������������
������������������������������������
����������������������
���������������
�����������
�������.���������
��������.���������
��������.�����������
�����������
����������������������
������
�������������
������������������
�
About WMRC's Electronic Publications:
This document was originally published in a traditional format.
It has been transferred to an electronic format to allow faster and broader access to important
information and data.
While the Center makes every effort to maintain a level of quality during the transfer from print
to digital format, it is possible that minor formatting and typographical inconsistencies will still
exist in this document.
Additionally, due to the constraints of the electronic format chosen, page numbering will vary
slightly from the original document.
The original, printed version of this document may still be available.
Please contact WMRC for more information:
WMRC
One E. Hazelwood Drive
Champaign, IL 61820
217-333-8940 (phone)
www.wmrc.uiuc.edu
r~
l~c i;' ?L=
DEPlIIlTMEN r OF
NAT"URAl
RESOURCES
WMRC is a division of the
Illinois Department of Natural
Resources
HWRIC RR031
lli SITU BIORECLAMATION OF CONTAMINATED
GROUNDWATER
by
Bruce E. Rittmann
Albert J. Valocchi
Joseph E. Odencrantz
Wookeun Bae
Environmental Engineering and Science
Department of Civil Engineering
University of Illinois at UrbanaChampaign
Urbana, Illinois 61801
Printed November 1988
Reprinted December 1988
Prepared for
the Illinois Hazardous Waste Research and Information Center
HWRIC Project No. HW88·026
and supported by funding from
the University of Illinois Water Resources Center, Urbana, Illinois
Grant No. S109
Reprinted by Authority of the State of Illinos 88/75.
This report is part of HWRIC's Research Report Series and as such has
been subject to the Center's external scientific peer review. Mention
of trade names or commercial products does not constitute
endorsement or recommendation for use.
CONIENTS
Item
Page No.
LIST OF TABLES
VI
LIST OF FIGURES
vii
LIST OF ABBREVIATIONS
xii
ABSTRACT
xiii
EXECUTIVE SUMMARY
xiv
CHAPTER 1. INTRODUCTION
1
CHAPTER 2. PROJECT OBJECTIVES
7
CHAPTER 3. EXPERIMENTAL METHODS AND RESULTS
9
3.1
PorousMedium Experimental System
3.1.1 Experimental Columns
9
3.1.2 Injection System and Dye Tracer Tests
3.1.2.1
9
PointSource Injection
11
11
3.1.2.2 LineSource Injection
11
3.1.2.3
14
Planar Injection System
3.2 Biologically Active Zone (BAZ) Experiments
3.2.1 Experimental Methods
17
17
3.2.1.1
Experimental SetUp
17
3.2.1.2
Characteristics of Column and Feed
Composition
21
Sampling afld Analytical Methods
25
3.2.1.3
3.2.2 Results for Column 1
26
3.2.3 Results for Column 2
31
3.2.4 Headioss Through the BAZs
38
iii
3.2.5 Determination of Kinetic Parameters
38
3.2.5.1
Determination of Y
41
3.2.5.2
Determination of k
45
3.2.5.3
Determination of K s
45
3.2.5.4
Determination of b
47
3 .3
Secondary Utilization of Halogenated Organic
Compounds in BAZ s "
3.3.1 Experimental Methods
49
3.3.1.1
Selection of Halogenated Organic Compounds .4 9
3.3.1.2
Experimental SetUp
49
3.3.1.3
Sampling and Analytical Methods
51
3.3.2 Results for Secondary Utilization Experiments
3.3.2.1
Removal of Halogenated Aliphatics in
Denitrification Columns
52
52
3.3.2.2
Removal of Dichlorobenzenes in Denitrification
Columns
61
3.3.2.3
Removal of Dichlorobenzenes with Hydrogen
Peroxide Injection
65
CHAPTER 4. CO.MPUTER MODELING
4.1
47
OneDimensional Solute Transport Model
73
73
4.2 Biofilm Phenomena and Kinetics
74
4.3 The Quasilinearization Technique
79
4.4 Treatment of Lateral Injection Ports
85
4.5
87
Development of the Secondary Utilization Model
iv
CHAP1ER 5. APPLICATION OF COMPUTER MODELING
91
5.1 SOC and N03 Profiles
91
5.1.1 OneBAZ Column
91
5.1.2 TwoBAZ Column
94
5.2
SecondarySubstrate Profiles
5.2.1 Carbon Tetrachloride
95
98
5.2.2 Bromoform, Ethylene Dibromide, Tetrachloroethene.
101
and Trichloroethene
5.3
Simulation of Bioreclamation Strategies
CHAPTER 6. CONCLUSIONS AND RECOMMENDATIONS
REIm:EN"CES
104
111
,.. 115
APPENDIXNOMENCLATURE
v
LIST OF TABLES
Page
Item
Table 3.1. Flow Rates Used for Dye Test
17
Table 3.2. Characteristics of Column Reactor
24
Table 3.3. Composition of Feed Solution for Denitrifying
Columns
24
Table 3.4. Secondary Utilization Experiments for Halogenated
Organic Compound Removal
50
Table 3.5. Composition of Feed Solution for Hydrogen Peroxide
Injection Column
51
Table 4.1. Parameters used for the Comparison of Numerical
Methods
Table 4.2.
Comparison of Efficiency for Traditional Time
Stepping and Quasilinearization Techniques
84
84
Table 5.1. Parameters used in SoluteTransport Modeling of
O'neBAZ Column
92
Table 5.2. Parameters used in SoluteTransport Modeling of
TwoBAZ Column
97
Table 5.3. K s and k Values of TeCE, EDB, and TCE Obtained
from Numerical Curve Fitting
.103
Table 5.4
104
Parameters used in Clogging Example Problem
VI
LIST OF FIGURES
Item
Figure 1.1.
Figure 3.1.
Figure 3.2.
Figure 3.3.
Page
Strategies for in situ bioreclamation of
contaminated groundwater.
3
Schematic of column reactor to establish
Biologically Active Zones (BAZs). .
10
Pointsource injection of thymolblue dye
solution
12
Linesource injection of thymolblue dye
solution
13
Figure 3.4.
Planar injection system with hypothetical
segments of area assumed to be completely mixed
15
with each orifice discharge. ..
Figure 3.5
Final design of planar injection system. .
Figure 3.6.
Planar injection of thymolblue dye solution at the
defined flow characteristics in Table 3.1 (Run 1)... 1 8
Figure 3.7.
Planar injection of thymolblue dye solution at the
defined flow characteristics in Table 3.1 ; (a) Run
2vertical transverse direction, (b) Run 219
horizontal transverse direction. ..
Figure 3.8.
Planar injection of thymolblue dye solution at the
defined flow characteristics in Table 3.1 (Run 3). ..20
Figure 3.9.
Locations of injection ports in Columns 1 and 2. .... 22
16
Figure 3.10. Experimental setup for biologicallyactivezone
(BAZ) experiments. .
23
Figure 3.11. Soluble organic carbon concentrations in Column
1
27
Figure 3.12. Relative effluent carbon concentrations, with
reference to input carbon, in Column 1. .
27
Vll
Figure 3.13. SOC profile at different times in Column 1.
29
Figure 3.14. BAZ distribution in Column 1 at day 351.
29
Figure 3.15 . Average SOC and N03 concentration profiles in
Column 1. .
30
Figure 3.16. N2gas accumulation in Column 1.
30
Figure 3.17. Gas accumulation and distribution in Column 1.
... 32
Figure 3.18. Soluble organic carbon concentrations in Column
2
34
Figure 3.19. SOC profiles in Column 2.
34
.
Figure 3.20. Gas accumulation and distribution in Column 2...... 3 6
Figure 3.21. N 2 gas accumulation in Column 2.
.
36
Figure 3.22. Average SOC and N03N concentration profiles in
37
Column 2. .
Figure 3.23. Biofilm distribution in Column 2 after 297 days
operation: distribution of dry cell mass (a),
biofilm thickness, Lf(b), and biofilm density, Xf
39
(c)
Figure 3.24. BAZ distribution in Column 2 at day 296. .
40
Figure 3.25. Locations of harvested cells for kineticparameter
determination
42
Figure 3.26. SOC utilization and cell growth in batch reactors. ..43
Figure 3.27.
Determination of average cell yield in Batch 5
44
Figure 3.28. Comparison of estimated biomass (Xo + Y(So S))
and measured biomass (X5) and in Batch 5
46
Figure 3.29. Exponential growth semilog plot for five batch
reactors. .
46
Figure 3.30. Determination of K s from Jl vs. S curve (Batch 5). .. 4 8
Vin
Figure 3.31. Cell decay in declining phase for Batch 12. .
48
Figure 3.32. SOC concentrations during Run 1 in a denitrifying
column. .
53
Figure 3.33. Profiles of halogenated aliphatic compounds in
denitrifying column at 50min. detention time
after nitrate injection (Run 1a).
.
a
55
Figure 3.34. Profiles of halogenated aliphatic compounds in a
denitrifying column at 125min. detention time
after nitrate injection (Run 1b). .
56
Figure 3.35. Profiles of halogenated aliphatic compounds in a
denitrifying column at 500min. detention time
after nitrate injection (Run Ic, day 406). .
57
Figure 3.36. Profiles of halogenated aliphatic compounds in a
denitrifying column at 500min. detention time
58
after nitrate injection (Run Ic, day 415)
Figure 3.37. Profiles of halogenated aliphatic compounds in
denitrifying column at 50min. detention time
after nitrate injection (Run Id). .
a
Figure 3.38. Profiles of halogenated aliphatic compounds in
denitrifying column at 50min. detention time
after nitrate injection (Run 2). .
a
59
62
Figure 3.39. Profiles of 1,2 and 1,3DCB in a denitrifying
column at 500minute detention time after nitrate
.
63
injection (Run Ic, day 415).
Figure 3.40. Profiles of 1,2 and 1,3DCB in a denitrifying
column at 50 minute detention time after nitrate
injection (Run Id). .
64
Figure 3.41. SOC concentrations in a H202 injection column
(Run 3)
66
Figure 3.42. Profiles of DCBs in an HzOzinjection column at 5068
minutes detention time (day 44 from Run 3). .
IX
Figure 3.43. Profiles of DCBs in an H202-injection column at 50minutes detention time (day 46 from Run 3). .
69
Figure 3.44. The effects of H202 on DCB removals in
nonbiological batch reactors. .
Figure 4.1.
Figure 4.2.
Figure 4.3.
Figure 5.1.
70
Conceptual Basis of the Biofilm Model (after
Rittmann and McCarty, 1980a). .
74
Comparison of traditional methods and quasilinearization for numerical solution of equation
(4.2) with the parameters given in Table 4.1
83
Comparison of upstream weighting and central
finite differencing on the lateral injection
prediction ability. .
86
Comparison of laboratory and numerical results
for the one-BAZ column. Zero distance indicates
the injection port. .
93
Figure 5.2.
Comparison of laboratory and numerical results
for the two-BAZ column. Nitrate injections are at
0.0 and 10.0 em. Lines represent model
prediction. .
96
Figure 5.3.
Numerical curve fit to the CTC profile at a
detention time of 50.0 min. The k and K s are
0.030 Ilg/mg cell-day and 4.5 Ilg/l, respectively.... 99
Figure 5.4.
Prediction of the eTC profile at a detention time of
125. min. and with k = 0.030 jlg/mg cell-day and
K s = 4.5 Ilg/l
100
Figure 5.5.
Prediction of the CTC profile at a detention time of
500 min and with k = 0.030 Ilg/mg cell-day and K s
= 4.5 Jlg/I
100
Figure 5.6.
Numerical fit to the BF profile at a detention time
of 125 min. The k and Ks values are 0.013 jlg/mg
cell-day and 9.5 jlg/l, respectively
102
x
Figure 5.7.
Prediction of the BF profile at a detention time of
500 min and with k = 0.013 IJ.g/mg cell-day and
K s = 9.5 Jlg/I
102
Figure 5.8.
SOC and nitrate profiles for one and three
injections of N03-
Figure 5.9.
106
Relative biofilm thicknesses comparing single and
107
multiple nitrate injections. .
Figure 5.10. Profiles of SOC and nitrate after their being
injected alternately.
.
108
Figure 5.11. Profile showing additional secondary utilization of
CTC after SOC is added by a second injection at 10
em
109
Xl
LIST OF ABBREVIATIONS
BAZ
BF
cpm
biologically active zone
bromoform
counts per minute
carbon tetrachloride
CIC
dichlorobenzene
IXB
FDB
ehtylene dibromide
soluble organic carbon
~
SMP
soluble microbial product
TeCE
tetrachoroethene
TCE
trichloroethene
l,l,l-TCA I, I, I-trichloroethane
1,2-DCB
1,2-dichlorobenzene
1,3-DCB
1,3-dichlorobenzene
1,4-DCB
1,4-dichlorobenzene
xii
ABSTRACT
This report summarizes the results of a research project aimed at
developing a better mechanistic understanding of the phenomena
controlling in situ biological activity.
A methodology involving
laboratory-column experiments and computer modeling was utilized
to investigate the formation of biologically active zones (BAZs) when
a limiting electron acceptor (N03 -) is injected along the flow path and ~
the secondary utilization of trace-level pollutants contained in the
water flowing through the BAZ. Laboratory experiments conducted
in a unique one-dimensional porous-medium column demonstrated
the relationship between lateral injection of N03 - and the location
and extent of BAZs when acetate was present as the sole carbon
source. BAZs established and sustained by acetate and N03 - were
able to degrade trace-level halogenated compounds.
Carbon
tetrachloride was nearly completely removed, while bromoform,
dibromomethane, trichloroethene, and tetrachloroethene were
removed to lesser degrees.
Trichloroethane was slightly removed.
Dichlorobenzenes, previously thought to be refractory in denitrifying
conditions, were removed by 20-30% during their passage through
the BAZ.
The fundamental phenomena of BAZ formation and the utilization
of limiting, nonlimiting, and secondary substrates were expressed
quantitatively in a computer model that coupled principles of onedimensional solute transport and steady-state-biofilm kinetics.
A
new, highly efficient solution algorithm was developed to solve
directly for the steady-state profiles of the limiting substrate and
biofilm mass, as well as for non-limiting and secondary substrates.
The predictive ability of the model was verified by successful
simulation of the laboratory experiments using independently
determined kinetic parameters.
The verified model was used to _
illustrate two possible strategies for field bioreclamation. First, theuse of multiple injection points can decrease aquifer clogging
potential by spreading out the extent of the BAZ.
Second, injection
of a supplementary carbon source can extend the length of the BAZ
in order to achieve greater removals of secondary substrates.
xiii
In Situ
Bioreclamation of Contaminated Groundwater
EXECUTIVE SUMMARY
In situ bioreclamation of contaminated groundwater is a
promising new technique for enhancing the clean-up rate of aquifers
contaminated with organic pollutants, such as chlorinated solvents,
gasoline constituents, and pesticides. In situ bioreclamation involves ~
injecting the materials necessary to significantly increase the
microbiological activity in the subsurface. The injected material is a
component that limits the growth of the desired microorganisms and
is usually an electron acceptor (e.g., oxygen or nitrate), a carbon
source, or a macro-nutrient (e.g., nitrogen or phosphorus). Injecting
the proper amount of the limiting material creates a region of
increased microbiological activity, called the Biologically Active Zone
(BAZ).
Creation of a BAZ offers major advantages for aquifer clean-up
because the removal agents, the bacteria, are in close proximity to all
the contaminants, including those in the water, those sorbed to
aquifer materials, and those in a nonaqueous liquid phase. Thus, the
relatively slow mechanism of flushing by water flow is replaced by a
degradation reaction very near the source of contaminants.
The direction of this study is towards an increased ability to
understand and quantitatively describe the key phenomena affecting
the formation of and reactions within a BAZ. The specific objectives
are:
1. Develop a laboratory-scale, porous-medium column that can be
used to create and study BAZs under well-defined conditions.
2. Evaluate the formation of one or more BAZs within the
laboratory-scale column when the electron acceptor, nitrate,
injected into the flow path.
IS
3. Using the laboratory-scale columns, evaluate the fate of
commonly found halogenated solvents as they passed through the BAZs.
XIV
4. Develop and test an efficient computer model for the formation of
BAZs and the utilization of substrates by the BAZs.
5. Apply the model to describe and interpret the formation of the
BAZ and the fates of the various substrates in the column
experiments.
6. Employ the model to evaluate strategies for in situ
bioreclamation in the field.
The laboratory-scale, porous-medium columns were constructed
of 2.5-cm inside diameter by 22.5-cm long glass tubes and were
filled with 3-mm glass beads. Ports were placed every 2.5 em to
allow for sampling and/or injection.
Special injection assemblies
were designed to allow for uniform planar injection of substrate into
the flow path. The systems gave an excellent approximation of onedimensional flow.
The organic source was sodium acetate, which was available as a
14CradioIabelled tracer. It was fed continuously to the inlet end of
the column from an elevated reservoir.
In most experiments, the
injected material was the electron acceptor, N03 . One or two BAZs
were established at and downstream of the injection ports. In order
to ensure that N03 was limiting, no other electron acceptors were
added to the feed medium, and extreme measures were taken to
preclude 02 entry in the reservoir, feed lines, and columns.
Welldefined BAZs developed from the injection ports and up to
7.5 em downstream of the injection ports. Photography of the intact
columns and of the beads in the columns demonstrated that the
bacterial growth was present as biofilms on the glass beads.
Photography and measurements of biofilm mass on the beads
confirmed that the amount of accumulated biofilm was greatest right _
after the injection port, and it gradually declined downstream.
Acetate (expressed as soluble organic carbon, SOC) and N03declined across the BAZs according to the expected stoichiometry,
0.67 mg N03 N/mg soc. For the column with two BAZs, removal of _
SOC was partial in the BAZ after the first injection port, because N03was depleted; however, SOC removal was essentially complete in the
xv
second BAZ, as sufficient N03 - was supplied in the second injection.
These results demonstrated that stoichiometric addition of an
electron acceptor could be used to remove an electron -donor
substrate to the degree desired.
Formation of
denitrification of
BAZs and caused
of SOC removal.
and demonstrated
within a confined
N2 gas bubbles occurred as
N03 -. These bubbles tended to
some short-circuiting, which led
Removal of the bubbles restored
the possible deleterious effects
aquifer.
a result of the
accumulate in the
to a deterioration
the SOC removal
of gas evolution
Eight trace-concentration halogenated solvents were applied to
the feed of the column having one BAZ. Two dichlorobenzenes were
added
together as a mixture, and six one- or two-carbon
halogenated aliphatics were added as another mixture.
Of the
halogenated aliphatics, carbon tetrachloride was removed the most
completely by the denitrifying BAZ. Tetrachloroethene, bromoform,
ethylene dibromide, and trichloroethene were removed to lower
degrees.
Trichloroethane was slightly removed.
1,2 and 1,3
dichlorobenzene also were 20-30% removed during passage through
the BAZ. Significant increases in the fractional removal were effected
as the liquid flow velocity was decreased, which increased the
contact time in the BAZ. These results are especially significant for
two reasons.
First, they show that common groundwater
contaminants were degradable in the BAZs induced by N03injection. Most interesting are the removals of the dichlorobenzenes
and trichloroethene, which were thought previously to be refractory
under denitrifying conditions.
Second, the results show that the
removals of each compound depended upon the degradation kinetics
of the particular compound and the contact time in the BAZ.
Modeling of the formation of a BAZ was based on application of biofilm kinetics to solute transport in porous media.
The steadystate-biofilm model, developed originally by Rittmann and McCarty
(1980a) and improved recently by Saez and Rittmann (1988), was
incorporated into a one-dimensional, steady-state, solute-transport
equation. The equation was transformed from the differential form
to one using discrete finite differences and solved numerically
directly for the steady-state profiles of substrate concentration and
XVI
biofilm accumulation.
Major modeling advancements were the
ability to have lateral injection sources at any point along the column
and the use of quasilinearization to give a highly efficient and direct
solution for the steady state. The quasilinearization technique, which
involves substituting a first-order Taylor series approximation for
the highly nonlinear reaction term, made the convergence to steady
state approximately ten times faster than by conventional methods.
Even greater improvements are expected for more complicated
geometries.
The modeling also was advanced by explicit coupling of the
steady-state-biofilm model solution, which solves for the
concentration profile of the limiting substrate and the amount of
biofilm, to models for a non-limiting substrate and for secondary
substrates. An example of a non-limiting substrate is N03 - when SOC
is limiting; the flux of N03 - into the biofilm was set equal to the flux
of SOC multiplied by a stoichiometric coefficient. Although the flux of
the non-limiting substrate was determined by the flux of limiting
substrate, it had its own rates of advection, dispersion, and injection.
A secondary substrate is, in this context, a trace-level contaminant
that is removed in the BAZ, even though its utilization provides
negligible or no benefit to the microorganisms. The flux of secondary
substrate was determined by its own kinetic characteristics and by
the amount of biofilm accumulated through utilization of the SOC.
The steady-state, solute-transport model for the limiting substrate
and the coupled transport model for the non-limiting substrate were
used to evaluate the experiments on the formation of BAZs. Kinetic
parameters for the utilization of the SOC were determined
independently; thus, model results were true predictions. SOC, N03 -,
and biofilm profiles matched the experimental results very well for
columns with one and two BAZs. Model predictions and experimental
results agreed quantitatively that removals of SOC and N03 - and --accumulation of biofilm were greatest in the first 2.5 cm beyond the
injection port.
Removal rates and biofilm accumulation declined
gradually in the next 5.0 cm, and substrate concentrations attained a
steady plateau value thereafter.
Predictions and experimental
results also concurred that injection of more of the limiting material
(N 03 - in this case) allowed formation of a second BAZ and renewed
removal of SOC. The model predictions correctly described all trends,
xvii
and absolute deviations between predicted and experimental results
were small in all cases.
The coupled transport model for secondary substrates also was
used successfully for describing the removal of the halogenated
aliphatic solvents.
Since the kinetic parameters for each secondary
substrate could not be determined independently, one set of results
from the column experiments was used to obtain a best-fit set of
kinetic parameters. These parameters were then used to predict theremoval across the BAZ for experiments with different liquid flow
velocities. Model and experimental results agreed well on the effect
of liquid flow velocity. When the flow velocity was decreased, the
contact time for the secondary substrates in the BAZ was increased
proportionally.
This increase in contact time allowed greater
removal.
For example, experimental and modeling results agreed
that the removal of carbon tetrachloride through a BAZ should
increase from 18% to 55% to 92% as the post-injection detention time
increased from 50 minutes to 125 minute to 500 minutes,
respectively.
The steady-state models were applied to investigate possible
strategies to be used in field bioreclamations. The use of multiple
injection wells was studied for its ability to decrease aquifer clogging
potential by spreading out the distance over which limiting substrate
is added.
Modeling results verified that the strategy of multiple
injections could reduce high densities of biofilm accumulation near
the injection well. Also investigated was the strategy of adding a
supplemental carbon source to extend the length of a BAZ.
The
modeling illustrated that such an extension of the BAZ could be
accomplished and could result in longer contact times of a secondary
substrate with the BAZ, thereby increasing the removal of the
secondary substrate.
xviii
CHAPTER 1. INTRODUCTION
Contamination of groundwater by organic materials -- such as
chlorinated solvents, petroleum products, and landfill leachates -- is
widely recognized as one of the most critical environmental problems
of recent times.
Currently, clean-up efforts usually involve
extraction of the contaminated water, followed by physical, chemical,
or biological treatment. Because the organic contaminants can adsorb
onto aquifer solids or can be trapped in regions of relatively lowpermeability, the volume of water required to be extracted is many
times larger than the pore volume that is contaminated; thus
conventional clean-up is very expensive and time-consuming.
In situ biological degradation is being proposed as a promising
alternative for aquifer restoration. In situ projects typically involve
a set of extraction and injection wells, which establishes a defined
flow field and permits inputs of seed microorganisms, electron
acceptor, carbon source, or other nutrients at one or more points
Being a very new technology, ins it u
along the flow path.
bioreclamation designs have been based on only a few simple
microbiological experiments aimed at testing biodegradation
potential and nutrient requirements.
Incorporation of realistic
biodegradation kinetics and groundwater hydraulics has not been
accomplished.
An initial requirement of any in situ decontamination technology
is that the flow field be defined. Otherwise, contaminants can escape
treatment by migrating out of the treatment site or by remaining in
isolated portions of the aquifer. For in situ bioreclamation, however,
more than a defined flow field is required: the water in that flow
field must pass through a biologically active zone before it is
extracted or leaves the treatment site. The biologically active zone in
an aquifer is made up almost completely by microorganisms attached
as biofilms to the large amount of surface area presented by the-=aquifer solids. Even in uncontaminated aquifers, bacteria are found
attached to aquifer solids; however, their densities are very low
( 106/gram of soil), and their metabolic capabilities are largely
undefined (Ghiorse and Balkwill, 1983).
Successful in si tu
bioreclamation requires that the attached biomass be increased
greatly from that normally found on aquifer solids. In some cases,
different types of microorganisms, having capabilities not found in
1
the natural community, should be added as seed. In almost every
situation, however, success requires that the microorganisms grow to
attached densities a hundred or more times that naturally present.
Cell growth and accumulation in an aquifer depend on the
availability of an electron donor, an electron acceptor, and several
other nutrients, such as nitrogen, phosphorous, and sulfur. Usually,
one factor is rate limiting and controls how much cell mass can be
accumulated. The growth-limiting factor can be called the limitingsubstrate (McCarty et aI., 1981). Enhanced in situ bioreclamation
usually involves adding the limiting substrate in such a manner that
the growth limitation is eliminated and significant quantities of
biomass are generated in the aquifer.
What the limiting substrate must be varies with the
contaminating situation. For instance, a leak or spill that creates high
organiccontaminant concentrations probably is limited by the
electron acceptor or a nutrient.
On the other hand, lowlevel
contamination of a drinking water supply by a distant source creates
a situation in which an organic electron donor is needed to allow
significant growth.
The objective of enhanced in situ bioreclamation is to establish a
biologically active zone by supplying the limiting substrate in such a
manner that no contaminant escapes biodegradation.
However,
biodegradable material added via an injection well to enhance in si tu
biodegradation often is consumed very rapidly near the injection
well (Rittmann and McCarty, 1980a; Rittmann et aI., 1980), creating
two significant problems: (1) biological growth is limited to only a
region very near the injection well, and (2) well clogging can occur.
The first problem is quite serious, since localization of biological
activity prevents adequate contaminant/microorganism contact
throughout most of the aquifer. The second problem also is serious
because clogging retards the input of the limiting substrate and may
force the groundwater flow to go around the biologically active zone.
The problem of localized biological activity can be solved, at least
in principle, by providing multiple injection wells perpendicular to
and/or along the flow path.
Figure 1.1 a depicts the case where multiple injection wells are placed laterally along the flow path to
create a sawtooth pattern of nutrient concentration, which allows
2
(a) enhancement of in situ biodegradation along the groundwater
flow path
injection
extraction
wells
DID
•
011
•
wells
natural
flow
lID. injection/extraction wells for hydraulic control of plume
migration
Ll injection wells for stimulation of in situ biological activity
•
biologically active zone
(b) enhancement of in situ biodegradation perpendicular
to groundwater
Figure 1.1. Strategies for in situ bioreclarnation of contaminated
grounowater.
3
the biological activity to extend the necessary distance to assure
Figure 1a also demonstrates the
adequate contaminant removal.
concept of coupled hydraulic/biological reclamation, since a network
of injection/extraction wells is utilized to hydraulically isolate the
contaminant plume from the natural groundwater flow regime.
To
date, all reported cases of in situ bioreclamation have included
hydraulic control measures; in fact, several projects did not utilize
multiple injection wells along the flow path and, thus, biological
activity was most likely concentrated in the vicinity of the hydraulic-control injection wells (Nagel et aI., 1982; Werner, 1985; Flathman et
aI., 1983, 1984).
The need for multiple injection wells along the flow path is most
acute for two commonly encountered situations.
The first occurs
when the limiting substrate is oxygen, a common electron acceptor.
Because of the low solubility of dissolved oxygen (about 9 mg/l when
exposed to the atmosphere) and its reactivity with reduced
materials, supplying dissolved oxygen from one injection point
cannot maintain sufficient dissolved oxygen throughout the flow path
when the amount of organic material to be degraded is more than a
Since degradation of certain common classes of
few mg/l.
compounds, especially including benzene derivatives, appears to
occur best (and likely exclusively) when oxygen is available, the
application of oxygen is likely to be a major vehicle for enhanced in
situ bioreclamation. Other oxygen sources are ozone and hydrogen
peroxide; although application of these materials overcomes some of
the solubility problems of dissolved oxygen, they are reactive with
reduced materials and are toxicants to microorganisms. Thus, they
cannot be applied in unlimited amounts.
The second common occasion when multiple Injections are needed
along the flow path occurs when an electron donor, usually an
organic compound, must be applied to allow increased growth of _
microorganisms that bring about contaminant removal throughsecondary utilization or co-metabolism (McCarty et aI., 1981;
Kobayashi and Rittmann, 1982; Stratton et al., 1983). Because the
electron donor can be utilized quickly near the injection well, small
input concentrations do not penetrate far into the aquifer, but large _
concentrations cause well clogging through biomass plugging or gas
binding (in methanogenic or denitrifying cases). Thus, the electron
4
donor input must be spread out along the flow path to give a
sufficient amount of microorganisms without plugging the aquifer.
Figure 1.1 b shows the case in which multiple injection wells are
placed perpendicular to the groundwater flow path to create a
biologically active zone through which all of a contaminant plume
must pass.
This bioreclamation scheme is probably less expensive
than that shown in Figure 1.1 a, since hydraulic control measures are
not utilized. Creating a biologically active zone perpendicular to thenatural groundwater flow path is a novel concept in the field of
aquifer restoration.
5
CHAP'IER 2. PROJECT OBJECTIVES
The overall objective of the project is to develop, evaluate, and
demonstrate a predictive modeling approach that combines realistic
phenomena for biofilm degradation and groundwater hydraulics and
that is suited to in situ bioreclamation schemes. The primary focus is
to investigate the fundamental mechanisms that act when an
electron acceptor is injected along the flow path of an electrondonorrich groundwater to establish a biologically active zone (BAZ).The most important mechanisms considered in this project are the
development of the biological activity within the porous medium and
the biodegradation of primary and secondary substrates in the
flowing water.
To accomplish the overall objective, the following
specific tasks have been performed:
1. A unique onedimensional biofilm reactor was designed and
developed to provide for substrate injection and sampling along
the flow path.
2. Experiments evaluating the formation of BAZs were conducted
with the biofilm reactors. Acetate was fed as the sole carbon
source and nitrate was injected as the limiting electron acceptor.
3. Secondary substrate utilization in BAZs was studied by conducting
experiments where various chlorinated solvents at low
concentration were fed into columns with established BAZs.
4. A new, highlyefficient numerical model that couples solute
transport mechanisms and biofilm kinetics was developed. The
model is capable of solving directly for the steadystate profiles of
primary limiting and nonlimiting substrates, secondary
substrates, and biomass.
5. The predictive ability of the model was verified by application to _
the laboratory experiments.
6. The model also was used to conduct numerical studies of the
impact of various hypothetical lateral injection schemes on the
overall efficiency of in situ bioreclamation.
7
CHAPTER 3. EXPERIMENTAL METHODS AND RESULTS
3.1
Porous-Medium Experimental System
3.1.1 Experimental Columns
In order to accomplish the research objectives defined in Chapter
2, a unique experimental set-up was designed with the following_
characteristics:
a. to provide a porous matrix having surface for biofilm
growth
b. to provide a well-defined, one-dimensional water flow
c. to feed electron donor (organic matter) and electron
acceptor (e.g., nitrate) in an independent manner
d. to have multiple electron-acceptor injection ports to
create biologically active zones (BAZs)
e. to take liquid samples along the length of the flow path
f. to measure the pressure drop along the length of the
column
Figure 3.1 is a schematic of the experimental system. A 2.S-em
diameter by 22.5-em long glass column packed with 3-mm diameter
glass beads was used to provide one-dimensional flow in a porous
matrix. Bulk flow was established by pumping in the electron-donor
feed solution at one end.
A key feature was that injection and
sampling ports were located along the length of the reactor.
Two
injection sites were arranged to satisfy purposes (c) and (d) above.
Six sampling ports and the effluent line provided syringe sampling
for substrate concentration (purpose (e)) and could also be used as
manometers to measure headlosses which arise when the biofilm
accumulated on the porous medium (purpose (f)). An injection port
also could be utilized as a sampling port or a manometer when it was
not used for injection.
The peristaltic, bulk-flow pump had a wide range of dispensing
rates (0.02 ml/min to 22 ml/min), such that the bulk flow velocity
could be manipulated to satisfy different experimental purposes.
A
multiple-channel syringe pump or a low-speed peristaltic pump was
used for electron acceptor injection; thus, independent application of
electron donor and electron acceptor was achieved.
9
-
Electron Acceptor
Injection
Feed from
Reservoir
/
Ports for Sampling
and Head Measurement
I \
Effluent
to
Waste
Biologically Active Zones
Glass-bead Medium
3-mm diameter
Glass Chromatographic Column
225 cm long by 2 5 cm in diameter
Figure 3.1. Schematic of column reactor to establish
Biologically Active Zones (BAZs).
10
3.1.2 Injection System and Dye Tracer Tests
In order that the mechanisms of biofilm accumulation, substrate
utilization, and clogging could be studied quantitatively, it was
necessary to eliminate complications arising from spatial variations
in concentration of injected material.
To preclude creating
concentration gradients perpendicular to the flow path and, thus, to
provide a satisfactory one-dimensional regime, the injection
arrangement was designed to give a uniform distribution of theinjected material across the cross-section of the column.
Dye tests were carried out to asses the hydrodynamic dispersion
of material injected from injection ports. A typical bulk flow rate of
0.5 ml/min (which corresponds to 0.1 cm/min of superficial flow
velocity or 0.25 cm/min of interstitial velocity) was adopted for
these tests.
An alkaline thymol-blue dye solution was introduced
through one injection port at a flow rate of one percent of the bulk
flow rate. When the alkaline thymol-blue solution, which was yellow
at acidic pH, mixed with the bulk flow, the thymol blue was exposed
to a pH higher than 9, and it turned to a blue color.
3.1.2.1
Point-Source Injection
Dye solution was injected through a stainless steel needle to the
exact center of the column. Figure 3.2 shows that the effect of crosssectional hydrodynamic dispersion (mechanical mixing + molecular
diffusion) was slow compared to advection (bulk flow); thus, a long,
funnel-shaped dye distribution was observed.
Clearly, the point
source did not provide a satisfac torily uniform injection.
3.1.2.2
Line-Source Injection
To provide a more uniform dye distribution, a line-source
injection was tested by using a closed teflon tube that had thirteen _
0.2-mm dia. orifices evenly spaced along the length of the tube.Although the dye distribution was strikingly improved in the vertical
transverse direction (see Figure 3.3a), Figure 3.3 b demonstrates that
the horizontal transverse distribution was nearly as poor as the case
of a point-source injection.
Thus, line-source injection was _
inadequate for establishing uniform input across the column's crosssection.
11
Figure 3.2. Point-source injection of thymol-blue dye solution. -=.
Flow characteristics were: superficial velocity = 0.1
em/min, interstitial velocity = 0.25 em/min, and dye
flow rate = 1% of bulk flow rate.
12
(a)
(b)
Figure 3.3. Line-source injection of thymol-blue dye solution; (a)
vertical transverse direction, (b) horizontal transverse
direction. The flow characteristics were the same as
in Figure 3.2.
13
3.1.2.3
Planar Injection System
Since a uniform distribution of electron acceptor from the point of
injection was essential, a planar-source modification of the injection
system was designed. A triplet of injection ports, each of which was
The system is shown
an injection needle, was provided.
schematically in Figure 3.4. Small orifices, 0.1 mm in diameter, were
spaced along the length of the injection needle. Since the goal of the
planar-injection system was to ensure uniform crosssectionalmixing, the orifices along each needle were spaced so as to provide
an equal injection rate per unit crosssectional area (see Figure 3.4).
The orifice spacing along the needle was determined by two factors.
The first was the unequal distribution of areas occupied by
successive annular segments in the crosssection.
In other words,
the outer annular segments had greater area per unit of radius than
did annular segments near the center, since area is a function of the
radius squared.
Second, the injection pressure at the top of the
needle was controlled by the injection pump, but frictional losses
caused the fluid pressure to decrease along the needle. Thus, orifice
flow rate diminished from the top to the bottom of the needle,
because orifice flow rate is a function of the pressure on the inner
side of each orifice.
An iterative calculation procedure was devised to compute the
spacing that guaranteed uniform crosssectional injection.
The
DarcyWeisbach equation (Daugherty and Franzini, 1977) for laminar
flow was used to compute the pressure loss along the needle. The
flow rate through each orifice was calculated from the remaining
pressure at the location of each orifice, using the same DarcyWeisbach equation. The areas of the annular segments in Figure 3.4
were determined in such a way that they were proportional to the
flow rates of corresponding orifices. The flow out of each orifice was
assumed to immediately and completely mix with its annular
segment of the crosssection, as shown in Figure 3.4.
The final design of the planar source is shown in Figure 3.5. In
order to avoid an overlap of orifices at the center of the cross section,
only the vertical needle (type A) had a center orifice. Stainless steel
syringe needles having D.84mm inside diameter (18gage) and Punctures with O.lmmO.22mm wall thickness were utilized.
diameter holes with an exact spacing were possible by using an
14
Needle
Needle
Section of the
glass column
Figure 3.4. Planar injection system with hypothetical segments of area assumed to be completely mixed with each
orifice discharge.
15
0
2.41
0
2.32
0
0
2 23
0
2.035
Type B
Type B
1 55
0
1.25
0
o 96
0
0.61
0
a 50
o 40
o 23
0
0
Type A
0
0
0.31
0
0
0
0
0
( a)
0
0
0.75
0
0
0
1 77
0
Type A
0
0
1 915
0
0
0
2 14
0
0
0
0
o 16
0.08
0
0
t
Type B
Distance from
the tip (cm)
(b)
Figure 3.5. Final design of planar injection system: arrangement
of needles (a), and orifices (b).
16
electrical discharge machine with the help of the Materials Research
Lab at the University of Illinois.
Several dye tests were performed to assess the mixing properties
of the planar-injection system.
As before, a thymol-blue dye
solution was used as the injection fluid; several experiments, at the
flow rates given in Table 3.1, were performed. Figures 3.6 , 3.7a and
3.8 show the dye distribution in the vertical transverse direction of
Runs 1, 2, and 3, respectively, and Figure 3.7b shows the horiz ntal~
transverse dye distribution for Run 2. All combinations showed v~ry
uniform cross-sectional dye distributions in all directions. Thus, the
planar source was successful for achieving a uniform cross-sectional
injection.
Table 3.1. Flow Rates Used for Dye Test
Run
Bulk flow rate per unit
cross-sectional area
(cm3 /cm 2 -min)
Planar injection
flow rate
(% of bulk flow)
0.05
0.1
0.2
2.0
1.0
0.5
1
2
3
3.2 Biologically Active Zone (BAZ) Experiments
3.2.1 Experimental Methods
3.2.1.1
Experimental Set-Up
Two columns were run for the biologically active zone (BAZ)
experiment: Column 1 and Column 2.
Column 1 had one planar
injection port for electron-acceptor input. One injection source led to
Column 2, on the other hand, had two sets of planar
one BAZ.
injection ports, which led to two BAZs. In practice, one goal of having
multiple injections is to evenly distribute the biomass, which
prevents excessive build-up of biomass in one location and reduces
17
Figure 3.6.
Planar injection of thymol-blue dye solution at the
defined flow characteristics in Table 3.1 (Run 1).
18
(a)
(b)
Figure 3.7. Planar injection of thymol-blue dye solution at the
defined flow characteristics in Table 3.1; (a) Run 2vertical transverse direction, (b) Run 2-horizontal
transverse direction.
19
Figure 3.8. Planar injection of thymol-blue dye solution at the-=.
defined flow characteristics in Table 3.1 (Run 3).
20
the hydraulic headloss which arises as the biofilm growth clogs the
pore space. Having two BAZs was a means to distribute the biomass
more evenly.
The locations of the injection ports in Column 1 and 2 are shown
in Figure 3.9. All the other substrates and nutrients were fed with
the bulk flow from the feed reservoir, which was deoxygenated by a
combination of boiling and nitrogen-gas purging before use.
Special efforts were needed to prevent reoxygenation of the
prepared
feed solution.
First, the feeding peristaltic pump was
located at the column outlet, and the connection tubing between the
feed reservoir and the column was shortened as much as possible.
Placement of the pump after the column was required, because the
flexible peristaltic-pump tubing was oxygen permeable.
Second, a
slight positive nitrogen gas pressure (about 103% of the ambient
pressure) was applied to the feed reservoir to prevent penetration of
oxygen from the air and to replace the volume of liquid dispensed by
the peristaltic pump. Third, all the sampling ports were capped with
serum caps. The columns were set in a vertical direction, and the
feed solution was pumped in from the bottom to the top. The overall
experimental set-up is shown in Figure 3.10.
3.2.1.2
Characteristics of Column and Feed Composition
Characteristics of the column reactors and liquid flow are shown
in Table 3.2. The flow rate of the electron-acceptor injection at each
injection port was adjusted to about one percent of the bulk-flow
rate, and this was not taken into account in detention time
calculations.
The feed composition is shown in Table 3.3. Acetate (CH3COO-)
was fed as the sole carbon source. The concentration was 20 mg/L as
COD, 18.4 mg/L as acetate, or 7.5 mg/l as SOC. A small amount of 14C-acetate was added to label the feed carbon.
Denitrifying one
mole of nitrate with acetate destroys up to one mole of H+,
potentially causing a pH increase. Thus, phosphate compounds were
added in such a way that the medium had sufficient buffering
capacity to maintain the pH between 6.9 and 7.1. Sodium molybdate (N a2M 0 04) was added at 0.25 mM to prevent the growth of the
sulfate-reducing bacteria (Bouwer, 1987).
21
f\03
,....
,....
I
......
Inj.
-
-
,....
,....
,....
~
COLUMN 1
0.0
25
50
75
10.0
125
150
175
200
22.5
DISTANCE, em
W3
-
I
-
f\03 Inj.
Inj.
,....
r-
,....
-I
-
..
COLUMN 2
00
25
5.0
75
100
125
15.0
17.5
200
225
DISTANCE, em
Figure 3.9. Locations of injection ports in Columns 1 and 2.
22
Injection Pump
Nilro~
~
I"m:wc e~
Electron
Acceptors
C\I
t::
E
:::l
o<.:>
Feed
Reservoir
Pressure
Regulator
Feeding Pump
(a)
(b)
Figure 3.10.
Experimental set-up for biologically
(BAZ) experiments; (a) a schematic,
left: injection pump, feeding pump,
2, feed reservoir, and nitrogen-gas
23
active zone
(b) a picture, fnJill
column 1, column
pressure regulator.
Table 3.2.
Characteristics of Column Reactor
Parameter
Column Reactor:
Length
Diameter
Volume
Glass-bead diameter
Porosity
Liquid flow:
Feed-flow rate
Feed-flow velocity
- Superficial
- Inters titial
Nitrate-injection rate
Detention time
Total
- After 1st injection
Table 3.3.
Unit
Value
cm
cm
cm 3
cm
22.5
2.5
110
0.3
0.4
mL/min
0.49
cm/min
cm/min
mL/min
min
mIn
0.10
0.25
0.006 for Column 1
0.012 for Column 2
90
60 for Column 1
70 for Column 2
Composition of Feed Solution for Denitrifying Columns
Compound
Concentration, mg/L
7.5 as SOC
170.0
108.75
88.5
3.4
11.0
27.5
0.15
51.5
Acetate (CH3COO-)
KHZP0 4
KZHP04
NazHP04
NI4CI
Mg S04
CaCh
FeC13
NazMo04
24
The columns were inoculated with a 1% dilution of
denitrification-reactor effluent from another study.
Feeding started
on March 12, 1987 for Column 1 and on June 18, 1987 for Column 2.
3.2.1.3
Sampling and Analytical Methods
Samples the soluble organic carbon (SOC) determination of the
feed solution and effluent stream were taken twice a week. Samples
for the determination of the SOC and nitrate profiles along the flowpath of the column were taken when the SOC removal in the column
reached a steady state.
All samples, except for the feed solution, were taken with a
peristaltic pump collecting sample at a rate equal to the feedflow
rate. When taking samples from the sampling ports, a syringe needle
was inserted into the center of the cross section of the column.
Samples for the feed solution were taken by hand using a syringe.
For each sample, approximately 10 mL of liquid was collected.
The SOC concentrations in the samples taken from the sampling
ports, effluent stream, or feed reservoir were measured by counting
14C. The liquid sample 'was passed through a 0.45J.lm membrane
filter to remove the suspended portion of organic carbon. Then, C02
was driven off by acidifying the sample to pH 2 or less with one drop
of IN HCI and shaking the vial for 10 minutes in a shaker. 14C was
counted with a Beckman liquid scintillation counter (Model LSI00).
Thus, a filtered and acidified sample contained only soluble organic
carbon.
The biomass in the liquid sample was estimated by taking the
difference between the filtered and unfiltered organic carbon
concentrations from acidified samples.
The total carbon concentration in the sampleSOC, biomass, and C02 - was estimated by counting the total 14C in the sample. In this case, the sample was collected in an airtight syringe which contained
a small amount (2.5% after sampling) of CarboSorb II (United
Technologies Packard), a strong base that absorbed C02 for
scintillation counting. The difference between the total carbon and _.
the unfiltered organic carbon was the C02.
25
Nitrate was measured using the chromotropic-acid method as
described in Standard Methods (American Public Health Association,
1981). Nitrite was also determined following Standard Methods.
3.2.2 Results for Column 1
Column 1 was operated by injecting a stoichiometrically sufficient
amount of nitrate through a single injection port. The performance
The fe d~
data for the entire column are shown in Figure 3.11.
concentration was the SOC in the feed reservoir, and the influent SOC
concentration was measured from the samples taken at the port
immediately upstream from the injection port.
Therefore, the
difference between the feed and influent samples was aerobically
degraded SOC. Its utilization was caused by residual oxygen in the
feed or oxygen that diffused through the connection tubing between
The SOC decrease from the influent
the reservoir and column.
sample to the effluent sample was achieved by a denitrification
reaction.
The location of the feeding peristaltic pump was changed from
column inlet to outlet, as described in Section 3.2.1.1 on day 71.
Also, feed solution was boiled during the N2 -gas purging to enhance
the deoxygenation after this day.
These provisions drastically
improved the quality of the influent, maintaining it almost at the
original feed concentration throughout the experiment.
The effluent SOC showed a gradually decreasing tendency for
about 120 days, after which it maintained a very low, steady-state
concentration, except for a few cases of fluctuations which were
caused by occasional system disturbances, e.g. gas removal from the
column. The average effluent SOC after day 120 was about 0.2 mgjL,
which corresponds to 97% removal of the influent SOC.
The relative carbon concentrations of SOC, biomass-C, C02 -C, and total-C in the effluent are presented, together with the input-C, in
Figure 3.12. Although there were a few irregular datum points, the
overall pattern was that 67% and 20% of the input C were converted
to C02 and biomass, respectively, while 3% of input C exited the
column as unused SOC. One tenth of the input C was not recovered in
the effluent carbon measurement and was retained biomass.
Thus,
most input C was mineralized, but a significant fraction was
26
o ~=;-"Tl'.IFLiJf&
r-I
o
80
40
160
120
200
240
280
320
TIME, days
Figure 3.11.
200
Soluble organic carbon concentrations In Column 1.
I::J
180
soc
•
0
160
D
«
a:
140
0
Z
120
+
BIOMASS-C
C02-C
TOTAL-C
FEED-C
0
Z
100
z
t=
t-
w
0
0
w
>
80
60
~..J
40
a:
20
W
Hf+tI +f+
•
++
0
•
iI+ ..
~,
+
.
0
-HHK+++.........
0
00
~
•
0
D
++0
•
*++1++++
0
0
[]
D
••
0
80 90 100110120130140150160170180190200210220230240
TIME, days
Figure 3.12. Relative effluent carbon concentrations, with
reference to input carbon, in Column 1.
27
converted to biomass that could be transported In the fluid flow or
retained in the column.
Figure 3.13 shows several SOC profiles along Column 1. Because
the SOC concentration at the nitrate injection port could not be
measured, it was assumed to be the same as the SOC of the
immediate upstream port.
The results were quite reproducible,
suggesting that the BAZ was approximately at steady state.
The
majority of the SOC removal took place in the 2.5-cm region ~
immediately downstream from the nitrate injection port; then the
rate of removal diminished toward the column outlet. Thus, the BAZ
was mainly contained within about 7.5 cm of the injection.
Figure 3.14 is a photograph of Column 1 at day 351. The backlighting emphasizes that most of the BAZ was located between the
injection port and the third sampling port, a distance of 7.5 cm. The
slight dark coloration throughout the reactor is evidence of some
attached biological activity, but the dense area shows where the
main BAZ was located.
Figure 3.15 superimposes the N03 N profile over the SOC profile.
Figure 3.15 shows that the SOC was the limiting substrate after the
injection, because nitrate was always present at concentrations of at
least 2.8 mg NIL. The upstream port (5cm location) before nitrate
injection showed a substantial nitrate concentration.
Since the dye
tests (see Figures 3.63.8) did not show any back diffusion of the
injected material, it should be attributed to a sampling error.
Subsequent samples which were taken at a reduced sampling flow
rate and did not show any significant nitrate concentration at this
port.
A considerable amount of nitrogen gas should be produced
during the denitrification energy reaction (McCarty, 1972).
Stoichiometrically, a complete oxidation of 1 mg of acetateC by nitrate produces 0.747 mL of Nz gas. As the feed solution was
already saturated with nitrogen gas, most of the nitrogen gas
produced should have been released to the gas phase.
The
photograph shown in Figure 3.16 demonstrates that nitrogen gas was
released and trapped in the column. Gas trapped in the pore space was measured by removing the liquid from the sampling and
injection ports before and after gas accumulation.
The volume
28
W Nitrate
8
Injection
lEI
6
••
....J
m
E
(,)'
Day 78
Day 146
Day 151
4
0
tn
2
0
0.0
2 5
5 0
7.5
10 0
12 5
15 0
17.5
20 0
22.5
DISTANCE, em
Figure 3.13. SOC profile at different times in Column 1.
Figure _3.14. BAZ distribution in Column 1 at day 351. The reactor
conditions were as defined in Table 3.2 and Table 3.3.
29
10
..J
......
C)
E
t
8
Nitrate Injection
I!l
Z
0
F=
•
6
<C
Avg. SOC
Avg. N03-N
a:::
I
Z
w
4
0
z
0
0
2
0
o0
2 5
5.0
7 5
10 0
12 5
15 0
17 5
200
22.5
DISTANCE, em
Figure 3.15. Average SOC and N03- concentration profiles in
Column 1 (data from day 146 and day 151).
':.
Figure 3.16. N 2-gas accumulation in Column 1. The reflections
at the top side of the column were produced as a
result of gas-bubble accumulation.
30
difference of the drained liquid was assumed to be equal to the gas
volume trapped in each segment of the column. Gas analysis with a
gas partitioner (Fisher Gas Partitioner, Model 1200) repeatedly
showed that N 2 -gas was the only detectable component in the
collected gas samples.
Figure 3.17 shows the gas distribution in Column 1 at day 233,
which was 72 days after the gas removal. Total gas accumulated in
the column was 11.1 mL, which corresponded to 38% of the totalpore volume after the injection port. Even though the denitrification
reaction occurred mainly in the first two segments after the injection,
there was not much gas in those two segments. More than half of the
gas was trapped in the 3rd and 4th segments after injection. No gas
accumulation occurred before the nitrate injection, confirming that
the gas was produced by denitrification in the BAZ.
The location of nitrogengas accumulation can be explained by the
following scenario. First, nitrogen gas was produced in the BAZ, but
it was in the liquid phase. Second, as more liquidphase nitrogen gas
accumulated, it was gasified to small bubbles. Third, the gas bubbles
agglomerated together, growing to larger bubbles as the water and
bubbles flowed downstream.
Finally, the large bubbles were
trapped and accumulated in the pore space. Probably, a steady state
occurred from a balance between gas bubble transport from
upstream and bubble shearoff to downstream.
The overall removal of SOC in Column 1 did not deteriorate in
spite of the gas accumulation, because the gas accumulation was not
significant in the BAZ. Thus, the residence time in the BAZ was not
affected significantly by the gas accumulation.
3.2.3 Results for Column 2
Column 2 was operated by tnJecting nitrate in such a manner that about one half of the SOC fed was removed in the first BAZ, and the
other half was removed in the second BAZ. The total nitrate injection
was the stoichiometrically sufficient amount required to completely
oxidize the fed acetate.
Initially, an equal amount of nitrate was
injected at the two ports. Later, the ratio was adjusted so that the upstream port injected 25% of the total and the downstream injected
75%.
31
1.0
r,
0.9
0.8
GLASS BEADS
0.7
0.6
0.5
0.4
0.3
tI
0.2
LIQUID
nitrate injection
0.1
~
0.0 1---_.....--,---F==::....---.....------r---..,..---_---1
0.0
25
50
7.5
10.0
12.5 150 17.5 20.0 225
DISTANCE, em
Figure 3.17.
Gas accumulation and distribution in Column 1
(day 233, or 72 days after gas removal).
32
The overall performance of the entire column is shown in Figure
3.18.
The effluent SOC reached an apparent steady-state within a
week and maintained a good removal until the injection ratio was
changed. The average SOC concentration between 8 to 25 days was
0.09 mg/L, corresponding to 98.6% removal of the influent SOC.
When the nitrate injection ratio was changed, the average effluent
SOC concentration increased to about 1 mg/L, which corresponded to
86% removal efficiency.
An important phenomenon observed from days 26 to 106 was a
cyclic fluctuation of the effluent quality.
Gas bubbles were also
observed during this period.
On day 107, the gas bubbles in the
column were removed by draining the liquid from the column. Then,
the column was refilled with liquid as the N2 gas was put under
negative pressure. Throughout this procedure, every precaution was
taken to minimize system disturbance and biofilm loss. Figure 3.18
shows that the substrate removal was greatly enhanced almost
immediately after the gas removal.
The effluent SOC decreased
within 3 days to 0.14 mg/L, which was comparable to the Column 1
effluent (0.2 mgjL). However, the SOC began to increase after 15
days of operation, and it reached a maximum effluent value after 50
days (2.4 mg-SOC/L). After that, it decreased again to the previous
low level.
The dynamic effects of gas accumulation caused the
changes in SOC removal, which are discussed in detail below.
Typical SOC profiles obtained at different operational conditions in
Column 2 are shown in Figure 3.19. Profile 1 represents the reactor
performance when an equal amount of nitrate was injected through
Most of SOC removal took place right after the first
each port.
injection, and the remaining SOC was removed after the second
The overall removal efficiency of this injection was
injection.
excellent, but it failed to create a balanced SOC removal, which was
necessary for balanced BAZ development. Profile 2 was obtained 66 _
days after the injection scheme was adjusted. The distribution of SOC removal between the two BAZs was satisfactory, but the overall
efficiency deteriorated considerably. Profile 3 was obtained on day
116, which was 9 days after gas removal. The distribution of SOC
removal and the overall removal efficiency were satisfactory after _
gas removal.
33
12
FEED
INFLUENT
EFFLUENT
10
•
-
8
*...... l1 l ti *~
..J
C)
E
6
en
0
..... +it-
4
+,,*+
+ ......
0
Gas measured
ratio changed
Gas removed
.. • ... .W
.........
.
2
+
++00
,0
Nitrate injection
0"
0
00
+,..
•
W
••
•
•
0
0
20
40
60
80
100 120 140 160 180 200 220 240 260 280
TIME, days
Figure 3.18.
Soluble organic carbon concentrations In Column 2.
10
Nitrate
[:J
t
8
•
•
Day 21
Day 91
Day 116
Nitrate
6
<5
0
en
4
..J
C)
E
2
...
ol-r,.~:
0.0
25
50
7.510012515.017520022.5
0151ANCE, em
Figure 3.19. SOC profiles in Column 2.
34
The effluent SOC concentrations and the SOC profiles make it clear
that gas accumulation was a key factor controlling SOC removal.
Therefore, the gas trapped in the pore space was measured using the
technique described in Section 3.2.2.
Figure 3.20 shows the gas
distribution in Column 2 at day 179, which was 72 days after the gas
removal.
The total gas accumulated in the column was 14.4 mL,
which corresponded to 42% of the total pore volume after the first
injection port. As for Column 1, relatively little gas was trapped near
the first injection port, although it was greater than in Column 1. ~
Accumulation was very large 2.5-7.5 cm downstream of the first
injection. Gas distribution after the second injection was relatively
even, and the volume constituted about 30% of the pore volume.
Figure 3.21 demonstrates the gas accumulation photo graphic ally.
Again, no gas was accumulated before the first injection.
Two features are particularly important in Figure 3.20. First, a
considerable volume of gas was trapped in the BAZs (compare the
gas accumulations within 5 cm from the injections with Column 1 in
Figure 3.17). Since the BAZs were the location of SOC removal, gas
accumulation in the BAZs seemed to cause the relatively poor
performance of Column 2. Gas accumulation caused a reduction in
liquid detention time in the BAZs and a loss of substrate/biofilm
contact.
Second, much
gas was contained between the first and
second BAZs; the peak gas volume amounted to 87% of pore volume.
If such a "body" of gas were to move downstream toward the outlet,
the second BAZ would be severely affected as the gas peak passed
through it. In such a case, the removal efficiency would be expected
to deteriorate temporarily. It seems plausible that movement of gas
"bodies" may have caused the large effluent-quality fluctuations (see
Figure 3.18).
Figure 3.22 shows a typical correlation between SOC and nitrate
concentrations in Column 2. Although nitrate was the rate-limiting _
substrate after the first injection, it was in surplus after the secondinjection, making SOC the ratelimiting substrate.
Characteristics of the biofilm in Column 2 were determined. After
297 days of operation, the column was taken apart, and 10 glass
beads from each injection or samplingport section, including the
inlet and outlet, were taken out to measure the biofilm dry weight
and thickness. The procedures were described previously (Namkung,
35
1.0..----------------------,
0.9
o
0.8
0.7
~
a:
0.6
:E
0.5
0.4
w
::;)
...I
o
:>
GlASS BEADS
1----------------------;
0.3
0.2
0.1
0.0
.....
t- f'- r- r- ., .- r- r-. . ,~-
o0
2 5
5.0
75
10.0
125 15.0 175 20.0 22.5
DISTANCE, em
Figure 3.20. Gas accumulation and distribution in Column 2
(day 170, or 72 days after gas removal).
Figure 3.21.
N2-gas accumulation In Column 2.
36
10
EI
-
8
Z
0
i=
6
-J
•
C>
~
E
avg SOC
avg N03·N
Nitrate
C3:
a:
t-
z
w
0
z
4
0
2
0
Nitrate
0
0.0
2.5
~
5.0
7.5
10 0 12.5 15 0
17 5 20.0 22.5
DISTANCE, em
Figure 3.22. Average SOC and N03--N concentration profiles In
Column 2 (data from day 114 and day 116).
37
1985). The results are shown in Figure 3.23. The distribution of dry
cell mass per glass bead along the length of the column clearly
visualizes the concept of the BAZ (see Fig 3.23-a). The amount of cell
mass on the glass beads increased sharply at each injection port and
decreased slowly downstream. A photograph of BAZ distribution in
Column 2 is shown in Figure 3.24.
A striking resemblance is
The decrease in cell
observed between Figures 3.24 and 3.23-a.
mass with distance from the injection was slower than might be
expected from the SOC profile (see Figure 3.22). This suggests that asignificant portion of downstream cell mass was sheared biofilm that
was transported from upstream and deposited on the glass beads.
The small amount of biofilm at the inlet was grown on oxygen which
was not removed from the feed solution or which permeated into
liquid through the connection tubing.
Biofilm thickness (see Figure 3.23b) showed a similar tendency of
increase and decrease along the column, but its distribution
corresponded more closely to the SOC distribution than did the dry
cell mass.
The biofilmdensity distribution in Figure 3.23 c was
Its
almost a mirror image of the biofilm thickness distribution.
3
val ues at the first and second injection ports, 11.9 mgcell/cm and
8.8 mgcell/cm 3 , respectively, were not very far from other results
in glassbead columns (Namkung, 1985).
However, the density
greatly increased downstream from the injection ports.
It appears
that the increased density was caused by gas accumulation, which
partially dried the biofilms.
3.2.4 Headloss Through the BAZs
No measurable headloss was observed for either column.
A
calculated headloss by the KozenyCarmen equation (Freeze and
Cherry, 1978), assuming a 75 Jlmthick biofilm in a 2.5cmIong BAZ,
was 0.07 mm, which also was immeasurable.
3.2.5 Determination of Kinetic Parameters
Four kinetic parametersnamely, the maximum specific substrate
utilization rate (k), the halfmaximum rate concentration (K s ), the
cellyield coefficient (Y), and the celldecay coefficient (b)were determined in one batch reactor. To consider potential physiological
differences of cells grown at different locations along the column,
38
30
6
"CI
CI:II
G)
.CI
:::::
20
t
NOi
~N03
G)
(,,)
I
Q)
::l
10
80
E
60
:i
40
=.
20
..,
30
c
E
(,,)
.......
CD
20
E
..:
x
10
0.0
2.S
s.o
7.S
10.0
12.5
01 STfiNCE.
1S.0
17.S
20.0
22.5
em
Figure 3.23. Biofilm distribution in Column 2 after 297 days
of operation: distribution of dry cell mass (a), biofilm thickness ,Lf (b), and biofilm density, Xf (c).
39
Figure 3.24. BAZ distribution in Column 2 at day 296. The reacto!'.:
conditions were as defined in Table 3.2 and Table
3.3.
40
five batch reactors were run in parallel with five different inocula, as
indicated in Figure 3.25. Batch 11 and Batch 12 used cells from the
first and second injection sites in Column 2, where the cell activity
was high.
Both electron donor and acceptor were relatively
abundant compared to other locations in the column, but each site
had a different donor/acceptor ratio (see Figure 3.22).
Batch 1-2
used cells from the second sampling-port after the first injection,
where cells were under nitrate limitation. Batch 2-2 used cells from
the second sampling-port after the second injection, where cells wereunder SOC limitation. Batch 5 used cells from the fifth samplingport
after injection in Column 1, where most of the easilybiodegradable
SOC was used up and the cells were under extremely SOClimited
conditions.
Five 120mL vials, equipped with airtight rubber caps, were
used as batch reactors. An aliquot of 100 mL of oxygenfree feed
solution which had the same mineral composition as the BAZexperiment feed filled the vials. The initial concentrations of acetate
(14C labeled) and nitrate were doubled to ensure exponential growth.
The headspace was filled with Nz gas.
The batch reactors were
shaken continuously in a shaker.
Samples were taken out by a
syringe, replacing the liquid volume with Nz gas. 14C in filtered and
unfiltered samples was coun ted to determine SOC and cell
concentrations.
The changes in SOC and cell concentrations with time from the five
reactors are shown in Figure 3.26. SOC and cell concentrations are
denoted by S and X, respectively.
3.2.5.1
Determination of Y
The cell yield coefficient, Y, was determined from
Y = baX/I1S
(3.1)
during the exponentialgrowth phase.
Y was calculated at each
sampling time, using cumulative ba X and ba S, and an average value
was taken. For example, Batch 5 computations of Yare shown in
Figure 3.27, which illustrate the convergence of Y to 0.36 mg cell C/mg SOC. The Y values for Batches 11, 12, 12, and 22 were 0.37,
0.36, 0.375, and 0.36 mg cellC/mg SOC, respectively.
The data
clearly show that Y was not a function of sampling location.
41
f\03
...-
I
r-
f'D"3 Inj.
Inj.
...
0
@
o0
25
5 0
-
r-
7.5
I
r-
......
®
100
125
150
r-
(3)
175
200
...-
r
I
......
Inj.
r-
r-
,......
......
r-
CD
o0
25
5.0
75
225
em
COLUMN 2
[\03
..
100
125
COLUMN 1
150
175
20 0
..
22 5
em
Figure 3.25. Locations of harvested cells for kineticparameter
determination
42
Zr>
0
~
~
<
~
~
z
~
U
Z
0
u
2000
1800
1600
1400
1200
1000
800
600
400
e 200
eo 2000
1800
1600
1400
1200
1000
800
600
400
200
•
•••
•
•
•
\
...
• 00
•
•
J>i
•
••
,0• I
•
•
1:1
Dc
\
.~
•
20
30
.
D
•
10
• S2-2
o X2-2
c S5
X5
•
[]
40
•
It
D
:. • • • •
50
60
70
[]
.
~
80
10 20
30
•
•
0
•
0
•
...
\
••
00
'.
40 50 60 70 80
TIME,hr
Figure 3.26.
SOC utilization and cell growth in batch reactors.
SOC and <;ells are denoted by S and X, respectivel y.
43
0
0
rn
m
E
0..!.
'ii
0
m
E
C
I
w
>=
I
I
w
0
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
m
iii
Effi
&:I BEla iii
0.36
liI
I:J
0.2
0.1
0.0
m
a
lEI
0
10
20
30
40
50
Time, hr
Figure 3.27.
Determination of average cell yield in Batch 5.
44
3.2.5.2
Determination of k
The maximum specific substrate utilization
determined for the exponenti al growth p hase with
rate,
k,
was
(3.2)
k = Jlm/Y
In which Jlm is a maximum specific cell growth rate which, In turn,
was computed using
Jlm
= In(Xt!Xo)!t
(3.3)
The measured X values, shown in Figure 3.26, gave reasonably good
estimation for Jlm, but the experimental deviations in cellconcentration measurements caused some scatter in the 11m values.
Thus, a different cell-mass estimating method, utilizing SOC
concentrations, was devised to smooth out the J-lm values. The
method utilized the fact that removed SOC was incorporated into cell
mass with proportionality of Y. Thus,
Xt = Xo + Y(So - St)
(3.4)
The Xt estimated from SOC data did not alter the shape of the cellmass curve, but gave a smoother profile from which to compute J-lm.
A comparison of the methods, presented in Figure 3.28 for Batch 5
data, shows that the new computation method did not alter the X
curve.
Cell-concentration vs. time was plotted on a semi-log coordinate to
compute flm .
Figure 3.29 shows the semi-log plots for the
exponential growth phase. The k values were determined from the
slopes of straight lines and from Y for each batch run. The k values
were 2.00, 2.24, 2.16, 1.83, and 2.22 mg-SOC!mg-cell,day for Batches
11, 1-2, 12, 2-2, and 5, respectively. The variations among batches-='
did not reflect any significant physiological differences among the
cells from different sampling locations.
3.2.5.3
Determination of Ks
K s was determined from a 11 (specific cell growth rate) vs. S curve.
The fl values were calculated based on
45
800
E
Q.
()
z
0
600
~
a:
I
z
w
(J
•c
I•
l
r-
400
W
r0-
0
(J
1:1
•
.II.Q
[]
z
~
!!Ii
200
Xo+Y(SoS)
X5
c
~
•
!II
..J
w
(J
i
0
I••
0
!II
I
10
.
I
I
I
20
30
40
50
Time, hr
Figure 3.28. Comparison of estimated biomass (Xo + Y(So - S))
and measured biomass (X5) in Batch 5.
1000
r-----------,,---.....,
E
a
Q.
en
•a
:!:
.6
()
CIf
<t
0
c
iii
100
10
5
11
12
12
22
& . . .........IL..&............"""L_..a..I..._""""__L.._'.......
20
30
40
50
60
Time, hr
Figure 3.29. Exponential growth semi-log plot for five batch
reactors.
46
(3.5)
Jl = (dX/dt)(I/X)
The dX/dt was estimated as .t~/X
A typical Jl vs. S curve is shown
in Figure 3.30. K s was estimated by finding the S value when Jl was
half of Jlm. Because a small fraction of SOC (less than 2% of input C;
0.23-0.25 mg/L) was refractory in all batches, that fraction of SOC
was subtracted from S values in the K s determination. The Kg values
determined for Batches 11, 1-2, 12, 2-2, and 5 were 0.78, 1.20, 0.78,
0.82, and 0.22 mg SOC/L, respectively. Perhaps the relatively low K s
value in Batch 5 reflected a cell adaptation to low SOC. That Batch 12 showed a relatively high K s value may indicate that a high carbon
affinity was not needed under nitrate limitation.
3.2.5.4
Determination of b
The cell decay coefficient, b, was determined for the declining
phase with
dX/dt
= bX
(3.6)
Integrating Equation (3.6) yields
logX t = bt/2.303 + logXO
(3.7)
x
vs. t curves were constructed on a semilog coordinate for the
declining phase. A typical curve is shown in Figure 3.31, in which
Batch 12 data are plotted. A characteristic of the celldecay curves
was a continuous decrease of the decay rate; thus, the curves look
almost diphasic, as is emphasized by the two curves in Figure 3.31.
The decay rate ranged from 0.07 to 0.12 dayl for the first phase,
and it decreased by at least one order of magnitude in the second
phase. A similar phenomenon was observed, or discussed by other
workers (Chang 1985, Casolari 1988). One explanation of a diphasic
response is heterogeneity of the microbial population. A fraction of cells decays quickly, but the other fraction decays very slowly and
remain to establish the second phase. A gradual adaptation of cells
to nutrient limitation can be a second explanation.
3.3
Secondary Utilization of Halogenated Organic Compounds in BAZs
The goal during in situ bioreclamation is usually expressed in
terms of specific hazardous contaminants which often constitute only
47
0.08
E:I
E:I
0.06
E:I
0.0626
EI
EI
...
.c
y-
:i
0.02
0.00
...IQ.LL.......L..........LL..L........a........L.........................L.................1....I
o
200
Kg
400
600
800 1000 1200 1400
S,cpm
Figure 3.30.
Determination of K s from Jl
VS.
S curve (Batch 5).
1000
E
Q.
(,)
Z
0
~
r::c
to-
z
W
C,,)
Z
0
C,,)
..J
..J
W
C,,)
100
0
200
400
600
800
Time, hr
Figure 3.31. Cell decay in declining phase for Batch 12. Lines
represent the initial rapid and later slow decay
rates.
48
a small fraction of the soluble organic carbon (SOC).
These
compounds can be efficiently degraded as secondary substrates
(Namkung et aI., 1983) in the BAZ when the amount of accumulated
biomass and the compound/biomass contact time are sufficiently
large.
Since the hazardous compounds frequently have slower
biodegradation kinetics than a compound such as acetate, removals
of the specific secondary compounds can be less than for general SOC.
The experiments reported in this section investigate the removal ~
of several common halogenated solvents by the BAZs established
through utilization of acetate as the primary substrate and electron
acceptors injected along the flow path.
The relative rates of
degradation in the BAZs and the effect of contact time are
emphasized.
3.3.1 Experimental Methods
3.3.1.1
Selection of Halogenated Organic Compounds
Six halogenated aliphatics and three chlorinated aromatics were
tested.
The halogenated aliphatics contained three sub-groups:
carbon tetrachloride (CTC) and bromoform (BF), which were
substituted methanes; l,l,l-trichloroethane (l,l,l-TCA) and ethylene
dibromide
(EDB),
which
were
substituted ethanes;
and
trichloroethene (TCE) and tetrachloroethene (TeCE), which were
substituted ethenes.
The chlorinated aromatics included 1,2dichlorobenzene (1,2DCB), 1,3dichlorobenzene (1,3DCB), and 1,4dichlorobenzene (1,4DCB). According to previous work by Bouwer
and McCarty (1983) and Bouwer (1987), the substituted methanes
(CTC and BF) should be relatively rapidly degraded under
denitrifying conditions, the substituted ethanes (1,1 ,1TCA and EDB)
should be relatively slowly degraded, and the substituted ethene
(TeCE) and the dichlorobenzenes should be refractory. Under aerobic
conditions, on the other hand, the halogenated aliphatics were=refractory, while the dichlorobenzenes were degradable (Bouwer and
McCarty 1985, Bouwer 1987).
3.3.1.2
Experimental Setup
Column 1 and Column 2 were used initially to test the removal of
halogenated organic compounds by a secondary utilization
mechanism in primarysubstrategrown BAZs (Run la and Run 2 in
49
Table 3.4).
The six halogenated aliphatics and three DCBs were
dissolved in the feed reservoir at a concentration of about 100 Jlg/L
each. The feeding started on January 23, 1988 for the compounds
except TCE and BF, which started on February 18, 1988, and for EDB,
All the other experimental
which started on March 7, 1988.
conditions were maintained the same as the previous BAZ
experiments.
Table 3.4. Secondary Utilization Experiments for Halogenated OrganicCompound Removal
Flow Detention Time ExperiRun Electron Primary Secondary Velocity after Injection mental
Acceptor Substrate Substrate (cm/min)
(min)
Period
1a
N03
1b
lc
Id
2
3
"
"
"
"
H20Z
Acetate Xaliphatics
and DCBs
"
"
"
"
"
"
"
"
DCBs
"
0.1
0.04
0.01
0.1
0.1
0.1
50
125
500
50
50
50
1988
1/233/23
3/234/27
4/276/11
6/116/16
1/23323
4/276/16
The effect of liquid detention time was tested by reducing the
flow rate in Column 1 to 40% and then to 10% of the original flow
rate.
Thus, the liquid detention time in the column was increased by
2.5 and 10 times, respectively. When the 10%flowrate experiment
was over, the flow rate was increased back to the original flow rate
to check for any additional cell adaptation to the halogenated
compounds. The series of experiments is summarized as Runs 1b to 1d in Table 3.4.
An aerobic column which had hydrogen peroxide injected as
electron acceptor was operated (Run 3 in Table 3.4) to test for
aerobic removal of dichlorobenzenes in the BAZ.
The reactor characteristics were identical to the BAZexperiment column which
was shown in Table 3.2. A 0.3% H20Z solution was injected to an
50
injection port located at 5.0 em from the inlet (the first injection port
in Figure 3.9-Column 2). Acetate was fed as a primary substrate at a
concentration of about 7.5 mg-SOC/L as before.
The mineral
constituents are given in Table 3.5.
The column was inoculated with a 1% dilution of settled primary
effluent obtained from Urbana Sewage Treatment Plant.
The
inoculum was adopted to DeBs for three weeks before use.
The
operation started on April 27, 1988.
Table 3.5.
Composition of Feed Solution for Hydrogen Peroxide
Injection Column
Compounds
Concentration, mg/L
Acetate (CH3COO-)
KH2P04
K2HP04
Na2HP04
7.5 as SOC
34.0
21.75
17.7
3.4
11.0
27.5
0.15
NH40
Mg S04
CaCl2
Fe C13
3.3.1.3
Sampling and Analytical Methods
Samples were taken by syringe pump from the sampling ports at
Approximately 12 mL of
the rate equal to the input feeding rate.
liquid was collected from each sampling port.
Syringe-pump
sampling had two advantages: keeping the exact upstream flow rate
and preventing volatilization during sampling.
Exactly 10 mL of sample was extracted with 1 mL of dodecane or
pentane in a 15-mL hypo vial by vigorous shaking for 3 minutes.
The head space was minimized by adding distilled water, and the
vial was tightly sealed with a teflon-faced silicone rubber cap. After
waiting 15 minutes for phase seperation, 2 JlL of the separated
solvent phase was injected into a gas chromatograph equipped with
51
an electron-capture detector (Hewlett-Packard Model 5710 A).
A
60/80 Carbopack B, 0.1 % sp-1000 glass column was used for
halogenated aliphatics, and a 1% sp-1000 on 100/120 Supelcoport
was used for the DCB s. The same extraction and injection procedure
was applied for standard solutions used for calibration.
Dodecane
was a superior extractant for the halogenated aliphatics, while
pentane was superior for the dichlorobenzenes.
H202 was determined using a titanium-chloride method (Parker1928).
3.3.2 Results for Secondary Utilization Experiments
3.3.2.1
Removal of Halogenated Aliphatics in Denitrification Columns
Six halogenated aliphatic compounds were fed into the oneBAZ
denitrification column, in which the liquid detention time after
injection was varied from 50 to 500 minutes. The primary substrate
(acetate) concentrations in the feed reservoir, influent, and effluent
throughout the experiments are shown as SOC in Figure 3.32. During
Run 1a, in which the detention time was not changed from the one
used to establish the BAZ, the effluent SOC concentration was low and
steady.
The effluent SOC was lower than 0.2 mg/L, which
corresponded to more than 97% removal. A representative sampling
for halogenatedcompound determination was made near the end of
this period, as indicated by arrow in Figure 3.32.
In Run 1b, in which the detention time was increased by 2.5
times, the effluent SOC increased slightly and the concentrations
were between 0.20.3 mg/L for most measurements.
Because a
lower effluent SOC concentration was expected with increased
detention time, 14C labelling of the feed acetate was removed for one
week to help elucidate the phenomenon. In spite of no input of 14C
in the feed, the effluent 14C level was not much changed (0.20 mg=.
SOCIL, on average).
When the 14Clabeling was restarted, the
average effluent SOC was 0.23 mg/L. These results show that the
portion of effluent SOC contributed by feed acetate was only 0.03
mg/L, while the majority (87%) came from another source, namely,
The accumulated and labeled biomass previously labeled biofilm.
was responsible for release of biomassassociated soluble microbial
products (N amkung and Rittmann, 1986).
With an increased
52
14 . . . . . - - - - - - - - - - - - - - - - - - - - - - ,
12
Run 1a
Run 1c
Run 1b
Run 1d
10
..J
"""'-
C)
8
E
cS
o
en
a
o
6
o
4
INFLUENT
EFFLUENT
14 C removed from feed
2
~I- r,_ ,_ r~,._; . ,I'_ r~;=_rI Wi.: _,~ .= _. a , . ,- f. _ r- r~-. o
00
•
FEB)
•
II
250 270 290 310 330 350 370 390 410 430 450
TIME, days
Figure 3.32. SOC concentrations during Run 1 in a denitrifying
column. Arrows indicate samplings for halogenated
organic compound determination.
53
detention time, the fed acetate was consumed only in the upstream
portion of the BAZ, and the biofilm in the downstream portion of the
BAZ was undergoing a starving condition.
Therefore, the biofilm
distribution in the column was not at a steady state.
Run Ie was initiated by increasing the liquid detention time by 10
times from the original. The effluent SOC increased even more at
first (about 0.5 mg/L), and then gradually decreased to about 0.3
mg/L. Again, this increase was caused by soluble microbial product~
formation. The influent SOC dropped severely due to the increased
retention time of liquid in the connection tubing between the
reservoir and the column. Oxygen diffusion through the tubing was
responsible for stimulating aerobic degradation of feed acetate
before it reached the denitrifying BAZ. This drop of the influent SOC
concentration was substantially eliminated by increasing the flow
rate in the tubing and by diverting most of the flow to waste. As
shown by the arrows for Run Ic in Figure 3.32, samplings made after
the flow diversion had an improved SOC concentration entering the
column.
In Run ld, the detention time was decreased back to the level of
Run la by increasing the flow rate through the column. The effluent
quality deteriorated initially after the change, then recovered
quickly. Due to a time constraint, the sampling was made before the
column reached a new steady state.
The results for the halogenated aliphatics, as well as SOC, are
shown in Figures 3.33 to 3.37 for Runs Ia to Id, respectively.
Sample concentrations are normalized to the measured concentration
at the sampling port just upflow of the nitrate injection, and the
initial concentrations of each compound are given in the figures.
Since the concentration at the nitrate injection port could not be
measured, it was assumed to be the same as the concentration of the
immediate upstream port.
In Figure 3.33, which is for the original (50-minutes) detention
time, out of six halogenated aliphatic compounds, only eTC showed
significant removal (28%) through the BAZ. Losses in all the other
compounds were not greater than usual experimental error.
54
120 - - - - - - - - - - - - - - - - - .
z
100
~
80
o
a:
I-
• 111TCA
• CTC
+ TCE
Z
W
(J
Z
o(J
w
>
~
w
a:
60
.
Ern
)(
SF
[] TeCE
40
A
s::x::;
20
OL..----L.--'---L-----L_....L-_L------L_....:I:...--I
0.0 2 5 5.0
7.5 10 0 12 5 15.0 17.5 20 022.5
DISTANCE, em
Figure 3.33. Profiles of halogenated aliphatic compounds in
a denitrifying column at 50-min. detention
time after nitrate injection (Run la). The
initial concentrations for the compounds were
(in Jlg/I) 84 for l,l,l-TCA, 81 for CTC, 95 for
TeE, 87 for EDB, 106 for BF, 79 for TeCE, and
6,600 for SOC.
55
120
it
Nitrate
100
z
0
i=
«
a:
80
!zw
0
z
60
0
~
111TCA
CTC
+
.
TCE
EDB
x SF
0
w
>
•
•
40
EJ
TeGE
A
SJC
...J
W
a:
20
oL.- L._ - L . l- .J : -=: jr : ~_.I
0.0255.07510012515.017.520022.5
DISTANCE, em
Figure 3.34. Profiles of halogenated aliphatic compounds in
a denitrifying column at 125-min. detention
time after nitrate injection (Run 1b). The
initial concentrations for the compounds were
(in Ilg/l) 99 for 1,1,1-TCA, 69 for CTC, 62 for
TCE, 69 for EDB, 57 for BF, 55 for TeCE, and
5,100 for SOC.
56
120
~
100
Nitrate
z
0
~
a:
80
•
to-
•
z
w
0
z
0
+
60
0
w
>
~
40
111TCA
eTe
TCE
EDI3
)(
SF
[]
TeGE
A
ax
....I
w
a:
20
0
0.0
2 5
5 0
75 10.012.515.017.5200225
DISTANCE, em
Figure 3.35. Profiles of halogenated aliphatic compounds in
a denitrifying column at 500min. detention
time after nitrate injection (Run 1c, day 406).
The initial concentrations for the compounds
were (in Jlg/I) 112 for 1,1,1TCA, 53 for CTC,
88 for TCE, 45 for EDB, 54 for BF, 50 for TeCE,
and 4,900 for SOC.
57
120
z
~
100
Nitrate
0
~
a:
•
80
•+
l-
zw
0
z
0
60
0
w
>
~
..J
40
111TCA
CTC
TeE
EIl3
x
SF
[]
TeCE
.&
gx
W
a:
20
0
0.0 2.5
5 0
7.5 10012515.017520022.5
DISTANCE, em
Figure 3.36. Profiles of halogenated aliphatic compounds in
a denitrifying column at 500-min. detention
time after nitrate injection (Run lc, day 415).
The initial concentrations for the compounds
were (in J..lg/l) 96 for l,l,l-TCA, 65 for CTC, 91
for TCE, 38 for EDB, 49 for BF, 84 for TeCE, and
3,200 for SOC.
58
120
~
100
Nitrate
Z
~
0
a:
80
0
60
•
•+
!zw
z
0
)(
c
0
w
>
~
...I
A
40
111TCA
eTC
TCE
EC8
BF
TaCE
SJC
W
a:
20
0
0.0 2 5 5.0
75 10.012515.017.520.022.5
DISTANCE, em
Figure 3.37. Profiles of halogenated aliphatic compounds In
a denitrifying column at 50-min. detention
time after nitrate injection (Run Id). The
initial concentrations for the compounds were
(in J.1g/l) 94 for l,l,l-TCA, 62 for CTC, 87 for
TeE, 45 for EDB, 59 for BF, 66 for TeCE, and
6,300 for SOC.
59
The results of Run 1b, which had a detention time 2.5 times longer
than Run 1a, are shown in Figure 3.34. The removal of CTC became
more significant (around 60%), and the other compounds, except
1,1,I-TCA, had 15%-20% removal. 1,1,I-TCA had the lowest removal
efficiency, about 10%. The concentrations of TCE and TeCE were
shown to be higher at 20-cm location than at 15-17.5 em.
The
percentage removals for these compounds dropped from 19 to 11 for
TCE, and 21 to 15 for TeCE. Since the percentage removal of the
primary substrate (acetate) was very stable (the coefficient ofvariation--l00% (standard deviation/mean)--was in the range of 24%; data not shown), the variation of the percentage removal of these
secondary substrates should be attributed to analytical error for
these compounds.
Figure 3.35 shows the results of the day-406 sampling in Run lc,
which had 10 times longer detention time compared to Run 1a, or 4
CTC was removed almost
times longer than that in Run 1b.
completely (94%), and its concentration decreased from 53 Jlg/L to 3
Jlg/L. Significant removals of TeCE (50%), BF (30%), and TCE (15%30%) occurred in response to the increased contact time between the
compounds and the BAZ. EDB and 1,1,I-TCA, however, showed only
comparable percentage removal (about 20% and 10%, respectively) to
the previous detention time in Run 1b. Figure 3.36 shows the results
in a subsequent sampling made on day 415.
Similar or slightly
increased removals were observed for TeCE (50%), TCE (30%), EDB
(20%), and 1,1,1-TCA (20%), but, less removals were made for CTC
(70%) and BF (20%). Even though there were some fluctuations in
removal efficiency between the two samplings, overall trends of
removal remained consistent, and the average removals for most
compounds in this extended detention time were substantially higher
than the removals in Run 1b. Another important finding in Run Ie
was that all of the six halogenated aliphatics tested were degradable
under denitrification conditions.
Figure 3.37 shows the results obtained from Run Id, which had
the same detention time as Run 1a.
The overall trends for the
halogenated compounds were very similar to those in Figure 3.33.
Thus, it became clear that the increased removal in Runs 1band 1c were not the effect of cell adaptation, but occurred because of the
increased detention time.
60
The results in these experiments, especially the results shown in
Figures 3.35 and 3.36, are partially consistent with, but still contain
substantial contradictions to those of Bouwer (1987), who showed
that CTC had the fastest removal, followed by BF, l,l,l-TCA, and EDB,
but TeCE was refractory. In this experiment, BF showed relatively
slow removal, but TeCE was relatively rapidly degraded in the BAZ.
Trace amounts of chloroform were produced after nitrate injection in
this experiment, which demonstrated a reductive dehalogenation of
CTC. Because no radioactive tracer study was performed, it was not~
demonstrated whether or not the removed portion of the
halogenated compounds was converted to C02. Bouwer and McCarty
(1983) demonstrated, however, that CTC was converted to C02 and
cell mass in a denitrifying biofilm column.
Run 2 was conducted under identical operating conditions to Run
la, except that this column had two BAZs, while Run I a was
performed in a one-BAZ column.
The results were qualitatively
similar to Run la, but showed less removal of CTC (see Figure 3.38).
This was, probably, due to the gas accumulation, as explained in
Section 3.2.3.
3.3.2.2
Removal of Dichlorobenzenes In Denitrification Columns
Profiles of 1,2-DCB and 1,3-DCB for the day-415 sampling from
Run Ie are shown in Figure 3.39. 1,4-DCB could not be determined
due to a technical problem: the peak for this compound overlapped
with that for BF in the Supelcoport column which was used for DeB
determination (BF was determined by a Carbopack column in which
1,4-DCB did not overlap). Figure 3.39 shows that 1,2-DCB decreased
from 35 Jlg/L to 23 Jlg/L, having about 30% removal across the BAZ.
A similar removal efficiency was observed in another sampling
which was made on day 406 (data not shown). The removal of 1,3DCB was slightly better than that of 1,2DCB in Figure 3.39, but it was
slightly less in the other sampling.
As with the halogenated aliphatics, detention time was a critical
parameter in the extent of DCB removal. At the reduced detention
time used in Runs 1a and Id, no significant removal was observed.
The result of Run Id is shown in Figure 3.40.
61
140
Nitrate
120
z
0
100
!zw
80
~
a:
•
111TCA
•+
TCE
60
x
SF
>
c
TeCE
«
..J
40
A
sx
0
z
0
0
w
i=
eTC
Ern
w
a:
20
0
o0
2 5
5.0
7.5 10 0 12 5 15 0 17.5 20 022 5
DISTANCE, em
Figure 3.38. Profiles of halogenated aliphatic compounds in
a denitrifying column at 50-min. detention
time after nitrate injection (Run 2). The initial
concentrations for the compounds were (in
flg/l) 158 for 1,1,1-TCA, 158 for CTC, 152 for
TCE, 130 for EDB, 119 for BF, 118 for TeCE, and
6,700 for SOC.
62
120
~
z
0
~
a:
....
z
100
Nitrate
80
w
0
Z
EJ
0
60
w
>
40
•..
0
~
..I
w
a:
20
0
0.0 2 5
13DCB6/10
12DCB6/10
SOC6/10
•
5.0
7.5 100 12515.017.520.0225
DISTANCE, em
Figure 3.39. Profiles of 1,2- and 1,3-DCB in a denitrifying
column at 500-minute detention time after
nitrate injection (Run Ic, day 415). Influent
concentrations were (in Jlg/l) 42 for I,2-DCB,
29 for 1,3-DCB, and 7,300 for SOC.
63
120 , . . . - . - - - - - - - - - - - - - - - . ,
z
~
o
a::
I-
z
~
Nitrate
100
80
W
(J
o(J
60
>
w
40
...J
20
Z
fi
W
EI
41/6~BCD31
•
..
SOC~6/14
12DCB~6/4
a::
OL.-----L.-......L-_.L...---.L._......L.._..&....-----L_.....L..---'
0.0255.07.510012515.0175200225
DISTANCE, em
Figure 3.40. Profiles of 1,2- and 1,3-DCB in a denitrifying
column at 50-minute detention time after
nitrate injection (Run Id). Influent
concentrations were (in flg/1) 35 for 1,2-DeB,
23 for 1,3-DCB, and 7,200 for SOC.
64
The observation of DCB removal in a denitrifying column is very
important, because these compounds were thought to be biologically
persistent under anoxic conditions (Bouwer and McCarty 1983, Kuhn
et al. 1985, Bouwer 1987).
To investigate any non-biological
reactions which might have been responsible for the removal, two
potential alternative pathways were examined: sorption and
volatilization.
DCBs have moderately high octanol-water partltIon coefficients, with a typical log K ow value around 3.4 (Miller et aI., 1985).
Therefore, DCBs could adsorb onto or absorb into hydrophobic parts
in cells produced from the primary substrate.
This was, however,
not the reason for the removal in this experiment. First, the liquid
samples were not filtered before DCB extractions; thus, any DCBs
Second and
sorbed to effluent cells would have been measured.
more important, the DCBs were fed continuously for the duration of
the test. Any sorption capacity of the cells in the column would have
been saturated long before samples were taken.
Volatilization of DCBs into the nitrogen gas, produced during the
denitrification reaction, could take place. But, the gas production rate
was trivial compared to the liquid flow rate (0.4% by volume) and
would not explain the substantial removal of DCBs.
Moreover,
Section 3.3.2.1 showed that 1,1,ITCA in Run lc had a much lower
percentage removal (15% on average of two samplings) than for the
DCBs, even though it had one order of magnitude higher Henry's law
constant (Lyman et aI., 1982). Lack of volatilization loss of 1,1,1TCA
supported the insignificance of volatilization for DCBs.
As there was no significant alternative pathway for DeB removal
in these experiments, the removal can be attributed to
biodegradation.
Further research is necessary to prove this
rigorously.
3.3.2.3
Removal of Dichlorobenzenes with HydrogenPeroxide
Injection
The feed, influent, and effluent concentrations of SOC in the
column injected with H202 (Run 3 in Table 3.4) are shown in Figure
3.41. The effluent SOC concentration decreased quickly and reached
a relatively stable concentration in a week. A slightly lower amount
65
14
12 -
.
10 -
,
I
.
m
E
cJ
0
8- EJ 1:J1El1EI[lJ
,
6-
1::1
.000 0 00
en
o
o
o
4
2
0
0
m Feed
o Influent
• Effluent
lEI 1::1 lEI
o
o
••
• • ••
•••
I
I
10
20
•
•
'V•
'!I
• ••
•
•
I '
I
I
40
50
30
•
60
TIME, days
Figure 3.41. SOC concentrations in a Hz02 injection column
(Run 3). Arrows indicate samplings for DeB
determinations.
66
of HZ 0 z than required for SOC oxidation was fed in order to avoid
toxicity from HZOZ. Thus, the reaction was under HzOZ limitation, and
the SOC degradation was not complete.
The hydrogen peroxide
disappeared faster than expected from the stoichiometry of SOC
oxidation. It is likely that the enzyme catalase degraded Hz02 to 02,
in order to reduce H20 2 toxicity, at a more rapid rate than the 02
was utilized.
DCB sampling was made on days 44 and 46, and results are shownin Figures 3.42 and 3.43, respectively.
Figure 3.42 shows
comparatively lower SOC removal, due to low H20 Z injection on that
day. In both figures, the three DCBs were removed with the same
pattern. 1,4DCB showed the lowest removal (10%) among the three
compounds, while the other two
had comparable degrees of
removal (20%30%).
Most of the DCB removal took place immediately after the HzO Z
injection. Because the SOC was removed beyond the first sampling
port, while DCB removal occurred before that port, some role of HzOz
in the removal of DCBs is implicated.
Hz Z was not directly
responsible for this removal, since a series of batch experiments (see
Figure 3.44) showed that a HzOz addition into a nonbiological reactor
did not cause any greater removal of DCBs than occurred in a control.
Although hydrogen peroxide did not play a direct role in DCB
removal, it is still possible that the biologically mediated hydrogen
peroxide decay to oxygen affected DCB removal, since hydrogen
peroxide was degraded before the first sampling port. Therefore, it
is not clear whether or not the normal aerobic degradation of DCBs,
as seen by other authors (Bouwer and McCarty 1985, Bouwer 1987,
Kuhn et al. 1985), was solely responsible for the DCB removals
observed in the HzOZ injection column.
°
The results in this experiment differ considerably from those of
Bouwer (1987), who employed a biofilm column with a 10min='
detention time
and aerobic conditions from contact with ambient
oxygen. Bouwer (1987) showed that 10 Jlg/L of 1,2DCB and 1,4DCB
were removed more than 97%, and the adaptation periods were less
than 3 weeks. 1,3DCB was removed 71 %, but the adaptation period
required was 500 days. The present study showed relatively lower
percentage removals, but 1,3DCB was removed with a much shorter
adaptation period.
These differences support the fact that the
67
120
z
~
o
a:
r-------------------.
100
!Zw
80
z
8w
60
)(
•
>
40
A
w
~
20
o
[:J
13DCB6/1 a
14DCB6/1 a
12DCB6/10
SOC6/10
a:
OL----.L.-....a...----II.....---'--..L-----I-..........-"""""------'
0.0 2 5
5.0
7.5 10.0 12.5 15.0 17.5 20.022 5
DISTANCE, em
Figure 3.42. Profiles of DCBs in an H202-injection column at
50-minutes detention time (day 44 from Run
3). The initial concentrations were (in Jlg/I) 44
for 1,2-DCB, 43 for 1,3-DCB, 42 for 1,4-DCB, and
6,650 for SOC.
68
120 - - - - - - - - - - - - - - - - ,
z
o
100
to-
80
~a:
Z
EI
W
o
z
60
>
40
..J
20
8w
~
W
x
•
A
13DCB6/12
14DCB6/12
12DCB6/12
SOC6/12
a:
O'........ . l I . . . . . . . a . I ......... "
oa
2.5
5.0
7.5 10.0 12.515.0 17.5 20 022.5
DISTANCE, em
Figure 3.43. Profiles of DeBs in an H202injection column at
50minutes detention time (day 46 from Run
3). The initial concentrations were (in j.lg/l) 33
for 1,2DCB, 35 for 1,3DCB, 46 for 1,4DCB, and
6,250 for SOC.
69
-
..I
m
30
:::1.
Z
)(
12DCBcontroi
0
+ 12DCBH202
a::
"
~
I
z
w
•
•
20
[]
0
z
0
13DCBcontroi
13DCBH202
14DCBcontroi
14DCBH202
0
10
a
10
20
30
40
50
60
TIME, min
Figure 3.44. The effects of HZOZ on DeB removals in nonbiological batch reactors. For this
experiment, 15mL extraction vials which
were sealed with teflonfaced silicone septa
were used as the batch reactors. A series of
vials which contained 20 mg/L of HZ 02, an
equivalent concentration that the column
reactor actually received, were filled with
feed solution without leaving headspace and
extracted at times 10, 30, and 50 minutes
after initiation. The time of extraction was
equivalent to the real detention times from
the Hz 0 Z injection port to the sampling ports
in Figures 3.43 and 3.44. A series of control
vials which did not contain H20 2 were
prepared and extracted same way.
70
removal mechanism in this study might have differed from that in
the aerobic column which Bouwer (1987) used.
Future study is
necessary to thoroughly evaluate the differences.
71
CHAPTER 4. COMPUTER MODELING
4.1
One-Dimensional Solute Transport Model
The governing mass balance on a biodegradable compound for
steady-state flow through a homogeneous, one-dimensional column,
such as that described in Chapter 2, has the form
(4.1 )
where S is the dissolved substrate concentration, E is the porosity, DH
is the hydrodynamic dispersion coefficient, v is the specific discharge
(superficial flow velocity), a is the specific surface area of the bed
particles, J is the substrate flux into the biofilm, and Qs is the
substrate source term due to lateral input through the injection
ports. Equation (4.1) can seldom be solved analytically to give S as a
function of t and x.
Hence numerical solution using a digital
computer is necessary.
For numerical solution in general, equation (4.1) is discretized in
time and space, and a finite difference approximation for both kinds
of derivatives are straight forward. The difference equations can be
solved at successive time steps until a given stopping point, as
defined by a particular problem.
Equation
(4.1) can be simplified for steady-state by setting the
time derivative to zero. The resulting equation,
(4.2)
is an ordinary differential equation, as opposed to equation (4.1),
which is a partial differential equation.
To discretize steady-state equation (4.2), no difference
approximation is needed for time.
The discretized steady-state
equation was chosen for two reasons.
First, it approximately
describes realistic scenarios of enhanced in situ bioreclamation: _
namely, the steady-state input of a limiting factor (the electron
acceptor here) into an aquifer containing a fairly constant pollutant
source. Second, the numerical solution of equation (4.2) provided an
73
opportunity to develop new, highly efficient solution techniques for
strongly non-linear ordinary differential equations.
The numerical
approach, based upon quasilinearization, also can be applied to other
groundwater situations involving nonlinear reaction terms.
The
numerical quasilinearization with finite differences is presented in
detail in Section 4.3.
4.2. Biofilm Phenomena and Kinetics
Because of the high specific surface area in an aquifer, almost all
of the biological activity is associated with the solids as biofilms or
microcolonies. This research utilizes the concept of a biofilm, which
is generally defined as a layer-like aggregation of microorganisms
attached to a solid surface. Modeling of biofilm kinetics has been
achieved by considering an ideal biofilm that is locally homogeneous
and planar.
The processes affecting substrates and biomass are
represented by a set of differential and algebraic equations which
must be satisfied simultaneously.
Figure 4.1 shows the conceptual basis of the biofilm model. The
biofilm, having thickness Lr, is composed of an idealized
homogeneous matrix of cells at a density of Xr.
The substrate
concentration changes nonlinearly across the biofilm thickness from
S s to Swat the attachment surface.
.Jx
z
Bulk
Liquid
Diffusion
Layer
S
L
Figure 4.1.
w
r
Substrate
Concentration
Conceptual basis of the biofilm model (after
Rittmann and McCarty, 1980a).
74
(S)
The substrate is transported from the bulk liquid across an
idealized layer, L, through which all the resistance to mass transfer
lies. This layer is referred to as the diffusion layer and represents
the resistance to mass transfer from the liquid to the biofilm. There
have been several published correlations for relating L to reactors
and substrate variables (Rittmann, 1982a).
The substrate
concentration varies linearly across this layer according to Fick's first
law
(4.3)
Substrate utilization within the biofilm is assumed to follow a
Monod relationship and is defined by the following equation
rut
-k Xr Sr
+ Sr
(4.4)
= Ks
where rut is the rate of substrate uptake per unit biofilm volume, k
is the maximum specific rate of substrate utilization, and K s is the
half-maximum-rate substrate concentration.
Equation (4.4)
represents the accumulation of substrate in the biomass due to
utilization, and it has a negative sense.
Molecular diffusion within the biofilm
second law and is represented by
rdirr
=
a2 Sf
Dr az 2
IS
described by Fick's
(4.5)
where rdiff is the rate of substrate accumulation due to diffusion, and
D f represents the molecular diffusion coefficient of the substrate
within the biofilm.
The total time rate of change of substrate within
the biofilm can be written as
k Xr Sf
K s + Sf
(4.6)
If the substrate profile is at steady-state,
75
equation (4.7) simplifies to
-
k Xr Sr
K s+ Sr
(4.7)
and is subject to the following boundary conditions:
1.
No substrate flux into the solid attachment surface
aSr _ 0
az -
2.
@ z =0 and t
~
(4.8)
0
Continuity of flux at the biofilm/diffusion layer interface
Sf = Ss
@
z = Lr and
t ~
0
(4.9)
There are three basic substrate concentration profiles that are
generally used to categorize biofilms. A deep biofilm is defined as
one in which the substrate concentration drops to zero at some point
within the biofilm.
A deep biofilm has the maximum substrate flux
for a given Ss value. The other extreme case is that of a fu 11 y
penetrated biofilm, or one in which the substrate concentration
equals Ss at all points within the biofilm. For all cases between fullypenetrated and deep, a biofilm is defined as shallow.
The remaining major aspect of the biofilm model is coupling mass
transport from the bulk liquid to the surface of the biofilm, equation
(4.3), with substrate utilization and diffusion in the biofilm, equation
(4.7). . The coupling and solution of the governing equations complete
the basics of biofilm modeling.
A steadystate biofilm is defined as one where all the time
derivatives are set to zero (Saez and Rittmann, 1988).
A mass
balance on the biofilm requires that cell losses are balanced by cell
growth. The growth rate of a biofilm per unit surface area is defined =.
by YJ, where Y is the cell yield coefficient and J is the substrate flux
into the biofilm.
The biofilm loss rate per unit surface area is
defined as LfX rbT, where Lr is the biofilm thickness, Xf is the
biofilm density, and bT is the overall firstorder loss coefficient. The
overall firstorder loss coefficient is comprised of two components, as
expressed in the following equation
76
bT = b + bs
(4.10)
where b is the cell maintenance and decay coefficient and b s is the
shear loss coefficient. The shear loss is a result of shear forces of the
flow passing by the biofilm and stripping pieces of it away with the
flow.
Rittmann (1982b) presented a simple model in which an
estimate of the shear loss coefficient can be made with knowledge of
the biofilm thickness and reactor parameters.
The equations
developed by Rittmann (1982b) are used in this research to estimatethe shearloss coefficient.
A final concept of steadystate biofilm modeling is a definition of
a threshold concentration below which no steadystate biofilm can
occur.
The threshold concentration is defined as Sm in, and
concentrations below it give a biofilm that is continuously losing
mass, which violates the steadystate assumption.
Smin is computed
as
K s bT
Smin = Y k bT
(4.11)
Steadystatebiofilm modeling involves the solutions of equations
(4.3) and (4.7), as well as the mass balance on biofilm mass.
Originally, complicated and time consuming numerical techniques
were used to solve the coupled equations.
The resulting solution
determined the flux, J, into the biofilm for a given set of kinetic
parameters (K s , k, bT, Y, Xr, L, D, and Dr) and a bulk substrate
concentration, Sb.
Such a solution technique encompasses all the
biofilm profiles Le. deep, shallow, and fullypenetrating.
Repetitious use of these sophisticated numerical models for
solving the governing equations
for steadystate flux determination
is not practical when the goal is to model a large system, such as for
aquifer bioreclamation.
As a result of this, several researchers=.
developed pseudoanalytical techniques that fit the numerical results
with algebraic equations (Rittmann and McCarty, 1980a; 1981; Saez
and Rittmann, 1988)
The solution presented by Saez and Rittmann (1988) is the most recent and accurate method available among the several in existence.
More accurate over a large range of
substrate concentration than
77
previous methods, the new pseudo-analytical technique is the best
option for steady-state biofilm modeling.
A short summary of the
model's structure is presented here for clarity; the reader is referred
to the original manuscript for complete details.
The first premise behind the pseudo-analytical technique is that
the actual flux to a steady-state biofilm is a fraction, f, of the flux
into a deep biofilm. This is represented mathematically by
J
=f
(4.12)
Jdeep
where Jdeep is the flux into a deep biofilm exposed to the same
concentration, Sb.
The second premise is that the solution is presented most
efficiently with dimensionless parameters.
The efficiency comes
about because the many dimensional parameters can be lumped
together to form a smaller number of dimensionless parameters,
which are indicated by an asterisk.
The key dimensionless
parameters are
*
bT
J* ==
~ _J
(KskXfDf) 1/2
S min = Yk - bT
K * == (D) [
L
Ks
:t
(kXfDf)J
1/2
For example, equation (4.12) can be rewritten In the dimensionless
regime as
J* = f J* deep
(4.13)
Saez and Rittmann (1988) found the value of f could be expressed
algebraically as
78
f
Ss*
. * 1)~]
= tanh[ex (s mIn
(4.14 )
where ex and f3 are defined by
ex = 1.5739 + 0.32075 ( - log Smin*)0.15213
a = 1.5739
~
= 0.5014
+ 0.37149 ( log Smin*)0.31344
+ 0.01985 ( -log Smin *)0.19476
f3 = 0.5014 + 0.02726 ( log Smin * )0.52256
. * -< 1 (4.15)
10- 4 <
- S mIn
1 ~ Smin * ~
103 (4.16)
. *<
10- 4 <
- S mIn
- 1
1~
Smin * ~
(4.17)
103 (4.18 )
Because Ss *, the dimensionless substrate concentration at the
biofilm surface, is not known apriori, it must be computed
iteratively, using a Newtons root finding technique, from
( 4.19)
Once the appropriate Ss * value is converged upon, the flux
calculated by Fick's first law
1S
(4.20)
which is easily transferred into the dimensional flux
definition of the non-dimensional flux
J
= J*
[KskX rDr] 1/2
using the
(4.21 )
4.3 The Quasilinearization Technique
The objective of this section is to present a summary of the
quasilinearization technique coupled with the finite-difference
solution technique.
The equation to be solved is the steady-state
one-dimensional transport equation with a biological reaction term,
which was presented in a previous section as equation (4.2).
79
Traditional approaches (Rittmann, 1982a) involve solving the
transient problem, equation (4.1), until steady-state is achieved.
Here the time derivative, in addition to the spatial derivatives, is
approximated using finite differences.
The time dimension adds
many more computations than are necessary if the steady-state
solution can be obtained directly.
Thus, the traditional approaches
are computationally inefficient and are not feasible for extension to
more complex problems. Here, a technique which bypasses all of the
intermediate calculations and solves directly for the steady-state isdeveloped. Handling the non-linearity of the reaction rate term (J) is
the focus of computational strategy, because the biofilm reaction rate
term approaches infinite reaction order at (S) values close to Smin.
The problems of non-linearity can be overcome by
quasilinearization (Lee, 1968).
The quasilinearization process
involves the use of a first-order Taylor's series approximation for the
non-linear substrate flux term. If sm is assumed to be the known
substrate concentration at an iteration level m, then the substrate
flux at the next iteration level can be approximated as
(4.22)
Equation (4.22) can be substituted into equation (4.2) to yield a
linear ordinary differential equation for sm+ 1,
d2
d
dJ
dJ
DH dx 2 sm+1 v dx sm+1 adS sm+1 = -Qs + aJ(sm) adS sm (4.23)
Finite differences are used to approximate the spatial derivatives
and yield a system of simultaneous linear algebraic equations which
can be solved for sm+ 1. In one dimension, the system of equations
has a tridiagonal matrix structure. The spatial domain is divided into
n grid-points (i = 1, n), where n is defined as
n=
LT
(4.24 )
~x
where LT is the total length of the column and ~x
is the grid spacing. A three-point finite difference approximation was used for the
dispersion term, which takes the form
80
Si+l - Si + Si-l
( 4.25)
8x 2
where i is the grid point at which the term was evaluated.
advective term is approximated by a central difference
dS
Si+l - Si-l
dx ==
28X
The
(4.26)
Substitution of equations (4.25) and (4.26) into equation (4.23) to
yield the discrete finite difference equation for a grid point i is given
by
(4.27)
where
DR v
e= 8x 2 +28X
_DR _v_.
c 8X 2 28x '
The discrete equations are subject to two appropriate boundary
conditions for the numerical method to be implemented.
The
influent condition (at x = 0) is
.
vS m
= vS
dS
- DHdx
(4.28)
in which sin is the substrate concentration at the inlet of the reactor
Rittmann (1982a). The boundary condition at the effluent end (x =
L) is
dS _ 0
dx -
(4.29)
When equation (4.27) is written for
appropriate boundary conditions are
equations for sm + 1 results.
each grid point and the
imposed, the system of
The key to implementing the numerical technique with
quasilinearization is an efficient and accurate evaluation of the dJ/dS
The new pseudo-analytic equations developed by Saez and
term.
81
Rittmann (1988) can be differentiated to yield an expression for
dJ/dS for the entire range of concentrations. Once the Ss value has
been converged upon, the expression is given by
dJ*
dSb *
= K*
dS *
( I-dS : * )
( 4.30)
where
dSb*
1
dS * = 1 + K*
s
~a
Ss*
S *min(S *min
{"-J 2[Ss *-In(1 +Ss *)]
1)~-1]
+
sech
2[ a(S Ss*
*min
Ss*
tanh[ aCS*min - 1)~]
~2(
A
- l)p·
Ss*
. Cl + Ss*)
Ss* - In(1 + S8*) )
} ( 4.3 1)
One interesting feature is the behavior of dJ/dS as the substrate
concentration approaches Smin, where the reaction order approaches
infinity. The expression dJ/dS can be simplified to a finite value at
Smin: the result is simply
dJ I
dS Smin
=
D
L
( 4.32)
Equation (4.32) can be understood on
biofilm must be fully penetrated when S
only be external mass transfer resistance
the flux is a first-order function of S,
constant.
intuitive terms. Since the
is near Smin, there should
to control the flux.
Thus,
with D/L as the reaction
The method of quasilinearization was implemented and proved to
be very accurate and efficient compared to previous methods. Figure
4.2 demonstrates the accuracy of the new technique, when no Qs
terms are included, by comparing the results to the previous method of time stepping.
Table 4.1 shows the parameters used for the
comparison of techniques. The new technique also was tested with
several other sets of kinetic and reactor parameters and had
similarly good accuracy and convergence. A comparison of the
computational efficiency for the techniques is demonstrated in Table 4.2. The convergence criterion was defined for both algorithms with
the following equation
82
7
6
-e
E
c
0
;::
•
5
:::::
C)
Previous Method
0
Quasilinearization
4
c
CI)
(,)
c
3
0
(.)
0
0
en
2
1
O-+~.,r"'T1lI
o
1
2
3
4
5
6
7
8
9
10
11
12
13 14
15
Distance Into Column (em)
Figure 4.2. Comparison of traditional methods and quasilinearization for numerical solution of equation (4.2) with the
parameters given in Table 4.1.
83
I (sm+l -sm)/sm 1< n
(4.33 )
where m is the current iteration level and n is the defined
convergence criteria (typically 0.1 to 0.001 %). Equation (4.33) had
to be satisfied for all grid points in the numerical domain.
Computational efficiency was characterized by the number of
iterations required to converge to the steady-state solution and by
the amount of execution time required to converge when a MicroVax computer was used.
Table 4.2 shows typical values for both_
efficiency measures.
The new technique was at least an order of
magnitude more efficient for modeling a one-dimensional problem.
Table 4.1
Parameters used for the Comparison of Numerical Methods
Parameters
Value
k
Kg
L
0.023 g SOC/gVSS-day
0.50 mg SOCII
0.0220 cm
360 cm/day
1.0 cm
0.30
0.0001 day
14 cm- 1
7.2 mg SOCII
0.5%
v
~x
€
~t
a
So
Convergence on S
Table 4.2.
Comparison of Efficiency for Traditional Time-Stepping
and Quasilinearization Techniques
Technique
Time-Stepping
Quasilinearization
Number of Iterations
182
11
Execution Time, secs
19.4
2.6
84
The computational advantage of quasilinearization should increase
dramatically as the problem is increased in size and complexity, such
as by including two and three dimensions.
4.4 Treatment of Lateral Injection Ports
The object of having multiple injections of the electron acceptor is
to spread out the biologically active zone, thus reducing the potential ~
for clogging. The Qs terms in equations (4.1) and (4.2) represents
injections along the flow path. Accurate and efficient solution of the
finite-difference equations becomes a more difficult problem when
lateral injections are allowed, because the inputs create local
numerical instabilities.
Therefore, special treatment is necessary to
incorporate the multiple lateral injections.
The approach used in the numerical formulation is to implement
local upstream weighting of the advection term at the lateral
injection port. This technique is a commonly used method to smooth
out numerical oscillations (Lapidus and Pinder, 1982).
Instead of
equation (4.26), the new finite difference approximation of the
advection term takes the form
v( Si+l - Si)
(4.34 )
at all segments which have inputs.
The discrete equation is
modified for the grid point of lateral injection
(4.35)
where
I
, DR
DH
e=
C
- ~x2
v
--+~x2
~x
Local upstream weighting corrects most of the problem of
upstream numerical dispersion; however, instability is still evident to
some degree. Figure 4.3 compares the results with and without local
upstream weighting for a sample situation. The parameters used are
the same as those shown in Table 4.1, with a lateral injection, Qs, of
85
8
a
Cental Difference
Upstream Weighting
6
=
.......
CD
E
c
-...
Q
4
a:::I
c
Q)
~
C
Q
U
2
o-l,.~r_;t:=f]
0.0
2 5
5.0
7.5
10.0
12.5
15 0
17 5
20 0
Distance from Injection (cm)
Figure 4.3. Comparison of upstream weighting and central finite
differencing on the lateral injection prediction ability.
86
6.5 mg/l at the same velocity of the main flow at grid point 4, which
is 10 em. downstream from the first injection of N03-.
4.5 Development of the Secondary Utilization Model
Secondary utilization is a concept which says that a trace-level
organic compound can be degraded by a biomass, even when the
concentration of the trace-level compound is less than its Smin. The
degradation is possible because the biomass is grown and su tained~
by its utilization of a more plentiful primary substrate (electron
donor), which allows and governs the accumulation of biomass
(Kobayashi and Rittmann, 1982; Namkung, et. al. , 1983). Namkung
et.al.(1983) demonstrated how to model the utilization of a
secondary substrate. Such a model requires knowledge of the kinetic
parameters of the secondary substrate, as well as the distribution of
biomass, which is determined by primary-substrate utilization.
The flux of secondary substrate is determined by its kinetic
parameters and the biofilm thickness.
A secondary substrate does
not effect the biomass thickness, the primary substrate controls Lf.
The secondary substrate enters the column at a certain concentration
and is subject to the same physical processes as the electron donor
and acceptor.
Therefore, another mass balance equation on each
individual secondary substrate must be performed. The steady-state
solute transport equation, equation (4.2), must be solved for each of
the secondary substrates.
The same numerical technique,
quasilinearization and finite-differences, can be utilized with certain
modifications.
Because secondary substrates do not affect Lf, the steadystatebiofilm model is not appropriate for determining its flux.
Instead,
the pseudoanalytical solution of Rittmann and McCarty (1981),
which was built upon the work of Atkinson and How (1974), must be
utilized to estimate the flux of the secondary substrate into the=.
biofilm. The distribution of biomass must be determined previously
from the method presented in Section 4.3 and is a necessary input
requirement for flux estimation for the secondary substrate.
The
details of the pseudoanalytical solution are presented in the original
paper; however a short summary is presented here for clarity.
87
The following dimensionless parameters are defined for use with
the pseudo-analytical solution of Rittmann and McCarty( 1981),
which is appropriate for biofilms at any thickness:
Lf*
= Lf
[(kXf)/(DfKs)] 1/2 ; L *
= L/'t
; Df*
= Df/D;
't ~=
2KsDr/kXf ,
Other dimensionless parameters that appear were defined in Section
4.2
The basic equation for the flux
J*
= 2Dr*Lf*1l S s ~s:
is given by equation (4.36)
1
(4.36)
where 11 = the effectiveness factor.
1. A starting estimate of an effectiveness factor 11 is required
Rittmann and McCarty (1981) suggested starting from
11 = tanh {{i Lf*} 1-{2 Lf*
2. A trial Ss * is estimated from
3. A trial flux is calculated from Ss *.
Ss*
J * = 2D*L*
f f11 Ss *+ 1
4. A checking Ss *' is calculated from the external mass transport
requirement
S s *'
= S * - J* L *
5. A value
cj>
=
cj>
is computed from
(1 + 28 s *') 1/2
6. Checking 11' is calculated from
cj>
88
1 -
tanh(~Lr*)
(
~Lr*
<j>
- 1 )
tanh
if <I>
~
1
<l>
or
1/<1> -
tanh(--aLr*)
-fiLr*
(
<j>
tanh
- 1 )
if <j> ~
1
<I>
7. If,,' and 11 are within 0.1 % of each other , then 11 has converged to
an acceptable value, and it is proper to proceed to the next step.
If not, it is necessary to go back to step 3 and repeat the process.
8. When an acceptable value of 11 is found, J* is calculated from
J*
=
2 D *L * Ss*
11 r r 1 + Ss*
9. The dimensional flux is then
J = J* (KsD/t)
The 11 iteration usually converges in no more than five iterations.
The internal iteration and complexity of the presented algorithm
does not permit an explicit expression for dJ/dS. However, the value
of dJ/dS at a particular bulk substrate concentration was estimated
using finite differences.
Forward differencing was used with an
interval, ~S,
of 0.1 to 0.01 % of Sb. The finite difference equation
used for the dJ/dS evaluation for the secondary substrate was
~
_ J(S +
dS ~S)
- J(S)
(4.37)
~S
The solute transport equation for the secondary substrate was
solved using the Lr distribution calculated from the primary
substrate at every grid point. The secondary substrate was assumed _
not to affect the overall growth rate of the biomass, which was
controlled solely by the primary substrate.
The method of
quasilinearization and finite differences was used identically as
89
before with the dJ/dS estimate given by equation (4.37).
The
concentration of secondary contaminant and its kinetic parameters as
well as the reactor parameters, were necessary inputs for the model.
The convergence and accuracy of the secondary utilization model
were very similar to those of the primary-substrate model presented
in Section 4.3.
90
CHAPTER 5. APPLICATION OF COMPUTER MODELING
5.1 SOC and N03- Profiles
The one-dimensional model presented in Chapter 4 was
evaluated for its ability to describe the laboratory results on SOC and
NO 3 - removal through the BAZs.
The laboratory results were
presented in detail in Chapter 3. The assumptions used for applying
the one-dimensional solute transport equation to the laboratory ~
column are that wall effects were negligible and the surface area due
to the sides of the column (less than one percent of the area of the
glass beads). need not be included.
The kinetic parameters (k, K s , Y, b, and Xf) were determined
independently, as explained in Section 3.2.5. The kinetic and reactor
parameters used to model the laboratory results are presented in
Table 5.1.
The value of Sm i n determined from the kinetic
parameters is slightly less than the plateau concentration of SOC
measured in the laboratory. This can be explained by the formation
of soluble microbial products (SMP) (Namkung and Rittmann, 1986)
which contained C 14.
Thus, SOC measurements toward the
downstream end of the column contained residual substrate and
some SMP, while the model p redictions are only for resi dual
substrate.
5.1.1 One-BAZ Column
The modeling procedure was a two-step process. First the SOC
profile was solved by assuming SOC was the rate-limiting substrate.
This yielded steady-state profiles of SOC concentration and JSOC, the
flux of SOC into the biofilm. Then, the N03- profile was obtained by
solving the solute-transport equation for N03 - when the rate of N03removal was equal to the flux of SOC multiplied by a stiochiometric
coefficient. The stoichiometry was found in the laboratory to be 0.67 .:.
mg N03N/mg Acetate as SOC. That is, equation (4.2) was solved for
the N03 concentration profile with JN03 = 0.67 JSOC. The numerical
values for flux of N03 were computed from stored values of the SOC
obtained with the primary substrate model and were calculated for
each grid point.
Since the rates of N03 were determined by multiplying the flux of SOC at each grid point by 0.67 mg N03 / m g
SOC, the governing transport equation was linear, so that
91
quasilinearization was not required. In other words, dJ/dS was zero
for N03 -, because J was a predetermined constant.
Table 5.1. Parameters used in Solute-Transport Modeling of OneBAZ column
Parameter
Acetate
So
L
Smin
k
Ks
Xf
y
b
IJsoc
DR
Units
Value Used in Model
as SOC
mgSOC/1
cm
mgSOC/l
mgSOC/mgcellday
mgSOC/1
mg cells/cm 3
mg cells/mg SOC
dayl
cm 2 /day
cm 2 /day
cm/day
cm 1
cm 3 /cm 3
as N
mg N03N/I
cm 2 /day
Value with
Alternative Units
6.5
0.02195
0.0131
2.22
4.17 mgSOC/mgcellCd
0.218
15.
8.0 mg cell C/cm 3
0.678 0.36 mg cell C/mgSOC
0.07
1.07
120.58
144.
20.0
0.30
7.32
1.40
The model results are compared with the experimental results in
The model and laboratory results compare very well. .
Figure 5.1.
Both substrates were removed rapidly in the first 5.0 cmdownstream of the injection.
They then approached a plateau
concentration beyond about 10 cm, as the SOC primary substrate
approached its Smin.
The correspondence between model and
experimental results for both substrates verifies that SOC was rate
limiting and that the stoichiometry between N03 and SOC removals
was correct. While there is nearly perfect agreement for the electron
acceptor, N03 , small deviations for SOC occur at 5.0 and 7.5 cm
92
8 __- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
7
~
6
,-..
•
Nitrate Laboratory
+
SOC Laboratory
Nitrate Numerical
Result
5
~
-e=
c
".Cl
4
~
b
d
~
Q,l
d
c
u
3
SOC Numerical
Result
2
1
+
0
o0
2.5
5.0
7 5
10.0
12 5
15.0
Distance from Injection (cm)
Figure 5.1. Comparison of laboratory and numerical results for
the oneBAZ column. Zero distance indicates the
injection port.
93
downstream of the injection port.
These deviations may be the
result of short circuiting due to nitrogen gas build-up, or they may
be caused by sampling error.
5.1.2 Two-BAZ Column
The two BAZ column was modeled using the solute transport
model and the same reactor and kinetic parameters as for the oneBAZ column. The same influent SOC concentration was used as in the~
one BAZ experiment; however, the electron acceptor was injected in
two locations, the second injection being ten centimeters downstream
from the first.
The same total amount of electron acceptor was
injected into both columns, but the two BAZ column received 25%
through the first port and the remaining 75% through the second
port, this corresponds to 1.92 and 5.52 mg N03 --Nil respectively.
This two-injection strategy caused N03 - to be the rate limiting
substrate in the first BAZ, where it was depleted to close to its Smin
just before the second injection.
At this point, there was
approximately 50 % removal of acetate. After the second injection,
N03- was in ample supply, and SOC (acetate) became the rate-limiting
substrate.
The change of rate-limiting substrate after the second injection of
NO 3 - presented an interesting modeling situation.
If the electron
acceptor had limited the growth throughout the length of the column,
the modified model with lateral injection ports could have been used
directly. In the case of a change of limitation, however, two coupled
solute-transport equations had to be used.
In the section of the column after the first injection,
quasilinearization and finite differences were used to solve the solute
transport equation for N03 -, the rate limiting substrate. Then, the
profile for SOC was obtained from the N03 - fluxes and stoichiometry,
as reported in the previous section.
At the point of the second ~
injection of nitrate a new solute transport equation had to be solved.
For the points downstream of the second injection, this new solute
transport equation was solved using SOC as the primary substrate; it
was coupled to the upstream segment of the column by considering
the continuity of SOC flux at the injection port. For N03 -, the upstream flux of N03 - was added to the flux through the injection
port, as it represented only approximately 0.18% of the flux through
94
the port. The N03 - profile after the second injection was obtained
from the SOC fluxes and stoichiometry.
Table 5.2 shows the parameters used in the solute transport
modeling. The kinetic parameters for N03- were not independently
measured in the laboratory and had to be estimated. The maximum
specific rate of substrate utilization, k, was taken from the k of SOC,
adjusted by stoichiometry. The K s value was varied until proper fit
of the laboratory data was obtained. The low value for K s for N03- is~
consistent for electron acceptors (Rittmann and Langeland, 1985).
The kinetic parameters for SOC, the primary substrate after the
second N03 - injection, were averaged from those measured
independently at different locations in the column.
Figure 5.2 shows the numerical results compared to the laboratory
data. The numerical results are in extremely good agreement with
the laboratory data. The stoichiometric values used in the numerical
work, 1.5 mgSOC/mgN03- and 0.67 mgN03-/mgSOC for the first and
second BAZ, respectively, allowed proper representation of both
substrate profiles in both BAZs. Thus, the choice of which substrate
was rate-limiting seems justified.
Comparison of Figures 5.1 and 5.2 demonstrates that having two
NO 3- injections spread out the distance over which a BAZ was
present. With two injections, the BAZ covered about 12.5 cm, while it
covered about 7.5 em for one injection.
5.2 Secondary Substrate Profiles
The laboratory profiles of the secondary substrates presented in
Chapter 3 for the one-BAZ column were modeled using the
framework summarized in Chapter 4. The primary substrate profile
was modeled first. From the results for the primary substrate, the
steady-state biofilm thickness was calculated at each grid point from '
the following equation (Rittmann and McCarty, 1980a)
(5.1)
where Ji is the flux into the biofilm at grid point i. The result was a
profile of biofilm thickness throughout the length of the column. The
L f values then served as key inputs to the model to estimate the
95
8
7
...
Nitrate Laboratory
+
SOC Laboratory
~
6
.
5
-.s
4
El
=I
~
s..
=I
Qj
~
Q
3
U
2
1
0
0.0
2.5
5 0
7.5
10.0
12.5
15.0
Distance from first Injection (em)
Figure 5.2. Comparison of laboratory and numerical results for
the two-BAZ column. Nitrate injections are at 0.0 and
10.0 cm. Lines represent model prediction.
96
17.5
flux of a particular secondary substrate into the biofilm (Rittmann
and McCarty, 1981; Namkung et aI., 1983).
Table 5.2.
Parameters Used in Solute-Transport Modeling of TwoBAZ column
Parameter
Nitrate
So
L
SminN
k
Ks
Xf
YN
b
DN
DHN
v
E
Acetate
So
k
Kg
Xr
Y
Smin
D
Units
Value
as N03 N
1.92, 5.52
mg N03/1
em
0.02195
mg N03/1
0.0090
mg N03/mg cellday 1.45
mg N03/1
0.146
3
mg cells/cm
15.
1.02
mg eells/mg N03dayl
0.07
2
1.07
em /day
2
em /day
120.58
144.
em/day
em 3 /cm 3
0.30
as SOC
7.09
mg SOCII
mgSOC/mgcellday
2.00
mgSOC/I
0.80
15.
mg cells/em 3
0.678
mg cells/mg SOC
mgSOC/I
0.0497
2
1.40
cm /day
The strategy for modeling the secondary substrates was to choose
k and Kg values that provided a good fit to the experimental results
for one experiment. This fitting exercise was needed, because the k
and K s values of these secondary substrates were not known
independently for the denitrification system.
The appropriate reactor parameters and the kinetic parameters of the primary
substrate. were known independently.
The fitted k and Kg values
97
were used to predict the results for different experiments with the
same secondary substrates.
As explained in Section 3.3, several secondary substrates were fed
continuously into the laboratory column. The original detention time
of 50 minutes showed only slight removal of one compound, CTC. To
obtain better removals, the detention time was increase at first to
Most of the compounds
125 minutes and then to 500 minutes.
showed significantly greater percentage removal as the det n ion~
time increased. Three categories of compounds resulted: (1) CTC was
rapidly removed, (2) BF, EDB, TeCE, and TCE were removed less
rapidly, and (3) TCA was not removed. To evaluate the secondarysubstrate modeling, CTC, BF, EDB, TeCE, and TCE were modeled.
5.2.1 Carbon Tetrachloride
The first secondary substrate modeled was CTC, which entered the
first BAZ at a concentration of 81.0 Jlg/l. The k and Ks values which
gave a good fit for a detention time of 50 minutes were 0.030 }lg/mg
cellday and 4.5 }lg/l, respectively. Figure 5.3 shows the results of
the numerical fitting for the 50 minute detention time. Clearly, all of
the points are well represented by the numerical result. In order to
demonstrate the predictive ability of the numerical model, the
profiles at the two other detention times (125 and 500 min.) were
calculated using the same k and Ks values. The superficial velocity
and influent concentration are suitably modified for each of the other
runs.
The 125 min. detention time was modeled by changing the
superficial velocity and influent concentration to 0.04 cm/min and
69 Jlg/I, respectively. Figure 5.4 shows the results of the numerical
prediction compared to the laboratory data. The two curves fit well
and show removal through all of the BAZ ( 7.5 em.). The SOOmin. _
detention time column was modeled using the laboratory obtainedconcentration of 53 Ilg/1 and an adjusted superficial velocity of 0.01
em/min. Figure 5.5 shows the results of the numerical prediction.
The results are encouraging, because the numerical and experimental
results have
the same trends of rapid decrease and approach a plateau concentration.
The absolute values of the plateau
98
90 - - . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
Numerical
80
:C)
:::
E;
..
Laboratory
70
r::::
0
:;::
-...
C'O
r::::
60
Q)
Co)
r::::
0
()
50
40
-+---....---.,....--,..---,..---.......---r__----,r---_--__- -__---I
0.0
2.5
5.0
7 5
10.0
Distance from Injection (cm)
Figure 5.3. Numerical curve fit to the eTC profile at a detention
celltime of 50.0 min. The k and K s are 0.030 ~g/m
day and 4.5 ,l/g~
respectively.
99
12.5
70
-
60
::::
C)
::l
.....-
•
Numerical
Laboratory
50
c
...
40
0
co
30
c
Q)
()
20
c
0
0
10
0
0.0
2.5
5.0
7.5
10.0
12.5
Distance from Injection (cm)
Figure 5.4. Prediction of the eTC profile at a detention time of
125. min. and with k = 0.030 J..Lg/mg cellday and Kg =
4.5 J..Lg/I.
60..,.....,
Numerical
50
Laboratory
C)
:i.
.....-
40
c
-...
0
co
30
c
Q)
()
20
c
0
0
10
0
0.0
2.5
5.0
7.5
10.0
12.5
Distance from Injection
(cm)
Figure 5.5. Prediction of the eTC profile at a detention time of
500 min and with k = 0.030 J..Lg/mg cellday and K s =
4.5 J..Lg/l.
100
concentration are slightly different, but are in the same order of
magnitude.
Before the secondary-substrate experiments were started, the one
BAZ column had been operated for over one year at approximately
By the
the same detention time and concentration of substrates.
time the secondary substrates were added to the column, a steadystate biomass distribution had been attained throughout the length
of the BAZ. One possibility is that the biomass distribution changed ~
during the experiments at longer detention times and, to some
degree, with different influent SOC concentrations. Reduction of the
detention time reduced the substrate (SOC) input to the BAZ and
should have created a situation of slower biological activity in the
BAZ. The time that the new detention times were maintained before
sampling of the profiles of the secondary substrates is a major factor
in the assumption made about the biomass distribution for the
secondarysubstrate measurements.
Because of the long time used
to establish the steadystate biofilm and the relatively short times
the reactors were run at the new detention times ( 6 weeks at 125
min. and 5 weeks at 500 min.), it was assumed that the biomass
distribution remained the same as that calculated for the 50 minute
detention time. Nevertheless, it is possible that the BAZ lost active
biomass during the detentiontime experiments. The data shown in
Figures 5.3 5.5 suggest that the loss of activity for eTC removal was
negligible.
5.2.2
Bromoform. Ehtylene Dibromide. Tetrachloroethene. and
Trichloroethene
The second secondary substrate modeled was bromoform (BF),
which entered the column at a concentration of 106 llg/l at a
detention time of 50 min. As opposed to CTC, BF had no detectable
removal at the 50 min detention time. This was due to insufficient
contact time with the biomass, and as a result, the detention time ofthe column was increased.
The 125min. detention time had an
influent concentration of BF of 57 Jlg/I, and BF removal was
observed. In this case, the kinetic parameters, k and K s , were found
for this detention time. Figure 5.6 shows the results of the numerical _
fitting between the experimental values and the model results using
k and K s values of 0.013 Jlg/mgday and 9.5 Jlg/l, respectively.
101
50
:::
tn
::1.
40
c
...
0
30
Numerical
Laboratory
A
aJ
c
20
G)
u
c
0
10
0
0
0.0
2 5
5 0
7 5
10 0
12 5
Distance from Injection (em)
Figure 5.6
Numerical fit to the BF profile at a detention time of
125 min. The k and Ks are 0.013 ~g/m
cell-day and
9.5 ,l/g~
respectively.
60
-...
-
50
A
tn
Numerical
Laboratory
::i.
40
c
0
aJ
c
G)
u
c:
0
30
20
10
0
00 . 0
5.0
2.5
7.5
10.0
12.5
Distance from Injection (em)
Figure 5.7. Prediction of the BF profile at a detention time of 500
min. and with k = 0.013 J.Lg/mg cell-day and K s = 9.5
~g/l.
102
The 500-min. detention time column had an influent
concentration of 54 Ilg/1 of BF. Figure 5.7 shows the results of the
numerical model using the kinetic parameters from the 125 minute
detention time.
The numerical model predicted a slightly lower
plateau concentration than was measured in the laboratory.
This
difference in removals perhaps can be attributed to biofilm loss
between the two sampling periods.
The assumption made was that
the biomass distribution remained constant throughout the duration
of the secondary-utilization experiments, even though some biomass ~
loss probably took place.
Modeling predictions for the 50-min. detention time gave a
prediction of only 1.5% removal of BF (results not shown).
This
negligible predicted removal was consistent with the undetectable
removal for the experiments. The kinetic parameters of
tetrachloroethene (TeCE),
ehtylene dibromide (EDB), and
trichloroethene (TCE) were measured at the intermediate detention
time of 125 min. Again, there were no detectable removals of these
compounds at the lowest detention time of 50 min. The numerical
model was fit to the laboratory data to obtain the k and K s values of
each compound. These values are shown in Table 5.3. Because the
numerical fit to laboratory data was similar to that shown for BF in
Figure 5.6, these curves are not presented. The same trend of
somewhat greater removals of substrate predicted by the numerical
model than measured in the laboratory applied to the greater
detention time results for these three compounds. The differences
are hypothesized to be the result of biomass loss.
Table 5.3. K s and k Values of TeCE, EDB, and TeE Obtained From
Numerical Curve Fitting.
(mg/mg-day)
Compound
Ks (mg/l)
k
TeCE
FDB
TCE
8.0
9.0
8.5
0.01760
0.00900
0.00935
103
5.3
Simulation of Bioreclamation Strategies
There are several possible strategies that can be used to achieve
maximum performance of a bioreclamation site.
Two very
interesting strategies are presented in this chapter in the form of
hypothetical examples. One strategy is to minimize the clogging from
biomass growth. A clogging problem is exacerbated by injection of
an excess amount of electron acceptor through one well or port. If
the electron-donor concentration is relatively large compared to tha ~
of the electron acceptor, clogging is likely to develop in a region close
to the point of injection. In order to reduce the potential for clogging,
lower concentrations of the electron acceptor can be added at several
locations along the flow path.
The clogging potential can be demonstrated by presenting an
example problem.
A model problem compares one injection of
nitrate at 10.0 mg/1 to a multiple injection of an equivalent amount
of nitrate. For this example, SOC is assumed to be present in excess,
so that nitrate is assumed to limit the growth throughout the length
of the column. The concentration of SOC is assumed to be 20.0 mg
SOCII as it enters the column. Kinetic parameters for denitrification
limited by N03 -, found in Table 5.2, were used for the model
problem. The reactor parameters used in this example are shown in
Table 5.3
Table 5.4.
Parameters Used in Clogging Example Problem
Parameter
Units
Nitrate
So
L
v
as N03-- N
mg N03-/l
cm
cmlday
cm 3/cm 3
em
cm- 1
as SOC
mg SOCII
E
dp
a
Acetate
So
Value
104
3.3, 3.3, 3.3
0.0289
180.
0.30
0.12
50.0
20.00
Figure 5.8 shows the nitrate and SOC profiles when the N03 - is
It is evident that the net
injected through one or three ports.
removal of SOC is nearly equivalent for either case, although it is
slightly better with three injections. A relative measure of clogging
potential can be estimated by considering the relative biofilm
thickness throughout the length of the column. The relative biofilm
thickness is the film thickness at a point in the reactor divided by
the particle diameter. It is clear by comparing the relative biofilm
that thethicknesses of the two scenarios, given in Figure 5.9,
multiple injection gives much less potential for clogging.
The second strategy involves enhancing the removal of organic
contaminants when the total available primary substrate is low.
Injection of additional SOC when the original SOC is depleted should
extend the BAZ and allow increased consumption of individual
secondary substrates.
In order to demonstrate the advantage of an additional injection
of carbon, a simple example was developed.
Consider a situation
such as the SOC profile in the oneBAZ column (see Figure 5.1). The
The
SOC was rate limiting throughout the length of the column.
injected electron acceptor was removed approximately 50%. At this
point, if additional acetate were added to the column, the BAZ could
be extended. This configuration was modeled numerically using the
laboratory kinetic parameters.
The initial concentration of nitrate
was 10.0 mg N03N/l, and the SOC was 6.5 mg SOC/I. At a distance
of ten centimeters downstream from the nitrate injection, 10.0 mg
SOCIl was injected. Figure 5.10 shows the resulting nitrate and SOC
profiles. The BAZ was extended another 7.5 cm from the second
injection.
The advantage of adding the SOC injection can be demonstrated
further by considering the fate of CTC applied to the column at a
concentration of 100 J1g/l (using the same k and Ks values obtained in =.
Section 5.2.1). Figure 5.11 shows the profile of CTC throughout the
length of the column. There is approximately 43% removal in the
first ten centimeters of the column. An additional 40 % removal is
calculated for the last ten centimeters and is due to the injection of
SOC. Thus, extension of the BAZ by addition of more biodegradable
SOC significantly enhanced removal of CTC.
105
20
B SOC Single
+
18
16
Q
CiJ
14
-=
12
S
''
.s=
Nitrate Single
B SOC Multiple
0 Nitrate
Multiple
~
....
q ,j
10
u
=
0
u
8
6
4
2
0
0.0
2.0
40
6.0
8.0
10.0
12 0
Distance into Column
(em)
Figure 5.8. SOC and nitrate profiles for one and three injections of N03·
106
1.00
0.90
0.80
fI.l
fI.l
0.70
c.;
.....
.c
0.60
Q.>
-a-
Single Injection
-+-
Multiple Injection
=
~
~
.§
=
<:>
0.50
=a
Q.>
.....i>
~
~
0.40
Q.>
0.30
0.20
0.10
0.00
0.0
2 0
6.0
4.0
8 0
10.0
Distance into Column
(cm)
Figure 5.9. Relative biofilm thicknesses companng single and
multiple nitrate injections.
107
12 0
-
10
SOC Profile
Nitrate Profile
8
g
c
t:lJl
6
.g==
~
~
"S
~
4
Cl
U
2
......___r_____I
...._ r_ t~_r _
O-+_,r.:s~;=Q
0.0
2 5
5 0
7 5
10.0
12 5
15 0
17 5
20
a
Distance into Column
(em)
Figure 5.10. Profiles of SOC and nitrate after being injected
alternately.
108
22.5
25.0
120 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
100
e
:::
m
::i.
80
-...
60
CTC
C
0
co
c
Q)
(,)
c
0
0
40
20
______,r__,.___r__-r__"""T"""___,_-r__...,.._......,...-r______1
0-+ ~-r_ " 'T" . ., .-r -" T" "'
0.0
2.5
5.0
7 5
10.0
125
150
17.5
20 0
22.5
25.0
Distance into Column (cm)
Figure 5.11.
Profile showing additional secondary utilization of
eTC after SOC is added by a second injection at 10
cm.
109
ClIAPTER 6. CONCLUSIONS AND RECOMMENDATIONS
This research project investigated the fundamental mechanisms
that can act when an electron acceptor is injected along the flow path
of an electron-donor-rich groundwater to establish a biologically
active zone (BAZ) for degradation of pollutants that serve as primary
and secondary substrates.
The research methodology consisted of
laboratory column experiments that were coupled with computer
modeling.
The laboratory experiments demonstrated that lateral injection of
NO 3- could be successfully utilized to control the location and extent
of BAZs in systems where acetate was fed as the sole carbon source.
Columns containing one and two BAZs were successfully operated,
and profiles of acetate and N03 - were determined.
Addi tional
measurements of steady-state biofilm thicknesses and densities gave
further evidence of the value of lateral injection for spreading out
biological activity along the flow path, which leads to enhanced
biodegradation capability and diminished clogging potential. These
experiments also demonstrated the deleterious effects of N2 gas
accumulation;
N2 gas bubbles that occurred as a result of
denitrification tended to accumulate in the BAZs, resulting in reduced
liquid contact times and lowered acetate removal efficiencies.
Laboratory experiments evaluating the secondary utilization of
eight trace-concentration halogenated solvents were also conducted.
Results of these experiments indicate that carbon tetrachloride was
removed
most completely by denitrifying BAZs,
while
tetrachloroethene, bromoform, dibromoethane, and trichloroethene
were removed to
lesser degrees.
Trichloroethane removal was
slight.
A significant result was that 1,2 and 1,3 dichlorobenzene
were 20-30% removed; these compounds have previously been
considered refractory under denitrifying conditions.
A highly efficient numerical model that couples solute transport
mechanisms and biofilm kinetics was developed.
Employing a
quasilinearization technique for the biofilm reaction term, the model
is capable of solving directly for the steady-state profiles of limiting
substrate, biofilm thickness, non-limiting substrates, and secondary The predictive ability of the model was successfully
substrates.
verified by simulating the results of the laboratory experiments
111
using independently determined kinetic parameters.
Independently
determined kinetic parameters did not exist for the secondary
substrates; in this case, one set of results from the column
experiments was used to obtain a best-fit set of kinetic parameters,
which were then used to predict the results for experiments
conducted with different liquid flow velocities.
The model
predictions correctly described all trends.
Absolute deviations
between predicted and experimental results were very small for
cases involving acetate and nitrate; systematic deviations for some 0[the secondary substrates occurred and probably were due to a loss
of biomass during the experiments conducted at the higher detention
times.
The steadystate models were applied to investigate possible
strategies to be used in field bioreclamations. The use of multiple
injection wells was studied for its ability to decrease aquifer clogging
potential by spreading out the distance over which the limiting
substrate is added.
Modeling results verified that the strategy of
multiple injections could reduce high densities of biofilm
accumulation near the injection well.
Also investigated was the
strategy of adding a supplemental carbon source to extend the length
of a BAZ. The modeling illustrated that such an extension of the BAZ
could be accomplished and could result in longer contact times for a
secondary substrate in the BAZ, thereby increasing the removal of
the secondary substrate.
The results of this research demonstrate that injection of limiting
substrates along the groundwater flow path is a viable means of
establishing spatially distributed BAZs for enhanced ins i t u
bioreclamation.
Tracelevels of hazardous secondary substrates can
The
be degraded as groundwater flows through the BAZs.
phenomena of formation of BAZs and substrate utilization within
BAZs can be quantitatively interpreted and predicted at the _
laboratory scale by rigorous mathematical models that coupleprinciples of solute transport and biofilm kinetics.
An ultimate goal is to develop the fundamental
coupled biological and hydrological processes to a
great that fieldscale in situ bioreclamation systems
reliably. A critical need is to extend the research
112
understanding of
level sufficiently
can be designed
reported here to
include transient and multi-dimensional aspects. In particular, the
following areas of additional research are recommended:
1. Use the combined experimental and modeling approach to study
transient biofilm kinetics and dual-substrate limitation.
2. Examine the use of alternative electron acceptors to establish
specialized BAZs capable of degrading specific pollutants.
3. Conduct further study of the basic mechanisms of dichlorobenzene
degradation under denitrifying conditions.
4. Examine the fundamental mechanisms of bacterial transport and
attachment and their role in the establishment and extension of
BAZs.
5. Study the effect of biological activity upon the hydraulic
properties of aquifers.
6. Extend the computer modeling to consider transient,
heterogeneous, multi-dimensional flow fields, as well as
transient biofilms.
7. Study the phenomena controlling the biodegradation of organic
contaminants which are strongly adsorbed or which form a
nonaqueous phase.
113
REFERENCES
American Public Health Association, Inc. 1981. Standard Methods for
the Examination of Water and Wastewater. 15th ed. Washington D.C.
Atkinson, B. and S.Y. How. 1974. "The Overall Rate of Substrate
Uptake (Reaction) by Microbial Films. Part II. Effect of Concentration
and Thickness with Mixed Microbial Films." Transactions. Inst. Chern.
Engrs. 52: 260-272.
Bouwer, E.J. 1987. "Biotransformations of Contaminants in
Groundwater. n Paper presented at the Third National Water
Conference, Philadelphia, PA, January.
Bouwer, E.J. and P.L. McCarty. 1983. "Transformation of Halogenated
Organic Compounds Under Denitrification Conditions." Appl. Environ.
Microb. 45: 1295-1299.
Bouwer, E.J. and P.L. McCarty. 1985. "Utilization Rates of Trace
Halogenated Organic Compounds in Acetate-Grown Biofilms." Biotech.
and Bioengr. 27: 1564-1571.
Casolari, A. 1988. "Microbial Death." In Physiological Models in
Microbiology, Vol. II. edited by M.J. Bazin and J.I. Prosser. 1-44. Boca
Raton, FL: CRC Press, Inc.
Chang, H.T. 1985. Mathematical Modeling and Scanning Electron
Microscopic Study of Biofilms on Adsorptive Media. Ph.D.
Dissertation. Department of Civil Engineering, University of Illinois,
Urbana, IL.
Daugherty, R. L. and J. B. Franzini. 1977. Fluid Mechanics with
Engineering Applications. New York: McGraw-Hill.
Flathman, P.E. et al. 1983. Biological Spill Cleanup. Proc. Technical
Seminar on Chemical Spills; October 25-27; Toronto. 117-130.
Flathman, P.E., J.R. Quince, and L.S. Bottomley. 1984. Biological
Treatment of Ethylene Glycol-Contaminated Groundwater at Naval
Air Engineering Center, Lakehurst, New Jersey. Proc. 4th National
115
Symposium on Aquifer Restoration and Ground Water Monitoring;
May 23-25; Columbus, Ohio. 23-25.
Freeze, R.A. and J.A. Cherry. 1979. Groundwater. Englewood Cliffs,
N.J.: Prentice-Hall.
Ghiorse, W.C. and D.L. Balkwill. 1983. "Enumeration and
Morphological Characterization of Bacteria Indigenous to Subsurface
Environments." Dev. Ind. Microb. 24.
Kobayashi, H. and B.E. Rittmann. 1982. "Microbial Removal of
Hazardous Organic Compounds." Environ. Sci. Techno!. 16: 170A181A.
Kuhn, E.P., P.J. Colberg, J.L. Schnoor, O. Wanner, A.F.B. Zehnder, and
R.P. Schwarzenbach. 1985. "Microbial Transformations of Substituted
Benzenes during Infiltration of River Water to Groundwater:
Laboratory Column Studies." Environ. Sci. Techno!. 19: 961-968.
Lapidus, L. and G. F. Pinder. 1982. Numerical Solution of Partial
Differential Equation in Science and Engineering. New York: J. Wiley
and Sons, Inc.
Lee, E. S. 1968. Quasilinearization and Invariant Imbedding. New
York: Academic Press.
Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt. 1982. Handbook of
Chemical Property Estimation Methods. New York: McGraw-Hill Book
Co.
McCarty, P.L. 1972. "Energetics and Bacterial Growth." In Organic
Compounds in Aquatic Environments. edited by S.D. Faust and J.V.
Hunter. New York: Marcel Dekker, Inc.
McCarty, P.L., B.E. Rittmann, and M. Reinhard. 1981. "Trace Organics
in Groundwater." Environ. Sci. Technol. 15: 40-51.
Miller, M.M., S. P. Wasik, G.-L. Huang, W.-Y. Shiu, and D. Mackay.
1985. "Relationship between Octanol-Water Partition Coefficient and
Aqueous Solubility." Environ. Sci. Technol. 19: 522-529.
116
Nagel, G., W. Kuehn, P. Werner, and H. Sontheimer. 1982. "Sanitation
of Groundwater by Infiltration of Ozone Treated Water." G WF Wasser/Abwasser. 123: 399407.
Namkung, E. 1985. Kinetics and Mechanisms of LowConcentration
Multisubstrate Utilization by Biofilms. Ph.D. Dissertation. Department
of Civil Engineering, University of Illinois, Urbana, IL.
Namkung, E. and B.E. Rittmann. 1986. Soluble Microbial Products
(SMP) Formation Kinetics by Biofilms. Water Research.20: 795806.
Namkung, E., R.G. Stratton, and B.E. Rittmann. 1983. "Predicting
Removal of TraceOrganic Compounds by Biofilms." J. Water Poll.
Control Fed. 55: 13661372.
Parker, G.A. 1928. In Colorimetric Determination of Nonmetals, edited
by D.R. Boltz and Z.A. Howell. New York: J. Wiley and Sons.
Rittmann, B.E. 1982a. "Comparative Performance of Biofilm Reactor
Types." Biotech. and Bioengr 24: 13411370.
Rittmann, B.E. 1982b. "The Effect of Shear Loss on Biofilm Loss Rate."
Biotech. and Bioengr. 24: 501506.
Rittmann, B.E. and W.E. Langeland. 1985. "Simultaneous
Denitrification with Nitrification in Single Channel Oxidation Ditches."
J. Water Poll. Control Fed. 57: 300308.
Rittmann, B.E. and P.L. McCarty. 1980a. "Evaluation of SteadyState
Biofilm Kinetics. Biotech. and Bioengr. 22: 23592373.
II
Rittmann, B.E. and P.L. McCarty. 1980b. "Model of SteadyState
Biofilm Kinetics." Biotech. and Bioengr. 22: 23432357.
Rittmann, B.E. and P.L. McCarty. 1981. "Substrate Flux into Biofilms of
any Thickness." J. Environ. Engr. 107: 831849.
Rittmann, B.E., P. L. McCarty, and P.V. Roberts. 1980. TraceOrganics
Biodegradation in Aquifer Recharge. Ground Water. 18: 236243.
117
Saez, P.B. and B.E. Rittmann. 1988. An Improved Pseudo-Analytical
Solution for Steady-State Biofilm Kinetics. Biotech. and Bioengr. 32:
362-368.
Stratton, R., E. Namkung, and B.E. Rittmann. 1983. "Biodegration of
Trace-Organic Compounds by Biofilms on Porous Media." J. Amer.
Water Works Assn. 75: 463-469.
Werner, P. 1985. "A New Way for the Decontamination of Polluted
Aquifers by Biodegradation." Water Supply. 3: 41-47.
118
APPENDIX--NOMENCLATURE
Fundamental
L
M
Ms
Mx
T
quantities
length
mass, in general
mass of substrate
mass of bacteria
time
English symbols
a
b
bs
bT
dp
D
DR
Df
f
J
Jdeep
J(Sffi)
k
Ks
L
Lf
Lfi
LT
n
Qs
rdiff
rut
surface area per reactor volume (L-l)
specific decay or maintenance-respiration coefficient (T-l)
coefficient of specific biofilm loss due to shearing (T-l)
overall first-order biofilm loss coefficient (T-l)
particle diameter (L)
molecular diffusion coefficient of substrate in water (L2T-l)
hydrodynamic diffusion coefficient (L2T-l)
molecular diffusion coefficient of substrate in biofilm (L2T-l)
ratio between the fluxes in actual and deep biofilms
substrate flux into biofilm (MsL -2T-l)
minimum substrate flux into a deep steady-state biofilm
(MsL -2T-l)
substrate flux at a gridpoint and iteration level m (MsL -2T-l )
maximum specific rate of substrate utilization (MsM x-I T-l )
half-maximum-rate substrate concentration (MsL -3)
thickness of effective diffusion layer (L)
biofilm thickness (L)
biofilm thickness at a gridpoint i (L)
total length of reactor (L)
number of gridpoints used in numerical model
substrate source due to lateral input through injection ports
(MsL -3T-l)
rate of substrate accumulation due to diffusion (MsL -3T-I)
rate of substrate accumulation due to substrate utilization
(M sL-3T-I)
rate-limiting substrate concentration (MsL -3)
bulk-liquid substrate concentration (MsL -3)
rate-limiting substrate concentration at gridpoint i (MsL -3)
119
sin
Smin
So
So
Ss
St
sm
t1S
t
t1t
v
x
.1. x
Xr
Xo
Xt
t1X
Y
z
substrate concentration at the inlet of reactor (MsL -3)
minimum bulk substrate concentration of the rate-limiting
substrate able to sustain a steady-state biofilm (MsL -3)
initial substrate concentration at time=O (MsL -3)
influent concentration of rate-limiting substrate (MsL -3)
substrate concentration at liquid-biofilm interface (MsL -3)
substrate concentration at time=t (MsL -3)
rate-limiting substrate concentration at iteration level m
(M sL-3)
change in substrate concentration (MsL -3)
time (T)
change in time (T)
superficial flow velocity (L T-l)
longitudinal distance into the reactor (L)
grid-spacing for the numerical model (L)
biomass density in the biofilm (MxL -3)
biomass concentration at time=O (MxL -3)
biomass concentration at time=t (MxL -3)
change in biomass concentration (MxL -3)
true yield of bacterial mass per unit substrate mass
utilized (MxM s)
distance normal to biofilm surface (L)
Dimensionless symbols
Dr*
J*
dimensionless molecular diffusion coefficient of substrate In
biofilm [= DflD)
dimensionless substrate flux into the biofilm
[=
(K s ki rD r)1/2]
J* deep dimensionless substrate flux into a deep biofilm
[
K*
=
Jdeep
]
(K skX rD f)1/2
dimensionless kinetic parameter
D
Ks
1/2]
[ = (L)
[(kXrDf)j
L*
Lr*
dimensionless diffusive layer thickness [= L/r,]
dimensionless biofilm thickness [= Lr [(kXr)/(DrKs)] 1/2]
120
Ss *
S min *
dimensionless substrate concentration at the liquid-biofilm
interface [== (Ss/K s)]
dimensionless minimum bulk substrate concentration of the
rate-limiting substrate able to sustain a steady-state biofilm
[=Ykb?, bT]
Sb *
dimensionless bulk substrate concentration [= (Sb/Ks)]
Greek symbols
a
~
€
11
J.lm
J.l
t
<j>
product coefficient in the factor f
exponential coefficient in the factor f
porosity or bed voidage
effectiveness factor, ratio of actual and fully-penetrated
substrate fluxes at a given Ss
maximum specific cell growth rate (T-1)
specific cell growth rate (T-1)
standard biofilm depth dimension (L) [="./ 2K sDflkXf ],
coefficient used in flux estimation of secondary substrates
[=
fi
Lf*
(1 + 28 s *')1/2]
convergence criteria of numerical method
121