 108/mL 2.4 2.4 5.0
Lower 95% Conﬁdence limits 1.0 1.0 2.0
(x108/mL)

Upper 95% Conﬁdence limits 9.4 9.4 20.0
(x108/mL)

vss, mg/L 15,267 16,250 29,833
vs, mg/L 19,700 31,600 39,600

 

 

5.5 Summary

A theory in previous anaerobic membrane bioreactor research suggests that the
shearing impact of the pump/membrane system acts to degrade the juxtaposition of the
syntrophic relationship (Brockmann 1995; Brockmann and Seyfried 1996; Brockmann
and Seyfried'1997; Ghyoot and Verstraete 1997; Stroot, McMahon et al. 2001). The
present research does indicate a reduction in the syntrophic interaction. However, with
the exception of the result for the MM AnMBR operating at a cycle time of 6, the
respirometer experiments suggest that the syntrophic relationship is still present and
functioning, although at a reduced efﬁciency as compared to the CMD. Kasper and
Wuhrmann (1978) reported that propionate-degrading systems were saturated to only 10
to 15% and hydrogen removal was less than 1% of the maximum possible rate. This
indicates that, provided the juxtaposition of acetogenic bacteria and hydrogentrophic
methanogens remains somewhat intact, there is signiﬁcant excess capacity available to
process hydrogen in anaerobic systems. The SEM photos for the SM AnMBR, the MM
AnMBR and the CMD provide evidence that the membrane systems maintain a degree of
biomass agglomeration (Figure 5.17) suggesting that hydrogen producers and hydrogen
consumers are able to maintain spatial proximity to each other.

Padmasiri et al. (2007) reported a reduction in microbial activity for an anaerobic
membrane bioreactor treating swine waste and proposed that the high velocity
environment promotes (due to the pumping action) an increase in the rate of hydrolysis
leading to a buildup of fermentation intermediates and ultimately a depression in
methanogenic activity. Padmasiri’s research was conducted at organic loading rates

between 1 and 3 g VS/L/day with unstable operation at loading rates greater than 3 g

131

VS/L/day characterized by VFA excursions greater than 4,000 mg/L. For each of the
operating conditions evaluated in the present research, the AnMBRs did not experience a
build-up of VFA. The AnMBR from the CMD/AnMBR research was operated at a cycle
time of 56 with an average volatile acid concentration of 232 mg/L as HAC (Chapter 3,
Figure 3.13, titration method outlined in Appendix B). The SM AnMBR and the MM
AnMBR operated at cycle times of 27 and 6 respectively and, at steady-state, there was
no volatile acid recognized by the titration technique of Appendix B. However, acetic,
propionic, butyric and Isobutyric acids were measured for the SM AnMBR and MM
AnMBR using a GCMS procedure (Appendix E) and resulted in a total VFA
concentration of 299, 271 mg/L respectively (Chapter 4, Table 4.10). Provided hydrogen
consuming methanogens maintain a sufﬁciently low H2 concentration, the fermentation
pathway results in the production of acetate, formate and H2. In the absence of these
scavengers, fermentation will proceed independent of methanogenic activity in the
direction of high molecular weight VFA (Hungate 1975; Bryant 1979). Neither the SM
AnMBR or the MM AnMBR (or the CMD) exhibited high concentrations of propionate,
butyrate or isobutyrate for Phase 2 of the research (Chapter 4, Table 4.10) suggesting
hydrogen was maintained at a sufﬁciently low concentration to avoid a build-up of higher
molecular weight VFA. This is consistent with the ﬁndings of Beaubien et al. (1996)
who found that the high shear stress generated by the operating condition of a membrane
bioreactor did not induce a signiﬁcant reduction on methanogenic speciﬁc activity at
organic loading rates (0.8-0.9 kg COD/kg VSS/day), similar to loading rates in the

present research.

132

 

Despite variations in metabolic activity for the three digesters in Phase 2, the methane
production was nearly the same for each system. This suggests that, though respirometer
testing indicated an apparent degradation of syntrophic activity, there remained an
adequate hydrogen consuming population capable of metabolizing H2 such that the fatty
acid oxidizing bacteria (acetogens) were not inhibited.

One possible theory for why the AnMBRs did not exhibit greater gas production
compared to the CMD is thought to be related to the rate of hydrolysis. Theoretically, the
longer SRT of the AnMBR provides a higher concentration of viable cells and a
corresponding increase in the rate of hydrolysis. However, if the rate of hydrolysis is not
inﬂuenced by the function of the AnMBR, it stands to reason that the number of viable
cells in the AnMBR will not differ signiﬁcantly from the CMD, as was observed for the
MPN testing (Table 5.8). Considering that all three systems of Phase 2 produced
methane equally (Chapter 4, Figure 4.22) without VFA build-up (Chapter 4, Table 4.10),
provides support for the theory that hydrolysis is controlling the production of
fermentation pre-cursors and thus limiting the growth potential for the downstream
anaerobic consortium. Furthermore, the glucose consumption test illustrated that all three
systems (SM AnMBR, MM AnMBR and CMD) operated at approximately the same
acidogenic rate per mass VSS (Figure 5.16), indicating that fermentation (of hydrolysis
breakdown products) is not the “bottleneck” in the process.

Apparently the AnMBR, despite the longer SRT, is not able to affect a higher rate of
hydrolysis than the CMD system. Munch et al. (1999) proposed a kinetic expression to
describe the hydrolysis rate as the ratio of:

Particulate concentration x Hydrolytic enzyme concentration
Acidogenic bacteria concentration

133

 

This indicates that the rate of hydrolysis is reduced at high biomass concentrations.
Myint et al. (2007) theorized that that the reduction in the rate of hydrolysis at high
biomass concentrations is likely due to limited surface area causing mass transfer
limitations of the hydrolytic enzyme. The theory suggests that hydrolysis is limited by
the particle surface area occupied by organisms secreting hydrolytic enzymes and, once
the surfaces are completely occupied, the maximum hydrolysis rate is deﬁned and
additional organisms cannot inﬂuence this rate. Methane production and most probable
number of viable cells for the Phase 2 SM AnMBR, MM AnMBR and the CMD suggest
that the available sites for particulate occupation are exhausted at the biomass
concentration found in the CMD. Therefore, the increased VSS concentration of the
AnMBR does not affect a higher rate of hydrolysis and, as a result, the viable cell

population reﬂects the substrate concentration available for metabolism.

134

 

Chapter 6

AnMBR DESIGN CONSIDERATIONS

6.1 Introduction

The purposed of this chapter is to bring together all of the AnMBR operational data, as
well as a qualitative understanding of the associated microbiology, so that design
considerations can be formulated for the treatment of liquid dairy manure. General
guidelines associated with the impact of cycle time, cross ﬂow velocity and membrane
conﬁguration, operating pressure, TS concentration and ﬂux rate, HRT and SRT, pump
selection, membrane pore size, membrane pore size, and membrane cleaning are
presented. The recommendations are a combination of speciﬁc ﬁndings of this research

with consideration for the previous work conducted by others in related areas.

6.2 Cycle Time

There is much research in support of the AnMBR as an effective, high rate process
capable of producing excellent efﬂuent quality while providing a very high level of
organic conversion. However, other research suggests consideration for microbial
inhibition due to the shearing impacts associated with turbulent transport of biomass
through the membrane system or other high shear applications (discussed in detail in

Chapter 2, Section 7).

Based on the ﬁndings presented in Chapters 5 and 6, recommended cycle times are

between 6/day and 27/day. Cycle time is the starting point for AnMBR design, providing

135

an engineering benchmark against which the other design parameters must fit. Therefore,

knowledge of acceptable cycle time limits is critical.

6.3 Cross Flow Velocity and Membrane Configuration

Cross-ﬂow velocity (CFV) and total solid content were identiﬁed in this research as the
most important factors in optimizing permeate ﬂux for liquid dairy manure. Baker et al.
(1985) observed higher permeate ﬂux rates at higher cross-ﬂow velocities for a mineral
slurry. They reported that permeate ﬂux was proportional to cross-ﬂow velocity raised to
the power of 0.6. Fane and Dell (1987) found that initial ﬂux declines were proportional
to the cross-ﬂow velocity raised to the power of 1.0; however, long-term steady-state
ﬂuxes were proportional to cross-ﬂow velocity to the power of 2.4 for bacterial
suspensions and, when fouled, the membrane exhibited negligible flux increases with
increasing cross-ﬂow velocity. According to F ane and Dell (1987), increasing the cross-
ﬂow velocity has the effect of decreasing the degree of polarization by increasing mass
transfer and other back-transport mechanisms.

Though it is clear that higher ﬂux rates occur with higher CVFS, there is also a
corresponding increase in pressure drop resulting in higher energy costs. Alternatively,
operation at a lower CFV requires less energy but results in a larger membrane surface
area to obtain the same permeate generation rate. The cost of a membrane system is
linear based on the membrane surface area requirement (determined according to the
design ﬂux rate). Consequently, a life cycle analysis is needed to ﬁnd the optimum CFV

to ﬂux to energy relationship. Because all wastes are unique, prior to selection of a

136

 

design CF V, ﬂux testing at various CFVs is suggested. Table 6.1 presents ﬂux and

related operating conditions for the systems evaluated in the present research.

Table 6.1 Flux summary

 

Digester Description Flux, L/mzlhr TS, % CFV, m/s TMP, kPa

 

 

 

 

 

AnMBR comparison 43 10.3 3.4 100
CMD/AnMBR 34 5.7 3.6 100
Phase 1 Single

118 2.6 4.5 86
Membrane
Phase 2, Single

53 4.9 4.5 86
Membrane AnMBR
Phase 2, 7-Element

24 5.0 2.9 240
AnMBR

 

 

 

 

 

 

 

Selection of the membrane conﬁguration is closely aligned with CFV and is based on
balancing the desired cycle time with system energy and capital cost constraints. Phase 2
(Chapter 4, Section 3.2) compared the MM AnMBR operated at a CFV of 2.9 m/s with
the SM AnMBR operated at a CFV of 4.5 m/s and it was shown that that higher CFV
generated a ﬂux rate that was approximately twice that of the lower CFV (Chapter 4,
Section 3.2 and table 4.6).

Equation 6.1, the Darcy Equation, states that head loss (or AP) is proportional to the

velocity squared.

HL = f *——*——— (6.1)

The following discussion is provided to illustrate that the Darcy relationship is valid

for this manure pumping application. During Phase 3, the SM AnMBR CFV was

137

reduced from 4.5 m/s to 3.5 m/s for a period of approximately 4 days and a summary is

 

 

 

presented in Table 6.2.
Table 6.2, Phase 3 CFV comparison
CFV, m/s Flux, L/mthr AP (:sz
4.5 51 76 20.25
3.5 31 49 12.25

 

 

 

 

 

 

Ratio of Change 1.6 1.7

 

According to the Darcy relationship, the ratio of AP should equate to the ratio of the
CFV2 and, in fact, the actual conditions are consistent with predicted expectation, as

illustrated in Table 6.2. Table 6.3, data from Phase 2, compares the ﬂux rate and cross-
ﬂow velocity of the SM AnMBR and the MM AnMBR. Due to the manifold
conﬁguration of the MM AnMBR (discussed in Chapter 4, Section 2.3), an accurate
measurement of the AP per element was not possible. However, considering the Darcy

relationship, the calculated AP is shown in Table 6.3.

Table 6.3, Phase 2 comparison of SM AnMBR and MM AnMBR

 

 

 

 

 

 

 

 

CFV, m/s Flux, L/mzlhr AP (2sz
4.5 53 76 20.25
2.9 24 32" 8.41

 

1
Calculated AP per element

138

The power required for pumping relationship is shown in Equation 6.2.
P4, = q-p-g-h+3.6x106 (6.2)
Where,

P4, = power (kW).

q = ﬂow capacity (m3/hr).

p = density of ﬂuid (kg/m3).

g = gravity (9.8 m/sz).

h = differential pressure head, (m).

Considering this relationship, the energy input difference between a CFV of 4.5 m/s and a
CFV of 2.9 m/s can be directly compared based on the fact that power required for
pumping is proportional to ﬂow rate x AP. Operating at a CFV of 2.9 m/s provides a
tremendous energy advantage compared to 4.5 m/s.

The ﬁndings of Chapter 4 suggest that the operating conditions of the MM AnMBR
did not negatively inﬂuence gas production and also indicated ﬂux was stable at 2.9 m/s.
Research presented in Chapter 4 also suggests there is negligible difference between gas
production at a cycle times between 6 and 27 (Table 4.6) and that operation at CFVS as
high as 4.5 m/s does not negatively impact methane production (Table 4.6). Therefore, it
is recommended that membranes be conﬁgured based on case of design and operation
within the general framework of a maximum pump discharge pressure of 480 kPa (Table

4.6) and a maximum cycle time of 27/day with the CFV selected to balance the capital

cost versus operating cost objectives of the project.

139

6.4 Operating Pressure

At operating pressures between 180 and 200 kPa and greater, Ghyoot and Verstraete
(1997) found ﬂux to be independent of pressure for sludge concentrations between 6.0
and 25.0 g TS/l. Strohwald and Ross (1992) found that membrane ﬂux was independent
of operating pressure above 260 kPa and a cross-ﬂow velocity of 1.9 m/s. Beaubien et al.

(1996) referenced Equation 6.3 from (Cheryan 1986):

J = , AP‘ (6.3)
“(R m "' BAPt)

 

Where,

J = permeate ﬂux (um/s)

AP, = Applied transmembrane pressure
p = Permeate viscosity

R’m = Resistance comprised of membrane-solute interactions presumed unaffected by
operating parameters

BAP, = Resistance related to operating conditions

Based on Equation 6.3, two regions of interest can be identiﬁed, a low pressure region
where the hydraulic resistance of the membrane dictates the ﬂux rate (R’ >>BAP,) and a
high pressure region where ﬂux is controlled by the operating conditions of the system
(BAPt >>R’m). The experimental work of Beaubien et al. (1996) showed two distinct
zones that depended on operating pressure. At operating pressures less than 80 kPa,
permeate rate was largely dependent on applied pressure and suspended solids
concentration. At operating pressures above 100 kPa, ﬂux is largely pressure
independent and permeate ﬂux was directly proportional to cross-ﬂow velocity. Fugere
et al. (2005) reported similar results indicating that ﬂux was relatively pressure

independent above 100 kPa; however, they noted that at pressures between 150 and 300

140

kPa, the rate of ﬂux change with increasing cross-ﬂow velocity was less than at pressures
between 50 and 100 kPa.

Identiﬁcation of optimum operating pressure was not a goal of the present research;
however, based on the results of previous research, and considering the necessary
transmembrane pressure (TMP) for a dairy manure AnMBR, the ﬂux will most likely be
pressure independent. As such, it is recommended that design be based on the minimum
pressure drop (based on membrane conﬁguration) to achieve the design cross ﬂow

velocity.

6.5 Total Solids Concentration and Flux Rate

Digester TS concentration was used in this research as a benchmark to compare
digester ﬂux conditions; however, it is common in the research to also reference total
suspended solids (TSS), mixed liquor total suspended solids (MLTSS),VSS or mixed
liquor volatile suspended solids (MLVSS). TS and VS were commonly used throughout
this research because they are accurate and easy to determine and were analyzed every
day. TSSN SS analysis is considerably more time consuming to analyze. Figure 6.1
illustrates that TS tracked closely with (TSS) for this research. A similar relationship
held for the comparison of VS to VSS (data not shown).

Ross et a1. (1992) found a constant ﬂux up to a suspended solids concentration of 40
g/L, after which ﬂuxed decreased rapidly for a maize-processing efﬂuent. Berube et al.
(2006) indicated that Saw et al. (1985) reported a log-linear decrease in the steady-state
permeate ﬂux with an increase in the concentration of suspended solids for digested

sludge. Kitamura et al. (1996) theorized that a decrease in membrane performance was

141

due to fouling caused by an increase in viscosity of the sludge at higher suspended solid

concentrations.

—o—Tss ”.53.. rs
60,000

 

50,000 4‘

 

40,000 ~

mg/I.

30,000 ~

20,000 1

 

 

10,000 -4'

 

Observations

Figure 6.1 TSS versus digester TS concentration

In general, based on the ﬁndings of this and previous research, there is a direct
relationship between digester TS concentration and membrane ﬂux rate between 2-10%
(Figure 3.6, Figures 4.20 and 4.21). This is consistent with the ﬁndings of (Li 1985;
Beaubien, Baty et al. 1996; Madaeni 1997). The initial research that compared the

AnMBR to the CMD resulted in relatively high ﬂux rates despite a very high digester TS

142

concentration (Table 3.3 and Figure 3.6). This is likely the result of the type of manure
used for this experiment. The manure was sand-separated and still contained large pieces
if undigested ﬁber. The AnMBR of the CMD/AnMBR system was fed the digested
CMD efﬂuent (feedstock to the CMD was sand-separated manure), resulting in a
homogeneous feedstock. Sand and solid-liquid separated manure was used as the
feedstock for Phases 1, 2 and 3 and, due to the solid-liquid separation process, provided a
consistent feedstock devoid of large particles. Madaeini (1997) indicated that smaller
particles sizes lead to lower ﬂux rates and this appears to be consistent with the outcome
of these experiments.

Flux rates trend down as TS concentration increases. However, from a design
perspective, it is the SRT, combined with the starting TS concentration of the wastewater,
that will deﬁne the digester TS operating concentration. Section 6.6 provides

recommendations relative to the design condition based on the ﬁndings of this research.

6.6 HRT and SRT

The experiments of the present research were conducted at HRTs of 12-20 days. Due
to the extended SRT that can be achieved with the AnMBR, HRTs lower than 12 days are
certainly possible. Dugba and Zhang (1999) operated two-stage anaerobic sequencing
batch reactors at 3 and 6 day HRTS with SRTs between 13-18 days and reported VS
destruction of 23-34% for a mesophilic system with organic loading rates of 2-4 g
VS/L/day of screened dairy manure. Zhang et al. (1997) operated an anaerobic
sequencing batch reactor on swine manure with VS reductions of 55-61% at an HRT of 3

days and organic loading rates of 1.0-5.5 g VS/L/day. Padmasiri et al. (2007) operated a

143

mesophilic AnMBR treating swine manure at a HRT of 6 days and organic loading rates
of 1.0-3.0 g VS/L/day (VS destruction not reported).

The selected OLR, in conjunction with the characteristics of the wastewater, deﬁne
the HRT (Equation 6.4). Di gester volume equals the required treatment volume per day

multiplied by the HRT.

HRT 1. VS concentration

6.4
OLRVS ( )

Based on the combination of digester volume, design SRT and design digester TS (or
VS, TSS, VSS) operating concentration, the digester wasting rate can be calculated. A
typical goal in the operation of an AnMBR is to maximize permeate production (and
minimize the total volume that is wasted from the digester) and this is accomplished by
operating the digester at the highest possible TS (or VS, TSS, VSS) concentration.

Equation 6.5 is used to calculate the required digester wasting rate.

Total mass dry solids in digester
SRT

 

Wasting rate, mass dry solids/day = (6.5)

The present research was conducted at OLRS of 1.0-4.0 g VS/L/day (with an
average of 2.2 g VS/L/day). Based on the present and related research, OLR rates of 2.0
— 4.0 g VS/L/day are suggested for sand and solid-separated dairy manure with digester
TS operating concentrations of approximately 1.2 to 2.5 times the feed TS concentration

and a design SRT in the range of 20-30 days. Di gester TS concentrations of 10-1 1% are

144

attainable; however, Speciﬁc work related to determining the maximum TS concentration

in relationship to ﬂux rate was not evaluated in this research.

6.7 Pump Selection

Kim et al. (2001) concluded that the activity of microorganisms was damaged more
severely and the microbial ﬂocs more easily destroyed with a positive displacement pump
compared to a centrifugal pump. He et al. (2005) reported that the mechanical shearing
impact of the pump used in an AnMBR for the treatment of food waste negatively
impacted the microbial activity of the system, particularly the methanogens and further
suggested the selection of a low shearing pump. Choo and Lee (1996) theorized that the
low viable suspended biomass concentration in the bioreactor (of an AnMBR) treating
alcohol-distillery wastewater was due to cell lysis caused by mechanical sheer stress from
the positive displacement recirculation pump, noting that the cells moved from the
bioreactor to the surface of the membrane.

A centrifugal pump was used in the present research. SEM images from the three
digesters used in this research (Figure 5.17) indicate that microbial ﬂocs are intact.
Metabolic testing suggests that the syntrophic interaction, though reduced compared to a
minimally mixed CMD, was intact (Chapter 5, Section 5). Positive displacement pumps
operate with close tolerances creating a greater potential for ﬂoc disintegration Madaeni
(1997) reported ﬂux decline with smaller particles. Considering these outcomes, it stands

to reason that a centrifugal pump is the most appropriate selection for system design.

145

6.8 Membrane pore size

Choo and Lee (1996) identiﬁed that the fouling tendency was at a minimum for a
membrane pore size of 0.1 pm for bacterial cells isolated from an anaerobic digester
system but also suggested that the size of the inﬂuent solid content was important when
selecting membrane pore size. Their line of reasoning followed that macrosolutes
smaller than the pore size of the membrane will easily pass through the membrane while
colloids that are considerably larger than the more size, will tend to remain on the
membrane surface but not penetrate the pores deeply and are thus easily swept away due
to the affect of CFV. Madaeni (1997) reported ﬂux decline with smaller particles.
Chang et al. (2002) reported that that Shimizu et al. (1990) correlated ﬂux with the pore
size for methanogenic wastes and illustrated that membranes with pore sizes in the range
of 005-02 um produced the maximum ﬂux among membranes ranging from 0.01-1.6
um.

Pore size may be an important factor with regard to nutrient and pathogen/virus
retention. For, example, the majority of the phosphorus content in raw and anaerobically
digested swine manure is linked to particles larger than 0.45 microns (Masse, Masse et al.
2005). The present research was conducted with a 0.03 pm membrane. Total
phosphorus reduction was 96%. Vogel et al. (Vogel 2003), operating a thermophilic
AnMBR on dairy manure, also reported a 96% reduction of total phosphorus using a
membrane with pore openings from 0.005 to 0.1 pm.

Wong et al. (Wong, Xagoraraki et al. 2009) evaluated the removal efﬁciency of the
CMD/AnMBR system from Chapter 3, Section 3.2 for E. coli, enterococci, C.

perfringens and coliphage with total loglo removals of 1.5, 1.2, 0.1 and 0.5 respectively

146

 

for the CMD and the 5.2, 6.1, 6.4 and 3.7 respectively for the AnMBR. The vast
majority of the removal was attributed to the AnMBR. The lowest removal efﬁciency
was for coliphage. This ﬁnding was not surprising considering viruses are typically
smaller than bacteria and can be as small as 0.01 pm. Nevertheless, the removal
efﬁciency for the coliphage was still 99.96% (3.7 loglo removal). Wong et al. (Wong,
Xagoraraki et al. 2009) also evaluated the AnMBR independent of the CMD. This
analysis illustrated that the AnMBR, in a stand-alone capacity, was capable of achieving
the same total pathogen and virus removal rates attributed to the combined
CMD/AnMBR system.

Considering the ﬁndings of this and other research, the optimum membrane pore size
appears to be in the range of 0.03 and 0.1 um. For future design, in the absence of new

information, a membrane pore size of 0.03 pm is recommended.

6.9 Cleaning Protocol

Vogel et al. (2003) used a caustic cleaning solution (3.5% NaOH) followed by a water
rinse and subsequent treatment with 3% phosphoric acid for a ceramic membrane used in
a dairy manure AnMBR. Zhang et a1. (2007) was able to recover 44% of the original
clean water ﬂux through a membrane used in a swine manure AnMBR cleaning with
EDTA at pH 2 and NaOH at pH 10. In addition, slightly better cleaning efﬁciency was
reported using HNO3 with the best results for both the EDTA and HNO3 occurring at
50°C. Zhang (2007) also concluded that the irreversible portion of the fouling (that
which could not be recovered with chemical cleaning) was most likely due to a rapid

process that cannot be avoided by weekly chemical cleaning using HNO3.

147

Several cleaning procedures were tested in present research. All included NaOH and
a citric acid cleaner at various pH levels and soaking times and some also included
soaking the membrane overnight in a 500 PPM bleach solution. Ultimately a consistent
approach that incorporated only NaOH and citric acid cleaner were used for membrane
cleaning. The cleaning was accomplished at temperatures of approximately 10°C to

25°C. The cleaning procedure included the following.

1. Isolate membrane from the digester.

2. Pump clean water from the CIP tank through the membrane to ﬂush membrane
and dispose of this material.

3. Re-ﬁll CIP tank with clean water, add NaOH to pH of 11.0 and then circulate
through the membrane returning to the CIP tank for approximately 30-45 minutes.

4. Add citric acid cleaner (used Citrajet® low foaming cleaner) to pH of 4.0 and
circulate for approximately 30-45 minutes.

5. Increase pH to 7.0 with NaOH.

6. Begin steady addition of clean water to CIP tank and direct efﬂuent from

membrane to disposal for approximately 15-30 minutes.

148

6.10 Summary

Considering the ﬁndings of this research and incorporating the findings of previous
researchers, Table 6.4 provides recommended design condition for AnMBR for the

treatment of liquid dairy manure.

Table 6.4 Design Consideration for AnMBR System

 

 

 

 

Design Parameter Recommended Value
Cycle Time <27
Cross Flow Velocity Up to 4.5 m/s
Operating Pressure As dictated by CF V membrane

geometry but less than 480 kPa

 

 

 

 

 

 

OLR 2.0 —- 4.0 g VS/L/day
SRT 20-30 days
Digester TS, % <10%
Membrane Pore Size 0.03 pm
Pump Selection Centrifugal
Membrane cleaning See Section 6.9

 

 

 

 

149

Chapter 7

ENGINEERING SIGNIFICANCE AND FUTURE WORK

The objective of this research was the development of a design approach for an anaerobic

membrane bioreactor for the treatment of liquid dairy manure. Evaluation of cycle time

at the pilot-scale and the metabolic level were central to this effort. A summary of the

basic ﬁndings of this research and recommended future work follows.

7.1 Summary of Research Findings

A summary of the ﬁndings of this research include the following.

1.

Cycle times greater than 86/day negatively impacted AnMBR performance as
measured by methane production.

Cycle times less than 27 do not negatively impact AnMBR performance as
measured by methane production.

The pump/membrane system of the AnMBR impacted syntrophic activity;
however, the biomass still exhibited approximately 35% of the activity on high
concentrations of propionate compared to a control CMD.

SEM imaging of the biomass indicated large groupings of microorganisms and
this supports the notion that the juxtaposition between acetogens and

hydro gentrophic methano gens stayed intact.

For the tested conditions, biomass concentration (or SRT) did not affect AnMBR
performance. The pilot-scale gas production was equal when the AnMBR was

operated at a VS concentration of 2.6% compared to less than 1.9% (when the all

150

permeate was returned to the digester tank such that the AnMBR was operated in
a complete mix conﬁguration).

6. A high CFV up to 4.5 m/S did not affect AnMBR performance as measured by
methane production.

7. The Increased biomass concentration of the AnMBR apparently did not increase
the rate of hydrolysis; hence, under the conditions tested, AnMBR performance

mirrored that of the control CMD.

Appendix D outlines an example case using typical operating conditions from this
research. The example case illustrates the SRT calculation and relates SRT to HRT and
digester volume and the impact of membrane conﬁguration on cycle time. Finally, the

example outlines the energy implications of the AnMBR system.

7.2 Future Work
Based on the AnMBR research conducted in support of this research, there are numerous

areas that are deserving of future effort.

7.2.1 Increased OLR

A ﬁnding of this research was that the estimated number of viable cells in the AnMBRs
was very similar to the number of viable cells in the CMD. This outcome was observed
despite the difference in biomass concentration between the CMD and the AnMBR.
Based on the pH and VFA data, all of the digesters functioned at low VFA concentration

suggesting there was not excess fermentation intermediates available for consumption. It

151

was theorized that, despite the higher biomass concentration, the AnMBRs did not
promote a higher level of hydrolysis due to mass transfer limitations. As a result, the
concentration of fermentable substrate was likely very similar for the AnMBRs as it was
for the CMD. This theory suggests that without a higher concentration of fermentable
substrate, the advantage of operating at a higher biomass concentration is negated.

One recommended course of action is to achieve steady—state operation for the
AnMBR systems and the CMD and then increase the OLR in a stepwise fashion through
the addition of a readily fermentable substrate (i.e. a substrate that does not require
hydrolysis such as glucose) to evaluate the effect on gas production and digester stability.
This would model the effect of adding an additional substrate such as ethanol plant syrup
to a manure-based digester.

A second recommendation is to increase the OLR by decreasing the HRT. The rate of
gas production of the AnMBR should follow the rate of hydrolysis. If the rate of
hydrolysis is signiﬁcantly impacted by the reduction in HRT, gas production will be
effected. Theoretically, the CMD should become unstable due to washout of acetotrophic
methanogens while the AnMBR should remain stable even at HRTS (potentially in the

range of 3 to 6 days).

152

7.2.2 Temperature Impact on Flux Rate

Ross et al. (1990) showed a ﬂux rate increase of 2% for each 1°C increase in operating
temperature. (Ross, Barnard et a1. 1990). Preez et al. (2005) reported that thermophilic
ﬂuxes, on average, were 29% higher than the Mesophilic ﬂuxes that were measured in
comparison research.

Under thermophilic conditions, microbes exhibit 2-3 times higher maximum Speciﬁc
growth rates compared to mesophilic microbes (Mladenovska and Ahring 2000). As a
result, the organic loading potentials of thermophilic anaerobic reactors are substantially
higher with improved process economy (Ahn and Forster 2002; Chackhiani, Dabert et al.
2004)

The thermophilic process is reported to be less stable to environmental changes than
the mesophilic process (Yu and Fang 2001). In general, methanogenic diversity (for 15
full-scale biogas plants operating under either mesophilic or thermophilic with either
manure or sludge as feedstock) was broader in plants operating at mesophilic
temperatures (Karakashev, Batstone et al. 2005).

Though thermophilic operation is believed to be less stable, high rate processes, such
as the AnMBR, maintain the advantage of long SRT (higher biomass concentrations) and
this may improve the stability of the operation. Further, because the capital costs are
driven by membrane ﬂux rates, the potential to improve these rates with thermophilic

operating conditions deserves evaluation.

153

7.2.3 Flux Recovery with Cleaning

A cleaning protocol was developed in this research that used NaOH circulated through
the membrane at pH of 11.0 for 30-45 minutes followed by the circulation of a citric acid
cleaner at pH 4.5 for 30-45 minutes followed by NaOH neutralization and freshwater
rinsing. Excellent results were obtained with this cleaning protocol and it could be
accomplished very quickly with minimal effort; however, quantiﬁcation of ﬂux recovery
was not conducted. It is recommended that experiments be performed with new
membranes to evaluate the ﬂux recovery that can be obtained. Lastly, based on the work
of Zhang et al. (2007), experimentation with cleaning at much higher temperatures than

used in this research is also suggested.

154

APPENDIX A

Wong, K., Xagoraraki, 1., Wallace, J ., Bickert, W., Srinivasan, S., Rose, J .B. (2009). Removal of Viruses
and Indicators by Anaerobic Membrane Bioreactor Treating Animal Waste. Jourgal of Environmental

Qualig, 38:In Press.

155

Removal of Viruses and Indicators by Anaerobic Membrane Bioreactor Treating Animal Waste

Kelvin Wong 1, Irene Xagoraraki 1., James Wallace 2, William Bickert 2, Sangeetha Srinivasan3, Joan B.

Rose 3

Department of C1sz and Envzronmental Engineering, Department of Agriculture and Biosystems

Engineering ,2 Department of Fisheries and Wildlife',3 Michigan State University

*Corresponding Author: Mailing address: Civil and Environmental Engineering, A124
Engineering Research Complex, East Lansing, MI 48824. Phone: (517)

353-8539. Fax: (517) 355-0250. E-mail: _rggorara@msu.cdu

156

Abbreviations: AnMBR, anaerobic membrane bioreactor; BAdV, bovine adenoviruses; BPyV, bovine
polymaviruses; CMAD, complete mix anaerobic digester; COD, chemical oxygen demand; MBR,
membrane bioreactor; TKN, total Kjeldahl nitrogen; TP, total phosphate; TS, total solids; VS, volatile

solids
ABSTRACT

Appropriate treatment of agricultural waste is necessary for the protection of public health in rural
areas since land-applied animal manure may transmit zoonotic disease. In this study, we evaluated the
potential of using a pilot anaerobic membrane bioreactor (AnMBR) to treat agricultural waste. The
AnMBR system, following a conventional complete mix anaerobic digester (CMAD), was able to achieve
high removals of both biological and chemical agents. The mean loglo removals of E. coli, enterococci, C.
perfringens and coliphage by the AnMBR were 5.2, 6.1, 6.4 and 3.7, respectively, and for the CMAD were
1.5, 1.2, 0.1, and 0.5, respectively. Compared to other indicators, coliphage was observed most frequently
and had the highest concentration in efﬂuent samples. Bovine adenoviruses (BAdV) and bovine
polymaviruses (BPyV) were monitored in this study using nested PCR methods. All of the CMAD inﬂuent
and CMAD efﬂuent samples were found positive for both viruses and three AnMBR efﬂuent samples were
found BPyV positive. The mean removals of total Kjeldahl nitrogen (TKN), total phosphate (TP), chemical
oxygen demand (COD), total solid (TS) and volatile solid (VS) by the entire system were 31%, 96%, 92%,
82% and 91%, respectively, but there was no removal of ammonium. When the AnMBR was operated
independent of the CMAD, AnMBR achieved similar E. coli and enterococci removals as the combined
CMAD/AnMBR system. The high quality of efﬂuent produced by the pilot AnMBR system in this study

demonstrated that such systems can be considered as alternatives for managing animal manure.

Keywords: Agricultural waste, manure, anaerobic membrane bioreactor, pathogen removal, indicators,

animal viruses, zoonotic pathogens

157

INTRODUCTION

With the increase of animal agriculture facilities, there is a growing concern regarding
transmission of enteric zoonotic pathogens via food and water. One of the most important sources of
microbiological pollution is fecal contamination from storage and management of manure (EPA 2006).
Campylobacter spp., Salmonella spp. (nontyphoid), Listeria monocytogenes, E. coli OlS7:H7,

Cryptosporidium parvum, and Giardia lamblia were identiﬁed by the Center for Disease Connol and

 

Prevention as causative agents most likely originating from farm sources (Gerba and Smith, 2005). l
Hepatitis E, Rotavirus, and Saprovirus have also been documented as zoonotic viruses (Gerba and Smith, D
i
i i
2005; Costantini et al. 2007). One of the most severe recent waterborne disease outbreaks occurred in L
‘i

Walkerton, Ontario, in May 2000 and was attributed to farm runoff. Seven people died and over 2,000 were

ill as a result of the outbreak (Holme 2003).

Proper treatment of agricultural animal waste and manure should not be overlooked especially in
terms of pathogen removal and inactivation. Animal manure is often land applied without prior treatment.
Although anaerobic digestion, aerobic digestion, and facultative lagoons are manure treatment alternatives

(Johnson et al. 2004), these systems do not necessarily remove zoonotic pathogens.

Membrane bioreactor (MBR) systems have become popular in the last couple of decades even
with the drawback of high capital investment and maintenance cost. Membranes provide a barrier for the
separation of pathogens and contaminants from wastewater and often provide a high quality efﬂuent. To
the best of our knowledge, there are currently no published studies evaluating the removal of pathogens and
pathogen indicators from animal waste using MBR systems. However, Cicek (2003) proposed that MBR

systems have great potential for agricultural waste treatment.

Ottoson et a1. (2006) compared the removal of indicators, Giardia cysts, Cryptospordium oocysts
and enteric viruses in a municipal wastewater by MBR and conventional treatment. Virus genomes were
removed equally by conventional and MBR treatment. However, MBR treatment removed microbial
indicators more efﬁciently than conventional treatment. There are also a number of published studies

focusing on the removal of MS2 coliphage and coliphage T4 by MBR systems and all of these studies

158

demonstrated high removals of coliphage (Shang et al. 2005; Lv et al. 2005; Zheng et al. 2005; Comerton

et al. 2005; Ahn et al. 2001; Ueda and Horan 2000).

Anaerobic membrane bioreactors (AnMBR) have been reported to generate high quality efﬂuent
(Fuchs et al., 2003; Fakhru’l-Razi, 1994). However, no study has been conducted on the removal of
biological agents in animal waste by AnMBR. The main objective of this study was to evaluate the removal
of pathogen indicators and animal pathogen viruses from agriculture waste using a pilot AnMBR following
a conventional anaerobic CMAD digester. The removals attributed by CMAD and AnMBR were
compared. Removals of E. coli and enterococci were evaluated when AnMBR was operated independent of

the CMAD. Finally, the removals of important chemical parameters are also presented in this study.
METHODS
CMAD and A nMBR Pilot Systems

The experiments were conducted in a pilot unit located at Michigan State University (Figure 1).
Sand-separated dairy manure was ﬁrst treated by a 100-L complete mix anaerobic digester (CMAD) and
the efﬂuent from the CMAD digester was further treated by 100-L AnMBR. The AnMBR was operated in
a cross-ﬂow conﬁguration. The system employed a centrifugal pump capable of approximately 35 Umin
at 200 kPa. The membrane was a 0.03 micron, 14.4 mm diameter, 0.126 m2 PVDF tubular product

manufactured by X-Flow, Inc. (Netherlands).

During the operating period, the CMAD was fed sand-separated dairy manure at an average
organic loading rate of 3.3 g (VS)/L-day. The efﬂuent from the CMAD was fed to the AnMBR , which
resulted in an average organic loading rate of 2.4 g (VS)/L-day. The permeate generation rate and pump
circulation rate were 64 ml/min and 35 L/min, respectively. The hydraulic retention time for both CMAD
and AnMBR was 9 days. The combined system hydraulic retention time was 18 days. The AnMBR solids
retention time averaged 28 days during the period of study. The system was operated under mesophilic

conditions.

Sand-separated dairy manure from Green Meadow Farms (Elsie, MI), a commercially operating
dairy farm, was the substrate for the CMAD. The efﬂuent from the CMAD was the substrate for the

159

AnMBR. The manure was pre-treated at the farm via sand separation. This process was essentially a grit

separator, where recycled water was added and the sand was settled from the manure.

Sampling and Sample Preparation

This study was conducted from February to April and June to August 2007. During the ﬁrst period,
the AnMBR was operated independent of the CMAD (AnMBR system alone, ﬂow bypassed CMAD). A
total of seven sampling events were conducted during the period. The sampling took place in
approximately one-week intervals. There were two sampling points: inﬂuent, and AnMBR efﬂuent (points
1 and 3 as shown in Figure 1). From June to August (second sampling period), the pilot unit was operated
as a combined CMAD/AnMBR system. There were three sampling points: CMAD inﬂuent, CMAD
efﬂuent/AnMBR inﬂuent (referenced throughout as CMAD efﬂuent) and AnMBR efﬂuent (points 1, 2, 3
as shown in Figure 1). A total of eight sampling events were conducted during this period and the sampling
took place in approximately 1 week internals. Only E. coli and enterococci were monitored during the ﬁrst
period. Six chemical parameters, four microbial indicators, and two animal enteric viruses were monitored

during the second period.

Both the CMAD inﬂuent and CMAD efﬂuent were grab samples. Due to large volumes needed for
microbiological analysis and low AnMBR permeate generation rates, the AnMBR effluent was collected as
a 24 hr conrposite sample. All of the samples were collected in sterilized disposable containers. Once the
samples were collected, they were placed in an ice-chest and transferred to the Water Quality Laboratory at
Michigan State University within 2 hours. All samples were stored in a 4°C refrigerator upon arrival to the

laboratory and were analyzed the day of collection. Any repeated testing was done the following day.

Due to the low concentration of viruses in the AnMBR efﬂuent, efﬂuent samples were concentrated
in order to achieve a larger equivalent volume during the PCR reaction. The concentration method used in
this study was developed by Haramoto et al (2005) except Arnicon Ultra (Millipore, Billerica MA) was
used to concentrate the NaOH eluent instead of Centriprep YM-50. The ﬁnal volume of concentrated eluent

was around 140 pl and was stored at -80 °C for DNA extraction. The literature reported virus recovery

160

percentage for this method was 56%132%. The equivalent volume of AnMBR efﬂuent for each PCR

reaction was about 20ml.

Indicator Analysis

E. coli, Enterococci, and C. perfringens were the bacterial indicators and somatic coliphage was
the viral indicator monitored in this study. Membrane ﬁltration (MF) technique was used for the detection
of indicator bacteria. E. coli and Enterococci were analyzed by EPA 1603 and 1600, respectively and the
analytical procedure used for C. perfringens was adopted by Bisson and Cabelli (1979). The CMAD
inﬂuent and CMAD efﬂuent samples were ﬁrst diluted 10, 100 and 1000 fold with phosphate buffer water.
Then, 1.0 ml of each dilution was aliquotcd for MP. The reported concentration was calculated from the
dilution that gave the most statistically accurate result (20 to 100 cfu per plate). For the AnMBR efﬂuent,

lL of sample volume was analyzed.

After ﬁltration, the membranes were placed on agar media for growing E. coli, Enterococci, and
C. perfringens, respectively. The incubation temperature for E. coli was at 35°C for 2.0105 hours and
445°C for 22011.0 hours. Enterococci and C. perfringens were incubated for 2412.0 hours at 41 and 45°C,

respectively.

Somatic coliphage was analyzed according to EPA 1602 single agar layer method. The host
culture was E. coli CN 13. Similar to bacterial indicator analysis, 10, 100 and 1000 fold dilutions were
analyzed for the inﬂuent and digester samples. The volume of AnMBR efﬂuent sample analyzed was 10
ml. The incubation procedure for somatic coliphage was 370°C for 24 12.0 hours. The dilution that gave
the most statistical accurate result, 20 to 200 pfu per plate, was used for calculating the reported

concentration.
Molecular Analysis

A stool extraction kit (Qiagen, Valencia, CA) was used for DNA extraction in this study. After
extraction, the DNA samples were stored in a -20 °C freezer before PCR analysis. The PCR reactions were
run in an iCycler thermal cycler (Bio-Rad, Hercules, CA). The two PCR assays for the detection of bovine

adenoviruses and polyomaviruses were selected from the nested PCR method published by Hundesa et al.

161

 

(2006). For adenovirus assay, the ﬁrst round PCR primers were 5’-GRT GGT CIY TRG ATR TRA
TGGA-3’ (forward primer) and 5’-AAG YCT RTC ATC YCC DGG CCA-3’ (reverse primer). The nested
primers were 5’-ATT CAR GTW CCW CAR AAR TIT TITGC-B’ (forward primer) and 5’-CCW GAA
TAH RIA AAR TTK GGA TC-3’ (reverse primer). The PCR cycles were increased to 40 instead of the 30

cycles in the published method for increasing the sensitivity of detection.

For the polyomavirus assay, the ﬁrst round PCR primers were 5’- GGTA TTC GCC CTC TGC
TGG TCA AG-3’(forward primer) and 5’- OCT GGC AAT GGG GTA TGG GT1“ CT-3’ (reverse primer).
The nested primers were 5’- ATP TCA AAG CCC CCT ATC ATC-3’ (forward primer) and 5‘- GCC TAC
GCC ATT CT C ATC AAG-3’(reverse primer). Aﬁer ampliﬁcation, selected positive samples were sent for
nucleotide sequencing to conﬁrm whether the bands were indeed the ampliﬁcation product of BAdV and
BPyV. Due to a strong non-target band in BAdV PCR product, MinElute Gel Extraction Kit (Qiagen,
Valencia, CA) was used to purify the target band for sequencing. No non-target band was observed in
BPyV PCR product; therefore, PCR product was puriﬁed by QIAquick PCR Puriﬁcation Kit (Qiagen,
Valencia, CA) before sequencing. All of the sequencing was performed at the Research Technology
Support Facility, Michigan State University. The sequence results were blasted using

http://www.ncbi.nlm.nih.gov/BLAST/.

Physical and Chemical Analysis

Total solids (TS) and volatile solids (VS) were measured according to AWWA Standard Methods
2540 B and 2540 E, respectively. Chemical oxygen demand (COD) was evaluated using Hach (Loveland,
Colorado) high range COD test kits. Total Kjeldahl nitrogen (TKN), ammonium nitrogen and total
phosphate (TP) were conducted according to “Recommended Methods of Manure Analysis”, Bulletin

A3769, University of Wisconsin Extension (2003).

Data Analysis
The concentrations of microbial indicators were described using loglocfu/ L. To determine

signiﬁcant differences of microbial and chemical concentration between three sampling locations, analysis

162

of variance (ANOVA) single test was performed using Microsoft Excel program. The p-values less than

0.05 indicated signiﬁcant difference.
RESULTS and DISCUSSION
Water Quality in the Combined CMAD/AnMBR System

The samples taken from June to August 2007 were analyzed for TKN, Ammonium, TP, COD, TS,
and VS. Number of samples, average concentrations, standard deviations, and removal percentage are
summarized in Table 1. More than 80 percent removal of TP, COD, TS and VS was achieved. COD
removal in this study is similar to COD removals observed in MBR studies treating other types of waste
(Cicek 2003). No ammonium removal was observed by the AnMBR. This may be due to the fact that
ammonium remained soluble throughout the entire system (typically ammonium concentration increases
during the anaerobic process as protein is degraded). On the other hand, phosphate tends to adhere to the
solid particles, which explains why the removal of TP by AnMBR was much higher than the removal of

TKN and ammonium.

The p—values obtained from ANOVA test showed there were signiﬁcant differences between
CMAD effluent and AnMBR efﬂuent in all chemical parameters except for ammonium (data not shown).
In this case, the signiﬁcant differences demonstrate effective reduction of chemical parameters in AnMBR

efﬂuent. CMAD treatment alone could signiﬁcantly lower the level of COD, TS and VS, but not TKN and

TP.
Microbial Indicator Concentrations and Removals in the Combined C MAD/A nMBR System

The microbiological indicator data collected from June to August 2007 are summarized in Table 2.
The mean and standard deviation were calculated based on all samples (for the samples that tested negative,
the analytical detection limit was used in the calculations) Enterococci had the highest mean concentration
in both CMAD inﬂuent and CMAD efﬂuent samples, but coliphage had the highest mean concentration
and occurrence in the AnMBR efﬂuent. Two, three and none of the AnMBR efﬂuent samples were tested
positive for E. coli, enterococci, and C. perfringens, respectively. The mean values for the E. coli and
enterococci in AnMBR samples were 0.31 and 0.51 loglocfu/L, respectively. The occurrence of E. coli and

163

enterococci in the AnMBR efﬂuent was likely due to passage of bacteria through membranes, which had
been documented in the literature (Delebecque et al. 2006). Five out of eight AnMBR samples were
positive for coliphage and the mean level of all samples was 2.47 loglopfu/L. Coliphage had the highest
occurrence frequency and concentration in AnMBR efﬂuent as expected since viruses are generally much

smaller than bacteria and the diameter of viruses could be as small as 0.01 pm.

The loglo removals of indicators by the CMAD and CMAD/AnMBR are illustrated in Figure 2.
The error bars in the ﬁgure represent the standard deviation between different sampling events. The logic
removals of E. coli, enterococci, C. perfringens and coliphage by AnMBR and CMAD were 5.2, 6.1, 6.4,
3.7 and 1.5, 1.2, 0.1, and 0.5, respectively. The total log", removals of E. coli, enterococci, C. perfringens
and coliphage by the entire system were therefore 6.7, 7.3, 6.5 and 4.2, respectively. These results
demonstrated that most of the overall removals were attributed to the AnMBR. One possible explanation
for the low removal efﬁciency of C. perﬁingens by the CMAD may be attributed to its tendency to exist in
the spore form under natural environmental conditions. The CMAD inﬂuent was the natural raw manure
and spores are known as extremely resistant to treatment processes. C. perfringens has been used as a

surrogate organism for parasites (e.g. Cryptospordium oocyst) (Yates, 2007).

Signiﬁcant differences between the microbial concentrations at the three sampling points were
analyzed by the ANOVA single test. Signiﬁcant reductions were observed for all indicator concentrations
between the CMAD efﬂuent and the AnMBR efﬂuent, as indicated by p-values in the range of 10'7 to 10''2
(data not shown). Although a signiﬁcant reduction in E. coli, enterococci and coliphage were also observed
after CMAD treatment, the extent of this initial reduction was far less than that attributed to AnMBR

treatment (Figure 2).

Coliphage removals observed in this study were compared with removals stated in the literature
using domestic wastewater. Lv et al. (2005), Zheng et al. (2005), Oota et al. (2005) and Ahn et al. (2001)
reported 98-100%, 90%, 100% and 100% of coliphage removals by an MBR, respectively. 2.3, 5.3, and 5.9
log", removals were observed in other studies (Ottoson et al. 2006; Comerton et al. 2005 and Ueda and

Horan 2000). All of these studies evaluated the performance of MBR systems in treating domestic

164

.._..]. I I I
r } ‘.

._l!‘

wastewater. These results were similar to our ﬁndings for coliphage removal by AnMBR (99.96% and 3.7

loglo removal).

Comparison of the Microbial Removals by the A nMBR System alone to the Combined CMAD/AnMBR
System

From February to April 2007, the AnMBR system was operated independent of the CMAD. The
log", removals of E. coli and enterococci by this system were 6.9 and 7.3. Figure 3 illustrates the
comparison of E. coli and enterococci removals by the AnMBR system alone to the combined
CMAD/AnMBR system. Results showed there was no difference in the removals of E. coli and
enterococci between these two systems. These results indicate that an AnMBR could achieve similar
microbial removal performance if challenged with the same feedstock received by the CMAD. However,
we’d like to note that coupling a CMAD with an AnMBR may provide an economic advantage due to the
reduced TS concentration of the substrate fed to the AnMBR since membrane bioreactor ﬂux rates decline
with increasing total solids concentration (Anderson et al. 1986; Beaubien et al. 1996; Ross et. al., 1990)

and the ﬂux rate directly impacts the energy required to operate the circulation pump.
Removal of Bovine Polyomaviruses and Adenoviruses in the Combined CMAD/AnMBR System

The number of measurements and occurrence frequency of animal enteric viruses in the samples
are summarized in Table 3. Eight samples for each sampling location were tested for bovine
polyomaviruses (BPyV) and bovine adenoviruses (BAdV). All of the CMAD inﬂuent and CMAD efﬂuent
samples were BPyV and BAdV positive. None of the AnMBR efﬂuent samples tested positive for BAdV

but there were three BPyV positive samples.

Interestingly, more samples also tested BPyV positive than BAdV by Hundesa et al. (2006), when
slaughterhouse wastewater and river water were tested. In their study, BAdV was detected in only one
sample but twenty-two samples were BPyV positive. These results may suggest the higher prevalence of
BPyV than BAdV in animal waste. Also, Polyomaviruses (35-40nm) are roughly half the size of the
adenoviruses (60-90nm) (Hurault de Ligny eta12000, Thomas 2004); the size difference between these

two viruses could be one of the factors explaining the fact that only BPyV were detected in AnMBR

165

effluent. However, this study used the same PCR methods as Hundesa et al. (2006). Similarities in the

proportion of virus types may also be attributable to the differences in the BPyV and BAdV PCR method

sensitivities.

In order to conﬁrm the PCR results, seven BAdV positive samples selected from CMAD inﬂuent
and CMAD efﬂuent and seven BPyV positive samples selected from all three locations were sent for
sequencing. The sequencing results showed all seven samples tested BPyV positive were 100% similar to
the nucleotide sequence of BPyV in the genebank. For BAdV, two samples were 85% to 86% similar to

BAdV type 2 and ﬁve samples were 99 to 100% similar to BAdV type 7.
CONCLUSIONS

The removal of pathogenic indicators and animal viruses from agricultural waste by an AnMBR
pilot system was evaluated. The mean logm removals of E. coli, enterococci, C. perfringens and coliphage
by both CMAD and AnMBR were 6.7, 7.3, 6.5 and 4.2, respectively but most of the removals were
attributed to the AnMBR. Three AnMBR efﬂuent samples tested BPyV positive but none tested BAdV
positive. The indicator that was found most frequently and had the highest concentration in the AnMBR
efﬂuent was coliphage. This suggests that coliphage or viruses would be suitable indicators for evaluating
the AnMBR performance. The AnMBR also demonstrated signiﬁcant removal of TKN, TP, COD, TS, and
VS. Overall, the CMAD digester, which is one of the suggested practices of treating animal wastes, had
much lower removals of indicators and animal viruses as well as physical and chemical parameters
compared to the AnMBR system. This study demonstrates that AnMBR systems could be considered as an
alternative treatment for animal waste especially when high removal of zoonotic pathogens is required. The
economic feasibility of such systems may be increased if a consortium of community users and farmers

could support the use of membrane systems for the co-treatment of human and animal waste.

ACKNOWLEDGMENTS

We thank Dr. Phanikumar Mantha for his advice on statistical analysis.

166

 

REFERENCES

Ahn K. H., K. G. Song, I. T. Yeom, and K. Y. Park. 2001. Performance comparison of direct membrane
separation and membrane bioreactor for domestic wastewater treatment and water reuse. Water Sci and
Tech. Water Supply. 1: 315-323.

Anderson, C. K., C. B. Saw, and M. I. A. P. Fernandes. 1986. Application of porous membranes for
biomass retention in biological wastewater treatment processes. Process Biochem. 21:174.

Beaubien, A. M., F. J. Baty, E. Francoeur, and J. Manem. 1996. Design and operation of anaerobic
membrane bioreactors: development of a ﬁltration testing strategy. J Membr Sci 109: 173 — 184.

Bisson, J ., and V. Cabelli. 1979. Membrane ﬁltration enumeration method for Clostridium perfringens.
Appl. Environ. Microbiol. 37:55.

Cicek, N. 2003. A review of membrane bioreactors and their potential application in the treatment of
agricultural wastewater, Can. Biosyst. Eng. 45: 637—646.

Comerton, A. M., R. C. Andrews, and D. M. Bagley. 2005. Evaluation of an MBR-R0 system to
produce high quality reuse water: microbial control, DBP formation and nitrate. Water Res. 39:3982-3990

Costantini, V. P., A. C. Azevedo, X. Li, M. C. Williams, F. C. Michel, Jr, and L. J. Saif. 2007. Effects
of Different Animal Waste Treatment Technologies on Detection and Viability of Porcine Enteric Viruses.
Appl. Environ. Microbiol. 73: 5284-5291.

Delebecque, N., C. Causserand, C. Roques, and P. Aimar. 2006. Membrane processes for water
disinfection: investigation on bacterial transfer mechanisms. Desalination. 199: 81-83.

Fakhru’l-Razi, A. 1994. Ultraﬁltration membrane separation for anaerobic wastewater treatment. Water
Sci. Technol. 30: 321—327.

Franson, M.A.H. 1998. Standard method for the examination of water and wastewater. 20th ed. American
Public Health Association, American Water Works Association, Water Environment Federation, United
Book Press, Baltimore, MD.

Fuchs, W., H. Binder, G. Mavrias and R. Braun. 2003. Anaerobic treatment of wastewater with high
organic content using a stirred tank reactor coupled with a membrane ﬁltration unit. Water Res. 37: 902—
908.

Gerba, C. P., and J. E. Smith, Jr. 2005. Sources of pathogenic microorganisms and their fate during land
application of wastes. J. Environ. Qua]. 34:42-48.

Haramoto, E., H. Katayama, K. Oguma, and S. Ohgaki. 2005. Application of cation-coated ﬁlter
method to detection of noroviruses, enteroviruses, adenoviruses, and torque teno viruses in the Tamagawa
River in Japan. Appl. Environ. Microbiol. 71:2403-2411.

Holme, R. 2003. Drinking water contamination in Walkerton, Ontario: positive resolutions from a tragic
event. Water Sci. Technol. 47:1-6.

Hundesa, A., C. Maluquer de Mates, 8. Bofill-Mas, N. Albinana-Gimenez, and R. Girones. 2006.
Identiﬁcation of human and animal adenoviruses and polyomaviruses for determination of sources of fecal
contarrrination in the environment. Appl Environ Microbiol 72:7886-93.

Hurault de Ligny, B., I. Etienne, A. Francois, O. Toupance, M. Buchler, G. Touchard, P. Lepogamp,

F. Comoz, T. Lobbedez, M. Godin, J. P. Ryckelynck, and Y. Lebranchu. 2000. Polyomavirus-induced
acute tubulo-interstitial nephritis in renal allograﬂ recipients. Transplant Proc. 32:2760-2761.

167

Johnson, G. (ed.) 2004. Membrane Filtration of Manure Wastewater; A Comparison of Conventional
Treatment Methods and VSEP, a Vibratory RO Membrane System. New Logic Research, Incorporated.
Emeryville, CA. (Available online at http://www.vsep.com/pdf/Membrane ﬁltra_t_h)n_oftmanure.pdﬁ
(accessed 8 May. 2007; veriﬁed 7 Dec. 2007).

Lv, W., X. Zheng, M. Yang, Y. Zhang, Y. Liu, and J. Liu. 2006. Virus removal performance and
mechanism of a submerged membrane bioreactor. Process Biochern 41:299-304.

Oota, S., T. Murakami, K. Takemura, and K. Noto. 2005. Evaluation of MBR efﬂuent characteristics
for reuse purposes. Water Sci Technol. 51 :441-446.

Ottoson, J ., A. Hansen, B. Bjorlenius, H. Norder, and T. A. Stenstrom. 2006. Removal of viruses,

parasitic protozoa and microbial indicators in conventional and membrane processes in a wastewater pilot
plant. Water Res. 40: 1449-1457.

Ross, W. R., J. P. Barnard, J. Leroux, and H. A. Devilliers. 1990. Application of ultraﬁltration

membranes for solids ~1iquid separation in anaerobic digestion systems: The Aduf Process. Water Sa.
16:85-91.

Shang, C., H. M. Wong, and G. Chen. 2005. Bacteriophage MS-2 removal by submerged membrane
bioreactor. Water Res. 39: 421 1—42 19.

Thomas, J.J., B. Bothner, J. Traina, W. H. Benner and G. Siuzdak. 2004. Electrospray ion mobility
spectrometry of intact viruses. Spectroscopy 18:31- 36.

Ueda, T., and N. Horan. 2000. Fate of indigenous bacteriophage in a membrane bioreactor. Water Res.
34:2151-2159

Peters, J. (ed.) 2003. Recommended methods of manure analysis. Cooperative Extension Publishing
Operations, Madison, WI. (Available online at http://cecommerce.uwex.edu/pdfs/A3769.PDF) (accessed 17
Mar. 2005; veriﬁed 25 Aug. 2006).

USEPA. 2001. Method 1602: Male-speciﬁc (F+) and Somatic Coliphage in Water by Single Agar Layer
(SAL). EPA 821-R-Ol-029. Ofﬁce of Water, US. Environmental Protection Agency, Washington, DC.

USEPA. 2002. Method 1600: enterococci in water by membrane ﬁltration using membrane-enterococcus
indoxyl-beta- D- glucoside agar (mEI). EPA 821-R-02- 022. Ofﬁce of Water, U. S. Environmental Protection
Agency, Washington, DC.

USEPA. 2002. Method 1603: Escherichia coli in water by membrane ﬁltration using modiﬁed membrane-
thermotolerant Escherichia coli agar (modiﬁed mTEC). EPA 821-R-02-023. Ofﬁce of Water, US.
Environmental Protection Agency, Washington, DC.

USEPA. 2006. Prepublication of the Groundwater Rule. Federal Register Notice. EPA-HQ-OW-2002-
0061; FRL-RIN2040-AA97. USEPA, Ofﬁce of Water, Washington, DC.

Yates, M. V. 2007. Classical Indicators in the let CenturyFar and Beyond the Coliform Water Environ.
Res. 79: 279-286.

Zheng, X., W. Lu, M. Yang, and J. Liu. 2005. Evaluation of virus removal in MBR using coliphages T4.
Chin. Sci. Bull. 50:862-867

168

TABLES and FIGURES

Table 1. Water quality in the combined CMAD/AnMBR system.

 

 

Removal by
CMAD Inﬂuent CMAD Efﬂuent AnMBR Efﬂuent Entire System
11 Mean SD Mean SD Mean SD Mean (%) SD
TKN (mg/L) 7 2,120 203 2,160 126 1,440 84 31.32 8.64
Ammonium
nitrogen 31.8
(mg/L) 7 1,240 469 1,300 91 1,330 89 -l6.55 9
TP (mg/L) 7 343 28 317 78 14 5.3 95.84 1.52

COD (mg/L) 7 44,900 12,100 31,800 6,560 3,440 705 92.03 1.78
TS (%) 7 4.54 0.69 3.36 0.64 0.80 0.15 81.71 5.25
VS (%) 7 2.97 0.32 2.21 0.45 0.27 0.05 90.66 2.35

N, number of measurements; SD, standard deviation; COD, chemical oxygen demand; TKN, total
Kjeldahl nitrogen; TP, total phosphate; TS, total solids; VS, volatile solids.

Table 2. Pathogen indicators occurrence in the combined CMAD/AnMBR system.

 

CMAD Inﬂuent CMAD effluent AnMBR efﬂuent

Mean SD F Mean SD F Mean SD F

 

E. coli (loglocfu/L) 7.01 0.51 8/8 6.46 0.76 8/8 0.31 0.58 2/8
Enterococci (loglocfu/L) 7.87 0.92 8/8 6.71 1.18 7/8 0.51 1.02 3/8
C. perﬁngens (logmcfu/L) 6.51 0.47 8/8 6.39 0.81 878 ND - 0/8
Coliphage (logmpfu/L) 6.64 0.54 8/8 6.14 0.37 8/8 2.47 0.43 5/8

 

 

 

 

F, fraction of positive samples; ND, none detected.

169

 

 

Table 3. Animal enteric virus occurrence in the combined
CMAD/AnMBR system.

 

CMAD CMAD AnMBR

 

Animal Viruses Inﬂuent Efﬂuent Efﬂuent
Bovine Adenoviruses 8/ 8 8/8 0/8
Bovine Polyomaviruses 8/8 8/8 3/8

 

 

170

 

.Bamomﬁ 05835 xi: 82988 .20
588.85 655565 658228 .Mmzé .m was :0. J 3:69 E 0:26 we? wczgﬁmm $83206. 20:3 2: ﬁnch: H53 Boa “mama/w 8 0:3. EOE

m

.m was H 350m E 0:8 mm? mEEEam .Qﬁzo @3393 Boa E9». 8 being 58m .BmaB EEEa Snob SE 589$ Bmoﬁozm 2:. ._ “warm

285m $25. a sauna 9:20 . oﬁzu 338:3 . neon mainam

.828 ._..... -w 6,66. .
, 7 . .. 28m
wage? ... .. L , , w . . u .
mvzom l w , , . . . . . . , .. iii/1
SEES.» ovocwmz .11.} U . .3:qu .
it. s ,7. 1 image.) . . - 1 ... .1 9.: can
4. 983 A l l, W .. 1. 1 coﬁsaom
, noun—:86 » . 95E . cam
, gum .... , mﬁxuz H86
, 0355280, 5on W EMA
, , . 1. 6F . 6.3
H. l .11.! 1 one H63
1 - _ m 830 8 $0
, >
055802
RD
5395. 1
58:2

86

171

I Removal by AnMBR

 

9
8 ‘ DRemovalbyCMAD
4
l
74.
E 6i
O
E :
é 5‘:
r: q
3 '1
l
2 ..
11
0H: . *'*r* #7.. —~.- 7

Eco“ Enterococci C. perﬁngens Coliphage

Figure 2. Log”) removal of Indicators by the combined CMAD/AnMBR system. Sampling period

from June to August.

172

DAnMBR system alone
9 ‘ ICombined CMAD/AnMBR system

LII ON
_1_. 1 _ .g.

 

 

 

T:

>

G

E 4

a:

2

go 3

.J
2
l
01_- .!w_

 

E. coli Enterococci

Figure 3. Comparison of logm removals of E. coli and enterococci by the AnMBR system and the
combined CMAD/AnMBR system. For light bars: sampling occurred from February to April; for

dark bars: sampling occurred from June to August.

173

APPENDIX B
Volatile Fatty Acid Procedure

The following is adopted from “A Direct Method for Differentiating Bicarbonate
and Acetate in Digester Control”, (Obrien and Donlan 1977).

1. Filter at least 50 m1 of Reactor Liquid.
2. Transfer 50.0 ml of ﬁltrate into 150 ml beaker and titrate to pH 3.3 with 0.10

N H2SO4. Record reading in ml. as reading #1 on Volatile Acids Analysis

 

Worksheet.
3. Cover Sample beaker with 65 mm watch glass and bring to boil for 60-90 sec.
4. Cool to room temperature and rinse watch glass into beaker with distilled
water.

5. Titrate sample to pH 4.0 exactly using 0.050 N NaOH. Record volume as #2

on work sheet.

6. Continue titration to pH 5.1 with 0.050 N NaOH. Record volume as #3 on

worksheet.
Calculations:
Total Alkalinity: (TA) as mg/l CaCO3 = (#1) x 100
Volatile Acids: (VA) as mg/l Acetic = (#3 - #2) x 100
Bicarbonate Alkalinity: as mg/l CaCO3 = (TA) — 0.83 x VA

174

APPENDIX C

Most Probable Number Methodology

Non-Selective Medium for Growth off Ceca] and Manure Anaerobic Bacteria (Adapted

from Caldwell and Bryant (1966).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Substrate % Contribution to 300 mL medium solution
Glucose 0.05 0.15 g
Cellobiose 0.05 0.15 g
Soluble starch 0.05 0.15 g
Xylose 0.05 0.15 g
Trypticase 0.2 0.6 g
Yeast extract 0.2 0.6 g
Mineral #1 . 3.75 11.25 ml
Mineral #2 3.75 11.25 m1
Rumen ﬂuid] 20.0 60 ml
Resazurin 0.1 ml/100 0.3 ml
ml

 

 

 

 

lRumen ﬂuid is clariﬁed by centrifugation at 15,000 x g for 20 minutes and autoclaved at 120°C at 15 psi
for sterilization prior to use in medium. Keep refrigerated until use.

Add enough distilled water to bring all of the above ingredients to a volume of 300 ml,
taking into account addition of the sodium bicarbonate and cysteine-sulﬁde solution
volumes (21 m1) below later.

Bring medium to gradual boil in a 500 m1 round bottom ﬂask (or 1000 m1 Erlenmeyer
ﬂask) under C02, until steam evolves and the medium changes to a reddish color. Cool
under ice to the touch and, while continuing to ﬂush the ﬂask with C02, add:

175

0 Sodium bicarbonate (8%) solution, 5 m1/100 ml for a total of 15.0 ml

0 Cysteine-sulﬁde (2.5% solution) 2 m1/100 ml for a total of 6.0 ml
Bubble C02 into the medium for a few minutes after adding the sodium bicarbonate and
cysteine-sulﬁde solution, then ﬂush headspace of the ﬂask with CO; while tubing into
Hungate tubes (9 ml per tube) with the tubes also under C02. Avoid blowing bubbles
into the tube during the pipetting process. Crimp the lids down tightly with a crimper and

autoclave the tubes at 120°C for 20 minutes at 15 psi.

 

Mineral Solution #1: 0.6% KzHPO4

Mineral Solution #2: To 100 ml distilled water add:

 

 

 

 

 

 

Compound Masig
KHZPO4 0.6
(NH4)2SO4 0.6
NaCl 1 .2
MgSO4-7H20 0.25
CaC12-2HZO 0.16

 

 

 

 

176

APPENDIX D

Example AnMBR Analysis

Operating Conditions:
SRT = 27 days

HRT = 12 days

VS destruction = 38%

Digester operating concentration = 3% VS

Membrane ﬂux rate = 40 L/mZ/hr = 960 L/mz/day

14.4 mm diameter membrane x 6000 mm, surface area per module = 1.89 m2
Cross ﬂow velocity = 4.5 m/s, required ﬂow rate per module = 310 L/minute

Desire 75% recovery of inﬂuent as UF permeate (40,000 L/day x 75% = 30,000
L/day)

0 Pressure drop per module = 100 kPa

 

38% VS Destruction
389 kg/day as Biogas
l
l ' . C" F”
40,000 L/day , 12.18681?!
vs = 2.6% ~-— -——--‘, Tank ,
Total vs = 1024 kg/day , . l m_ _ _ _-
1480,000 Liters 1
;vs = 3% l

'iTotal VS=14,400 i I

30,000 1../day
530 kg VS = 0.35%
Total VS = 105 kg/day

SRT Calculation

The SRT can be set to a deﬁned value based on HRT or based on digester operating VS
concentration. In the example above, SRT = 14,400 kg + 530 kg = 27 days. If the HRT
were decreased from 12 days to 6 days with all else remaining constant, the SRT would

decrease to 13.5 days. Alternatively, if a HRT of 6 days is desired with a corresponding

177

SRT of 27 days, this condition requires that the digester be operated at a VS
concentration of 6%. There will be an energy penalty for operating at the higher VS
concentration.

Cycle T3115 Calculation

Required Membrane Su ace Area
30,000 L/day + 960L/m -d = 31.25 m2, require 31.25 m2 + 1.89 m2 = 16 modules

Membrane Conﬁguration
Assume modules operated in parallel, 310 L/min x 16 modules = 7,142,400 L/day

Cycle time = 7,142,400 L/day + 480,000 L = 15 cycles/day
If modules are placed so that there are 8 sets of 2 in series, the cycle time would be

decreased to 7.5 cycles per day.

Ener Calculations ased on 14.4 mm x 6 000 mm module :

Pressure drop/module at approximately 3% VS (approximately 5% TS) = 100 kPa
Assume with other system losses, the design pressure = 200 kPa (20.4 m H20 @
20°C)

Pump efﬁciency = 83%

Motor efﬁciency= 92%

Flow rate through membrane system, 7, 142, 400 L/day (297. 6 m 3/hr)

Liquid density— — 1000 kg/m3

Methane production rate, 261 L CH4/kg VS actual conditions

NIST standard conditions, T= 20°C, P = 101.325 kPa

Methane production at STP = 232 L CH4/kg VS

34.6 MJ/m3 methane (assumes LHV)

178

 

Ph = q-p-g-h-I-3.6x106
Where,

Ph = power (kW)

q = ﬂow capacity (m3/hr)

p = density of ﬂuid (kg/m3)

g = gravity (9.8 m/sz)

h = differential pressure head, (m)

P., = 297.6 m3/hr x 1000 kg/m3 x 9.8 m/sz x 20.4 m + 3.6x106 = 16.5 kW

 

 

Total power required = 16.5 kW + 0.83 (pump eff.) + 0.82 (motor eff.) = 21.6 kW

 

Comparison of potential energy from biogas:

1024 kg VS fed x 232 L CH4/kg VS fed = 237,568 L CH4 produced per day

Converting to electricity equivalent and assuming 35% conversion efﬁciency
34.6 MJ/m3 x 237.568 m3 = 8220 MJ

8,220 M] x 0.2778 kW-hr/MJ = 22,283 kW-hr

 

 

2,283 kW-hr +24 hr/day x 35% (conversion to electricity efﬁciency) = 33.3 kW

 

179

 

 

APPENDIX E
Gas Chromatograph Mass Spectroscopy Protocol

Instrument

Agilent 5973 inert mass selective detector with autosampler

Column

30 meter DBWAX, 0.25 mm inner diameter x 0.25 pm ﬁlm thickness

Temperature Program

50°C increasing at 20°C per minute to 120°C where held for 5 minutes, followed by 2°C
increase each minute to 130°C, followed by 40°C increase per minute to 240°C. The
total run time for the heating routine was 16.25 minutes. Total volume was 1 111 injected

in a splitless mode and a constant ﬂow of 1.5 mL per minute, scanning masses 10 through

300.

Quantification
Extracted ion chromatograms of a m/z (mass-to-charge-ratio) indicative for each
compound (acetic, propionic, butyric and Isobutyric acids) were integrated and the areas

entered into a Microsoft Excel program using a standard curve of 0 through 500 ng/ 111 or

ug/mL.

180

 

Appendix F

Photographs of Pilot Digesters

 

Complete Mix Digester

 

View of SM AnMBR (grey tube is the single membrane housing)

181

 

View of identical digester tanks, the tank to the left was used for MED, MMED and
MM AnMBR, the tank to the right for SMD and SM AnMBR

182

LIST OF REFERENCES

Ahn, J. H. and C. F. Forster (2002). "The effect of temperature variations on the
performance of mesophilic and thermophilic anaerobic ﬁlters treating a simulated
paperrnill wastewater." Process Biochemistry 37(6): 589-594.

Amon, T., Amon, B., Kryvoruchko, V., Zollitsch, W., Mayer, K., Gruber, L. (2007).
"Biogas production from maize and dairy cattle manure - Inﬂuence of biomass

composition on the methane yield." Agriculture, Ecosystems and Environment
118: 173-182.

Anderson, G. K., A. James, C.B. Saw (1986). Crossﬂow microﬁltration - A membrane
process for biomass retention in gaerobic digestion. World Filtration Conference,
Ostend, Belgium.

Anderson, G. K., B. Kasapgil, et al. (1994). "Comparison of porous and nonporous media
in upﬂow anaerobic ﬁlters when treating dairy waste-water." Water Research
28(7): 1619-1624.

 

Anderson, G. K., B. Kasapgil, et al. (1996). "Microbial kinetics of a membrane anaerobic
reactor system." Environmental Technology 17(5): 449-464.

Anderson, G. K., C. B. Saw, et al. (1986). "Applications of porous membranes for
biomass retention in biological waste-water treatment processes." Process
Biochemistry 21(6): 174-182.

Archer, D. B. and G. E. Powell (1985). "Dependence of the speciﬁc growth-rate of
methanogenic mutualistic cocultures on the methanogen." Archives of
Microbiology 141(2): 133-137.

Baker, R. J ., A. G. F ane, et a1. (1985). "Factors affecting ﬂux in cross-ﬂow ﬁltration."
Desalination 53(1-3): 81-93.

Baker, R. W., Ed. (2000). Membrane Technology and Applications. West Sussex,
England, John Wiley & Sons.

Beaubien, A., M. Baty, et al. (1996). "Design and operation of anaerobic membrane
bioreactors: Development of a ﬁltration testing strategy." Journal Of Membrane
Science 109(2): 173-184.

Beede, D. (2003). "Michigan Dairy Review." Michigan Dairy Review 8(1): 3-5.

Berube, P. R., E. R. Hall, et al. (2006). "Parameters governing permeate ﬂux in an

anaerobic membrane bioreactor treating low-strength municipal wastewaters: A
literature review." _VlLater Environment Research 78(8): 887-896.

183

Bickert, W. (2006). Professor - Emeritus. Integrated manure management discussion. J.
Wallace. East Lansing. '

Boone, D. R. and M. P. Bryant (1980). "Propionate-degrading bacterium,
syntrophobacter-wolinii sp-nov gen-nov from methanogenic ecosystems."
Applied And Environmental Microbioltgy 40(3): 626-632.

Brockmann, M. (1995). Crossﬂow-Microﬁltration systems in wastewater treatment - a
biological approach. Filtech Europa/95 conference, Karlsruhe, Germany.

Brockmann, M. and C. F. Seyfried (1996). "Sludge activity and cross-ﬂow microﬁltration
a non-beneﬁcial relationship." Water Science And Technology 34(9): 205-213.

Brockmann, M. and C. F. Seyﬁied (1997). "Sludge activity under the conditions of
crossﬂow microﬁltration." Water Science And Technology 35(10): 173-181.

Brockmann, M. a. S., C. (1995). Cross-ﬂow microﬁltration systems in wastewater
treatment - a biolog'cal approach. Filtech Europa/95 Conference, October 10 - 12,
1995, Karlsruhe, Germany. '

Bryant, M. P. (1979). "Microbial methane production - Theoretical aspects." Journal Of
Animal Science 48(1): 193-201.

Cadi, 2., H. Huyard, et al. (1994). "Anaerobic-digestion of a synthetic waste-water
containing starch by a membrane reactor." Environmental Technology 15(11):
1029-1039.

Caldwell, D. R. and M. P. Bryant (1966). "Medium without rumen ﬂuid for nonselective
enumeration and isolation of rumen bacteria." Applied Microbiology 14(5): 794.

Castrillon, L., I. Vazquez, et al. (2002). "Anaerobic thermophilic treatment of cattle
manure in UASB reactors." Waste Management & Research 20(4): 350-356.

Chackhiani, M., P. Dabert, et al. (2004). "16S rDNA characterisation of bacterial and
archaeal communities during start-up of anaerobic thermophilic digestion of cattle
manure." Bioresource Technology 93(3): 227-232.

Chang, I. S., P. Le Clech, et a1. (2002). "Membrane fouling in membrane bioreactors for

wastewater treatment." Journal Of Environmental Engineering-Asce 128(1 1):
1 01 8-1029.

Chapman, J. M., V. R. Phillips, et al. (1990). "Mesophilic anaerobic-digestion of dairy-
cow slurry on a farm scale - maintenance requirements and reliability." J oumal Of
Agricultural Engineering Research 47(4): 277-285.

184

 

Chayovan, S., J. B. Gerrish, et al. (1988). "Biogas production from dairy manure - The
effects of temperature perturbations." Biological Wastes 25(1): 1-16.

Cheryan, M. (1986). Ultraﬁltration Handbook. Lancaster, PA, Technomic.

Choate, V. T., Houldsworth, et al. (1983). Membrane enhanced anaerobic digesters. 37th
Industrial Waste Conference, Purdue University, West Lafayette, IN.

Choo, K. H. and C. H. Lee (1996). "Effect of anaerobic digestion broth composition on
membrane permeability." 1N_ater Science And Technology 34(9): 173-179.

Choo, K. H. and C. H. Lee (1996). "Membrane fouling mechanisms in the membrane-
coupled anaerobic bioreactor." _W_ater Research 30(8): 1771-1780.

Chynoweth, D. P., C. E. Turick, et al. (1993). "Biochemical methane potential of biomass
and waste feedstocks." Biomass & Bioenergy 5(1): 95-111.

Cicek, N., J. P. Franco, et al. (1998). "Using a membrane bioreactor to reclaim
wastewater." Journal American Water Works Association 90(11): 105-113.

Converse, J. C. a. K., KG. (2002). Nutrient and solids separation of ﬂushed dairy manure
by settling. ASAE Annual Intemational Meeting / CIGR XVth World Congress,
Chicago, IL.

Dague, R. R., R. E. McKinney, et al. (1970). "Solids retention in anaerobic waste
treatment systems." Journal Water Pollution Control Federation 42(No.2, Part 2):
R29-R46.

Demirer, G. N. and S. Chen (2004). "Effect of retention time and organic loading rate on
anaerobic acidiﬁcation and biogasiﬁcation of dairy manure." Journal Of Chemical
Technology And Biotechnology 79(12): 1381-1387.

Demirer, G. N. and S. Chen (2005). "Two-phase anaerobic digestion of unscreened dairy
manure." Process Biochemistry 40(11): 3542-3549.

Dolﬁng, J. and W. Bloemen (1984). "Characterization of anaerobic sludge by
determination of the methane production-rate." Antonie Van Leeuwenhoek
Journal Of Microbiology 50(1): 86-86.

Dolfrng, J. and W. Bloemen (1985). "Activity measurements as a tool to characterize the
microbial composition of methanogenic environments." J oumal Of

Microbiological Methods 4(1): 1-12.

du Preez, J ., B. Norddahl, et al. (2005). "The BIOREK((R)) concept: a hybrid membrane
bioreactor concept for very strong wastewater." Desalination 183: 407-415.

185

Dugba, P. N. and R. H. Zhang (1999). "Treatment of dairy wastewater with two-stage
anaerobic sequencing batch reactor systems - thermophilic versus mesophilic
operations." Bioresource Technology 68(3): 225-233.

Eastman, J. A. and J. F. Ferguson (1981). "Solubilization of particulate organic-carbon
during the acid phase of anaerobic-digestion." Journal Water Pollution Control
Federation 53(3): 352-366.

Epstein, T. and B. Korchin (1981). Cheese wlm ultraﬁltration with ﬂat plate pplysulfone
membranes. 9lst National Meeting, AICHE, Detroit, MI.

Fakhru'l-Razi, A. (1994). "Ultraﬁltration membrane separation for anaerobic wastewater
treatment." Water Science And Technology 30(12): 321-327.

Fane, A. G. and C. J. D. Fell (1987). "A Review Of Fouling And Fouling Control In
Ultraﬁltration." Desalination 62: 117-136.

Fuchs, W., H. Binder, et al. (2003). "Anaerobic treatment of wastewater with high
organic content using a stirred tank reactor coupled with a membrane ﬁltration
unit." Water Research 37(4): 902-908.

Fugere, R., N. Mameri, et al. (2005). "Treatment of pig farm efﬂuents by ultraﬁltration."
Journal Of Membrane Science 255(1-2): 225-231.

Gao, M. C., Z. L. She, et al. (2007). "Performance evaluation of a mesophilic (37 degrees
C) upﬂow anaerobic sludge blanket reactor in treating distiller's grains
wastewater." Journal of Hazardous Materials 141(3): 808-813.

GarciaMorales, J. L., E. Nebot, et al. (1996). "Methanogenic and acidogenic activity test
in an anaerobic thermophilic reactor." Biotechnology Techniques 10(4): 249-252.

Gavala, H. N., H. Kopsinis, et al. (1999). "Treatment of dairy wastewater using an
upﬂow anaerobic sludge blanket reactor." Journal Of Agricultural Engineering
Research 73(1): 59-63.

Ghaly, A. E. and E. A. Echiegu (1992). "Performance evaluation of a continuous-ﬂow

no-mix anaerobic reactor operating on dairy manure." Energy Sources 14(2): 113-
134.

Ghaly, A. E. and E. A. Echiegu (1993). "Performance evaluation of an intermittent ﬂow
continuous mix anaerobic reactor operating on dairy manure." Biomass &

Bioenergy 4(3): 185-197.

Ghyoot, W. R. and W. H. Verstraete (1997). "Coupling membrane ﬁltration to anaerobic
primary sludge digestion." Environmental Technology 18(6): 569-5 80.

186

 

Grethlein, H. E. (1978). "Anaerobic digestion and membrane separation of domestic
wastewater." Journal Water Pollution Control Federation 50(4): 754-763.

Hamdi, M. and J. L. Garcia (1991). "Comparison between anaerobic ﬁlter and anaerobic

contact process for fermented olive mill wastewaters." Bioresource Technolggy
38(1): 23-29.

Harper, S. R., Pohland, F.G. (1985). "Recent developments in hydrogen management
during anaerobic biological wastewater treatment." Biotechnology And
Bioengineering 28: 585-602.

Hawkes, F. R., T. Donnelly, et al. (1995). "Comparative performance of anaerobic
digesters operating on ice-crearn waste-water." _Water Research 29(2): 525-533.

He, Y. L., P. Xu, et al. (2005). "High-concentration food wastewater treatment by an
anaerobic membrane bioreactor." Water Research 39(17): 41 10-4118.

Henze, M. and P. Harremoes (1983). "Anaerobic treatment of wastewater wastewater in
ﬁxed ﬁlm reactors - a literature review." Water Science And Technology 15(8): 1-
101.

Hickey, R. (2007). Organic loading rate of anaerobic contact process and UASB solids
tolerance discussion. J. Wallace. East Lansing.

Hills, D. J. (1979). "Effects of carbon - nitrogen ratio on anaerobic-digestion of dairy
manure." Agricultural Wastes 1(4): 267-278.

Hills, D. J. and D. W. Roberts (1981). "Anaerobic-digestion of dairy manure and ﬁeld
crop residues." Agricultural Wastes 3(3): 179-189.

Hoffrnann, R. A., M. L. Garcia, et a1. (2008). "Effect of shear on performance and
microbial ecology of continuously stirred anaerobic digesters treating animal
manure." Biotechnology And Bioengineeripg 100(1): 38-48.

Hungate, R. E. (1975). "The rumen microbial ecosystem." Annual Reviews.

Ianotti, E. L., Kaﬂtewitz, P., Wolin, M.J., and Bryant, MP. (1973). "Glucose
fermentation products of Rurninococcus albus grown in continuous culture with

Vibrio succinogenes: changes caused by interspecies transfer of H2." L
Bacteriol(114): 1231-1240.

Ince, B. K., O. Ince, et a1. (2001). "Assessment of biogas use as an energy source from
anaerobic digestion of brewery wastewater." Water Air And Soil Pollution 126(3-
4): 239-251 .

187

 

Ince, B. K., O. Ince, et a1. (2000). "Inert cod production in a membrane anaerobic reactor
treating brewery wastewater." Water Research 34(16): 3943-3948.

Ince, 0. (1998). "Potential energy production from anaerobic digestion of dairy
wastewater." Journal Of Environmental Science And Health Part A-
Toxic/Hazardous Substances & Environmental Engineering 33(6): 1219-1228.

Ince, 0., G. K. Anderson, et al. (1995). "Effect of changes in composition of
methanogenic species on performance of a membrane anaerobic reactor system
treating brewery waste-water." Environmental Technology 16(10): 901 -914.

Inglis, S. F., Gooch, C.A., Jones, L.R., Aneshansley, D. (2007). Cleanout of a plug-410w
anaerobic digester after ﬁve years of continuous operation. International

Symposium on Air Quality and Waste Management for Agriculture, Broomﬁeld,
CO.

James, A., C. A. L. Chernicharo, et a1. (1990). "The development of a new methodology
for the assessment of speciﬁc methanogenic activity." _W_ater Research 24(7): 813-
825.

Karakashev, D., D. J. Batstone, et a1. (2005). "Inﬂuence of environmental conditions on
methanogenic compositions in anaerobic biogas reactors." Applied And
Environmental Microbiology 71(1): 331-338.

 

Kaspar, H. F. and K. Wuhrmann (1978). "Kinetic-parameters and relative turnovers of
some important catabolic reactions in digesting sludge." Applied And
Environmental Microbiology 36(1): 1-7.

 

Kim, J. S., C. H. Lee, et al. (2001). "Effect of pump shear on the performance of a
crossﬂow membrane bioreactor." Water Research 35(9): 2137-2144.

Kitamura, Y., T. Maekawa, et a1. (1996). "Treatment of strong, nitrogenous wastewater
by an anaerobic contact process incorporating ultraﬁltration." Applied
ﬁgineering in Agriculture 12(6): 709-714.

Klomparens, K., S. L. Flegler, et al., Eds. (1986). Procedures for transmission arm
agnningplectron Microscopy for biological and medical science. Burlington, VT,
Ladd Research Industries.

Knight, B. I. (2003). The Manure Management Challenge. Anaerobic Digester

Technology Applications in Animal Agricutlure - A National Summit", Raleigh,
NC.

Lettinga, G., A. F. M. Vanvelsen, et al. (1980). "Use of the upﬂow sludge blanket (usb)
reactor concept for biological wastewater-treatment, especially for anaerobic
treatment." Biotechnology And Bioengineering 22(4): 699-734.

188

 

Li, A., and J .J . Corrado (1985). Scale-up of the membrane anaerobic reactor system. 40th

Annual Purdue Industrial Waste Conference, Purdue University, West Lafayette,
IN, USA.

Lo, K. V., N. R. Bulley, et al. (1983). "The effect of solids-separation pretreatment on
biogas production from dairy manure." Agricultural Wastes 8(3): 155-165.

Lo, K. V. and P. H. Liao (1985). "High-rate anaerobic-digestion of screened dairy
manure." Journal Of zigricultural Engeering Research 32(4): 349-358.

Lo, K. V., P. H. Liao, et al. (1984). "Mesophilic anaerobic-digestion of screened and
unscreened dairy manure." Agricultural Wastes 11(4): 269-283.

Madaeni, S. S. (1997). "The effect of operating conditions on critical ﬂux in membrane

ﬁltration of latexes." Process Safety And Environmental Protection 75(B4): 266-
269.

Mahmoud, N., G. Zeeman, et al. (2004). "Anaerobic stabilisation and conversion of

biopolyrners in primary sludge - effect of temperature and sludge retention time."
Water Research 38(4): 983-991.

Martin, J. H., Roos, KR (2007). Comparison of the performance of a conventional and a

modiﬁed plug-ﬂow digester for scraped daiﬂ manure. International Symposium
on Air Quality and Waste Management for Agriculture, Broomﬁeld, CO.

Masse, D. I. and R. L. Droste (2000). "Comprehensive model of anaerobic digestion of

swine manure slurry in a sequencing batch reactor. " Water Research 34(12):
3087-3 106.

Masse, L., D. I. Masse, et a1. (2005). "Size distribution and composition of particles in

raw and anaerobically digested swine manure." Transactions Of The Asae 48(5):
1943-1949.

McCarty, P. L. and D. P. Smith (1986). "Anaerobic waste-water treatment."
Environmental Science & Technology 20(12): 1200-1206.

Mclnemey, M. J ., Bryant, MP. and Pfennig, N. (1979). "Anaerobic bacterium that

degrades fatty acids in syntrophic association with methanogens." Arch.
Microbiol. 122: 129-135.

McMahon, K. D., P. G. Stroot, et al. (2001). "Anaerobic codigestion of municipal solid
waste and biosolids under various mixing conditions - II: Microbial population
dynamics." Water Research 35(7): 1817-1827.

189

 

Miron, Y., G. Zeeman, et al. (2000). "The role of sludge retention time in the hydrolysis
and acidiﬁcation of lipids, carbohydrates and proteins during digestion of primary
sludge in CSTR systems." Water Research 34(5): 1705-1713.

Mladenovska, Z. and B. K. Ahring (2000). "Growth kinetics of thermophilic
Methanosarcina spp. isolated from full-scale biogas plants treating animal
manures." Fems Microbiology Ecology 31(3): 225-229.

Moller, H. B., S. G. Sommer, et al. (2004). "Methane productivity of manure, straw and
solid fractions of manure." Biomass & Bioenergy 26(5): 485-495.

Moser, M. A., R. P. Mattocks, et al. (1998). Beneﬁts, costs and operating experience at
seven agricultural anaerobic digesters. Bioenergy '98, Expanding Bioenergy
Partnerships. Madison, WI.

Myint, M., N. Nirmalakhandan, et al. (2007). "Anaerobic fermentation of cattle manure:
Modeling of hydrolysis and acidogenesis." _W_ater Research 41(2): 323-332.

Nagano, A., E. Arikawa, et a1. (1992). "The Treatment Of Liquor Waste-Water
Containing High-Strength Suspended-Solids By Membrane Bioreactor System."
Water Science And Technology 26(3-4): 887-895.

Noike, T., G. Endo, et al. (1985). "Characteristics of carbohydrate degradation and the

rate-limiting step in anaerobic-digestion." Biotechnology And Bioengineering
27(10): 1482-1489.

Obrien, J. E. and R. J. Donlan (1977). "Direct method for differentiating bicarbonate and
acetate in digester control." Abstracts Of Papers Of The American Chemical
Society 173(MAR20): 121-121.

Oliver, B., B. F. Pain, et al. (1986). "Mesophilic anaerobic-digestion of dairy-cow slurry
on a farm scale - economic-considerations." Journal Of Agg'culttaal Engineering
Research 34(3): 229-243.

Oremland, R. S. (1988). "Biogeochemistry of methanogenic bacteria." Biology of
Anaerobic Microorganisms. J. B. A. Zehnder. new York, John Wiley & Sons:
641-705.

Owen, W. F., D. C. Stuckey, et al. (1979). "Bioassay for monitoring biochemical
methane potential and anaerobic toxicity." Water Research 13(6): 485-492.

Padmasiri, S. 1., J. Z. Zhang, et al. (2007). "Methanogenic population dynamics and

performance of an anaerobic membrane bioreactor (AnMBR) treating swine
manure under high shear conditions." Water Research 41(1): 134-144.

190

 

Pain, B. F., V. R. Phillips, et al. (1988). "Mesophilic anaerobic-digestion of dairy-cow
slurry on a farm scale - energy considerations." Journal Of Agricultural
Engineering Research 39(2): 123-135.

Pain, B. F., R. West, et al. (1984). "Mesophilic anaerobic-digestion of dairy-cow slurry
on a farm scale - lst comparisons between digestion before and aﬁer solids
separation." Journal Of Agricultural Engineering Research 29(3): 249-256.

Pierkiel, A. and J. Lanting (2005). "Membrane-coupled anaerobic digestion of municipal
sewage sludge." Water Science And Technology 52(1-2): 253-258.

Powers, W. J ., A. C. Wilkie, et al. (1997). "Effects of hydraulic retention time on
performance and efﬂuent odor of conventional and ﬁxed-ﬁlm anaerobic digesters
fed dairy manure wastewaters." Transaction_s Of The Asae 40(5): 1449-1455.

Qureshi, A., K. V. Lo, et al. (2006). "Dairy manure treatment, digestion and nutrient
recovery as a phosphate fertilizer." J Environ Sci Health B 41(7): 1221-35.

Rittmann, B. E. and P. L. McCarty (2001). Environmental Biotechnology: Principles and
Applications. New York, New York, McGraw Hill.

Ross, B., and H. Strohwald (1994). "Membranes add edge to old technology." Water
Qaulity Intemationjal 4: 18-20.

Ross, W. R., J. P. Barnard, et al. (1990). "Application of ultraﬁltration membranes for

solids - liquid separation in anaerobic-digestion systems - The ADUF process."
Water Sa 16(2): 85-91.

Ross, W. R., J. P. Barnard, et al. (1992). "Practical application of the ADUD process to
the full-scale treatment of a maize-processing efﬂuent." Water Science And
Technology 25(10): 27-39.

Samsoon, P. ,R. E. Loewenthal, et al. (1987). "Hypothesis for pelletization 1n the upﬂow
anaerobic sludge bed reactor. " Water _Sa 13(2): 69- 80.

Saw, C. B., G.K. Anderson, A. James and MS. Le (1985). A membrane technique for
biomass retention in a_naerobic waste treatment process. 40th Annual Purdue
Industrial Waste Conference, Purdue University, West Lafayette, IN, USA.

Schroepfer, G. J., W. J. F ullen, et a1. (1955). "The anaerobic contact process as applied to
packinghouse wastes." Sewage and Industrial Wastes 27(4): 460-486.

Shimizu, Y., M. Rokudai, et al. (1990). "Effect of membrane resistance on ﬁltration

characteristics for methanogenic waste." Kagaku KogakiRonbunshu 16(1): 145-
1 5 1 .

191

 

 

Smith, P. H. and R. A. Mah (1966). "Kinetics of acetate metabolism during sludge
digestion." Applied Microbiology 14(3): 368-&.

Soto, M., R. Mendez, et a1. (1993). "Methanogenic and non-methanogenic activity tests -
Theoretical basis and experimental set-up." Water Research 27(8): 1361-1376.

Speece, R. E. (1996). Anaerobic biotechnology for industrial wastewater. Nashville, TN,
Archae Press.

Strohwald, N. K. H. and W. R. Ross (1992). "Application of the ADUF process to

brewery efﬂuent on a laboratory scale." Water Science And Technology 25(10):
95-105.

Stroot, P. G., K. D. McMahon, et al. (2001). "Anaerobic codigestion of municipal solid
waste and biosolids under various mixing conditions - 1. Di gester performance."
Water Research 35(7): 1804-1816.

Summers, R., P. N. Hobson, et a1. (1987). "Stirred-tank, mesophilic, anaerobic-digestion
of fattening-cattle wastes and of whole and separated dairy-cattle wastes."
Biological Wastes 20(1): 43-62.

Sutton, P. M., P. Berube, et al. (2004). Membrane bioreactors for anaerobic treatment of
wastewaters. W. E. Federation. British Columbia, The University of British
Columbia.

Valcke, D. and W. Verstraete (1983). "A practical method to estimate the acetoclastic
methanogenic biomass in anaerobic sludges." J oumal Water Pollution Control
Federation 55(9): 1191-1195.

Vandenbe.L, C. P. Lentz, et a1. (1974). "Assessment of methanogenic activity in
anaerobic digestion - Apparatus and method." Biotechnology And Bioengineering
16(11): 1459-1469.

Vavilin, V. A., S. V. Rytov, et al. (1996). "A description of hydrolysis kinetics in
anaerobic degradation of particulate organic matter." Bioresource Technology
56(2-3): 229-237.

Veeken, A., S. Kalyuzhnyi, et al. (2000). "Effect of pH and VFA on hydrolysis of organic
solid waste." Journal Of Environmental Engineering-Asce 126(12): 1076-1081.

Vesilind, P. A. and J. J. Peirce (1983). Environmental Pollution and Control. Stoneham,
MA, Butterworth Publishers.

Vogel, D. S. (2003). Therrnophilic anaerobic digestion with ultraﬁltration for dairy
manure. M.S. thesis, Marquette University, Milwaukee, USA

192

von Munch, E., J. Keller, et a1. (1999). "Mathematical modelling of preferrnenters - 1.
Model development and veriﬁcation." _W_ater Research 33(12): 2757-2768.

Wong, K., I. Xagoraraki, et al. (2009). "Removal of viruses and indicators by anerobic

membrane bioreactor treating animal waste." Journal of Environmental Qaulity
IN PRESS.

Wright, P., Inglis, S., Jianguo, M., Gooch, C., Aldrich, B., Meister, A., Scott, N. (2004).
Comparison of Five Anaerobic Digestion Systems on DairyLF arms. ASAE/CSAE
Annual International Meeting, Ottawa, Ontario, Canada.

Yao, H. and R. Conrad (2001). "Thermodynamics of propionate degradation in anoxic
paddy soil from different rice-growing regions." Soil Biology & Biochemistry
33(3): 359-364.

Young, J. C. and P. L. McCarty (1969). "Anaerobic ﬁlter for waste treatment." Journal
Water Pollution Control Federation 41(5P2): R160-&.

Yu, H. Q. and H. H. P. Fang (2001). "Acidiﬁcation of mid- and high-strength dairy
wastewaters." Water Research 35(15): 3697-3705.

Zhang, B., P. J. He, et al. (2007). "Extracellular enzyme activities during regulated
hydrolysis of high-solid organic wastes." Water Research 41(19): 4468-4478.

Zhang, J ., S. I. Padmasiri, et al. (2007). "Inﬂuence of cleaning frequency and membrane
history on fouling in an anaerobic membrane bioreactor." Designation 207(1-3):
153-166.

Zhang, R. H., J. Tao, et al. (2000). "Evaluation of two-stage anaerobic sequencing batch
reactor systems for animal wastewater treatment." Transactions Of The Asae
43(6): 1795-1801.

Zhang, R. H., Y. Yin, et al. (1997). "Anaerobic treatment of swine waste by the anaerobic
sequencing batch reactor. " Transactions Of The Asae 40(3): 761-767.

Zitomer, D. H., T. C. Bachman, et al. (2005). "Thermophilic anaerobic digester with

ultraﬁlter for solids stabilization." Water Science And Technology 52(1-2): 525-
530.

193

 

   

"lllllllllllllllllHillﬂlllllllﬂlﬂllllll“