Fluorescence detection of inhibition in anaerobic digestion


Material Information

Fluorescence detection of inhibition in anaerobic digestion
Physical Description:
Owens, John Martin, 1959-
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 30335683
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
    Table of Contents
        Page iv
        Page v
    List of Tables
        Page vi
    List of Figures
        Page vii
        Page viii
        Page ix
        Page x
        Page xi
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
    Chapter 2. Review of literature
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
    Chapter 3. Materials and methods
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
    Chapter 4. Results and discussion
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
    Chapter 5. Summary and conclusions
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
    Reference list
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
    Biographical sketch
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
Full Text








It is with great pleasure that I express my appreciation

to all who contributed to this work, and I regret that not

everybody who, in some way, gave support can be named or even


I would like to thank Dr. David Chynoweth for acting as

my committee chairman while at the same time sharing his

genuine thoughts, feelings, and experiences as a true friend.

I wish to thank Dr. Spyros Svoronos for his untiring and

excited interest in the subject leading to many helpful

insights. Additionally, I wish to thank Dr. Roger Nordstedt,

Dr. Jim Jones, and Dr. Ben Koopman for serving on my committee

and showing the stamina to scrutinize this extensive document,

as well as proposing appropriate advice.

I would also like to thank my colleague, Pratap

Pullammanappallil, for enduring long nights of methods

development, programming, data analysis, and general


I also wish to thank Dr. Gerry Isaacs, our former

department chairman, for shaping the Agricultural Engineering

Department into the world class institution which it is today,

and for convincing me to accept his offer of admission. And

finally, I wish to thank Dr. Otto Loewer and wish him well in

guiding the department to higher heights of achievement.

This project was partially funded by a grant from the

State of Florida High Technology and Industry Council with

matching industrial funding from BioChem Technology, Inc. and

with additional support provided by the Institute for Food and

Agricultural Sciences, University of Florida. The author

wishes to express his gratitude to Dr. William Armiger,

President of BioChem Technologies, for providing both expert

advice and use of fluorescence instruments.


ACKNOWLEDGEMENTS . . . . . . . . .. ii

LIST OF TABLES . . . . . . . . . .. vi

LIST OF FIGURES . . . . . . . . . .. vii

Abstract . . . . . . . . . . . . x

CHAPTER 1 INTRODUCTION . . . . . . . . 1
Objectives . . . . . . . . . . 1
Background . . . . . . . . . . 2
Statement of Problem . . . . . . . 4

Anaerobic Digestion . . . . . . . 6
Microbiology of Anaerobic Digestion . . .. 11
Inhibition and Toxicity . . . . . .. 18
Phenol . . . . . . . . .. 20
Sulfite . . . . . . . . .. 22
Fluorescence . . . . . . . . .. 26
Previous Studies in Anaerobic Digestion ... . 31

Fluorescence Probes . . . . . . .. 34
Flow Cells . . . . . . . . . .. 35
Standardization Set-up . . . . . . .. .36
Digester Set-up . . . . . . . .. 37
Digester Operations . . . . . . .. 40
Analytical Procedures . . . . . . .. 42
Statistical Methods . . . . . . .. 43

Standardization Experiments . . . . .. 44
Pure Fluorophores . ................. 44
Media Components, Inhibitors, and Other
Soluble Reagents . . . . . .. .51
Particulates . . . . . . . .. 55
Cell Culture, Supernatant, and Pellet . . 62
Temperature Effects . . . . . . .. 64
Model Predictions . . . . . . . .. 71
Experimental Runs . . . . . . . .. 82
Overloading Experiments . . . . .. 82

Underloading Experiments . . . . .. 87
Phenol Experiments . . . . . .. 93
Sulfite Experiments . . . . . .. .101
Final Discussion . . . . . . . .. 111

Summary . . . . . . . . . .. 115
Conclusions . . . . . . . . .. 117
Suggested Research . . . . . . . .. .120

REFERENCE LIST . . . . . . . . . .. 121

BIOGRAPHICAL SKETCH . . . . . . . . .. .129


Table 3-1. Standard glucose media used in digester
under normal operation . . . . . .... 41

Table 3-2. Mineral solution, S4, as modified from Owen
et al., (1979) . . . . . . . . .. 41

Table 4-1. Linear regression coefficients of NADH and
oxidized F,,20 probe signals to concentrations of
pure coenzymes in phosphate buffer at ambient
temperature (23C). . . . . . . . 46

Table 4-2. Fluorescence signal change due to 1 g L-1
addition of some media components . . . .... 52

Table 4-3. Fluorescence signal responses due to
inhibitors and other soluble reagents. . ... 54

Table 4-4. Fluorescence signal responses to
particulate materials . . . . . . .... 59

Table 4-5. Linear regression estimates of fluorescence
response to fractions of digester culture ..... 62

Table 4-6. Variable steady-states and model parameter
estimates . . . . . . . . . .. 74

Table 4-7. Non-linear response curve parameter
estimates for simulation and experimental runs
subject to a doubling of feed strength ..... 85

Table 4-8. Non-linear response curve parameter
estimates for simulation and experimental runs
subject to a halving of feed strength ....... 92

Table 4-9. Non-linear response curve parameter
estimates for simulation and experimental runs
subject to phenol addition . . . . ... 100

Table 4-10. Non-linear response curve parameter
estimates for simulation and experimental runs
subject to sulfite additions . . . ... 110


Figure 2-1. Molecular structure of coenzyme NAD/NADH. 15

Figure 2-2. Molecular structure of coenzyme F420 .... 17

Figure 2-3. Role of NADH and F42, in anaerobic
digestion . . . . . . . . . .. 18

Figure 2-4. Schematic of mechanism for fluorescence. 27

Figure 3-1. Conceptual drawing of fluorescence probe
developed by BioChem Technology, Inc . . ... 35

Figure 3-2. Schematic of flow cell fabricated for use
with fluorescence probes . . . . . .... 36

Figure 3-3. Apparatus for testing the signal response
of fluorescence probes to various compounds. . 37

Figure 3-4. Schematic of CSTR digester and data
acquisition system . . . . . . .... 39

Figure 4-1. Response of NADH and F420 probes to pure
fluorophores in phosphate buffer . . . ... 46

Figure 4-2. Response of probes to incremental
increases in pure NADH in phosphate buffer
followed by addition of sulfite. . . . ... 49

Figure 4-3. Response of NADH and F42, probe signals to
addition of 10.14 pM of NADH on top of 1.23 pM F42,0
followed by addition of sulfite. . . . ... 50

Figure 4-4. Fluorescence response to step-wise
addition of cellulose to phosphate buffer
containing MADH, followed by addition of sulfite
and finally, PAC . . . . . . . .... 56

Figure 4-5. Fluorometer signal responses to
particulate additions of cellulose followed by
PAC . . . . . . . . . . . .. . 58

Figure 4-6. Transmittance spectra for excitation and
emission filters for MADH and F42, probes. ... . 60

Figure 4-7. Fluorescence response of addition of
glucose culture fractions to phosphate buffer. . 63

Figure 4-8. NADH fluorescence versus flow cell
temperature . . . . . . . . ... 66

Figure 4-9. Temperature corrected NADH signal compared
to flow-cell temperature fluctuations ....... 67

Figure 4-10. Correction of fluorescence signals for
room temperature variations and NADH correction
using F42o . . . . . . . . . .. 69

Figure 4-11. Analytical approximation for 24 h and
numerical solution for 100 h following a step-
increase in feed strength from 40 to 80 g COD L-1. 78

Figure 4-12. Analytical approximation for 24 h and
numerical solution for 100 h following a step-
decrease in feed strength from 40 to 20 g COD L-1. 79

Figure 4-13. Analytical approximation for 24 h and
numerical solution for 100 h following the
addition of inhibitor (phenol) to feed at 40 g
L- . . . . . . . . . . . . 80

Figure 4-14. Response of on-line and off-line
measurements to step-increase in feed
concentration in Experiment 2 . . . . .... 83

Figure 4-15. Response of on-line and off-line
measurements to a step-increase in feed
concentration in Experiment 4. . . . ... 84

Figure 4-16. Response of on-line and off-line
measurements to a step-decrease in feed
concentration in Experiment 1 . . . . .... 89

Figure 4-17. Response of on-line and off-line
measurements to a step-decrease in feed
concentration in Experiment 9 . . . . .... 90

Figure 4-18. Response of on-line and off-line
measurements to a step-decrease in feed
concentration in Experiment 10 . . . ... 91

Figure 4-19. Response of on-line and off-line
measurements to addition of 40 g L1 of phenol in
the feed in Experiment 3 . . . . . .... 97


Figure 4-20. Response of on-line and off-line
measurements to addition of 40 g L' of phenol in
the feed in Experiment 7 . . . . . .... 98

Figure 4-21. Response of on-line and off-line
measurements to addition of 40 g L' of phenol in
the feed in Experiment 8 . . . . . .... 99

Figure 4-22. Response of on-line and off-line
measurements to addition of sulfite in the feed in
Experiment 5 . . . . . . . ... 107

Figure 4-23. Response of on-line and off-line
measurements to addition of sulfite in the feed in
Experiment 6 . . . . . . . ... 108

Figure 4-24. Response of on-line and off-line
measurements to addition of sulfite in the feed in
Experiment 11 . . . . . . . . ... 109

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy




May, 1993

Chairman: David P. Chynoweth
Cochairman: Spyros A. Svoronos
Major Department: Agricultural Engineering

Anaerobic digestion is a process by which a complex

mixture of symbiotic microorganisms transforms organic

materials into biogas, nutrients, and cell matter. This

microbial process is frequently limited by imbalances caused

by toxic feed components. Fluorescence monitoring provides

a non-invasive means to measure reduced (NADH) or oxidized

(F420) intracellular coenzyme concentrations, which are

altered by process imbalances. The objective of this

research was to compare responses of on-line fluorescence

probes with on-line methane production rates for detecting

inhibition by phenol and sulfite. Two on-line fluorescence

monitoring probes (BioChem Technologies, Inc., King of

Prussia, Pennsylvania) were studied. The performance of the

probes on test solutions indicated that the F420 probe was

not useful since it suffered from interference due to

reflection from particulates. While the NADH probe also

responded to particulate reflection as well as auto-

fluorescence of other compounds, the detection of NADH over

background signals was possible. However, dark suspended

materials impeded the usefulness of either probe. The

probes were instrumented to a 6-L continuously stirred

computer controlled anaerobic digester fed a synthetic

glucose medium. Data of methane production rate,

fluorescence, and conventional measurements were obtained

from the system subject to imposed perturbations including

step-changes in feed concentration and additions of phenol

and sulfite. Results indicated that responses of the NADH

probe to step-changes in feed concentration were not

reliable. Results of phenol inhibition experiments

indicated that responses of the NADH probe were significant

and occurred prior to significant changes in methane

production rates. In the experiments inhibited by sulfite,

the NADH probe responses were significant but did not

precede responses of methane production rates. The results

of the inhibition experiments are important and indicate

that the detection of intracellular changes caused by

inhibition is possible using the NADH probe.



The objective of this research was to compare the

responses of on-line fluorescence probes with that of on-line

methane production rate for detecting inhibition by phenol and

sulfite. This objective was selected in order to assess the

utility of using fluorescence monitoring systems in process


The specific objectives of the study were to:

1. Verify the response of the NADH and F420 fluorescence

probes to pure fluorophores.

2. Examine the role of media components, inhibitors, and

particulates in interfering with fluorophore detection by the

fluorescence probes.

3. Develop a simple dynamic model to predict the behavior

of the NADH fluorescence signal to disturbances applied in the

experimental runs.

4. Compare the response of on-line fluorescence

monitoring, on-line methane production rate, and model

predictions to process disturbances including step-changes in

feed concentration and addition of phenol and sulfite.


The increase in fuel prices and concern for impending

scarcity which highlighted U.S. energy policy in the 1960s and

1970s contributed to expanded research of viable alternative

energy sources. In addition, rigorous environmental

legislation was developed to control the deterioration of

land, air, and water sources. Together these actions promoted

the application of anaerobic digestion for its ability to

benefit both of these objectives. Yet, the full potential of

this technology has not yet been realized due to perceived

limitations in process stability perpetuated by poor designer

and operator understanding of the biological principles

inherent to this process. Methods for on-line monitoring of

digester stability for use in process control may overcome

these limitations strengthening the tendency to apply the


A means for on-line monitoring of digester stability is

offered by fluorescence technology. Fluorescence monitoring

of intracellular concentrations of reduced nicotinamide-

adenine dinucleotide (NADH) involves illumination of the

culture with an excitation wavelength of 350 nm and measuring

the resulting emissions at 460 nm. Since all living cells

utilize oxidized nicotinamide-adenine dinucleotide (NAD+) and

NADH redox pairs for shuttling of electrons generated in

substrate catabolism, NADH fluorescence reveals both the

general redox state of the illuminated cultures (specific

fluorescence), as well as the culture density (total

fluorescence). In contrast, coenzyme factor 420 (F420), a

flavin derivative found only in Archaebacteria (the kingdom

which includes the methanogens), strongly absorbs photons at

420 nm and fluoresces at 460 nm. Fluorescence monitoring has

proven useful in controlling pure culture processes including

fermentations and aerobic metabolism (Armiger et al., 1986a).

Its application to mixed-culture anaerobic digestion systems

has recently been studied for digesters fed with soluble

substrates(Peck and Chynoweth, 1990 and 1992; Owens et al.,

1991 and 1992). Fluorescence monitoring provides a non-

invasive means to directly measure reduced (NADH) and oxidized

(F420) coenzyme concentrations, indicating the relative redox

state of the culture. If incorporated into process control

schemes, it could improve the performance of anaerobic

treatment systems.

The anaerobic digestion is a multi-step process involving

a consortium of microorganisms. Frequently, these steps

become uncoupled when triggered by differential responses to

changes in feed concentration and the addition of inhibitors.

This may lead to instability in the fermentation and sometimes

even complete failure of the process. Techniques currently

available to assess the real-time performance of anaerobic

digestion (e.g. biogas production, methane production, pH) are

limited, and by the time a response is detected, fermentation

failure is frequently inevitable and difficult to reverse.

The delays associated with off-line measurements (e.g.

volatile organic acids, alkalinity) are unacceptable. There

is a need for on-line systems which provide (in real-time)

early detection of process disturbances.

Statement of Problem

Under a recent contract with the State of Florida High

Technology and Industry Council entitled "Fluorescence

Monitoring and Control of Methanogenic Fermentations,"

preliminary testing of N1ADH and F420 fluorescence probes

indicated they might aid in the detection of process

inhibitors found in industrial waste treatment streams. This

project was initiated to eventually develop an expert system

process control strategy for anaerobic digestion using on-line

fluorescence monitoring of intracellular NADH along with

conventional digester performance measurements. BioChem

Technology, Inc. graciously donated the use of both NADH and

F420 fluorometric probes, data acquisition hardware, and

computer software for this research.

Preliminary work (Peck and Chynoweth, 1990 and 1992)

using this probe in intermittently-fed anaerobic digestion of

soluble substrates resulted in qualitative responses to

feeding events, various substrates, and several inhibitors.

The results showed promise for early detection of process

imbalance due to overloading and inhibition. Characterization


of the fluorometric probe responses to inhibitor additions in

continuously-fed systems is the first step toward application

of expert system controls using fluorometry.

The hypotheses are as follows:

1) Compounds which inhibit the anaerobic digestion

process uncouple the microbial consortium causing a build-up

of intermediates and influence the redox state of the culture

which can be detected by fluorescence.

2) Fluorescence monitoring can be used to detect

response to inhibition prior to detection by methane

production rate measurements.


Anaerobic Digestion

Anaerobic digestion is a process by which a complex

mixture of symbiotic microorganisms transforms organic

materials into biogas, nutrients, and additional cell matter,

leaving salts and refractory organic matter. In waste

treatment, this process can provide a source of energy while

reducing the pollution potential of the substrate. Unlike

fossil fuels, use of these renewable resources represents a

closed carbon cycle and thus does not contribute to increases

in atmospheric concentration of carbon dioxide. Replacement

of fossil fuels also reduces atmospheric pollutants

responsible for acid rain.

An Italian physicist, Volta, is credited with discovering

the evolution of biogas from organic matter in 1776 (Barker,

1956). By 1897 septic tanks in Montunga, Bombay, produced

biogas which fueled engines (Buswell and Hatfield, 1938).

While use of the process for waste treatment continued, the

discovery of large fossil fuel reserves displaced it for

energy production.

The current resurgence of anaerobic technology includes

a flurry of research into microbiology and reactor design.

While the oil embargo of 1973 contributed to this resurgence,

strict environmental legislation encourages new approaches to

wastewater treatment. The ultimate disposal of aerobic sludge

is no longer ignored in plant design, and the increasing cost

of sludge handling and landfilling generates interest in

minimizing the production of sludge. The ability of the

anaerobic process to remove organic with a minimum of sludge

production and aeration demand, continues to spur the

development of new reactor designs applied to a variety of

wastewater streams.

A drawback of anaerobic wastewater treatment entails the

low growth rate of anaerobic bacteria central to the process,

which command longer process start-up and recovery periods

following an upset. The principal upsets are caused by

organic overloading and the presence of inhibitors in the

waste stream. In either case, the capacity of methanogenic

bacteria to utilize acetic acid and hydrogen causes a build-up

of intermediates, a drop in pH, and a pickling of the process.

Concern over process stability warrant careful consideration,

and in fact it is these characteristics which have restrained

anaerobic processes from becoming the preferred treatment

method for a variety of wastewaters.

Applications of the anaerobic digestion process can

generally be divided into low-solids wastewater pretreatment

systems and slurry or high solids systems. The slurry and

high solids systems operate in a batch, fed-batch, or

intermittently-fed mode, while wastewater pretreatment systems

are normally continuously fed and operated at a higher

dilution rate, providing an opportunity for automated control

of the feed rate.

Recently, anaerobic wastewater pretreatment has enjoyed

extensive acceptance for a variety of industrial wastewaters

associated with food processing, beverages, breweries,

distilleries, and most recently pulp and paper production

(Totzke, 1991). Batch operation of the production sequence is

common in these industries, producing a wastewater of variable

strength and quantity, complicating the operation of a

continuous biological treatment system. Additionally, the

intermittent presence of microbial inhibitors in the

wastewater stream can lead to serious process upsets and even


Currently, the control of feed rate to the anaerobic

digester most often relies on off-line measurements of

volatile organic acids to prevent process upset through manual

intervention (Russel et al., 1985). Several investigators

have advocated control schemes based on biogas production rate

(Podruzny and Van den Berg, 1984), alkalinity (Rozzi et al.,

1985), liquid phase hydrogen (Whitmore and Lloyd, 1986;

Dochain et al., 1991), pH (Denac et al., 1988), and digester

substrate concentration (Renard et al., 1988).

Recently, the use of on-line methane production rate has

been suggested as a means to control reactor feed rate to

maintain a more uniform organic loading rate when feed

concentrations are variable (Pullamnmanappallil et al., 1991).

However, such a scheme could interpret the presence of an

inhibitor which reduced the methane production rate as a

decrease in waste strength and respond by increasing feed rate

leading to process failure. Through an expert system

algorithm, Pullammanappallil (1993) avoided this tendency, but

his controller is conservative, cutting off the feed upon

detection of a threshold drop in methane rate and then slowly

increasing the feed rate while testing for a drop in yield

which identifies the presence of an inhibitor. This approach

results in some unnecessary down-time whenever there is a

significant drop in feed strength and this could be avoided if

inhibition could be ruled out sooner.

In the simplest reactor design, a continuously-fed

continuously-stirred digester (CSTR), the hydraulic retention

time (HRT) is dictated by the microbial growth rate, since

mixing causes bacterial washout equal to the net growth. The

extremely slow growth of methanogens requires a rather high

HRT of 10 to 20 days to avoid washout in a CSTR. In spite of

this disadvantage, uniform reactor concentrations and the

applicability of simple chemostat models give the CSTR an

important role in research and process development.


A number of digester designs have been developed for

various waste types and have been reviewed previously

(Chynoweth, 1987). McCarty (1982) demonstrated that waste

treatment in a reactor in which effluent solids are recycled

was dependent on sludge age or solids retention time (SRT)

rather than HRT as in a CSTR. This discovery allowed him to

develop the "anaerobic filter" which, like a trickling filter,

uses a stationary support material to trap bacteria and retain

them in the reactor. Diffusional limitations, channeling and

clogging which plague this reactor led Switzenbaum and Jewel

(1980) to develop the "expanded bed" process which used

lighter inert support material that could be expanded in the

fluid flow up through the reactor.

Lettinga (1978) developed the upflow anaerobic sludge

blanket (UASB) reactor as a modification to the anaerobic

filter. Recently, Van den Berg et al. (1985) developed a

hybrid anaerobic filter/sludge bed reactor which he indicates

resists toxic overloads. These novel reactors allow an HRT as

low as 1 hr, and thus were the first high-rate anaerobic

treatment for dilute wastewaters. With such short liquid

detention times, these high rate digesters are subject to

rapid increases in inhibitor concentrations when transient

pulses of inhibitors enter the feed stream.

Microbiology of Anaerobic Digestion

In the 1920's, Buswell and Hatfield (1938) demonstrated

the treatability of a range of wastes and emphasized the

concept of an acid versus a methane stage, showing the

importance of volatile organic acids as intermediates in the

process. They also demonstrated the applicability of a

stoichiometric equation which balanced carbon, hydrogen and

oxygen. Later, Buswell and Sollo (1948) used "4C tracers to

show that acetate was indeed cleaved to form methane and

carbon dioxide.

Of great importance was the discovery of Bryant et al.

(1967) that the conversion of ethanol to methane was

accomplished with a mixed culture, by isolating the elusive

"S-organism" from Methanobacterium omelianskii. The discovery

of other co-cultures quickly followed, and the number of

species isolated in pure culture increased.

With the identity of closely coupled syntrophic

co-cultures of methanogens with other species, the hypothesis

of an acid phase followed by a methanogenic phase developed

into a more descriptive scheme based on four trophic groups

(Zeikus, 1982).

First the fermentative, or hydrolytic bacteria, hydrolyze

complex organic polymers and ferment them to acids, alcohols,

hydrogen and carbon dioxide. The hydrogen-producing

acetogenic bacteria ferment the larger acids and alcohols to

a combination of acetic acid, one-carbon compounds, hydrogen


and carbon dioxide. The homoacetogenic bacteria synthesize

acetic acid utilizing hydrogen and carbon dioxide, unicarbon

compounds, or by hydrolyzing multi-carbon compounds.

The methanogenic Archaebacteria uniquely catabolize

acetic acid and unicarbon compounds to methane. The substrate

for methanogenesis divides them into two groups; the

acetoclastic methanogens which cleave acetic acid and the

hydrogen utilizing methanogens which utilize hydrogen and

unicarbon compounds. This distinction is not always useful

since several species fall in both groups.

The acidogenic or fermentative group of organisms derive

their energy from substrate level phosphorylation via the

Embden-Meyerhof-Parnus (glycolysis) pathway. The hydrolyzed

compounds are transported into the cells and fermented to

pyruvate, producing ATP. Reducing potential in the form of

NAD+ is regained by converting pyruvate to a number of

substances dependent on the metabolic pathways available to

that species (Mclnerney and Bryant, 1981). Hydrogen, acetic,

propionic and butyric acids are the major products of these

primary fermentations.

The standard free energy change associated with

converting propionate and butyrate to acetate and hydrogen is

+76.1 and +48.1 kJ mole-' when compounds are present in molar

concentrations. For the catabolism of propionate to occur,

hydrogen partial pressures must be kept below 9.1 x 10-3 kPa

(Mclnerny et al., 1980). Thus, acetogenic organisms can only


grow in the presence of hydrogen utilizing organisms such as

methanogens or sulfate reducing bacteria.

Zeikus (1980) estimated the cell density of acetogens in

typical sludge digesters at 10' per mL, one of the lowest

populations for trophic groups in digesters. The mechanisms

for metabolism of these organisms is unknown. The importance

of these organisms is exemplified by the "soured" digester in

which a build-up of hydrogen prevents these bacteria from

converting the higher acids to acetate and hydrogen, resulting

in a build-up of acids that inhibit hydrogen-utilizing

bacteria which promotes more hydrogen accumulation.

The homoacetogenic bacteria are also important for

maintaining low concentrations of hydrogen as required by the

acetogens since they synthesize acetate from hydrogen and

carbon dioxide. The free energy change for this conversion is

extremely favorable, and thermodynamic efficiency is reported

to be 57% by Lynd et al. (1982).

Species associated with this group include Clostridiumn

thermoaceticum and Butyribacterium methylotrophicum (Zeikus,

1980) both of which are able to produce heat resistant spores.

Some of these organisms may also function as members of other

trophic groups. Cell densities in sludge digesters are

reported to be 105-10' cells per mL. (Zeikus, 1980). Although

the use of carbon isotopes shows only a small portion of the

methane is derived through this path, the function of these


bacteria, keeping the partial pressure of hydrogen low, may be


The methanogens are obligate anaerobes which can pick up

electrons from dead end fermentations, through inter-species

hydrogen transfer, and shuttle these electrons through a

unique form of respiration which results in the reduction of

carbon dioxide to methane. The organisms which use hydrogen

to reduce CO2 are commonly regarded as the earliest life forms

due to their chemoautotrophic abilities (Brock et al., 1984).

Astonishingly, this diverse group of organisms is known

to metabolize a very limited number of substrates including

acetate, format, methanol, acetone, methylamines, carbon

monoxide, and H2/CO2. All morphological forms are represented

among the methanogens including rods, cocci, spirals,

sarcinae, and filamentous organisms (Zeikus, 1982).

Common to these procaryotes is a unique cell wall

structure which lacks peptidoglycan, as well as an electron

transport system lacking cytochromes, flavins and quinones

(Brock et al., 1984). Still it is believed that they perform

a unique form of anaerobic respiration which includes an

electron transport phosphorylation process. A cofactor F,420,

involved in this system has been identified as well as

coenzymes F430, which is a nickel-containing tetrapyrrole, and

F342, which is a pterin named methanopterin.

Coenzymes central to the redox reactions associated with

cellular energy production play a central role in process

Nicotinamide-adenine dinucleotide
+ +- NADH
NAD +2e'+ H -

^ NH 2 I\NH

HO O O _-"
o 0
0 0 N NH
HO P=0 H H ':I

0 0

Figure 2-1. Molecular structure of coenzyme NAD+/NADH.

stability. The pyridine nucleotide coenzymes nicotinamide-

adenine dinucleotide (NAD, Figure 2-1) and nicotinamide-

adenine dinucleotide phosphate (NADP) are common electron

carriers in all organisms. In there reduced forms, NADH and

NADPH, these coenzymes fluoresce when exposed to light of 340

nm wavelength. In bacteria, NADPH is principally involved in

supplying reducing potential in biosynthetic cellular

processes and is found at concentrations 10-fold less than NAD

(London and Knight, 1966). The principal pyridine nucleotide,

NAD, is involved in fermentative and respiratory oxidation

reduction reactions and is found in roughly equivalent

concentrations of 5 pM/g dry weight in anaerobic organisms.

The dehydrogenase enzymes which catalyze electron transfers to

NAD cannot substitute NADP, thus cells maintain NAD

principally as NAD+ and NADP principally as NADPH to suit

their respective roles in metabolism (Smith et al., 1983).

Factor 420, (F,420), is 7,8-didemethyl-8-hydroxy-5-

deazaflavin derivative coenzyme (Figure 2-2), which acts as an

electron carrier in the reduction of carbon to methane in

methanogenic bacteria (Peck and Archer, 1989). In its

oxidized form, F420 is highly fluorescent when exposed to light

of 420 nm wavelength (Eirich et al., 1978). With the

development of methods for F420 extraction and quantification

(Van Beelen et al., 1983), the compound was considered for use

in enumeration of viable methanogen populations in mixed

culture systems and for estimating potential methanogenic

activity (Binot et al., 1981). The quantity of F420 found in

hydrogen-utilizing methanogens is 1 pM/g dry weight, but in

acetoclastic methanogens its level is considerably less at 0.3

pM/g dry weight (Dolfing and Mulder, 1985). In mixed cultures

fed sugar wastes, F420 was found at 0.15 pM/g sludge (Dolfing

and Mulder, 1985).

The use of F420 for estimating acetoclastic methanogenic

activity has been criticized due to a wide variation in F420

content found in sludges grown on various substrates (Dolfing

and Mulder, 1985). Peck (1989) found that the F,420 content in

pure cultures of methanogens varied over their batch growth

Coenzyme F420

F4+2e'+2H + 420- H 2

Coo* 0' OH OH O OH
C-1NH -C-CH -0-P-O-CH -0CH -04- CH04


N 0
CH 2

H r R

.0 -coo -

Figure 2-2. Molecular structure of coenzyme F420.

cycle with considerable amounts present in the supernatant.

Pause and Switzenbaum (1984) found the variations of F,420

content of sludges decreased with increasing solids retention

time of daily fed digesters, with a different effect depending

on substrate.

With the central role played by NAD and F420 coenzymes in

the energy yielding metabolism of organisms involved in

anaerobic digestion and with their redox-dependent fluorescent

nature, there is a potential for monitoring process changes

prior to detectable changes in extracellular concentrations of

intermediates (which require off-line analysis) or in changes

Sugars, Proteins, and Fats
mi )

Acetic Acid, Hydrogen, and 00
4 2

: Methanogenesis

Methane and 002

F4^*20 -H2

Fluorescing form of coenzyme

Figure 2-3. Role of NADH and F420 in anaerobic digestion.

in methane production rate. Figure 2-3 shows the turnover of
the two coenzymes with respect to their role in anaerobic
digestion. Disturbances such as inhibitors which affect the
relative pool sizes of these redox pairs could be detected in
real-time using a suitable on-line fluorometer which is
sensitive and specific to these intracellular fluorophores.

Inhibition and Toxicity

Concern over toxicity in anaerobic wastewater treatment
is not due to a greater sensitivity of the microbes to




inhibition compared to aerobic organisms, but instead to the

greater consequence of a process upset since anaerobic

processes require longer start-up periods. Henze and

Harremoes (1982) proposed that anaerobic systems are thought

to be more sensitive to toxic substances because of the

combination of small design safety factors and lack of proper

process control.

Many of the studies relating to inhibition in anaerobic

digestion have been performed in small batch assays similar to

the anaerobic toxicity assay (ATA) described by Owen et al.

(1979). Over similar time periods, experiments measuring the

effect of inhibitor concentration on metabolism of different

substrates can yield information including lag-time,

acclimation-period, methane production rate, trophic groups

affected and inhibition as a percent of control. Often the

concentration exhibiting 50% of control activity is estimated

from these data and reported as a 50% inhibition


The effects of inhibitors on continuous or semi-

continuous digesters have received less study, due to the

rather tedious nature of operating such reactors and the

length of time required to achieve steady-state operation.

Speece (1985) compared inhibition in CSTR's and anaerobic

filter digesters using several potential inhibitors and found

the filter reactors proved to be more resistant to upset by

toxin addition. These studies monitored the transient gas


production and effluent COD responses to impulse injections of

inhibitors directly into the digesters.

On-line monitoring of digesters for detection of

inhibitors is rather novel. Pullammanappallil et al. (1991)

developed an expert system that uses on-line measurements of

methane production rate (Harmon et al., 1990) to prevent onset

of digester imbalance when its regular operation is upset by

changes in substrate feed concentration or the presence of

inhibiting factors in the feed. Crucial to this system is

detection of an initial transient change in methane production

rate. Peck and Chynoweth (1992) reported promising results

using an NADH fluorescence probe to detect the addition of 2-

bromoethane sulfonic acid (BES), a methanogen specific

inhibitor, in a semi-continuous daily fed digester.

A number of substances have been found to inhibit the

digestion process. Speece (1985) assayed 52 petrochemicals

with unacclimated acetate enriched methanogen cultures and

found chloride substitution, aldehydes, double bonds, and

benzene rings were common to various inhibitors. Two

inhibitors of industrial significance include phenol and



Phenol is a monomeric constituent of many naturally

occurring polymers including tannins, lignins, and humus. In

the processing of foods (eg. olives, coffee, and vegetables)


and forest products, these compounds may become solubilized

and hydrolyzed in the wastewater stream to form phenols and

substituted phenolic products which are known to act as

inhibitors to the anaerobic digestion process (Field, 1989).

Phenol is also a constituent of proteins containing the amino

acid tyrosine. In addition, phenol has been used as a

disinfectant in animal production facilities where it may

contaminate the waste stream. Thus, phenol is useful as a

model compound to study inhibition of many related compounds.

The inhibition caused by phenol is attributed to the

apolarity of the compound which enables it to be partially

solubilized in bacterial membranes inhibiting membrane

function. The apolarity of phenol also contributes to its

interaction with hydrophobic regions of enzymes, enabling

inhibition of enzyme activity. Phenol has also been shown to

be biodegradable in anaerobic conditions, such that

acclimation of a particular inoculum can be expected to

increase concentrations resulting in inhibition.

One of the earlier studies of phenol inhibition of

methane bacteria was reported by Speece and Parkin (1983).

Using unacclimated inoculum they found that 26 mM (2444 mg/L)

of phenol resulted in 50% inhibition of batch anaerobic

toxicity assays (ATA) compared to positive controls.

In a study of phenol inhibition of acetoclastic

methanogens, Wang et al. (1991) estimated 50% activity

reduction occurred at a concentration of 1250 mg/L in batch


assays. Sierra-Alvarez (1990) found 50% activity occurred at

1100 mg/L using batch assays on granular sludge from

distillery wastes. Field (1989) estimated 50% inhibitory

concentrations of phenols at 1500 mg/L using volatile organic

acid (VOA) mixtures as substrates in batch assays.

While phenols have been shown to be biodegradable in

anaerobic digestion, the rather long lag times found in batch

studies indicate that degradation should not be significant in

short term studies of transient inhibition in continuous

digesters. The available data indicate that aiming at 2000

mg/L phenol should produce a significant inhibition response.

Thus, media concentrations which provide this concentration in

the digester in 24 h will be used in this study.


Sulfite (S032-) is commonly used in industry for

processing in pulp and paper production, food processing, and

in boiler water treatment, from which it may enter their

wastewater stream. The sources for sulfite include sulfur

dioxide, sulfite salts, bisulfite salts, and meta-bisulfites

all of which are readily soluble and reach an equilibrium

based on the equation 2-1 (Green, 1976).

H20 + SO 'Il H2SO3 HSO; + H+ V SO?-+ 2H+ (2-1)

Cooking elemental sulfur yields sulfur dioxide which, if

passed through an aqueous medium, dissolves and combines with


water to form sulfurous acid which then dissociates, according

to pH, to bisulfite and sulfite ions. When exposed to air,

sulfite reacts with oxygen to form sulfate.

Much of the studies on biological effects of sulfites are

associated with its use in food preservation, where it was

noted for simultaneously acting as a microbial inhibitor,

preserving vitamins such as ascorbic acid, and inhibiting

enzymatic browning reactions. Rehm (1964) noted that in spite

of extensive use in foods, the mechanism of its antimicrobial

action was not well understood. Rhem demonstrated that some

of the antimicrobial action of sulphurous acid was due to its

action on respiration and fermentative metabolism. He found

that sulfite formed addition products with NAD, destroyed

thiamine and split cystine into cysteine.

In sulfite semi-chemical pulping, a process common in the

pulp and paper industry, elemental sulfur or sulfur containing

minerals (iron pyrites) are burned in air yielding sulfur

dioxide gas which is absorbed in water containing base

(Gorden, 1970). This cooking liquor is introduced into a

batch of wood chips which is heated under pressure for a 5 to

12 hour period. The pulp may then undergo a peroxide

bleaching process prior to entering the paper production

process. The waste sulphite cooking liquors are either

blended with waste bleaching liquors or evaporated to produce

sulphite evaporator condensate requiring further treatment.

In both cases, a wastewater stream amenable to anaerobic


treatment is produced but may intermittently contain high

concentrations of sulfite.

When waste sulfite liquors are blended with spent

peroxide bleaching liquors, excess peroxide rapidly reacts

with any excess sulfite producing sulfate. Through increasing

effluent recirculation in response to a rise in oxidation-

reduction potential (ORP), transient increases sulfate

concentrations were tolerated in digesters treating these

wastes (Andre de Vegt, Paques, Inc., personal communication).

However, when excess sulfite was not oxidized to sulfate and

entered the anaerobic digester, system upsets were noted even

with effluent recirculation.

When spent sulfite liquors are evaporated, sulfite is

carried over into the condensate as sulfur dioxide where it

again dissolves. Full-scale anaerobic treatment of evaporator

condensate from a sulfite pulp mill posed problems which were

partially attributed to sulfite contained in the wastewater

stream (Saslawsky et al., 1988). They found concentrations of

250-600 mg LI total SO2 in the condensate.

Compounds containing sulfur play a complex role in

anaerobic digestion since it is an essential nutrient, it can

act as an electron acceptor for sulfate reducing bacteria

competing for carbon sources, it can be a source of

inhibition, as hydrogen sulfide it is partially removed in the

gas phase, and it can precipitate with heavy metals. Speece

(1987) reported that at 9 mM all inorganic sulfur compounds


except sulfate inhibited methane production with sulfite less

inhibitory than sulfide. In a review of methanogenic toxicity

of forest industry wastewaters, Sierra-Alvarez (1990) reported

the 50% inhibitory concentrations for sulfite and total

hydrogen sulfide of 125 and 530 mg/L (1.6 and 15.6 mM),


In batch assays and an anaerobic filter, Eis et al.

(1983) found continuous feeding of sulfite was not inhibitory

until 1000 mg/L (12.5 mM). Cohen (1992), using inoculum from

digesters treating wool-scouring effluents, found that 2500

mg/L (24.03 mM) of sodium meta-bisulfite caused 44% inhibition

of gas production in short-term (27 hr) assays.

Sarner (1990) discussed the methanogenic toxicity of high

sulfate/sulfite wastewaters associated with pulp and paper and

molasses-based fermentation industries, but attributed the

inhibition to the resulting sulfide production. Sarner (1990)

was so concerned about the toxicity of sulfide formed from

anaerobic treatment of spent sulfite pulping liquor, he

designed a pretreatment process centered on an anaerobic

trickling filter which promoted growth of sulfate reducing

bacteria, and utilized gas scrubbing with gas recirculation to

remove the sulfide from the wastewater stream.

Based on the variation of reported inhibitory

concentrations of sulfite found in the literature there seems

to be a certain level of disagreement on the extent of sulfite

inhibition. This discrepancy is likely due to the unstable


nature of sulfite and the inaccuracy of analytical methods

used in its measurement. lodometric titration techniques

commonly applied in sulfite determination were designed for

analysis of boiler water and are subject to significant

interference when high concentrations of soluble organic

contaminate the sample (Clesceri, et al., 1989).


Fluorescence, a type of luminescence, is a photochemical

phenomenon in which a compound absorbs high energy photons,

propelling certain electrons to a higher energy state, from

which they emit a photon, normally of lower energy, upon

returning to ground state (Guilbault, 1973). Absorption and

emission spectra are characteristic of a particular molecule.

This mechanism of fluorescence, shown in Figure 2-4, was first

described by G. G. Stokes in 1852. With the advent of various

instrumentation, fluorescence techniques were developed for

analysis of many compounds in clinical pathology, inorganic

analysis, agricultural chemistry and public health.

The first studies on the use of fluorescence for the

detection of intracellular coenzymes was reported by Chance

(Chance and Thorell, 1959) after he noted the similarity of

fluorescence spectra of living cells to that of the reduced

coenzymes, NADH and NADPH. With the potential for on-line

monitoring and control of bioreactors, instrumentation was

developed for commercial on-line sensors based on culture


Electron Energy Level

Excited state -

Intermediate state -

UV Aborpon


Figure 2-4.

Infrared easem

ViAble fluorescence


Ground state I V
Schematic of mechanism for fluorescence.

fluorescence (Ingold and BioChem Technologies, Inc., Humphrey
et al., 1989). In 1986, U.S. Patent 4,577,110 was issued for
an "optical apparatus and method for measuring the
characteristics of materials by their fluorescence" (MacBride
et al., 1986).
Much of the research using fluorescence sensors was
concerned with correlating signal levels with viable biomass
concentrations and metabolic state in pure culture reactor
systems (BioChem Technology, Inc., 1987).


The basic equation relating fluorescence to concentration

of pure fluorophore in a non-absorbing solvent is given by 2-2

(Guilbault, 1989).

F = 0 (1-e-OL (2-2)

where: F is the fluorescence intensity in all directions:
(watts cm-2)
is the fluorescent yield or ratio of emitted to
absorbed energy: (unitless)
I0 is the incident radiant power: (watts cm-2)
a is the molar absorptivity: (M-' cm-1)
L is the path length: (cm)
C is the molar concentration of the fluorophore: (M)

A disadvantage of fluorescence is its strong dependence

on environmental factors such as temperature, pH, ionic

strength, and viscosity, as well as the influence of other

absorbing compounds in complex solutions. Quenching reduces

fluorescence emissions by a competing deactivation process

(Guilbault, 1989). In temperature quenching, increases in

temperature increase molecular collisions which compete with

the deactivation process and lower fluorescence emissions.

Decreased fluorescence emissions are also caused by inner

and outer filter effects as well as turbidity caused by the

presence of particulates. Inner filter effects are caused by

compounds which absorb excitation light attenuating the energy

reaching the fluorophore. Outer filter effects are caused by

compounds which absorb emitted light attenuating the energy

reaching the sensor. Particulates may lower the fluorescence


signal by a combination of reflection and absorbance of both

excitation and emission light.

The detection of a specific fluorophore in a complex

media is complicated by auto-fluorescence of other compounds.

Many aromatic compounds including amino acids and other

cofactors may also fluoresce when exposed to light. Such

interference causes a background signal on top of which the

fluorophore must be detected.

The monochromators in the fluorometers used in this study

employ a set of band pass filters to refine the spectrum of

light passing through the filter to wavelengths near the

desired excitation and emission wavelengths. The spectrum of

transmitted light through these band pass filters is described

by a Lorentzian line shape. The slit-width of the spectral

transmittance shape dictates how much of the excitation light

might be detected as emission light by the overlap of

excitation and emission filter shapes. In addition, the

transmittance through these filters can overlap with the

emission fluorescence of competing fluorophores.

In the linear range of the instrument, fluorescence of

intracellular NADH in a complex fermentation can most simply

be described as the sum of background fluorescence with a

linear function of the biomass concentration and activity

(specific growth rate) product as shown in Equation 2-3.

F = F, + KpX

where: F



is the fluorescence signal: arbitrary
fluorescence units (AFU)
is background fluorescent signal: (AFU)
is the specific growth rate of bacteria: (d'1)
is the viable biomass concentration: (g VSS L-1)
is a linear coefficient: (AFU g VSS-1 L d)

Assuming a noncompetitive inhibition model for microbial

growth, y, the functional relationship of growth to substrate

and inhibitor concentration is given by equation 2-4.


S( /S+1) (/ +)max
(K.IS+l) (I/K1+l1)

where: P



is the specific growth rate of bacteria: (d-')
is the maximum specific growth rate: (d-')
is the substrate saturation coefficient:
(g COD L-1)
is the substrate concentration: (g COD L-1)
is the inhibitor concentration: (g L"1)
is the inhibition coefficient: (g L-')

Together with the simple mass balance relationships of

substrate, biomass, and inhibitor in a simple chemostat,

Equations 2-3 and 2-4 can be used to predict the dynamics of

fluorescence of such a system to step changes in feed

substrate or inhibitor concentrations.


Previous Studies in Anaerobic Digestion

Studies using on-line fluorescence technology in

anaerobic digestion have been limited to intermittently-fed

systems and culture dilutions (Peck and Chynoweth, 1990,1992,

Samson et al., 1988). In unpublished preliminary research,

BioChem Technology, Inc., found that the commercial NADH probe

designed to be specific for NADH responded to oxidized F420

which might affect its application to methanogenic

fermentations. A prototype probe for F420 was developed with

the intent to account for the contribution of F420 to the NADH

fluorescence signal.

Samson et al. (1988) confirmed the cross response of the

NADH probe to pure F420 and concluded that the new prototype

F420 fluorometer was highly specific. Applying the F420 probe

to non-diluted anaerobic sludge proved inconclusive due to the

dark nature of the sludge. With a 1/32 dilution of the

anaerobic sludge they were able to detect additions of pure

F420 using the F420 fluorometer. Their study did not address

using fluorometers to monitor changes in digesters subject to


The most significant results using the NADH and F420

fluorometers, are found in a series of studies by Peck and

Chynoweth (1990, 1992). The cross-response of both

fluorometers to pure additions of both fluorophores were


documented. Using intermittently-fed glucose digesters, they

verified that signal pulses of the NADH probe after discrete

feeding events were due to the glucose addition and not other

feed constituents.

Peck and Chynoweth (1992) also obtained signal responses

to discrete feeding events using several process intermediates

including acetate, propionate, and format. They also step-

increased daily discrete feedings until process failure and

produced results which appeared to show that failure was

indicated by a drop in NADH probe response prior to other

measured process variables including VOA's, gas production

rate, gas quality, pH, and ORP. They also subjected their

system to inhibition by BES, which disrupted the NADH probe

signals response to the discrete feeding events following the

addition of the inhibitor.

While the studies by Peck and Chynoweth yielded a wealth

of information, the dynamic nature of daily fed glucose

reactors makes quantification of the signal response

difficult. Using a truly continuous CSTR subject to step

changes in feed substrate and inhibitor concentrations will

yield responses which may be fit to simple first order models

with dead time to quantify the rate and magnitude of

fluorescence response.

In other previous work, Pullammanappallil et al. (1991)

developed an expert system that uses on-line measurements of

methane production rate (Harmon et al., 1990) to prevent onset


of digester imbalance when its regular operation is upset by

changes in substrate feed concentration or the presence of

inhibiting factors in the feed. This control system is

conservative (i.e., it sacrifices performance) because

interpretation of changes in methane production rate do not

enable one to determine whether the change is caused by a

decrease in feed substrate concentration or by inhibitors in

the feed. On-line measurement of changes in intracellular

coenzyme concentrations to process disturbances may prove more

rapid than on-line methane production rate measurements and

allow a faster method for detection of inhibition. If the

NADH and F420 probes can be used to differentiate between the

presence of an inhibitor and a drop in the substrate

concentration, they have the potential to provide more robust

control of the process.


Fluorescence Probes

The two on-line fluorescence probes, used in these

studies were developed by BioChem Technology, Inc., King of

Prussia, Pennsylvania. Both probes are configured as

reflectance spectrofluorometers where excitation and emission

light paths are opposed at 180. The basic configuration for

both probes is illustrated in Figure 3-1.

The commercially available NADH probe (Model #400-352)

contains a low pressure mercury lamp UV light source which

passes through a bandpass filter to give a mean excitation

wavelength of 351 nm (+/- 40 rnm). The emitted light passes

through an additional bandpass filter (460 +/- 25 nm), and

illuminates photodiodes which are tuned to detect

intracellular NADH emission at a wavelength of 460 nm. The

F420 probe is similarly constructed but has a bandpass

excitation filter centered at 406 nm (+/- 34 nm) with emission

set for receiving 465 nm (+/- 20 nm). The probe housing

contains circuitry which compensates for source lamp drift and

transmits a 4 to 20 mA signal to an interface box connected to

an external module analog converter which delivers a 0-2 V



Figure 3-1. Conceptual drawing of fluorescence probe
developed by BioChem Technology, Inc.

signal to a MetraByte model DASCON-1 A/D data acquisition card

installed in an IBM XT compatible computer. The external

model allowed up to 6 additional analog inputs to the data

acquisition system.

Flow Cells

A flow cell, mounted on each probe, was required for

monitoring the recirculated reactor medium and eliminating

interference from external light sources. The flow cells were

machined and fabricated from PVC. Figure 3-2 shows a

schematic of the completed flow cell design. The flow cell

Figure 3-2. Schematic of flow cell fabricated for use with
fluorescence probes.

allowed the probe to be easily adapted to different digesters

operating under varying conditions.

Standardization Set-up

Standardizing the response of the fluorescence probes to

additions of pure NADH (Eastman Kodak Company), F420, media

components, inhibitors, and particulates verified the

instrument's response to the fluorophores and interfering

components. The NADH and F420 probes were inserted into flow

cells which were attached to a stirred flask and a peristaltic

pump as depicted in Figure 3-3.

Figure 3-3. Apparatus for testing the signal response of
fluorescence probes to various compounds.

The probe and flow cell arrangement was tested for its

response to NADH additions and changes in flow rate. Standard

additions of NADH, equivalent to a step change in

concentration of 2 MM, were employed to analyze the signal

response and standardize the value of a normalized

fluorescence unit (NFU) as calibrated by BioChem Technology

using thioflavin S.

Digester Set-up

A continuously-fed continuously-stirred (CSTR) digester

was instrumented for automated data acquisition of methane

production rate, temperature, and NADH and F420 fluorescence.


During this investigation the CSTR was, at times, also

instrumented with on-line pH, oxidation-reduction potential

(Eh), flow cell temperature, and ambient temperature

(nominally at 23 C). The digester was constructed from 15.2

cm ID iron pipe coated with epoxy. A stirring rod with three

paddles was fixed through a sealed bearing in the top of the

vessel and was operated continuously at 150 rpm. The digester

was fed by a computer controlled peristaltic pump and

temperature controlled by a PID temperature controller (Omega

CN200, +/-0.1 C) with set-point control provided by the

computer. A schematic of the set-up is detailed in Figure


The CSTR set-up employed an IBM compatible AT computer

instrumented with a digital I/O interface board for monitoring

a float switch on a U-tube gas meter. Two serial ports on the

computer were used for the peristaltic feed pump control and

temperature set-point supervisory control. The peristaltic

feed pump was a Cole Parmer Masterflex (Model 7550-90)

computerized drive pump equipped with a size 14 pump head

(Masterflex Model 7014-20) and size 14 Norprene tubing (6042-

14). The feed pump was operated at a flow rate of 2.5 mL min'

with a cycle time of 0.25 min in order to simulate continuous

feeding. Custom software written in QuickBasic was used to

acquire the methane production data and to control temperature

set point and feed pump cycling.

Figure 3-4. Schematic of CSTR digester and data acquisition

A second computer (IBM XT) with two four-channel A/D data

acquisition boards, provided by BioChem Technology, was

connected to a pH meter (Fisher Acumet Model 810), an Eh probe

(Omega PHCN-36 pH/ORP controller), thermocouples (Omega OMNI

I, copper-constantine), and NADH and F420,, probes through an

interface box. The IBM XT executed FERMAC software developed

by BioChem Technologies for fluorescence monitoring of


bioprocesses. A second peristaltic pump (Cole Parmer 7553-30)

fitted with two size 18 pump heads (Model 7018-20) and size 18

Norprene tubing circulated the digester contents through the

fluorescence flow cells and through a flow cell containing the

pH and Eh probes.

Digester Operations

The 6-L digester was operated at a 20 day retention time

(dilution rate = 0.05 d'1) at a nominal loading rate of 2 g COD
L-'1 d-'. Biogas leaving the reactor passed through a soda lime

absorber for CO2 removal prior to displacing the liquid in the

gas meter. Glucose feed medium was sterilized and kept

refrigerated during digester operation. Digester effluent was

collected from an overflow tube.

The normal glucose medium was prepared using the

ingredients outlined in Table 3-1. The trace nutrient mineral

solution used is outlined in Table 3-2. The feed medium was

prepared by dissolving all ingredients less the mineral

solution and diammonium phosphate in a 2-L polypropylene flask

while sparging with nitrogen gas. The feed flask was then

capped with a stopper fashioned with 3 glass tubes, one of

which reaches to the bottom of the flask. After autoclaving

at 20 psi for 10 min, a tedlar gas bag was attached to one of

the inlet tubes while the solution cooled. After cooling the

mineral solution and diammonium phosphate solution were added

via sterile syringe.

Table 3-1. Standard glucose media
under normal operation.

used in digester

Reagent Amount
Glucose (dextrose) 28.8 g
Yeast extract (Difco) 3.2 g
Casamino acids (Difco) 3.2 g
NaHCO3 3.0 g
NaCH3CH2COOH 0.7 g
Mineral solution, S4 15.0 mL
(NH4)2HP04 solution (26.7 g L-1) 5.0 mL
Distilled water to 1.0 L

Table 3-2. Mineral solution, S4, as modified from Owen
et al., (1979).

Compound Concentration (g L'1)
CaCl2 2H20 16.7
NH4C1 26.6
MgCL2* 6H20 120
KC1 86.7
MnCl2*4H20 1.33
CoCIl- 6H,0 2.0
HB03 0.38
CuCl2 2H20 0.18
Na2MoO4,2H20 0.17
ZnCl2 0.14
NiCl2 6H20 0.15
H2WO4 0.007

Analytical Procedures

Volatile organic acids were measured on an FID gas

chromatograph (Shimadzu GC-9AM). The samples were prepared by

centrifugation followed by sample acidification using 20%

phosphoric acid spiked with an internal standard of iso-

butyric acid. The samples were injected onto a 2 m long by 2

mm id glass column packed with 80/100 chromosorb 1200 WAW

coated with 3% H3PO4. A 1 iL volume was injected at an inlet

temperature of 180 C with column temperature ramped from 130

C to 170 C over 5 min and a detector temperature of 200 C

and N2 carrier gas.

Phenol was measured after centrifugation using absorbance

at 270 nm on a spectrophotometer (Shimadzu UV-160). NADH and

F420 in phosphate buffer were quantified by absorbance at 340

nm and 420 nm respectively using extinction coefficients from

the literature.

Total solids (TS) and volatile solids (VS) were

determined in accordance with Standard Methods (Clesceri et

al., 1989). Total suspended solids (TSS) and volatile

suspended solids were determined by the difference between

whole and centrifuged samples. Biomass concentrations were

determined by optical density (Milton Roy, Spectronic 21D) at

600 nm after calibration using samples of known TSS and VSS



Statistical Methods

Data from standardization and digester experiments were

imported into a computer spreadsheet. All linear regressions

associated with temperature corrections and standardization

studies were analyzed using linear regression functions

contained in QuatroPro 3.0.

The transient responses of methane production rate and

NADH fluorescence measurements from experiments subjected to

step changes in feed concentration and inhibitor content were

fitted to a general first order response with dead time using

non-linear parameter estimation included in SigmaPlot 4.1

(Jandel Scientific, 1990) which uses the Marquardt-Levenberg



Standardization Experiments

Prior to installing the probes in the flow-cells in the

CSTR set-up, it was critical to assure that the instruments

performed satisfactorily and that their behavior was not

impeded by the flow-cell design. In addition, the

investigation of the potential for media components,

inhibitors, and particulates to effect probe measurements was

essential. To achieve these objectives, a series of

experiments were performed with the probes in their flow-cells

in the standardization set-up outlined in Materials and


Pure Fluorophores

Using the standardization set-up described in Chapter 3,

incremental additions of pure fluorophores in phosphate buffer

were circulated through the flow-cells. Standard additions of

NADH were approximately equivalent to a final step change in

concentration of 2 yM. Additions of oxidized F420 in 0.2 pM

increments, were also employed. The results were used to

analyze the signal responses, verify that both the NADH and


F420 probes responded normally in the flow cells, and

standardize the value of mV (for the F420 probe) and normalized

fluorescence unit (NFU) as calibrated by BioChem Technology

using the standard fluorophore, thioflavin S for the NADH


Table 4-1 lists the linear regression coefficients of

both probes in response to additions of pure coenzymes in

phosphate buffer and Figure 4-1 shows the data and regression

lines. While each 1 pM increase in NADH concentration

produced an increase in the NADH probe signal of 7.29 NFU,

there was no significant response of the F420 probe to NADH

additions. In contrast, each 1 pM increment of oxidized F420

produced significant responses in both probes, yielding 267

NFU on the NADH probe and 19.1 mV on the F420 probe.

These results verify those obtained by Peck and Chynoweth

(1992) where 1 pM changes in oxidized F420 concentration

produced 271 NFU and 18.3 mV increases in signals from the

NADH and F4,20 probes, respectively. Such similar responses

indicate that both probes in their current flow cell

arrangements were performing normally. The ancillary response

of the NADH probe to additions of oxidized F420 can be

explained by the absorbance spectra of oxidized F42,0 and the

rather broad slit width of the excitation filter employed in

the NADH fluorometer (351 +/- 40 nm). While the absorbance

optima of oxidized F420 is at 420 nm, the half height peak

width (slit width) is about 30 nm. In addition, as with most

_ 200


0 100



Z 300


* 1oo




r I I- 7I 850
) 2 4 6 8 10 12 14 16




F420 (uM)
Figure 4-1. Response of NADH and F420
fluorophores in phosphate buffer.

probes to pure

Table 4-1. Linear regression coefficients of NADH and
oxidized F420 probe signals to concentrations of pure
o-. 4... .. .l...-U 4-.. k. '. .. ~4- ...kJ...4. -^4-- ^ l^'o<



coenzymues n.J p ospateu u.c UUeJr a), enLilk tempi~erature %j 11
Coenzyme Added
PNADH F420 Units

slope 7.29 (0.14)- 267 (11.0) NFU uM-'
constant 94.4 (2.03) 107 (12.1) NFU

slope 0.02 (0.02) 19.1 (0.89) mY PM-1
constant 861 (0.26) 858 (0.97) mV

ns o:



fluorophores, oxidized F420 has a second absorbance peak with

optima at 300 nm (Eirich et al., 1978) which also causes

fluorescence. Furthermore, the NADH probes excitation filter

has a broad slit width of 80 nm around its optimal

transmittance of 351 nm. This overlap allows the NADH probe's

excitation light to substantially overlap F420's absorbance

spectra, and since both fluorophore emissions occur at a

similar wavelength, the NADH probe detects the fluorescence of

oxidized F420. Apparently, the fluorescent yield of F420 is

significantly greater than NADH, since pM quantities of F420

produces a signal over 36 times that of an equimolar quantity

of NADH.

Dynamic analysis of flow cell, pump, and instrument

response allowed the development of a measurement transfer

function which was used to determine if the response was fast

enough to measure changes at rates expected in the process.

Neither probe showed a signal response due to changes in pump

flow rate. Data from a standardization experiment, with pump

recirculation at 41 mL min-', was converted to deviation

variables and a first order instrument response with dead time

was fitted to the data using nonlinear regression. At this

flow rate, both the short dead time (td 0.01 h) and low time

constant (tri = 0.03 h) ensure that the instrument will

adequately respond to expected digester dynamics which are

dominated by a time constant equal to the dilution rate of

0.05 d-' (Owens et al., 1991).


Due to the pronounced response of NADH fluorescence in

the CSTR digester subjected to the introduction of Na2SO3 in

the feed, addition of Na2SO3 to pure NADH in phosphate buffer

was investigated at the end of the standardization experiment.

Figure 4-2 shows a temporal plot of the response of both

probes to incremental increases in NADH concentration followed

by addition of Na2SO3. The downward spikes of the NADH probe

signal and the upward spikes of the F420 probe signal

correspond with times when the flow cells were filled with air

prior to an addition event. The fact that the F420 probe

signal increased when liquid was removed from the flow cell

indicates that a substantial portion of the signal is

comprised of reflected excitation light which is detected by

this probe. When the phosphate buffer fills the cell, the

increase in light absorption by the liquid lowers the F420

probe signal.

After the step increases in pure NADH, and the

corresponding increases in the NADH probe signal were

observed, 312 pM Na2SO3 was added to the buffer. This yielded

the surprising result of a quenching of the NADH probe signal

over a 1 h period. In a similar standardization experiment

employing pure oxidized F420, addition of Na2SO3 caused no

significant change in either the NADH or F420 probe signal

levels. In Figure 4-3, 10.14 pM of NADH was added into a

phosphate buffer previously containing 1.23 pM F420, followed

by an addition of 312 pM of Na2SO3. Again, the added sulfite

200 1--r

180 -

160 -
5- 140 -

S120 -
z 100
2 I I

I40I >/ I I
940 /
Air Peaks
920 --


860 .* I ----

S20 i i I | 500 .
15 _400 a
10 ___ 200
0 0
I I I I0
------ ^,777 1 1 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Time (h)
Figure 4-2. Response of probes to incremental increases in
pure NADH in phosphate buffer followed by addition of sulfite.

resulted in quenching the signal caused by the added NADH over

a 1-h period; yet the signal due to F420 appeared unchanged.

Apparently, some type of reaction of sulfite ions with

the reduce NADH impairs the fluorescence of the resulting
compound. Rehm (1964) stressed the reactive nature of

,-. 400 -

.. 300

< 200

100 I -l --

1.23 uM F420
-- 920 -

900 n

860 -- 1 1 -

S15 l '- 500 -
400 a
10-: 300
5 200 o"
S0 -100 Q2
i I I 0
1 2 3 4 5 6
Time (h)
Figure 4-3. Response of NADH and F4,20 probe signals to
addition of 10.14 pM of NADH on top of 1.23 yM F420, followed
by addition of sulfite.

sulfites with aldehyde and keto groups to form sulfoxide

groups suggesting that the sulfites could combine with the

keto group located on the nicotinamide moiety of the NADH
receiving hydrogen and electrons during reduction of NAD, and

is the component of the molecule responsible for fluorescence.


These standardization experiments showed that only the

NADH probe responds to pure NADH, while both the F420, and NADH

probes respond to F42, additions as found by Peck and Chynoweth

(1992). A new phenomenon illustrated by these experiments is

the effect of sulfite in quenching fluorescence due to NADH

and the ability of sulfites to disclose the portion of a

compound fluorescent signal due specifically to NADH.

Media Components. Inhibitors, and Other Soluble Reagents

The standardization setup was also employed to test the

response of the probes to various media components,

inhibitors, and soluble reagents employed in this study.

Glucose, casamino acids, and yeast extract were separately

added to phosphate buffer and circulated through the flow

cells to examine their contribution, if any, to the

fluorescence signal and the results are shown in Table 4-2.

The fluorescence signal response due to glucose was not

significant as noted by the magnitude of the standard

deviation. In contrast, both casamino acids and yeast extract

exhibited a significant amount of intrinsic fluorescence based

on measurements by the NADH fluorometer. The F42,0 probe showed

a minimal response to these two media components.

Both tyrosine and tryptophan are known to be fluorescent

(McAvoy et al., 1992) and can be expected in appreciable

quantities in casamino acids. The yeast extract, in addition

to containing proteins comprised of these amino acids, has a


Table 4-2. Fluorescence signal change due to 1 g L'
neidiitA on of some media components.

SValues in parentheses are standard aeviatlons.

complex mixture of vitamins and cofactors which includes

compounds which are comprised of the fluorescence nicotinamide

group. BioChem Technology, Inc. (1984) also subjected the

NADH probe to several media components and found that addition

of 1% yeast extract (MD 1492) exhibited 128 NFU's of

background fluorescence while a 1% addition of casein peptone

(Type B MD 8098) gave an 81.9 NFU increase.

The extent to which these medium components contribute to

the background fluorescence of an active glucose digester is

not known, but in a continuously-fed CSTR with a 20-day HRT,

most of the proteins contained in these amendments will be

degraded and will likely be over-shadowed by extracellular

enzymes which are leaked or excreted by the bacteria culture.

Peck and Chynoweth (1992), using their daily-fed CSTR,

compared the addition of media, with and without glucose, to

assure that observed fluorescence responses were due to

glucose stimulated changes in NADH concentrations rather than

Media Component Probe Signal Response Units
Glucose NADH -0.18 (0.41)" NFU (g L-1)-1
F420 -1.08 (3.50) mV (g L-')-1
Casamino Acids NADH 10.5 (0.33) NFU (g L-) -1
F420 0.33 (0.18) mV (g L-1)-1
Yeast Extract NADH 97.4 (2.65) NFU (g L-)-'
_____F420 0.36 (0.33) mV (g L-')-'


changes in background fluorescence caused by other media

amendments. They observed no fluorescence response when they

added media containing yeast extract and casamino acids but no

glucose, indicating these media components did not contribute

to background fluorescence in active glucose digesters.

To assess whether there is any significant probe response

to the inhibitors themselves, both phenol and sodium sulfite

and a number of other soluble reagents were tested using the

standardization set-up. The results for these experiments are

shown in Table 4-3. None of the signal responses due to these

compounds are significant as seen by the size of the standard

deviation of the estimates. While strongly absorbing,

phenol's absorbance peak with a maximum at 270 nm is distinct

enough and far enough from the excitation wavelengths not to

cause any substantial inner filter effect on fluorescence

signals detected by the probes.

While sodium sulfite reacts with pure NADH, it showed no

significant effect on fluorescence signals in a phosphate

buffer. Other oxidized products of sulfur and nitrogen were

also assayed and they exhibited no significant intrinsic

fluorescence. Potassium sulfate, potassium nitrite, and

sodium nitrate were all tested to see if they shared the

ability to quench NADH fluorescence exhibited by sodium

sulfite and the results showed no significant decrease over

natural decay rates due to the unstable nature of pure NADH.

Table 4-3. Fluorescence signal responses due to inhibitors
and other soluble reagents.


The reactive nature of sulfites to NADH is unique among the

reagents tested.

The test of probe responses to sucrose was instigated by

results, shown in the next section, of significant signal

responses to cellulose. Sucrose, a disaccharide, exhibited no

significant signal responses, indicating the signal increases

due to cellulose were not solely a property of the polymeric

nature of this glucose polymer.

Inhibitors Probe Signal Response Units
Na2SO3 NADH 1.88 (2.22)" NFU mM-'
_______ F420 -0.29 (1.43) mV mM-'I
Phenol NADH -0.21 (0.32) NFU (g L-')"
_____F420 0.02 (0.25) mV (g L-')-1
Other Reagents_____________
K2S04 NADH -1.12 (1.62) NFU miM-1
_______ F420 1.02 (1.56) mV mM-1i
KNO2 NADH -1.79 (2.74) NFU mM-1
________ F420 1.19 (1.16) mV mM-1i
NaN03 NADH 0.16 (1.71) NFU mM-1
________ F420 1.61 (1.68) mV mM-1i
Sucrose NADH -0.40 (0.61) NFU (g L-)-)
______F420 -0.18 (0.68) mV (g L-1)-'

j.1in p.rn .[ e areIr

s: a. n ar. Q.i 0. J .. L.J ,~[1 ,


Tests of particulate substances in the standardization

set-up were investigated to ascertain to what degree

particulates in solution could interfere with fluorescence

detection. Many digesters receive feed containing significant

concentrations of suspended material. In addition, both

viable and dead bacteria are particulate in nature and may

effect fluorescence detection of intercellular fluorophores.

Using the standardization set-up, aliquots of a 10%

cellulose slurry were added step-wise to a solution containing

18.1 pM NADH in phosphate buffer, to examine how the cellulose

particulates interfere with the signal. The results are shown

in Figure 4-4. As in previous experiments, the downward

spikes of the NADH probe signal corresponds with periods when

the flow-cells were pumped dry of the buffer solution.

Contrary to expectations, the step-wise additions of cellulose

caused increases in the signals from both probes with the F420,

probe signal reaching values higher than seen in any previous


Additions of cellulose were followed by adding sodium

sulfite, which, over a 1-h period, quenched the signal due to

NADH. After this, the NADH probe still exhibited a large

signal due to the cellulose particulates in solution. The

white color of cellulose particulate indicates a high

reflectivity over the visible spectrum for these particulate

surfaces. This leads to the interpretation that a portion of






) 20

0 -. 15
4 '4 10
o Mg 5 i- ---
W 0
3 4 5 6
Time (h)
Figure 4-4. Fluorescence response to step-wise
cellulose to phosphate buffer containing NADH,
addition of sulfite and finally, PAC.






addition of
followed by

the excitation light is reflected off of the cellulose

particulates and becomes detected as fluorescence.

At the end of this experiment, 1.6 g L'1 of powdered

activated carbon (PAC) was added to the cellulose solution to

assess the effect of a black particulate matter on the

18.1 uM NADH
/ '^ ~1.6 gL1 PAC

fluorometer signals. Addition of PAC completely quenched the

NADH probe signal caused by the cellulose. The black color of

PAC particles indicate the strongly absorbing nature of PAC

surfaces over the visible spectrum with a corresponding low

reflectivity coefficient. Thus, it may be concluded that PAC

has the potential to mask fluorometer signals due to

particulate reflections, as well as fluorophore fluorescence.

In another experiment, cellulose was added stepwise to a

phosphate buffer followed by stepwise additions of PAC. The

results are shown in Figure 4-5. Again, without any

contribution of known fluorophores in the buffer solution, the

step-wise addition of cellulose caused corresponding step-wise

increases in fluorescence probe signals. Step-wise additions

of PAC resulted in step-wise decreases in both probe


The effect of two other white particulates, talc and

dextran, were also tested and the probes maximum reflective

signal response was determined by covering the probes with

aluminum foil. The response of the probes to these materials,

cellulose and PAC are given in Table 4-4. Of all the white

particulates tested, cellulose exhibited the greatest signal

increase on a g L-' basis. Talc gave the next greatest effect

with dextran only exhibiting a marginal effect on the NADH

fluorometer and no significant effect on the F420, probe signal.

The fine particulate size of the dextran resulted in

suspensions which appeared translucent compared to the milky








V w

0 1 2 3

1.5 '



0.0 u

Time (h)

Figure 4-5. Fluorometer signal responses to particulate
additions of cellulose followed by PAC.

suspensions produced by the other particulate materials, so

reflectivity of the resulting solution can be expected to be


Except for the dextran, the effect of these particulates

on the F420 probe signal are markedly higher than their effects


Table 4-4. Fluorescence signal responses

b no data.


to particulate

on the NADH probe. When both probes are covered by an almost

purely reflective material, aluminum foil, the signals rise to

a maximum of 127 NFU for the NADH probe and 1200 mV for the

F420 probe. The contribution of this reflective signal to NADH

probe measurements of active glucose cultures, though

significant, was not great enough to impede NADH detection

since these signals were often more than 300 NFU (Peck and

Chynoweth, 1992). In contrast, the F420 probe, rarely produced

signals greater 900 mV in active glucose digesters, so the

signal could be completely masked by interference from

reflective particles in solution.

The 180 configuration inherent in the design of these

probes allows them to be configured as reflectometers if

Particulates Probe Signal Response Units
Cellulose NADH 4.83 (0.28) NFU (g L-)-)
F_____420 10.4 (0.40) mV (g L-1)-1
PAC NADH -55.1 (17.8)" NFU (g L-1)-
F420 ______-109 (17.8) mV (g L-')-'
Talc NADH 0.58 (0.04) NFU (g L-)-'
F420 4.83 (0.07) mV (g L'-1)-1
Dextran NADH 0.08 (0.04) NFU (g L-)-1
F420 0.06 (0.27) mV (g L-)-')
Aluminum Foil NADH 127 (nd)b NFU
______ F420 1200 (nd) mV
rta+a in nrantheca r ar atinar1 rd viati nns.

1.0 NADH Probei

0.8 -/ Emission

0.6 .
Excitati ni
o 0.4 -

S0.2 -

Product Spectrum

E. 1.0
F4., Probe
0 42 Emission

S0.86 Excitation



200 300 400 500 600
Wave Length (nm)

Figure 4-6. Transmittance spectra for excitation and
emission filters for NADH and F420 probes.

excitation and emission band-pass filters are removed
(MacBride et al., 1986). But even with the filters in place,
there is a potential for detection of reflected excitation
light. Based on the slit widths of the bandpass filters,
Figure 4-6 shows the transmittance spectra of excitation and
emission filters for both probes, generated from Lorentzian

equations using the cited optima and slit widths. While the
excitation filter spectral peak of the F420 probe is not as


broad as that of the NADH probe, it is not as well separated

from the emission filter spectral peak of the F420 probe

compared to the separation for the NADH probe. The product

spectrum for the two probes shows the extent of overlap of the

excitation and emission filter spectra, and indicates that a

greater proportion of reflected light is detected by the F420


The role of particulates in interfering with fluorescence

measurements using the BioChem Technology fluorometers is

complicated since the reflective nature of some particulates

can add to the signal, while other light-absorbing

particulates may lower the signal by increasing the turbidity

of the solution and effectively lowering the path length of

the fluorescence flow cell. Still, the experiments showed

that the NADH probe can detect pure NADH in spite of suspended

cellulose particles, and the quenching nature of sulfite was

not impeded by the presence of cellulose particulates. The

usefulness of the F420 probe as currently configured is in

doubt since it suffers from both low sensitivity to pure F42,0

in combination with substantial interference by the presence

of particulates. Analysis of F42, fluorometric data from this

probe should be limited in light of these findings.

Cell Culture. Supernatant, and Pellet

Since active anaerobic glucose cultures contain both

particulate matter (viable and dead cells) and extracellular

compounds (eg. enzymes), a portion of the NADH probe signal

should be attributed to these compounds. In addition, it is

conceivable that NADH is present in the supernatant from

leaking or lysed cells. In order to investigate these

contributions to the NADH signal, experiments with culture

fractions were performed.

Active culture from a glucose digester was removed and

added to phosphate buffer in the standardization set-up to

examine the contribution of soluble and suspended components

to the background fluorescence signals. The results are shown

in Figure 4-7, with linear regression coefficients listed in

Table 4-5.

Table 4-5. Linear regression estimates of fluorescence
response to fractions of digester culture.
Culture Fraction Probe Signal Response Units
Whole NADH 121 (3.84)' NFU (gVSS L'p)-l
F420 13.6 (0.84) mV (gVSS L-1')-
Supernatant NADH 106 (2.53) NFU (gVSS L-')-'
F420 5.15 (0.61) mV (gVSS L-2)-l
Pellet NADH 29.7 (0.96) NFU (gVSS L-)-'
_______________ F420o 3.19 (0.50) mV (gVSS L-')-1
Data in parentheses are standard deviations.

150 ,1--,---
140 V Whole v- --
1 4 v Supernatant /
g 130 0 Pellet

1 110





0.0 0.1 0.2 0.3 0.4 0.5 0.6
Culture Addition (g VSS L- 1)

Figure 4-7. Fluorescence response of addition of glucose
culture fractions to phosphate buffer.

Centrifugation and resuspension of the pellet can be

expected to cause some damage to viable bacteria, due to

mechanical shear and the introduction of oxygen. In spite of

efforts to minimize this damage, the data for the NADH probe

indicated that a major portion of the fluorescence signal from

the whole culture was exhibited by the supernatant. Unlike

the treatment of the pellet, the supernatant was not diluted

back to the original sample volume (90 mL instead of 100 mL)

so the estimates for its NADH response are slightly high.

Still, this data suggests that soluble extracellular

components in the culture have a significant impact on

background fluorescence signals.

After addition of each of these culture fractions, 3 mM

of Na2SO3 were added. In all cases there was no significant

change in the NADE fluorescence signal, indicating that the

signals from the different fractions were not due to NADH


These experiments show that extracellular components from

active glucose cultures can contribute to background

fluorescence signals, yet the presence of extracellular NADH

appears not to be significant. While the F420 probe showed a

significant response to culture additions the signal appears

equally split between supernatant and pellet fractions, but in

light of the assessment of this probe's utility, no specific

conclusions can be drawn from the data.

Temperature Effects

In analyzing the data from the first three experimental

runs on the digester system, a diurnal pattern in the

fluorescence signals was noted even during steady state

operation of the digester. Even though the digester was

operated under precise temperature control at 35.0 +/- 0.1 C

poor temperature control in the room housing the system was

thought to be responsible for the daily oscillations in the


measurements; either due to changes in the culture temperature

in the circulation loop to the flow-cell or to an effect on

the electronics.

A second temperature probe was instrumented to the

FluroMeasure System and inserted in the recirculation loop

containing the flow cells. The computer controlling the

digester temperature controller was programmed to cycle the

digester through temperatures between 33 and 37 C, to

determine the correlation between fluorescence signal and

temperature. Figure 4-8 shows a plot of NADH fluorescence

versus flow cell temperature over a 20-h period. As expected

for the fluorescence phenomenon, increasing temperature caused

a decrease in fluorescence, presumably caused by temperature

quenching of the fluorophore activation. The correlation is

linear over the temperature range applied with a slope of

-1.97 NFU C'1 (std = 0.08, r2 = 0.91). Thus, to correct for

this temperature dependant variation, the NADH signal was

corrected to 35 C using Eq. 4-1.

NADH35 = NADHT 1.97-(35-T) (4-1)

where: NADH,,35 is corrected NADH signal
NADH, is uncorrected NADH signal
T is measured flow cell temperature

Figure 4-9 shows the application of Eq. 4-1 to correct

the NADH signal caused by temperature variations of the liquid

as measured in the flow cells. The correction appears to

410 I-
4100 1

405 Slope -1.97 NFU C-1


Z 395

390 I
28 30 32 34

Flow Cell Temperature (C)

Figure 4-8. NADH fluorescence versus flow cell temperature.

remove the majority of signal variation due to flow cell


After several additional experimental runs, a diurnal

variation in corrected NADH signal was still present. Further

testing indicated the variations were due to room temperature

changes rather than flow cell temperature changes. It

appeared that radical room temperature fluctuations were

affecting the instrumentation. The temperature probe was

subsequently located to monitor air temperature changes in the

room to account for this effect.

Regressions of NADH and F420 data with room temperature

during steady-state digester operation yielded linear

regression coefficients of -9.53 NFU C-' (std 0.42) for the

NADH probe, and -1.14 mV C-' (std = 0.05) for the F42,,0 probe.

Equations 4-2 and 4-3 were used to correct fluorescence data

for runs when room temperature data were available.

430 i

I 410 NADH (uncorrected

o 1 400-
S 390

S 3NADH., (corrected

z 370 II I-
72 76 80 84
36 -- 1 ,

34 -

c 30

0 28

26 I-
72 76 80 84
Time (h)

Figure 4-9. Temperature corrected NADH
flow-cell temperature fluctuations.

88 92

signal compared to

NADH- = NADH,, 9.53-(20 T,) (4-2)

Where: NADHc is the temperature corrected signal
NADH, is the uncorrected signal
TIX is the measured room temperature

F420 = F420 -1.14-(20 Tr,) (4-3)

Where: F420-c is the temperature corrected signal
F420-R is the uncorrected signal
T_ is the measured room temperature


to 3S5 C)


88 92

7 V


Since the variations in the F4,0 signal appeared primarily

to be an effect of temperature variations on instrumentation

and since room temperature effects are expected to effect both

fluorescence instruments similarly, experimental runs lacking

corresponded temperature data used F420 variations to correct

for NADH variations due to temperature. A regression of NADH

data on to F420 data during a steady state period resulted in

a linear regression coefficient of 7.11 NFU mV'1 (standard

deviation (std) = 0.30). The equation for this temperature

correction is given in 4-4.

NADH. = NADH + 7.11-(904.42 F420) (4-4)

Where: NADH is the temperature corrected NADH signal
NADH is the uncorrected NADH signal
F,20 is the uncorrected F420 signal

In Figure 4-10 the results of these temperature

corrections are shown for a 40-hour period during which the

digester was operated at steady state. The digital conversion

of temperature and F420 signals yields data with a precision of

+/- 0.5 C and 0.3 mV for the room temperature measurements

and F420 signal, respectively. When these data were used to

correct the NADH signal some additional variation (noise) was

introduced, while the diurnal variations due to temperature

were minimized.

While temperature variations cause significant

fluctuations in fluorescence signals, given accurate

monitoring of temperature, the signals can be corrected when

appropriate correlations are made. The temperature dependence


480 -

460 -

440 -


400 -
500 -





400 -





900 -

22 -

- 20 -

18 -


80 90 100
Time (h)

110 120

Figure 4-10. Correction of fluorescence signals for room
temperature variations and NADH correction using F420.


Corrected by Room T

I I \ I

NADH (uncorrected)

Corrected by F420

F420 (uncorrected)


of fluorescence could be approximated with linear regression

coefficients in the temperature ranges encountered. While

flow cell temperature variations affect the fluorescence

signals, room temperature fluctuations through its effect on

instrumentation was found to be the greater source of the


Model Predictions

In order to develop an appreciation for the expected

behavior of NADH fluorescence responses to perturbations in

feed strength and inhibitor additions in CSTR glucose

digesters a simple dynamic model was developed. Using a

single-substrate, single-population, non-competitive

inhibition model for growth, the mass-balance around a CSTR

yields a system of three differential equations.

Neglecting inhibitor consumption, the equations for

biomass (microbial), substrate, and inhibitor are given in 4-

5, 4-6, and 4-7.

dX = X-DX (4-5)

dS = 'X + (So _-S}D (4-6)
dt YZs

dI = (Jo I) (4-7)

where: X is the microbial concentration: (g VSS L-1)
S is substrate concentration: (g COD L-1)
Y,/, is biomass to substrate yield: (g VSS g COD-1)
I is the inhibitor concentration: (g L-')
S0 is the feed concentration: (g COD L-1)
I0 is the feed inhibitor concentration: (g L'1)
P is the specific growth rate: (d'1)
D is the dilution rate of the digester: (d-')

For a step-change in feed concentration from S0 to S2 at

time 0, the function for So is given in Eq. 4-8. And for a

step change in inhibitor concentration in the feed from 0 to

I, at time 0, Eq. 4-7 is independent and linear so we can find

the analytical solution for the inhibitor concentration, I, in

the digester given in Eq. 4-9.

s = for t<0 (4-8)

0 for t-O (4-9
I= 1'.(j-e-D) for tO (4-9)

Three basic cases were investigated in the experimental

runs and in the modeling efforts. They include:

1) Step-increase in feed concentration to double strength

(overloading) equivalent to S1=40 g COD L', S2=80 g COD L-1,

and I0=0 g L-';

2) Step-decrease in feed concentration to half strength

(underloading) equivalent to S1=40 g COD L', S2=20 g COD L"1,

and I0=0 g L-1;

3) Inhibition by adding inhibitor to feed to achieve

50% inhibitory concentration in digester within 24 h. For

the phenol this is equivalent to S1=S2=40 g COD L-1, and I,=40

g L-.

The auxiliary equations used for specific growth rate,

methane production rate, and NADH fluorescence are given in

Eq. 4-10, 4-11, and 4-12.

S= max
S(KJ/S+1) (I/K+I)


P= YP/x iX (4-11)

F = KFy lX (4-12)

where: P is the specific growth rate
p_ is the maximum specific growth rate
K, is the saturation coefficient
S is the effluent substrate concentration
I is the inhibitor concentration
KI is the inhibition coefficient
P is the methane production rate
Yp/x is the methane to biomass yield
F is the fluorescence signal
K, is a linear coefficient

Steady-state results obtained from the digester operated

at a 20 d HRT (D = 0.05 d'), exhibited a biomass concentration

of 4 g VSS L', an effluent substrate concentration of 0.1 g

L-1 (as VOA), and methane production rate of 0.6 L CH4 L-1 d"1.

Using these steady-state estimates and assuming a maximum

growth rate for the combined culture of 0.1 d', the

differential equations and auxiliary equations can be solved

at steady state to calculate values for K., Yx/s, Yp/x, and K7.

For modeling purposes the steady-state fluorescence was chosen

as 1 arbitrary fluorescence unit (AFU). Assuming that the 50%

activity of the culture occurs at an inhibitor concentration

of 2 g L-1 (as for phenol) than this value is a reasonable

estimate for K.. The steady-state values for the variables

and estimates used for the parameters with the appropriate

units are given in Table 4-6.

Using this model with the given parameter estimates, the

behavior of the state variables and auxiliary variables was

investigated when the system was subjected to step changes in

Table 4-6.
atimata. -

Variable steady-states and model parameter

feed concentrations and inhibitor additions. Dynamic

simulation using numerical integration of the differential

equations can approximate solutions for the system. In

addition, linearizing the model allows analytical estimates of

the behavior of state variables near their steady-states.

For all three cases, Eq. 4-5 and 4-6 were approximated

using the first linear terms of a Taylor's series expansion,

and transformed to the Laplace domain using deviation

variables. After eliminating variables, partial fraction

expansions of the solutions were transformed back to the time

Variable Value Units
X 4.0 g VSS L-'
S 0.1 g COD L-1
So 40.0 g COD L-1
P 0.05 d-1
P 0.6 L CH, L-d-1
F 1.0 AFU
D 0.05 d-1
P_ 0.1 d-1
K, 0.1 g COD L-1
K, 2.0 g L-1
YX/, 0.1 g VSS g COD-'
YP/X 3.0 L CH4 g VSS-'
K, 5.0 AFU (g VSS L-' d-1)-1


domain yielding analytical estimates for the transient state

variable responses near their steady states.

From the approximate analytical solutions for biomass

concentration (X) and reactor substrate concentration (S)

estimates for the behavior of auxiliary variables, methane

production rate (P) and for NADH fluorescence (F) were

calculated numerically for a 24 h period following the

disturbances. Numerical integration using the Euler method

with a time-step of 0.01 d, also was employed to solve the

system for the first 100 h after the disturbance. The results

for the three cases are plotted in Figures 4-11, 4-12, and


For the case when feed strength was doubled, the

approximate analytical solutions for substrate and biomass

concentrations are given in Eq. 4-13 and 4-14.

S = 0.1 -0.2 (l-e-0-05t) +0.2 (l-e-9.5t) (4-13)

X = 4 +4.03(I-e-005t) -0.02 (1-e-995st) (4-14)

The initial transients predicted by these equations are

dominated by two eigenvalues (inverse time constants), one

identical to the dilution rate, D, of 0.05 d-1 and the other

much larger at 9.975 d-1, incorporating the intrinsic dynamics

of the system parameters. Both from these equations and from

the plots in Figure 4-11, it is apparent that the behavior for

the substrate concentration is strongly affected by the faster

constant, while the rate of change of the biomass


concentration is principally dominated by the much slower

dilution rate of the system. The initial transients predicted

for NADH fluorescence, F, and methane production rate, appear

also to be strongly affected by the faster constant of the


For the case where the feed strength was reduced to half

strength, the approximate analytical solutions for substrate

and biomass concentrations are given in Eq. 4-15 and 4-16.

S = 0.1 +0.1(l-e-0-05t) -0.1(l-e-9'975t) (4-15)

X = 4 -2.015 (1-e-0-05') +0.01 (1-e-9975t) (4-16)

Again the dynamics of the initial transient predictions

are comprised of the same two constants, and although the

signs of the coefficients are reversed from the preceding

case, the rate of the drop in X is dominated by the slower

dilution rate while the rate of decrease in effluent

substrate, S, is strongly affected by the faster time

constant. The plots in Figure 4-12, again illustrate the

similarity of the transient response of both NADH

fluorescence, F, and methane production rate with that of the

effluent substrate concentration, S.

For the case in which an inhibitor was added to the feed,

the approximate analytical solutions for substrate and biomass

concentrations are given in Eq. 4-17 and 4-18.

S = 0.1 +4.02 (1-e-0-05t) -0.02 (Il-e-9'975t) (4-17)

X = 4 -4.03 (I-e-0-05t) +0.002(-1 e-9"975) (4-18)

For the inhibition case, the exponential terms again

contain the same two constants. Both Eq. 4-17 and Figure 4-

13, no longer indicate that effluent substrate concentration

dynamics are as strongly affected by the faster constant.

Effluent substrate, S, appears instead to rise more slowly

corresponding with the slower time constant associated with

the dilution rate and, in this case, the dynamics of inhibitor

concentration in the digester. The predicted biomass

concentration dynamics are still dominated by the dilution

rate of the system. Both of these approximate solutions are

only valid at t<
Examining the numerical simulation solutions also shown

in Figures 4-11, 4-12, and 4-13, it is notable that, although

the magnitudes of these transients are quite different than

indicated by the analytical approximations, the initial rates

and quasi-first order behavior of the variables are quite

similar. Based on these results, a first order model for the

response of the on-line measurements of NADH fluorescence and

methane production rate is a reasonable operating assumption,

and allows a model framework for fitting response parameter

estimates to on-line data in order to compare simulation and

experimental results.

4 1-- 5.0

3 4.5 *

2 -4.0

1 3.5

0 3.0
1.4 -i i


I 1.0
0 1 0.5
k 0.
d 0.6
L) 0.4 -


~ 0.8 ~
n 0.6 -
M o
U 0.4 -

0.2 -

0.0 I i
100 I

20 -
0 --- I -- I --- --- --- I -
-20 0 20 40 60 80 100
Time (h)

Figure 4-11. Analytical approximation for 24 h and numerical
solution for 100 h following a step-increase in feed strength
from 40 to 80 g COD L-'.

1.4 -i 4.5
1.0 4.0
S0.8 W,
-3.5 rn
< 0.6 >
S0.4 3.0 b
0.2 -
0.0 I I 2.5
1.0 -- i i -


4. 0.6
'0 1
o 0.4
d. ,.-------

0.20 ----


m 0.10

-. 0.05

50 ------- --- -- ---
v 40 I
0 30
4w,. Z 20

~ 0
0 I- I I I- -

-20 0 20 40 60 80 100
Time (h)
Figure 4-12. Analytical approximation for 24 h and
numerical solution for 100 h following a step-decrease in
feed strength from 40 to 20 g COD L-'.







1 0 ,-- i, --





n n I-I --

40 I I

-20 0 20 40
Time (h)


4.0 7





-04 0 100

60 0 100
60 80 100

Figure 4-13. Analytical approximation for 24 h and numerical
solution for 100 h following the addition of inhibitor
(phenol) to feed at 40 g L-'.






Since the simplified model of the glucose digester system

ignored intermediates which might accumulate in the reactor,

and dead times associated with pumping and measurement, adding

a dead time to the first order model used for estimating these

response parameters will account for delays not considered in

the simulations. Therefore, the final response model used for

analyzing NADH and methane rate data from experimental runs is

given in Eg. 4-19 where the disturbance is implemented at t=0.

Y = {Yss for t4t, (4-19)
=Ys + C(1-e-k('-t )) for tttd

Where: Y,, is the steady state value
C is the coefficient of the response
k is the inverse time constant
td is the dead time of the response

Experimental Runs

Overloading Experiments

Only two experimental runs in which the feed

concentration was doubled were performed in the digester set-

up (Exp. 2 and Exp. 4). The plots for these runs are shown in

Figures 4-14 and 4-15, respectively. Neither the behavior of

the NADH signal nor the methane production rate were

consistent between these two runs. Table 4-7, gives the

results of the nonlinear regression estimates of the

parameters of the process response model shown in Eq. 4-19.

In Experiment 2, the NADH fluorescence response showed

only a 20 NFU increase after 70 h from the step-increase in

feed concentration, while the methane production rose slowly

to its maximum at 60 h from the disturbance. In Experiment 4,

in contrast, the NADH probe signal increased by 200 NFU after

100 h from the feed concentration increase, while the methane

production rate increased abruptly just after the change in

feed. In both runs the VOA and pH profiles appeared similar,

with VOA reaching 400 mg COD L-' after 40 h from the change in


While both experiments were started after at least two

days of steady state operation in which fluorescence, methane

production rate, VOA, and pH appeared to be constant,

fundamental differences in the make-up of the microbial

consortia could explain the differences displayed. If the

500 i-,-,l,-- ,1000

t.450..-----'----^ -"^950 .
Z 450
U 900
S400 rs
z 850

350 -- l -- -- 800

ST' 1.2

o i 1.0

1__________, ,--
5.. 7.- 4 0* -

7.5 ,600

7.0 400-

6.5 200 r

6. 100 i I I I i
o *. 80

PO 60
U 40 --
f ^! 20
-20 0 20 40 60 80 100
Time (h)

Figure 4-14. Response of on-line and off-line measurements
to step-increase in feed concentration in Experiment 2.

700 ,0,-, - l 1000
850 950
600 0
z E
u 550 00
S500 -.
., 850
Z 450
400 --- -- l -- l --- --- -- BOOJ i 0
400I I 800


~ .~ 1.2
o 1.0

V, 0.8


7.5 1--- -- I -- i --- --- --- 600o

:4 7.04;

6.5 2 005

6.0 0
6 100 I I I I i
o 80
f'. .2 20-
0 I i i I i
-20 0 20 40 60 80 100
Time (h)

Figure 4-15. Response of on-line and off-line measurements
to a step-increase in feed concentration in Experiment 4.

Table 4-7. Non-linear response curve parameter estimates
for simulation and experimental runs subject to a doubling
of feed strength.
Data Source Parameter Estimates
Yss C k t'
(h-i') (h)

F 1.00 1.03 0.090 0.00
(AFU) (na) (0.01). (0.007) (0.70)
P 0.60 0.62 0.091 0.00
(L LI d-') (na) (0.00) (0.008) (0.69)
Exp. 2
NADHT 402 111 0.003 5.55
(NFU) (2.03) (65.1) (0.003) (1.06)
CH4 Rt. 0.75 0.77 0.032 2.12
(L L-1 d-') (0.05) (0.01) (0.001) (0.25)
Exp. 4
NADHc 446 237 0.025 6.75
(NFU) (9.34) (0.76) (0.001) (0.56)
CH,4 Rt. 0.66 0.33 0.608 0.00
(L L-1 d-1) (0.03) (0.00) (0.041) (0.08)
SValues in parentheses are standard deviations of the
na not applicable for model steady states.

organisms responsible for the major portion of the NADH signal

were already at their maximum activity before the increase in

feed strength, then increases in fluorescence would only occur

as the population increased, a relatively slow process. This

could explain the minimal increase in fluorescence and slower

rise in methane production rate observed in Exp. 2.


Except for NADH fluorescence, the response of the

measurements in Exp. 4 was quite similar to predictions of the

simulation model shown in Figure 4-11, if the simulation model

effluent substrate concentration (S) is equated to the

experiment's VOA measurements. Fundamental in the model

equations, both methane production rate and NADH fluorescence

are assumed to be linearly dependent on biomass concentration

and growth rate, so their response behavior is identical in

the simulation. However, in Exp. 4 the fluorescence signal

does not exhibit the abrupt increase shown by the rapid change

in methane production rate.

The response curve parameter estimates in Table 4-7,

indicate that the first order constant (k) for the

fluorescence signal in Exp. 2, is not significant indicated by

its large standard deviation. For k = 0, there is no time

dependence and hence no significant signal change. If the

constant is zero then the value for the estimate of the

exponential coefficient (C) is not reliable as well, since the

product of the two terms will be zero when either of these

parameters are zero. This means that there was no significant

NADH signal response in the experiment. In contrast, the NADH

probe response in Exp. 4 was not trivial and both the first

order constant (k) and coefficient (C) are highly significant.

Underloading Experiments

Three experimental runs in which the feed concentration

was halved from 40 g COD L' to 20 g COD L-1 were performed in

the digester set-up (Exp. 1, Exp. 9, and Exp. 10). The plots

for these runs are shown in Figures 4-16, 4-17, and 4-18,

respectively. The behavior of the NADH signal was not

consistent between these three runs. Table 4-8, gives the

results of the nonlinear regression estimates of the

parameters of the process response model shown in Eq. 4-19.

While the initial and final steady state values for

methane production rate varied between the three runs due to

measurement and calibration errors, the behavior of the

methane production rate responses following the underloading

event were quite similar. The numerical simulation prediction

of the methane production rate response to an underloading

exhibited the rather fast drop to the new steady state. The

methane production rate dropped almost immediately with dead

times less than 1 h for all cases. While the numerical

simulation response exhibited a first order constant of 0.99

h', the estimates for the experimental data were all lower,

0.68, 0.33, and 0.73 h-1, for Exp. 1, 9, and 10, respectively.

In contrast, the NADH probe signals exhibited little

change in the experiments. While Exp. 1 and 9 showed the

signal dropping slightly after the step-decrease in feed

concentration, in Exp. 10 the signal actually increased over

the whole experiment. The response curve parameter estimates


in Table 4-8 indicate the response in Exp. 9 was not

significant (k not significantly different than 0). While the

fits for Exp. 1 and 10 are significant, they are in opposite

directions (C=-8.56 and 15.6 NFU, for Exp. 1 and 10,

respectively) and their magnitudes are not significantly

different than fluctuations caused by temperature corrections

(+/- 10 NFU).

In Experiment 1, no VOA data were available, while in

Exp. 9, the volatile acids were dropping, prior to the feed

change indicating the digester may not have been at a true

steady-state prior to start of the experiment. In Experiment

10, the VOA levels were extremely low (<10 mg L-') throughout

the experiment, indicating little change after the

underloading. It is possible that the viable biomass

concentration in Exp. 10 was increasing throughout the

experiment causing a steady increase in NADH probe signal

during this experiment.

Overall, the underloading experiments do not exhibit any

major responses in NADH signal, for the conditions of the

experiments, in contrast to predictions by the simulation

model, and the discrepancy between experiments indicates the

detection of underloading by NADH fluorescence was not


450 i 1000

> -950 -
v 400 900


350 Boo BOO
1.0 ,-i ,

.~' 0.8

f 0.4 -

S 0.2 -

0.0 I -- -- I I
7.5 i i ,i

S7.0 -

6.5 -

6.0 -I i i I
50 I
S 40
o 30
0 o --- -- l -- I -- l -- I -
-20 0 20 40 60 80 100
Time (h)

Figure 4-16. Response of on-line and off-line measurements to
a step-decrease in feed concentration in Experiment 1.