United States
Environmental Protection
Agency
Robert S. Kerr Environmental
Research Laboratory
Ada, OK 74820
EPA/600/2-89/033
July 1989
Research and Development
&EPA
In-Situ Aquifer
Restoration of Chlorinated
Aliphatics by Methanotrophic
Bacteria
-------
EPA/600/2-89/033
July 1989
IN-SITU AQUIFER RESTORATION OF CHLORINATED ALIPHATICS
BY METHANOTROPHIC BACTERIA
by
Paul V. Roberts, Lewis Semprini, Gary D. Hopkins,
Dunja Grbic-Galic, Perry L McCarty, and Martin Reinhard
Research Staff:
Constantinos V. Chrysikopoulos, Mark E. Dolan, Franziska Haag,
Thomas C. Harmon, Susan M. Henry, Robert A. Johns, Nancy A. Lanzarone,
Douglas M. Mackay, Kevin P. Mayer, and Robert E. Roat
Department of Civil Engineering
Stanford University
Stanford, California 94305
CR-812220
Project Officer
Wayne C. Downs
Processes and Systems Research Division
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma 74820
U.S. ENVIRONMENTAL PROTECTION AGENCY
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
ADA, OKLAHOMA 74820
U.S. Environmental Protection Agency
Regions, Library (PL-12J)
12ih
-------
DISCLAIMER NOTICE
The information in this document has been funded wholly or in part by the United States Environmental
Protection Agency under Cooperative Agreement CR-812220 to Stanford University. It has been
subjected to the Agency's peer and administrative review, and it has been approved for publication as
an EPA document. Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
-------
FOREWORD
EPA is charged by Congress to protect the Nation's land, air, and water systems. Under a
mandate of national environmental laws focused on air and water quality, solid waste management and
the control of toxic substances, pesticides, noise and radiation, the Agency strives to formulate and
implement actions which lead to a compatible balance between human activities and the ability of
natural systems to support and nurture life.
The Robert S. Kerr Environmental Research Laboratory is the Agency's center of expertise for
investigation of the soil and subsurface environment. Personnel at the Laboratory are responsible for
management of research programs to: (a) determine the fate, transport and transformation rates of
pollutants in the soil, the unsaturated and the saturated zones of the subsurface environment; (b) define
the processed to be used in characterizing the soil and subsurface environment as a receptor of
pollutants; (c) develop techniques for predicting the effect of pollutants on ground water, soil, and
indigenous organisms; and (d) define and demonstrate the applicability and limitations of using natural
processes indigenous to the soil and subsurface environment for the protection of this resource.
This report describes research conducted to develop, evaluate, and demonstrate the efficacy of
enhanced biotransformation of chlorinated organic contaminants for in-situ aquifer remediation. The
research shows how methanotrophic bacteria can be employed to degrade compounds such as vinyl
chloride, 1,2-dichloroethylene isomers, and trichloroethylene, which are widely encountered as ground
water pollutants.
Clinton W. Hall
Director
Robert S. Kerr Environmental
Research Laboratory
in
-------
EXECUTIVE SUMMARY
This project undertook the ambitious task of evaluating the potential of an innovative approach to in-
situ aquifer restoration: enhanced biotransformation of chlorinated solvents. The chlorinated solvents
investigated included trichloroethylene (TCE), cis- and trans-1,2-dichloroethylene (cis- and trans-DCE),
and vinyl chloride (VC). Biotransformation was achieved by creating conditions that promote the growth of
a group of bacteria known as methanotrophs, microorganisms that use methane as a source of food and
energy under aerobic conditions. The growth of these bacteria can be stimulated by providing methane
and oxygen in the proper amounts. These bacteria are able to transform the target chlorinated solvents as
secondary substrates through a process termed cometabolism.
This report presents the results of a multidisciplinary investigation that evaluated the technical
feasibility of biostimulating native populations of methane-oxidizing bacterial communities to degrade the
target chlorinated compounds. The evaluation encompassed a small-scale demonstration of the
method's effectiveness under natural conditions at a field site, as well as detailed investigations of the
bacterial community's growth and transformation kinetics.
The laboratory studies improved understanding of the growth and transformation kinetics of the
microbial population and also characterized the aquifer solids from the field site with respect to important
transport properties. These laboratory investigations played an essential role in providing a foundation for
designing and interpreting the field demonstration. Mixed cultures of methane-oxidizing bacteria and
heterotrophs enriched from samples of the aquifer solids were capable of degrading TCE rapidly and
completely under favorable conditions. Pure cultures of methane-oxidizers were isolated that also trans-
formed TCE. It was demonstrated that nutrient media formulations could increase substantially the TCE
transformation rates by mixed and pure suspended cultures. Studies with labelled compounds indicated
that the transformation by mixed cultures proceeded to form CC-2 and cell mass as the main products,
although there was a minor amount of intermediate product of trans-DCE transformation. Column
experiments showed that high methane concentrations slowed the transformation of the target com-
pounds through competition for enzyme sites. Sorption experiments showed that the equilibrium relation
was approximately linear within the concentration range studied, and sufficiently strong to account for
retardation factors on the order of two to ten for the various target organic compounds.
In the field experiments, biostimulation was accomplished by feeding the native population
methane and oxygen. Methane utilization commenced rapidly, within ten days in the first biostimulation at
the test site, and within one day in subsequent biostimulation episodes. Biotransformation of the target
organic compounds ensued immediately after commencement of methane utilization, and reached steady
state values within approximately two weeks. The extents of transformation at steady state were as fol-
lows: approximately 95% for VC, 85% for trans-DCE, 40% for cis-DCE, and 20% for TCE. These amounts
of transformation were achieved in a relatively small biostimulated zone, with travel distances of 1 to 4
meters and residence times of 8 to 25 hours. Mathematical modeling of the transport and transformation
processes confirmed that the behavior observed in the field demonstration was entirely consistent with
the results of the laboratory research and theoretical expectations. The excellent agreement between the
observed behavior and model predictions strengthens confidence in the validity of the results and con-
clusions.
This research has confirmed that biostimulation of a native population of methanotrophs is capable
of substantially enhancing the transformation of halogenated aliphatic contaminants. The transformation
rates are moderately high, with half lives on the order of several hours to several days under the conditions
iv
-------
that could be created and maintained in the subsurface environment. Transformation rates of this
magnitude are substantially greater than the rates of most natural degradation processes in the sub-
surface, and if achieved over an extensive domain, would lead to complete degradation over the time
spans corresponding to typical aquifer residence times.
Biostimulation of natural methanotrophic bacteria to achieve biotransformatipn of halogenated
alkenes deserves full consideration as an alternative means of groundwater remediation in cases where
this group of chemicals constitutes a major part of the contamination. This technology has been demon-
strated to be effective in continuous operations in a real subsurface environment at small scale, and is
ready for demonstration at a real contamination site where conditions are favorable.
-------
CONTENTS
Foreword iii
Executive Summary iv
Figures ix
Tables xv
Acknowledgments xvii
1. Introduction 1
Background 1
Research objectives 10
Report organization 11
2. Summary and Conclusions 12
Field demonstration methodology 12
Site characterization 13
Field demonstration of biostimulation and biotransformation 13
Identification of intermediate product 14
Sorption 15
Growth and transformation rates 15
Mathematical modeling 16
Project integration 17
3. Recommendations 18
Recommendations for process research 18
Recommendations for application 19
4. Field Experiment Methodology and Site Characterization 21
Experimental methodology 21
Characterization of the field site 24
Geologic characteristics 25
Site instrumentation 36
The automated Data Acquisition and Control system 39
Analytical system performance 41
Summary 44
5. Results of Tracer Tests 45
Natural gradient tracer tests 45
Induced gradient tracer tests 47
Pulsed injection 61
Summary 63
6. Results of Biostimulation and Biotransformation Experiments 65
Results of the first season of field testing 67
Results of the second season of field testing 73
Results of the third season of field testing 82
7. Formation and Fate of trans-Dichloroepoxide 91
Chemical synthesis of trans-DCO 91
Results 92
Summary 90
8. Sorption 96
Introduction 96
Moffett solids 96
Moffett solids sorption studies 102
Summary 108
vii
-------
9. TCE Transformation by Mixed and Pure Groundwater Cultures 109
Introduction 109
Methods 109
TCE Degradation experiments 110
Results 113
Conclusions 125
10. Batch Exchange Soil Column Studies of Biotransformation by Methanotrophic Bacteria 126
Introduction 126
Materials and Methods 126
Results 131
Discussion 143
Conclusions 145
11. Continuous Flow Column Studies 147
Introduction 147
Methods 148
Results 151
Summary 159
12. One-Dimensional Solute Transport in Porous Media with Well-to-Well Recirculation 162
Background 162
Model formulation and solution 163
Parameter estimation methodology 165
Application to field experiments 166
Summary 168
13. Biostimulation and Biotransformation Modeling 172
Introduction 172
Model development 172
Model simulations of biostimulation experiments 177
Model simulations of biotransformation experiments 188
Discussion 194
14. In-situ Biotransformation Methodologies 197
Prototype scenarios 197
Contamination characterization 200
Comparison of laboratory and field results 200
Model simulations of restoration scenarios 203
References 205
viii
-------
FIGURES
Number Page
1.1 Oxidation of methane by methanotrophs and the initial transformation step in TCE
oxidation 4
1.2 Initial oxygenation of TCE by methanotrophs and the transport of the epoxide
from the cell 6
1.3 Degradation of chlorinated alkenes by mixed methane-grown microbial enrichments
from soil 7
1.4 Anaerobic transformation pathways for selected chlorinated aliphatic compounds 8
1.5 Possible systems for biologically treating a contaminated aquifer 10
4.1 Vertical section of the test zone 22
4.2 Location of the field site, SU-39, at the Moffett Naval Air Station, Mountain View,
California 24
4.3 Map of the well field installed at the field site 26
4.4 Fence diagram constructed from inspection of cores and well logs 26
4.5 Particle size distribution of the aquifer core samples based on standard sieve analysis 27
4.6 Automated regression match of the leaky aquifer model to pump test observation
well response 29
4.7 Schematic of the injection system 38
4.8 Chlorinated organics delivery system 38
4.9 Schematic of the automated Data Acquisition and Control system 39
4.10 Daily variations in 1,1,1-TCA concentrations due to diurnal temperature fluctuation 42
5.1 Results from the Tracer2 natural gradient tracer test 46
5.2 Results from the Tracers natural gradient tracer test 46
5.3 Bromide tracer breakthrough and elution in the Tracer4 experiment 49
5.4 Bromide tracer breakthrough and elution in the Tracers experiment 49
5.5 Initial breakthrough of bromide in the Tracers experiment 50
5.6 DO breakthrough in the Tracer4 experiment 52
ix
-------
Number Page
5.7 DO and bromide breakthroughs at the S2 well in the Tracers experiment 52
5.8 Normalized breakthrough of bromide and TCE at the S1 well in the Tracers experiment 54
5.9 Breakthrough and elution in the Tracers experiment 54
5.10 Normalized breakthrough of bromide, trans-DCE, cis-DCE, and TCE at the S1 well
in the Tracerl 1 experiment 56
5.11 Normalized breakthrough of TCE during the Tracers experiment 56
5.12 Response at the S1 and S2 wells to the two-step vinyl chloride addition 57
5.13 RESSQ simulations of the injected fluid fronts which develop under induced
flow conditions of the tracer experiments with no regional flow 60
5.14 RESSQ simulations of the injected fluid fronts which develop under induced
flow conditions of the tracer experiments with a regional flow of 300 m/yr 60
5.15 Fit of the 1 -D advective-dispersion transport model to the breakthrough of DO
at the S2 observation well during the Tracer4 test 61
5.16 Comparison of predicted and observed effects of dissolved oxygen pulsing 63
6.1 Dissolved oxygen (DO) concentration response at the observation wells and
extraction well due to biostimulation of the test zone 68
6.2 Methane and DO response at the S2 observation well due to the biostimulation
of the test zone 69
6.3 The response at the S2 observation well resulting from 8- and 4-hr alternate
pulses of DO and methane 69
6.4 The response at the S1 observation well resulting from 8- and 4-hr alternate
pulses of DO and methane 70
6.5 Steady-state normalized TCE fractional breakthroughs during the first season's
biotransformation experiment, Biotrans4 71
6.6 Normalized bromide tracer breakthrough for the steady-state period shown in
Figure 6.5 71
6.7 Response at the S1 well of methane, DO, and trans-DCE in the second season's
biostimulation-biotransformation experiment, Biostim2 75
6.8 Response at the S2 well of trans-DCE, cis-DCE, and TCE in the second season's
biostimulation-biotransformation experiment, Biostim2 75
6.9 Fractional breakthrough at the S2 well of trans-DCE, cis-DCE, and TCE under
steady-state biotransformation conditions at the end of Biostim2 experiment 77
6.10 Production of trans-DCE oxide (epoxide), an intermediate of trans-DCE
biotransformation in the Biostim2 experiment 77
-------
Number Page
6.11 Response at the S1 well of TCE and cis-DCE to the reduction (100-200 hrs) and
termination (275-475 hrs) of methane addition 78
6.12 Response at the S1 well of trans-DCE and trans-DCE oxide (epoxide) to the
reduction (100-200 hrs) and termination (275-475 hrs) of methane addition 79
6.13 Steady-state fractional breakthroughs of the chlorinated prganics during the peroxide
addition experiment (0-553 hrs) and methane reduction (553-750 hrs) experiments 80
6.14 Decreases in normalized concentration of vinyl chloride, trans-DCE, and cis-DCE
at the S2 well in response to biostimulation in the third season 84
6.15 Transient responses of vinyl chloride and trans-DCE at the S1 well due to methane
pulsing 85
6.16 Transient responses of vinyl chloride and trans-DCE at the S3 well due to methane
pulsing 85
6.17 Decreases in cis-DCE and trans-DCE concentration during the first 380 hrs
of biostimulation in the third season 87
6.18 Methane and trans-DCE concentrations at the S1 well for periods of methane (0-350 hrs)
and formate (350-500 hrs) addition 87
6.19 Response of trans-DCE and cis-DCE to injection of: 1) methane, 2) formate,
3) methane, 4) methanol, and 5) no electron donor 88
7.1 Pseudo-first-order plots for trans-1,2-DCE epoxide hydrolysis at pH values of
5.0, 7.9, and 10.6 93
7.2 kODS as a function of pH 94
7.3 Concentrations of trans-DCO at observation wells S1, S2, S3, and the extraction
well in the Biostim2 experiment 95
8.1 Location of cores used in laboratory sorption studies 97
8.2 Particle size distribution corresponding to the synthesized bulk samples; observed in
aquifer solids from core SU-39-6 97
8.3 BET plots for nitrogen and krypton adsorption by silica standard, and nitrogen
adsorption by Moffett 20-40 fraction 99
8.4 Nitrogen adsorption/desorption isotherm for Moffett 20-40 fraction 101
8.5 Combined nitrogen adsorption and mercury intrusion cumulative pore size
distribution for Moffett 20-40 fraction 101
8.6 Flame-sealed ampule method for batch sorption studies 103
8.7 Sorption isotherm at increasing times for Moffett bulk solids and TCE 103
8.8 Increase of apparent K
-------
Number Page
8.9 One-day and thirty-day apparent Kj values for Moffett size fractions and TCE 105
8.10 Sorption isotherms and K2, and cell fractions by mixed cultureMMI (dry weight = 0.40 mg/ml) 120
9.3 Effects of methane concentration on percent TCE transformed into the nonvolatile
aqueous, CO2, and cell fractions by mixed culture MM2 (dry weight = 0.33 mg/ml) 121
9.4 Effects of methane concentration on percent TCE transformed into the nonvolatile
aqueous, CO2, and cell fractions by Methytomonas MM2 (dry weight = 0.12 mg/ml) ....121
9.5 Biomass concentration/mass transfer limitation study 123
10.1 Column design 127
10.2 Experimental design for sample mixing and column feeding 128
10.3 Dissolved oxygen concentration versus time for the control column (oxygen and TCE) 132
10.4 Methane concentration versus time for column 4 132
10.5 TCE degradation and concentration in the initial effluent from the control column
(oxygen and TCE) 133
10.6 TCE degradation in column 6 (low methane plus nutrients) 134
10.7 TCE concentration in the initial effluent from column 6 (low methane plus nutrients) 134
10.8 TCE degradation in column 3 (high/low methane plus nutrients) 135
10.9 TCE degradation in column 5 (high/low methane no nutrients) 135
10.10 TCE degradation in column 4 (pulse high/low methane plus nutrients) 136
10.11 The 1,2-DCA degradation and concentration in the initial effluent from column 7
(1,2-DCA: low methane plus nutrients) 137
xii
-------
Number Page
10.12 Percent degradation for TCE and 1,2-DCA in columns 6 and 7, respectively
(low methane plus nutrients for both TCE and 1,2-DCA) 137
10.13 Vinyl chloride removal efficiency in biostimulated column versus control column 138
10.14 Methane utilization versus VC removal in a methanotroph enriched column 139
10.15 Effect of hydrogen peroxide addition on methane and oxygen consumption
in batch exchange soil columns 141
10.16 Mass balance of 14C-labeIed TCE in batch exchange soil columns with fourteen
days between exchanges of pore liquid 142
11.1 Design of the continuous flow column 149
11.2 Results of column transport experiments and model simulations 153
11.3 The results of the first biostimulation-biotransformation experiment 155
11.4 The net methane and DO consumption in the biostimulation experiment 156
11.5 Fractional transformation of TCE during methane variation experiments 158
11.6 Labeled CO2 produced during methane variation experiments 158
11.7 Methane stimulation experiment with trans-DCE 160
11.8 trans-DCE epoxide concentration history 160
12.1. Comparison of the approximate analytical, semi-analytical, and case of no recirculation
solutions: V = 0.5 m/d, D = 0.02 m2/d, R = 1, Cp = 1.0 mg/l, x = 2.0 m,
I = 6.0 m, tp = 40.0 days, and q = 0.1 166
12.2. Bromide concentration breakthrough data of experiment Tracers observed at
S1 (squares), and simulated concentration history (solid curve) 169
12.3. Bromide concentration breakthrough data of experiment Tracers observed at
S2 (squares), and simulated concentration history (solid curve) 169
12.4. Bromide concentration breakthrough data of experiment Tracerl 1 observed at
S1 (squares), and simulated concentration history (solid curve) 170
12.5. Bromide concentration breakthrough data of experiment Tracerl 1 observed at
S2 (squares), and simulated concentration history (solid curve) 170
12.6. Curve matching with the two-parameter (V and D) classical advection-disperion model
to Tracers data observed at (a) S1, (b) S2, and with Cp as an additional
fitting parameter, (c) S1, and (d) S2 171
13.1 Arrival of fluid at distances along a direct path from the injection to observation wells under
hydraulic conditions of the field experiment based on simulations using RESSQ 178
13.2 Breakthrough of bromide, methane, and DO at the S1 observation well and the fit to
equation (13-4) 178
xiii
-------
Number page
13.3 Model simulation and observed methane and DO response at the S2 observation well
in the Biostiml experiment 181
13.4 Snapshots of the predicted distribution of methane-utilizing biomass during Biostiml
experiment 181
13.5 Model and field response at the S2 well resulting from the change from short to long
alternating pulse cycles 183
13.6 Model and field response at the S1 well resulting from the change from short to long
alternating pulse cycles 183
13.7 Predicted biomass concentration at the node 2.2 meters from the injection well due to
biostimulation with short and long pulse cycles 185
13.8 Snapshot of the predicted spatial distribution of DO and methane during a DO pulse
cycle after a steady-state biomass is achieved 185
13.9 Predicted steady-state biomass distributions achieved at two different pulse cycle
lengths 186
13.10 Simulated and observed DO and methane response at the S1 observation well during
the Biostim2 experiment 186
13.11 Simulation and field response of trans-DCE, cis-DCE, and TCE at the S1 well during
the second season's biostimulation experiment (Biostim2) 189
13.12 Simulations and field response of vinyl chloride, trans-DCE, and cis-DCE at the
S2 well during the third season's biostimulation experiment (BiostimS) 191
13.13 Simulation of vinyl chloride response at the S2 well using the equilibrium sorption
model 191
13.14 Simulation of the aqueous- and sorbed-phase vinyl chloride concentrations at the
S1 well using the non-equilibrium sorption model 193
13.15 Simulation of the aqueous vinyl chloride response at the S1 well using the equilibrium
sorption model 193
13.16 Simulation of the trans-DCE response at the S1 well to methane pulsing and formate
substitution for methane 194
14.1 Two possible bioremediation systems 198
14.2 Simulation of the aqueous phase concentration (boxes) and sorted phase
concentration (crosses) of vinyl chloride at the S2 well resulting from a
two step addition of vinyl chloride 202
14.3 Comparison of biorestoration versus pump-and-treat of rapidly degrading, moderately
sorbed compounds, such as trans-DCE, based upon the scheme illustrated
in Figure 14.1a 204
14.4 Comparison of biorestoration versus pump-and-treat remediation of TCE
contamination with and without biostimulation based upon the scheme
illustrated in Figure 14.1b 204
xiv
-------
TABLES
Number Page
1.1 Biological Processes and Environmental Conditions Under Which Different
Compounds May Be Transformed by Bacteria 3
4.1 Sequence of Experiments and Processes Studied in the Field Evaluation 23
4.2 Parameter Estimates from Regression Analysis Using the Leaky-Aquifer Model 29
4.3 Comparison of Aquifer Parameters Derived from Leaky-Aquifer and Constant-Pressure
Models 30
4.4 Groundwater Chemistry: Major Ions and Other Parameters 32
4.5 Trace Chemical Composition of the Groundwater from the SU-39 Site 33
4.6 Organic Carbon Content of Moffett Aquifer Solids 35
4.7 Summary of Analytical Methods and Detection Limit 41
4.8 Summary of Analytical and Injection Performance 41
4.9 Comparison of Variability of Absolute and Normalized Concentration Data for
Observation Well S2 43
5.1 Estimates of Regional Velocities Based on the Results of the Natural Gradient
Tracer Experiments 47
5.2 Comparison of Bromide Tracer Tests Under Induced Gradient Conditions 48
5.3 Summary of Induced Organic Tracer Tests Performed 53
5.4 Residence Times and Retardation Factors for the Chlorinated Organic Compounds
Based on the Time Required to Achieve 50% Fractional Breakthrough 58
5.5 Residence Times and Retardation Factors for the Chlorinated Compounds
Based on Center-of-Mass Estimates 58
6.1 Experiments and Processes Studied 65
6.2 Tracers Experiment - Percentage Breakthrough of the Organic Solutes and Bromide
at the Observation Wells 74
6.3 Percentage Biotransformation in the Second Season's Biostimulation-Biotransformation
Experiments 81
xv
-------
Number Page
6.4 Tracer! 2 Experiment Percentage Breakthrough of Chlorinated Solutes and
Bromide at the Observation Wells 83
6.5 Percentage Biotransformation - Third Field Season 89
8.1 Mass Fraction of Particle Sizes Used to Prepare Synthesized Bulk Samples 98
8.2 Specific Surface Area and Internal Porosity for Moffett Field Size Fractions 100
8.3 Partitioning Coefficients and Aqueous Solubilities of the Solutes Studied with
Moffett Bulk and Synthesized Bulk Solids 106
9.1 TCE Transformation Rates 119
9.2 Rate Coefficients for Methane-Oxidizing Mixed Cultures 119
9.3 Mixed Culture TCE Transformation Rates 122
9.4 Methylomonas MM2 TCE Degradation Rates 124
9.5 . Mixed Culture MM 1 TCE Degradation Rates 124
10.1 4 The Original Experimental Design 131
10.2 The Distribution of 14C in the Initial Effluent Removed from Column 6 after 14TCE
Addition to the Influent Was Stopped 139
10.3 Feed Conditions for Peroxide Study 140
11.1 Experiments and Processes Studied 152
11.2 Model Results: Mobile-Immobile Zone Model 153
11.3 TCE Biostimulation Results 161
11.4 trans-DCE Biostimulation Results 161
12.1 Estimated Transport Parameters for the Bromide Breakthrough Data of Experiments
Tracers and Tracer! 1 167
12.2 Estimated Transport Parameters Obtained by the Classical A-D Model for the Bromide
Breakthrough Data of Experiment Tracers 168
13.1 Basic Features of the Non-Steady-State Biotransformation Model 173
13.2 Input Parameters Used in the Biostimulation Model Simulations 179
13.3 Model Setup Parameters 180
13.4 Comparison of Adjusted Model Parameters for the Biostimulation Experiments
(WellS2) 187
13.5 Model Parameters for Simulation of Chlorinated Organics in Biostim2 189
13.6 Model Parameters for Simulation of Chlorinated Organics in BiostimS 192
xvi
-------
ACKNOWLEDGMENTS
The authors thank the personnel of the U.S. Navy, especially the Public Works Department at the
Moffett Naval Air Station for allowing the Field Site to be located on their base. They have cooperated fully
in helping us solve the many logistical problems associated with performing a field study of this type.
Public works officer, John Heckman, has been especially helpful in this regard.
We would also like to thank the staff at the Oakland Office of the California Regional Water Quality
Control Board for permitting us to perform these experiments. Thomas Berkins, Steve Morse, and Steve
Ritchie have provided helpful suggestions which have aided in the design of the experiments.
Members of the Kerr Laboratory of EPA have also provided input to the experimental design and
the characterization of the test zone, and have conducted laboratory studies which have helped to guide
the field experiments. John Wilson, Michael Henson, and Barbara Wilson provided helpful technical
information. We also thank Jack Keeley and William Dunlap, who served as Project Officers during the
early stages of this work.
Graduate students and postdoctoral students, who have made significant contributions to the
development and characterization of the field site include, in addition to those listed as authors, the
following: William Ball, Christoph Buehler, Helen Dawson, Meredith Durant, and Barton Thompson.
xvii
-------
SECTION 1
INTRODUCTION
Lewis Semprini, Dunja Grbic-Galic, Perry McCarty, and Paul Roberts
The in-situ remediation of aquifers contaminated with halogenated aliphatic contaminants, com-
monly known in water supply as chlorinated solvents, is a promising alternative in efforts to protect
groundwater quality. Chlorinated aliphatic compounds are frequently observed in groundwater. In a sur-
vey of 945 water supplies, Westrick et al. (1984) found trichloroethylene (TCE), tetrachloroethylene
(PCE), cis- and/or trans-1,2-dichloroethylene (cis-, trans-DCE), and 1,1-dichloroethylene to be the most
frequently appearing compounds other than trihalomethanes. Approaches for the restoration of aquifers
contaminated by these compounds based on extracting the contaminated groundwater by pumping and
subsequently treating at the surface have been shown to be effective, but often entail great expense and
also a risk of transferring the contaminants to another medium, i.e., the atmosphere. To circumvent these
difficulties, in-situ treatment of the contaminants has become a potentially favorable alternative, with
investigations centering on promoting biotransformation of the contaminants.
Our group at Stanford University has assessed under field conditions the capacity of native
microorganisms, i.e., bacteria indigenous to the subsurface environment, to metabolize halogenated
synthetic organic contaminants, when proper conditions are provided to enhance microbial growth.
Specifically, the growth of methanotrophic bacteria was stimulated in a field situation by providing ample
supplies of dissolved methane and oxygen. Under biostimulation conditions, the transformation of repre-
sentative halogenated organic contaminants, such as trichloroethylene (TCE), cis-1,2-dichloroethylene
(cis-DCE), trans-1,2-dichloroethylene (trans-DCE), and vinyl chloride (VC), was assessed by means of
controlled addition, frequent sampling, quantitative analysis, and mass balance comparisons.
The field demonstration study was conducted at Moffett Naval Air Station, Mountain View, CA, with
the support of the Kerr Environmental Research Laboratory of the U.S. Environmental Protection
Agency, and with the cooperation of the U.S. Navy. To provide guidance for and confirmation of the field
work and to obtain a more basic understanding of key microbial and physical processes, laboratory
experiments were also performed at Stanford University's Water Quality Control Research Laboratory.
This report summarizes the results of both the field study and the associated laboratory studies.
BACKGROUND
The in-situ restoration of aquifer contaminated with hydrocarbons is not a new idea. Raymond
(Raymond, 1974; Raymond et al., 1976) pioneered the development of the process for the in-situ
reclamation of aquifers contaminated by liquid fuels. This work indicated that by promoting the proper
conditions in the subsurface (i.e., by the addition of oxygen and nutrients), a native population of
microorganisms was stimulated that degraded the hydrocarbon contaminants. The microorganisms used
the hydrocarbon contaminants as primary substrates for growth.
In-situ biorestoration of aquifers contaminated by halogenated aliphatic compounds requires a
somewhat different approach, since in most cases the halogenated aliphatic compounds cannot be
-------
utilized by native microorganisms as primary substrates for growth. However, they can be degraded as
secondary substrates by microorganisms which utilize another primary substrate for growth. The in-situ
bioremediation of aquifers contaminated with organic compounds has been thoroughly reviewed recently
(Lee et al., 1988; McCarty, 1988; and Wilson et al., 1986). These articles review the basic microbial,
chemical, and physical processes that affect in-situ bioremediation and provide an adequate overview of
the basic information on in-situ bioremediation, so that information is not repeated here.
Substrate Utilization Concepts
In this work, the evaluation of the in-situ restoration of aquifers contaminated with chlorinated
aliphatic compounds was studied. McCarty (1988) reviewed biological processes and environmental
conditions under which different organic compounds are transformed. Contaminants may be transformed
as 1) primary substrates for growth, or 2) by co-metabolism or secondary substrate transformation.
Primary Substrate Utilization-
When the contaminant is utilized as a primary substrate, it is a source of energy and carbon for the
microorganisms. Under these conditions in-situ biotransformation may be promoted by supplying the
appropriate electron acceptor and nutrients to the subsurface.
Secondary Substrate Transformation-
The cells grown on the substrate can sometimes degrade other compounds, called secondary
substrates, even though the secondary substrates do not afford sufficient energy to sustain the microbial
population (McCarty et al., 1981). However, when the contaminants themselves do not act as a suitable
substrate for cell growth, it is sometimes possible to grow a bacterial population by providing an easily
metabolized substance, called the primary substrate. To transform contaminants as secondary substrates,
both the primary substrate and the electron acceptor must be added to the treatment zone, if one or the
other is not present naturally. This results in an even more complex system than when contaminants are
degraded as primary substrates.
Cometabolism is one form of secondary substrate transformation in which enzymes produced for
primary substrate oxidation are capable of degrading the secondary substrate fortuitously (Brock et al.,
1984). We will use the more general term, "secondary substrate transformation," in this report to describe
the biotransformation of the chlorinated aliphatics, although the specific transformations dealt with in this
work are thought to result from cometabolism.
Transformation Overview--
Table 1.1 summarizes environmental conditions and biological processes under which different
compounds may be transformed. Transformations as primary substrates and by co-metabolism
(secondary substrate transformation) are presented. While a few chlorinated compounds can be used as
primary substrates for growth, the majority are transformed as secondary substrates. Less chlorinated
compounds are shown to be more easily transformed by oxidation processes, while more highly chlori-
nated compounds are more easily transformed by reduction processes.
The transformation of the chlorinated organics by methanotrophic bacteria is an oxidation process.
As shown in Table 1.1, TCE can be transformed by either oxidative or reductive processes. One of the
advantages of the oxidation process is that complete mineralization of the contaminant can result,
whereas the reductive process often produces less chlorinated compounds, some of which are more
difficult to degrade under reducing conditions.
The present work, to our knowledge, was the first attempt to demonstrate in-situ biorestoration of
chlorinated organics as secondary substrates. The success of this venture depended largely upon the
existence of a microbial community indigenous to the subsurface environment, that is capable of metabo-
lizing organochlorine compounds as secondary substrates.
-------
TABLE 1.1. BIOLOGICAL PROCESSES AND ENVIRONMENTAL CONDITIONS UNDER
WHICH DIFFERENT COMPOUNDS MAY BE TRANSFORMED BY BACTERIA
Primary Substrates
Aerobic and anaerobic: Glucose, acetone, isopropanol, acetate, benzoate, phenol
Aerobic primarily: Alkanes, benzene, toluene, xylene, vinyl chloride,
1,2-dichloroethane, chlorobenzene
Co-Metabolism (Secondary Substrates!
Oxidations: Trichloroethylene, dichloroethylene, dichloroethane, vinyl
chloride, chloroform
Reductions: 1,1,1-Trichloroethane, trichloroethylene, tetrachloroethy-
lene, dichloroethylene, dichloroethane, carbon tetrachlo-
ride, chloroform, DDT, lindane, polychlorinated biphenyls
Source: McCarty (1988).
Methanotrophic Bacteria
Wilson and Wilson (1985) reported the transformation of trichloroethylene by aerobic microbial
communities from soil which were grown on natural gas. Since the main constituent of natural gas is
methane, this experiment focused attention on methanotrophs, a physiological group of aerobes which
utilize methane (and a limited number of other C1 compounds) as their carbon and energy sources.
Characteristics and Occurrence-
Methanotrophs have been studied extensively, not only owing to their growth substrate specificity
and the resulting specialized ecological niche they occupy in the environment, but also because of their
capability to cometabolically oxidize a whole range of various organic non-growth substrates.
The characteristics of methanotrophs (and methylotrophs as a broader physiological group) have
been summarized in numerous reviews (Quayle, 1972; Anthony, 1975; Colby et al., 1979; Wolfe and
Higgins, 1979; Hanson, 1980; Higgins et al., 1981; Hou, 1984a).
In nature, methanotrophs can be found in boundary regions between anaerobic habitats, which are
the sources of methane, and the aerobic ones from which oxygen for respiration and methane oxy-
genation can be extracted. One could expect these bacteria to be present in subsurface environments,
where aerobic and anaerobic "pockets" frequently coexist in relatively small areas. This expectation was
confirmed in our field investigations.
Since the supplies of their major carbon and energy sources, as well as of oxygen, are frequently
uncertain, numerous methanotrophs have developed the capability of forming resting stages (exospores
or cysts) which enable them to survive the periods of famine (Hou, 1984a). This ability makes them
excellent candidates for subsurface existence.
Depending on fine intracellular membranous structures and the type of carbon assimilation pathway,
methanotrophs can be divided into two broad groups: type I, with ribulose monophosphate biosynthetic
pathway and bundles of vesicular membranous discs distributed throughout the cell; and type II, with
serine biosynthetic pathway and paired membranes aggregated at the periphery of the cell, running
throughout it parallel with its outer membrane. There are discrepancies from these two general descrip-
tions, and microorganisms can be found which combine characteristics of both types.
-------
Enzymes--
Regardless of the type, energy generation from the major substrate, methane, is initiated through
the activity of a powerful enzyme, methane monooxygenase (MMO). MMO incorporates one atom from
the oxygen molecule into methane to create methanol. The other oxygen atom needs to be reduced to
water, and therefore, the MMO enzyme requires a reduced electron/proton carrier (a requirement typical
for monooxygenases in general).
The enzymes that catalyze subsequent oxidations include methanol dehydrogenase, which
catalyzes formation of formaldehyde from methanol; formaldehyde dehydrogenase, which converts
formaldehyde to formate; and formate dehydrogenase, which produces CO2 (upper row in Figure 1.1).
Formaldehyde is a branching point in methanotrophic metabolism, since it is used not only for energy
generation, but can be channeled into biosynthesis as well.
Tonge et a!. (1975) first isolated and partially purified a MMO (from Methylosinus trichosporium, a
type II obligate methanotroph); it consisted of three components (one soluble and two particle-bound),
contained both iron (cytochrome c) and copper as cofactors, and could oxidize ethane, propane, butane,
and carbon monoxide in addition to methane. Subsequently, it was shown that cell-free systems derived
from resting cell suspensions of methane-grown methanotrophs epoxidized alkenes to the correspond-
ing 1,2-epoxides, and hydroxylated alkanes to the corresponding secondary alcohols and methyl
ketones; the MMO activity resided mainly in the particulate fraction of methanotrophs (Patel et al., 1979).
Stirling et al. (1979) demonstrated hydroxylation of mono- and dichloromethane by the MMO from this
bacterium.
CELL CONSTITUENTS
:H3OH HCHO _HCOOH
NADH NAD X XH NAD NADH NAD NADH
TCE 7=r> TCE Epoxide
E! - METHANE MONOOXYGENASE
E, « METHANOL DEHYDROGENASE OR
' ALCOHOL OXIDASE
E3 - FORMALDEHYDE DEHYDROGENASE
E4 - FORMATE DEHYOROGENASE
X = a proton and electron carrier
Figure 1.1. Oxidation of methane by methanotrophs and the initial transformation step in TCE oxidation
(adapted from Hou, 1984a, and Henry and Grbic-Gali6, 1986).
-------
In addition to these paniculate fraction activities, a soluble MMO system was purified from Methy-
lococcus capsulatus (Bath), a type I obligate methanotroph, and shown to catalyze the oxygenation not
only of methane, but also of chlorinated and iodinated methanes, as well as nonhalogenated alkanes,
alkenes, ethers, cyclic compounds, alicyclics, and aromatics (Colby et al., 1977). This enzyme was also
resolved into three components: "A", containing iron and acid-labile sulfide; "B", a protein of undefined
function; and "C", a flavoprotein containing non-heme iron and acid-labile sulfur, which could accept elec-
trons from NADH. Later, Stirling and Dalton (1979) found a soluble MMO in Methylosinus trichosporium as
well; this enzyme also had a very broad substrate specificity. Methylobacterium sp. CRL-26, a facultative
methanotroph, was demonstrated to contain both a paniculate and a soluble methane monooxygenase
activity; the soluble MMO oxidized alkenes, including methylated and brominated alkenes, CO, ethers,
cyclic and aromatic hydrocarbons; and hydroxylated alkanes (C2 to C8), chloro-, fluoro-, bromo-, and nitro-
methanes, and nitro-alkanes (Patel et al., 1979,1982; Patel, 1984).
The dispute about the importance and specificity of paniculate versus soluble MMO has not been
resolved yet, but it seems the expression of the two enzyme classes depends on the growth conditions.
For example, paniculate MMO activity is associated with low oxygen tension and copper excess, whereas
soluble MMO seems to be expressed when oxygen is plentiful but copper is limiting (Hou, 1984a).
From the above discussion, it becomes obvious that MMO is a remarkable enzyme with an unusually
broad substrate specificity. Yet, methanotrophs, especially the obligate ones, can grow only on a very
limited range of compounds. The process which attacks the non-growth substrates is termed cometabo-
lism (first suggested by Horvath, 1972), and results only in a partial change of the compounds.
Competition between the growth substrate and the non-growth substrate for the active site of the
relevant enzyme is very likely to happen, and can greatly influence the transformation of the non-growth
compound. If the transformation occurs in a complex natural microbial community, which consists of vari-
ous populations (e.g., methanotrophs and heterotrophs) working together, complete degradation of a
cometabolic substrate can be achieved if its early transformation products can be further transformed by
subsequent members of the food chain. This seemed to be the case in the experiment described by
Wilson and Wilson (1985), where TCE was completely degraded by a microbial community from soil. The
phenomenon warranted further investigation, because of its potential applicability.
TCE Transformation-
Parallel investigations on the capability of methanotrophic- heterotrophic communities to degrade
TCE were started in 1985 in the Kerr Research Laboratory (EPA, Ada, Oklahoma), in the Environmental
Engineering and Science Program (Civil Engineering, Stanford University), in the Environmental
Sciences Division of Oak Ridge National Laboratory (Oak Ridge, Tennessee), and in a consulting com-
pany, Cambridge Analytical Associates (Bioremediation Systems Division), Boston, Massachusetts.
Henry and Grbi£-Gafi£ (1986) at Stanford University suggested a possible mechanism of initial
transformation of TCE (Figure 1.1, second row; Figure 1.2) which was based on the well-known capability
of methanotrophs to epoxidize alkenes (Hou et al., 1979a; Hou, 1984b) and which involved epoxidation
of the TCE molecule by methane monooxygenase, and subsequent transport of the epoxide
intermediate out of the methanotrophic cell, where it could be subjected to various other transformations.
The suggested mechanism was accepted by Little et al. (1987a, b; 1988) at Oak Ridge National
Laboratory, who used it to explain transformation of TCE by pure cultures of methanotrophs, and incorpo-
rated it into the general pathway of TCE transformation by methanotrophic-heterotrophic mixed cultures,
involving epoxidation of TCE by methanotrophs, abiotic hydrolysis of the epoxide to nonvolatile products,
and subsequent heterotrophic degradation of the products to CO2, chloride, and water. The details of
the pathway are still being investigated by the research group at Stanford, using pure cultures of methan-
otrophs and mixed methane-grown cultures. There seem to be differences in transformation intermedi-
ates depending on the microbial groups involved.
-------
BACTERIAL CELL
AQUEOUS ENVIRONMENT
TCE
CO
CARBON MONOXIDE
HCOO-
FORMATE
CI2HC-COO-
DICHLOROACETATE
Figure 1.2. Initial oxygenation of TCE by methanotrophs and the transport of the epoxide from the cell
(from Henry and Grbic-Gafic, 1986).
Fogel et al. (1986) at Cambridge Analytical showed that methane-grown enrichments from fresh-
water sediments could degrade TCE, cis- and trans-1,2-dichloroethylene, vinyl chloride, and vinylidene
chloride. Early degradation products were water-soluble intermediates; the authors theorized that these
intermediates could be further mineralized to CC>2 and other inorganic products. The more highly chlori-
nated (and therefore highly oxidized) compound, tetrachloroethylene (PCE), was not transformed.
Henson et al. (1987,1988) at Kerr Research Laboratory demonstrated that a range of halogenated
methanes, ethanes, and ethylenes (except PCE) could be completely degraded by mixed methane-
utilizing cultures derived from soil (Figure 1.3). However, the details of the methanotrophic transformation
process still remained unclear. Important questions remained regarding competition between methane
and TCE for the MMO active site, nutritional requirements of mixed cultures and methanotrophs
themselves, and other important aspects of this microbial activity. Many of these questions have been
resolved during our work on this project, and will be described and explained in the subsequent sections.
Secondary Transformation of TCE by Other Groups of Bacteria
In addition to methanotrophs, other microbial groups show capabilities of transforming halogenated
alkenes as non-growth substrates. Henson et al. (1987) used propane as the growth substrate to enrich
TCE-degrading mixed cultures from soil. Henry et al. (1988) obtained ethylene-grown cultures which
could degrade TCE, but at much lower rates than methanotrophs.
An entirely different group of microorganisms-pseudomonads-were shown by Pritchard and
coworkers at Environmental Research Laboratory, EPA, Gulf Breeze, Florida (Nelson et al., 1986,1987,
1988) to transform TCE while growing on aromatic compounds (toluene or phenol); the microorganisms
were enriched from freshwater samples. Some of the laboratory strains of pseudomonads capable of
growing on toluene showed similar activity. Depending on the microbial strain, either a dioxygenase
(toluene dioxygenase), or a monooxygenase is involved in the TCE oxidation process.
-------
100
CD
20
0
0
10 15 20
TIME (DAYS)
30
Figure 1.3. Degradation of chlorinated alkenes by mixed methane-grown microbial enrichments from
soil (from Henson et al., 1987).
It could be theorized that numerous and diverse microorganisms, if only they contain an oxygenase
with relatively broad substrate specificity, could transform halogenated aliphatic compounds. If this were
true, this transformational capability could be quite widespread in natural habitats, which would open
exciting possibilities for bioreclamation of environments contaminated by chlorinated solvents. This
hypothesis warrants further investigation aimed at understanding the active microorganisms, including
their physiology, environmental and nutritional requirements, adaptability, and biotic-abiotic interactions,
before full advantage can be taken of cometabolism for purposes of aquifer restoration.
Anaerobic Transformations of Halogenated Aliphatic Compounds
Halogenated compounds that are common groundwater contaminants are frequently transformed
by both abiotic and biotic processes. Such transformations generally occur most readily under active
anaerobic conditions that result from simultaneous contamination with other organic chemicals that are
readily susceptible to anaerobic transformations, especially methane fermentation. These reductive
transformations can be of significance to the methanotrophic process for in-situ biodegradation that is the
subject of this report. Anaerobic transformations lead to the formation of less halogenated compounds
that are most readily degraded by methanotrophs. Thus, a brief review of anaerobic transformations is in
order.
The occurrence of reductive transformation of halogenated aliphatic compounds in groundwater
was first demonstrated in 1981 (Bouwer et al., 1981). Since then, several investigations have elucidated
this process so that the environmental conditions required and the transformation products to be
expected are now much better understood. In general, anaerobic transformations of halogenated alkanes
and alkenes lead to the production of a wide variety of less-halogenated products (Bouwer and McCarty,
1983; Gossett, 1985; Parsons and Lage, 1985; Vogel and McCarty, 1985,1987; Barrio-Lage et al., 1986;
Belay and Daniels, 1987), and the rates of transformation are faster under conditions where methane is
formed. Methane-producing bacteria (methanogens as opposed to methanotrophs) are implicated in
-------
many of these transformations (Belay and Daniels, 1987), but other anaerobic bacteria can participate in
them as well.
Anaerobic transformations of halogenated solvents follow pathways illustrated in Figure 1.4.
Compounds such as tetrachloroethene (PCE) and trichloroethene (TCE) are sequentially reduced to form
1,2-dichloroethene (both cis- and trans-isomers), and then vinyl chloride. Vinyl chloride can also be trans-
formed anaerobically, but the rate is very slow. Another common solvent, 1,1,1-trichloroethane (TCA),
can be transformed abiotically into 1,1-dichloroethene and acetic acid (Vogel and McCarty, 1987). TCA
can also be reduced biologically to 1,1-dichloroethane (1,1-DCA), and then into chloroethane. However,
1,1-DCA is relatively stable and the rate of transformation into chloroethane is slow.
As reported by others, the reduced products such as 1,2-dichloroethene and vinyl chloride are
much more readily biodegraded by methanotrophs than the parent compounds, PCE or TCE. Thus, it
may be desirable to have the anaerobic transformations occur first in an aquifer before remediation with
methanotrophic bacteria is considered. Indeed, there may be circumstances under which it is desirable to
encourage the reductive process. In other cases, one may simply take advantage of naturally occurring
reductions that occur in aquifers under the proper conditions.
Implementation of Secondary Substrate Transformation for Biorestoration
Our study evaluated an oxidation process in which methanotrophic bacteria initiate the transforma-
tion of the chlorinated aliphatics. In this case, methane and oxygen must both be added to the treatment
zone in order to enhance selectively a native methanotrophic population that transforms the chlorinated
organics as secondary substrates.
cci2 =cci2
*
CHCI=CCI2
I
CHCI=CHCI CH2 =CCI2
\ /
CH2 =CHCI
cci3cci3
Figure 1.4. Anaerobic transformation pathways for selected chlorinated aliphatic compounds [after
Vogel et al. (1987) and McCarty (1988)]. Arrows with "a" indicate abiotic transformations;
other arrows represent biotic transformations.
-------
Rate Concepts-
Numerous models have been developed to describe the rate of biotransformation of halogenated
organics as secondary substrates. One of the simplest models used when the contaminant concentration
is low is given by McCarty (1984):
dC2/dt = - (k/Ks)XaC2 (1-1)
where C2 is the concentration of the secondary substrate, Xa is the microorganism concentration, and
k/Ks is a ratio of constants that is equivalent to a second-order rate constant. The k value (time'1) repre-
sents the maximum specific substrate utilization rate per unit mass of microorganisms per unit time, and Ks
is the half velocity constant (mass/volume) which represents the organisms' affinity for the substrate. At
low substrate concentrations, the rate of transformation depends on the concentrations of both microor-
ganisms and contaminant.
Biostimulation-
The basic objective of enhanced in-situ biorestoration by secondary substrate transformation is to
increase transformation rates in the treatment zone by stimulating a large bacterial population. Equation
(1-1) illustrates how this may be accomplished. One means is to increase the concentration of the
microbial population of interest. This process, referred to as biostimulation, is accomplished by adding the
appropriate electron donor and electron acceptor. In the field experiment the biostimulation of a naturally
occurring methanotrophic population required the addition of methane and oxygen.
Enhanced transformation rates might also be achieved through the addition of minor nutrients,
adjusting chemical conditions, such as pH, or physical conditions such as temperature. Selection of a
specific type of native methanotroph or inoculating the treatment zone with a type that rapidly transforms
the chlorinated organics is another means of increasing rates. The information required for such methods
of enhancement was not available, and therefore was not evaluated in our field study. Some of the
laboratory studies performed as part of this research work focused on obtaining this information.
A third factor which affects the rates of transformation is the contaminant concentration. Physical
processes such as sorption onto the aquifer solids act to lower aqueous concentrations, thus decreasing
transformation rates, especially in the case where only aqueous-phase transformation occurs. If biotrans-
formation rates are rapid compared to desorption rates from the aquifer solids, the rate at which both the
aqueous and sorbed contaminants are transformed may be reduced. To account for this effect, laboratory
sorption studies using aquifer solids from the test zone were performed as an important component of this
work.
Approaches to Treatment-
The enhanced in-situ biotransformation approach taken in this work required creating an in-situ
treatment zone in an aquifer that represented conditions of a real contamination incident. Conceptual
models for in-situ treatment systems are given by Lee et al. (1988) and McCarty (1984). The conceptual
model of McCarty (1984) is shown in Figure 1.5. Several different forms of biological treatment are shown:
1) surface treatment, 2) a well bore reactor, and 3) treatment in the contaminated aquifer. Our evaluation
represented treatments near the well bore and in the contaminated aquifer.
The process evaluated, however, is not limited to subsurface treatment. Surface bioreactors can
also be integrated with in-situ treatment. Since in-situ treatment would most likely require the extraction of
contaminated groundwater, surface treatment in combination with in-situ treatment would most likely be
used in practice.
-------
SUBSTRATE
AND
NUTRIENTS
1
— "
*
\ ABOVE GROUND
BIOREACTOR
BYPASS
L
^ AQUIFER
-*— BIOREACTOR
y// v//
**—
_^
\
%
Wf
1
51
Nl
y y//
^ — "
EXTRACTION
WELL
SUBSTRATE
AND
NUTRIENTS
WELL-CASING
BIOREACTOR
Figure 1.5. Possible systems for biologically treating a contaminated aquifer (McCarty, 1984).
RESEARCH OBJECTIVES
The overall objective of this work was to assess the efficacy of the proposed method for enhancing
the in-situ degradation of the halogenated aliphatic compounds. The specific objectives of this research
project were to:
1) Demonstrate whether the proposed method of promoting the biodegradation of
chlorinated aliphatic compounds is effective under controlled experiments performed
in-situ, in an aquifer representing conditions typical of groundwater environments;
2) Quantify the rate of decomposition, and identify intermediate transformation prod-
ucts, if any;
3) Bracket the range of conditions under which the method is effective, and establish
criteria for dependable treatment of a real contamination incident;
4) Determine factors that affect biodegradation rates by means of basic microbiological
studies on methanotrophic bacteria in the laboratory;
5) Quantify the sorption of the chlorinated aliphatics on the aquifer solids in controlled
laboratory experiments; and
6) Simulate the in-situ biodegradation process using a mathematical model that
incorporates key biological and transport processes, and adapt suitable models for
this purpose.
10
-------
In order to meet these objectives a combined field, laboratory, and modeling study was performed.
The field studies focused on objectives 1, 2, and 3, while the laboratory studies focused on objectives 2,
3, 4, and 5. The modeling effort of objective 6 permitted comparisons between the field and laboratory
results.
REPORT ORGANIZATION
In keeping with EPA's required format, overviews of the report's contents and findings are provided
in the Executive Summary (preceding the Table of Contents), and in the Summary and Conclusions
(Section 2) and the Recommendations chapter (Section 3).
Section 4 presents the methodology of the field experiments that was developed to provide a
convincing and objective demonstration of the proposed method. Details of the Automated Data
Acquisition and Control system used to continuously monitor the field experiments are presented in this
section. The geologic, hydrogeologic, chemical, and microbiological characteristics of the field site are
also summarized.
Section 5 presents the results of tracer experiments performed in the test zone. The results of
tracer tests under natural gradient and induced gradient conditions are presented. These tests charac-
terized the transport of the chemicals of interest in subsequent biostimulation and biotransformation
experiments and quantified the solute residence times.
Section 6 presents the results of biostimulation and biotransformation experiments conducted at
the field site; this section contains the principal findings of the field demonstration phase of the project. In
Section 6, the response of the natural methanotrophic community following biostimulation with methane
and oxygen, and the resulting transformations of the target organochlorine compounds are described.
Section 7 deals with the chemistry of the single transformation intermediate identified in the field work.
Section 8 presents the results of laboratory experiments that quantified the sorption characteristics
of the aquifer solids and other properties affecting transport. The laboratory studies of microbial growth
and transformation of the target compounds are summarized in Sections 9-11. Research conducted with
enriched and pure cultures are covered in Section 9. Column studies using aquifer solids from the Moffett
site are presented in Sections 10 (batch columns) and 11 (continuous-flow columns).
A new mathematical modeling approach for simulating transport with recycle of extracted solutes is
presented in Section 12. The simulation of the transport, biostimulation, and biotransformation observed
in the field demonstration are presented in Section 13, together with the development of the mathemati-
cal models employed. An orientation toward practical application of this biorestoration approach is offered
in Section 14, which describes a scenario for application to aquifer remediation.
11
-------
SECTION 2
SUMMARY AND CONCLUSIONS
Paul Roberts, Dunja Grbic-Gali6, Perry McCarty, Gary Hopkins, and Lewis Semprini
This project demonstrated conclusively the efficacy of enhanced in-situ biotransformation of
chlorinated alkenes by microbial communities comprising methanotrophic and heterotrophic bacteria. It
proved easy to stimulate the growth of the native population of methanotrophic bacteria by providing
oxygen and methane in the proper amounts. Once stimulated, the mixed methane-grown communities
metabolized the target chlorinated compounds at rates that ranged from moderately rapid (with a half life
on the order of a few days) to very rapid (with a half life of less than one day). In most cases, transforma-
tions appeared to progress completely to stable, harmless end products, although in one instance a
transitory intermediate product was identified.
The project was organized as a multidisciplinary effort, which encompassed evaluations at a small-
scale field demonstration site, as well as detailed studies in the laboratory. The major conclusions of the
various aspects of the research, and their interrelationships, are summarized below.
FIELD DEMONSTRATION METHODOLOGY
An effective methodology was developed to evaluate objectively and quantitatively the efficacy of
the biorestoration approach for stimulating the growth of the desired bacterial population and transforming
the target organic compounds under natural conditions at a field site. The methodology entails creating a
flow field dominated by pumping from an extraction well, while introducing solutes in known amounts at a
nearby injection well and measuring concentrations regularly at the injection, extraction, and intermediate
observation points (Section 4). Interpretation of biotransformation behavior could then be made by
qualitative examination of the concentration histories of the various solutes at the several monitoring
points, comparing results under biostimulation conditions with results obtained under similar conditions in
the absence of biostimulation measures (Section 6). These interpretations could then be substantiated
by quantitative mass balances.
A custom-designed, automated data acquisition and control system (Section 4), constructed by the
project team for this purpose, provided continuous records of accurate data over sustained periods that
enabled us to compute mass balances with relative errors of only a few percent.
Incorporating experimental controls and quantitative mass balances to the extent possible is an
absolute prerequisite for meaningful experimentation in the field as well as in the laboratory. Only field and
laboratory experimentation of this kind can provide a reliable engineering scientific basis for evaluating and
designing in-situ biorestoration strategies.
12
-------
SITE CHARACTERIZATION
The site chosen for the field demonstration, at Moffett Naval Air Station, offered a near-ideal com-
bination of characteristics (Section 4). The site was representative of a typical situation of groundwater
contamination in the San Francisco Bay area and elsewhere, in which a shallow sand-and-gravel aquifer is
contaminated by chlorinated aliphatic compounds widely used as solvents. Drilling logs revealed that the
aquifer at the test site consisted of a layer of silt, sand, and gravel, approximately 1.2 m thick, at shallow
depth (approximately 5 m below the ground surface), well confined above and below by a silty clay layer of
low permeability. The formation groundwater was also of appropriate composition for the field experi-
ments. The water was moderately saline and was substantially contaminated by chlorinated organic com-
pounds, mainly 1,1,1-trichloroethane, but was devoid of the chlorinated alkenes (TCE, 1,2-DCE, and VC)
chosen as target compounds for this study.
Sustained pump tests showed that the transmissivity was sufficiently high (approximately
100 m2/day) to permit extracting water at the design rate (approximately ten liters per minute) without
excessive drawdown at the extraction well. Natural gradient tracer tests showed that the local groundwater
velocity was approximately two meters per day. Preliminary mathematical modeling of the flow field,
imposing a forced gradient on the natural flow field to simulate injection/extraction operations, showed
that injection and extraction rates of approximately one liter per minute and ten liters per minute,
respectively, would be sufficient to satisfy the two main requisites for the field experiment from the
hydraulic point of view: 1) complete permeation by injected fluid of the aquifer in the observation zone
between the injection and extraction points (i.e., minimum dilution by native groundwater in that zone);
and 2) complete recovery of the injected fluid at the extraction well (to assure accurate mass balances).
Extensive tracer tests (Section 5) undertaken to quantify transport velocities and residence times in
the test zone confirmed that the aquifer was permeated virtually completely by the injected fluid in the
observation zone, as evidenced by complete breakthrough of bromide tracer at the observation wells,
under the chosen experimental conditions. Further, the overall mass balances, comparing the amounts of
tracer injected and extracted, demonstrated that the tracer recovery in the extracted water was essentially
complete. This was necessary to assure the validity of the experimental approach.
The hydraulic residence times between the injection well and the three observation wells in the test
zone, quantified by tracer tests under the forced gradient conditions, were found to be in the range of 8 to
27 hrs; the residence time between the injection and extraction well was 25 to 40 hrs, depending on the
pumping rate. These residence times were later found to be suitable for quantifying the transformation
rates of interest in this work. The retardation factors for the organic solutes, evaluated from relative
mobility data obtained in the field, were in the range of two to twelve.
FIELD DEMONSTRATION OF BIOSTIMULATION AND BIOTRANSFORMATION
The biostimulation and biotransformation evaluations conducted in the field (Section 6) were con-
sistent in all major respects with expectations based on the laboratory results and theory.
It was confirmed that a native population of methane-oxidizing bacteria could be stimulated by
introducing dissolved methane and oxygen into the aquifer in proper amounts, without any other
supplementary nutrients. Within ten days, the population of methane utilizers had grown to the point of
utilizing substantial amounts of methane, and within another five days methane utilization was complete.
Clogging of the injection well and borehole could be controlled effectively by alternately pulsing methane
and oxygen, a strategem which also served to spread the microbial growth more uniformly over a larger
domain around the injection point. The ratio of oxygen consumption to methane consumption was
2.5 g/g, which is consistent with literature data and laboratory results on methanotrophic growth.
13
-------
Transformation of the organic target compounds ensued immediately following the beginning of
methane utilization, increasing with time as the bacterial population grew, and ultimately reaching a steady-
state value that differed among the compounds. The steady-state transformations observed during the
final year's field work, quantified by normalization to the bromide fractional breakthrough, were as follows:
TCE, 10 to 29%; cis-DCE, 33 to 45%; trans-DCE, 85 to 90%; and VC, 90 to 95%. Of the values cited, the
lower end of the range represents the nearest observation point (1 m distant, 8 h residence time),
whereas the upper end of the range represents more distant observation points with longer residence
times (2 to 4 m; 16 to 27 h). The injected concentrations of the target compounds were in the range of 50
to 100 u.g per liter. A chlorinated alkane present as a background contaminant, 1,1,1-trichloroethane
(TCA), was not degraded to any appreciable extent.
GC analysis of water samples during active biotransformation of trans-DCE provided evidence of an
intermediate transformation product identified in laboratory studies (Section 7) to be the epoxide of trans-
DCE, which was present in amounts equivalent to a few percent of the parent compound. No other
intermediate products were identified.
Termination of the methane feed was followed by cessation of transformation activity on approxi-
mately the same time scale as that of organic transport. This suggested that the microbial population
remained active in the absence of methane for only a short time before ceasing to transform the target
organic compounds (Section 6). However, the concentration oscillations in response to the alternate
pulsing of methane and oxygen suggested methane inhibition. Close examination of the concentration
variations indicated that the organic compounds were transformed more fully when the methane
concentration was lower.
Substitution of either formate or methanol for methane effectively diminished the effects of meth-
ane inhibition and increased transformation rates of organic compounds (trans- and cis-DCE) temporarily
(Section 6). However, the substitute electron donors could not sustain growth of the active microbial
population indefinitely, which was demonstrated by the decrease in transformation activity after approxi-
mately 50 to 100 hrs without methane.
After a prolonged period without methane feed, the population of active methane utilizers declined
substantially. However, even after eight months without methane feed an appreciable fraction of the
population grown during biostimulation continued to survive. When the test zone was restimulated after
the eight-month starvation period, methane utilization commenced immediately and continued to increase
until utilization was complete within three days.
Employing peroxide as a means of increasing the electron acceptor dose permitted operating at a
higher rate of methane feed and increased biological growth, but did not enhance the rate of transfor-
mation of the target organic compounds.
Overall, the field results confirmed the existence of a natural population of methane oxidizers that
could be stimulated by introducing methane and oxygen. Moreover, it was demonstrated that quantitative
comparisons could confirm the extent of transformation within five percent. Finally, it was observed that
substantial transformation of TCE, cis- and trans-DCE, and VC occurred within a distance of a few meters
and residence times on the order of several days.
IDENTIFICATION OF INTERMEDIATE PRODUCT
The intermediate product observed in the field work was definitely identified as the epoxide of trans-
DCE, namely trans-1,2-dichlorooxirane (1,2-DCO) in the laboratory (Section 7). The compound was
synthesized and purified to provide a standard for quantitative analysis. The compound was found to
degrade via a hydrolysis mechanism, with a half-life of approximately 4 days at 18°C.
14
-------
SORPTION
The retardation factors quantified from the field data were consistent with the results of laboratory
studies of sorption (Section 8). The sorption of the organic solutes by aquifer core samples from the
Moffett site confirmed that sorption equilibrium was approximately linear, justifying the use of a distribution
coefficient for interpreting and reporting the sorption equilibrium data. Sorption was strongest for TCE
and weakest for VC, among the compounds studied. The retardation factors calculated from the labora-
tory sorption data agreed closely with those estimated from the transport experiments conducted in the
field. The extent of sorption was approximately equal for all grain size fractions, but equilibrium was
reached much more slowly in large grains than in small ones. This finding illustrates that the deviations
from sorption equilibrium owing to rate limitations may be an important factor influencing transport and
biotransformation behavior. The slow rates of adsorption and desorption need to be taken into account
by incorporating the appropriate rates into transport and biotransformation models used for simulation and
design.
GROWTH AND TRANSFORMATION RATES
Biotransformation studies of several kinds were conducted to characterize the populations of
methanotrophic bacteria at the field site. These included studies with enriched mixed cultures and iso-
lated pure cultures grown on nutrient media (Section 9), as well as experiments with the natural population
grown on aquifer solids under conditions simulating the field experiments, in batch exchange soil columns
(Section 10) and a continuously fed column (Section 11).
The experiments with mixed cultures enriched from Moffett samples (Section 9) evaluated the ability
of populations grown on several substrates-methane, propane, and ethylene~to transform TCE as the
target compound. Methane oxidizers transformed TCE about one hundred times faster than ethylene
oxidizers; propane oxidizers showed no ability to transform TCE. The transformation of TCE in both the
methane- and ethylene-oxidizing mixed cultures was complete, although about five percent (methane-
grown cultures) to ten percent (ethylene-grown cultures) of the original TCE remained in a nonvolatile
aqueous fraction that has yet to be completely differentiated. Pure cultures of both methane- and
ethylene-oxidizing organisms were isolated from the corresponding mixed cultures, and were shown to
be capable of transforming TCE. Acetylene inhibited both methane oxidation and TCE transformation,
implying that the methane monooxygenase (MMO) enzyme was responsible for both processes.
Experiments with varying methane concentration revealed that high methane concentration slows or
stops the transformation of TCE, presumably through the competition between methane and TCE for the
MMO enzyme active sites. The properties of the various cultures enriched from the Moffett aquifer
material differed somewhat with respect to transformation rates and the effects of environmental variables
on rates. In some, but not all, cultures, TCE concentrations above 10 mg/l were found to inhibit the rates
of both methane oxidation and TCE transformation. Cultures containing storage compounds (PHB
granules) were able to transform TCE as rapidly in the absence of methane as in the presence of low
methane concentrations. This observation suggests the importance of the availability of reducing power
in sustaining the normal functions of MMO. Extremely high concentrations of oxygen exercised a slight
inhibitory effect as well.
Batch experiments with cultures grown on Moffett solids (Section 10) largely confirmed the results
of the experiments with cultures grown on nutrient media, and served to demonstrate the applicability of
the results to the specific case of the aquifer at the Moffett site. The ability to conduct concurrently a
number of parallel experiments, including controls, enabled the project team to evaluate the effects of
parameters such as methane concentration, nutrient requirements, and the choice of electron acceptor
and electron donor, under conditions simulating the real subsurface environment.
The experiments showed conclusively that a population of the native methanotrophic community
could be stimulated in a porous medium consisting of Moffett aquifer material, without the addition of
microbial seed or nutrients. The subsurface environment contained sufficient nitrate and phosphate as a
15
-------
nutrient source; the column experiments showed that transformation rates were not enhanced by
supplying additional nitrogen and phosphorus.
Columns fed with methane and oxygen began to utilize the methane within 7 days, and partial ICE
transformation ensued within 80 days, reaching approximately 20% after a year. No significant amounts of
intermediate transformation products of TCE were found. Mass balances on columns previously
saturated with sorbed TCE and then purged with water for prolonged periods, with and without
biostimulation, showed that the TCE was removed from the solids twice as fast by the combination of
biodegradation and desorption as by desorption alone. In similar experiments conducted with
1,2-dichloroethane (DCA), the compound degraded to about the same extent as TCE, but concentrations
responded more rapidly, as DCA was less strongly sorbed. Vinyl chloride (VC) degraded still more rapidly
than DCA, being removed about one-half as fast as methane itself. Within two days, VC degradation was
essentially complete.
The concentration observations generally support the hypothesis of enzyme competition, and
showed that methane should not be present at too high a concentration. It was further demonstrated that
methane does not have to be added continuously for TCE degradation to proceed; TCE transformation
persisted for several days after methane depletion, and indeed seemed to be more rapid at very low
methane concentrations.
The rate of TCE transformation by suspended mixed and pure methanotrophic cultures increased
by an order of magnitude when the growth medium contained the complexing agent EDTA, compared to
a growth medium containing nutrients but no EDTA (Section 9). The field results (Section 6) also suggest
that TCE degradation can be increased by intermittently supplying an alternative source of reducing
power (e.g., a single-carbon compound such as formate or methanol), to maintain transformation activity in
the absence of methane. Use of a high concentration of hydrogen peroxide (85 mg/l) as an alternative
source of oxygen (electron acceptor) proved to be inhibitory to both methane utilization and TCE
transformation (Section 10). The batch growth and transformation experiments allowed evaluation of
alternative strategies and operation modes for the field experiments. Results from these efforts helped
reconcile microbiological theory and research conducted with cultures grown on nutrient media, on the
one hand, with the growth and transformation behavior observed in an aquifer environment, on the other.
The continuous flow column experiments (Section 11) simulated closely the conditions of the field
experiment. The experiments were conducted with continuous feed of methane and oxygen (unlike the
batch column experiments, Section 10), with a hydraulic residence time of one day, corresponding
approximately to the travel times between the injection well and the observation wells at the field site. In
the initial biostimulation with methane and oxygen, substantial methane utilization commenced 20 days
after beginning the methane feed, increasing rapidly over the next 5 days to the point where methane was
completely utilized. The mass ratio of oxygen consumption to methane consumption was approximately
2.5:1. Following attainment of complete methane utilization, transformation of TCE began, ultimately
reaching approximately 20%. The extent of transformation of TCE was not improved by raising the
influent methane concentration from 4.5 to 6.5 mg/l. On the other hand, TCE transformation did increase
substantially (from 22% to 29%) by temporarily ceasing the methane input for a period of up to 20 days.
The extent of trans-DCE transformation under similar conditions was much greater than that of TCE (85%
vs 22%). Transformation of trans-DCE in the continuous column persisted unabated for more than 40
days after cessation of methane input.
MATHEMATICAL MODELING
A non-steady-state model developed for simulating the results of the field experiments proved
useful in interpreting field data and comparing with laboratory results (Section 13). The model
incorporated advection, dispersion, sorption with and without rate limitation, and the microbial processes
of substrate utilization, growth, halogenated aliphatic transformation, and competitive inhibition. The
transport was simplified by assuming one-dimensional, uniform flow, as a computational compromise to
permit more rigorous representation of the biological processes. Input parameters were estimated based
16
-------
on the results of the laboratory research, or on values from the literature. Only the initial population of
methane-utilizing bacteria was allowed to vary as an unconstrained fitting parameter.
The model was able to simulate the dynamic behavior of the system very closely, including the
concentration oscillations stemming from the pulsed addition of methane and oxygen. The observed
transient responses of the target organic compounds were matched closely by the model simulations,
using rate parameters that were consistent with the values inferred from rate experiments conducted in
the laboratory. Model simulations of the effects of competitive inhibition and rate-limited sorption and
desorption also agreed well with the observed behavior, showing substantial attenuation of the organic
solute concentrations.
PROJECT INTEGRATION
The coordination between concurrent laboratory and field work proved extremely beneficial. This
arrangement facilitated the cooperation between researchers from the relevant disciplines-microbiology,
biochemistry, organic chemistry, process engineering, and hydrology-in a manner that assured proper
consideration of all of the crucial factors in designing and interpreting the experimental program. This
interactive mode also promoted a flexible approach to periodically reevaluating research priorities in the
various research tasks: laboratory work could be redirected to address relevant mechanistic questions
raised in the field work, and the field demonstration could be modified to take advantage of potential
improvements suggested by the laboratory research. For example, the field demonstration program was
modified to evaluate the transformation of the DCE isomers and VC after laboratory experiments had
shown that these compounds were transformed more rapidly than TCE.
Moreover, the results and conclusions from laboratory research and field work were in general
agreement. This provided greater confidence in describing the governing mechanisms and relevant
processes than otherwise would have been possible. For example, the column experiments agreed with
the results of the field biostimulation and biotransformation in virtually every respect. In particular, the
column experiments (Sections 10 and 11) and the field observations of biostimulation and biotransforma-
tion (Section 6) agreed remarkably well. The areas of close agreement encompassed the delay in the
onset of methane utilization during the initial stage of biostimulation (20 vs 10 days), the observed ratio of
oxygen consumption to methane consumption (2.8 vs 2.5 g/g), and the extents of transformation of TCE
(approximately 20%), trans-DCE (approximately 85%), and VC (approximately 95%). The batch column
results confirmed the field results with respect to the effects of variable methane concentration, and
replacement of oxygen with peroxide. However, the laboratory and field tests differed with respect to the
effect of terminating the methane feed. The laboratory column experiments indicated that transformation
of the halogenated organic compounds persisted for weeks, whereas the field observations exhibited a
much more rapid decline in secondary substrate transformation. The laboratory results from the sorption
equilibrium and rate experiments were consistent with the parameters estimated from the dynamic tests
conducted in the field, which confirmed that sorption is a major factor governing the mobility and availability
of organic contaminants in an aquifer. Thus sorption considerations need to be taken into account in
evaluating aquifer restoration strategies.
In establishing such connections between laboratory investigations of processes and field
investigations of behavior under natural conditions, mathematical modelling was an essential tool in facil-
itating the transfer of information. In developing appropriate models, it was essential to strike a judicious
compromise between the competing goals of accurate process representation and computational feasi-
bility. The mathematical model chosen for the present application stressed relatively complete
representation of the relevant biological processes, and compensated with a highly simplified model for
advective/dispersive transport. The rate parameter values estimated from the mathematical model simula-
tions of the field data agreed well with the growth rates and TCE transformation rates observed in the
laboratory.
17
-------
SECTION 3
RECOMMENDATIONS
Paul Roberts, Dunja Grbic-Gali6, Perry McCarty, Gary Hopkins, and Lewis Semprini
This project has focused on evaluating the potential of aquifer restoration by enhancing the bio-
transformation of halogenated alkenes, e.g. TCE, cis- and trans-DCE, and VC, through the biostimutation
of a community of methane-oxidizing bacteria.
The laboratory research improved the understanding of the processes of microbial growth and
transformation, as well as their interrelationships with other processes. It also provided the scientific basis
for employing these processes in the subsurface environment. One group of recommendations identi-
fies directions for extension of this kind of applied research to further strengthen our understanding of
this potentially useful microbial community, as well as other subsurface phenomena related to aquifer
restoration.
The demonstration of the methodology under field conditions confirmed the efficacy of the
biostimulation approach. Accordingly, another group of recommendations points out steps toward apply-
ing this technology for large-scale cleanup, and for further refining methodologies for field evaluations.
RECOMMENDATIONS FOR PROCESS RESEARCH
1. There is a need to further clarify the roles of the various organisms in the methanotrophic com-
munity, including the heterotrophic members. Achieving this goal requires further investigation into
the physiology and metabolism of the various organisms, focusing on both enriched mixed cultures
and pure cultures. The aim of this research is to be able to specify completely and unambiguously
the optimum conditions for growth and secondary substrate transformation. The formation and
function of the methane monooxygenase enzyme needs to be elucidated more completely, as
does the phenomenon of competition between methane and other compounds for enzyme sites.
The potential for sustaining enzyme activity for prolonged periods in the absence of methane, by
adding sources of reducing power such as formate or methanol or by biostimulating in such a
manner as to create internal sources, deserves further attention as a means of enhancing the
degradation of the targeted chlorinated organics.
2. It is essential to delineate the degradation pathways of halogenated compounds more fully, and to
reach quantitative understanding of the transformation rates. The roles of abiotic transformation
reactions of the epoxide intermediates, such as hydrolysis, need to be appreciated more fully, as
well as the biotransformation processes. The kinetics of the individual transformation processes
and the influence of environmental factors on the relative rates of alternative transformation
processes must be thoroughly comprehended to enable confident prediction of the intermediate
and final products. Further, it is crucial to understand the effects of both abiotic and biotic
processes in mediating the geochemical conditions that, in turn, govern which processes can
transform the contaminants at appreciable rates. Future research should be committed to
understanding the interrelations among geochemical conditions and chemical and biological
transformations.
18
-------
3. It is necessary to continue research on the sorption of the targeted organic compounds on real
aquifer solids. Biotransformation processes cannot be adequately understood without sufficient
understanding of other processes that govern solute transport, distribution, and local chemical
conditions in the subsurface environment. Sorption is a major factor influencing the organic
contaminants' relative mobility and distribution between the solids and the pore water, and is an
important factor to consider in evaluating the prospects for successful in-situ biotransformation.
Research must be continued to provide sufficient insight into the sorption equilibrium, the distri-
bution of the sorbed solute within the solid phase, and the relative rates of sorption and desorption,
and to demonstrate the validity of the principles for the full spectrum of prospective target
chemicals. These insights should permit accurate formulation of expressions representing the
sorption/desorption phenomena in mathematical models used to predict transport and
transformation. Research in this direction is crucial to successful application of biorestoration
technology in the usual situation in which the contamination is largely associated with the solids
rather than in solution.
4. It is also essential to verify more conclusively the accuracy of the mathematical model formulation of
the biotransformation process, and to refine this formulation if necessary. In particular, the
questions related to prolonged transformation activity in the absence of methane and to competi-
tion for the enzyme and methane inhibition of cometabolic transformation need to be clarified, and
appropriate rate models must be developed.
5. Research and development should be undertaken to adapt the methanotrophic biotransformation
technology for application to remediation in the vadose zone. There is every reason to believe that
this modification is possible, as it is inherently easier to introduce methane and oxygen in the gas
phase than in solution.
RECOMMENDATIONS FOR APPLICATION
6. The approach evaluated here merits full consideration for application to real aquifer remediation
cases. This technology should be considered as an alternative where the contamination consists in
large part of the compounds for which methanotrophic transformation has been shown effective in
the demonstration phase of the present work: namely, VC, trans- and cis-DCE, and possibly TCE.
An initial application where the rapidly degraded compounds trans-DCE and VC are present as major
contaminants would be ideal. The success of such applications of course will depend on the
prevailing conditions and the complexity of the situation. Candidate scenarios for early application
of this innovative technology should be well characterized with respect to hydrology, geochemistry,
and contaminant distribution. The candidate sites should be reasonably homogeneous, and the
contaminant distribution should be well-defined and reasonably uniform throughout an extensive
domain; however, it is not essential to restrict the application to a shallow zone such as that
investigated at the Moffett site; indeed a deeper zone would be favorable because it would permit
operation at higher pressures and correspondingly higher concentrations of oxygen and methane.
Any application should be preceded by thorough characterization: not only the usual coring, water
sampling and analysis, field hydrogeology and geochemistry, but also supporting studies of the
kind illustrated by the laboratory tasks of this research: i.e., batch and continuous column studies of
biotransformation rates conducted with core materials from the site and quantification of sorption
equilibrium and rates. Such laboratory studies should be continued concurrent with the aquifer
restoration operation to aid in evaluating its effectiveness and to assess the experience for
generalization to subsequent operations of this kind.
7. Additional development and optimization should be conducted at small scale at a well-instrumented
site; the Moffett site is ideally suited for this purpose. Specific tasks to be undertaken in the near
term include the following: a) determine whether TCE transformation can be further enhanced by
the addition of a chelated mineral or EDTA directly to the test zone; b) evaluate the efficacy of
substituting a primary substrate other than methane, such as methanol, once the methanotrophic
population has become established; and c) assess alternative methods of methane addition that
19
-------
would circumvent or minimize enzyme competition by avoiding high methane concentrations in the
biostimulated zone (for example by adding methane and oxygen at different locations and allowing
them to blend in the aquifer). This additional applied research should be undertaken prior to, or
concurrent with, any large-scale application of the methanotrophic approach for aquifer
biorestoration.
8. Further development of mathematical models incorporating all processes and phenomena relevant
to biorestoration of large-scale subsurface domains is necessary to support aquifer remediation for
different treatment scenarios, designs, and operating policies. Models for this purpose will have to
account better for the possible effects of aquifer heterogeneity~in physical, chemical, and biological
terms~and ultimately will have to be modified to extend to two and three dimensions.
9. A national survey of contamination sites should be undertaken to establish a data base on the
ubiquity of methanotrophic communities to permit evaluation of the potential for widespread
application of aquifer restoration with natural populations.
10. For application to cases where such natural populations are absent, methodologies should be
developed and demonstrated to achieve colonization of the subsurface environment with intro-
duced cultures. Criteria should be elaborated for the preparation of these seed cultures.
11. A technoeconomic evaluation should be undertaken to evaluate the cost-effectiveness of the
biorestoration technology described in the present work. The technoeconomic evaluation should
encompass a range of scenarios chosen to represent typical conditions at real contaminated sites,
and should combine the practical experience of consultants with operations experience in aquifer
remediation with the scientific insights embodied in the applied research developed from the
present study. This evaluation will serve to assess the promise of the proposed alternative in the
overall context of the national remediation effort, and will help identify the kinds of scenarios for
which the methanotrophic biorestoration is best suited.
12. The methanotrophic biotransformation approach deserves further development and evaluation as a
means of surface treatment, alone or in combination with subsurface application of the same
technology.
20
-------
SECTION 4
FIELD EXPERIMENT METHODOLOGY AND SITE CHARACTERIZATION
Gary Hopkins, Lewis Semprini, Douglas Mackay, Paul Roberts, and Robert Johns
This section will discuss the experimental methodology used in the field evaluation. Results of
the detailed site characterization will be presented and the field site instrumentation will also be described.
EXPERIMENTAL METHODOLOGY
The experimental methodology developed to meet the goals of the field study was as follows:
A. Select a representative demonstration site based on available information regarding
regional hydrology and geochemistry, and considering practical and institutional
constraints;
B. Characterize the site by means of coring, pump tests, sampling and analysis of the
native groundwater;
C. Construct a system of wells for injection, extraction, and monitoring of water at the
demonstration site;
D. Design and install an automated system for sampling and analysis of the ground-
water at the demonstration site;
E. Determine the velocity and direction of groundwater flow under natural gradient
conditions, by means of bromide tracer tests;
F. Assess the mobility of the chlorinated aliphatics, relative to bromide tracer, at the
demonstration site and quantify residence times in the system under injection/
extraction conditions.
G. Stimulate the growth of native methane-oxidizing organisms by injecting dissolved
methane and oxygen (biostimulation mode); and
H. Assess the transformation of the chlorinated aliphatics under biostimulation condi-
tions.
This methodology provided a staged approach for evaluating the proposed technology. The ini-
tial phases of the study (A-E) focus on selecting the field site and characterizing its physical, chemical,
microbiological and hydraulic properties. The later phases of the experiment involve biostimulating
methane-oxidizing bacteria in the test zone and evaluating the degree of transformation of specific con-
taminants of interest.
21
-------
The information obtained during the early phases of the experiments was critical to the success of
subsequent evaluation experiments, which were dependent on the ability to run controlled experiments
in the subsurface. The hydraulic information obtained in pump tests and tracer experiments was required
in designing a fluid injection and extraction system used to create the in-situ treatment zone. The
chemical, physical and microbiological characteristics of the test zone also indicated whether favorable
conditions existed for the Die-stimulation of a native population of methane-oxidizing bacteria. These data
were necessary for determining whether a controlled evaluation of the proposed technology could be
performed at the selected site.
The basic approach of the evaluation experiments was to create a test zone in the subsurface. This
was accomplished in our experiments as follows. A series of injection, extraction, and monitoring wells
were installed within a shallow confined aquifer, as shown in Figure 4.1. An induced flow field was created
by the injection and extraction of fluid. The chemicals of interest for a specific experiment were metered
into a stream comprising 10 to 15% of the extracted groundwater and then reinjected. The
concentrations of the specific chemicals were monitored at several locations, including the injected fluid,
the three monitoring wells, and the extracted fluid. The spatial and temporal responses of the chemicals in
the test zone were determined by frequent monitoring, using an automated data acquisition and control
system located at the test site.
The sequence of field experiments using this approach is outlined in Table 4.1, where the row
numbers are referenced as Stages 1 through 5 in the following text. The initial experiments (Stage 1)
studied the transport of bromide ion as a conservative tracer. The experiments determined fluid resi-
dence times in the system, the amount of dispersion, and the recovery of the injected fluid at the extrac-
tion well. In later experiments (Stages 2 and 3), bromide, dissolved oxygen and the target chlorinated
compounds were injected simultaneously. The retardation factors of the different chemicals with respect
to bromide, owing to sorption, were determined. The transformation of the chlorinated aliphatic com-
pounds in these experiments was evaluated based on comparisons with the bromide tracer by comparing
the steady-state fractional breakthrough achieved at the monitoring wells. These Stage 3 tracer experi-
ments therefore served as quasi controls, permitting a comparison of the observed responses before and
after the the test zone was biostimulated. Results of these transport experiments are presented in
Section 5.
SAMPLING
INJECTION WELLS
WELL
EXTRACTION
WELL
SAMPLING
WELLS INJECTION
• WELL
0 -i
1^
M 4
I -
CL
SJ
1
,
AY
!/ /// t
>ND AND
IAVEL
/ y
\
yxx/yv
177/7 y/V/yV/ryVyV yyyy/yv/yyyyyyyyyyy/yyy/y
CLAY SI SI S2 S3 P N3 N2 Nl NI
* 1 i i i i i i i i r | i i
Distance from well SI, m
Rgure 4.1. Vertical section of the test zone.
22
-------
TABLE 4.1. SEQUENCE OF EXPERIMENTS AND PROCESSES STUDIED IN THE
THE FIELD EVALUATION
Injected Chemicals
Process Studied
1) BT
2) Br
3) Br'+TCEa + O2
4) CH4 + O2 + (nutrients) + TCEa
5) Transient experiments
Advection/Dispersion
Retardation/Dispersion (TCA - Elution)
Retardation (Transformation)
Biostimulation + Biotransformation
Biotransformation + Competitive Inhibition
aTCE, cis-DCE, trans-DCE, vinyl chloride.
The biostimulation and biotransformation experiments were performed in Stage 4. Biostimulation
involved the addition of methane, oxygen, and nutrients (if required), to stimulate the growth of methane-
consuming bacteria in the test zone. The transient response of the different chemical components was
monitored, as previously discussed. This experiment determined: 1) how easily the methane-oxidizing
bacteria were stimulated and whether nutrients were required, 2) stoichiometric requirements of oxygen
to methane, 3) information on the kinetics and the rate of growth, and 4) the areal extent over which
biostimulation was achieved.
Biotransformation in response to biostimulation was directly demonstrated by measuring the
simultaneous decrease in concentration of the chlorinated aliphatics at observation locations. In the sec-
ond and third seasons of testing, the chlorinated organics were continuously added after quasi-steady-
state conditions were achieved in Stage 3 of the experiments. Thus, the degree of biotransformation was
assessed as concentration decreased to steady-state values during transport through the biostimulated
zone.
The final stage of the experiments (Stage 5) involved transient experiments to determine how
changes in operating conditions affected biotransformations. Transient experiments included terminating
methane addition, substituting formate and methanol for methane, and adding hydrogen peroxide
instead of oxygen into the test zone. The transient responses of the chlorinated organics to these
changes were monitored at observation locations. Results of Stages 4 and 5 are presented in Section 6.
The extents of biotransformation of the chlorinated organics were determined using several
methods: 1) comparisons of steady-state fractional breakthrough concentrations at monitoring points,
2) comparisons of steady-state breakthroughs obtained before (Stage 3) and after biostimulation, and
3) complete mass balances on the amounts injected and extracted. In methods 1 and 2 the steady-state
fraction (normalized) breakthroughs were calculated by dividing the compound's observed concentration
by the injection concentration. Steady-state conditions were operationally defined as an experimental
period of several days to a week during which concentrations remained essentially constant. This period
followed a transient period, in which concentrations were changing in response to a stimulus, such as a
change in input concentration.
Estimates of the degree of biotransformation for methods 1 and 2 were made using equation 4-1:
Percent biotransformed = (1 - Cf)0rg/Cf ,br) x 100% (4-1)
23
-------
where for method 1, Cfi0rg is the mean steady-state fractional breakthrough of the organic solute after
biostimulation and Cf,t>r is the mean steady-state fractional breakthrough of bromide over the same time
interval. For method 2, the fractional breakthrough of the organic solute before biostimulation is
substituted for fractional bromide breakthrough in equation 4-1.
The field evaluation described above consisted of a series of stimulus-response experiments. In
order to perform these experiments, a test zone with controlled hydraulic conditions was created in the
subsurface through the injection and extraction of groundwater. The stimulus in the experiments was
achieved by injecting known quantities of the chemicals of interest in a controlled manner into the test
zone. The response was measured in terms of the chemical concentrations of fluid samples taken at
observation wells and the extraction well.
In order to perform controlled field experiments, the site was carefully characterized with respect to
its physical, chemical, and biological properties. The site was instrumented to permit injection and
monitoring of the chemical concentrations of interest.
CHARACTERIZATION OF THE FIELD SITE
Field Site Description
After a reconnaissance study of several sites, a location at the Moffett Naval Air Station, Mountain
View, Ca., was chosen (Figure 4.2). The site, designated SU-39, located on the lower part of the Stevens
Creek alluvial fan, is approximately 3 km south of the southwest extremity of San Francisco Bay. The sur-
face elevation at the site is 8.5 m above mean sea level.
Figure 4.2. Location of the field site, SU-39, at the Moffett Naval Air Station, Mountain View, California.
24
-------
The experimental site is located in a region where the groundwater is contaminated with several
organic solutes for which this biorestoration method might be applied. The area of groundwater contami-
nation shown in Figure 4.2 represents the 1 mg/l TCE contour of the "A" Aquifer delineated in January,
1983 (Canonie Engineers, 1983). The plume contains 1,1,1-trichloroethane (TCA) and trichlproethylene
(TCE) at concentrations up to 100 mg/l, measured at points 700 and 1000 m from the SU-39 site. Thus, if
effective, the treatment method may have direct use in the area where it was evaluated.
GEOLOGIC CHARACTERISTICS
The geologic characteristics of the test zone were examined using core samples and drilling well
logs. Figure 4.3 shows the location of the wells installed at the test site. A series of exploratory wells
(1,SI,3,4,5,6) were installed in July 1985, using the hollow stem auger drilling method. Cores were
obtained using 2" pitcher barrels that were pushed ahead of the drill bit. The 6 test drillings identified a
shallow, confined aquifer which is known as the "A" Aquifer, the shallowest of several in the region. Weil
togs indicate that the aquifer is confined between silty clay layers, and is approximately 1.2 m thick; the
top border is located 4.4 to 4.6 m below the ground surface, and the bottom ranges from 5.3 to 5.7 m
below the surface.
Figure 4.4 is a fence diagram constructed from cores and well logs of the fully penetrating wells SI,
P, Nl, 5, and 6. Well SI is the injection well and P is the extraction well used in the experiments. The well
logs generally show similar lithologic profiles. The uppermost 0.5 m consists of silty sand with pebbles up
to 8 cm in diameter. This surface layer is underlain by approximately 4 m of silt and clay of a brownish-black
to olive gray color, indicating that the sediment contains organic material. The bottom of this sequence of
the upper confining layer is marked by a clayey sand that commonly separates the silt and clay overburden
from the underlying aquifer.
The aquifer consists of fine- to coarse-grained sand and appears poorly sorted in most cores. The
aquifer, as indicated by the slotted well screens in Figure 4.4, is located 4.3 to 5.8 m below the surface.
Gravel lenses with pebbles up to 2.5 cm in diameter occur in some cores within the sand layers. Due to
the presence of gravel, intact cores were difficult to obtain. Cores were often lost over the depth interval
from 4.7 to 5.2 m below the surface. This zone is considered to have the highest gravel fraction.
The aquifer thickness and composition differs spatially. Along the north-south series of wells (SI, P,
Nl) the aquifer is composed of a mix of sands and gravels, of fairly uniform thickness. However, profiles in
the areas of wells 5 and 6, to the east, show less gravel and more sand, indicating some aquifer thinning
compared to the SI through Nl region. Cores obtained from wells farther east (not shown) showed that the
permeable layer in that location consisted primarily of sand, with no gravel present.
A layer of dark greenish-gray silty clay underlies the aquifer (top at 5.9 m below the surface). While
no well was drilled through this clay/silt layer at the project site, other studies in the vicinity have shown that
this layer is approximately 7 m thick and is underlain by another aquifer (Canonie Engineers, 1983).
The particle size distributions of aquifer cores are shown in Figure 4.5. Core samples taken from
wells 4 and 6 at a depth of 5.5 m and 5.3-5.5 m, respectively, exhibit similar distributions of particle sizes,
with a large fraction of the solids being coarse to medium sands and gravel. The core sample from well 5 at
a depth of 4.1-4.3 m has a greater fraction of fine sand and silt, which is consistent with well log
observations. Petrographic analysis shows that the aquifer solids consist of rock fragments of the parent
rock of the Santa Cruz Mountains. These include include graywackes, cherts, and volcanics of
eugeosynclinal (slope) origin (Franciscan Series).
. The observations at the test site are consistent with geologic studies in the region. The inter-
layering of coarse and fine sediments in the Santa Clara Valley results from changes in sea level caused by
world-wide climate fluctuations (Atwater et al., 1977). During times of high sea level (warm periods), fine-
grained estuarine sediments were deposited in the valley, resulting in the clay and silt aquitards. During
25
-------
o
SCALE, meters
Figure 4.3. Map of the well field installed at the field site.
Slotted screen
I2m
Figure 4.4. Fence diagram constructed from inspection of cores and well logs.
26
-------
Grave
100 p
80 -
60 -
I
.
«
j
TH
Sand
Coarse (o
medium
Fine
SMI
U.S standard sieve sizes
i * i i i
. i
II
i
|,
\
^
\
L. .
X '
'|
i
— t -
" s "i
o
• Well-5(1 3.5-14') A
-^
^ 1
'' s^1
i !
_i_ .
i
i
T" '
1
it
1
1
4 i
X. .. I
Clay
§ S5 i s i §
000 ° 0 °
Grain diameter, mm
WelM(18') • Well-6 (17,5-19*)
Figure 4.5. Particle size distribution of the aquifer core samples based on standard sieve analysis.
times of low sea level (glaciation in northern latitudes), these sediments were covered by coarser-grained
alluvial deposits that form the aquifers. At the study site, the aquifer consists of alluvial sediments
deposited during the last 5000 years. The aquifer is spatially heterogeneous, with the composition
varying appreciably over short distances. The test zone appears to have the structure of a buried stream
channel, containing sand and gravel in some areas and only sand in others. This structure is common in
alluvial aquifers, which are characterized by deposition from multiple channels with constantly shifting loci
of deposition, resulting in discontinuous lenses of sand and gravel (Press and Siever, 1974).
Hydraulic Characteristics
Maps of the regional piezometric surface of the "A" Aquifer have been reported by Canonie Engi-
neers (1983). The hydraulic gradient is northward at about 4.5 m per km. Piezometric measurements
that were made with the original wells of the test zone indicated that the aquifer was confined with a
piezometric surface 2.5 m above the top confining layer (6.4 m above mean sea level). The magnitude
and direction of the gradient in the test zone was in the range of regional values. The original gradient
estimates had a large level of uncertainty due to the short spatial resolution. Wells 11,12,13, installed in
August 1986, provided a more accurate estimate of the local gradient due to the greater distances
between wells. A gradient of 0.0032 in a northerly direction was estimated using these wells.
A series of pump drawdown tests was performed to characterize hydraulic properties of the aquifer.
The tests determined the transmissivity, which permitted estimates of the hydraulic conductivity and the
natural gradient groundwater velocity. The possible influence of leakiness, barriers, and abnormalities was
also examined.
The drawdown tests were conducted using a Hydrologic Analysis System Model SE 200A well test
device (In-Situ, Inc.) that consisted of a central mini-computer and downhole pressure transducers. Six
27
-------
transducers were placed in the wells (SI, P, Nl, P1, El, 6) during the tests. Nine tests were conducted,
varying in duration from 30 to 3500 min. Water was extracted at a steady rate from well P while the
drawdown versus time was recorded in the extraction well and the observation wells. In one test, test 8,
well El was used as the extraction well. Pressure transducer measurements, accurate to ± 0.3 cm, were
recorded as frequently as every second.
A preliminary analysis of the pump tests was made using the graphical semi-log technique and the
log-log-type curve-matching method (Semprini et al., 1988). The graphical analysis of the pump tests
pointed to the following conclusions: (1) the pump test response matched that of a confined leaky
aquifer system; (2) the aquifer had an average transmissivity of 140 m2/day, a storativity of 0.00013, and
an r/B value of 0.05; and (3) the presence of directional anisotropy was not clearly indicated. The study
also recommended the use of unbiased statistical methods in analyzing the pump tests.
A multiple non-linear regression method based on a technique presented by Golub and Pereya
(1973), entitled VARPRO (Stanford Computer Science Department, 1973), was used subsequently for
automated analysis of the pump test data. VARPRO computes a weighted least-squares fit of the
observed data to the chosen aquifer model using a modified version of the Levenberg-Marquardt
algorithm. The Laplace space solutions of the well functions were used to represent the aquifer model in
the regression analyses; the Stehfest algorithm was used to invert the Laplace space solutions. Details of
the methods used are presented by Johns et al. (1989).
The Theis model was the first model evaluated using the non-linear regression method. There was
a poor match between observations and the Theis model due to the nearly steady-state drawdown
condition reached at late times.
The second model evaluated was that for leakage across the confining layer (Hantush and Jacob,
1955). Figure 4.6 illustrates the excellent agreement between the field data and the leaky aquifer model
for observation wells Nl and SI. The goodness of fit to the leaky aquifer model is indicated by the low
weighted residual variance, only 3.5x10"4m2. Subsequently, all the observation well data from tests 5, 6,
and 7 were regressed to this model. The parameter estimates obtained are summarized in Table 4.2.
Values of transmissivity, storativity, and r/B are in the range obtained by the visual log-log matching of the
type curves. The regressions, however, showed larger spatial differences with greater variations in trans-
missivity estimated using the regression fits, compared to visual fitting.
Recharge from a nearby boundary is another phenomenon which could provide pressure support
to the "A" aquifer. A solution for a well in a semi-infinite aquifer with a recharge boundary was generated
from the superposition of a pumping well and an image injection well. The resulting regression of the field
observations yielded an excellent match to the recharge boundary model. Table 4.3 presents a
comparison of the parameter and variance estimates for both the leaky-aquifer and constant-pressure
models and their fits to the field observations. The variance and transmissivity values for the recharge
boundary model and leaky aquifer model are essentially identical. Therefore, it is not possible to
distinguish on a statistical basis which of the two alternative explanations is a more likely cause for the
deviation from the Theis model.
Average transmissivity values in the western portion of the well field range from 130 m2/d at well Nl
to 151 m2/d at well SI. Lower average values were noted to the east, ranging from 74 m2/d to 112 m2/d
in wells 6, El, and P1. Aside from this west-east trend there is also a trend from south to north, illustrated
by both the decrease from well SI to well Nl in the western segment and the decrease from well 6 to well
P1 in the eastern segment. Superimposing these two variations suggests an overall southwest to
northeast trend in transmissivity, with values decreasing from 151 m2/d in the southwest to 74 m2/d in
the northeast. The decline in transmissivity values towards the northeast suggests a decrease in
conductivity and/or aquifer thickness in this direction. The average r/B values are greater in the east
compared to the west, with respective values of 0.33 and 0.10. This observation is consistent with the
decrease in transmissivity to the east.
28
-------
0.1
0.01
0.1 1 10 100 1000
Log T(min)
Figure 4.6. Automated regression match of the leaky aquifer model to pump test observation well
response.
TABLE 4.2. PARAMETER ESTIMATES FROM REGRESSION ANALYSIS USING
THE LEAKY-AQUIFER MODEL
Tests
Tests
Test?
Well
Nl
SI
6
El
P1
Nl
SI
6
El
P1
Nl
SI
6
El
P1
Transmissivity
(m2/d)
132
165
159
*
it
131
148
104
*
*
127
141
72
90
74
Storativity
0.0011
0.0019
0.0034
*
*
0.0011
0.0019
0.0025
«•
*
0.0015
0.0026
0.0043
0.0022
0.0024
r/B
0.10
0.10
0.20
*
*
0.06
0.07
0.25
*
*
0.11
0.13
0.31
0.33
0.42
* Noisy data, regression routine did not converge.
29
-------
TABLE 4.3. COMPARISON OF AQUIFER PARAMETERS DERIVED FROM
LEAKY-AQUIFER AND CONSTANT-PRESSURE MODELS
Well S1, Test 7
Leaky-Aquifer Model
Constant-Pressure Model
Transmissivity
(m2/d)
140.6
143.1
Storativity
0.003
0.002
Variance
(m2)
0.00035
0.00036
The higher transmissivity in the west is consistent with site geology as indicated in the fence
diagram, Figure 4.4. More gravel was found in the western section (SI, P, Nl), while more sand and clay
lenses were found in the eastern section. Although quantitative area! variations cannot be estimated with
the models used in this study, the agreement between the well-site geology and the areal trends in the
regression-matched pump test results suggests spatial variations in the parameters.
The average Storativity value determined from the leaky aquifer model for the test zone is 0.0023,
with a standard deviation of 40%. However, despite the large variability in the Storativity parameter, none of
the estimates were even within an order of magnitude of the storage values expected for phreatic
conditions (10~1). Thus, it is clear that confined conditions were observed for the duration of the tests.
Choosing between leakage or a recharge boundary as a model for the aquifer at this test site is
difficult; both models fit the pump test data extremely well. Thus, the solution to the model identification
problem considered here appears to be non-unique. Subjective criteria, such as the geologic description
at the well site, must be used to evaluate which aquifer model is more likely correct. For example, the
recharge boundary model estimates the boundary to be 31 m from the pumping well; however, the
nearest natural recharge source, Stevens Creek, is located 350 m from the pumping well. Thus, geologic
support for a recharge boundary within 35 m of the test site is questionable. Reviewing other studies in
the Santa Clara Valley (Harding Lawson Associates (HLA), 1986), there is evidence to support leakage
across the aquitard below the "A" aquifer. This gives strong indications that a leaky aquifer model is better
suited for the Moffett Field test area.
The leakage parameter estimates from the leaky aquifer model ranged from 0.06 to 0.42, with an
average value of 0.19. Using these results, the fraction of fluid leaking across the aquitard was
determined at the 6-m radius for the best, worst, and average cases. For the best case, r/B = 0.06
(B = 100 m), essentially none of the injected or extracted fluid crosses the aquitard. For the worst case,
r/B = 0.33 (B = 18m), only 10% of the fluid crosses the aquitard. On the average, r/B = 0.12
(B = 50 m), only 1% of the fluid crosses the aquitard. Since the well layout for the experiments puts all
process and observation wells within this 6-m radius, leakage through the aquitard within the process area
should be negligible. Thus, despite the low-to-moderate leakage coefficient determined in the regres-
sion analysis, the small scale of the pilot area and the high aquifer transmissivity result in minimal leakage at
the test site.
The high transmissivity results in an estimated hydraulic conductivity of 100 m/d (based on an
aquifer thickness of 1.4 m). The hydraulic conductivity is in the range of values given by Bouwer (1978)
for coarse sand (20-100 m/d), gravel (100-1000 m/d), and sand-gravel mixes (20-100 m/d), which is
consistent with the aquifer cores as indicated by the particle size distributions shown in Figure 4.5.
The long-term pump tests show that steady-state drawdowns were achieved, and that the aquifer
was capable of supplying water at the rates required for the experiments, with less than a one-meter draw-
down at the extraction well. The long-term pump tests did not detect any abrupt barriers to flow.
30
-------
The pump tests indicated that the site had several favorable hydraulic features: 1) high transmis-
sivity would permit the required pumping and injection of fluids into the test zone; 2) loss of permeability
by clogging, which might result from biological growth or chemical precipitation, would be limited, owing to
the original high permeability; 3) vertical leakage is insignificant, because the test zone is fairly well
bounded above and below; and 4) the aquifer is capable of supplying groundwater at rates required for
the experiments with less than one meter of drawdown at the extraction well.
One potential problem with the high hydraulic conductivity is that the velocity of the groundwater
under natural gradient conditions is high. A velocity of 1 m/d was estimated based on the hydraulic con-
ductivity of 100 m/d, the measured hydraulic gradient across the field of 0.0032, and an estimated poros-
ity of 0.33. This high groundwater velocity limits the control of fluid residence times, since the induced
flow field must be operated in such a manner as to overcome the natural flow in order to assure capture of
the injected solutes.
Chemical Characteristics
Samples of the groundwater from the A-aquifer were obtained during the pump test program to
determine the the background concentrations of inorganic and organic components. The analyses pro-
vided information on the quality of the groundwater in the area of the test zone and determined whether
the aquifer was contaminated with chlorinated aliphatics of interest.
Inorganic Composition--
Table 4.4 presents the major anions and cations, along with other parameters. The charge balance,
as well as the measured and calculated TDS, confirms that all of the major ions have been identified. The
major cations, in decreasing milliequivalent concentrations, are as follows: calcium > magnesium > sodium
> potassium. The major anions are, in declining order: sulfate > bicarbonate > chloride > nitrate. The
groundwater hardness is 920 mg/l, based on the calcium and magnesium concentrations, and would be
classified as very hard water. Bicarbonate is the major form of alkalinity at the measured groundwater pH of
6.5. The dissolved oxygen content of the groundwater is below 0.2 mg/l.
The analysis of the major chemical components indicates that the test zone is suitable for the
experiments. The chemical composition, including the pH, is suitable for the microbial growth. However,
because the concentration of dissolved oxygen in the groundwater is very low, all of the oxygen required
for microbial growth must be added to the test zone. The presence of high nitrate and low ammonia
concentrations indicate that the aquifer is not strongly anaerobic. Thus, major problems associated with
the change in the oxidation state by the addition of oxygen were not anticipated, at least from the
microbiological point of view. The high calcium concentration was a potential problem, e.g., the precipita-
tion of sulfates and carbonates with changes in fluid chemistry. The chemical composition of the
groundwater indicates that the pore water concentrations are close to the solubility limits of gypsum
(CaSO4) and calcite (CaCO3). Owing to the high sulfate concentration, the groundwater is not consid-
ered of drinking water quality; the poor quality of the formation water facilitated obtaining regulatory
approval to perform the experiments.
Trace Chemical Analysis-
Analysis for trace element composition was performed by Inductively-Coupled Argon Plasma
Spectrometry at the Robert S. Kerr Environmental Research Laboratory (Bledsoe, 1985, unpublished
data). Table 4.5 shows that concentrations of all inorganic elements were below 1000 u.g/1, and in most
cases below the detection limit of the analysis. Concentrations were below levels that would be
considered toxic to microorganisms, and indicate that the addition of trace nutrients may be required to
promote effective microbial growth.
Analyses were conducted to determine the type and concentrations of trace organic compounds at
the field site. Four volatile organic compounds were detected, as shown in Table 4.5. The highest con-
centrations in the native groundwater were found for 1,1,1-trichloroethane (TCA), which is present at an
31
-------
TABLE 4.4. GROUNDWATER CHEMISTRY: MAJOR IONS AND OTHER PARAMETERS
MAJOR IONS
CATIONS
Na+
K+
Ca++
Mg++
NH4+
TOTAL
ANIONS
ci-
Br
HC03-
N03"
P043"
S042'
TOTAL
CHARGE BALANCE
Concentrations
(mg/l)
Lab1a
53.
2.6
200.
100.
<0.1
356.
Lab_La
42.
0.6
270.
6.9
0.1
750.
1070.
ERROR = 2%
(mg/l)
Lab 2a '
44.
1.5
216.
93.
nd
355.
Lab_2a
39.
<0.2
227.
14.9
<0.1
699.
980.
Milliequivalents
Calculated from Lab 1 results
(meq/l)
2.3
<0.1
10.0
8.2
<0.1
20.5
1.2
<0.1
4.4
0.2
<0.1
15.6
21.4
OTHER PARAMETERS
Total Dissolved Solids {TDS, mg/l)
Measured = 1456 ±15 (by gravimetric analysis)
Calculated = 1426 (from major ion analyses)
Estimated = 1000-1400 (from specific conductance)
pH = 6.5 (measured in the field)
DO < 0.2 mg/l
Temperature = 18°C (measured in the field)
a Major ion analyses conducted by different laboratories. Lab 1, Lab 2, and Lab 3 refer to Sequoia
Analytical Laboratory, Kerr Environmental Research Laboratory, and Stanford University Civil
Engineering Laboratory, respectively.
32
-------
TABLE 4.5. TRACE CHEMICAL COMPOSITION OF THE GROUNDWATER
FROM THE SU-39 SITE
TRACE INORGANIC CONSTITUENTS3
Element
Fe
Mn
B
Zn
Sr
Ba
Al
As
Be
Ag
Cd
Co
Cr
Cu
Hg
Li
Mo
Nl
Pb
Ti
Se
n
V
DISSOLVED
(ng/i)
nd
300
150
10
67
20
<100
<30
<3
<10
<3
<10
<10
<10
<30
<10
<10
<10
<20
<100
<30
<20
<10
TOTAL
(WO
540
310
200
30
76
20
<100
<30
<3
<10
<3
<10
<10
<10
<30
<10
<10
<10
<20
<100
<30
<20
<10
TRACE ORGANIC CONSTITUENTS1*
Concentration
Chemical (u.g/1)
1,1 -dichloroethylene (1,1 -DCE) 14
1,1-dichloroethane(1,1 DCA) 0.5
1,1,2-trichloro-1,2,2-trifluroethane (Freon 113) 9.4
1,1,1 -trichloroethane (TCA) 97.4 ± 30
Determined by Inductively-Coupled Argon Plasma Spectrometry; when results were
below detection limit (d.l.), the results are listed as less than (<) the d.l. for the method.
Determined by gas chromatography or gas chromatography/mass Spectrometry. Val-
ues listed are averages of duplicate determinations, except for TCA. TCA analyses
were conducted on seven samples taken during the period 7/9/85-11/10/85; the TCA
concentrations in the samples ranged in concentration from 56 to 131 ug/l.
33
-------
average concentration on the order of 100 u.g/1, varying over a range of 56-131 uxj/l for analyses
conducted over several months. Trace amounts of other halogenated compounds are present, as shown
in Table 4.5. The organic solutes chosen as targets in the biostimulation experiments (Section 6)--
trichloroethylene (TCE), cis- and trans-1,2-dichloroethylene (cis- and trans-DCE), and vinyl chloride (VC)-
were not detected in the native groundwater samples.
Analyses were as performed for purgeable aromatics. No such compounds (e.g., benzene, xylene,
toluene, chlorinated aromatics), were detected. Total (non-purgeable) organic carbon was determined to
be approximately 2 mg/l, within the range of 0.1-10 mg/l reported for groundwaters due the presence of
natural humic and fulvic acids (Freeze and Cherry, 1979).
These analyses showed that the native groundwater in the test zone had the following important
characteristics with respect to chlorinated compounds:
1) The groundwater was contaminated with halogenated organics at low concentrations.
This was considered an important criterion for obtaining regulatory approval to
conduct the experiments. The concentrations were low, however, and would not be
toxic to the native bacteria.
2) The concentrations of the target organic compounds (TCE, cis-, and trans-DCE, VC)
were below the detection limit. Thus, controlled experiments could be performed by
adding small but measurable quantities of the target compounds to the test zone.
The results of the initial inorganic and organic analyses indicated that the groundwater was of a
suitable chemical composition for performing the experiments. The chemical composition would not
inhibit the stimulation of the methanotrophic bacteria, and it appeared feasible to inject and transport dis-
solved oxygen in the test zone without undue consumptive losses.
Aquifer Solids Analysis
Core samples of the aquifer material were obtained in order to characterize the aquifer material's
physical, chemical, and microbiological properties. Some of the core material was to be used for microbi-
ological studies in the laboratory. Aseptic procedures as outlined by Wilson et al. (1983) were used for
obtaining the core samples and transferring the materials to storage containers.
Microbial Enumeration-
The acridine orange-epifluorescence procedure of Ghiorse and Balkwill (1983) was used to enu-
merate the active bacteria attached to solid samples from the test zone. The analysis indicated that the
microorganisms were typicallv attached to particles of organic matter. The bacterial numbers per gram of
dry solids varied from 2x16° to 39x106, within the range of values of 1x106 to 50x106 obtained in
subsurface investigations of Ghiorse and Balkwill (1983), Wilson et al. (1983), and Webster et al. (1985).
No apparent trend with depth was indicated, but the highest value was observed in the sand and gravel
zone, 5.2-5.4 m below the surface. The higher bacteria counts may be associated with the high per-
meability of this zone and a correspondingly greater flux of nutrients.
The presence of methanotrophic bacteria was not established using this enumeration procedure,
since the method is not type-specific. The presence of methane-consuming bacteria on aquifer solids
was, however, demonstrated in column experiments which were discussed by Wilson et al. (1987). In
these studies, columns were packed with core solids obtained from well S1. After exchanging the pore
water with water containing methane and oxygen, oxygen and methane consumption was observed.
Results of mixture culture and soil column studies, presented in Sections 9-11, also demonstrated the
presence of methane-consuming bacteria. These studies and the bacteria enumeration study indicated
that the test zone had an indigenous microbial population that could be successfully biostimulated.
34
-------
Organic Carbon Content-
Trie organic carbon content of the Moffett aquifer material was determined by measurement on a
Dohrmann DC-80 organic carbon analyzer following pretreatment consisting of acidification with H3PO4
and heating under vacuum to remove carbonate, addition of K2S2O8, and autoclaving at 121°C for 4 hrs
in sealed ampules to oxidize the organic matter to CO2 (Ball et al., 1989). The ampules were then broken
into the oxygen stream of the DC-80 analyzer, and the CO2 production was quantified by a Horiba
nondispersive IR spectrometer. Coarse-grained samples were preground for 10 seconds in a tungsten
carbide mill to facilitate complete removal of inorganic carbon and complete recovery of the organic
carbon.
Results are summarized in Table 4.6. For the bulk material, the average value was 0.11% carbon,
with no significant influence of pregrinding. The organic matter appeared to be concentrated in the clay
fraction, which had an organic carbon content six times that of the bulk material, whereas the coarse-
grained fractions have organic carbon contents as much as 40% less than the bulk average.
Based on these measurements, it appeared likely that the Moffett aquifer material would exhibit
substantial sorption capacity, significantly greater than observed at the Borden site in our previous field
experiment (Roberts et al., 1986; Curtis et al., 1986), where the organic carbon content was measured as
0.02%.
TABLE 4.6. ORGANIC CARBON CONTENT OF MOFFETT AQUIFER SOLIDS
Size Fraction
Organic Carbon Content3
Percent, mean + std. dev.
ground
not ground
Bulk
Clay-top
Clay-bottom
U.S. Std. Mesh
<200
-100+200
-60+100
-40-60
-20+40
-8+20
-4+8
0.112±0.020b 0.110 ±0.014
0.649 ± 0.039
0.638 ± 0.090
0.161 + 0.014C
0.113 ±0.009
0.087 + 0.005
0.100 ± 0.008
0.062 ± 0.005
0.095 ± 0.009
0.082 ± 0.007
a
b
c
4 replicates, unless otherwise noted.
3 replicates.
6 replicates.
35
-------
SITE INSTRUMENTATION
The Well Field
Figure 4.1 presents a vertical section of the test zone and the well field used in the experiments.
The well field was originally designed to permit simultaneous experiments by creating two test zones
through the injection of fluids at both the south (SI) and north (Nl) injection wells, and extraction at the
central extraction well (P). The operation of the extraction well was intended to dominate the regional flow
field in the study area in an approximation of radial flow. The injection wells were located 6 m from the
extraction well. The monitoring wells were located 1.0, 2,2 and 4.0 m from the injection wells. This
spacing was chosen to result in roughly equivalent fluid residence times between monitoring wells if radial
flow conditions existed.
The extraction and injection wells were constructed of 2" PVC wellstock which is slotted over a
1.5-m screened section. The screened section was positioned 4.3 to 5.8 m below the surface in order to
fully penetrate the aquifer. After installation with a hollow stem auger, the borehole around the screened
section was back filled with sand (Monterey #8). The internal volume of each injection well was reduced by
60% by inserting a 1.5-m length of 1.75"-OD PVC hollow bar with a packer attached at the top. The hollow
bar had twenty 1/8" holes drilled along its length to distribute flow and was plugged at the bottom. The
injection solution was delivered via a 1/4"-OD stainless steel (SS) tubing connected to the top of the PVC
hollow bar packer.
The monitoring wells were 1.75"-OD stainless steel well casing with a 0.6-m screened drive point
(Johnson Wirewound #35 slot). The wells were installed with minimal disturbance of the aquifer by
augering to a depth of 4 m and hand-driving the wellpoint to a depth of 4.7 to 5.3 m (± 1 cm). The
0.6-m screen section was placed to intercept what was considered to be the most permeable zone
consisting of sands and gravels. A 1/4"-OD stainless steel tube containing a series of orifices along the
slotted interval was placed into the sandpoint. The sandpoint was filled with 3-mm glass beads to the top
of the screen, which reduced the internal pore volume by 60% and produced a porosity of 38%. A packer
was then slid over the 1 /4" sampling tube.
The Automated Data Acquisition and Control System as well as the injection system were housed in
a control shed adjacent to the well field. Samples from the test zone were pumped to the surface with a
Cole-Parmer, multihead peristaltic pump, located in the control building. In order to prevent losses by
volatilization and sorption, the fluid injection and sampling lines were fabricated with 1/4"-OD stainless
steel tubing. The total maximum volume of the sampling lines and the tubing to the well screen was
approximately 300 ml.
The Extraction System
The groundwater extraction system was designed to maintain a very constant rate of fluid withdrawal
and permit changes in rates, if desired. The central extraction well was equipped with a shallow well jet
pump, essentially a combination of centrifugal and eduction systems. The inlet to the suction pipe was
located at a depth of 5.1 m, the center of the screened section. The extracted water was delivered to the
instrument/control house where the flow rate was controlled using a pressure regulator and needle valve.
Flow rate was measured by an electronic paddle-wheel sensor backed-up by a standard rotameter.
Induced gradient conditions were created by extracting groundwater at a rate approximately 7 to 8 times
greater than the injection rate. This was required in order to dominate the regional groundwater flow.
Extraction rates ranged from 8 to 10 l/min in the experiments. Results of induced gradient bromide tracer
tests and modeling studies, described in Section 5, were used to determine the injection and extraction
rates.
In order to meet discharge requirements, the excess extracted water was air stripped before it was
discharged to a storm sewer. The air stripper was an 8"- ID column packed to a depth of 1.4 m with 5/8"
polypropylene Flexrings (Koch). The air stripper was composed of glass segments (Corning) which were
36
-------
easily acid cleaned to remove carbonate deposits that resulted from air stripped. Air flow was provided by
a 600 SCFM high-pressure blower. The air stripper removed more than 95% of the measurable
chlorinated aliphatics, and was capable of achieving the discharge requirement of 5 u.g/1 for each com-
pound.
The Injection System
The chemical injection system was designed to achieve: 1) constant injection rates, 2) constant
chemical concentrations, 3) alternate pulse injection of the groundwater containing methane or oxygen,
and 4) sampling with minimal disturbance of the system. The system was also designed to automatically
shut down when injection water was not being supplied due to a system malfunction.
Figure 4.7 is the schematic of the injection system. The extraction water before air stripping was
used as the injection supply water. This greatly reduced the buildup of carbonates in the system. A gear
pump pumped the supply water through a nominal 5u. polyester filter and a UV disinfection unit (rated at
4.5 log reduction of E. col's at 4 l/min). Rotameters metered the water at a rate of 2 l/min into each of two
counter current gas sorption columns, one for oxygen and the other for methane.
The countercurrent sorption columns, fabricated from 4"-OD plexiglass, were 40" long and were
packed to a depth of 34" with 1/2" ceramic bert saddles(Koch). Methane and oxygen, as pure gases, were
introduced into each of the columns at 330 ml/min. At this flow rate, the columns achieved effluent con-
centrations ranging from 16 to 20 mg/l for methane and 33 to 38 mg/l for oxygen, about 75% to 80 % of
the saturation values. The columns had overflow drains equipped with syphon-type gas traps to keep the
gas from flowing out the bottom of the columns, thus decreasing the explosive hazards of working with
pure oxygen and methane. Excess gas was vented outside of the instrument/control house.
The injection system also included solenoids and a pulse timer that permitted the alternate injec-
tion of groundwater containing either methane or oxygen. This system was adapted in order to avoid bio-
logical clogging near the injection well by keeping the electron donor and acceptor separated in that
region. This was accomplished by installing a timing clock (not shown in Figure 4.7), that controlled the
two pulse solenoids connected to the effluent of the two gas sorption columns. The interval over which
each pulse cycle length could be varied ranged from a fraction of an hour up to a full day. By adjusting the
pulse cycle duration, the ratio of the amount of methane to oxygen injected could also be regulated.
The groundwater containing either methane or oxygen was pumped by a gear pump to a rotame-
ter and mixer, where the spike solutions containing the chlorinated organics were added. The in-line
mixer was a low volume (approximately 20 ml), magnetically stirred mixer. Each spike solution was added
as an independent stream to avoid any combined solubility problems. Inorganic spike solutions (bromide
tracer, hydrogen peroxide) were made up as batches and pumped into the injection stream via a peristaltic
pump.
The system for delivering spike solutions of the chlorinated organics was designed to: 1) maintain
constant injection concentrations; 2) change the injected concentration, if desired; 3) add an aqueous
spike solution free of any co-solvent such as methanol; and 4) permit the addition of several chlorinated
organics, simultaneously. Figure 4.8 shows the solution preparation system, which consisted of a
refutable water reservoir, a solute saturation flask, and a multichannel peristaltic pump to inject the spike
solutions at the desired flow rates. A separate saturation flask was used for each chlorinated organic
compound. Water was drawn from the refillable reservoir through the solute saturation flask via the
peristaltic pump. The solute saturation flask, containing a sufficient quantity of pure chlorinated organic
solute to have an immiscible phase present, was mixed by a magnetic stirrer to form a saturated aqueous
solution of the chlorinated organic. The flask was immersed in a water bath maintained at 21 °C.
The flow rate of the solute-saturated spike was controlled by selecting the diameter of the
peristaltic pump tube and the pump's motor speed. Stainless steel tubing, 1/16", was connected to the
peristaltic pump tube and to transport the solute spike to the in line mixer shown in Figure 4.7.
37
-------
Gas
Sarpt 1 an
Columns
Supp1y
Filter
and
UV
Disinfection
To
Constant
Head
To
Sample <-|
Aon 1 To Id
Pulse
Soleno1ds
To
Inject
Veil
Al
xer
Spike Sols
Pressure
Swltch
Figure 4.7. Schematic of the injection system.
So 1ute
SaturaL i on
Flask
Aain
F 1 ow
Stream
Injector
Water
Reserve i r
Aulii
Channe1
Per i sia111c
Pump
Figure 4.8. Chlorinated organics delivery system.
38
-------
Upon leaving the mixer the bulk of the injection solution was pumped by a gear pump through the
injection rotameter and to the injection well. A pressure switch, connected to the rotameter influent,
automatically shut down the three gear pumps and the multichannel peristaltic pump if the rate of injection
water supply decreased significantly due to an upstream malfunction. The switch also sent a signal to the
automated data acquisition system to alert the field operator of the shut-down.
The side stream of the injected s
tube. This side stream flow occurs only
stream goes to a constant head
keeps the system balanced, maintaining
being sampled.
solution was connected to the sampling manifold via a 1/4" SS
when the injected fluid is sampled. The flow from the third side
reservoir that overflows into the air stripper. The constant head reservoir
constant injection rates during periods when the injected fluid is
THE AUTOMATED DATA ACQUISITION AND CONTROL SYSTEM
Data Acquisition and Control
experiments. The system permits the
the concentrations of the bromide trace
solved oxygen, and pH.
system (DAC) was devised at the test site to implement the field
continuous measurement of the experiment's principal parameters:
methane, halogenated aliphatic compounds of interest, dis-
Figure 4.9 is a schematic
operated by the DAC system, an ion
chromatograph equipped with a electron
chlorinated organics analysis, a gas
the methane analysis, a dissolved oxygen
fabricated as part of the automated system
representation of the DAC system. The following instruments were
chromatograph (1C) for the bromide tracer analysis, a gas
capture (GC-ECD) and a Hall detector (GC-Hall) for the
chromatograph equipped with a flame ionization detector (GC-FID) for
meter, and a pH meter. The ion and gas chromatographs were
at the field site.
I nJec11 on
Pressure
Sw i tch
D1ssa1ved
Oxygen
Aeier
Samp 1 1n
Aon 1 fa 1
Ault i -Channel
Pen i sta11 i c
Pump
Figure 4.9. Schematic of the automated Data Acquisition and Control system.
39
-------
The system was driven by a 6 Mhz microcomputer. The computer is equipped with a Techmar Lab
Master A/D board for transforming analog to digital response. The system also includes a Techmar Mega-
Function board, CGA composite monitor, 20 Mb hard disk, and modem. The program controlling the
analytical system was written and compiled with Microsoft's Quick Basic.
The DAC system was run in either manual or automated mode. In manual mode, the operator could
1) select individual sample locations, 2) create a sample table for automated sampling mode, 3) calibrate
instruments, 4) produce real-time graphics of current experiments, or 5) analyze stored data and
recalculate erroneous data (such as misidentified chromatographic peaks).
The DAC system functioned in automated mode as follows. The DAC system selected from a pre-
programmed table the location of the next sample. The multihead Cole Farmer peristaltic pump, used for
sampling, was activated through the interface. The sampling manifold was flushed with rinse water
(activated carbon treated, nominal 1 p.m filtered, deionized water) for 30 sec at a flow rate of 200 ml/min.
The electrically driven solenoid, for the selected sampling location, was opened and pumping was
initiated. Pumping was continued for a preset time (approximately 4 min), to obtain a representative
sample.
The sample was split into two streams, one flowed through a cell containing the DO probe and the
other through a cell containing a pH probe. After pumping at the sampling location for the desired time,
the DAC stopped the sampling pump and collected readings from the DO meter and the pH meter.
Analog signals were converted to digital output through the Lab Master A/D board.
The DAC then started processing the IC.GC-ECD, GC-Hall and FID analyses. The sample manifold
solenoid was held open and the sample was pulled through the chromatograph sample loops by a multi-
channel Technicon pump. The sample lines and loops were fabricated of either stainless steel or glass to
minimize sample losses due to sorption.
All the analyses were performed simultaneously, with the DAC controlling the analysis and the col-
lection of the data through the Lab Master A/D board. A Spectra-Physics 4270 integrator, which was
added to the system in the spring of 1987, processed the output from the GC-ECD, and the GC-Hall anal-
ysis. An integrator programmed into the DAC system as a subprogram was used to process the output
fromthelCandGC-FID.
The data were stored in the systems data base both as integrated peak areas and as computed
concentrations. The storage as integrated peaks permitted recalculation of concentrations if a calibration
was in question or a peak was misidentified. The stored concentrations could then be plotted by the DAC
system providing real time monitoring capabilities.
Upon completion of the analysis, the DAC the system automatically proceeded to the next sample.
System interprets could be made at this time in order to enter manual operation.
The system calibrations were performed in manual operation using external standards. The external
standards for the GC calibrations required the preparation of a solution containing known concentrations
of the compounds of interest in a 100 ml (Spectrum) gas-tight syringe. The syringe solution was first satu-
rated with methane. The chlorinated organics were when added to the syringe solution as standards dis-
solved in methanol. The standard syringe solution was processed in the same manner as a field sample.
This was accomplished by attaching the syringe to the GC sample valves, where the sample manifold
normally was connected. The standard solution was then pulled through the sampling loop by the
Technicon Pump, in the same manner as the field sample. Similarly, the 1C is calibrated with a standard
solution fed at the point where it normally connects to the sample manifold. During normal operation the
system was calibrated several times a week. The system was also calibrated after maintenance of the
analytical equipment.
40
-------
ANALYTICAL SYSTEM PERFORMANCE
The DAC system proved capable of sustained and reliable operation, and permitted continuous
operation and oversight capabilities that would have been unachievable in a manual mode. During 1987,
for example, the DAC system collected data over 85% of the year and processed approximately 9,300
samples, yielding approximately 93,000 individual data points. The detection limits for the various analy-
ses are summarized in Table 4.7. The detection limits are generally two orders of magnitude lower than
the injected concentrations of the field experiment.
In order to evaluate in the analytical system and the injection system, the average injection con-
centrations, standard deviations, and coefficients of variation of injection concentrations for one experi-
ment are summarized in Table 4.8. The compounds tabulated in Table 4.8 are broken into four groups
based upon type of analysis.
TABLE 4.7. SUMMARY OF ANALYTICAL METHODS AND DETECTION LIMIT
Component
Organic Solutes
(Freon113,TCA,TCE)
(trans-, cis-DCE)
(trans-, cis-DCE, vinyl chloride)
Methane
Method of Measurement
Gas Chromatography-ECD
Gas Chromatography-ECD
Gas Chromatography-Hall
Gas Chromatography-FID
Detection Limit
Dissolved Oxygen
PH
Anions (Br, NOa)
Probe
Probe
Ion Chromatography
0.1 mg/l
NA
0.5 mg/l
0.5 u.g/1
5ug/l
0.5 ug/l
0.2 mg/l
TABLE 4.8. SUMMARY OF ANALYTICAL AND INJECTION PERFORMANCE
Compound
Mean
Standard
Deviation
Number of
Observations
Coefficients of
Variation (%)
Freon 113
trans-DCE
cis-DCE
TCA
TCE
DO
Methane
Bromide
Nitrate
PH
4.3
121
140
46.0
47.3
35.0
16.9
45.9
56.8
6.5
0.8
19.8
23.3
10.8
6.9
2.4
2.3
3.0
3.3
0.2
169
167
164
169
169
61
27
81
82
89
18.4
16.3
16.6
23.6
14.5
6.9
17.6
6.5
5.8
NA
41
-------
The first group are compounds measured by the GC-ECD, including the three spiked chlorinated
solutes, trans-1,2-dichloroethylene (t-DCE), cis-1,2-DCE (c-DCE), and TCE. The concentrations range
from 4 u.g/1 for Freon as a background contaminant to 140 ug/l for cis-DCE. The coefficients of variation fall
within expected bounds for all cases, and do not exceed 25%. In the ECD group, the compounds native
to the groundwater, Freon 113 and TCA, show the greatest coefficient of variation. This is probably due
to unequal gas flows through the two gas sorption columns, which strip a fraction of these recycled
compounds.
The second group consists of methane and DO, which were alternately pulse-injected at a ratio of
1:2 (methane:oxygen). In actual operation methane gas flow was stopped while an oxygen pulse was
being injected. This may contribute to some variability in the measured methane concentrations.
Nonetheless, the coefficient of variation achieved for the methane analysis, 18%, conforms to the
precision expected for this type of GC analysis. The dissolved oxygen measurements are very consistent,
as indicated by the low coefficient of variation of 6.9%.
The third group of compounds consists of the anions, which are measured by 1C. Bromide was
added as a conservative tracer, whereas nitrate was native to the groundwater. The 1C data were very
reproducible, with standard deviations of approximately 6 to 7%. The final parameter measured was pH,
which was also very reproducible, with a precision of ± 0.2 pH unit.
The compounds that were measured by GC analysis were found to have the greatest variability.
Daily changes in temperature at the site, which caused diurnal variations in measured concentrations, is
one reason for this. In Figure 4.10, minimum values in TCA concentrations are shown to occur at about
4 AM while maximum values occur at 3 PM. This is in response to nightly cooling and warming in the
afternoon. Although considerable effort was made to reduced these variations, they could not be
completely eliminated.
Diurnal Temperature Fluctuations
SInject4—h Slo-a S2x-x S3O-O Extract
60
0.0
24
48 72
Time (Hours)
96.0
120
Figure 4.10. Daily variations in 1,1,1 -TCA concentrations due to diurnal temperature fluctuations.
42
-------
Normalizing Data
In the analysis of the experimental results, the concentrations at the observation locations were
normalized to the injected concentrations. This was done to: 1) provide an easy means of showing the
fractional breakthrough at observation locations of the injected chemicals; and 2) show the degree of bio-
transformation that was achieved during transport through the biostimulated zone. In addition, normaliza-
tion provided a means of reducing some of the variability that resulted from temperature variations in the
analytical system. This was accomplished by normalizing to the last injected concentration obtained prior
to the particular measurement being normalized.
Table 4.9 presents a raw data set and a normalized data set from one experiment. Concentrations at
the observation location had achieved near-steady-state fractional breakthroughs during the period ana-
lyzed. The coefficient of variation of the normalized data is lower than that of the raw data. A 40%
reduction in the coefficient of variation was achieved by the normalization. This results mainly from
reducing the daily variations caused by temperature fluctuations.
TABLE 4.9. COMPARISON OF VARIABILITY OF ABSOLUTE AND NORMALIZED
CONCENTRATION DATA FOR OBSERVATION WELL S2
Chemical
VC
trans-DCE
cis-DCE
TCE
Chemical
VC
trans-DCE
cis-DCE
TCE
Mean
Concentration
(Mfl/l)
41.9
50.9
96.7
33.8
Normalized
Concentration
0.91
0.95
0.99
0.96
Standard
Deviation3
(WO
5.2
8.6
14.2
4.7
Standard
Deviation
0.082
0.098
0.092
0.082
Coefficient of
Variation13 (%)
12.3
16.9
14.7
13.9
Coefficients of
Variation15 (%)
9.0
10.3
9.4
8.5
Number of
Observations
18
39
39
39
Number of
Observations
18
39
39
39
a Standard deviation of individual measurements.
b Coefficient of variation of individual measurements.
The coefficients of variation of the normalized concentrations at the observation locations are less
than 10%. When the data are used to estimate the degree of transformation achieved, based on equation
4.1, the mean value and the standard error of the mean are used. The standard error of the mean reflects
the large number of samples used in the estimate, where the standard deviation is divided by the square
root of the number of observations. Thus, for example, the standard error of the mean of the cis-DCE
concentration given in Table 4.9 is 14.2/390-5 = 2.3 ug/l. This is a very small standard error, compared to
the average value of 96.7 ug/l. This illustrates the advantage of the DAC system: the large number of
precise observations minimize the relative errors associated with the mean estimates.
43
-------
SUMMARY
A field site was selected for the demonstration that offered a suitable combination of attributes from
the standpoint of hydrogeology, geochemistry, logistics, and institutional acceptability. A shallow, con-
fined aquifer was tested and shown to have adequate permeability and acceptable chemical composition.
The well test data also provided a basis for preliminary design of the injection/extraction system. The
general approach to the experimental program is also summarized.
The quality assurance checks demonstrated that the automated data acquisition was capable of
providing a large number of measurements with high precision. This ability permitted the field experi-
ments to be performed as a series of stimulus-response tests, with the response of the system to the
stimulus precisely monitored. The injection system used in the field experiments delivered controlled
amounts of the chemicals, which was required in these experiments. Coefficients of variation in the mean
observed concentrations were less than 15%. Normalization reduced the coefficient of variation to less
than 10%.
44
-------
N2x-x N3O-O
D)
c
o
•1-1
4J
(D
C_
-P
C
0)
u
c
o
u
0.0 30 60 90
Time (Hours)
120
150
sgure 5.1. Results from the Tracer2 natural gradient tracer test.
80 120
Time (Hours)
160
200
Figure 5.2. Results from the Tracers natural gradient tracer test.
46
-------
SECTION 5
RESULTS OF TRACER TESTS
Lewis Semprini, Gary Hopkins, and Paul Roberts
Tracer tests were performed under both natural gradient and induced gradient conditions to
characterize fluid and contaminant transport in the test zon|. Natural gradient tracer tests were conducted
to determine the direction and magnitude of groundwater flow. Induced gradient tracer tests provided
information on the transport of chemicals that were used in the later biostimulation and biodegradation
experiments. _'*
S ', " '
"~ "\ - -
NATURAL GRADIENT TRACER TESTS \. "~ > . -. ;
Two natural gradient tracer tests were conducted, Tfacer2 and Tracers. The tests were performed
as follows: a slug of 460 liters of bromide tracer was injected over a period of 3 to 4 hrs into a well along the
main line of wells S1 through Nl, and then allowed to drift under natural gradient conditions. Responses at
monitoring wells encompassed both the breakthrough and the elution of the bromide tracer. In the
Tracer2 test (Figure 5.1), well P was used to inject the tracer and wells N3, N2, and N1 were used as
monitoring wells. In the Tracer3 test (Figure 5.2), well S1 was used to inject the tracer, and all the wells
along south to north legs were monitored.
The experiments indicated that the groundwater flow-Was primarily in a northerly direction. Figures
5.1 and 5.2 show responses at the monitoring wells for the Tfacer2 and Tracers tests, respectively. The
response curves are skewed in shape, with a sharp rise in concentration followed by a gradual decrease,
or tailing, to background concentrations. The areas under the response curves are seen to decrease as
the distance from the injection well increases, especially for the Tracers test. The maximum concentra-
tions are significantly lower than the injected concentrations. The decrease in area with distance and the
low maximum concentrations suggest either a f tow direction that deviates slightly from being parallel to the
line of the observation wells and/or a large amount of lateral dispersion.
Table 5.1 summarizes the results from the natural gradient tracer tests. The skewed shape of the
response curves are indicated by the greater time associated with the center of mass of the response
curves compared to the time to the maximum observed concentration. The groundwater velocity esti-
mates based on the time corresponding to the center of mass of the response curve are in good
agreement for the Tracer2 test. An average value of 2.5 m/d was obtained. The results obtained from the
Tracers test are more variable, with values increasing with increased distance from the injection well to the
observation well. The higher velocities are associated with a decrease in area under the response curves.
The rapid transport in the test zone is typified by the initial response at the S3 monitoring well, which
precedes that of the S2 well, even though the latter well is located closer to the injection well for this test.
This earlier breakthrough is reproduced in all the tracer experiments performed to date. These data
suggest that the aquifer is quite heterogeneous. The high permeability zones rapidly convey the tracer to
the distant wells, while the responses at observation wells closest to the injection well represent contri-
butions from a range of permeability zones. The observation wells are not fully penetrating. Thus, if there
were layering and vertical structure in the test zone, the monitoring wells may be sampling different zones,
45
-------
TABLE 5.1. ESTIMATES OF REGIONAL VELOCITIES BASED ON THE RESULTS OF THE
NATURAL GRADIENT TRACER EXPERIMENTS
Trace r2
Tracers
Well
N3
N2
N1
S1
S2
S3
Distance
from the
Inj. Well
(m)
2.0
3.8
5.0
1.0
2.2
3.8
Time
Max.
Cone.
(hrs)
8.8
27.8
32.8
16.4
32.5
12.9
Time
Center
of Mass
(hrs)
17.9
38.6
50.5
32.9
44.3
20.0
Velocity3
(m/d)
2.6
2.4
2.4
0.7
1.2
4.8
Area Under
Response
Curve
(mg-hr/l)
1555
1059
1250
3658
2131
1019
a Velocity based on center of mass.
especially along the south experimental leg, where the variations in estimated velocity are great. More-
over, the extensive tailing in the response curves would suggest multi-permeability zones, as discussed
byMolzetal. (1986a).
The results of the two natural gradient tests indicate a fairly high groundwater velocity at the site:
approximately 2.4 m/d. The velocity is higher than the 1 m/d value obtained from the measured gradient
and hydraulic conductivity estimated from pump tests, but. nonetheless of the same order of magnitude.
The hydraulic conductivity, however, is based on an aquifer thickness of 1.5 m. If the thickness were
less, higher estimates of groundwater velocity would result.
INDUCED GRADIENT TRACER TESTS
Induced gradient tracer tests were performed using bromide as a conservative tracer, DO as a
potentially reacting tracer, and the chlorinated organics studied in later biotransformation experiments.
The hydraulic conditions approximated closely those of the later biostimulation-biotransformation
experiments. The bromide tracer tests provided information on the (1) fluid residence times, (2) degree
of breakthrough at observation locations, (3) dispersion, and (4) extent to which the injected fluid was
captured by the extraction well. The chlorinated organic transport experiments provided information on
(1) the extent of retardation of chlorinated solutes due to sorption onto the aquifer solids, (2) the rate
processes affecting sorption and desorption, and (3) whether transformation occurred before the test
zone was biostimulated. While providing information on transport, the experiments also served as quasi-
controls before the test zone was biostimulated.
The southern experimental leg of wells-including injection well SI, observation wells S1, S2, S3,
and extraction well P-were used in these experiments and all subsequent biostimulation and
biotransformation experiments. By injecting and extracting groundwater in this configuration, the induced
flow was in the same direction as the natural groundwater flow, thus promoting effective capture of the
injected groundwater by the extraction well.
The induced gradient tracer tests were performed with continuous chemical addition, using the
system described in Section 4. Fluid extraction rates ranged from 8 to 10.5 l/min, while injection rates
ranged from 0.7 to 1.5 l/min. The injection and extraction rates used in the different experiments are
presented in Table 5.2.
47
-------
TABLE 5.2. COMPARISON OF BROMIDE TRACER TESTS UNDER INDUCED GRADIENT CONDITIONS
Test3 TR4
Season 1
TR5
1
TR8
2
TR11
3
TR12
4
Injection Rate (l/min) 1.1 0.66
Extraction Rate (l/min) 8.0 8.0
Percent Steady-State
Breakthrough WellSI 95 94
Well S2 95 72
Well S3 80 57
Extraction 9 5
Time to 50% Break-
through (hrs) WellSI 8 9
Well S2 20 17
Well S3 20 7
Extraction 26 20
Percentage Recovered
at the Extraction Well 66 59
1.36
10.0
100
98
84
13
7.5
16
20
30
105
1.5
10.0
102
100
96
14
9
23
27
40
94
1.5
10.0
100
99
95
15
8
21
26.5
42
ND
a TR4 = Tracer4 experiment, etc.
The induced-flow tracer tests were conducted in each of the field seasons. The first season's tests
included studies with bromide, TCE, and DO. The second season's studies included bromide, TCE,
trans-DCE, and cis-DCE; vinyl chloride was included in the third field season as well.
Bromide tracer tests provided the initial information on the transport characteristics of the test zone.
Bromide tracer results from the Tracer4 test (first season), and the Tracers test (second season) are
shown in Figures 5.3 and 5.4, respectively. Concentrations are normalized to the injected concentration
for both experiments. Bromide was injected for 107 hrs in the Tracer4 test and for 190 hrs in the Tracers
test. Both sets of data show a tightly spaced temporal response with approximately eight samples at each
observation point per day. The tracer breakthrough at observation locations, due to constant bromide
injection and its elution from the test zone after its addition was stopped, is apparent. The normalized
concentration values of the Tracers test are shown to have less variability compared to the Tracer4 test.
This results from an experimental modification for adding the bromide tracer that resulted in a more
constant rate of bromide addition in the Tracers, and subsequent, experiments.
The steady-state fractional breakthrough at observation locations and the time required to achieve
50% of this breakthrough are summarized in Table 5.2 for several key tracer tests. Comparing the Tracer4,
Tracers, and Tracers experiments, the highest degree of fractional breakthrough was observed in the
Tracers experiment, while the lowest was observed in the Tracers test. The higher degree of break-
through results from the factor of 2 increase in the injection rate in the Tracers test compared to the
Tracers test. Breakthroughs of less than 100% result from dilution by the native groundwater. This
dilution is most apparent at the extraction well as a result of the greater rate of extraction compared to
injection. The S3 observation well, farthest from the injection well, always showed some dilution of the
injected fluid by native groundwater.
48
-------
CO
I!
(O
c
CB
01
DO
S
5.'
CD
CT
CD
0)
S
(Q
0)
Q.
CD_
£
o'
CD
H
1
3
CD
X
Norm. Concentration (C/Co)
(Q
C
CD
O1
CO
CD
S
3
51
CD
1
CD
SO
5
(O
Q.
CD_
S.
o'
3
CD
CD
X
1.
Norm. Concentration (C/Co)
-------
The mean fluid residence time is approximately 8 hrs to the S1 well, and 18 hrs to the S2 well. The
mean fluid velocity based on transport times to both wells is approximately 3 m/d. The time required to
achieve 50% breakthrough does not change significantly between the Tracer4 and Tracers experiment,
despite the different injection and extraction rates. This result indicates the strong influence of the natural
groundwater flow (1 to 2 m/d) on transport in the test zone. Due to the dilution and more dispersed
response at the S3 well, the 50% breakthrough time is probably less representative of the mean
residence time, which is also more variable at that well.
Figure 5.5 shows the initial breakthrough of the bromide tracer during the initial 200 hrs of the
Tracers test. The initial breakthrough at the S3 observation well, located 3.8 m from the injection well
preceded that at well S2, 2.2 m from the injection well. This response indicates the short-circuiting of flow
or, since the observation wells are only partially penetrating, the possible existence of vertical variations in
permeability. As time proceeds, the fractional breakthrough at the S2 well surpasses that of the S3 well,
and eventually reaches 100%, while at steady state, the S3 well only achieves 84% breakthrough. Due to
the behavior of the S3 well in these tracer tests, less emphasis is placed in its response in the
interpretation of the results of later biostimulation and biodegradation experiments.
The bromide breakthrough observed during the Tracers test shows a relatively rapid increase to
80% fractional breakthrough at the S1 and S2 wells, followed by a much slower approach to 100%
breakthrough. This gradual increase at later time results in part from a gradual increase in the injected
bromide concentration, due to the recycling of bromide in the extracted water. The recycle also resulted
in some extended tailing in the elution part of the tracer test. In Section 12 of this report, an analytical
model is presented that takes into account advective and dispersive transport, as well as recycle, in a
uniform flow field. The analytical model's fit to these tracer data are presented in Section 12.
1.2
Extract A-A
A&A-&A&A&—AAAA AAAA
40
80 120
Time (Hours)
160
200
Figure 5.5. Initial breakthrough of bromide in the Tracers experiment.
50
-------
Numerous bromide tracer tests were performed during the three years of field testing as part of
organic solute transport experiments, and biostimulation-biotransformation experiments. The tests all
showed very reproducible responses, similar in form to that shown in Figure 5.4. In all tests, the initial
breakthrough at the S3 well always preceded that at the S2 well. Transport times were very reproducible
from test to test, and similar degrees of fraction breakthrough at observation locations were obtained.
As summarized in Table 5.2, in the second and third seasons of testing, steady-state fractional
breakthroughs at the S1 and S2 wells of 98% to 100% were always obtained, while some dilution was
always observed at the S3 observation well. In the third season, the time to 50% breakthrough at the S2
and S3 observation wells increased, although injection and extraction rates were nearly unchanged. The
third-season tests were performed during the second season of drought, which may have reduced the
natural gradient component of the groundwater velocity. The biostimulation of the test zone in two
successive seasons of field testing also may have resulted in some changes in flow characteristics of the
test zone.
The degree of capture of the bromide tracer in the first season of testing ranged from 59% to 66%.
In the second and third seasons, 94% to 105% of the injected bromide was captured by the extraction
well. The mass balance of greater than 100% is within the experimental error. The increase in the tracer
recovery most likely results from the higher extraction rates used in the later seasons. This ability to
recover essentially all the bromide tracer reduced the error in mass balances used to estimated the degree
of biotransformation of the chlorinated organics. It also enabled us to obtain regulatory support to evalu-
ate the transformation of vinyl chloride in the test zone.
The DO transport experiments evaluated whether DO consumption occurred before the test zone
was biostimulated, and if DO was transported at the same velocity as the bromide tracer. Since DO
concentration of the native groundwater is very low, DO in the injected groundwater and its breakthrough
at observation locations was monitored and compared to the the bromide tracer.
Figure 5.6 shows the concentration breakthrough of DO observed in the Tracer4 experiment. The
S1 observation well concentration was higher than the injection concentration. This results from diffusion
of atmospheric oxygen through the Teflon tubing used to deliver the injected water to the test zone. The
Teflon tubing was therefore replaced with stainless steel tubing for the subsequent experiments.
Dissolved oxygen response was essentially the same as that of the bromide tracer (Figure 5.3). The
results indicated little utilization or retardation of DO during transport. Thus, the ability to transport
required DO through the aquifer was demonstrated.
DO concentrations were monitored as part of the Tracers experiment performed during the second
season of field testing. This test preceded the biostimulation tests of the first season. The normalized
concentration of DO compared to bromide at the S2 observation well is shown in Figure 5.7. The
breakthrough of DO was delayed compared to bromide, and the fractional breakthrough of DO reached
approximately 90% of the injected concentration, compared to the 100% achieved by bromide. Both
observations indicated some DO consumption occurred during transport through the test zone. Based
on the DO injection concentration of 15 mg/l and the fractional breakthrough being 10% lower than
bromide, an estimated 1.5 mg/l of DO was consumed. Most of this consumption occurred within the first
meter of transport, where most of the biomass was stimulated, as will be discussed in detail in Sections 6
and 13. The DO consumption probably results from the oxidation of organic matter that remains after the
biostimulation of the test zone.
Induced flow tracer experiments with the chlorinated organics solutes were performed in all three
seasons of field testing. Table 5.3 summarizes the organic solute tracer tests performed. In the first
season both the breakthrough and elution of TCE was studied. In the second season, the transport of
TCE, cis-DCE, and trans-DCE was studied, while in the third season the transport of vinyl chloride in
addition to the prior year's solutes, was also studied. In the second and third seasons the biostimulation-
biotransformation experiments immediately followed the tracer experiments. Thus, only data on the
breakthroughs before biostimulation were obtained. The elution of these compounds from the test zone
was monitored at the end of the biostimulation experiments.
51
-------
Ol
ro
3
(Q
3
Ol
O
O
QJ
Norm. Concentration (C/Co)
Concentration (mg/L)
01
3
CL
o
CT
a
SB
3
i
CO
SI
^^
3"
CD
CO
ro
I
CD
i
T3
CD
CD
CO
I
en
05
o
O
CT
3
0>
(Q
CD
CD
X
T3
CD
CD
-------
TABLE 5.3. SUMMARY OF INDUCED ORGANIC TRACER TESTS PERFORMED
Experiment Season Chemicals
Studied
Tracers 1 TCE
Tracers 2 TCE
tran-DCE
cis-DCE
Tracer! 1 3 TCE
trans-DCE
cis-DCE
Tracer12 3 VC
Average
Cone.
(MS/0
165±30
48+10
1 1 2±39
110±36
47±6
50±7
85±13
44±7
Processes
Studied
Retardation
Elution
Mass Balance
Retardation
Steady-State
Breakthrough
Retardation
Steady-State
Breakthrough
Retardation
Steady-State
Breakthrough
In the first season of field testing, the transport of TCE was studied in the Tracers experiment before
the test zone was biostimulated. TCE was injected continuously for 240 hrs at an average concentration
of 160 u.g/1, after which its injection was terminated and its elution from the test zone was monitored.
Figure 5.8 shows the normalized breakthrough of both bromide and TCE at the S1 observation well.
TCE is shown to be retarded with respect to bromide. The time to achieve 50% breakthrough for TCE
(assuming TCE would achieve the same as that obtained by bromide) is 42 hrs. Comparing to the value of
8 hrs for bromide, indicates that TCE is retarded by a factor of approximately 5. Thus, due to the sorption
onto the aquifer solids, the TCE takes a much longer time to achieve steady-state concentrations at the
observation locations.
In order to complete system mass balances, the elution of TCE was monitored in the Tracers
experiment. Based on the breakthrough response, extended tailing was expected during elution, thus
prolonging the experiment. The injection of TCE was therefore stopped before steady-state
breakthrough concentrations were achieved. Figure 5.9 shows both the breakthrough and elution of
TCE from the test zone. The much slower increase at the S2 well due to the longer flow path is apparent.
Very extended tailing in both the TCE breakthrough and elution from the test zone was observed at the
observation wells and the extraction well.
The response of TCE to breakthrough and elution can be compared with the bromide tracer, using
the Tracers test for comparison purposes (Figure 5.4; note the factor of two difference in the time scales).
The TCE response does not conform to that expected for the processes of advection, dispersion, and
retardation, assuming equilibrium sorption between the aquifer solids and the aqueous phase. Besides
retardation, the TCE data show greater spreading (dispersion) compared to the bromide. Part of this
spreading may be associated with recycle of the injected TCE. However, other factors probably contribute
to the spreading, including (1) sorption, if the process is rate-limited, as will be discussed in Section 8;
and (2) layers of differing hydraulic conductivity in the aquifer.
53
-------
Tl
<5'
I
en
bo
01
CO
i
01
CO
CD
S
£0
£
3-
S
CQ
3-
03
Q.
CD
O
m
CD
if
CD
T3
CD
a.
CD
Norm. Concentration (C/Co)
Norm. Concentration (C/Co)
I
CT
CD
CD
m
SI
i-*
CD
O)
...A
CD
CD
3
S
5i
CD
.
CD
S
-------
Extended tailing resulted in concentrations in the 1 u.g/1 range remaining at observation wells 600 hrs
after the mass balance was completed. Since pumping was virtually continuous during that period, the
TCE captured by the extraction well is estimated to be close to that of bromide. Thus, the results of this
tracer test suggest little transformation of ICE before biostimulation.
In the Tracers and Tracer! 1 experiments, performed in the second and third seasons of field
testing, trans-DCE and cis-DCE were added along with TCE. The experiments were also performed at
higher injection and extraction rates, and at a lower TCE injection concentration (50 ug/l), compared to
the Tracers experiment. One of the goals of these experiments was to achieve steady-state chlorinated
organics concentrations at observation locations before the test zone was biostimulated. To accomplish
this, the chlorinated organics were continuously injected for approximately 1000 hrs.
Figure 5.10 shows bromide, trans-DCE, cis-DCE, and TCE responses at the S1 well in the Tracerl 1
experiment. At early times, trans-DCE appears to be more strongly retarded than TCE, while at later times
TCE appears to be more strongly retarded. Cis-DCE is clearly the least strongly retarded of the
compounds tested. Based on the results of batch sorption studies (Section 8), TCE was anticipated to
be the most retarded compound. One possible explanation for the delayed breakthrough of trans-DCE at
early time is that trans-DCE was partially transformed during transport through the previously biostimulated
test zone. However, we do not have independent evidence to confirm this transformation. The later time
response, indicating greater retardation of TCE, is consistent with the results of batch laboratory studies,
presented in Section 8.
The shape of the cis-DCE, trans-DCE, and TCE breakthrough responses in the Tracers and the
Tracer! 1 experiments was observed to be similar in form. All responses showed a greater amount of
spreading than observed with the bromide tracer.
Figure 5.11 shows the normalized TCE concentrations at observation locations in the Tracers
experiment. The very gradual approach of TCE to the injection concentrations is apparent, especially at
the S2 well and the extraction well, due to the longer distances traveled. With prolonged injection in both
the Tracers and Tracerl 1 experiments, cis-DCE, trans-DCE, and TCE reached 90% to 95% of the
injection concentrations at the S1 observation well. These results indicated little transformation of the
these compounds before the initiation of the biostimulation-biotransformation experiments.
Tracerl2, the final organic tracer experiment, studied the transport of vinyl chloride. Results of the
previous chlorinated organic transport experiments demonstrated that prolonged periods of addition were
required in order to achieved steady-state breakthrough concentrations. To shorten this time, a two-step
vinyl chloride addition method was used. In the first step, vinyl chloride was injected at 1.5 times the final
desired concentration. After 50% fractional breakthrough was achieved at the S2 well, the injection
concentration was reduced to the final operating concentration. Results of model simulations, presented
in Section 14, indicated that a much faster approach to steady-state conditions would be achieved by the
two-step approach, especially if the sorption-desorption process were rate-limited.
Results of the vinyl chloride addition experiment are presented in Figure 5.12. After injecting at a
concentration of 69 ug/l for 48 hrs, the injection concentration was reduced to 44 u.g/1. The
concentrations at both the S1 and S2 wells approached steady-state levels rapidly. The results indicated
that the two-step method of organic addition resulted in a rapid attainment of steady-state conditions. The
results also showed vinyl chloride was not transformed before the test zone was biostimulated.
In this experiment, and in the later stages of the Tracerl 1 experiment, DO was not being injected
into the test zone. Therefore background nitrate served as the electron acceptor. The results indicated
little transformation under these anoxic conditions. However, unlike the Tracers experiment, where the
cis-DCE, trans-DCE, and cis-DCE, were added along with DO, vinyl chloride was not added in the
presence of DO. Thus, it is not known whether transformation of vinyl chloride would have been
observed in the presence of DO.
55
-------
(Q
§
Ol
(O
CD
01
Norm. Concentration (C/Co)
in
N
CD
Q.
CD
I
3
(Q
m
a.
I
a.
CD
X
.
CD
CD O
CD CD
^°-
3 cr
5 3
_?s ^-
It
3 a
Q.'
CD
o
m
Q.
c/>
i
O
O
m
m
Norm. Concentration (C/Co)
a
21
CD
CO
-------
S2x-x
0.0
40
80 120
Time (Hours)
160
200
Figure 5.12. Response at the S1 and S2 wells 1
o the two-step vinyl chloride addition.
Retardation Estimates
The retardation of the chlorinated organics with respect to bromide tracer was estimated using two
methods: 1) the time required to achieve 50% of the steady-state fractional breakthrough, and 2) the
center of mass of the breakthrough response, as described by Roberts et al. (1980). The second method
is considered to be a better representation of the field response, since it takes into account the long
tailing in the breakthrough response. However, maintaining constant conditions over the long experiment
required for this estimate proves to be difficult in practice. Thus, both methods of estimating retardation
factors will be evaluated and compared.
The 50% breakthrough method assumes that the chlorinated organics eventually achieved the
same degree of fractional breakthrough as the bromide tracer. The retardation for TCE is given by the ratio
of (T50o/0 JCE :T50% Br)- Tnis method is expected to give a conservative estimate of the retardation factor
due to the asymmetric shape of the observed breakthrough response.
Table 5.4 presents the estimates of 50% residence times, and estimated retardation for the organic
addition experiments. Bromide residence times used in the retardation estimates were presented in
Table 5.2. The residence time tor 1,1,1-TCA, a background contaminant in the test zone, was based on
its elution from the test zone during the Tracer4 experiment.
The estimated retardation factors for TCE were found to be fairly reproducible from test to test. No
observable increase in retardation of TCE resulted from the biostimulation of the test zone in the previous
year. Average retardation values of 6 and 8.3 were obtained for the S1 and S2 wells estimates,
respectively. Vinyl chloride and 1,1,1-TCA were the least retarded, with an average retardation value of
1.8. The retardation of cis-DCE and trans-DCE differ by about a factor of 2, having respective values of 4
and 10. Results for cis-DCE were reproducible in the two tests performed as well. Trans-DCE estimates,
however, differed by almost a factor of two in the different tests. The higher retardation estimate for trans-
DCE in the Tracerl 1 test may have resulted from partial transformation at early time.
57
-------
TABLE 5.4. RESIDENCE TIMES AND RETARDATION FACTORS FOR THE
CHLORINATED ORGANIC COMPOUNDS BASED ON THE TIME
REQUIRED TO ACHIEVE 50% FRACTIONAL BREAKTHROUGH
Experiment
Tracer4
Tracers
Tracers
Tracerl 1
Compound
1,1,1-TCA
TCE
TCE
trans-DCE
cis-DCE
TCE
trans-DCE
cis-DCE
Well S1
*50%
(hrs)
10
40
60
50
30
50
120
45
Well S2
t50%
(hrs)
30
160
150
150
70
175
280
90
R
(S1)
1.3
5
7
6
3
6
13
5
R
(S2)
2.0
9
8
8
4
8
12
4
Tracer12 Vinyl chloride 13 42 1.6 2.0
Table 5.5 presents estimates of mean residence times, t, and retardation factors for the Tracers
experiment using the center-of-mass method. The mean residence times are longer than those based on
the 50% breakthroughs (Table 5.4). Bromide residence times are increased by almost a factor of 2,
compared to those based on the 50% breakthrough. Part of this increase is related to recycle of the
injected bromide, which caused some of the extended tailing. The residence times of the chlorinated
organics were larger by factors of 2 to 5 compared to the 50% breakthrough results, due to the very
asymmetric response observed. The retardation factors estimated using this method are as much as a
factor of 2 greater than those achieved based on the 50% breakthrough. For this case, TCE was the most
strongly sorbed compound, followed by trans-DCE and cis-DCE.
TABLE 5.5. RESIDENCE TIMES AND RETARDATION FACTORS FOR THE
CHLORINATED COMPOUNDS BASED ON CENTER-OF-MASS
ESTIMATES
t t t
Substance WellSl WellS2 Well S3 MeanR
(hrs) (hrs) (hrs) Factor
Bromide 14 27 23 1
TCE 160 300 300 12
cis-DCE 145 200 170 8
trans-DCE 155 250 225 10
58
-------
The higher retardation factors based on the center-of-mass calculation are expected, since this
method would better capture the long-term effects of rate-limited sorption, where the amount sorbed on
the aquifer solids increases with exposure time. These results are in qualitative agreement with the results
of batch laboratory studies reported in Section 8. There is some uncertainty in the estimates of retardation
due to the difficulties in performing the long-term transport experiments. However, for the purposes of
the field experiments and modeling exercises, the retardation effects were fairly well established.
The results of the field experiments are also in qualitative agreement with the results of the batch
laboratory experiments. The rank order of retardation of TCE > trans-DCE and cis-DCE > vinyl chloride
(Table 5.5), is consistent with Kd values given in Table 8.3, based on the headspace method.
Mathematical Simulation of the Tracer Test Results
Preliminary mathematical modeling of the results of the tracer experiments has been performed
using 1-D and 2-D models. The semi-analytical model, RESSQ, developed by Lawrence Berkeley
Laboratory and described by Javandel et al. (1984) was used to simulate 2-D advective transport under
the injection, extraction and natural gradient conditions of the tracer experiments. 1-D analytical solutions
were used to estimate dispersion coefficients and to determine if a 1-D modeling approach could be used
in the development of a numerical model to simulate the biostimulation and biotransformation processes.
The RESSQ model was used to estimate (1) the area! extent of the injected fluid front that develops
around the injection well and observation wells, (2) the fluid residence times from the injection well to the
observation wells, and (3) the degree of recovery of the injected fluid at the extraction well.
Simulations were performed to illustrate the original design of the well field to permit simultaneous
experiments along three experimental legs. The model input parameters were a fluid injection rate of
0.5 l/min at three wells, an extraction rate of 8 l/min, regional flow velocity of zero, a porosity of 0.35, and
an aquifer thickness of 1.2 m. Figure 5.13 shows the results of the simulations. An injected fluid front of
uniform size develops around each of the three experimental legs. The maximum width of the front is
approximately 1.6 m in the vicinity of the S1 and S2 observation wells.
Figure 5.14 shows the fronts that develop when a regional groundwater velocity of 300 m/yr in a
northerly direction is imposed on the simulation discussed above. The front around the east injection well
is shifted northward due to the groundwater flow. The regional flow leads to a thinning of the front along
the southern leg and a broadening along the northern leg. These results indicated that the southern leg
(SI,S1,S2,S3) should be used in the experiments for the following reasons: 1) the injected fluid
supplying the nutrients becomes less dispersed, and hence a more dense microbial population can be
stimulated, and 2) the injected tracers and chlorinated hydrocarbons can be most effectively recovered at
the extraction well by injecting upstream of the natural groundwater flow. The area dominated by the
injected fluid does become smaller, however, which helps explain the dilution of the injected fluid by the
native groundwater that was observed in the tracer experiments.
Simulations were performed with the RESSQ model to determine whether the predicted fluid
residence times are in the range of values estimated by the tracer tests. The model predicted fluid resi-
dence times of 8 hrs and 21 hrs for wells S1 and S2, respectively. These results were in agreement with
the tracer test values given in in Table 5.2. The following aquifer properties were used in the simulation: a
regional fluid velocity of 300 m/yr, a porosity of 0.35, and an aquifer thickness of 1.25 m. These values
are in agreement with the measured and estimated values. The simulations indicate that the injected fluid
should be totally captured by the extraction well under these conditions. The tracer tests, however,
indicated that only 60 to 70% of the bromide was captured. The reason for this lower degree of capture is
unknown, but heterogeneities in aquifer properties is a probable cause.
The simulations indicate that the region near the injection well does not conform to uniform flow, but
that the flow is nearly uniform at distances of more than 0.5 m from the point of injection, and hence in the
region of the observation wells. To determine the degree of dispersion required to fit the observed
59
-------
•I
NO
GRADIENT
I
(meters)
Figure 5.13. RESSQ simulations of the injected fluid fronts which develop under induced flow
conditions of the tracer experiments with no regional flow.
«r
8-
8
• '
300 M/YR
I 1
1
(meters)
Figure 5.14. RESSQ simulations of the injected fluid fronts which develop under induced flow
conditions of the tracer experiments with a regional flow of 300 m/yr.
60
-------
breakthrough response at the S1 and S2 wells, 1-D simulations were performed. The nonlinear least-
squares fitting program described by van Genuchten (1981) was used in fitting the data to the solution to
the 1 -D convective-dispersive transport equation.
Figure 5.15 shows the fit to the DO breakthrough response at the S-2 observation well in the
Tracer4 experiment. A reasonably good fit is obtained with the 1-D model, with a resulting Peclet number
(Pe) of 6.6, which corresponds to an aquifer dispersivity of 0.33 m (Length/Pe). Model fits of the data
from the Br, DO, and methane experiments were performed for the S1 and S2 wells. The best-fit Peclet
number based on the S1 well ranged from 2.7 to 4.0 with an average value of 3.1. The values based on
data from the S2 well ranged from 3.4 to 6.6 with an average value of 4.4. The resulting average
dispersivities were 0.32 and 0.45 m for the S1 and S2 wells, respectively. The 1-D analysis resulted in
best-fit dispersivity values similar to those obtained from the analysis of the S1 and S2 data. The results
indicate that 1-D transport modeling is of value in the initial stages of experimental design and data
interpretation, when complex biostimulation and biotransformation processes must be taken into
consideration. A more detailed analysis of the bromide tracer data is presented in Section 12.
PULSED INJECTION
To enhance the effectiveness of biostimulation, it was decided to introduce the methane (primary
substrate) and oxygen (electron acceptor) as alternating, timed pulses. This decision was based upon two
essential requirements: 1) the need to avoid clogging of the injection well and borehole interface, and 2)
the need to achieve as uniform a distribution of microbial growth as possible throughout a substantial
portion of the aquifer. Failure to fulfill the first requirement would cause loss of hydraulic capacity and
premature termination of our experiments since the drastic chemical measures such as chlorination or
V=0.105 m/hr D=.035 m2/hr L=2.2 m
1.1
1 -
0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0
a a
—r—
20
40
TIME (MRS)
Figure 5.15. Fit of the 1-D advective-dispersion transport model to the breakthrough of DO at the S2
observation well during the Tracer4 test.
61
-------
strong acid treatment, that are customarily employed to rejuvenate clogged wells, would interfere with
biostimulation. Failure to satisfy the second requirement would lead to conditions of extremely high
microbial densities near the injection point and low bacterial populations elsewhere. This would inhibit
degradation of halogenated aliphatic compounds. It was anticipated that introduction of methane and
oxygen as alternating timed pulses would assure their separation in the injection well and borehole, thus
discouraging biological growth in that critical region. Subsequently, gradual mixing of the methane and
oxygen would occur, owing to the action of hydrodynamic dispersion and associated mixing processes
during transport through the aquifer. This would stimulate growth of methanotrophic bacteria over the
mixing zone. In designing the pulsed injection system, two important variables had to be selected: 1) the
ratio of the individual pulses of methane and oxygen, and 2) the overall pulse length.
The ratio of the individual pulses of methane and oxygen can be estimated approximately from
knowledge of the stoichiometry of methane oxidation. The oxygen requirement for complete oxidation of
methane is 2 moles oxygen per mole methane, which corresponds to a mass ratio of 4 g O2 per gram
methane. In choosing the pulse lengths, the concentrations achieved by the saturation columns for oxy-
gen and methane also must be taken into account.
The overall pulse length was evaluated by employing a transport model that incorporates a periodic
input (Valocchi and Roberts, 1983). The form of periodic input that corresponds most closely to the case
of alternating inputs of methane and oxygen is the rectangular pulse, or saw-toothed wave. The model of
Valocchi and Roberts (1983) takes into account the effects of advection, dispersion, and sorption on
transport and mixing of rectangular pulses under conditions of uniform flow. Although the situation at the
Moffett Feld site certainly differs appreciably from the simple case of uniform flow in a homogeneous
medium, the model computations based on the idealized case are instructive in exploring the effects of
pulse length on mixing, and serve as a point of departure for experimental design.
In the absence of reaction, the normalized amplitude ratio is the most convenient measure of the
degree to which the pulses remain separated during transport, or conversely the degree to which mixing
has occurred. The amplitude ratio is the ratio of observed magnitude of concentration fluctuations mea-
sured at an observation a distance x removed from the injection point to the magnitude of the fluctuations
measured at the injection point. The amplitude ratio varies from zero to unity: a value near zero means that
concentration fluctuations are damped nearly completely, and signifies virtually complete mixing over the
distance traversed, whereas a value near unity implies negligible mixing.
Model computations were conducted under conditions simulating those at the Moffett site. The
important variables were the integral distance, x; the pore water velocity, u; and the Peclet Number for dis-
persion, Pe. The values for the simulation were chosen as x = 1 m, u = 0.12 m/h, and Pe = 5
(dimensionless), to correspond to the results at the nearest observation, S1, based on the results of the
early tracer tests., i.e., the dissolved oxygen breakthrough in the initial stages of the Tracer4 set. The
computation's results (Figure 5.16) indicated that substantial mixing over a transport distance of 1 m (the
distance from SI to S1) would be attained using a pulse length on the order of several hours, and that
pulse lengths on the order of several tens of hours would prevent adequate mixing prior to the first obser-
vation well.
To test the model, toward the end of the Tracers experiment the dissolved oxygen injection was
switched to an on/off mode, with pulse lengths chosen to span the range of potential choices for experi-
mental operation, i.e., less than one hour to 12 hrs. The observed values are shown in Figure 5.16 as
open circles.
The observations show qualitatively the kind of trend predicted by the model: with short pulses
(< 1 hr), the mixing is complete within the first meter, but, as the pulse period is increased to several
hours, substantial concentration fluctuations begin to appear at the observation well, indicating that mixing
is incomplete. The prediction does not agree quantitatively with the data, as the onset of substantial
observed concentration fluctuations occurs at a lower critical value of the pulse period. Indeed, the value
of the Peclet Number must be chosen as 100, rather than the observed value of 5, to simulate the pulsing
62
-------
o
cc
LU
Q
ID
H
_l
Q.
X = im
U =O.I2m/h
O OBSERVED
PREDICTED
Pe = 5
Pe=IOO
I 10
PERIOD, 2T(h)
Figure 5.16. Comparison of predicted and observed effects of dissolved oxygen pulsing.
data satisfactorily. These deviations may well be caused by deviations from the model assumptions of
uniform flow in a homogeneous medium. Nonetheless, the qualitative agreement between predicted and
measured values for the effect of pulsing was deemed adequate as a framework for experimental design
of the biostimulation phase.
SUMMARY
The tracer experiments provided the needed information on the transport characteristics of the test
zone. The tests also demonstrated that controlled, reproducible experiments could be performed at the
field site.
Natural gradient tracer tests indicated the presence of a strong regional component to fluid flow in a
northern direction. The tests combined with RESSQ model simulations indicated that the induced flow
experiments would be best performed by injecting into the upgradient SI well and extracting from the P
well downgradient.
Induced gradient bromide tracer tests established fluid residence times in the test zone. The tests
also demonstrated that reproducible transport experiments could be performed at the test site. The tests
showed that by increasing extraction rates, complete recovery of the injected bromide could be achieved.
This ability to completely recover the injected fluid provided confidence in conducting subsequent
biotransformation experiments with vinyl chloride.
The dissolved oxygen transport tests demonstrated that DO was not consumed and was
transported like bromide. Due to the presence of organic matter after the test zone was biostimulated, a
minor amount of DO consumption was observed.
63
-------
The chlorinated organic transport experiments showed the retardation of these compounds
compared to the bromide tracer. The rank order of retardation was as follows: TCE > trans-DCE and cis-
DCE > vinyl chloride. Retardation estimates based on the center of mass of the breakthroughs curves
were a factor of 1.5 to 2 greater that those based on the time to reach 50% breakthrough. The higher
values are probably more representative of the actual retardation since they include the effects of long-
term sorptive uptake onto the aquifer solids. Thus, the organic transport experiments provided required
information on the sorption and retardation process of the organics in the test zone.
Modeling studies helped confirm the results of the tracer tests. Simulations using the model
RESSQ indicated that the fluid residence times observed in the field are reasonable, when the fluid
injection and extraction are superimposed on the strong regional flow field. The 2-D simulations indicated
that uniform flow conditions in the direction of the monitoring wells were approached a short distance from
the injection well.
Simulations assuming 1-D advection-dispersion transport, provide a good match to the observed
DO breakthrough at observations wells. The match indicates that the transport to the observation wells
can be reasonably approximated by the assumption of uniform flow.
Simulations of the response to pulsing of DO agreed with theoretical predictions. The requirement
of lower dispersion coefficients (higher Peclet numbers) to match the pulse response, compared to those
required to match the complete breakthrough suggests deviations from the assumption of uniform flow.
Nonuniform flow was probably the result of aquifer heterogeneities.
64
-------
SECTION 6
RESULTS OF BIOSTIMULATION AND BIOTRANSFORMATION EXPERIMENTS
Lewis Semprini, Gary Hopkins, and Paul Roberts
Biostimulation and biotransformation experiments were performed in three successive field
seasons. Performing the experiments in this manner permitted the results of the previous field season,
and information gained in laboratory studies to be incorporated into the next season's experimental
design. Table 6.1 includes the series of experiments that were performed in each of the field seasons. In
the latter years the number of chlorinated aliphatics studied increased, and the sequence of experiments
was changed. These changes over the course of the experiments enabled the in-situ biotransformation
process to be effectively demonstrated, and supplied detailed information on processes governing the
rates of biotransformation.
The first season of testing focused on the ability to biostimulate a native population of methan-
otrophic bacteria, and on the biotransformation of TCE in the stimulated test zone. The biostimulation
experiment was initiated by introducing methane and oxygen in the absence of TCE. After biostimulation
was achieved, TCE was added, together with methane and oxygen, to assess TCE biotransformation.
TABLE 6.1. EXPERIMENTS AND PROCESSES STUDIED
Experiment
First Season
Biostiml
Biotranl
Biotran4
Duration
9/05/86-
9/30/86
9/30/86-
10/21/86
12/10/86-
12/31/86
Chemicals
Injected
Methane
DO
Bromide
Methane
DO
TCE
Methane
DO
Bromide
TCE
Average3
Cone.
(mg/i)
5.9±1.1b
20.7±4.3
16614.5
5.7±1.2
22.2+1.7
0.097±0.024
5.210.9
23±1.5
159116
0.05110.010
Process
Studied
Biostimulation of native methane-
utilizing bacteria. Alternating
pulse injection of methane and DO.
Biotransformation of TCE with active
biostimulation. Non-steady-state
conditions.
Biotransformation of TCE with
active biostimulation. Steady-state
conditions.
TABLE 6.1 cont.
65
-------
TABLE 6.1 (cont.)
Second Season
Tracers
Biostim2
Decmeth!
Peroxid2
Third Season
Tracerl 1
Tracer! 2
BiostimS
7/06/87-
8/15/87
8/17/87-
1 0/26/87
10/27/87-
11/08/87
11/30/87-
1 2/23/87
8/10/88-
1 0/1 0/88
10/20/88-
1 0/20/88
10/20/88-
11/23/88
DO
Bromide
TCE
cis-DCE
trans-DCE
Methane
DO
Bromide
TCE
cis-DCE
trans-DCE
DO
TCE
cis-DCE
trans-DCE
H2O2
Methane
TCE
cis-DCE
trans-DCE
Bromide
TCE
cis-DCE
trans-DCE
Bromide
TCE
cis-DCE
trans-DCE
VC
Methane
DO
Bromide
TCE
cis-DCE
trans-DCE
VC
Formate
Methanol
14.3±1.3
78±6
0.048±0.010
0.110+0.036
0.112±0.039
5.3±0.9
23.4±2.0
44±4
0.036±0.006
0.091+0.025
0.092±0.026
24.5±1.1
0.045+0.005
0.136±0.022
0.095+0.013
45
10.6±1.5
0.054±0.006
0.143±0.026
0.10±0.02
72±5
0.047±0.006
0.085±0.013
0.050+0.007
44±3
0.042±0.003
0.100±0.011
0.054+0.007
0.04410.007
6.6±0.7
21.3±0.7
45±2
0.046±0.003
0.100±0.015
0.052±0.009
0.034±0.007
73
16.9
Transport and breakthrough of
bromide, TCE, cis-DCE, and
trans-DCE before biostimulation.
Simultaneous biostimulation and
biotransformation fof TCE, cis-DCE,
and trans-DCE.
Test if biotransformation occurs
without active methane
utilization.
Test if increased biomass improves
biotransformation by substituting
hydrogen peroxide for oxygen.
Transport and breakthrough of
bromide, TCE, cis-DCE, and
trans-DCE before biostimulation.
Transport and breakthrough of
bromide and vinyl chloride while
continuing injection of TCE,
cis-DCE,, and trans-DCE
Simultaneous biostimulation-bio-
transformation of TCE, cis-DCE,
trans-DCE, and vinyl chloride.
Transient testing of formate and
methanol as substitute for
methane as electron donors.
a Average values for methane, DO, H2O2, formate, and methanol are time-averaged due to pulsing.
b Standard deviation of the injection concentrations.
66
-------
In order to provide a more direct and convincing demonstration of the in-situ biotransformation
process, the experimental sequence was changed, and additional compounds were studied during the
second field season. This was accomplished by first adding TCE, cis-DCE, and trans-DCE before
biostimulation, to achieve nearly complete breakthrough of these compounds at the observation wells.
The test zone was then biostimulated, while continuing to add the halogenated organics. This sequence
permitted the direct observation of the effect of biostimulation on biotransformation. In the second
season, several supplementary transient experiments were also conducted: 1) hydrogen peroxide was
introduced into the test zone as a source of oxygen to enhance a greater biomass of methane-utilizing
bacteria, and 2) the methane concentration was varied stepwise to evaluate its effect on the biotransfor-
mation rates of the chlorinated aliphatics.
The sequence of field experiments in the third season was similar to that of the second, except that
the biotransformation of vinyl chloride was also studied. Transient experiments included the substitution
of formate and methanol for methane after biostimulation was achieved to investigate the influence of
methane inhibition on the rates of biotransformation of the chlorinated aliphatics.
RESULTS OF THE FIRST SEASON OF FIELD TESTING
Biostimulation Experiment
The first biostimulation experiment (Biostiml) was conducted to determine (1) the ease of stimu-
lating indigenous methane-oxidizing bacteria, (2) stoichiometric requirements for methane and oxygen,
(3) information on the rate of growth, (4) the spatial extent over which biostimulation could be achieved,
and (5) the effectiveness of the alternate pulsing of methane and DO for distributing microbial growth.
Groundwater containing either methane or oxygen was pulse-injected alternately at the SI injection
well. The injection system used is described in detail in Section 4. The pulse cycle was varied during the
course of the experiment, from less than 1 hr during start-up (to ensure that pulsing would not interfere
with growth) to a 12-hr period during the latter stages of the experiment. This approach ensured that
growth was distributed in the test zone. The pulse length ratio was 1:2 (methane:oxygen), resulting in
time-averaged injected fluid concentrations of 5.4 and 19 mg/l for methane and oxygen, respectively. No
additional nutrients (N or P) were added to the injected fluid.
Figure 6.1 shows the dissolved oxygen (DO) concentration as a function of time at the three
observation locations: S1 (1 m), S2 (2.2 m), and S3 (4 m). During the first 140 hrs of the experiment
there was little evidence of DO consumption. Within 50 hrs of injection, DO concentrations reached
maximum steady-state values, which decreased slightly with distance from the injection well. During this
period, the bromide tracer and methane achieved the same degree of fractional breakthrough as DO,
indicating that this initial decrease in concentration with distance resulted primarily from dilution by the
indigenous groundwater and not from microbial consumption. The maximum dilution occurred at the
extraction well, due to the 1:8 ratio between the injection and extraction flow rates. This early response
indicated that microbial activity in the test zone was sufficiently low so that no measurable DO utilization
resulted.
The first signs of an onset of DO consumption were observed in the extraction well and the S3
observation well after approximately 200 hrs of injection. Owing to the increasing utilization removal by
microorganisms with distance, the decrease in DO was greatest at the observation wells farthest from the
injection well.
67
-------
SlQ-Ei S2X-X S3O-O Extract A-A
0.0
100
200 300
Time (Hours)
400
500
Figure 6.1. Dissolved oxygen (DO) concentration response at the observation wells and extraction well
due to biostimulation of the test zone.
Figure 6.2 shows the similar response between methane and DO at the S2 observation well,
decreasing as expected for methane oxidation by methanotrophs. At early time (0-50 hrs) methane and
DO broke through similarly to the bromide tracer, indicating no retardation and minimal DO consumption.
The relatively rapid decrease in the methane and oxygen concentrations over the period between 200 to
430 hrs indicates microbial growth. The methane concentrations at well S2 decreased below the
detection limit (0.25 mg/l) after 430 hrs of injection, while a residual oxygen concentration of approximately
3.5 mg/l was maintained. Based on these values, the mass ratio of oxygen to methane consumed was
approximately 2.5, which is significantly lower than the ratio of 4 that would be required for complete
methane oxidation. The lower measured ratio was also expected, however, since some of the methane
carbon utilized would be associated with biological growth.
The decrease of methane concentration below the detection limit at the S2 observation well after
430 hrs indicated that microbial growth was becoming concentrated near the injection well. The pulse
cycle period was then lengthened to 12 hrs in order to prevent biofouling near the wellbore. The resulting
response at the S2 well is shown in Figure 6.3. Peak methane values then increased from below detec-
tion to maximum values of approximately 1 mg/l. Peak methane concentrations occurred when minimum
DO concentrations were observed, consistent with transport theory. The response to pulsing at the 81
well is shown in Figure 6.4. Peak methane and oxygen concentrations are less strongly attenuated than
those at the S2 well. This was expected; since the S1 well is closer to the injection well than S2, less
dispersive mixing occurs. Long pulse cycles were continued throughout the first year of experiments.
Based on the low levels of methane that were observed consistently at the monitoring wells, the pulsing is
believed to have promoted a spatially distributed microbial population in the test zone. Biofouling of the
wellbore and sand pack was thus limited by the pulsing methodology.
68
-------
DOO-O Methane
0.0 100 200 300
Time (Hours)
400
500
Figure 6.2. Methane and DO response at the S2 observation well due to the biostimulation of the test
zone.
24
3 20
en
S IB
c.
-2 12
tO
t.
-P
c.
0)
CJ
c
o
CJ
DOO-O MethaneA-A
400
440
480 520
Time (Hours)
560
600
Figure 6.3. The response at the S2 observation well resulting from 8- and 4-hr alternate pulses of DO
and methane.
69
-------
DO 0-0
Methane A—A
— 20-
S 16-
C
o
•l-t
-p
ID
C_
•P
C
0)
CJ
C
o
CJ
400
440
480 520
Time (Hours)
560
600
Figure 6.4. The response at the S1 observation well resulting from 8- and 4-hr alternate pulses of DO
and methane.
Biotransformation Experiments
The initial biotransformation experiments were performed after the test zone was biostimulated.
After biostimulation TCE was injected continuously over a three-month period. The initial concentration of
TCE averaged 97 u.g/1. Methane and oxygen were continuously pulse-injected during this period in order
to maintain biostimulated conditions. The results of these initial TCE experiments are discussed in detail
in a previous EPA report (Semprini et al., 1988), so they will be briefly summarized only.
In these experiments, estimates of biotransformation were assessed by: 1) comparing with TCE
breakthrough in the pseudo-control experiment (Tracers); 2) comparing the steady-state fractional
breakthroughs of TCE to those of bromide as a conservative tracer; and 3) computing mass balances on
the amount of TCE injected and extracted, and normalizing with respect to bromide mass balances. The
three methods were found to agree closely, with all indicating that 20 to 30 percent of the TCE added to
the system was biotransformed.
Comparisons with bromide as a conservative tracer are considered the most accurate estimate of the
degree of transformation achieved. Before the extent of biotransformation was estimated by this method,
the TCE injection concentration was lowered from 97 to 51 u.g/1 to minimize sorptive loss due to slow
uptake of TCE onto aquifer solids. The resulting steady-state fractional breakthrough of TCE is shown in
Figure 6.5. Biotransformation is indicated by the lower breakthrough at the S2 well compared to the S1
well, with both wells having a fractional breakthrough significantly lower than a value of unity. The results
of a bromide tracer test performed during this same time period are shown in Figure 6.6, for comparison.
The steady-state fractional breakthroughs for bromide at each observation well (S3 not shown) are
significantly higher than those of TCE, indicating transformation of TCE. Using equation (4-1) the degree
of biotransformation of TCE was estimated based on the steady-state fractional breakthroughs of the two
70
-------
3
CO
i
>
O
OJ
Q;
(D
3
8
CT
CD
s
I
CO
CD
a
CD
I
CO
CD
I
CO
I
Norm. Concentration (C/Co)
o>
in
I
p>
01
2". CO
Norm. Concentration (C/Co)
21
o »
3^
CD CD
5' 3
3 O
CD 3
X 3
T3 D)
CD «
3. N
3.
CO
O
-------
chemicals. The estimated transformation based on each observation location is as follows: S1,17%; S2,
28%; S3, 23%; and extraction, 16%. Biotransformation was indicated only in the zone of active methane
utilization. In the area between S2-S3 and the extraction well, no methane was present to support
bacterial growth, and no additional degradation of TCE was observed.
The concentration of 1,1,1-trichloroethane (TCA), found as a background contaminant in the test
zone, also was monitored during the biotransformation experiments. During the first season the average
field concentration of TCA, based on extracted well samples, was 65 u.g/1. The average concentration of
TCA recycled in the injected fluid was 56 u.g/1, due to partial stripping in the DO and methane sorption
columns. During the same steady-state period, as presented in Figure 6.5 for TCE, the average TCA
concentrations at monitoring wells S1 and S2 were 53 and 55 u,g/l, respectively. These concentrations
were essentially identical to the injected levels. This suggests that little biotransformation of TCA occurred
during transport through the test zone. After normalizing for the degree of mixing of the injected fluid with
the native fluid (based on bromide tracer test data shown in Figure 6.6), over 95% of the estimated TCA
concentration was observed at the S1 and S2 observation wells. The estimate indicates minimal
transformation of TCA. There is a large error associated with this estimate, however, due to the fluctuating
concentration of TCA in the indigenous groundwater.
At the end of the first year's biostimulation experiment, TCE addition was stopped. Extraction and
monitoring of the TCE elution continued for an additional 3 months. The observations indicate a slow
release of TCE sorted onto the aquifer solids. A TCE mass balance over the complete biotransformation
experiment showed 10.1 g of TCE were injected and 4.5 g were extracted. This represents a recovery of
45%. During this same overall period, 65 to 70 percent of the bromide tracer was recovered. The lower
recovery of TCE compared to bromide provides additional evidence that 25-30% of the injected TCE was
biodegraded.
It could be argued that increases in the sorption capacity of the aquifer as a result of increased
organic carbon due to biostimulation is responsible for these observations. We disagree on the following
grounds. First, the amount of biomass produced by biostimulation was too small to account for the
observed effect. Model estimates of the steady-state biomass stimulated are presented in Section 13.
Based on these estimates the biostimulated organic carbon would increase the organic carbon content of
the aquifer by only 1%, which is a very minor amount. Also the results of tracer experiments presented in
Section 5 showed no significant increase in TCE retardation over the three years of testing, consistent
with the above estimate.
The experimental approach taken also minimized the effects of sorption interactions. Injection
concentrations of the chlorinated solutes were lowered before assessing the extent of biotransformation,
so that desorption and not sorption was likely occurring. In making mass balances, the elution of the
chlorinated solutes from the aquifer solids was monitored for several months in order to complete mass
balances.
Summary of the First Season's Results
The first season of field testing demonstrated that indigenous methanotrophic bacteria could be
easily biostimulated by the introduction of methane and oxygen to the test zone. Methane and DO were
found not to be strongly sorbed and were transported like the bromide tracer. No additional nutrients (N,
P) were required. Biostimulation of methanotrophs was demonstrated by the uptake of methane and DO
after a lag period of approximately 2 weeks. The stoichiometric ratio of methane and DO consumption was
consistent with expectations for an active methanotrophic community. Under the flow conditions pre-
vailing in the test zone, complete methane utilization occurred over a flow path of 2 m, as long as the
pulse cycle duration was short (< 1 hr). Alternate pulsing of methane and DO with long cycles (12 hr)
72
-------
helped distribute methane, and presumably the microbial biomass, throughout the test zone, which mini-
mized biofouling near the injection well.
TCE biotransformation experiments performed after the test zone was biostimulated indicated that
20 to 30% of the TCE was transformed during transport through the biostimulated zone. Different
methods of estimating the degree of transformation yielded similar results. Measurements of TCA as a
background contaminant indicated insignificant transformation of TCA.
Despite the fact that only partial TCE transformation was observed, the results of the first year of
testing demonstrated that, if sufficient control and care are exercised in experimentation, quantitative evi-
dence of degradation can be obtained in the field under conditions representative of real aquifer systems.
RESULTS OF THE SECOND SEASON OF FIELD TESTING
In the second season of field testing, the experimental design was modified to provide a more direct
and convincing demonstration of biotransformation. The chlorinated organic solutes of interest were first
added to the test zone to achieve steady-state fractional breakthroughs before biostimulation. The test
zone was biostimulated subsequently through the addition of methane and oxygen, while continuing to
add the chlorinated solutes. The resulting change in concentration of the chlorinated organic solutes due
to biostimulation was observed directly. The number of contaminants studied was also increased. Cis-
DCE and trans-DCE were added together with TCE. Laboratory studies of Henson et al. (1987, 1988)
indicated methane-utilizing mixed cultures degraded cis-DCE and trans-DCE more rapidly than TCE while
TCA was degraded more slowly. Fogel et al. (1986) presented similar results from mixed culture studies. It
was desired to see if similar results would be obtained in the field demonstration.
Transient experiments were performed as well to study the effect of methane concentration on bio-
transformation and to determine whether continuous methane addition was required for transformation to
occur. Finally, experiments were performed in which peroxide was substituted for oxygen in order to
increase the amount of methane injected.
The Organic Addition Experiment
The objective of this experiment (Tracers) was to establish steady-state concentrations of the
organic contaminants in the test zone before restimulation. The experiment provided information on the
transport of the three organics relative to bromide and the degree of capture of injected bromide at the
extraction well. The results of these transport experiments are discussed in Section 5.
Before injecting the chemicals, a background contaminant was observed in the test zone's
groundwater that co-eluted with trans-DCE during GC analysis. The peak area of this background con-
taminant was equivalent to 16 to 27 u,g/l of trans-DCE. GC-MS analyses identified this background
contaminant as 1,1-dichloroethane (1,1-DCA). Later results showed that 1,1-DCA was not transformed to
a significant extent in the test zone. Thus, the concentration of trans-DCE was corrected for the presence
of this co-eluting compound by subtracting an average background concentration as measured in wells
N1, N2, N3 outside the test zone during the second season's experiments.
The average concentrations injected were 112,110, and 48 ug/l for trans-DCE, cis-DCE, and TCE,
respectively. Aerobic conditions were maintained by the addition of dissolved oxygen in the injected fluid
at an average concentration of 14 mg/l. Gradual increases toward injected concentrations in TCE, cis-
DCE, and trans-DCE were observed over the 40 day injection period. The rank order of retardation was as
follows: TCE > trans-DCE > cis-DCE.
73
-------
Table 6.2 summarizes the quasi-steady-state fractional breakthroughs achieved for bromide and the
chlorinated organic solutes. Complete bromide breakthrough was observed at the S1 and S2 wells, indi-
cating negligible dilution of the injected fluid by the indigenous groundwater. Mass balances, (Section 5),
indicated 100 % recovery of the injected bromide by the extraction well. The organic solutes reached
quasi-steady-state fractional breakthroughs of 90% to 95% at the S1 well, indicating negligible transfor-
mation by biotic or abiotic processes during this quasi-control stage before biostimulation.
Cis-DCE reached the highest fractional breakthrough, followed by trans-DCE, and TCE. This order
is consistent with the degree of retardation, with TCE being the most strongly sorbed and having the low-
est degree of fractional breakthrough. TCE concentrations at the S2 well and the extraction well increase
very slowly due to advective, dispersive, and sorptive processes. Thus TCE, and to lesser extents cis-
and trans-DCE, probably did not achieve a maximum steady-state breakthrough concentration at all
locations before the start of the biotransformation experiment.
TABLE 6.2. TRACERS EXPERIMENT - PERCENTAGE BREAKTHROUGH OF THE CHLORI-
NATED SOLUTES AND BROMIDE AT THE OBSERVATION WELLS
Substance WellSI WellS2 Well S3 Extraction
Bromide
TCE
cis-DCE
trans-DCE
100 + 0.73
94 + 2
94 + 3
94 + 3
98 + 3.0
84 + 2
9413
9313
83 1 0.7
6812
7213
72 + 5
13.610.1
10.310.2
12.410.5
11.61N.D.
a Standard error of the mean.
The Biostimulation-Biodegradation Experiment
The combined biostimulation-biotransformation experiment (Biostim2) immediately followed the
organic solute addition experiment. Operating conditions of this experiment are presented in Table 6.1.
While injection of organic solutes continued, methane and oxygen were added in short pulses of 20 and
40 min, respectively. Average methane and DO injection concentrations were in the range of those used
in the first season's tests. Methane and DO uptake occurred very rapidly, with essentially no lag observed.
The response indicated that some of the methane utilizers stimulated in the first season were still present
to immediately initiate methane utilization at the start of the second season.
Figure 6.7 shows the simultaneous response of methane, DO and trans-DCE at the S1 well. The
decrease in all three components due to biological activity is apparent after about one day of injection.
Methane decreased below the detection limit after 72 hrs of injection, while dissolved oxygen and trans-
DCE showed a continued gradual decrease in concentration. The reduction in trans-DCE concentration,
coincidental with the consumption of methane, provides direct evidence of its in-situ biotransformation in
response to the biostimulation of the methane-oxidizing bacteria.
Figure 6.8 shows the S2 responses of all three organics resulting from biostimulation of the test
zone. Bromide tracer results at S2 are also shown for comparison. A five-point running average is pre-
sented to clearly show the trends for each compound. The decreases in concentration from normalized
values near unity are apparent, especially for cis-DCE and trans-DCE. During this period, the bromide
tracer showed complete breakthrough to injected concentrations, demonstrating that decreases in the
74
-------
1.2
t-DCEo-e DOO~0 Methane A-A
E
C_
o
z
40
eo iso
Time (Hours)
160
200
Figure 6.7. Response at the S1 well of methane, DO, and trans-DCE in the second season's biostimu-
lation-biotransformation experiment, Biostim2.
1.2
t-DCEo-Q c-DCEx-x TCE<&-O Bromide
80
160 240
Time (Hours)
320
400
Figure 6.8. Response at the S2 well of trans-DCE, cis-DCE, and TCE in the second season's biostimu-
lation-biotransformation experiment, Biostim2.
75
-------
organic concentrations were not related to advective or dilution losses in the system, but were the result
of biotransformation. Trans-DCE is shown to decrease in concentration most rapidly, followed by cis-DCE
and TCE. The more rapid decrease in trans-DCE concentration, compared to cis-DCE and TCE, results
primarily from its having a faster rate of transformation. Other processes affecting the transient response
include sorption and microbial growth. These processes will be discussed in greater detail in the biotrans-
formation modeling chapter.
In order to distribute the microbial growth population more evenly throughout the test zone, the
pulse time was increased to the standard 12-hr cycle used in most experiments. Steady-state bio-
transformation conditions were achieved during this period of longer pulse cycle, as shown in Figure 6.9.
Fractional breakthroughs of the chlorinated aliphatics at the S2 observation well indicate the following
extents of transformation: TCE, 20%; cis-DCE, 50%; and trans-DCE, 80%.
Production of a Transformation Intermediate
At the onset of the second year's biotransformation experiment, an intermediate product was
detected during the chlorinated organic's GC analysis. The sensitivity to electron capture detection indi-
cated that this transformation product was halogenated. No peaks had appeared with a similar GC reten-
tion time during the previous year's study with TCE, indicating that the intermediate was associated with
either trans-DCE and/or cis-DCE transformation. Janssen et al. (1987) reported the formation of a rela-
tively stable trans-DCE oxide (epoxide) from the biodegradation of trans-DCE by a consortium of methan-
otrophic bacteria. It is plausible that such an compound might be formed as an intermediate product of
trans-DCE degradation. Figure 6.10 shows the increase in the intermediate product's concentration that
occurred simultaneously with the decrease in trans-DCE concentration. The production of the epoxide
appears to be associated with the transformation of trans-DCE. Results of GC-MS analyses, discussed in
Section 7, confirmed that this intermediate was indeed the trans-DCE oxide (epoxide).
Quantification of the epoxide concentration was made, using an epoxide standard synthesized in
our laboratory. Stored peak areas were converted to concentrations based on the relative response of
the epoxide and trans-DCE to ECD detection. There is some uncertainty in the epoxide concentration
estimated, since the epoxide standard had to be synthesized and purified. Despite this uncertainty in the
absolute concentration, the epoxide appears to represent only 5 to 10% of the trans-DCE degraded.
Transient Methane Addition Experiments
After steady-state biotransformation conditions were achieved, a series of transient methane addi-
tion experiments were performed to assess to what degree lower methane concentrations influence
biotransformation rates. Moreover, results were used to determine if biotransformation continued after
methane addition was temporarily ceased. Figure 6.11 shows the response of TCE, cis-DCE and
methane at the S1 well due to these perturbations. During the period of 100-200 hrs, lower injected
methane concentrations were produced by shortening the methane pulse length by a factor of two. Over
this short study period no significant change in the degree of biotransformation resulted. Over the period
of 275-475 hrs, methane addition was ceased, while input of the chlorinated solutes and DO continued.
Gradual increases towards injected concentrations of cis-DCE, TCE, and trans-DCE (Figure 6.12),
resulted. The increases indicated that transformation diminished in the absence of methane. The slow
increase, due to sorptive retardation, makes it difficult to determine if biotransformation stopped
immediately after methane was no longer available. After 475 hrs, methane addition was restarted to
restimulate the test zone. A rapid decrease in concentration of the halogenated aliphatics to previous
levels occurred as biotransformation commenced. These data demonstrate that the biostimulated
population of methane-utilizing bacteria required active methane utilization for the biotransformation of
halogenated aliphatics to occur.
76
-------
c
o
•ft
-p
(0
c_
4J
C
-------
c-DCEx-x TCEO-O Methane
ISO
240 360
Time (Hours)
480
600
Figure 6.11. Response at the S1 well of TCE and cis-DCE to the reduction (100-200 hrs) and termination
(275-475 hrs) of methane addition.
Figure 6.12 shows the response at the S1 well of the epoxide intermediate and trans-OCE during
these transient experiments. A very rapid decrease in the epoxide concentration resulted after methane
addition was stopped. The response, which was much faster than the increase in trans-DCE concentra-
tion, indicates that trans-DCE epoxide formation also ceased immediately after methane addition was
stopped. This rapid decrease also suggests that the epoxide is much less strongly sorbed than trans-
DCE, and/or that transformation of the epoxide continues after methane addition was stopped.
Peroxide Addition Experiment
A transient experiment was performed to determine if greater transformation rates could be
achieved by increasing the methane-utilizing biomass through the addition of greater amounts of
methane to the test zone. To achieve this, the methane pulse length had to be increased and the DO
pulse length decreased. This was accomplished by injecting hydrogen peroxide (H2C-2) as a source of
dissolved oxygen under active biostimulation conditions.
Hydrogen peroxide was injected at a concentration of 272 mg/l, which upon complete breakdown
would produce 128 mg/l of dissolved oxygen. This reduced the electron acceptor pulse time from 8 hrs
(with oxygen) to 2 hrs with hydrogen peroxide, while the methane pulse was increased from 4 hrs to
10 hrs. Based on mass balances, the rate of methane addition was increased by a factor of 2, compared
to the previous studies with oxygen. Theoretically, a corresponding increase in the methane-utilizing
biomass should result.
78
-------
120
t-DCEo-e t-Epoxidex-x
0.0
ISO 240 360 480
Time (Hours)
600
Figure 6.12. Response at the S1 well of trans-DCE and trans-DCE oxide (epoxide) to the reduction
(100-200 hrs) and termination (275-475 hrs) of methane addition.
Analyses for hydrogen peroxide breakthrough at observation wells were performed on samples
obtained manually. No breakthrough of hydrogen peroxide was observed, indicating that the H2O2
reacted rapidly to form oxygen within the first meter of travel in the test zone. Measurements at the S2 well
indicated that methane was transformed completely during periods when oxygen was in sufficient excess.
The stoichiometric ratio of methane to DO consumption appeared to be unchanged compared to earlier
results. Hydrogen peroxide therefore did not appear to inhibit the activity of methane-utilizing bacteria.
Figure 6.13 shows the steady-state fractional breakthroughs of the chlorinated organics at the S2
well during the peroxide addition experiment, corresponding to the first 553 hrs of the graph. The break-
throughs, compared to those in Figure 6.9, indicate no significant enhancement in the degree of trans-
formation was achieved with the increased methane addition made possible by switching to hydrogen
peroxide as a concentrated oxygen source.
Upon completion of the peroxide addition experiment, the injection of oxygen and methane was
resumed, using the standard 12-hr cycle. The time-averaged injection concentration of methane was
reduced to 5.2 mg/l, a factor of 2 lower than the previous experiment with peroxide. These lower methane
conditions correspond to the time interval of 553 to 750 hrs in Figure 6.13. The degree of transformation
remained essentially the same as that achieved during the peroxide addition experiment.
The results of these transient experiments indicated that the addition of greater quantities of
methane to increase methane-utilizing biomass did not result in greater extents of biotransformation. One
possible explanation is that the higher methane concentrations inhibited the rates of transformation of the
chlorinated aliphatics.
79
-------
E
C_
o
2
1.2
t-DCEQ-B c-DCEx-x TCE 0-0
150
300 450
Time (Hours)
600
750
Figure 6.13. Fractional breakthroughs of the chlorinated organics during the peroxide addition
(0-553 hrs) and methane reduction (553-750 hrs) experiments.
Degree of Biotransformation
The degree of biotransformation of the chlorinated solutes was estimated for periods when quasi-
steady-state conditions were achieved. These biotransformation estimates were based on (1) compari-
sons with bromide as a conservative tracer, as given in equation (4.1), and (2) comparison with the
fractional breakthroughs of the organic solutes before biostimulation. For the latter estimate, the fractional
breakthrough of each chemical before biostimulation, as measured during the Tracers experiment (Table
6.2), was substituted for the bromide fractional breakthrough in equation (4.1). The steady-state fractional
breakthroughs of bromide and the chlorinated organic compounds used in the estimates in Table 6.2
were based on as many as 40 measurements at each observation location. The great number of
measurements reduced the mean standard error of the estimate.
The estimates of percentage biotransformation are presented in Table 6.3, with the range of the
95% confidence interval given for each location and chemical. The estimated range of minimum to maxi-
mum percentage biotransformation obtained by the two methods described above, based on the data
from the farther observation wells (S2 and S3), were as follows: TCE, 4 to 30%; cis-DCE, 46 to 58%, and
trans-DCE, 58 to 76%. With TCE, a significantly lower removal resulted from the comparison with the
Tracers results, perhaps because TCE had not reached its true steady-state level in that experiment. It is
believed that the higher removal estimated from the comparison with bromide provides the more accurate
value.
80
-------
TABLE 6.3. PERCENTAGE BIOTRANSFORMATION IN THE SECOND SEASON'S
BIOSTIMULATION-BIOTRANSFORMATION EXPERIMENTS
Comparison with Bromide Fractional
Breakthroughs3-6
Comparison with Fractional Breakthrough
Before Biostimulationb-c
Halogenated
Ethene
TCE
cis-DCE
trans-DCE
TCE
cis-DCE
trans-DCE
Well S1
(%)
10-14
34-40
61-67
0-10
29-37
55-64
Well S2
(%)
21-25
50-58
70-76
4-12
46-54
62-72
Well S3
(%)
23-30
50-58
67-73
3-11
40-48
58-66
Extract
(%)
20-26
55-72
(N.D.)
-6-1
51-69
(N.D.)
a Percentage biotransformation estimates (95% confidence interval shown).
b Values over a time interval of last 150 hrs of the experiment Biostim2 shown in Figure 6.9.
c Fractional breakthrough before biostimulation given in Table 6.2.
Estimates at well S1 indicate that essentially all of the biotransformation of trans-DCE occurred within
the first meter of travel in the zone between the injection and the S1 observation well. This corresponds
to the zone of most active methane utilization. Cis-DCE and TCE appeared to have undergone additional
transformation between the S1 and S2 wells. The extent of TCE transformation was consistent with the
results from the first season of testing.
The extent of transformation of trans-DCE during the steady-state portion of the experiment,
evaluated above, was lower than that achieved during the earlier stage, where greater than 80% transfor-
mation was achieved (Figure 6.8). Cis-DCE, however, shows the opposite result. It is not clear what
caused these changes. The main difference in the experimental conditions was the change to longer
methane-oxygen pulse lengths during the latter steady-state period.
The degree of biotransformation was also determined based on mass balances of TCE and cis-DCE
injected over the course of the experiments, compared to the amount removed from the system by the
extraction well. The concentrations of the extracted fluid were measured for a period of three months after
organic addition was stopped in order to complete the mass balances. Biostimulation conditions were
maintained throughout this period. Estimates are not presented for trans-DCE, since the presence of the
co-eluting 1,1-DCA in the extracted groundwater resulted in a large error in the estimate.
Mass balances showed that 30% of the TCE and 54% of the cis-DCE injected were not recovered
by the extraction well, while 100% of the bromide was recovered over the same period. This lower
recovery of the organic solutes suggests biotransformation and not hydraulic losses from the system. The
complete mass balance estimates of biotransformation are in good agreement with those based on the
steady-state fractional breakthroughs given in Table 6.3. Thus, the estimates derived from the S2 and S3
data are believed to represent the degree of transformation that is occurring over the entire test zone, as
determined using the extraction well data.
Summary of the Second Season's Results
Results of the second season of field testing directly demonstrated that biostimulation of methane-
utilizing bacteria results in the biotransformation of trans-DCE, cis-DCE, and TCE. Trans-DCE was most
rapidly degraded, followed by cis-DCE, and TCE. The maximum extent of transformation achieved in the
81
-------
2-m biostimulated zone are as follows: trans-DCE, 80%; cis-DCE, 55%; and TCE, 20%. Similar estimates
of the degrees of transformation were obtained from complete mass balances on the amount of
chlorinated solute injected and extracted from the test zone.
Transient experiments demonstrated that the transformation of the chlorinated solutes rapidly
ceased upon termination of methane addition in the test zone. Thus, biotransformation in the test zone
required active utilization of the electron donor.
Transient experiments under active biostimulation conditions showed similar degrees of transfor-
mation, despite a two-fold variation in amount of methane injected. Methane-utilizers could be maintained
with hydrogen peroxide used as a source of oxygen; however, higher degrees of transformation, as a
result of stimulating a greater biomass, did not occur. These data suggest that the higher methane con-
centrations may have inhibited the rates of transformation of the chlorinated alkenes.
The formation of trans-DCE oxide resulted from the transformation of trans-DCE. The formation of
the epoxide was found to be very sensitive to biostimulation conditions. A rapid decrease in the epoxide
concentration occurred soon after methane addition was stopped, further confirming the close
association of biotransformation with active methane utilization.
RESULTS OF THE THIRD SEASON OF FIELD TESTING
The biotransformation process was studied in greater detail in the third season of field testing . The
experiments focused on the biotransformation of vinyl chloride along with TCE, trans-DCE, and cis-DCE.
Based on the mixed culture studies of Fogel et al. (1986) and our soil column studies (Section 10), we
anticipated that vinyl chloride would be nearly completely transformed. The incorporation of a Hall detec-
tor in the automated data acquisition system permitted more accurate quantification of the concentrations
of the less chlorinated organics and eliminated the problem of the interference of the trans-DCE mea-
surement, due to the co-elution of 1,1-DCA. Thus, a repeat of the second year's experiments with a more
accurate detector permitted confirmation of those results.
Along with the initial biotransformation experiment, several key transient experiments were per-
formed to determine whether methane was inhibiting the transformation of the chlorinated ethenes. In
the previous year's tests, increased methane addition failed to enhance transformation. These results,
and our laboratory results, indicated that high methane concentration inhibits transformation rates of the
chlorinated ethenes as secondary substrates.
The third year's transient experiments were designed to determine if transformation was enhanced
by the absence of methane. To accomplish this, formate and methanol, which are intermediates in
methane utilization (Anthony, 1982), were substituted for methane once the test zone was biostimulated.
Anthony (1982) in reviewing the oxidation of hydrocarbons by whole cells of methanotrophs discussed
the requirement of a reductant (usually NADH) for the initial oxidation of the hydrocarbon. He indicated
that NADH in methanotrophs is generated by further metabolism of the hydroxylation product, or by
metabolism of a second oxidizable substrate (e.g., methanol or formaldehyde), or by oxidation of internal
storage polymers. Stirling and Dalton (1979) also observed more rapid oxidation of several hydrocarbons
in the presence of formaldehyde. The slower oxidation of these compounds in the absence of
formaldehyde was believed to result from poor generation of NADH. Since formaldehyde is a hazardous
substance, formate was chosen for use in our experiments. It was assumed that these electron donors
(formate, methanol) would not inhibit transformation and would keep the MMO enzyme system active.
Thus, if methane did inhibit transformation, then enhanced transformation should be observed when
these electron donors were introduced.
82
-------
The sequence of field experiments was similar to that used in the second season of field testing.
The chlorinated organics were first added to the system to achieve steady-state breakthroughs before
biostimulation. Methane and DO were then added to biostimulate the test zone while continuing to add
the organics. After steady-state biotransformation conditions were attained, the transient experiments,
including the addition of formate and methanol, were performed.
The chlorinated organics were added in experiments Tracer11 and Tracer! 2, as discussed in Sec-
tion 5. Table 6.4 presents quasi-steady-state fractional breakthroughs of bromide and the chlorinated
aliphatics at the end of the Tracer12 test. The chlorinated solutes reached 90% to 95% of their respective
injection concentrations at the observation wells, after adjusting for dilution by indigenous groundwater.
The results indicate minimal transformation of the chlorinated organics during this period when neither
methane or oxygen was being added. As was the case in the second year's organic addition experiment,
TCE, the most strongly sorbed compound, had the lowest fractional breakthrough before the start of the
biotransformation experiment. The fractional breakthroughs were lowest at the extraction well, indicating
that steady-state conditions had not been achieved in all areas of the test zone before the biostimulation
experiment was initiated.
TABLE 6.4. TRACER12 EXPERIMENT PERCENTAGE BREAKTHROUGH OF CHLORINATED
SOLUTES AND BROMIDE AT THE OBSERVATION WELLS
Well Bromide VCa t-DCE c-DCE TCE
S1
S2
S3
Ext
100l0.6b
9910.8
10010.5
1510.5
9511.8
9212.0
10011.8
9212
10011.2
9611.6
9711.6
9713
10011.2
9911.5
9811.3
9112
9510.9
8811.3
8611.2
7012
a Chlorinated organics adjusted for dilution by native groundwater based on the bromide fractional
breakthrough present in column 1.
b Standard error of the mean.
Results of the Biostimulation-Biotransformation Experiment (Biostim3)
The operating initial conditions of the BiostimS experiment were similar to that of the second season
of testing, as indicated in Table 6.1. Methane and DO addition was initiated using a 3-hr pulse cycle,
slightly longer than the previous year. Methane was utilized immediately upon addition. After methane-
saturated water had been injected for one day, concentrations remained below 0.5 mg/l at the S1 well and
below detection at the farther observation wells, S2 and S3. The limited breakthrough indicated that a
significant numbers of methanotrophs from the prior field season had survived and were immediately able
to consume essentially all the methane within the first meter of transport. After injection of methane and
DO for one day, the pulse cycle length was increased to the standard 12-hr period to distribute methane in
the test zone.
Figure 6.14 shows the corresponding decrease in concentrations of vinyl chloride, trans-DCE and
cis-DCE at the S2 well, in response to biostimulation. Biotransformation started immediately upon
methane addition, demonstrating that active methane utilization was required to initiate the transformation
of the chlorinated organics. As in previous tests, the bromide tracer breakthrough reached unity,
demonstrating that decreases in organic solute concentrations result from biotransformation. Vinyl
83
-------
1.2
VCH—h t-DCEQ-e c-DCEx-x Bromide A-A
40
80 120
Time (Hours)
160
200
Figure 6.14. Decreases in normalized concentration of vinyl chloride, trans-DCE, and cis-DCE at the S2
well in response to biostimulation in the third season.
chloride was the most rapidly transformed, followed by trans-DCE, and cis-DCE. TCE (not shown) was the
most slowly transformed. The concentrations of both vinyl chloride and trans-DCE were observed to
decrease most rapidly within the first 50 to 80 hrs, followed by a more gradual decline. Cis-DCE decreased
more gradually with time. The rank order of rates of biotransformation, trans-DCE > cis-DCE > TCE, is
consistent with the second season's results. The decrease in trans-DCE concentration was more rapid
than observed in the second-season tests (Figure 6.8), which is also consistent with the more rapid
methane uptake. This result indicated that more methane-utilizing bacteria were initially present in the test
zone, resulting in faster removal rates.
System Response to Methane Pulsing
Direct evidence of methane inhibition was obtained when the pulse cycle time was increased from 1
to 12 hrs. The transient response of methane, trans-DCE and vinyl chloride at the S1 well is shown in
Figure 6.15. During the first day, using short pulses of methane, an 80% reduction in vinyl chloride and a
60% reduction in trans-DCE occurred with minimal methane breakthrough. Upon initiation of the longer
methane pulses, distinct cyclic oscillations in vinyl chloride and trans-DCE concentrations were observed,
which correlated with methane pulses. The degree of attenuation of the pulses in chlorinated organics
were related also to methane pulse heights. At the S2 well, shown in Figure 6.14, the amplitudes of the
vinyl chloride and trans-DCE concentration cycles were damped, as methane concentrations remained
near the detection limit. The response at the S3 well, presented in Figure 6.16, shows definite
oscillations of all three components, which are reduced in amplitude compared to those observed at the
S1 well. Based on the response at the three observation wells, the degree of attenuation in aqueous-
phase concentration of the chlorinated organics is shown to be directly related to the attenuation in the
methane pulse heights. Oscillations in concentrations observed in the region near the injection well were
strongly attenuated during transport, by the processes of dispersion, sorption, and transformation.
84
-------
1.2
t-DCEQ-e Methane
0.0
40
80 120
Time (Hours)
160
200
Figure 6.15. Transient responses of vinyl chloride and trans-DCE at the S1 well due to methane pulsing.
1.2
t-DCEo-o Methane
40
60 120
Time (Hours)
160
200
Figure 6.16. Transient responses of vinyl chloride and trans-DCE at the S3 well due to methane pulsing.
85
-------
Methane inhibition is the most likely explanation for this behavior. Higher methane concentrations
reduce the rates of transformation of the chlorinated aliphatic compounds, and hence the aqueous-phase
concentrations of the chlorinated organics increase during periods of high methane concentration as the
reaction rates decrease. The reverse would be true for periods of low methane concentration. The rapid
changes indicate that the methane-utilizers respond very quickly to changes in methane concentration.
Model simulations discussed in Section 13 will discuss the interactions of inhibition and rate-limited sorp-
tion on this observed behavior.
Figure 6.17 shows the gradual decrease in concentration of cis-DCE and trans-DCE at the S2 well
that occurred over the first 400 hrs of the experiment. Vinyl chloride injection was stopped at 200 hrs.
The gradual decrease results from increase in the microbial mass with time, as well as from the slow
desorption of these compounds from the aquifer solids. Model simulations presented in Section 13 will
discuss these processes in greater detail.
Transient Experiments With Formate and Methanol
In order to study the competitive inhibition phenomenon in greater detail, a series of transient
experiments were performed in which different electron donors were substituted for methane. The series
of experiments included the following measures: 1) the substitution of formate for methane; 2) the
reintroduction of methane; 3) the substitution of methanol for methane; and 4) the termination of injection
of methanol, i.e., no electron donor injection.
In the first phase of the transient experiments, formate addition was initiated after quasi-steady-state
biotransformation of trans-DCE was achieved after 350 hrs of methane addition. Formate (NaCOOH) was
injected at a concentration of 218 mg/l, using a pulse duration of 4 hrs, the same as that used for methane
addition. This resulted in a time-weighted concentration of 73 mg/l. These conditions were chosen in
order to maintain the same DO consumption in the test zone, assuming complete oxidation of formate.
During the test, formate concentrations were not measured, but DO consumption was used as an
indicator of formate utilization. Upon adding formate, a slight increase in DO concentration was observed,
confirming that the stoichiometric model was appropriate and that complete formate utilization was attained
in the test zone.
Figure 6.18 shows methane and trans-DCE concentrations at the S1 well for the periods of
methane and formate addition. During the period of methane addition, oscillations in concentration are
superimposed on a gradual trend of decrease in concentrations, to quasi-steady-state values. A clear
response of trans-DCE to formate addition was observed. Trans-DCE concentrations decreased rapidly,
while the oscillations in concentration essentially disappeared. Both responses indicated that inhibition
had ceased. During the early stages of the formate addition, concentrations decreased to levels slightly
lower than the minimum achieved when no methane was present with pulsed methane additions. These
results indicate that, during early stages of formate addition, trans-DCE transformation proceeded at an
enhanced rate that was not inhibited due to the presence of methane. The results also indicate that,
during pulsed methane addition, the microbes responded quickly to changes in methane concentration,
attaining near-maximum rates of secondary substrate transformation as the methane concentration
temporarily receded below the detection limit.
As the formate addition experiment progressed, trans-DCE concentrations at the S1 well gradually
began to increase. Apparently, the ability to keep the system stimulated with formate was gradually lost.
This decrease may result from the following possible causes: 1) the loss in microbial mass, with microbial
decay proceeding at a faster rate than microbial growth, 2) the inability of formate to keep the MMO
enzyme system activated, or 3) competition of other heterotrophs for formate.
86
-------
1.2
VCH-+ t-DCE&-e c-DCEx-x
80
160 240
Time (Hours)
320
400
Figure 6.17. Decreases in cis-DCE and trans-DCE concentration during the first 380 hrs of biostimulation
in the third season.
1.2
t-DCEo-s Methane
0.0
100
200 300
Time (Hours)
400
500
Figure 6.18. Methane and trans-DCE concentrations at the S1 well for periods of methane (0-350 hrs)
and formate (350-500 hrs) addition.
87
-------
The subsequent three transient experiments included the following phases: 1) reintroduction of
methane; 2) methanol addition; and 3) no electron acceptor addition. The responses of trans-DCE and
cis-DCE at the S1 observation well for all four phases of transient experiments are shown in Figure 6.19.
During the preceding period of methane addition, cis-DCE concentration variations were not strongly
correlated with the pulses in methane concentration. With the addition of formate, the cis-DCE response
was similar to that of trans-DCE; the concentrations decreased for a short period of time, and then
increased. Variations in cis-DCE concentrations, however, did not decrease with formate addition; thus,
methane inhibition was not clearly demonstrated. However, the decrease in cis-DCE concentration for the
initial period of formate addition suggests that cis-DCE transformation rates may be inhibited by the
methane. Like trans-DCE, concentrations of cis-DCE increased with prolonged formate addition, indi-
cating a decrease in transformation rate with time.
Upon reintroducing methane (520-660 hrs), decreases in both trans-DCE and cis-DCE concentra-
tions resulted, demonstrating restimulation of the test zone. Oscillations in the trans-DCE concentration
appeared, indicating a resumption of competitive inhibition by methane. The restimulation with methane
indicates that methanotrophs were not growing on formate, and formate may not have kept the MMO
enzyme activated.
Methanol was cut off and methanol addition was started at 660 hrs (Figure 6.19). Like formate,
methanol was injected at a concentration of 51 mg/l, to maintain a DO consumption equivalent to the
complete oxidation of 20 mg/l of methane. Following the switch to methanol, trans-DCE concentrations
decreased during formate addition and the oscillations in concentration were greatly reduced. Cis-DCE
showed an initial reduction in concentration, similar to that achieved with formate. The methanol experi-
ment was conducted for only 100 hrs, not long enough to determine if trans- or cis-DCE concentrations
would increase with prolonged addition. The limited data suggest that cis-DCE concentrations were
increasing at the end of the methanol addition.
t-DCEo-Q c-DCEx-x Methane
£
C
o
420
520 620
Time (Hours)
720
620
Figure 6.19. Response of trans-DCE and cis-DCE to injection of: 1) methane, 2) formate, 3) methane,
4) methanol, and 5) no electron donor.
88
-------
The final transient test performed was termination of electron donor addition. Methanol addition was
terminated at 774 hrs, while the addition of DO and the chlorinated organics continued. As illustrated in
Figure 6.19, an abrupt increase in trans-DCE concentration was observed, indicating that transformation
ceased soon after methanol addition was stopped. The increase is shown to be much more rapid than the
increase observed during formate addition. The results are consistent with those observed in the second
field season (Figures 6.11 and 6.12), and clearly demonstrate that the biotransformation of the chlorinated
organics in the test zone required the utilization of an appropriate electron acceptor for MMO enzyme
activity. Along with methane, both formate and methanol appear suitable for this purpose.
Degree of Transformation
The degree of biotransformation was determined based on 1) comparisons with bromide as a
conservative tracer, and 2) comparisons with the fractional breakthrough of the chlorinated organics
before the test zone was biostimulated (Table 6.4). The degrees of transformation estimated by both
methods are presented in Table 6.5 for the various compounds.
Vinyl chloride was biotransformed more than 95% within 2 m of travel in the test zone. The lower
degree of biotransformation of vinyl chloride, based on the extraction well estimates, probably results from
not having achieved steady-state concentrations at that location in the 200 hr evaluation period.
Transformation greater than 95% might have been achieved if vinyl chloride addition had been continued
for a longer period. Vinyl chloride concentrations were reduced to approximately 1 u.g/1, demonstrating
that drinking water standards could be satisfied by this process.
Ninety percent of the trans-DCE was biotransformed during transport through the biostimulated
zone. Greater than a 90% reduction would have been achieved with prolonged treatment of the test
zone. The degree of biotransformation was greater than the 80% estimated in the second season of
testing. The difference probably resulted from the larger error associated with the second year's estimate
due to the presence of a 1,1-DCA peak that co-eluted with trans-DCE in the GC analysis and the lower
sensitivity of the ECD detector.
TABLE 6.5. PERCENTAGE BIOTRANSFORMATION - THIRD FIELD SEASON
Comparison with Bromide Fractional
Breakthroughs3
Comparison with Fractional Breakthrough
Before Biostimulation
Halogenated
Ethene
TCE
Cis-DCE
trans-DCE
vinyl chloride
TCE
cis-DCE
trans-DCE
vinyl chloride
Well S1
(%)
7-13
28-34
83-87
81-90
2-10
25-37
81-89
78-90
Well S2
(%)
15-19
38-44
89-91
94-98
1-9
36-44
86-94
92-98
Well S3
(%)
16-22
41-45
89-91
93-97
6-14
38-46
86-94
91-99
Extract
(%)
24-30
46-52
68-72
85-91
-9-9
33-53
64-76
84-91
a Percentage biotransformation estimates (95% confidence interval shown).
89
-------
As shown in Figure 6.17, the cis-DCE concentration gradually deceased with time. Steady-state
conditions required for cis-DCE transformation estimate were probably not achieved during the 400 hr
time period of these experiments. The 45% reduction in cis-DCE concentration is in the range of values
determined in the second year's tests. If steady-state conditions had been achieved, a higher degree of
transformation might have been achieved.
The TCE transformation estimates based on the bromide comparison are in agreement with values
from earlier field seasons. Lower estimates result from comparisons with the fractional breakthrough of
TCE before the test zone was biostimulated, consistent with the second year's study. As previously
discussed, the bromide estimates are probably more correct, since TCE is the most strongly retarded
compound and steady-state fractional breakthroughs were probably not achieved before the start of the
biostimulation experiment.
Most of the transformation occurs within the first meter of travel in the test zone. Slight additional
decreases are observed in the second meter of travel, as indicated by the greater degrees of transforma-
tion at the S2 well, especially for cis-DCE and TCE, the least degraded compounds.
Summary of Third Season Results
The third season of field testing demonstrated that nearly complete transformation of vinyl chloride
could be achieved in the test zone. Upon introducing methane and oxygen, a 95% reduction in vinyl
chloride concentration was observed within 200 hrs. Over 90% of the trans-DCE was transformed within
350 hrs. The extents of transformation of cis-DCE and TCE, approximately 50% and 20% respectively,
were similar to those achieved in the first and second seasons of testing.
Oscillations in vinyl chloride and trans-DCE concentrations were correlated with pulses in methane,
providing evidence that methane inhibits the transformation of the chlorinated organics. The effect of
methane inhibition also was observed in transient experiments, when formate and methanol were
substituted for methane. Maximum transformation rates resulted when methane concentrations were low,
or when a suitable electron donor is substituted for methane. When no electron donor was added, the
transformation ceased rapidly, consistent with the results of previous experiments.
SUMMARY
The biostimulation and biotransformation experiments conducted at the Moffett field site demon-
strated conclusively the feasibility of the in-situ biostimulation of an indigenous methanotrophic commun-
ity and the cometabolic biotransformation of the target organic compounds. Methane utilization
commenced rapidly, and the resulting methanotrophic community was easily restimulated following long
periods of starvation. Cometabolic transformation of the target compounds ensued immediately after the
onset of methane utilization, and was clearly linked to the presence of an active methanotrophic commun-
ity. The extents of transformation of the target compounds were consistent from year to year. The rate
and extent of transformation was greater for compounds that are less chlorinated: VC > DCE > TCE. The
cometabolic transformations appeared to be inhibited by high methane concentration. Substitution of an
alternative electron donor, e.g. formate or methanol, temporarily enhanced the biotransformation of the
target compounds but could not sustain it in the long term. Transformation of the target compounds
ceased soon after the addition of the electron donor was terminated. The region of cometabolic
transformation extended over the entire spatial domain occupied by methane-oxidizing bacteria.
90
-------
SECTION 7
FORMATION AND FATE OF trans-DICHLOROEPOXIDE
Martin Reinhard, Franziska Haag, and Gary Hopkins
Laboratory studies using methanotrophic consortia have indicated that trans-1,2-dichloroethene
(trans-1,2-DCE) is transformed via epoxidation to trans-1,2-DCE epoxide (also referred to as trans-1,2-
dichlorooxirane) (trans-DCO) (Janssen et al., 1987). High concentrations of trans-DCE were found to
inhibit both growth of methanotrophs and co-metabolic transformation of trans-1,2-DCE. Thus, formation
of trans-DCO may be a complicating factor in bioremediation schemes using methanotrophs. Moreover,
formation of epoxides constitutes a public health concern. Ethylene epoxide, for instance, is a carcino-
gen and a reproductive toxicant. The formation and fate of such compounds must be understood before
biostimulation of methanotrophs can be confidently applied for aquifer restoration.
In this section we report a procedure for the chemical preparation of trans-DCO at the milligram scale.
A trans-DCO standard was needed for compound verification, calibration and for laboratory transformation
studies. We also report data on the hydrolytic transformation rate of trans-DCO at different pH values and
discuss the observed formation of trans-DCO in the biostimulated aquifer.
CHEMICAL SYNTHESIS OF trans-DCO
We synthesized trans-DCO by adapting the procedure used by Liebler and Guengerich (1983) for
the epoxidation of vinylidene chloride. An aliquot of 2 ml trans-1,2-DCE was added to 7 g m-chloroper-
benzoic acid (Aldrich) in 50 g p-dichlorobenzene (Aldrich). The mixture was reacted at 55°C for 24 h. The
reaction mixture was distilled under vacuum and the distillate (approximately 1 to 1.5 ml) was collected in a
condenser cooled with dry ice.
GC/MS analysis of the distillate indicated the presence of trans-1,2-DCO and impurities such as
trans-DCE, cis-DCE, cis-DCO, dichloroacetaldehyde and p-dichlorobenzene. Both cis-DCE and cis-DCO
are suspected rearrangement products formed during the epoxidation reaction (Griesbaum et al., 1975).
The structure of trans-DCO was verified using 1H-NMR (Varian EM360, 60 MHZ) and GC/MS (Model 5070,
Hewlett-Packard, Palo Alto, CA). Both NMR and MS spectra agreed with those reported by Griesbaum et
al. (1975).
The product trans-DCO was separated from the impurities on a GC (Packard Model 437A) equipped
with a 3 ft x 1/4 in packed column (3 % SP-1500, 80/120 Carbopack B, mR54966, Supelco). The outlet of
the column was connected to a programmable Valco gas switching valve (Valco, Houston, TX). The
fraction containing trans-DCO was trapped at the column outlet using a cold finger condenser submerged
in liquid N£. Five 100-u.l portions of the distillate were injected. After warming the trapped trans-DCO to
room temperature, the concentrated trans-DCO was weighed (5.4 ±1.1 mg) and dissolved into 1 ml
pentane. The purity of the solution was checked using GC/MS.
91
-------
Initial experiments indicated that dichloroacetaldehyde coeluted with trans-1,2-DCO on the
Carbopack column. To circumvent this problem, a procedure was developed to chemically remove
dichloroacetaldehyde from the mixture. This was accomplished using p-dichlorophenylhydrazine to form
the hydrazone derivative (March, 1985). However, in samples stored for periods of six to nine months,
dichloroacetaldehyde decomposed and p-dichlorophenylhydrazine treatment was unnecessary. The
procedure was tested as follows: To 10-u.l distillate in 1 ml pentane were added 0.5 g dichlorophenyl-
hydrazine. The reaction mixture was reacted for one hour at ambient temperature. After one hour, alde-
hyde was removed. The epoxide was relatively nonreactive toward this agent. After 3-1/2 hrs in the pres-
ence of excess dichlorophenylhydrazine, only a small fraction reacted.
Analytical Procedures
Using the purified standard and an electron capture detector (ECD), the response factor of trans-
DCO was determined to be 150 times greater than that of trans-DCE, permitting us to detect very low
concentrations. This ratio was used to calculate the trans-DCO concentrations in the field. In the
automated field system, trans-DCO was separated from all other known peaks and was apparently
detected as a single peak. However, the field system was not calibrated using the trans-DCO standard.
The reported field data thus assumes the same recovery for trans-DCE and trans-DCO, and similar
response factors for the field and laboratory ECD detection. The relative response factors were
determined on the GC/MS system also using total ion current (TIC) detection. The TIC response for the
trans-DCO was stronger than cis-DCE by a factor of 1.7. Estimates using GC/MS data agreed with the field
data.
Laboratory Hydrolysis Study
Hydrolysis of trans-DCO was measured at 25°C at pH 5.0,7.9 and 10.6. Trans-DCO in pentane was
added to 40 ml of 0.03 M phosphate buffer in a 60 ml volumetric flask to give an initial concentration in the
range of 310 to 315 mg/L After vigorously shaking the flask, a 1-ml aliquot was added to glass ampules
containing 12 ml buffer solution. The ampules were then flame-sealed with less than 1 ml of headspace
(Burlinson et al., 1982), placed into a 25°C water bath, and analyzed at regular intervals over a period of
240 hrs.
RESULTS
Confirmation of trans-DCO in Field Samples
The formation of a biotransformation intermediate was observed during the field biotransformation
experiment Biostim2. Initially, only the peak area of the unknown was recorded, but the peak could not be
identified on the basis of retention time alone.
To confirm the suspected presence of trans-DCO in field samples, 100-ml samples from the
stimulated zone were analyzed using closed-loop-stripping/desorption analysis (Graydon et al., 1983). A
GC/MS equipped with a 60 m thick-film capillary column (DBS, 0.32 mm i.d., 1-u.m film thickness; J & W
Scientific, Rancho Cordova, CA) was used. Both trans-DCO and 1,1,1-TCA (which was present in the
formation water) coeluted on this column. The trans-DCO was present at much lower concentration than
TCA, but positive identification of trans-DCO was possible using chemically prepared trans-DCO standard.
Hydrolysis Kinetics
The rate data of trans-DCO transformation obtained at 25°C is shown in Figure 7.1. The observed
first order rate constants and the 95% Cl for pH 5, 7.9,10.6 are 0.0135 ±0.0003 Ir1,0.0145 ±0.0002 Ir1,
92
-------
-7
240
O pH 10.C
Figure 7.1. Pseudo-first-order plots for trans-1,2-DCE epoxide hydrolysis at pH values of 5.0, 7.9, and
10.6.
and 0.0291 ±0.0006 h'1, and the corresponding half-lives are 51.3 hr, 47.8 hr, and 23.8 hr, respectively.
The rate at pH 10.6 was approximately twice that determined at pH 5.0 and at 7.9, indicating that hydrolysis
at pH 5.0 and 7.9 is due to reaction with water. The average of the pseudo-first order rate constants for
the neutral (water) reaction is 0.0140 rr1, with a half-life of 49.5 hr at 25°C.
The increase in transformation rate at pH 10.6 by a factor of 2.1 indicates some reaction with the
hydroxide ion. By invoking the following relationship,
-] = KW/10"PH
(7-1)
we obtain a value of 37.9 mol/l-sec for the second-order reaction constant of the base (OH') promoted
process. This rate is too small to be significant in the environmentally relevant pH range 6 to 9; and hence,
the pH is not a significant variable under field conditions. The pH profile of DCO hydrolysis derived from
our data is shown in Figure 7.2.
Janssen et al. (1987) have reported a half-life of 30 h at 30°C, corresponding to a reaction rate of
0.023 h'1. From these data and our rate constant of 0.014 h'1 at 25°C, an activation energy (Ea) of 17.8
kcal/mol may be estimated. The typical ground water temperature at the Moffett field is 18°C. Using the
estimated Ea, we estimate a pseudo-first-order rate constant and a half-life of 0.0068 rr1, and 114 hrs
(4.3 days), respectively, at 18°C.
93
-------
-U.O -
-0.9 -
-1 -
-1.1 -
-1.2-
-1 J -
-1.4-
J-1.5 -
-1.6-
f -1.7 -
-13 -
-1.9 -
-2 -
-2.1 -
-13. •
-2J -
T
i
i
i
i
t
i
i
a >
1
1
1
1
/
1
A
i 1 i 1 1 i
11
Vote.
MUM
Figure 7.2. k0bs as a function of pH. Open symbols are measured data, full symbols are calculated
values. The lines indicate the neutral (full line) and the base-promoted component (dashed
line) of the measured overall pseudo-first-order hydrolysis constant, k'0bs-
trans-DCQ Formation During Biostimulated Transformation of trans-DCE
Formation of trans-DCO was observed during Biostim2 when trans-DCE was one of the secondary
substrates. Figure 7.3 shows the concentration of trans-DCO at S1, S2 observation wells and at the
extraction well during the 600 h experiment. Formation of trans-DCO was believed to occur in the
immediate vicinity of the injection well, where growth of methanotrophs and consumption of methane was
occurring. After an initial lag phase of approximately 20 to 30 hrs, the trans-DCO concentration began to
increase rapidly at S1 and S2 to a range of three to five u.g/1. This rise was observed at the extraction well
also, although the trans-DCE concentration remained lower due to dilution by indigenous groundwater.
The absence of a measurable delay in the appearance of trans-DCO indicates that this compound was not
sorbed significantly. The concentrations measured at S1 and S2 for the most part followed the same time
profile.
In the extraction well, the expected attenuation factor due to dilution was 7.35. This estimate is
based on an injection rate of 1.36 l/min, an extraction rate of 10 l/min, and the assumption of complete
recovery of the injected water. The average trans-DCO concentration and the 95 % confidence interval at
S1 were 4.99 (± 0.38) \sg/\. At S2 and at the extraction well, the trans-DCO concentrations (and 95% Cl)
were 4.54 (± 0.31) and 0.56 u,g/l (± 0.03), respectively. Thus, the average trans-DCO concentrations at
S2 and the extraction well were lower by 9.0 and 89 %, respectively. The expected concentration at the
extraction well based on the dilution ratio was 0.68 u,g/l. Comparison with the observed value of 0.56 u.g/l
indicates that 18 % may have transformed during transport between S1 and the extraction well.
Based on an average residence time of approximately 10 hrs between S1 and S2 and 24 hrs
between S1 and the extraction well, and using the measured hydrolysis rates, we would expect hydrolysis
of 7% and 15 % of the trans-DCO, respectively. Hence, predicted and observed transformation rates are
consistent with the observed concentration decreases in the field, although the relatively close agree-
ment may be fortuitous. In any case, there does not seem to be a transformation pathway other than
hydrolysis for trans-DCO outside the biostimulated zone.
94
-------
0.0
120
240 360
Time (Hours)
480
600
Figure 7.3. Concentrations of trans-DCO at observation wells S1, S2, S3, and the extraction well in the
Biostim2 experiment. The data range of 300-600 hours was used for estimating hydrolysis
rates.
Within the biostimulated zone, where the concentration of microorganisms is greater, other and
more effective transformation pathways seem to exist. The data shown in Figure 9.12 indicate rapid
decrease (within approximately 8 hrs) after the methane injection was terminated. Part of this decrease
results from advective transport from the test zone. Another component of this decrease may be reaction
with nucleophiles present in the biomass of the test zone. Trans-DCO, which is an electrophile, may react
with nucleophiles, such as sulfur and nitrogen compounds, and covalently bind to enzymes and other cell
material. This questions merits further investigation.
Summary and Conclusions
Formation of trans-DCO, which was previously detected in laboratory experiments, was confirmed in
field biostimulation experiments using GC/MS and a standard compound. On a mass basis, approximately
5 to 10% of the trans-DCE was converted into trans-DCO. Laboratory studies have indicated a reaction
rate and half-life of 0.014 Ir1 and 49.5 hrs, respectively, at 25°C, and the reaction rate was found to be
independent of pH in the pH range 5 to 10. Using literature data, the Ea was estimated as 17.8 kcal and
the half-life extrapolated to 18°C was estimated as 4.2 d.
Although the hydrolysis rate of trans-DCO is relatively fast, formation of this potentially harmful by-
product may have to be considered in remediation schemes using methane, particularly in cases where
the treated water is recycled rapidly into the water supply. Assuming a t-)/2 of six days (15°C), a decrease
from 100 u.g/1 to 2 u.g/1 (the USEPA standard for vinyl chloride in drinking water) will take approximately
1 month.
95
-------
SECTION 8
SORPTION
Thomas Harmon and Paul Roberts
INTRODUCTION
The ultimate goal in solute transport modeling in groundwater is the independent estimation of the
relevant parameters. The Moffett field site has provided an outstanding opportunity for advancement
towards this goal. With the monitoring capabilities in the field, laboratory techniques for parameter
estimations can be rigorously evaluated, and employed subsequently in cases where the field charac-
teristics are less fully known.
The focus of the Moffett solids sorption studies was to increase the understanding of the
interrelationship between a complex natural aquifer environment and a greatly simplified laboratory
experiment. More specifically, experiments have included the estimation of laboratory-scale equilibrium
parameters (partitioning coefficients) for comparison with observed field parameters. Field observations
(Section 5) have indicated that the time scale of sorption processes is large enough, relative to that of
groundwater flow, to cause significant deviations from local equilibrium. Thus, laboratory scale studies
have included an investigation of nonequilibrium modeling parameters (apparent diffusivities for transport
within solid grains, and equivalent first-order rate constants).
MOFFETT SOLIDS
Solid samples were obtained from cores taken near the test zone. Figure 8.1 provides a plan view
of the field site, and the exact location of the cores used for the sorption studies. One core (B2) held
together sufficiently to yield a sample from what appeared to be loose material composing the aquifer of
the test zone. The remaining cores provided material from above or below the high conductivity zone, or a
mixture of solids from different layers (drilling slough).
A description of the solids characterization and sorption studies is facilitated by a clarification of
nomenclature. The few samples obtained from the aquifer zone in core B2 were used in one set of
sorption experiments involving 'bulk'" solids. Because no portion of the majority of the cores appeared to
be representative of the aquifer zone, the remaining solids were sieved into fractions. Table 8.1 provides
a list of the U.S. standard sieve sizes used, and the mass fraction of each particle size from the aquifer
zone sample of cores SU-39-4 and 6 (see Figure 8.1). The particle size distribution from which these
mass fractions were derived is shown in Figure 8.2, for the aquifer zone sample of SU-39-6. The particle
size distribution for SU-39-4 is similar. The fractions combined according to this particle size distribution
were used in a second set of sorption experiments involving 'synthesized bulk' solids.
Solids Preparation
Each of the cores was extruded carefully to separate the interior core solids from those near the
walls, where oxidation of the steel core barrels appeared to have altered the solids. The high fraction of
silts and clays in the nonrepresentative samples necessitated the wet sieving of slurries prepared in a ratio
96
-------
N
(SU-39B2)D
Control
x Shed
a
(SU-39-B1)
X Synthesized bulk solids
Y Bulk solids
P Solids for panicle size distribution
QS3
X
Q
(SU-39-B3)
Scale, meters
Rgure 8.1. Location of cores used in laboratory sorption studies.
Grain dliimttr, mm
Figure 8.2. Particle size distribution corresponding to the synthesized bulk samples; observed in
aquifer solids from core SU-39-6.
97
-------
TABLE 8.1. MASS FRACTION OF PARTICLE SIZES USED TO PREPARE
SYNTHESIZED BULK SAMPLES
(Determined Using Distribution in Figure 8.2)
U.S. Standard
Sieve No.
+4-10
10-20
20-40
40-60
60-80
80-120
120-200
<200
Particle Diameter
(mm)
4.75-2.00
2.00-0.85
0.85-0.425
0.425-0.25
0.25-0.18
0.18-0.125
0.125-0.075
< 0.075
Mass Fraction
(-)
0.45
0.24
0.10
0.057
0.025
0.032
0.026
0.07
of about two liters of Moffett groundwater to one kilogram of solids. To avoid dissolution of calcium
carbonate (CaCOs), groundwater was equilibrated with reagent grade CaCOs prior to slurry preparation.
Solids retained on the No. 4 mesh sieve (> 4.75 mm) were discarded. Solids retained on each subse-
quent sieve were then re-sieved in amounts no greater than about 100 g. Portions of the fractions
designated for surface area and pore analyses were rinsed with a solution of calgon (sodium hexameta-
phosphate) to disperse persistent clays that might affect the results.
Bulk and fractionated solids were oven-dried at 50-60°C. To avoid biasing samples towards
particular particle sizes, solids were riffle-split down to sample size, as described by Ball et al. (1989).
Particle Characterization
Methods of particle surface area and pore size characterization have been refined recently in our
laboratory for the analysis of sandy aquifer material of relatively low specific surface area (< 3 m2/g) (Ball et
al., 1989). Surface area methods used include low-temperature nitrogen adsorption, low-temperature
krypton adsorption, and ethylene glycol monoethyl ether (EGME) adsorption. Particle porosity was
characterized using low-temperature nitrogen desorption and mercury porosimetry.
Low-temperature adsorption was interpreted by the BET approach, as developed by Brunauer et al.
(1938). The krypton method was used to surmount the possible limitations of the nitrogen-BET method
for low-surface solids and has been applied successfully to well-characterized solids such as glass and
quartz (Beebe et al., 1945; Gaines and Cannon, 1960; Sing and Swallow, 1960). However, the krypton-
BET method shares a known limitation of the nitrogen-BET method: interlayer surfaces of expanding clay
minerals are not measured. The EGME method of Carter et al. (1965) utilizes a polar adsorbate to measure
both the external and internal surfaces of soils and clays.
Mercury porosimetry measures a wide range of pore sizes, but is limited at the lower end by the high
pressures required. Nitrogen desorption is capable of measuring smaller pore sizes, but is limited at the
upper end by very large errors in measurement at high partial pressures of N2. The range of pore radii
accessible by a combination of the two methods is approximately 1.5 to 10,000 nm.
Specific Surface Area-
Figure 8.3 contains BET plots for the results of krypton and nitrogen surface area determinations of
a silica standard (No. 2008, 5.29 ± 0.08 m2/g, Quantachrome Corporation) by Ball et al. (1989). For
98
-------
0.3
~ 0.2
e
" 0.1
0.0
Nitrogen Silica Std
Krypton Silica Std
Nitrogen Moffett +20-40
0.0
0.1
0.2
0.3
P/Po
Figure 8.3. BET plots for nitrogen and krypton adsorption by silica standard, and nitrogen adsorption by
Moffett 20-40 fraction.
triplicate analyses, the figure shows the nitrogen approach is accurate and precise: results of a linear
regression based on the data yield a reasonably precise surface area estimate of 5.20 ± 0.23 m2/g.
Triplicate krypton analyses were also accurate, and yielded a more precise surface area estimate of 5.26 ±
0.11 m2/g.
The BET plot in Figure 8.3 for the +20-40 mesh Moffett solids shows that the nitrogen method is
less accurate for these natural materials, but still provides reasonable accuracy. The specific surface area
estimate for each of the Moffett size fractions is provided in Table 8.2. The results show relatively high
surface areas~3.6 to 7.0 m2/g~ even though the internal surface areas of expanding clays were excluded.
The krypton method, accurate for total sample surface areas ranging from about 0.1 to 5.0 m2, failed for
what was considered a representative sample mass of the Moffett solids (> 1 g).
The EGME method applied to the same silica standard yielded a specific surface area estimate of
3.92 m2/g. Only one sample was analyzed because of the substantial amount of standard required
(approximately 5 g) by the method. Results from analyses of the Moffett fractions are shown in Table 8.2.
The EGME surface areas range from 18 to 45 rr^/g, and in all cases are at least three times higher than the
N2 BET value for the respective size fraction, indicating the presence of a substantial fraction of
expanding clay minerals. Surprisingly, the EGME surface area is nearly as large for the coarsest fraction as
for the finest fraction, and twice as large as for the intermediate fractions. This suggests that the large
particles consist in part of agglomerates of clay-sized particles.
The mass fraction of clay-sized particles which may exist in the Moffett fractions was estimated using
the BET and EGME surface areas. The estimates indicate that if the surface area of the clays is on the
order of 800 m2/g, then this mass fraction is in the range of 2 to 4%. Surface areas of this magnitude
99
-------
TABLE 8.2. SPECIFIC SURFACE AREA AND INTERNAL POROSITY FOR
MOFFETT SIZE FRACTIONS
Specific Surface Area (rr^/g)
Intra-Particle Porosity
Fraction N2 BET EGME
4-10
10-20
20-40
40-60
60-80
80-120
120-200
<200
5.9
6.0, 4.1
6.8, 5.6
3.6
5.0
4.8
7.1
7.1
38
28, 36, 36
22
19
18
26
21
45
NO
0.029
0.022
0.015
0.014 (60-100)
0.017, 0.015 (100-200)
have been measured in smectites, which are common at the site. The possibility of contamination of the
fractions with clays from the confining layers seems unlikely. The mass fraction of even the upper clay
(157 m2/g) would have to be at least 30% to explain the EGME surface areas for some of the fractions.
Intra-Particle Porosity-
The nitrogen adsorption/desorption isotherm for the +20-40 mesh fraction is shown in Figure 8.4.
This type of isotherm, with its characteristic hysteresis loop, has been well-documented in the physical
adsorption literature (e.g., Gregg and Sing, 1982). Its shape is attributed to differences in capillary
condensation in constricted pores along the adsorption and desorption branches of the isotherm. Using
the Kelvin equation to relate adsorbate gas pressure to meniscus radius, and assuming a pore geometry,
the pore volume and the pore size distribution can be determined from the isotherm. The simplest and
most commonly applied model of pore geometry is that of cylindrical, open-ended, non-intersecting
pores. For the pressures associated with the nitrogen method, the pore radii sampled correspond to the
lower third of the mesopore range (2-15 nm) and the upper portion of the micropore range (1.5-2 nm), as
categorized by Gregg and Sing (1982).
Figure 8.5 combines the pore size distributions derived from mercury intrusion and nitrogen
desorption. The two methods have been discussed in detail elsewhere (Gregg and Sing, 1982; Ball et al.,
1989). Here and in general, the mercury data describes a much larger range of pore sizes than does the
nitrogen data. The distribution derived from nitrogen does, however, include pores smaller than 3 nm in
radius (the lower limit of the mercury method) which are approaching the molecular size of the solutes of
this study.
The intra-particle porosity of the Moffett solids was relatively higher for the larger size fractions
than for the smaller ones. The results of mercury intrusion studies, included in Table 8.2, show the
internal porosity ranges from 0.029 and 0.022 cm3/g for the +10-20 and +20-40 fractions, respectively, to
an average of about 0.015 cm3/g for all of the smaller fractions. Particles larger than 2 mm (No. 10 mesh)
were too large for the analysis.
Discussion of Solids Characterization
Discussion of the characteristics of the Moffett solids is facilitated by comparison to a more homo-
geneous aquifer solid. The Borden aquifer solids, which have been studied in great detail (Curtis, 1984;
Ball and Roberts, 1985,1987; Ball et al., 1989), consist of a relatively clean, sandy material, with a size
100
-------
M
"5
T3
Desorptlon
Adsorption
0.0 0.2 0.4 0.6 0.8 1.0
P/Po, [-]
Figure 8.4. Nitrogen adsorption/desorption isotherm for Moffett 20-40 fraction.
0.03
u
u
~ 0,02 H
«r
e
u
0.01 -
0.00
" Nitrogen
Mercury
10 100 1000 10000
Pore Radius, [nm]
Figure 8.5. Combined nitrogen adsorption and mercury intrusion cumulative pore size distribution for
Moffett 20-40 fraction.
101
-------
distribution dominated by medium to coarse sizes (0.42-2 mm), and a bulk value of organic carbon content
of about 0.0002. The Moffett material is much more heterogeneous mineralogically, and composed of a
greater fraction of coarse sands and gravels (> 2 mm) and silts and clays. The organic carbon content of
the Moffett solids (approximately 0.001) indicates a higher organic partitioning potential. The BET surface
area for the Moffett fractions ranges from about two times greater than that of the Borden sand, for the
largest fraction, to about an order of magnitude greater for the fine fractions. However, the two materials
exhibit very similar values of intra-particle porosity, providing two indications: 1) the Moffett solids are
characterized by a much rougher, more weathered surface than the Borden sands (electron microscopy
has confirmed this suspicion), and 2) the Moffett solids have a potential for the diffusion-limited uptake of
solute similar to that observed in sorption experiments with Borden sand. The results from the EGME
surface area analyses suggest that a fraction of expanding clay minerals, probably weathering products,
are present in the Moffett fractions. Because the sampjes for surface analysis were rinsed with a
dispersant, it may be that these clays are present deep within the internal pores of the particles. Electron
microscopy seems to support this hypothesis by indicating the presence of tiny particles along the internal
pore walls of the Moffett particles. The effects of such clays are uncertain, but their presence must be
taken into account in understanding sorption processes.
MOFFETT SOLIDS SORPTION STUDIES
Background
The procedure for the batch sorption studies using radio-labeled (14C) compounds has been used
previously in this laboratory (Curtis, 1984; Ball and Roberts, 1985). The method, outlined schematically in
Figure 8.6, entails adding known quantities of aquifer solids and water to glass ampules, spiking the
resultant samples with a known quantity of radio-labeled contaminant, flame-sealing the ampules, and
incubating the samples for a given period of time. At the end of the incubation period, the samples are
centrifuged, and a portion of the supernatant is collected in liquid scintillation cocktail for counting. Using
control samples (containing no aquifer solids) to account for volatilization losses, the concentration of
sorbed contaminant is taken as the difference between the mass added and the mass in the aqueous
phase. The method was adapted for headspace analyses with non-labeled compounds by equilibrating a
portion of the sample supernatant in headspace vials, and analyzing the vapor phase with the Hall
electroconductivity detector.
The procedure employing radio-labeled compounds has the advantage of accuracy (> 95%
recovery on controls). Such accuracy is necessary for an investigation of sorption rate limitations. The
disadvantages of the method are primarily associated with the expense of the compounds. While the
headspace modification of the method is less accurate, and probably not suitable for detailed investiga-
tions, it offers the advantage of enabling the simultaneous analysis of several compounds.
Solids Sorption Studies
The batch sorption studies were undertaken to characterize the equilibrium behavior of the
contaminants of interest with the Moffett solids. More specifically, the studies were used to investigate
the linearity of the equilibrium behavior, and to quantify the distribution coefficients (Kd) and the rate at
which equilibrium is achieved.
Partitioning of 14C-Labeled TCE--
Early studies were aimed at measuring the apparent equilibrium isotherm of TCE for equilibration
times corresponding to the contact time of the contaminants in the field. Figure 8.7 shows such an
isotherm for TCE equilibrated with Moffett bulk material for 10 days. The data exhibit a slightly nonlinear
behavior, and are best fitted by a Freundlich isotherm model. However, as shown by the regression line in
the plot, a linear approximation agrees reasonably well; the resulting value for Kd using the data obtained
with 10 days equilibration is 2.0 ml/g.
102
-------
WMK
4
|S|
Solid
Extract
Figure 8.6. Flame-sealed ampule method for batch sorption studies (Curtis, 1984; Ball and Roberts,
1985, 1987).
200
L.
O
in
•o
01
t-
o
at
100 -
0 20 40 60 80 100 120
Aqueous Concentration, [ug/I]
Figure 8.7. Sorption isotherm at increasing times for Moffett bulk solids and TCE.
103
-------
Similar studies of ICE sorption by bulk Moffett solids for 1,3, and 5 days resulted in similar behavior,
with apparent K
-------
10
8-
6 •
g
2-
• Bulk
• Synthesized Bulk
] 1
1 10
Time, [days]
100
Figure 8.8. Increase of apparent Kd with time for sorption of ICE on bulk and synthesized bulk solids.
10
8 -
6-
I
<
2 -
.01
.1 1
Particle Diameter, [mm]
10
Figure 8.9. One-day and thirty-day apparent K
-------
0.3
O
-------
Diffusion Rate--
Trie dependency of sorption on particle size has been observed previously (Ball and Roberts,
1985; Wu and Gschwend, 1986; Ball and Roberts, 1987), and has been attributed to mass transfer
(diffusional) limitations. Ball and Roberts (1985,1987) employed a model based on diffusion of solute
through intra-particle pores that is retarded by equilibrium partitioning. For the case of linear sorption
isotherms, Pick's Second Law becomes
St 8r L8J 5r (8-1)
where Cr = the solute concentration at a point within the pore space of a particle, [M/L3]; r = a point along
the radius of the particle, [L]; and Da = the apparent diffusivity of the solute within the particle pores,
[L2/T].
The apparent diffusivity is defined as follows:
(8-2)
where De = the effective pore diffusivity, that is, the bulk diffusivity adjusted for the tortuosity of the pores,
[L2/T]; p = the grain density, [mass solid/volume particle]; and e = the intra-particle porosity, [volume
pore/volume particle].
By assuming that the ultimate uptake of solute has been reached within 30 days, the TCE sorption
data of the synthesized bulk in Figure 8.8 were transformed to provide the plot of fractional uptake with
time in Figure 8.11. The fit shown is for the analytical solution of the problem of diffusion from a well-stirred
solution of limited volume by Crank (1975). The fit provides a rough estimate of the apparent diffusivity,
averaged over a wide range of particle sizes.
For comparison with field scale behavior, the best estimate of the apparent diffusivity (5x1 D'12 m2/s)
was converted to an equivalent first-order rate parameter, or mass transfer coefficient, for one-dimensional
transport. This approximation greatly simplifies the mathematics by enabling the substitution of a first-
order differential equation for equation (8-1).
Wu and Gschwend (1988) studied the effects of particle size distribution on the first-order
approximation of the diffusion model in batch systems. They concluded that the approximation was
appropriate only for a relatively narrow range of particle sizes (within an one order of magnitude). With the
intent of estimating the field scale behavior, we applied the approximation over a relatively wide range of
particles. An empirical method provided by van Genuchten (1985) uses the following conversion:
_ 22.7 Da
a2 (8-3)
where a is an equivalent first-order mass transfer coefficient, [d'1], and a is the particle radius. A variation
used by Goltz (1986) employs the method of moments to derive a slightly different conversion:
a2 (8-4)
Using equations (8-3) and (8-4), the estimates of a range from 0.4 to 0.6 d'1.
107
-------
Fractional Uptake
> o o — -
j 'CD (o b -
1
Q „.*-"**'
x"
x' Kd = 7.4 ml/g
jp" d = 40 urn
1 Da=5(10"12)cm/s
10 100
Time, [days]
Figure 8.11. Sorption rate of TCE by synthesized bulk, showing diffusion model fit.
SUMMARY
Laboratory studies using solids from Moffett cores have focused on quantifying sorption equilibrium
and rate parameters at the particle scale. Results of solids characterization studies indicate that the Moffett
particles provide a complex system for sorption processes. The particles seem to be extremely weathered
aggregations of grains, possibly containing clay-sized particles. The relatively high organic carbon content
of the solids within, and adjacent to, the test zone seems to vary by as much as a factor of four. Applying
the methods developed in our laboratory for a relatively clean, homogeneous sand, we observed a fast
fraction of contaminant uptake by the solids, followed by a fraction of uptake governed by a rate that is
orders of magnitude slower. This indicated that, although the solids are quite complex, the suspected
sorption mechanism-organic partitioning limited by mass transfer-adequately explains the rate behavior.
The transport parameters derived from laboratory scale studies provide a basis for independently
estimating field-scale transport parameters. The results presented in this section will be used in Section
13 to interpret field observations of transport behavior.
108
-------
SECTION 9
TCE TRANSFORMATION BY MIXED AND PURE GROUNDWATER CULTURES
Susan Henry and Dunja Grbic-Gali6
INTRODUCTION
Although TCE transformation by methanotrophic mixed cultures (consortia) and pure cultures has
been well-documented, research on this topic is still in its nascent stages, and it may be premature to
generalize. What appears to be emerging, however, is that different methanotrophs possess different
TCE-transformation capabilities, and respond differently to different conditions.
In order to develop and optimize methanotrophic bioremediation methodologies, and to develop
predictive capabilities, we must achieve a fundamental understanding of the transformation process. This
includes defining the effects of operational parameters on TCE transformation, and defining the
physiological bases for given responses. Since it is evident that not all methanotrophs possess the same
capabilities, the greatest value lies in evaluating many methanotrophic cultures. In that manner, it should
be possible to define the characteristics of a methanotrophic culture that is most suitable for a given
application.
In our research, we have obtained TCE-degrading bacterial cultures from the Moffett Field
groundwater aquifer. These cultures have been evaluated for their TCE transformation capabilities by
radiolabeling experiments and by gas chromatography, and have been examined by scanning and
transmission electron microscopy. One of the objectives of our work is to characterize the behavior of
these cultures under conditions that may be considered operational parameters in either in situ
bioremediation or waste treatment applications. We have evaluated the effects of methane concentration,
availability of reducing power, oxygen concentration, TCE concentration, and mineral medium formulation
on TCE transformation by these cultures; the physiological bases for some TCE transformation
phenomena; and the intermediates and products resulting from the methanotrophic transformation of
TCE.
METHODS
Aquifer Cultures
Enrichment of Cultures-
Mixed cultures were enriched from aquifer material or groundwater collected from the Moffett field
test site, or from effluent from a soil column. The aquifer material used was gravelly, silty sand from the
core SU-39-2 (B series) aseptically collected on August 27,1985, before the test site had been enriched
with methane (Section 4). The groundwater was withdrawn from the test site July 1987, after the site had
been enriched with methane. The soil column had been packed with aquifer material from core SU-39-2,
and had been enriched on methane and transformed TCE (Section 10).
Continuously-stirred reactors containing one liter of a phosphate-buffered nitrate salts mineral
medium formulation, here termed "regular mineral medium" (per liter: 1 g NaNOa, 0.5 g K2HPO4, 0.5 g
109
-------
KH2PO4, 0.2 g MgSO4-7H2O, 0.02 g CaCI2-2H2O, 0.005 g FeSO4-7H2O; trace metals: 70 u.g
ZnSO4-7H2O, 17 u.g MnSO4-H20, 20 u.g HaBOa, 100 ng CoCI2-6H2O, 10 u.g CuCI2-2H2O, 20 ug
NiCl2-6H2O, 30 u,g Na2MoO4-2H2O) were inoculated with approximately one gram aquifer material or
approximately 20 ml groundwater or column effluent, and incubated under a continuous flow of
approximately 25% gaseous substrate (methane, ethylene, or propane) in air. When turbidity was
observed, 10 ml of the culture was transferred to a new reactor and incubated as before. This process was
repeated 5 to 10 times, depending upon the culture, over the period of several months, until culture
stability, as determined by light microscopy and consistent colony morphology, appeared to have been
achieved.
Isolation of Pure Cultures-
Pure cultures were isolated on agarose plates made with regular mineral medium. Plates were
incubated in dessicators filled with approximately 25% substrate (methane or ethylene) in air. Purity was
confirmed on the basis of repeated colony isolation; constant, homogeneous colony morphology; and
constant, homogeneous cell morphology.
The methane-oxidizing isolate was tested for growth on multicarbon substrates in liquid and on solid
media, including blanks (no substrate); blanks + vitamins; glucose; glucose + casamino acids; fructose;
fructose + casamino acids; succinate; succinate + casamino acids; casamino acids; formate; glyoxylate;
and dichloroacetate. Lack of growth on any of these substrates in liquid culture was a confirmation of
purity. Background growth on solid media, regardless of substrate, was not an indication of lack of purity.
The isolate, after seven sequential transfers on blank, glucose, glucose + casamino acids, casamino acids,
and methanol plates, would still grow on methane when transferred back to liquid or solid media, and
maintained consistent colony morphology throughout. It is assumed that the growth on solid media is due
to C-1 contaminants in the solid media.
Enzyme Assays-
The methane-oxidizing pure culture was tested for enzymes characteristic of type I and type II
methanotrophs. The hydroxypyruvate reductase assay (Large and Quayle, 1963) tested for the serine
pathway of carbon assimilation. The hexulose phosphate synthase assay (Mary Lidstrom, pers. comm.)
tested for the ribulose monophosphate (RMP) pathway of carbon assimilation.
Microscopy-
The cultures were evaluated by light microscopy and by scanning and transmission electron
microscopy. The cultures were examined by phase contrast for motility and morphology, and by
differential staining, including the Gram stain and staining with sudan black B for poly-p-hydroxybutyrate
(PHB) granules (Norris and Swain, 1971).
For electron microscopy, the cultures were fixed with glutaraldehyde and stained with osmium
tetroxide and uranyl acetate, and dehydrated in increasing strengths of ethanol (Hayat, 1981). For
scanning electron microscopy the samples were then dried with hexamethyldisilazane (HMDS,
Polysciences, Inc.) (Nation, 1983). For transmission electron microscopy the samples were imbedded in
VCD-HXSA, a low viscosity imbedding medium (Oliveira et al., 1983) with 0.5% silicone added for ease of
sectioning (Fran Thomas, pers. comm.). Thin sections were stained with lead citrate. Scanning electron
micrographs (SEMs) were taken on a Philips 505 scanning electron microscope, and transmission
electron micrographs (TEMs) were taken with a Philips 410 transmission electron microscope.
TCE DEGRADATION EXPERIMENTS
Cultures were raised to mid-log phase of growth in continuously-stirred reactors under a continuous
stream of methane in air, then transferred to 100 ml serum bottles or 250 ml screw cap bottles for
degradation studies. In early experiments cultures were raised under a stream of 20-25% methane in air,
which was later changed to 35-45% methane to enhance the growth rate. Cultures were raised in
"regular" mineral medium unless otherwise specified. Cultures were raised on a formulation here termed
"Whittenbury" mineral medium (Whittenbury et al., 1970) for some experiments. The pure culture
110
-------
received vitamins (20 u.g/1 biotin, 20 p.g/1 folic acid, 50 u.g/1 thiamine.HCI, 50 u.g/1 calcium pantothenate, 1
u,g/l B12, 50 u.g/1 riboflavin, 50 p.g/1 nicotinamide) (Mary Lidstrom, pers. comm.) when grown in liquid
culture. All degradation experiments were conducted in shake flasks, which were incubated upside-down
on a rotary shaker in a 21 °C environmental chamber.
Radiotracer Experiments
Degradation experiments with I4c-iabeled TCE were conducted to positively confirm that TCE was
being biologically transformed, and to determine the percentage of the TCE carbon in the cell, the CC>2,
and the aqueous intermediate fractions. The shake flasks were sacrificed for analysis using a liquid
scintillation counting (LSC) assay. Three types of controls, autoclaved cultures, gamma irradiated
cultures, and sterile mineral medium blanks were evaluated. Gamma irradiation kills the cultures but does
not rupture the cells as autoclaving does. Gamma-irradiated cultures were used to evaluate sorption of
TCE to the cells. Less than 3% of the 14C was associated with the gamma irradiated cultures after a one-
week incubation. It was therefore concluded that sorbed TCE represents a negligible fraction of the 14C
that is associated with TCE-transforming cultures. No significant differences were observed for the three
different kinds of controls. Autoclaved cultures were used for most experiments, and mineral medium
blanks were used for some.
LSC Assay--
A modification of a liquid scintillation counting assay, described in Section 10, was used to evaluate
the amount of 14C carbon in the TCE, CC-2, nonvolatile aqueous, and cell fractions. The method entails
nitrogen stripping 1 ml fractions to which base or acid has been added. In order to pull into solution the
CO2 in the headspace of the shake flask, base was added to the flasks before samples were withdrawn for
stripping (Figure 9.1). The method was modified because it was observed that raising the pH of the shake
flasks to pH 9.5-10 resulted in an artificially high aqueous intermediate radioactivity and an artificially low
cell radioactivity, most likely because the high pH was causing cell lysis. The modification involved adding
base to one set of replicate bottles to determine the TCE and CO2 radioactivity, and separating the cells
and the aqueous fraction in another set of replicate bottles by centrifugation, and acidifying the
supernatant after the cells had been removed.
Fractions were combined with 10 ml Insta-gel (Packard) liquid scintillation cocktail and evaluated by
liquid scintillation counting.
Preparation of Stock Solution-
It was discovered that the TCE arrives from the manufacturer with an approximately 6% nonvolatile
contaminant fraction. Therefore, it was necessary to prepare aqueous stock solutions in such a manner
that the nonvolatile contaminant fraction was not transferred to the stock solution. The radiolabeled TCE
stock solutions were prepared by diffusing the 14C-TCE (Sigma/Pathfinder) into sterile Milli-Q water. The
break-seal ampule containing the gaseous TCE was first dipped into liquid nitrogen to freeze the TCE into
the bottom of the sealed ampule. The break-seal was then broken with a sterile glass rod, and the ampule
tube was quickly inverted and attached with a Teflon/stainless steel swagelok fitting onto a flask containing
100 ml sterile Milli-Q water and a stir rod. The flask was secured to a stir plate in a 4°C chamber and stirred
for two days, permitting the TCE to diffuse into the water, while the nonvolatile contaminant fraction
remained behind in the break-seal ampule. A 14C-TCE stock solution of 99.3% purity was thereby
achieved, and the recovery of the TCE was approximately 90%.
Time-Series Experiments for Rate Determinations
Time-series analyses of TCE concentrations were conducted to generate rate coefficients. Cul-
tures were incubated with unlabeled TCE and the headspace concentration was analyzed according to a
method developed by Griddle et al. (1989) on a Tracor gas chromatograph equipped with an electron
capture detector. Sorption of TCE to cells was found to be insignificant and mineral medium blanks were
found to behave identically to autoclaved controls, so mineral medium blanks were used as controls. TCE
calibration standards were made up by weight in methanol and stored in a freezer.
111
-------
Two serum
bottles,
25 ml. culture
Add NaOH to pH 10
Shake
I
Centrifuge
1
Remove
Supernatant
1 ml
HC1
BASE ACID
Strip with nitrogen
I \
10 ml Liquid Scintillation Cocktail
FB
NON-STRIPPED BASE ACID
TCE CO2 Aqueous
CO2 Aqueous
Aqueous /
\/ \/
N - B = TCE B - A = CO2
Two serum
bottles,
25 ml. culture
I I
Centrifuge
Remove Supernatant
1 ml HC1
pH 3
CELLS ACID
Rinsed Strip w/ nitrogen
Re-centrifuged
*
Resuspended
I /
10 ml Liquid Scintillation Cocktail
FB
FB
CELLS
AQUEOUS
INTERMEDIATES
CELLS
AQUEOUS
INTERMEDIATES
Figure 9.1. Diagram of LSC assay.
112
-------
Cultures were diluted as necessary with either phosphate buffer or the same mineral medium
formulation in which they had been cultured, so that the rate of change of the TCE concentration in the
shake flasks did not exceed mass transfer rates. Cultures grown in Whittenbury mineral medium were
diluted approximately 1:20, and those grown in the Whittenbury without EDTA or FeNaEDTA,
approximately 1:2. Cell biomass was determined on a dry weight basis using pre-weighed rinsed and
dried 0.2 u.m Supor (Gelman) filters. Filters were dried overnight at 105°C and cooled in a desiccator over
phosphorous pentoxide.
The data generated by these experiments were modeled on a mass basis using the Monod
equation of substrate utilization kinetics, -dS/dt = kXS/(Ks + S). Mass was converted to aqueous
concentration on the basis of the headspace and aqueous volumes, using a Henry's constant of 0.33
(dimensionless ratio of mass concentrations). To generate the second-order rate coefficient, k/Ks,
degradation experiments were conducted at initial TCE concentrations of 35 to 60 u.g/1, and it was
assumed that Ks » S, yielding an equation of the form -dS/dt = kXS/Ks. The maximum utilization rate
k was determined at 3 mg/l TCE, at which concentration it was assumed that S » KS) yielding
-dS/dt = kX. Given k/Ks and k, Ks was then generated. The correlation coefficient R2 that describes the
fit of the data to the model was greater than 0.95 for all data reported here. Biomass was assumed to be
constant throughout, which proved to be a reliable assumption. Most rate experiments were conducted
with no methane added. When methane was added, it was at less than 0.4 mg/l, and given the low
methane concentration and the slow growth rate of the organisms, no significant increase in the biomass
was detected during the duration of the transformation experiments.
Analysis for TCE Transformation Products
Products of the oxidation of TCE by the pure culture Methylomonas MM2 were evaluated in the
nonvolatile aqueous fraction and in the headspace. To obtain an aqueous fraction with a high
concentration of products, but with a low anion concentration (NO" PO~ SO^", Ch : resulting from the
mineral medium), cells were centrifuged and resuspended in a 1/8th dilution ol regular mineral medium.
Dense cultures were incubated with 3 to 10 mg/l TCE, followed by centrifugation. The supernatant was
then filter-sterilized. Acidic break-down products in the supernatant were evaluated by ion
chromatography on a Dionex lonalyzer using 5 millimolar borate buffer. Volatile chlorinated products were
evaluated by gas chromatography. Carbon monoxide was evaluated by headspace analysis on a Trace
Analytical RGD2 reduction gas detector (hydrogen analyzer).
RESULTS
Cultures
Propane-Oxidizers--
Two propane-utilizing mixed cultures were enriched from aquifer material. The cultures did not
transform TCE and were not studied further.
Ethylene-Oxidizers-
Mixed culture-An ethylene-degrading mixed culture, EM1, was enriched from aquifer material. The
culture contained Gram variable bacilli as well as Gram negative bacilli and coccobacilli (Plate 9.1 a).
Pure culture-A pure culture, "Ethyl", was isolated from the ethylene-oxidizing mixed culture. This
pure culture was a Gram variable bacillus approximately 0.5 x 2 urn in size (Plate 9.1b). It formed slow-
growing bright yellow slimy colonies on solid media. Ethyl grows best on ethylene, and grows less well on
fructose, succinate, and tryptone-glucose. Fructose and succinate inhibit growth on ethylene. No growth
was observed on methane, methanol, or ethanol. When incubated under a 5% oxygen and 95% nitrogen
headspace with succinate in mineral medium lacking nitrate or ammonia, no growth was observed,
113
-------
indicating that Ethyl is not a Xanthobacter sp. Ethyl appears to fit the description of Mycobacterium sp.,
but this has not been confirmed.
Methane-Oxidizers--
Mixed cultures-Three different methane-oxidizing mixed cultures were obtained by enrichment.
The cultures are different, on the basis of macroscopic and microscopic appearance. MM1, enriched from
aquifer material, consisted of various Gram negative bacilli, coccobacilli, cocci, and prosthecates (Plate
9.2a). MM1 contains a type II methanotroph, an irregular sphere that is less than 1 urn in diameter (Plate
9.3a). MM2, enriched from column effluent, contained predominantly Gram negative motile bacilli and
Gram negative coccobacilli, as well as some yeasts (Plate 9.2b). Two different methanotrophs, a type I
(Plate 9.3b) and Type II (Plate 9.3c) were observed in this culture. MM3, enriched from groundwater,
contained primarily Gram negative bacilli and coccobacilli (Plate 9.2c). MM3 contained a type II methan-
otroph, an irregular sphere even smaller than the one in MM1 (Plate 9.3d).
Pure culture--A pure culture, Methylomonas MM2 ("NAL") was isolated from mixed culture MM2
(Plate 9.2d). This pure culture contains the stacked internal membranes characteristic of type I
methanotrophs (Plate 9.4a). Methylomonas MM2 is pigmented pink on liquid and solid media, turning
yellow with age. It is a Gram negative motile bacillus. Methylomonas MM2 assimilates carbon by the
ribulose monophosphate pathway, and does not test positive for enzymes of the serine pathway. It is
microaerophilic and grows well on methane and methanol. It exhibits background growth on solid media,
but exhibits no growth on multicarbon substrate in liquid media.
An unusual pattern of lysis, uncharacteristic of old age or nutrient deprivation, was sometimes
observed. Transmission electron micrographs of the lysed cells revealed electron-dense, regularly-
shaped hexagonal spots approximately 60 nm in diameter, within the perimeter of lysed cells (Plates 9.4b
and 9.4c) as well as attached to the surface of some live cells (Plate 9.4d). The spots, on the basis of the
fixing and staining protocol, could only be nucleic acids. All of this strongly suggests that the culture had a
phage associated with it. Attempts to obtain plaques on Methylomonas MM2 were unsuccessful. This
cannot be taken as an indication the phage are not present, however. It is likely that the culture was
already lysogenized by the phage, rendering it immune to to infection and subsequent lysis.
TCE Transformation--
Table 9.1 summarizes a range of TCE transformation rates observed for the cultures when grown in
regular mineral medium and incubated with and without methane under various conditions. The TCE
concentrations used in these experiments were 40 to 60 u.g/1.
Propane-oxidizers-The propane oxidizers did not transform TCE.
Ethvlene-oxidizers-Both the ethylene-oxidizing mixed culture EM1 and the pure culture Ethyl
transformed 100% of the TCE. Transformation occurred with and without ethylene present, and high
ethylene concentrations (tested at 13 mg/l) inhibited transformation. The mixed culture transformed most
of the TCE to CO2 and cell fractions, with about 10% remaining in the nonvolatile aqueous fraction. The
pure culture transformed the TCE to CO2, cell, and nonvolatile aqueous fractions. The pure culture was
tested for growth on TCE, cis-DCE, and trans-DCE. No growth was observed. The rate of TCE
transformation by the ethylene-oxidizers was approximately two orders of magnitude less than the rates
exhibited by the methane-oxidizers (Table 9.1).
Methane-oxidizers-AH the methane-oxidizing cultures transformed TCE. Rates were variable,
depending upon growth conditions and incubation conditions. All exhibited what has been termed
"stationary transformation", that is, they transformed TCE in the absence of methane. Unless impaired by
incubation conditions, 100% of the TCE was transformed. In mixed cultures, the TCE was transformed
primarily to CO2 and cells, with about 5-10% remaining in the nonvolatile aqueous fraction after extended
incubation. In pure cultures, approximately 5-10% was transformed to CC>2,10-20% to cells, with 70-80%
remaining in the nonvolatile aqueous fraction. Acetylene inhibited methane oxidation and TCE
transformation, indicating that the methane monooxygenase enzyme catalyzed the oxidation.
114
-------
Plate 9.1. Scanning electron micrographs of ethylene-oxidizing cultures. Mixed culture EM1 (9.1 a) and
pure culture 'ethyl' (9.1b).
115
-------
Plate 9.2. Scanning electron micrographs of methane-oxidizing cultures. Mixed cultures MM1 (9.2a),
MM2 (9.2b), MM3 (9.2c), and pure culture Methylomonas MM2 (9.2d).
116
-------
MM2
0,3 Mm
Plate 9.3. Transmission electron micrographs of the methane-oxidizing mixed cultures. (9.3a) Type II
methanotroph in MM1. (9.3b) Type I and (9.3c) type II of MM2. (9.3d) Type II of MM3.
Methanotrophs of MM1 (9.3e), and MM3 (9.3f) containing PUB granules.
117
-------
1um
•9.4d; ..? 'V.: /-
Plate 9.4. Transmission electron micrographs of pure culture Methylomonas MM2 showing Type I
membranes (9.4a), and cells infected with phage (9.4b, 4c, 4d).
118
-------
TABLE 9.1 TCE TRANSFORMATION RATES
k/Ks (l/mg-day)
Methane-oxidizers
mixed culture MM1 0.039-0.047
mixed culture MM2 0.000-0.010
mixed culture MM3 0.043-0.048
Methylomonas MM2 0.000-0.17
Ethvlene-oxidizers
mixed culture EM1 0.0002-0.0006
pure culture ("Ethyl") 0.0008-0.0009
Rate coefficients were determined for the three mixed cultures. Using the first and second-order
coefficients generated from degradation studies conducted at 3 mg/l and 60 ug/l respectively, the half-
saturation coefficient Ks was calculated (Table 9.2). The fit of the data to the model is described by the R2
value, an R2 value of 100 describing a perfect fit.
TABLE 9.2 RATE COEFFICIENTS FOR METHANE-OXIDIZING MIXED CULTURES
3 mg/l TCE 0.07 mg/l TCE
R2 k/Ks (l/mg-day) R2 Ks(mg/l)
Mixed Culture
MM1
MM2
MM3
0.0060
0.0088
0.0097
0.97
0.99
0.90
0.041
0.01
0.046
0.99
..a
0.99
0.15
0.9
0.21
a The k/Ks value used for this calculation is the average of several experiments.
Effects of Variables on TCE Transformation by Methane-Oxidizers -
Methane concentration-Mixed cultures MM1 and MM2 and pure culture Methylomonas MM2 were
tested for the effects of 3 different methane concentrations-no methane; 2% of headspace (1.3 mg total,
0.45 mg/l aqueous); 20% of headspace (13 mg total, 4.5 mg/l aqueous)~on TCE transformation.
The LSC assay method was used to evaluate the fraction of the TCE transformed into the
nonvolatile aqueous, COa, and cell fractions by determining the 14C in those fractions. Results are
presented in Figures 9.2, 9.3, and 9.4. The percent of the TCE transformed at the given time in hours is
represented by bar graphs. The percent of methane and oxygen in the headspace, representing a mean
of four bottles, is also given. Determinations were done in replicate as described in the methods section.
119
-------
All three cultures transformed TCE when no methane was present. However, mixed culture MM1
was the only one in this experiment to transform TCE significantly faster when no methane was present.
Rates of transformation for MM2 and the pure culture when no methane was present were significantly
slower than when methane was provided.
The enzyme system that oxidizes TCE, methane monooxygenase (MMO), is the same enzyme that
oxidizes methane. Since the TCE and methane compete for the same enzyme, methane will competi-
tively inhibit TCE transformation. In all cultures tested, high methane concentrations had an inhibitory
effect. This effect was particularly pronounced in mixed culture MM1: at 1/2 hour, 70% and 60% of the
TCE had been transformed under 'no' and 'low methane' respectively, and only 10% had been trans-
formed under 'high methane'. In other experiments with the pure culture, Methylomonas MM2, 2 mg/l
methane resulted in a 25% reduction in the initial TCE oxidation rate.
Availability of reducing power--All of the cultures transformed TCE in the absence of methane, a
desirable capability in light of the competitive inhibition. However, the response to methane deprivation
was variable. Depending upon the duration of deprivation, TCE transformation capability was lost. This
effect relates to the availability of reducing power. The MMO requires a source of reducing power to carry
out its oxidative function. When methanotrophs are cometabolizing TCE in the absence of a substrate,
depletion of the source of a source of reducing power can halt the cometabolic transformation.
TIME
Figure 9.2. Effects of methane concentration on percent TCE transformed into the nonvolatile
aqueous, COg, and cell fractions by mixed culture MM1 (dry weight = 0.40 mg/ml).
120
-------
ro
(Q
CO
£ ED
II
c to
-wo
o-
0) 3-
a§
a •
= 8
o 3
11-
PERCENT TCE TRANSFORMED
II
-
"S 3
^ "
< 2.
1 5
<. 3
P5'
T' O
(Q
C
CO
CO
1
CO
-wo
0) -3
a§
O CD
ii
to 0)
gCD
33
I!
PERCENT TCE TRANSFORMED
222
O * i8* i8* i8* I8* r I |
I. a
i
CD
w
CD"
-------
Mixed cultures MM1 and MM3 were found to transform ICE as rapidly in the absence of methane as
in the presence of low concentrations of methane. Mixed culture MM2, and pure culture Methylomonas
MM2 transformed TCE less well in the absence than in the presence of mi Hi molar concentrations of
methane. In one experiment, when Methylomonas MM2 was incubated without methane for 24 hours and
then tested for TCE oxidation, it had completely lost its ability to transform TCE. We had hypothesized
that the PHB granules may act as a source of reducing power for the oxidation of non-metabolizable
compounds by the methane monooxygenase enzyme, and that the methanotrophs of mixed and pure
cultures MM2 and Methylomonas MM2 lacked such storage granules. Transmission electron micrographs
revealed that the methanotrophs of mixed cultures MM1 and MM3 contain PHB granules (Plates 9.3e and
9.3f), whereas Methylomonas MM2 and the methanotrophs of mixed culture MM2 do not contain PHB
granules.
Table 9.3 presents the results of an experiment evaluating the effects of starvation on TCE
transformation. The mixed cultures MM1, MM2, and MM3 were incubated for 15 hours in the absence of
methane, and then evaluated for TCE transformation capability in the presence and absence of methane.
The change in the TCE concentration was evaluated in duplicate bottles. MM1 and MM3, the two cultures
with methanotrophs containing PHB granules, transformed TCE as well, whether or not methane had
then been added to the shake flasks. Mixed culture MM2, however, could no longer transform TCE
unless methane was added, and even when methane was added the culture did not recover its full
capabilities.
TABLE 9.3. MIXED CULTURE TCE TRANFORMATION RATES
k/Ks (l/mg, day)
No Methane 0.4 mg/l Methane
Mean ± Std. Dev. Mean ± Std. Dev.
Mixed Culture
MM1
MM2
MM3
0.041±0.003
0
0.046±0.001
0.046±0.001
0.004+0.001
0.046±0.004
Oxygen concentration-Oxygen is required by methanotrophs both for respiration and for the
oxidative function of the MMO. Many methanotrophs, however, have been described as microaerophilic,
and grow best at oxygen tensions below atmospheric. High oxygen concentrations at 50% of headspace
were inhibitory, reducing TCE transformation by approximately 20%, in all cultures tested. Oxygen at 35%
was not inhibitory. When cultures were incubated under near-anoxic conditions (< 1% 02), very little
methane or TCE was oxidized. When oxygen was depleted, TCE transformation stopped.
TCE concentration--TCE and its oxidation product, the TCE epoxide, are cytotoxic. Toxic inhibition
of TCE transformation can be anticipated at some concentration in the mg/l range. The mixed cultures
MM1, MM2, and MM3 were evaluated for TCE transformation and methane consumption at 5,10,44, and
87 mg/l TCE using radiolabled TCE. All of the TCE was transformed and all of the methane was consumed
at 5 and 10 mg/l TCE. At 44 and 87 mg/l TCE, very little or none of the methane was consumed and 20%
122
-------
or less of the ICE was transformed, depending upon the culture. Rate studies with Methylomonas MM2
suggest that TCE or some of the oxidation intermediates may exert a subtle inhibitory effect on TCE
transformation by the pure culture at concentrations as low as 3-5 mg/l. Such an effect could complicate
the determination of the rate constant k and the half-saturation constant Ks using Monod kinetics, and
may require the inclusion of a term to describe the inhibitory effect.
Biomass concentration/mass transfer limitations-ln early experiments with radiolabeled TCE, mass
transfer may have limited transformation rates. To test for this, a degradation study with mixed culture MM1
was conducted at four different cell densities using radiolabeled TCE. Mixed culture MM1 was raised to a
cell density of 0.41 mg/ml dry weight and diluted, achieving the following cell densities: 0.014 mg/ml,
0.078 mg/ml, 0.24 mg/ml, and 0.41 mg/ml. After dilution, cultures were transferred to bottles for
degradation studies. Each bottle received 1.3 mg methane, achieving an initial aqueous concentration of
0.45 mg/l, as well as 6 ug TCE, achieving an initial aqueous concentration of 105 u.g/1 TCE. In Figure 9.5,
the natural log of the percent TCE remaining is plotted against time. If no other factor were affecting TCE
transformation, then the amount of TCE transformed would have increased proportionately with
increasing biomass. However, at 2 hrs and again at 8 hrs, the amount of TCE transformed was not
significantly different for 0.24 and 0.41 mg/ml dry weight cells. Mass transfer of oxygen, or more likely
TCE, must have been inhibiting transformation rates.
In subsequent studies, care was taken to ensure that TCE transformation rates did not exceed TCE
mass transfer rates. The shaker table speed was doubled to greater than 250 rpm and cultures were
diluted, as described in the Methods subsection.
Mineral medium formulation-Pure culture Methylomonas MM2 and mixed culture MM1 were
evaluated for the effects of growth in different mineral media on TCE transformation rates. Tables 9.4 and
9.5 summarize the rates of TCE transformation in the absence of methane at a TCE concentration of
approximately 40 u.g/1. Each value represents the mean of two to four bottles. Whittenbury mineral
medium is different from regular medium in that it lacks copper, but contains nickel, EDTA and FeNaEDTA.
°T
eg -0.2.
=
§
0>
gj -0.6.
U -0.8 .,
! ^
«* -1.2 . .
-1.4 ..
-1.6
18
TIME, Hours
Figure 9.5. Biomass concentration/mass transfer limitation study.
123
-------
TABLE 9.4 METHYLOMONAS MM2 TCE DEGRADATION RATES
k/Ks (l/mg, day)
Mean ± Standard Deviation
Mineral Medium Formulation
Regular 0.17 ±0.03
Whittenbury 0.50 ± 0.03
Whittenbury + copper 0.57 ± 0.03
Whittenbury, no EDTA or FeNaEDTA 0.01 ± 0.002
Whittenbury, no EDTA 0.01 ± 0.002
TABLE 9.5. MIXED CULTURE MM1 TCE DEGRADATION RATES
k/Ks (l/mg, day)
Mean ± Standard Deviation
Mineral Medium Formulation
Regular 0.041 ± 0.003
Whittenbury 0.61 ± 0.02
For both cultures, when EDTA was included in the growth medium, there was a significant increase
in the TCE degradation rates. Methylomonas MM2 grown in Whittenbury mineral medium degraded TCE
three times faster than when it was grown in regular mineral medium, and fifty times faster than when it was
grown in Whittenbury without EDTA. Mixed culture MM1 grown in Whittenbury mineral medium degraded
TCE fifteen times faster than when it was grown in regular mineral medium. The presence of copper was
not responsible for the lower rates observed for the culture grown in regular mineral medium: addition of
copper to the Whittenbury formulation did not decrease the rate. Neither is increased iron availability
responsible for the increased rates observed when the culture is grown in the Whittenbury formulation:
omitting EDTA but including FeNaEDTA in the formulation does not increase the rates.
The literature reports (summarized by Hou, 1984a) that the soluble methane monooxygenase
enzyme (MMO), of the three described methanotrophs that do express a soluble MMO, has a broader
specificity, and it has been proposed that the soluble MMO may have a greater affinity for TCE than the
paniculate MMO (R. S. Hanson, pers. comm.). The soluble MMO in these three strains is expressed
under conditions of high density and copper limitation. Copper is thought to control the expression of the
soluble MMO at the level of transcription (Mary Lidstrom, pers. comm.), and when copper is available, the
methanotroph expresses the membrane-bound "paniculate" MMO that is thought to have a lower affinity
for TCE. Since addition of copper to the Whittenbury medium does not decrease the TCE transformation
124
-------
rate, however, and in fact may increase it, it is unlikely that such a mechanism of control is operative in
Methylomonas MM2. Moreover, transmission electron micrographs reveal that Methylomonas MM2
produces intracellular membranes regardless of growth conditions, which further indicates that the
organism is not expressing a soluble form of the MMO.
TCE Transformation Intermediates-
Based on studies of the products resulting from the breakdown of TCE epoxide under aqueous
conditions (Henschler et al., 1979; McKinney et al., 1955; Miller and Guengerich, 1982), carbon
monoxide, formate, glyoxylate, and dichloroacetate have been proposed as the principal products
resulting from the methanotrophic oxidation of TCE and the subsequent abiotic rearrangement of the
epoxide (Henry and Grbi6-Gali6 1986, Little et al., 1988). Using ion chromatograpy to evaluate the
nonvolatile aqueous intermediate fraction produced by his pure cultures, Little trapped radiolabeled
fractions eluting from the ion chromatograph that had the same retention time as glyoxylate and
dichloroacetate standards (Little et al., 1988, C. D. Little, pers. comm.).
Methylomonas MM2 does produce acids from the oxidation of TCE. Either glyoxylate or formate or
both were present in the supernatant (the glyoxylate and formate peaks overlapped so it was not possible
to quantify these acids), but no mono-, di-, or tri-chloroacetate were present. There were, however, three
large peaks that could represent di-acids, and/or chlorinated di-acids, and/or chlorinated acids or di-acids
with more than two carbons. The culture supernatant will have to be derivatized and evaluated by gas
chromatography/ mass spectrometry to distinguish between formate and glyoxylate and to identify the
unknown acid peaks. No volatile chlorinated products were detected.
No carbon monoxide was detected in the headspace while Methylomonas MM2 was oxidizing TCE.
However, this does not confirm that carbon monoxide is not a transformation intermediate. Methylomonas
MM2 oxidized carbon monoxide, and it is possible that the carbon monoxide was oxidized before the
bottles were evaluated. The effect of carbon monoxide as a competitive inhibitor of TCE oxidation is
being investigated.
CONCLUSIONS
These findings contribute to the emerging definition of the range of capabilities that can be
anticipated in a methanotrophic treatment application, and help delineate some of the desirable
characteristics of a methanotrophic community suitable for treatment applications. A suitable
methanotrophic community should be able to transform TCE in the absence of methane, and should
contain storage granules as a source for regenerating reducing power in the absence of substrate, such
as the type II methanotrophs in mixed cultures MM1 or MM3. It should be able to tolerate variations in
oxygen concentration, as well as high TCE concentrations, without significant cytotoxicity or impact on
TCE transformation rates. Especially promising is the discovery that a given mineral medium formulation
can increase rates by one to two orders of magnitude. The ideal methanotroph would be one capable of
the rapid TCE transformation rates exhibited by pure culture Methylomonas MM2 and by mixed culture
MM1 when grown on Whittenbury mineral medium.
125
-------
SECTION 10
BATCH EXCHANGE SOIL COLUMN STUDIES OF BIOTRANSFORMATION
BY METHANOTROPHIC BACTERIA
Nancy Lanzarone, Kevin Mayer, Mark Dolan, Dunja Grbic"-Gali6, and Perry McCarty
INTRODUCTION
Laboratory-scale columns containing aquifer material were found effective for studying the potential
of aquifer and soil organisms to biotransform halogenated compounds (Wilson and Wilson, 1985; Siegrist
and McCarty, 1987). However, the strong sorption of halogenated organics by aquifer material (Mackay et
al., 1986) complicates the interpretation of laboratory studies (Wilson and Wilson, 1985; Siegrist and
McCarty, 1987). Sorption retards the transport of pollutant? in groundwater and may affect their
transformation as well.
The objective of this study was to evaluate the use of laboratory-scale soil columns for estimating
the important factors in the development of in-situ treatment processes. The significant questions to be
answered when considering in-situ bioremediation include: 1) are native bacteria present in the aquifer
that are capable of growing on an added primary substrate, 2) what is the period of time required to
increase the population of such bacteria to an adequate level, and 3) following the population increase of
such bacteria, are they capable of transforming the contaminants of concern? The contaminants of inter-
est in the batch exchange soil columns were TCE, 1,2-DCA, and VC; the primary substrates (electron
donors) were propane and methane. Electron acceptors included dissolved oxygen and hydrogen
peroxide as a substitute for dissolved oxygen. Assuming that contaminant transformation could be
achieved, it was desired to investigate the effects of the primary substrate concentration and nutrient
supplements (NH3-N and PO4-P) on the rate and/or extent of aerobic TCE degradation. The extent of
degradation of TCE that had been sorted to the aquifer solids also was studied.
MATERIALS AND METHODS
Column Preparation
The aquifer solids were obtained in July 1986 from the Moffett Naval Air Station, Santa Clara Valley,
California, using a hollow-stem auger drilling rig. A 6" pitcher barrel that had been flamed with ethanol to
reduce contamination of the cores was driven ahead of the drill bit. The material, which was recovered
from a depth of 14.5-19.5 ft, consisted primarily of fine through coarse sand, gravel, and minor clay. To
remove the inner core material without contamination, a 4"-diameter pipe that had been flamed with
ethanol was hammered into the aquifer material inside the pitcher barrel. All tools that might come in con-
tact with the aquifer material were either autoclaved or flamed with ethanol. The material obtained was wet
sieved with an autoclaved number 10 sieve (openings < 2 mm) and stored in sterile mason jars until used.
The laboratory columns (Figure 10.1) (Siegrist and McCarty, 1987), were autoclaved prior to being
filled with the aquifer material. A Bunsen burner was operated during the packing to maintain the airflow in
an upward direction and thus help prevent airborne contamination. Aquifer material was added with a
scoop while the columns were lightly tapped with a plastic rod to help the material settle evenly.
126
-------
Teflon Connector!
\
^J2.8 cm
em
7 em
, 1
i
D
¥::
iS
: •:
f
1
x%
x::
•:•:•
'•*•*
1
r
T«flon Tublnp
Rutabor Stoowr
SIMl NMdli
Glut Wool
Aoulfor liiurlil
Total Volumo
V, » 133 ml
Pan Valuim
Vw» 60 ml
Glm Column
1.0. « 2 cm
Gin* Wool
Rubber SIODDtr
Figure 10.1. Column design.
Groundwater from the field site that had been air-stripped and autoclaved was pumped in an upflow
direction at approximately 8-10 ml/min to help eliminate fines that otherwise might cause clogging. The
columns were visually inspected for uniformity of packing, covered with aluminum foil to prevent the
growth of photosynthetic organisms, and kept fully water-saturated and at room temperature (22°C) for the
duration of the experiment.
Chemicals and Stock Solutions
The 14C-trichloroethylene (14TCE) used (Pathfinder Laboratories Inc., St. Louis, Mo.) had a specific
activity of 4.11x104dpm/u.g. Radiolabelled 14C-1,2-dichloroethane (141,2-DCA) in methanol (Amersham
Corp., Grover Heights, II.) was diluted with unlabeled reagent grade 1,2-DCA (J. T. Baker Chemical Co.,
Phillipsburg, N. J.), yielding a specific activity of 1.38x104 dpm/u.g. A nutrient stock solution was prepared
in Milli-Q water with (NH3)2HPO4 (0.170 g/l) and NH4CI (0.244 g/l) (J. T. Baker Chemical Co., Phillipsburg,
N. J.). The vinyl chloride (VC) used (Alltech Associates Inc., Deerfield, IL) was a 1027 ppm VC in nitrogen
gas 2-component NBS traceable Scott specialty gas equilibrated under pressure in Milli-Q water.
Column Operation
The column fluids were exchanged approximately once each week with 126 ml (st. dev = 2 ml;
n = 187) of new feed solution. A syringe pump (Sage Instruments; Division of Orion Research Inc.,
Cambridge, Mass) with two 100-ml gas-tight syringes with adjustable plungers (Spectrum, Houston, Tx.)
was used to exchange the liquid in an upflow direction at a flow rate of 5 ml/min. Dissolved oxygen break-
through curves were similar for all of the columns (data not shown). These indicated that the first 18-20 ml
of liquid removed from a column during an exchange would not be significantly contaminated by short-
circuiting of influent fluid.
Three different samples, collected and analyzed during each exchange, were termed as follows:
influent, initial effluent, and final effluent (Figure 10.2). The influent sample was obtained from the feed
syringes before any contact with the column occurred. The initial effluent sample represented the first 18
127
-------
Influent Sample
Syringe Pump
5 ml/min
Two 100 ml
Syringes
0-18 ml
Initial Effluent
19-106 ml
Discarded
106-126 ml
Final Effluent
Figure 10.2. Experimental design for sample mixing and column feeding.
ml of liquid removed from the column during an exchange. The initial effluent was assumed to represent
the composition of the pore fluid within the column just prior to an exchange, and was compared with the
influent from the previous exchange to determine oxygen demand, primary substrate consumption, TCE,
1,2-DCA, or VC degradation, and mass balances. The final effluent sample represented the last 18-20 ml
removed from the column during an exchange.
Feed Solution Preparation
Groundwater from the field site that had been air-stripped was used to exchange the columns and
was stored in the dark at room temperature (22°C) until used. Prior to exchange, the water was filter-steril-
ized using 0.2-u.m sterile cellulose nitrate filters (Micro Filtration Systems, Dublin, Ca.) and placed in auto-
claved glass bottles that were connected to gas cylinders. The volume of the bottle attached to an oxy-
gen cylinder was 4.01, that to a methane cylinder was 2.0 I, and that to a propane cylinder was 1.01.
Some of the filtered water also was placed in an autoclaved 2-I glass jar. The appropriate gas was bubbled
through each bottle of water using a gas diffuserfor one hour prior to exchange.
An appropriate amount of methanated or propanated water and 2 ml of nutrient supplement solution
were drawn into one 100-ml syringe and mixed back and forth through Teflon tubing to another 100-ml
syringe. If a column received a low concentration of methane (1.5 mg/l) rather than a high concentration
(4.5 mg/l), water from the 2-I glass jar was added for dilution to give the same total volume (50 ml).
The 50-ml solution was then divided evenly between the two syringes, and both were filled to
100 ml with oxygenated water. The syringe fluids were mixed again through connecting tubing to create
a single uniform feed. Influent samples for dissolved gas analyses were taken with a 20-ml glass syringe
attached to the connecting tubing, and analyzed immediately. The syringes then were separated and
emptied to 75 ml, and either 0.5 ml of the 14TCE stock solution, 0.10 ml of the 141,2-DCA stock solution,
or 5 ml of the VC stock solution was added. The syringes were reconnected, and the solutions mixed.
128
-------
Next the syringes were reconnected with Teflon tubing and placed on the syringe pump. After the
influent samples for gas chromatographic analysis and scintillation counting were taken with a 20-ml glass
syringe, the 100-ml syringes were attached to the column, and the column fluid was exchanged in an
upflow direction at a flow rate of 5 ml/min.
Microcosm Batch Experiments
A set of microcosm batch experiments was designed to supplement the aquifer column studies of
trichloroethylene degradation by a mixed culture of methanotrophic bacteria. The 35-ml microcosms were
sacrificed at various time intervals to measure the growth rate and to examine the relationship between the
growth stage of the methanotrophic community and the degradation of TCE.
Initial experiments were conducted using sets of 35-ml vials inoculated with 1.5 g of well-mixed
aquifer solids (sand) from columns that had been operated as draw-and-fill sequential batch reactors. In
the second of the two initial experiments, additional inoculum was added in the form of 2 ml of effluent that
had been collected while operating the draw-and-fill aquifer material columns. A mineral solution
containing 4.5 mg/l of dissolved methane, 24 mg/l of dissolved oxygen and 50 u.g/1 of TCE was added to
each vial. An equal number of vials were established as controls by omitting the dissolved methane from
the solution. The microcosms were sealed to exclude any headspace, then placed in a rotary shaker in
the dark at room temperature (22°C). Triplicate microcosms were sacrificed at various time intervals, as
often as every 4 hours during periods of rapid growth.
Each vial was analyzed for methane, dissolved oxygen, and 14C-labeled TCE and CC>2 activity.
Acridine orange direct counts (Ghiorse and Balkwill, 1983) and total protein determination (Kennedy and
Fewson, 1968) were used to estimate microbial population.
Analytical Methods
Dissolved Oxygen--
Dissolved oxygen (DO) concentrations were measured with a Model 54A oxygen meter (Yellow
Springs Instruments, Yellow Springs, OH) that had been calibrated in air-saturated water. The coefficient
of variation for DO less than 20 mg/l was 4% (n = 8), and for higher DO was 1.3% (n = 10).
Dissolved Methane and Propane Analyses--
Dissolved methane and propane concentrations were determined by headspace analysis. Bottles
of known volume (approximately 13 ml) and weight were sealed with rubber stoppers, and 10 ml of air was
withdrawn twice with a 10-ml glass syringe to create a vacuum. From the 20-ml sample syringe, 5 ml of
sample was placed in an evacuated bottle. The bottles were shaken vigorously by hand for 30 sec,
equilibrated with air to atmospheric pressure using a syringe, and 0.5 ml of the gas phase was removed
with a 1-ml gas-tight syringe for analysis on a Hewlett Packard 5730A gas chromatograph (Hewlett
Packard, Avondale, Pa.) equipped with an FID detector and a 60/80 Carbosieve column (5ftxl/8";
Supelco Inc., Bellafonte, Pa.). Quantification was achieved by injecting 0.5 ml of calibration gas (1000
ppm) of either methane or propane (Alltech Associates, Deerfield, IL) and comparing the relative area with
a Spectra-Physics 4020 calculating integrator (Spectra-Physics, Sunnyvale, Ca.). The coefficient of
variation for methane based on three sets of data was 9%.
TCE, 1,2-DCA, and VC Analyses--
Concentrations were determined by gas chromatography. Samples collected in 20-ml syringes
were transferred immediately to 14-ml bottles, sealed without headspace with Teflon-faced septa and alu-
minum crimp tops, and extracted with 1 ml of degassed iso-octane (after Henderson et al., 1976) using
bromochloropropane as an internal standard. A 5 u.l sample of the extract was injected into a packed col-
umn (10% squalene on Chromosorb W/AW, at 60°C) on a GC equipped with a linearized Ni-63 electron
capture detector. Quantification was achieved by injecting 10, 50, and 100 u.g/1 standards that had been
treated like samples and comparing the relative areas with a Spectra-Physics 4020 calculating integrator.
The influent was analyzed for each exchange, but the initial effluent was analyzed only periodically. The
analytical coefficients of variation for TCE and 1,2-DCA were ± 5% and ± 6%, respectively.
129
-------
Vinyl chloride concentrations were determined by headspace analysis using gas chromatography.
Approximately 5 ml of the collected samples were transferred immediately into 15-ml glass vials and sealed
with Teflon-faced septa and screw top closures. The vials were inverted, placed on a shaker table, and
allowed to equilibrate for 30 min. A 300-u.l sample of the headspace gas was injected into a capillary
column (J&W fused silica megabore DB 624) on a Finnigan 9610 GC fitted with a Hall electrolytic
conductivity detector. Quantification was achieved using standard curves created by injecting NBS
traceable vinyl chloride in nitrogen gas standards with peak areas integrated using the Nelson Analytical
3000 Series chromatography data system. The influent, initial effluent, and final effluent were analyzed
for each exchange.
Radtoactfvtty Anar/ses-
Carbon-14 activity was assayed using a Tricarb Model 4530 scintillation spectrometer (Packard
Instrument Co., Downers Grove, IL) which made counting efficiency corrections with the external standard
channels ratio method (Harrocks, 1974). For each sample, three separate aliquots were counted. A
1.0-ml sample was injected into a glass counting vial containing 3 drops of 1N HCI, another 1.0 ml into a
vial containing 6 drops 1N NaOH, and a third 1.0 ml into a vial containing 10 ml of liquid scintillation cocktail
(Instagel; Packard Instrument Co., Downers Grove, IL). The first two vials were stripped with nitrogen gas
(100 mYmin for 15 min), and then 10 ml of liquid scintillation cocktail was added. This procedure allowed
making a presumptive test for the production of 14COo, which would not be stripped at high pH but would
be at low pH. The volatile 14C-activity, which was air-stripped at any pH, was assumed to represent
nontransformed TCE or 1,2-DCA, or VC.
Hydrogen Peroxide Determination-
The presence of hydrogen peroxide, in the experiments where hydrogen peroxide was used as an
alternate source of dissolved oxygen, was detected by peroxide test strips (EM Science). The
concentration of hydrogen peroxide was measured by diluting 10-ml samples with deionized water,
adding 5 ml of 10% sulfuric acid, and three drops of ferroin indicator, and titrating the solution with 0.1N
eerie sulfate to a blue endpoint. Two moles of cerium are reduced in an acidic solution by each mole of
hydrogen peroxide present. A series of standard solutions and deionized water blanks were analyzed
along with the samples.
Evaluation of Transformation and Degradation-
The term biotransformation generally represents a change in the chemical structure of a compound,
while degradation represents the combination of mineralization to inorganic end products and conversion
to cellular products. In this study, no volatile transformation products were found. Moreover, since GC
analyses for TCE and 1,2-DCA corresponded closely with the concentration of 14C measured volatile
compounds, no measurable quantities of volatile transformation products were formed. Thus, the
measured transformation products were in the form of either nonvolatile organics or CO2.
Measurement of compound degradation presented some difficulties. The 14TCE was contami-
nated such that about 8% of the activity was nonvolatile organics or CO2, with about 90% being the
former. The uncharacterized nonvolatile fraction was completely biodegradable and could be transformed
to 14CO2 while in the column. To reduce errors in interpretation, degradation is defined here to be the
change in the 14C activity of the nonvolatile plus CO2 fractions between the initial effluent from the current
exchange and the influent from the previous exchange (NaOH sample activityjnjtja| effluent ~ NaOH sample
activityjnf|uent). Degradation thus represents the change in the nonvolatile plus CO2 fractions rather than
just the production of CO2. This procedure may underestimate the extent of mineralization that may have
actually occurred since the TCE that was incorporated into biomass could not be determined for the
columns, because it would have required their destruction.
130
-------
RESULTS
Initial Column Studies on TCE. 1.2-DCA. and VC Decomposition
Approach--
Seven columns were initially operated as described in Table 10.1, and the subsequent modifi-
cations that were made are described below. The columns were exchanged 6 times before TCE was
added, and 7 times before 1,2-DCA was added. Day 0 on all graphs represents the first day that the
chlorinated compound was added.
TABLE 10.1. THE ORIGINAL EXPERIMENTAL DESIGN
Column Number Original Operating Conditions3
1 Dissolved oxygen and TCE (Control)
2 Dissolved oxygen, dissolved propane, nutrient supplements, and TCE.
3 Dissolved oxygen, 4.5 mg/l dissolved methane, nutrient supplements, and TCE
4 Same as 3
5 Dissolved oxygen, 4.5 mg/l dissolved methane, and TCE
6 Dissolved oxygen, 1.5 mg/l dissolved methane, nutrient supplements, and TCE
7 Dissolved oxygen, 1.5 mg/l dissolved methane, nutrient supplements, and 1,2-DCA
alnfluent concentrations were: DO - 25 to 35 mg/l; TCE - 25 u.g/1; 1,2-DCA - 80 u.g/1.
Electron Donor and Acceptor Utilization-
Dissolved oxygen (DO) was consumed in the control column (Figure 10.3) and in all of the columns
fed methane and propane (data not shown). The oxygen demand was greater than expected from the
stoichiometry for complete oxidation of methane and propane, and averaged 17 mg/l (st. dev. = 7 mg/l). It
may have been associated with reduced aquifer solid components, but the actual source was not found.
An active methane-utilizing population was present within the aquifer solids, as evidenced by the
decrease in the initial effluent methane concentration within 7 days (data not shown). This corresponded
closely with the two-week period required for methane utilization to begin in the field studies (Figure 6.2).
Propane did not stimulate the growth of an active biomass capable of its own oxidation, even after several
months of operation (data not shown). Thus, a propane-utilizing population did not appear to be present
in the aquifer material. No TCE degradation occurred in this column.
The Effect of Methane on TCE Degradation-
Two methane concentrations were evaluated: 4.5 mg/l (high) and 1.5 mg/l (low). A pulsing strategy
also was investigated in which one column was fed an alternating pulse of methane and oxygen one week
followed by oxygen alone in the next. Profiles of influent and effluent methane concentrations before
and after pulsing are illustrated in Figure 10.4.
The influent TCE concentration for all columns was approximately 25 u.g/1. Negative values some-
times calculated for TCE degradation probably reflect errors due to the initial sorption of the 14C-labeled
nonvolatile fraction prior to the initiation of its biodegradation, or analytical errors as indicated by the slight
fluctuation around zero for the control column (Figure 10.5). Degradation is believed not to have occurred
in the control column and certainly was so small as to be insignificant.
131
-------
I
<5
1
Influent
Initial Effluent
Final Effluent
0 50 100 150 200 250 300 350
Time (Days)
Figure 10.3. Dissolved oxygen concentration versus time for the control column (oxygen and TCE).
f
1
Influent
Initial Effluent
Final Effluent
0 50 100 150 200 250 300 350
Time (Days)
Figure 10.4. Methane concentration versus time for column 4.
132
-------
16
I
I
I
g
Degradation
WtEffTCE
50 100 150 200 250 300 350
Time (Days)
Figure 10.5. ICE degradation and concentration in the initial effluent from the control column (oxygen
andTCE).
Column 6, with 1.5 mg/l methane and nutrients, was the first to show TCE degradation, which
became measurable after about 80 days (Figure 10.6). The extent of TCE degradation there shown, as
well as the pore water TCE concentration based upon the volatile 14C activity (Figure 10.7), rose
throughout the experiment.
Little or no TCE degradation occurred in columns 3 and 5 when operated initially with 4.5 mg/l
methane. Subsequently, when the methane was reduced to 1.5 mg/l, degradation began in both
columns (Figures 10.8 and 10.9).
The high methane feed to column 4 was pulsed after 79 days, and TCE degradation began (Figure
10.10). When the pulsed methane concentration was 4.5 mg/l, the extent of degradation clearly was
related to the pulsing, and more degradation occurred during the weeks in which no methane was fed to
the column. Later, when the methane concentration was reduced to 1.5 mg/l and pulsed, the difference
in the extent of TCE degradation between one week and the next was less evident. Furthermore, the
extent of degradation here was the highest achieved (about 4 u.g/1).
The Effect of Nutrient Supplements on TCE Degradation-
Columns 3 and 5 were operated identically throughout the entire experiment, except that nutrients
were added only to the former column (Figures 10.8 and 10.9). A permutation test (Bowker and
Lieberman, 1972) indicated that the TCE degradation was statistically higher in column 3 than in column 5
(99%, one-tailed test). However, this test was insufficient to prove that the nutrient addition was
responsible. It is possible that other factors, such as differences in the nature of the aquifer materials or
microbiaf cultures in the columns, was the cause.
133
-------
6-
5
4
3-
2"
1-
o-
-1
TCE
Addition
Stopped
0 50 100 150 200 250
Time (Days)
300 350
Figure 10.6. TCE degradation in column 6 (low methane plus nutrients).
0 50 100 150 200 250 300 350
Time (Days)
Figure 10.7. TCE concentration in the initial effluent from column 6 (low methane plus nutrients).
134
-------
50 100 150 200 250
Figure 10.8. TCE degradation in column 3 (high/low methane plus nutrients).
50 100 150 200 250
Time (Days)
Figure 10.9. TCE degradation in column 5 (high/low methane no nutrients).
135
-------
Pulsed with
[CH4] = 4.5mg/l
Pulsed with
[CH4]»1.5 mg/l
0 50 100 150 200 250 300 350
Time Pays)
Figure 10.10. TCE degradation in column 4 (pulse high/low methane plus nutrients).
Time Studies--
Experiments were conducted with column 6 to determine the approximate rates of oxygen con-
sumption, methane utilization, and TCE degradation. One-, two-, and three-day exchange intervals were
used. Utilization of DO, methane, and TCE appeared complete within the first one to two days after an
exchange and was similar to that at the end of a seven-day exchange period. It would appear from these
results that when methane was completely consumed, TCE utilization terminated. However, in the
pulsing studies, TCE degradation occurred following an exchange with only dissolved oxygen in the feed
solution. Other factors which are not yet understood appear to be involved.
1,2-DCA Degradation-
The 1,2-DCA degradation (Figure 10.11) began sooner than TCE degradation, and the extent of
degradation was higher. However, the influent 1,2-DCA concentration (80 u.g/1) was also significantly
higher than that of TCE (25 u,g/l). The percentage degradations based upon the influent concentration
for TCE and 1,2-DCA are shown in Figure 10.12.
VC Degradation-
After completion of the 1,2-DCA and TCE experiments, columns 5 and 7 were used to study VC
degradation. Weekly feeding of the columns with methane (1.5 mg/l) and oxygen continued while VC
was introduced at influent concentrations of approximately 125 u.g/1. An additional column, prepared
similar to the other columns, was employed as a control column by feeding it only oxygen and vinyl
chloride.
In columns 5 and 7 only trace levels of VC (< 5 u.g/1) remained after the one-week exchange
period, while approximately 25 u.g/1 VC remained in the control column (Figure 10.13). The presence of
VC in the initial effluent of the control column and its absence from the initial effluents in columns 5 and 7
indicate biodegradation of the VC by the enriched methanotrophic cultures.
136
-------
•a
35
30
25
20
15
10
5-
Degradation
Init Eff Cone
0 50 100 150 200
Time (Days)
Figure 10.11. The 1,2-DCA degradation and concentration in the initial effluent from column 7
(1,2-DCA: low methane plus nutrients).
40
30-
20-
10
Mr
d
V/ .
0 50 100 150 200 250 300
Time (Days)
Figure 10.12. Percent degradation for TCE and 1,2-DCA in columns 6 and 7, respectively (low methane
plus nutrients for both TCE and 1,2-DCA).
137
-------
120
100-
o
O
UJ
15
B Control
• Column 5
246
Time Between Exchanges (days)
Figure 10.13. Vinyl chloride removal efficiency in biostimulated column versus control column.
The time between exchanges was varied from 3 hrs to 1 week to determine the minimum time
necessary for complete VC removal from columns 5 and 7. VC was removed at about one-half the rate of
methane utilization. The methane in columns 5 and 7 was utilized within 1 day of feeding, while VC was
found to remain for approximately 2 days (Figure 10.14). The presence of methane did not seem to inhibit
VC transformation, as most of the VC was removed in the first day when methane concentrations were
highest.
Mass Balances of TCE and 1,2-DCA-
The pore water concentrations of TCE and 1,2-DCA (Figures 10.7 and 10.12) always were lower
than the influent concentrations. Conversion to CO2 accounts for some of the net loss of TCE and 1,2-
DCA from the pore water, but other processes, such as sorption and incorporation into biomass, probably
were involved as well.
An approximate 14C mass balance for TCE and 1,2-DCA was made by dividing the total 14C activity
in the initial effluent by the total 14C in the influent. Even after nearly one year of operation with TCE, less
than 50% of the influent 14C was present in the initial effluent. A better mass balance was achieved with
1,2-DCA (70%). It is apparent from this and other studies that 1,2-DCA did not sorb as strongly to the
aquifer solids as the TCE. In order to make better quantitative mass balances, the extents of sorption to
the solids and incorporation into the biomass need to be evaluated.
Degradation of TCE That Had Been Sorbed-
The degradation of TCE that had been previously sorbed to the aquifer material was investigated in
column 6 by discontinuing the addition of TCE, but continuing the exchange of the column fluid. Here,
14C activity detected in either the initial or the final effluent sample resulted from 14C-labeled TCE or
intermediate products that had desorbed from the solids or from biological solids decomposition. The
results, assuming that all of the 14C-nonvolatile compounds and 14CO2 in the initial effluent resulted from
TCE degradation between exchanges, are shown in Figure 10.6.
138
-------
CO
a
•
u
c
o
o
HI
"a
VC as %
CH4
20-
0 1 2
Time Between Exchanges (days)
Figure 10.14. Methane utilization versus VC removal in a methanotroph enriched column.
About 55% of the initial effluent 14C was in the form of nonvolatile compounds plus CC>2 and the
remainder was TCE (Table 10.2). Interestingly, the initial effluent concentration of the 14C nonvolatile plus
CC>2 fractions was about the same after the"*C addition was stopped as when it was added, but the TCE
concentration was lower (Figure 10.7).TCE
TABLE 10.2. THE DISTRIBUTION OF 14C IN THE INITIAL EFFLUENT REMOVED FROM COLUMN 6
AFTER 14TCE ADDITION TO THE INFLUENT WAS STOPPED3
Number of Exchange
'Without TCE Addition
1
2
3
4
5
6
CO2
(WO
5.52
5.01
4.05
4.76
4.63
3.69
Nonvolatile
Products
(MQ/1)
0.708
0.430
0.557
0.607
0.531
0.430
C02 +
Nonvolatile
Products
(ug/l)
6.22
5.44
4.61
5.37
5.16
4.12
TCE
(ug/i)
4.98
4.76
4.3
3.36
3.14
4.48
Total Removed
as CO2 + Nonvolatile
Products + TCE
(ug/l)
11.2
10.2
8.90
8.72
8.30
8.60
Average
4.61
0.544
5.15
4.17
9.32
aThe results are expressed as u,g/l-TCE.
139
-------
Distribution in the Columns--
The lower portion of the columns, which was nearer the influent port, was exposed to a higher TCE
concentration during the column fluid exchange than the upper portion because some sorption occurred
during the exchange. To illustrate this, the control column was operated for the last four exchanges in a
downflow direction. The TCE concentration in the pore water near the lower port, as measured by the first
18 ml of effluent was significantly higher (Figure 10.5), but the other parameters, such as dissolved
oxygen, did not change appreciably. This uneven TCE distribution perhaps affected to some degree the
mass balances and the measured extent of degradation. This is a difficult problem to overcome in column
studies of this type.
Additional Column Studies on TCE Degradation and Peroxide Addition
Approach--
Six glass columns were packed with completely mixed aquifer material for this study. Each of the
columns received 50 u.g/1 of 1^C-labeled TCE. Four of the columns were fed dissolved methane and a
source of oxygen. Two of these columns were operated with filtered groundwater containing 5 mg/l
dissolved methane and 22 mg/l dissolved oxygen - sufficient to maintain aerobic conditions following
oxidation of all the methane. An additional two columns received hydrogen peroxide as a source of
dissolved oxygen to permit a higher concentration of dissolved methane in the feed solution. The
concentration of this mixture was set at 10 mg/l methane and 85 mg/l peroxide (equivalent to 40 mg/l
dissolved oxygen). A gas phase formed in the columns when a greater concentration of hydrogen
peroxide was added. One of the control columns received a solution of dissolved oxygen and TCE but no
methane. The other control column was established as a control for the methane/peroxide treatment by
omitting the methane from the influent solution. The feed conditions are summarized in Table 10.3.
The laboratory columns were operated as sequential batch reactors, as previously discussed. To
test the effect of the feeding schedule on the bacteria population and TCE degradation, exchanges were
performed at intervals of 3 days, one week, or two weeks.
Breakthrough curves using bromide, dissolved oxygen and dissolved methane as tracers showed
that the first 25 ml of effluent were unmixed with the fresh influent solution and could be considered a
representative effluent sample. The effluent sample was analyzed for methane concentration (gas
chromatography), dissolved oxygen (dissolved oxygen probe), number of bacteria [acridine orange direct
count (Ghiorse and Balkwill, 1983)], TCE and other1 ^-labeled compounds (scintillation counting and gas
chromatography).
TABLE 10.3. FEED CONDITIONS FOR PEROXIDE STUDY
Concentration in
Column
Number
P1
P2
P3
P4
P5
P6
CH4
5
5
10
10
—
—
02
22
22
-
—
22
-
Batch Feed (mg/l)
H2O2
—
85
85
—
85
TCE
0.05
0.05
0.05
0.05
0.05
0.05
140
-------
the methane concentration decreased after
Methane and Dissolved Oxygen Utilization-
The presence of methanotrophs in the aquifer material was demonstrated within several weeks of
the commencement of methane and dissolved oxygen treatment. Operating with a 3-day residence time,
each exchange. After the third 3-day treatment period, the
methane concentration had fallen by two orders of magnitude, from 5 mg/l to 0.05 mg/l. Complete
methane consumption (below the 0.01 mg/l detection limit) occurred within three days for all subsequent
conditions, except for the experiments using hydrogen peroxide as the oxygen source.
Dissolved oxygen concentration declined in stoichiometric ratios with the methane utilization. The
minimum dissolved oxygen concentration approached the sensitivity limit of the measurement system,
but remained in the 0.5 to 1 mg/l range. The control column, with no methane added, exhibited an
oxygen demand as well. When the time between pore solution exchanges was lengthened to two weeks,
dissolved oxygen dropped from 30 mg/l to 20 mg/l, which was still above the concentration at equilibrium
Figure 10.15).
with the atmosphere (approximately 8 mg/l)
The addition of hydrogen peroxide as
At least 40 mg/l of hydrogen peroxide reached
exchange, exposing all portions of the system
completely degraded to oxygen within seve
activity of the methanotrophic bacteria. More
Since the dissolved oxygen concentration
peroxide remained, it was thought tha
methanotrophs for oxygen (Figure 10.15).
10 mg/l to 7 mg/l should have had the effect
but the residual methane remained in the 1
with heterotrophic bacteria may be combined
methanotrophic activity.
Methane
the source of oxygen partially inhibited methane utilization.
the effluent port of the columns during each solution
to this strong oxidant. Although the peroxide had
ral hours, the high level of oxidant may have disrupted the
than 1 mg/l of methane remained after the treatment period.
lad dropped to between 1 and 2 mg/l and no measurable
the non-methanotrophs could be out-competing the
Decreasing the methane concentration in the influent from
of increasing the oxygen available to the methanotrophs,
to 2 mg/l range. It seems likely that competition for oxygen
with some toxic effect of hydrogen peroxide to limit
Oxygen
Oxygen
(Control Column)
Consumed
(1008-5.2mg/i) (1008 - 23mg/L) (1008 = 30mg/L)
Dissolved Oxygen as Oxygen Source
Methane
Figure 10.15.
(100X-9.4mg/l)
Hydrogen
Effect of hydrogen peroxide
soil columns.
Oxygen Oxygen
(Control Column)
Consumed
(1007. = 38mg/L) (1007. = 38mg/L)
Peroxide as Oxygen Source
methane and oxygen consumption in batch exchange
141
-------
TCE Transformation-
After three months of operating the methane/dissolved oxygen columns, unequivocal evidence of
TCE transformation was established. The added TCE was strongly sorbed to the aquifer solids so that only
half of the labeled TCE could be recovered in the effluent after the treatment period. From the first
exchange period, 14C-labeled CO2 was detected in the effluent of both the methane-fed columns and
the control columns. However, only in the methane-fed columns did the activity of the labeled CO2 in the
effluent surpass the activity added in the form of the non-volatile contaminant in the TCE solution (Figure
10.16). After nearly one year of operation at two-week exchange intervals, the methanotrophic bacteria
transformed 15 to 20% of the 14C-labeled TCE to 14C-labeled CO2, while there was no evidence of TCE
transformation in the control column. Labeled carbon that had been incorporated into biomass could
conceivably double the measured transformation, but confirmation of the rate of incorporation awaits
destructive sampling of the columns.
No transformation products other than CO? could be detected. The only 14C-labeled components
in the effluent were TCE, CC>2 species and a negligibly small amount of non-volatile compounds.
The columns using hydrogen peroxide, rather than oxygenated water, as a source of oxygen
received 14C-TCE for over three months. Degradation of TCE remained less than 5% under these
conditions, so the addition of hydrogen peroxide was discontinued.
Reaction Time--
By decreasing the time between exchanges, more primary substrate (methane) is made available to
the methanotrophs and theoretically a greater population could be maintained. The three-day reaction
period was sufficient for complete methane utilization, but 14C-labeled CO2 production remained at 15 to
20%. For each of the reaction periods tested, the experiments were continued for at least twelve
exchanges to allow sufficient time for a new equilibrium to be established.
14C-TCE Recovered
Remaining in Column
COLUMNS RECEIVING METHANE
(Influent Concentration = 10075)
MC-TCE Recovered
Remaining in Column
CONTROL COLUMN
Figure 10.16. Mass balance of 14C-labeled TCE in batch exchange soil columns with fourteen days
between exchanges of pore liquid.
142
-------
Bacterial Population-
Bacterial population in the effluent samples has been examined by direct microscopic counting,
using epifluorescence of acridine orange-stained samples under blue light. This method does not
distinguish methanotrophic from heterotrophic bacteria, nor does it provide direct information on the
bacteria attached to aquifer solids.
Bacteria in the effluent from the methane-fed columns averaged approximately 8x105 bacteria per
milliliter, with substantial variability even in replicate samples. The control column tended to have slightly
fewer bacteria in the effluent but the difference was not statistically significant. The shorter time between
exchanges did not noticeably affect the numbers of bacteria in the effluent. Populations in effluent from
the hydrogen peroxide columns fell in the same range as populations from the columns using oxygenated
water.
Batch Microcosm Results
In the first batch microcosm experiment, methane was slowly consumed for the first five days,
followed by a relatively rapid and complete utilization by the sixth day. No TCE degradation was observed
until after all methane had been depleted. However, TCE degradation (as measured by labeled carbon
dioxide production) continued for approximately 5 days, until 6% of the initial TCE had been degraded to
carbon dioxide.
The mean direct count of the final bacterial population in the methane-fed microcosms was
approximately 106 organisms per ml, about twice the average counted in the initial samples. Although
triplicate samples were analyzed for bacteria using acridine orange direct counting and protein
measurements, the variability remained too high to statistically distinguish the final population numbers
from either the initial population or the population in the control microcosms which received no methane.
The bacterial population estimated in sequential batch aquifer material columns (the source of inoculum
for the microcosms) was approximately twice the number of the final microcosm populations. This seems
to provide a likely explanation for the diminished TCE degradation.
In the second experiment using 35-ml vials, slow methane utilization continued for only two days
before a 24-hour period of rapid consumption. After all the methane had been consumed, a 5-day period
of TCE degradation began. Only 3% of the initial TCE had been recovered as carbon dioxide by the end
of the experiment. Results of bacterial population observations were nearly identical to those of the first
microcosm experiment.
The temporal separation of methane utilization and TCE degradation was one of the more intriguing
results of these experiments. This may be due to insufficient biomass for measurable TCE degradation
until the period of rapid growth and increase of the population. Another likely explanation is the effect of
competition between methane and TCE for the active sites on the monooxygenase enzyme.
Another important observation was the continuation of TCE degradation for several days after the
primary substrate had been consumed. One conclusion would be that the methane monooxygenase
enzyme remains active for some time after methane is depleted, utilizing some stored reducing power
(Henry et al.,1988). A potential method for enhancing TCE degradation, besides increasing the
population of methanotrophic bacteria, is to support the activity of the bacteria during the period
immediately following methane depletion. Building on the results of recent field experiments at Moffett
Field (Section 6), it may be possible to supply additional reducing power to the methanotrophic bacteria
without competing for the methane monooxygenase enzyme.
DISCUSSION
Environmental factors that contribute to the persistence of a pollutant include the lack of a
degrading population and substrate concentrations so low that sufficient energy for microbial growth can-
not be obtained. Halogenated aliphatic compounds, such as TCE , 1,2-DCA, and VC, often are present at
143
-------
very low concentrations in groundwater and therefore may not be able themselves to support the growth
of an active transforming population. Furthermore, no organism has been identified yet that can obtain
energy for growth from ICE degradation. Transformations observed to occur are generally considered to
result from co-metabolism. In this manner, trace concentrations of TCE, 1,2-DCA, and VC were aerobically
biodegraded in the studies reported here when methane was used as a primary substrate.
If indigenous groundwater microorganisms cannot utilize a primary substrate which is supplied, then
population growth and organic decomposition cannot occur. The same holds true if the primary substrate
does not encourage the production of organisms with the necessary enzymes for co-metabolism. The
Moffett-site aquifer material did not appear to contain a propane-using microbial population, but did con-
tain a methane-using population with the capability to degrade TCE and 1,2-DCA. The latter has been
confirmed in the Moffett Field demonstration (Section 6), where the extent of TCE degradation was similar
to that found in this laboratory evaluation. This illustrates the value of column studies for evaluating the
potential for in-situ biodegradation.
The laboratory column studies were also useful for evaluating several important factors. In this
research the dissolved methane concentration had the greatest effect on the extent of TCE degradation.
Some methane was necessary to stimulate the growth of a TCE-degrading culture, but the methane did
not have to be added continuously, and TCE degradation occurred during pulsing intervals when only
oxygen was added. Methane monooxygenase (MMO) is probably responsible for the initial oxidation of
TCE (Little et al., 1988). However, in previous studies (Hou et al., 1979a; and Patel et al., 1982) methane
was shown to compete with other possible substrates for MMO activity. In this study, little or no TCE
degradation occurred when the added methane concentration was 4.5 mg/l, but did occur with 1.5 mg/l,
and this might have been due to the competition between methane and TCE for MMO. Even so, in the
columns with high methane concentration, it would have been expected that methane might have only
initially inhibited TCE degradation. Once methane was consumed, and the concentration became low
enough to no longer be inhibitory, not enough dissolved oxygen perhaps remained for TCE degradation
to occur. Therefore, both the dissolved methane and dissolved oxygen concentrations might have had
an effect on the extent of TCE degradation.
Column 6, which was fed low methane (1.5 mg/l) and nutrients throughout the experiment, was the
most successful in degrading TCE. Pulsing methane with oxygen followed by oxygen only was also an
effective treatment for stimulating TCE degradation. When the high methane (4.5 mg/l) was pulsed, more
degradation occurred during the weeks when no methane was added, perhaps because substrate
competition for MMO was reduced. When the pulsed methane concentration here was reduced (1.5
mg/l), the difference in the extent of degradation between exchanges was less apparent, and was also the
highest observed. Perhaps the larger population resulting from the higher initial methane concentration
additions was a factor in the subsequent higher rates.
Time studies with column 6 showed that the methane was consumed within one day and most of
the TCE degradation occurred within two days after an exchange. In the column that received pulsing, the
methane probably also was gone after one or two days. However, when only dissolved oxygen was
added to the feed solution the following week, the methanotrophic bacteria again initiated TCE degrada-
tion. Why TCE degradation stopped within one or two days after an exchange and began again when only
oxygen was supplied needs to be clarified. One possibility is that the oxygen concentration after two days
was insufficient for continued TCE utilization. More study is needed here.
The effect of the methane concentration on the extent of TCE degradation has important implica-
tions for in-situ treatment systems. These results suggest that the methane should not be present at too
high a concentration. In addition, methane does not have to be added continuously for degradation to
occur. Pulsing should help reduce the operating costs because less methane would be used. These
results , however, need to be confirmed with other cultures, as well as in actual field experiments where
operating conditions would necessarily be somewhat different than in this laboratory evaluation.
The three contaminants studied in this research behaved differently. VC was completely degraded,
whereas 1,2-DCA and TCE were partially degraded. The 1,2-DCA degradation began sooner and was
more complete than TCE degradation. The 1,2-DCA appeared to sorb less strongly to the aquifer solids,
144
-------
and more of the 1,2-DCA could be accounted for in a crude mass balance. In both the 1,2-DCA column
and the ICE columns, the ratio of the compound concentration remaining after degradation to the
concentration of the halogenated organic in the pore water was approximately the same. This was not the
case with VC degradation, where pore water concentrations fell below the detection limit (2 ug/l).
A limitation of this research is that better quantitative mass balances on ICE and 1,2-DCA were not
obtained. The concentrations of TCE and 1,2-DCA in the pore fluids always were significantly lower than
in the influent feed. Two possible sinks for the halogenated compounds are sorption and incorporation
into biomass. To the extent that degraded TCE and 1,2-DCA were incorporated into biomass, the
degradation reported in these studies has been underestimated. Furthermore, if the extent of degrada-
tion is proportional to the concentration of the contaminant, more degradation might have occurred in the
bottom portion of the columns, where a higher concentration of the halogenated organics appears to
have been present due to sorption. Another limitation, due to the lack of radiolabeled VC, was the inability
to determine the ultimate fate of the removed VC. It was uncertain how much of the removal of VC from
the control column could be attributed to biodegradation and how much was due to non-destructive fates,
such as sorption and volatile losses.
The study in which the degradation of TCE that had been sorted to the aquifer material was inves-
tigated was the most realistic in terms of simulating an in-situ treatment of a contaminated aquifer. Based
on the results of this limited study, twice as much TCE was removed from the column through the
combination of biodegradation and desorption as would have been removed by desorption alone during
passage of water through the column. These results suggest that the addition of methane and oxygen to
water circulated through a groundwater system to remove TCE for above ground removal could result in at
least a doubling of the rate of removal, and at the same time accomplish degradation of about one half of
the TCE removed. This encouraging observation should be pursued further.
Use of sealed microcosms permitted an accurate accounting of the total mass of methane, oxygen
and TCE in the systems. The precision of bacterial population measurements proved less satisfactory. No
TCE was degraded while detectable methane remained in the microcosms. While it is possible that this
was an effect of low population during the initial growth stages of the batch systems, the complete
absence of TCE degradation in the late stages of growth (when more than 50% of the methane had been
consumed) suggests that the presence of methane inhibits degradation. Population density does seem
to have an effect on the efficiency of TCE degradation, since the percentage of degradation was less
during the microcosm experiments than during sequential batch experiments which apparently support
higher populations. It was not possible to quantify the relationship between bacterial numbers and TCE
degradation in this study.
Temporal separation between methane consumption and TCE degradation was well documented in
the batch microcosm study. The methanotrophic community transformed TCE for several days after the
primary source of energy had been exhausted, implying the utilization of a stored energy source and the
continued viability of the monpoxygenase. This suggests the potential for a sequence of bacterial
activation with a methane solution followed by support of the TCE-degradation activity by an alternative
energy source which does not compete for the methane monooxygenase enzyme. The effect of adding
single-carbon compounds (methanol, formate, and/or formaldehyde) to the microcosms immediately after
methane has been consumed should be investigated further to assess whether the rate of the duration of
TCE degradation can be increased by supplying an alternative source of reducing power to an active
methanotrophic community.
CONCLUSIONS
The 20% transformation of TCE to CO2 under aerobic saturated conditions, with methane as the
primary substrate, offers encouragement for in-situ treatment of TCE-contaminated aquifers.
Microorganisms enriched from Moffett core material were able to degrade TCE, although they had been
enriched from an aquifer previously unexposed to halogenated alkenes. Laboratory studies performed
on carefully prepared columns of aquifer material permit evaluation of the potential of the indigenous
microbial population for biotransformation of the contaminant.
145
-------
The results suggest that approximately half of the TCE added to the columns may be sorted to the
aquifer solids during the treatment period. In the methane-fed columns, incorporation of labeled carbon
into biomass may account for a portion of the missing TCE.
A greater efficiency of TCE degradation was not obtained by increasing the rate of methane addition
from once every 14 days to once every 3 days, a fourfold increase. Further studies are needed in order to
explain this phenomenon.
Inhibition of methanotrophic activity resulted from the use of hydrogen peroxide as an oxygen
source in the feed solution. Less TCE was degraded under these conditions even though the dissolved
methane concentration of the influent could be doubled.
Sealed microcosm studies and some of the column studies indicated temporal separation of
utilization and TCE transformation, which suggests competition between CH4 and TCE for the MMO active
site, and greater affinity of the enzyme for CH4, resulting in competitive inhibition of TCE transformation in
the presence of detectable quantities of CH4.
146
-------
SECTION 11
CONTINUOUS FLOW COLUMN STUDIES
Robert Roat, Lewis Semprini, and Paul Roberts
INTRODUCTION
Experiments with a continuous flow column were undertaken to model the conditions of the field
tests. As a working hypothesis for these experiments, we aimed to prove that a continuous flow column,
filled with screened aquifer solids, may be used to simulate transient and steady-state biodegradation field
results for trichloroethylene(TCE) and trans-1 ,2-dichloroethylene (trans-DCE). The objectives of the
study were to determine:
1) the lag time required for biostimulation of methanotrophs as indicated by uptake of
methane;
2) the ratio of methane to oxygen use in the biostimulated column;
3) the degree of biodegradation of TCE and trans-DCE in a continuous flow laboratory
column;
4) the amount of TCE converted to CO2 using radiolabeled carbon (1 ^-TCE,
5) the relationship between methane uptake, TCE degradation, and CO2 production;
6) the transient response of the TCE and 14C-CO2 concentrations following stimulation of
a methanotrophic bacterial population;
7) the concentrations of any degradation intermediates and the relation between the
appearance of these intermediates and the onset of biotransformation of trans-DCE;
8) the effect of varying methane concentrations on degradation rates; and
9) the long-term effect of the shutdown of methane on concentration responses of TCE,
trans-DCE and their intermediates and degradation products following termination of
methane feed.
The use of a continuous flow column in the laboratory combines several advantages: 1) the con-
tinuous flow system has transport characteristics similar to those at the field site, allowing observation of
transient responses; 2) the continuous flow columns asymptotically approach steady-state effluent condi-
tions; and 3) the laboratory setting permits the detection of degradation products such as CO2 using
radiolabeled compounds. The first of these advantages is not shared by the semi-batch columns (Sec-
tion 10). The last advantage could not be realized in the field work (Section 6) since radiolabeled
compounds could not be used.
147
-------
METHODS
Column Description
A glass column packed with aquifer solids was used in the experiments. The column, shown in
Figure 11.1, was 40 cm long and 2.5 cm ID, with a ground-glass stopper at the top and a taper at the
bottom. Side sampling ports were affixed to the column to allow sample collection at intermediate
locations. The inflow and outflow ports, along with the side ports, were plugged with Teflon-coated butyl
rubber stoppers to minimize chemical losses.
The column was packed with aquifer material from the test zone at Moffett Field. Aseptic
procedures (Section 10) were used in obtaining the core samples. To eliminate particles larger than 4 mm,
the aquifer material was passed through a number 8 Tyler sieve. The column was packed using proce-
dures described in Section 10. This procedure elutriated any silt or clay which might clog the column.
After obtaining a clear effluent, a glass wool filter was placed at the level of the inflow port to evenly
distribute the influent solution over the top of the column. The ground-glass plug was then clamped in
place to prevent losses from the top of the column.
Setup
The column was configured for downward flow. A syringe pump (Orion Research, Cambridge, MA)
drove two 100-ml gas-tight syringes(Spectrum, Houston.TX) containing aquifer water amended with
chemicals of interest. In order to minimize release of radio-labeled TCE into the lab, the excess effluent
was collected in a Teflon gas collection bag (Alltech) mounted at the same height as the inflow syringe.
This physical arrangement also minimized the risk of inadvertent draining of the column during syringe
changes. To prevent chemical losses, 1/16"OD stainless steel tubing was used at the inflow, outflow and
sample port manifolds. For physical flexibility, inch-long 1/16" OD Teflon lines connected the syringes to
the stainless steel lines at the top of the column.
Analytical Techniques
The concentrations of dissolved oxygen, methane, TCE, trans-DCE, CO2, and trans-DCE epoxide
(DCO, Section 7) were measured. The analytic techniques used for each of these data sets are
mentioned here: a complete discussion of each of these methods may be found in Section 10. Sample
collection techniques are discussed below in detail, since they differ from those described in Section 10.
Dissolved oxygen (DO) was measured using an oxygen electrode (Yellowsprings Instruments,
model 54A). After meter calibration, 3 ml of water collected from the column were exposed to the probe
and allowed to reach equilibrium. Concentrations in the range of 0.5-29 mg/l were measured, although
the probe appeared to underestimate concentrations above 25 mg/l.
Methane concentrations were measured using head space analysis on a Hewlett-Packard 5730A
gas chromatograph with an FID detector and 60/80 Carbosieve column. Samples of approximately 2 ml of
column liquid were collected, stored and measured as described in Section 10.
TCE concentrations were determined by two techniques. Scintillation counting on a scintillation
spectrometer (Tricarb model 4530, Packard Instrument Co., Downers Grove.lL) was used for influent and
effluent column measurements of carbon-14 (14C) labeled TCE (see Section 10). In order to obtain nor-
malized TCE and CO2 concentrations from the activity data of the scintillation counter, three 1-ml samples
of column influent or effluent were required. One sample, designated "neutral", was placed directly in
10 ml of scintillation cocktail. The second, designated "acid", was injected into a vial containing 5 drops of
1N HCI, while the third, called "base" was injected into 5 drops of 1N NaOH. The acid and base samples
were purged with nitrogen for 10 minutes. 10 ml of scintillation cocktail were then added to each and the
scintillation count performed.
148
-------
Continuous Flow Column
inject
..Sample Port
Outflow
-------
Pentane extraction and analysis on a Tracer GC with a packed column and electron capture detector
(BCD) permitted correlation of scintillation counts with concentration in the source syringe and the
influent. The extraction and chromatography techniques are described more completely in Section 10.
Scintillation counting required 3 ml of sample, whereas pentane extraction required 14 ml.
Concentrations of 14C-CO2 also were determined using the scintillation spectrometer. Radiola-
beled carbon dioxide was assumed to be the product of TCE degradation, since TCE was the only signifi-
cant source of carbon-14. These samples were generated simultaneously with the TCE scintillation
counts.
Concentrations of trans-DCE were determined using a headspace analysis technique with the
Tracor ECD-GC and packed column. This technique, described in Section 10, required 2 ml of solution
from the column. Standard solutions of trans-DCE in water were prepared and analyzed to determine both
the residence time and a calibration curve for concentration versus peak area.
Sample Collection
As Figure 11.1 shows, samples could be collected at the influent, sideports, or effluent of the
column. The influent sampling protocol comprised three steps. First, the effluent valve near the Teflon
collection bag was closed. Next the influent sample port was opened and a 10-ml glass syringe was
attached to the port using swagelok fittings and a 2 cm, 1/16" OD Teflon line. After the syringe and Teflon
line were purged of trapped air, the pump speed was increased to quickly fill the syringe with liquid, after
which the speed was returned to normal. The syringe was then disconnected. To prevent the trapping of
air in the influent lines, the effluent valve was then reopened and the collection bag was squeezed to
force a few drops of column water and any entrained air bubbles out of the influent collection port. The
port was then recapped and normal flow continued.
Effluent samples were collected from the column in much the same manner. The effluent valve was
closed and a syringe attached to the effluent sample port. Following the purging of any air bubbles, the
syringe was allowed to fill at the normal 66 ml/day flow rate.
The solution collected in the glass syringe was used for three separate analyses. Aliquots of 1 ml
each were placed in three flasks containing either scintillation cocktail, 1N HCL, or 1N NaOH, for use in the
detection of TCE and CO2- If methane stimulation was underway, 2 ml were placed in a 14-ml
preweighed vial for use in methane analysis. Finally, a 3-ml sample was placed in the DO sampler and
tested. If trans-DCE samples were required, a second syringe was allowed to fill, then 2-ml samples were
added to a Teflon-capped 14-ml vial.
In order to obtain effluent samples for the pentane extraction ECD-GC analysis discussed above, a
flow-through collection loop was used. The sample collection method was adapted from that described in
Roberts and Mackay (1986). To collect a sample, the effluent valve was closed and an empty 14-ml
sample bottle clamped in line. The effluent sample port was then opened, the effluent valve reopened,
and the bottle was allowed to fill from the effluent collection bag. After the bottle had filled, the effluent
port was closed, and at least 2 volumes of column effluent were allowed to flow through the bottle. The
sample was then removed, capped and refrigerated. A new sample bottle was placed in line. This collec-
tion method allowed the relatively large sample sizes needed for the pentane analysis to be collected
without losses due to leakage of the glass collection syringe. It also allowed unattended overnight sample
collection.
Stock Solution Preparation
14C-trichloroethylene(14C-TCE) was obtained from the same source as Section 10. The original
stock solution of 14C-TCE had a specific activity of 4.11 x 104 dpm/u.g. The continuous flow studies
required the addition of unlabeled TCE, so as to increase the inflow concentration of TCE to field concen-
trations of about 40 u,g/l, while maintaining an ideal influent scintillation count of 1000 dpm/ml. This unla-
beled TCE spike reduced the specific activity of the stock solution to 2.55 x 104 dpm/u,g.
150
-------
In addition, experiments involving the use of unlabeled (trans-DCE) (Aldrich Chemical Co.) at an
inflow concentration of 800 \ig/\ required that trans-DCE also be added to the stock solution after final
adjustment of the TCE concentration.
This spiked stock solution was stored in a refrigerated gas-tight syringe and used as a source for the
influent syringes for the column. The stock solution allowed the maintenance of consistent inflow TCE
activity and trans-DCE concentration overtime.
Column Operation
The column was operated at a continuous flow rate of 66 ml/day with a continuous feed of TCE and
oxygen. This flow rate equals one pore volume (PV) per day through the column: it approximates the
residence time of tracer between the injection well and well S2 in the field study. In both the TCE and
trans-DCE experiments, the halogenated solvents were fed continuously to the column. Biostimulation
experiments were not started until the effluent and influent concentrations of the organic compounds had
reached equilibrium.
The inflow syringes held a total of 200 ml of Moffett site water amended with oxygen, TCE, and at
times methane and trans-DCE. The inflow was changed every third day. Proper concentrations of
methane and oxygen were added by the method described in Section 10, while TCE and later trans-DCE
were added by spiking the inflow syringes with small amounts of the concentrated stock solution. The
chemically-amended water was carefully mixed between the two inflow syringes to obtain equal concen-
trations in both syringes. The activity, DO concentration and methane concentration of the resulting
mixture were then sampled and analyzed. The use of two sets of inflow syringes allowed the new solution
to be prepared with minimum disruption of the column flow.
Column effluent was sampled on a regular basis, ranging from once every several hours during
periods of major transformation activity to once every other day.
RESULTS
Operation History
Table 11.1 is a chronological summary of the experiments performed with the continuous flow
system. The gap between the TCE addition and Biostim experiments represents a 4-month period during
which the stock solution and inflow concentration procedures were refined to improve the precision of
inflow concentration measurements. The period also represents the time required for the TCE effluent
and influent concentrations to reach equilibrium.
In general, the column was operated non-stop for the duration of the experiments. Two major
exceptions to this generality should be mentioned. The column was shut down for 25 days during the
methane variation experiment from 1/1/88 to 1/26/88. The column was also shut down for 28 days
between 7/20/88 and 8/17/88. In addition, several minor shutdowns of several hours to a day occurred
during occasional pump failures.
Column Transport Experiments: Tracer and TCE Addition
It was desired to operate the column with fluid residence times similar to those observed in the field.
The initial experiments, therefore, focussed on the determination of critical transport properties. These
experiments were performed prior to biostimulating the column. Tritium was assumed to be a conservative
tracer, while lissamine, an organic dye, also was tested to determine if it could serve as a non-sorbing
conservative tracer, in conjunction with radiolabeled TCE. After initial results were obtained with tritium
and lissamine, radiolabeled TCE addition was started.
151
-------
TABLE 11.1 EXPERIMENTS AND PROCESSES STUDIED
Experiment
Tracer
TCE
Addition
Biostim.-
Biotrans
Methane
Variation
Methane
Shutdown
trans-DCE
Addition
trans-DCE
Biostim
and
Biotrans.
Methane
Shutdown2
Duration
2/19/87-
2/23/87
2/23/87-
3/1/87
7/23/87-
11/17/87
11/17/87-
3/2/88
3/3/88-
6/28/88
6/28/88-
1 1/2/88
11/2/88-
12/10/88
12/10/88-
12/27/88
Pore Chemicals
Volume Injected
1-4 tritium
lissamine
(dye)
4-1 1 14C-TCE
02
120-190 14C-TCE
02
CH4
243-346 14C-TCE
02
CH4
346-466 14C-TCE
02
CH4
466-593 14C-TCE
02
CH4
trans-DCE
593-631 14C-TCE
02
CH4
trans-DCE
631-648 14C-TCE
02
CH4
trans-DCE
Average
Cone, (mg/l)
(95% Conf. Int.)
0.040+0.003
27.0±1.3
0.040±0.003
25.9±1.1
4.510.2
0.040±0.003
17.911.8
6.5±0.3
0.04010.003
17.911.3
0.0
0.04010.003
17.911.3
0.0
0.79610.057
0.04010.003
22.5+1.0
4.5+0.3
0.79610.057
0.04010.003
17.9+1.3
0.0
0.79610.057
Process
Studied
Transport,
dispersion.
Retardation,
tailing.
Biostimulation of native
methanotrophs,
Biotransformation of TCE
and production of CO2.
Effect of methane con-
centration variation on the
degree of degradation
and production of CO2.
Effect of removal of
methane as a primary
substrate on the
degradation of TCE.
trans-DCE uptake by
column with no
biostimulation.
Biostimulation of
methanotrophs and
biotransformation of
trans-DCE.
Effect of methane
shutdown on trans-
DCE degradation and
epoxide production.
Figure 11.2 shows the tracer experiment data, expressed as normalized concentration as a function
of the number of pore volumes eluted. Note that, although the TCE results are presented in the same
graph as those of the tritium and lissamine experiment, TCE addition actually occurred at a later time in
order to avoid confounding of tritium and TCE scintillation counts.
152
-------
(MOBILE - IMMOBILE ZONE MODEL)
Figure 11.2. Results of column transport experiments and model simulations.
As expected of a conservative tracer, tritium broke through most quickly, followed by lissamine and
ICE. However, the tritium data show exhibited substantial tailing, since the influent concentration was not
reached for four PV, although 80% of influent concentration was reached in only 1.2 PV. Lissamine
showed more severe tailing, as it reached only 62% of the inflow concentration in 1.2 PV. TCE shows a
high degree of tailing. In fact, the TCE data reached only 28% of inflow concentration in 1.2 PV before
the rate of increase in concentration slowed dramatically.
The tailing evident in Figure 11.2 is not consistent with the usual advection-dispersion model. The
large amount of tailing may result from immobile (or slow flow) zones in the column due to non-uniform
packing. The results were therefore compared to the mobile-immobile zone model of van Genuchten
(1981). Results of this modelling exercise are shown in Table 11.2. The results are plotted in Figure 11.2
as solid lines. The mobile-immobile zone model appears to describe the experimental data quite well.
TABLE 11.2. MODEL RESULTS: MOBILE-IMMOBILE ZONE MODEL
Peclet Number:
Mobile Zone:
Pore Volume:
Mass Transfer Coeff.:
Retardation Factor:
Tritium:
Lissamine:
TCE:
16
70% of pore volume
66ml
1
1.8
6
153
-------
The most important result of this experiment from an operational standpoint was the determination
of the pore volume. The mobile-immobile zone modeling, fitted to the tritium data, provided an estimate
of the pore volume. This finding allowed the flow rate to be set at 66 ml/day, or 1 pore volume per day, in
order to simulate flow conditions at the Moffett field site, which exhibited hydraulic residence times
between the injection and observation points that were on the order of one day (Section 5). Thus, in all of
the following graphs, units of pore volume and days may be used interchangeably.
Biostimulation Experiments (Biostim-Biotrans)
The first set of biostimulation experiments was performed to characterize the ability of indigenous
methanotrophic populations to transform TCE as a secondary substrate, with methane as the primary sub-
strate, under continuous flow conditions.
The objectives of the Biostim-Biotrans sequence of experiments were to 1) determine the lag time
between methane breakthrough and the onset of methane utilization; 2) measure the ratio of dissolved
oxygen to methane consumed; 3) observe the transient transformation of TCE and production of labeled
C02 in response to biostimulation; and 4) estimate the degree of TCE degradation after steady state
conditions were achieved.
Prior to the stimulation of the indigenous bacterial community, an extended period was allowed for
equilibration of the TCE concentration and stabilization of the column operation procedure. During this
120-day period, the column was operated at a flow rate of 66 ml/day, an influent TCE concentration of
40 u,g/l, and a dissolved oxygen concentration of 28-30 mg/l.
The biostimulation of methanotrophs was undertaken by introducing 4.5 mg/l of methane into the
influent syringes. Since this concentration was achieved by mixing ratios of methane- and oxygen-satu-
rated water, the influent concentration of dissolved oxygen was reduced to 22.5 mg/l. Inflow TCE con-
centration remained steady at 40 |o.g/l.
Figure 11.3 shows normalized column effluent concentrations for methane, as well as the amounts
of TCE transformed and radio-labeled CO2 produced. Methane concentrations were found directly using
an externally calibrated headspace technique for the HP Gas Chromatograph described in Section 10.
The TCE and CO2 data were obtained from the scintillation counting of carbon-14 activity described
above. This activity (in units of disintegrations per minute(DPM)) has three components: 14C associated
with TCE; 14C associated with CO2; and 14C associated with a non-volatile contaminant in the TCE source.
As mentioned in the methods section, the three samples measured were designated neutral, base and
acid. The neutral fraction corresponds to the total carbon-14 activity in the 1-ml fluid sample. The base
fraction provided the activity of the labeled CO2 and the non-volatile contaminant, while the acid fraction
contributed scintillation counts for the non-volatile component alone. The 14C-TCE activity was defined,
therefore, as the difference between neutral and base counts; 14C-CO2 activity was calculated as the dif-
ference between base and acid counts.
Figure 11.3 presents data termed "TCE Transformed"," Net CO2 Produced" and "Effluent CH4".
The normalized "TCE Transformed" values represent the difference between the average inflow 14C-TCE
activity and the effluent 14TCE activity, all divided by the average inflow TCE activity: ( -
TCEout)/. The normalized "Net CO2 Produced" data are defined by the difference between the
effluent 14CO2 activity and the average background effluent 14CO2 activity, all divided by the average
14TCE inflow activity: (CO2o - ) / . Background effluent 14CO2 is defined as
the quantity of radiolabeled CO2 producedTprior to biostimulation. This CO2 production, which averages
8% of the inflow concentration, is attributed to the mineralization of the non-volatile fraction in the stock
TCE. The background 14CO2 production is factored out of the data in Figure 11.3 by the preceding
equation.
154
-------
1.1
1 -
0.9 -
0.8 -
0.7 -
0.6 -
0A -
0.4 -
0.3 -
0.2 -
0.1 -
CH4 STARTUP
I X
. XX'
. X
k*
x n
x x
x
X x X EFFLUENT CH4
X
X
X
X
X
X
X
TCE TRANSFORMED
n D D
D
00 0°* NET C02 PRODUCED
OOo-M
120
140 160
Doy( POT* Volumes)
180
Figure 11.3. The results of the first biostimulation-biotransformation experiment.
Methane addition began at PV 127, i.e. on day 127 of column operation. As the first 6 data points
of Figure 11.3 show, neither TCE transformation nor net CO2 production was observed between the time
at which TCE saturation was attained and the time at which methane utilization commenced. Methane
breakthrough to the extent of 80% of the inflow concentration occurred within 2 PV. Following
breakthrough, the normalized methane concentration fluctuated within the range of 0.7 to 1.1, showing
essentially no uptake in the column for 20 days. After PV 147, methane utilization was observed. Effluent
concentrations rapidly declined below a normalized value of 0.7 and were below detection levels (0.1
mg/l) within 5 days (PV 152).
Prior to PV 152, virtually no TCE was degraded between column inflow and effluent. Over the same
period, no net radiolabeled CO2 was produced. Not until all of the methane had been consumed did
degradation of TCE and production of CO2 in excess of the contaminant fraction begin. TCE transforma-
tion began immediately following the complete consumption of methane. Carbon dioxide production
lagged behind TCE transformation, suggesting that some carbon from the TCE may have been incorpo-
rated into cell mass or intermediates before complete mineralization was attained.
Between PV 152 and 184, the average TCE transformation shown in Figure 11.3 reached 25.5%.
CC-2 gradually climbed to a high of 15% (average 12%, PV 160-184), then dropped below 10% after PV
184. The discrepancy between TCE disappearance and CC-2 production suggests that approximately
one-half the labelled carbon was not accounted for, at least in the short term. It is conceivable that some of
the missing 14C was stored in the column in the form of carbonate, possibly as a result of precipitation or
isotope exchange with pre-existing carbonate. It was not feasible to test this hypothesis by analyzing for
14C in the solids, since the column would have to be destroyed.
Figure 11.4 demonstrates the relationship observed between the consumption of methane and
oxygen as the biostimulation commenced and reached steady state. The steady state is defined here as
a prolonged period in which influent conditions were maintained constant and the effluent methane
155
-------
120
160
MY(POI« VOJUME)
180
Figure 11.4. The net methane and DO consumption in the biostimulation experiment.
concentration did not change appreciably over time, e.g., beginning at PV 150, in Figure 11.4. Data were
collected for inflow and effluent samples over the course of the experiment. Figure 11.4 presents these
data for CH4 consumption, as the difference between inflow and effluent concentrations(CH4
Consumption = CH4 inf|OW - CH4 eff|uent) and for biological DO demand, calculated as the difference
between the average background effluent DO concentration prior to biostimulation and the individual
effluent DO values (DO Demand = ( - DOeffiuent)- Note tnat although the
same methane data are portrayed in Figures 11.3' and 11.4, the presentation of those data differs. The
methane information in Figure 11.3 is presented as the difference between inflow and effluent
concentrations rather than as the ratio of effluent to inflow concentrations. This alternative presentation
method allows direct visual comparison of the ratio of methane consumption to oxygen demand.
The DO values prior to methane addition (before PV 127) indicate a 12 mg/l oxygen demand in the
column, independent of the biostimulation of methanotrophs. Abiotic oxidation reactions may explain this
demand.
Coincident with the uptake of methane between day PV and 152, the DO demand in the column
increased. Figure 11.4 shows a rapid increase in biological oxygen demand at PV 148, from 0 mg/l to
9.0 mg/l, followed by a more gradual increase between PV 152 and 167 to 12.7 mg/l. The demand then
decreased slowly to 6.0 mg/l at PV 178. The change from rapid to gradual increase in oxygen demand
coincided approximately with the complete utilization of methane at PV 152. This behavior seems
consistent with a rapid uptake due to methane utilization by a growing population of methane utilizing
bacteria, followed by a slow increase in demand due to cell biomass decay.
Recalling that the inflow methane concentration averaged 4.5 mg/l, the biostimulated column
required about 12.7 mg/l O2 (Figure 11.4). As the stoichiometry of the balanced equation for complete
methane mineralization
CH4 + 2O2 -> CO2 + 2H2O
156
-------
implies, 2 moles (64 g) of O2 are theoretically required to completely convert 1 mole (16 g) of CH4 to
CO2. On a mass basis this implies an ideal ratio of 4:1 C^CH^. The O2:CH4 consumption ratio inferred
from the experimental data is 12.7:4.5 or 2.82:1. The tower DO demand compared to the stoichiometric
ratio for complete oxidation, probably is explained by cell growth and incorporation of methane into cell
mass as a carbon source. It also was observed that the DO probe underestimated values of oxygen
concentration when the concentration exceeded 20 mg/l, causing a negative bias in the DO demand.
Methane Variation/ Methane Shutdown Experiments
The objectives of the transient methane variation experiments were to: 1) determine the effect of
methane variations on the steady state percent of degradation of TCE and production of CO2; and
2) observe the transient effects of the methane variation on the TCE degradation and CO2 production.
After allowing the transformation to equilibrate for 90 PV after the onset of methane utilization, the
inflow methane was first increased from 4.5 to 6.5 mg/l (PV 242, Figures 11.5 and 11.6). The increased
methane feed was continued until PV 272, when the column flow was shut down completely for 44 days.
Subsequently, flow was restarted with a methane feed concentration of 6.5 mg/l and continued through
PV 299. At that time the inflow methane concentration was reduced to 0 mg/l. The cessation of the
methane feed continued to the end of the experiment at PV 346 and for an additional 120 days. The 120
day operation without methane feed allowed the TCE transformation in the column to stop completely.
Figures 11.5 and 11.6 show the normalized effluent TCE concentrations and CO2 concentration
history, respectively. The data are presented as the ratio of the instantaneous value of the respective
parameters to the average TCE inflow concentration. The CO2 data were adjusted to remove the
contribution of the degraded non-volatile contaminant fraction.
No trace of the increased inflow methane concentration was ever detected in the effluent methane
analysis. DO demand decreased from 16.7 to 11.9 mg/l. As Figure 11.5 shows, the increase had virtually
no effect on the effluent TCE concentration. TCE transformation remained at about 23%, and, as Figure
11.6 indicates, CO2 production stayed relatively constant.
The 44-day column shutdown, which is represented as the break at PV 272 in Figures 11.5 and
11.6, caused a temporarily high TCE transformation and CO2 production. These values quickly returned
to the averages seen before shutdown. The reduced TCE concentrations might have also resulted from
slow sorptive uptake during the 44-day shutdown.
Elimination of inflow methane caused a more pronounced change. With no methane present in the
inflow, TCE transformation increased slightly to a maximum of 34.8% during the subsequent 20 days
(days 303-323, Figure 11.5). Following that time, the amount of degradation began to decrease. Within
70 days after methane shutdown, the TCE degradation ceased and the effluent TCE concentration again
equaled the inflow concentration. CO2 production changed more substantially (Figure 11.6). Immediately
following the cessation of methane, the production of radiolabeled CO2 doubled, reaching 14% of the
influent radiolabeled TCE (average, PV 303-313). This increase lasted for 10 days, then decreased over
the next 30 days to background levels.
Biostimulation Experiments with trans-DCE
The objectives of the trans-DCE Biostim/Biotrans experiments were to 1) determine the time
required to restimulate the methanotrophic population; 2) quantify the steady-state transformation of
trans-DCE in the column; 3) observe the production of degradation intermediates; and 4) assess the
effect of methane shutdown on the degradation of trans-DCE.
In view of the field experiment results for the second season, presented in Section 6, a higher
degree of degradation was expected for trans-DCE than for TCE. In addition, formation of an epoxide
intermediate was anticipated. Field results also raised the expectation of rapid restimulation of bacteria
and rapid shutdown of trans-DCE degradation following shutdown of methane.
157
-------
0.8 -
0.7 -
0.6-
OJt -
0.4 -
0.3 -
0.2 -
0.1 -
n -
4.5 mg/l
CH4
D D
nrP
cr^i c
Q3Q
n
6.5 mg/l
CH4
1
I
n
"
bo
on
D
D C
D
< — COLUMN
fa SHUTDOWN
°
oa
00 £
n Q
LJ
Q Cr ^J~*
a D
i
0 mg/l
CH4
tn a
a
T>6 DO ff
tH^jp D u
tj
D
D C
EFFLUENT n
D
rn a
D
Tl
220 240 260 280 300
PORE VOLUMES EXCHANGED
320
340
Figure 11.5. Fractional transformation of TCE during methane variation experiments.
C02 Produc«d/TCE Inflow
o^t* -
0.22 -
0 .2 -
0.18 -
0.16-
0.14 -
0.12-
0.1 -
0.08 -
0.06 -
0.04-
0.02 -
£
4.5 mg/l
CH4
D
D
D D
a n
a
D
n
6.5 mg/l
CH4
[
n
]
a
DD Q
a a j-p
D
< — COLUMN
SHUTDOWN
D
j
a
D '
D DD
D
D
0 mg/l
CH4
D
DD_n
D
ODD
D a
a a
3
D
n
D
I r — ' — T i i i r ii i i i
20 240 260 280 300 320 340
PORE VOLUMES EXCHANGED
Figure 11.6. Labeled CO2 produced during methane variation experiments.
158
-------
For this final experiment, unlabeled trans-DCE was added to the column inflow in addition to labeled
TCE. The trans-DCE addition began following the cessation of TCE degradation in the methane
shutdown experiments. During the start-up and stabilization of trans-DCE concentration in the column,
the operating conditions were as follows: trans-DCE, 796±57 u.g/1; TCE, 40±0.3 u.g/1; and dissolved
oxygen, 17.9±1.3 mg/l. After a 120-day stabilization period, methane addition was restarted.
Field results had indicated immediate degradation of trans-DCE upon reintroduction of methane.
Instead, degradation did not become observable for 7 days( Figure 11.7). After 18 days of stimulation
(Day 18), 85% of the influent trans-DCE was being degraded. At day 20, the methane was once again
shut down; despite the absence of primary substrate, the degradation of trans-DCE continued
unaffected, remaining in the 80-90% range, for 30 days. Only after day 50 did the degradation of trans-
DCE decrease.
Degradation of trans-DCE coincided with the production of an intermediate suspected to be trans-
DCE epoxide( Figure 11.8). The identification of the unknown intermediate resulted from the sudden
appearance of a new peak in the gas chromatograms for the effluent, beginning at approximately 10 pore
volumes. The peak residence time was 3.9 minutes, or 1.4 minutes after the trans-DCE peak and 4.2
minutes before the TCE peak. This finding corresponded to the residence time observed in other labora-
tory work (Section 7).
Figure 11.8 shows the epoxide concentration history: the data have been converted from peak
areas to concentrations using a sensitivity of 150 times that of trans-DCE based on response factors given
in Section 7. Results show a maximum epoxide concentration of 40 u.g/1 during the period of methane
consumption(day 7 through 20), and a maximum concentration of 69 u.g/1 during the 30 days (days 20 to
50) before trans-DCE degradation shut down. The epoxide concentration represents approximately 10%
of the trans-DCE degraded.
SUMMARY
The experiments permitted a direct comparison with the field results, unlike the batch column
experiments(Section 10). Tables 11.3 and 11.4 summarize the major findings of the continuous-flow
experiments. The steady-state differences in concentrations (influent -effluent) of TCE and trans-DCE
following biostimulation agreed well with field results. The extent of biotransformation of TCE and trans-
DCE estimated from the continuous column experiments was 25.5% and 85%, respectively. Both field
and column work with trans-DCE showed the appearance of an intermediate thought to be trans-DCE
epoxide coincident with the disappearance of trans-DCE. Major differences between field and
continuous flow laboratory experiments were observed regarding the time required to restimulate the
bacterial population needed to degrade the trans-DCE. The column required a much longer restimulation
time, probably as a result of decay due to prolonged DO addition during the period between the
biostimulation experiments. In addition, the column experiment continued to show trans-DCE
transformation for several weeks after the cessation of methane addition, whereas in the field experiment
transformation began to decrease within a few days.
159
-------
1.2 -
1.1 -
1 -
0.9 -
0.8 -j
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
D
oDo
'if ff
DD 0 D
D
CJ,
n
D
D
%
n
Methane on
at PV=0
I
1
B
ED D
Methane off Q D
n ° DD o n
a ° a
20
40
60
^
1
Figure 11.7. Methane stimulation experiment with trans-DCE.
70
60 -
30 -
40 -
30 -
20 -
10 -
Methane on
at PV=0
D
0
D B
D
Methane off
20
40
PORE VOLUME
TD
a
60
Figure 11.8. trans-DCE epoxide concentration history.
160
-------
TABLE 11.3. TCE BIOSTIMULATION RESULTS
Methane
Cone.
4.5
4.5
6.5
0.0
TCE
Loss3
(%)
25.5±6.8
29.2±5.1
23.1±5.3
34.817.4
Net DO
Consumeda'b
(mg/l)
11.3±1.3
7.5±0.3
11.5±0.3
7.6±0.6
Net CO2
Produced3-6
(%)
12±3
7.1±0.9
3.311.4
14.2+1.3
Time Period0
152-184
184-240
240-299
303-323
a Mean 195% confidence limit.
b Adjusted by subtracting values measured before biostimulation.
c Expressed as an interval of days, or pore volumes.
TABLE 11.4. trans-DCE BIOSTIMULATION RESULTS
Methane
Cone.
4.5
0.0
trans-DCE
Loss3
(%)
84.019.0
85.413.4
Maximum
trans-DCE Epoxide
Produced
(ng/i)
40
69
Time Period5
0-23
23-50
3 Mean 195% confidence limit.
b Expressed as an interval of days, or pore volumes.
161
-------
SECTION 12
ONE-DIMENSIONAL SOLUTE TRANSPORT IN POROUS MEDIA WITH
WELL-TO-WELL RECIRCULATION
Constantinos V. Chrysikopoulos and Paul Roberts
This section summarizes the development, testing, and application of a mathematical model
adapted to account for the recycling of solute between an injection well and an extraction well. This kind
of coupling through recycle from the extraction well to the injection well is inherent to the experiments
conducted at the Moffett site, as described in Section 4.
BACKGROUND
The transport of nonreactive solutes through homogeneous porous media consisting of imper-
meable particles commonly has been characterized by the classical advection-dispersion equation, which
is based on the empirical relation of Pick's diffusion law (Fried and Combarnous, 1971; Bear, 1972). For
sorbing solutes, the advection-dispersion equation has been modified to incorporate the effects ol
adsorption (Hashimoto et al., 1964; Lindstrom et al., 1967) and hysteresis (van Genuchten et al., 1974)
However, in order to simulate the asymmetry and tailing of breakthrough curves observed in several
experimental studies of solute transport, first-order rate models have been developed to account for
solute exchange between zones of mobile and completely mixed immobile water (Coats and Smith,
1964; van Genuchten and Wierenga, 1976; Rao et al., 1980), while physical nonequilibrium models
incorporate solute transfer into immobile regions of various geometries using the second law of diffusion
(Rasmuson and Neretnieks, 1980; Goltz and Roberts, 1986a). Most mathematical models in current use
for simulating the transport of sorbing solutes are based on chemical equilibrium isotherms rather than on
kinetic sorption relationships, because of their computational simplicity. However, the validity of the local
chemical equilibrium assumption has been questioned in studies of sorbing solute transport through
laboratory columns (Nkedi-Kizza et al., 1983; Hutzler et al., 1986; Miller and Weber, 1986) and through
aquifers (Goltz and Roberts, 1986b; Roberts et al., 1986).
There are certain cases where direct application of the existing solute transport models is not
adequate. For example, the conditions of the transport experiments at the Moffett site, where a flow field
is induced between an injection-extraction well-pair and chemicals of interest are introduced into a fraction
of the extracted fluid which is reinjected, cannot be simulated accurately without taking into account the
feedback due to recirculation.
In order to represent realistically the field conditions, a semi-analytical solution and an approximate
analytical solution are derived for a model describing one-dimensional solute transport through porous
media under local equilibrium conditions and solute recirculation between the extraction-injection well-
pair. The solutions are developed for a flux-type inlet boundary condition in a semi-infinite medium.
Although one-dimensional models may not be adequate for every field situation, this particular model
provides a starting point for investigating the effects of well-to-well recirculation. Furthermore, for one-
dimensional models, analytical or semi-analytical solutions are more likely obtainable, leading to more
efficient and accurate computations than the purely numerical solutions of multi-dimensional models. The
model is applied to the field-data of experiments Tracers and Tracerl 1 (Section 5).
162
-------
MODEL FORMULATION AND SOLUTION
The transport of a sorbing solute through one-dimensional porous media under steady-state flow
conditions is governed by the following partial differential equation (Lapidus and Amundson, 1952):
p ac*(t,x) 82C(t,x) aC(t,x)
+e at -u~~~v ( ]
where C is the liquid-phase solute concentration (M/L3), C* is the solid-phase concentration of the
adsorbed solute (M/M), D is the hydrodynamic dispersion coefficient (L2/t), V is the average interstitial fluid
velocity (L/t), x is the spatial coordinate in the direction of flow (L), t is the time (t), p is the bulk density of
the solid matrix (M/L3), and 0 is the porosity (L3/L3). For linear, reversible, instantaneous sorption, the
equilibrium relationship between the solute substance in the aqueous and solid phases is given by
C*(t,x) = KdC(t,x) (12-2)
where Kd is the partition or distribution coefficient (L3/M). The distribution coefficient is a measure of
solute retention by the solid, and is expressed as the ratio of solute concentration on the adsorbent to
solute concentration in solution. Combining equations (12-1) and (12-2) leads to
ac(t,X)
v {12'3)
where the dimensionless variable R is the retardation factor defined as
R = 1 +£ Kd (12-4)
For a semi-infinite system and pulse input conditions that take into account the effect of solute
recirculation, the appropriate initial and boundary conditions are
C(0,x) = 0 (12-5a)
•qC(t,L)l 0tp
^=0 (12-5C)
where Cp is the pulse-type injection concentration (M/L3), and tp is the duration of the solute pulse (t).
The condition (12-5a) corresponds to the situation in which the solute is initially absent from the one-
dimensional porous medium. The third-or flux-type boundary condition (12-5b) for pulse injection
implies concentration discontinuity at the inlet and leads to material balance conservation (Brigham,1974;
Kreft and Zuber, 1978). The upstream boundary condition (12-5b) includes on the right hand side the
term qC(t,L), where C(t,L) is the solute concentration at the extraction location L, and q is the fraction of
the recirculating solute mass, 0 < q < 1. This term accounts for solute recirculation between the
extraction-injection locations by adjusting instantaneously the inlet concentration in proportion to the
163
-------
solute concentration at the exit. The downstream boundary condition (12-5c) preserves concentration
continuity for a semi-infinite system.
The solution to (12-3), subject to initial and boundary conditions described by equations (12-5a, b
and c) is obtained by taking the Laplace transforms of these equations with respect to the time variable t
and the space variable x. The details of the mathematical derivations have been reported by Chrysiko-
poulos et al. (1989). Here, only the solution to the well-to-well recirculation model is presented.
The semi-analytical solution in the Laplace domain is given by
Q(t,x)-Q(t-tp,x) t>tp
(12-ea)
where
-1 (C(s,x)}
(12-6b)
C(s,x, - -
sD(M-N)
D(N-M)
M = -55-
(12-6d)
D 4D2
s is the Laplace domain time variable and L"1 is the Laplace inverse operator. Numerical inversion of the
Laplace transform can be obtained by techniques such as the Stehfest algorithm (Stehfest,1970).
The approximate analytical solution is obtained by employing Maclaurin's approximation to simplify
equation (12-6c) so that its analytical inversion from Laplace time variable s to real time t is achievable,
and it is given by
(t,x)-<&(t-tp,x) t>tp
(12-7a)
where
= A(t,x) + £ am(t,x) * G(tJL)
m=1
164
-------
CCi(t,x) = A(t,x)
Otm(t,x) = Om.i(t,x)*G(t,L) m>2
(12-7C)
(12-7d)
A(t,x) =
.vl
D DR
JtDR/
exp -
(Rx - Vt)2
4DRt
2(DRt)
1/2
(12-7e)
0(0,) = q
r2 \
1/2
RU
exp -
(RL - Vt)'
4DRt
12DR)
exp
VLlErfc
D-l [2(DRt)
RL + Vt
xl/2
(12-71)
The "*" signifies convolution, and the nested convolution integrals are easily determined by numerical
integration techniques.
Figure 12.1 shows plots of dimensionless concentration versus time for the approximate analytical
solution (12-6), the semi-analytical solution (12-7), and the case of no recirculation (q = 0). The semi-
analytical solution was obtained by numerical inversion of the Laplace transform utilizing the Stehfest
algorithm (Stehfest, 1970), while the convolution integrals of the approximate analytical solution were
evaluated by Simpson's rule. For this illustrative comparison only the first three terms of the infinite series
(12-7b) were taken into account. The predictions of the analytical and semi-analytical solutions are
indistinguishable. The model without recirculation predicts a steady-state breakthrough concentration Cp,
while the recirculation model doesn't approach a steady state, but rather the predicted concentration
increases with injection time. For q «1, the approximate analytical solution can be utilized efficiently with
just a few terms of the infinite series, because at late stages of a broad pulse injection the breakthrough
concentration increases only very slowly. However, for q ~ 1 the number of the required summation terms
in the infinite series is determined by the number of injected fluid recirculations, or equivalently the
number of pore volumes passing through the system during the experimental period. For example, if q =
1, the breakthrough concentration at the end of k fluid recirculations is approximately kCp.
PARAMETER ESTIMATION METHODOLOGY
The model developed above contains four parameters, namely the retardation factor, the dispersion
coefficient, the fluid velocity, and the fraction of the recirculating solute mass. In applying the model to
laboratory or field data, it is necessary to best estimate these parameters and to quantify their uncertainty.
There are several approaches available for parameter determination. Here, the nonlinear least-squares
regression method has been adopted.
In general, the objective of the nonlinear least-squares method is to obtain estimates of the model
parameters which minimize the residual sum of squares between simulated and observed data. For this
work, the Standards Times Series and Regression Package (STARPAC) (Donaldson and Tryon, 1983)
was used for parameter estimation. STARPAC includes an adaptive nonlinear least-squares algorithm
developed by Dennis et al. (1981). This algorithm adaptively decides when to use the computationally
expensive second-order part of the least-squares Hessian, which accounts for its reliability and efficiency
when the residuals are large or the model is highly nonlinear.
165
-------
1.2
1.0
0.8
5" 0-6
0.4
0.2
0.0
Eq. 12-6
Eq. 12-7
q=0.0
20
40
60
80
Time (d)
Figure 12.1. Comparison of the approximate analytical, semi-analytical, and case of no recirculation solu-
tions: V m 0.5 m/d, D = 0.02
tp = 40.0 days, and q = 0.1.
m2/d, R = 1, Cp = 1.0 mg/l, x = 2.0 m, I =6.0 m,
The uncertainty of the estimated nonlinear parameters is quantified by approximate 95% confi-
dence intervals, which are based on the single-variate Student's t distribution, assuming normally distrib-
uted parameter estimates. This may be criticized because these limits do not yield joint confidence
regions. However, although the concept of confidence region construction is intuitively simple, it can lead
to considerable difficulties. The results of a Monte Carlo study on approximate confidence region
evaluations for nonlinear least-squares parameter estimates conducted by Donaldson and Schnabel
(1987) have shown that the simple and most frequently used linearization methods often grossly
underestimate confidence regions. On the other hand, the likelihood and lack-of-fit methods are
considered reliable. Since the utilization of such methods is a computationally demanding task, only the
approximate 95% confidence intervals are presented in the following example.
APPLICATION TO FIELD EXPERIMENTS
The bromide breakthrough data observed at monitoring wells S1 and S2 from the experiments
Tracers and Tracerl 1 were used to validate the well-to-well recirculation solute transport model. For each
data set, the best estimates of the three unknown parameters V, D, and q were obtained by the estimation
methodology previously described. The fourth parameter of the model, R, was set equal to unity,
because bromide was considered as a conservative tracer. Since the estimation of the model parameters
depends critically on the early and late parts of the observed breakthrough responses, to improve
convergence only these portions of the breakthrough histories were employed. The parameter esti-
mates, together with relevant statistics, are given in Table 12.1. The relatively narrow 95% confidence
limits of the estimated parameters, as well as the close agreement of the parameter estimates obtained for
the two observed tracer concentration profiles, indicate that the model can simulate adequately the bro-
mide transport and the solute recirculation effects, at least within the experimental sub-zone.
166
-------
TABLE 12.1 ESTIMATED TRANSPORT PARAMETERS FOR THE BROMIDE BREAKTHROUGH
DATA OF EXPERIMENTS TRACERS AND TRACER11
Tracers: Monitoring Well S1
Approximate
Parameter Estimate Standard 95% Confidence Limits
Deviation Lower Upper
Tracers: Monitoring Well S2
Tracerl 1 : Monitoring Well S1
Tracerl 1 : Monitoring Well S2
V (m/h)
D(m2/h)
q(-)
V (m/h)
D(m2/h)
q(-)
V (m/h)
D (rr^/h)
q(-)
V (m/h)
D (mfrh)
q(-)
0.119
0.032
0.138
0.123
0.032
0.154
0.113
0.032
0.135
0.094
0.021
0.181
0.003
0.004
0.005
0.002
0.002
0.005
0.003
0.004
0.009
0.001
0.002
0.008
0.113
0.024
0.127
0.120
0.027
0.145
0.107
0.024
0.116
0.091
0.018
0.165
0.125
0.040
0.149
0.127
0.037
0.164
0.119
0.040
0.154
0.096
0.025
0.197
The actual bromide breakthrough responses of experiment Tracers observed at sampling locations
S1 and S2, together with the model simulated profiles, are shown in Figures 12.2 and 12.3, respectively.
Similarly, the breakthrough responses of experiment Tracerl 1 observed at sampling locations S1 and S2,
together with the model simulated profiles, are shown in Figures 12.4 and 12.5, respectively. Good
agreement between the experimental data and the simulated concentration history is shown for both
cases. Clearly, the observed data incorporate some experimental error caused mainly by slight
inconsistencies in daily calibrations of the analytical apparatus. Such variations in the observed data
cannot be simulated. Furthermore, the one-dimensional, well-to-well recirculation solute transport model
cannot account for the inhomogeneities and the three-dimensional nature of the real environment.
Nonetheless, for experiment Tracers there is remarkably good agreement between the parameter
estimates for V and D at the two different monitoring wells, S1 and S2, as seen in Table 12.1. Good
agreement is also observed between the model parameter estimates for experiments Tracers and
Tracerl 1 at monitoring well S1. However, the parameter estimates for the two different experiments at
monitoring well S2, as well as the parameter estimates for the two different monitoring wells, S1 and S2, of
experiment Tracerl 1 differ by 20 to 50 percent. For Tracerl 1 the experimental data at monitoring well S2
indicate a reduction in both V and D, and an apparent increase in q. This might be attributed to the
changes in the permeability of the aquifer caused possibly by the biostimulation experiments conducted
during the one-year time lag between the two tracer experiments.
To demonstrate the advantage of the well-to-well recirculation model, the classical advection-dis-
persion model is employed to fit the bromide breakthrough responses of experiment Tracers. The
simulated concentration histories at sampling locations S1 and S2 for the case of fixed injection concen-
tration of bromide are shown in Figures 12.6a and 12.6b, respectively. For the case where the injection
concentration is considered as an additional fitting parameter, the simulated profiles at sampling locations
S1 and S2 are shown in Figures 12.6c and 12.6d, respectively. For both cases, the fitted values for V at
S1 and S2 are reasonable (see Table 12.2), but the estimated 95% confidence limits are broader than
167
-------
those obtained by the well-to-well recirculation model. However, the values of the dispersion coefficients
are approximately double the values estimated when recirculation is properly taken into account. It should
be mentioned that, for the second case, the fitted injection concentration is approximately 30% higher
than its true value. Figures 12.6c and 12.6d show visually-good curve matches to the experimental data;
however, the estimated parameter values are unrealistic since the classical advection-dispersion model
does not fully represent the physical system.
SUMMARY
A semi-analytical and an approximate analytical solution to the one-dimensional advection-disper-
sion transport model accounting for well-to-well recirculation have been presented. Solutions are given
for a flux-type inlet boundary condition and a semi-infinite medium. Sorption is assumed to be governed
by a linear equilibrium isotherm.
Bromide breakthrough concentration profiles obtained from the transport studies at the Moffett site
(Section 5), were used to validate the model. Parameter estimates for the velocity, dispersion coefficient,
recirculation rate, and the associated 95% confidence intervals were determined by nonlinear least-
squares regression. Good agreement was shown between the observed tracer breakthrough responses
and the simulated concentration history.
TABLE 12.2 ESTIMATED TRANSPORT PARAMETERS OBTAINED BY THE CLASSICAL A-D
MODEL FOR THE BROMIDE BREAKTHROUGH DATA OF EXPERIMENT
TRACERS
Approximate
Parameter Estimate3 Standard 95% Confidence Limits
Deviation Lower Upper
Monitoring Well S1
V(m/h)
DOr^/h)
Monitoring Well S2
V (m/h)
D(m2/h)
0.123
(0.112)
0.062
(0.114)
0.128
(0.115)
0.083
(0.130)
0.019
(0.007)
0.034
(0.022)
0.013
(0.006)
0.034
(0.023)
0.086
(0.097)
0.005
(0.069)
0.103
(0.103)
0.016
(0.084)
0.161
(0.126)
0.129
(0.158)
0.155
(0.127)
0.150
(0.176)
a Estimates in parentheses are obtained with injection concentration as an additional fitting parameter.
168
-------
D)
E
c
o
c
0>
o
c
o
o
100
200
300
400
Time (hours)
Figure 12.2. Bromide concentration breakthrough data of experiment Tracers observed at S1 (squares),
and simulated concentration history (solid curve).
o>
E
c
o
c
o
o
c
o
o
100
200
300
400
Time (hours)
Figure 12.3. Bromide concentration breakthrough data of experiment Tracers observed at S2 (squares),
and simulated concentration history (solid curve).
169
-------
a
O
100
80
60
40
20
0 100
-i 1 L
200 300
Time (hours)
400 500
Figure 12.4. Bromide concentration breakthrough data of experiment Tracer! 1 observed at S1
(squares), and simulated concentration history (solid curve).
100
a.
O
O
0 100 200 300 400 500
Time (hours)
Figure 12.5. Bromide concentration breakthrough data of experiment Tracer11 observed at S2
(squares), and simulated concentration history (solid curve).
170
-------
100
too
100
200
300
400
Time (hours)
100
200
300
400
100
80
100 200 300
Time (hours)
400
100
200
300
400
Time (hours)
Time (hours)
Figure 12.6. Curve matching with the two-parameter (V and D) classical advection-disperion model to
Tracers data observed at (a) S1, (b) S2, and with Cp as an additional fitting parameter, (c) S1,
and (d) 52.
171
-------
SECTION 13
BIOSTIMULATION AND BIOTRANSFORMATION MODELING
Lewis Semprini and Perry McCarty
INTRODUCTION
A non-steady-state model was developed for simulating the results of the field experiments. The
model includes basic processes of microbial growth, utilization of the electron donor and electron
acceptor, and the biotransformation of the chlorinated aliphatics as secondary substrates. Transport
processes of advection, dispersion, and sorption in porous media are included in the model formulation.
Model simulations provide a quantitative means of evaluating the field results. The model also
permits comparisons of field results with studies performed in the laboratory. Simulations of possible
restoration scenarios can be made also using the calibrated model.
MODEL DEVELOPMENT
The non-steady-state model that was developed has features similar to those described by Molz et
al. (1986b) and Borden and Bedient (1986). The basic features included in the model are summarized in
Table 13.1. The contaminants and the electron donor (methane) and electron acceptor (oxygen) are
assumed present as dissolved or sorbed onto the aquifer solids. The model was developed for several
flow geometries, including linear (uniform) flow and radial flow. Based on the results of tracer simulations
and 2-D modeling using RESSQ (discussed in Section 5), the flow between the injection wells and the
monitoring wells was simplified to 1-D uniform. The model presented here and the resulting simulations
are for the uniform flow case.
The model incorporates the basic microbial rate processes into the partial differential equations
describing solute transport in porous media. Microbial growth and electron donor and acceptor utilization
are modeled using Monod kinetics, assuming that rates are functions of aqueous substrate
concentrations. The biomass is assumed to be an attached shallow biofilm that is fully-penetrated by the
substrate, i.e., there are no substantial concentration gradients within the film. Sorption of components is
modeled as either an equilibrium or non-equilibrium process.
The rates of microbial growth and decay are assumed functions of both electron donor and acceptor
(Kissel et al., 1984; Molz et al., 1986b; Borden and Bedient, 1986):
-bxA
dt \KSD+CD/\KSA + CA/ \KSA + CA/ (13-1)
where Xa = cell concentration (mg/l), k = maximum utilization rate (g donor/g cell-d), Y = yield coefficient
(g cells/g donor), KSo = donor saturation constant (mg donor/I), KSA = acceptor saturation constant (mg
acceptor/I), b = cell decay coefficient (d'1), and CD and CA are the concentration of the appropriate elec-
tron donor and the electron acceptor (mg/l), respectively. The values of CD and CA are identical to local
concentration in the advecting pore water, owing to the assumption of a shallow biofilm.
172
-------
TABLE 13.1. BASIC FEATURES OF THE NON-STEADY-STATE BIOTRANSFORMATION MODEL
1 -D Transport
Advection, Dispersion, Sorption
Linear, Radial, and Variable Volume Geometries
Monod Kinetics
Electron Donor and Electron Acceptor
Shallow Biofilm of Microorganisms
Sorption
Equilibrium
Non-Equilibrium
Secondary Substrate Biotransformation Kinetics
Monod Kinetics
Competitive Inhibition Kinetics
Boundary Conditions Which Permit Cyclic Pulsing of Methane and Oxygen
Rates of utilization of electron donor and acceptor are given by equations (13-2) and (13-3),
respectively:
dt \KSD + CD/ \KSA + CA/ (13-2)
_ _ kFXa_Cp
dt \KSD + CD/\KSA + CA/ \KSA + CA/ (13-3)
where F is the ratio of electron acceptor to electron donor utilization for the biomass synthesis
(g acceptor/g donor), dc = cell decay oxygen demand (g O2/g cells), and f
-------
dC2
dt
-Xak2
KS2
(13-5)
With this model the rate of transformation of the secondary substrate would be inhibited by the
presence of methane, the electron donor, as an alternative substrate. This inhibitory effect is represented
by the last term in the denominator of equation 13-5. Inhibition increases with increasing CD and
decreasing KSD-
The criteria of Rittmann and McCarty (1980) and Suidan et al. (1987) indicate that the assumption of
a fully-penetrated biofilm without external or internal mass-transfer limitations, implied in the above
formulations, is appropriate for the conditions of the field experiments. This permits a greatly simplified
model. The assumption that the biomass is essentially all attached to aquifer material also appears
appropriate (Harvey et al., 1984). Thus, transport of the microbial mass is not incorporated and equation
(13-1) is directly used to model the change in microbial mass with time.
The 1-D uniform transport of the electron donor, electron acceptor, and the secondary substrate is
governed by:
*i + Pb3C = Dh92C _V3C
* 0 dt 3x2 3x (13-6)
where C is the concentration in the liquid phase (mg/l), C is the concentration of the sorbed solute on the
solid phase (mg/kg), Dh is the hydrodynamic dispersion coefficient (m2/d), V is the average interstitial fluid
velocity (m/d), x is the spatial coordinate (m), pb is the bulk density of the solid matrix (kg/I), and 6 is the
porosity. Based on our laboratory studies, sorption onto the aquifer solids was modeled as linear and
reversible, with the equilibrium sorbed-phase concentration given by:
(13-7)
where Kd is the partition coefficient (I/kg).
For the case of equilibrium sorption, substitution of equation (13-7) into (13-6) leads to the following
transport equation in terms of the liquid-phase concentration:
where R is the retardation factor for the solute (Hashimoto et al., 1964):
6 (13-9)
For the non-equilibrium case, the simple first-order linear nonequilibrium model was used:
^•= a(KdC-C) (13-10)
at x '
174
-------
where a is the rate coefficient for mass transfer between the phases (d~1). This simple model represents a
reasonable approximation of more complex sorption models that include diffusive transfer between
mobile and immobile zones (van Genuchten, 1985).
Substituting equation (13-10) into equation (13-6) yields:
Equations (13-10) and (13-11) must be solved to completely describe transport with non-equilibrium
sorption.
The kinetic expressions for utilization of the electron donor and acceptor, and the biotransformation
of the secondary substrate, are added to the transport equations. For the case of equilibrium sorption the
rate expressions (13-2, 13-3, and 13-4 or 13-5) are added to equation (13-8):
acp _ kXa / CD \ CA
\ [
I I
RD 8x2 RD ax RD \KSD + CD/\KSA + CA/ (13-12)
_ kFXa Cp \ / CA _ dcfdXa / CA
at RA dx2 RA 3x RA \KSD + CD/\KSA + CA/ RA \KSA + CA/ (13-13)
V aC2 k2Xa/- C2
at ~ R2 Ix^" R2 ax R2 KS2 + C2 ( 3 '
C(x,0) = f(x)
-Dhff + VC = Vg(t)
f
-------
Numerical Method
For the equilibrium sorption case, a set of four nonlinear partial differential equations (13-1 , 13-12,
13-13, 13-14), and in the nonequilibrium case a set of 8 equations, must be solved subject to the initial
and boundary conditions given in equations (13-15a,b,c). The equations are coupled with as many as
three of the dependent variables contained in an equation. Numerical methods are required to solve
these equations. Methods for solving these types of equations are given by Molz et al. (1986b), Borden
and Bedient (1986), and Speitel et al. (1987).
This model was formulated using the finite difference method that was solved by numerical inte-
gration. The method is similar to the one used by Borden and Bedient (1986). The method involves:
1 ) Discretization of the spatial terms on the right side of the equation using finite differ-
ences to produce a set of ordinary differential equations.
2) Simultaneous solution of the resulting sets of ordinary differential equations by
numerical integration, using a fixed-step-size, fourth-order accurate, Runge-Kutta
technique (Gear, 1971).
The Quick formulation (Leonard, 1979) was used for the spatial discretization, which significantly
reduces the degree of numerical dispersion. The method uses central differences for the dispersion term
and, for this 1-D case, a four-node formulation for the advection term given by:
dx Ax (13-16)
The solution is explicit in nature and easy to program and modify. The Runge-Kutta method of
numerical integration was chosen since it executes significantly faster than other methods, while achiev-
ing sufficient accuracy (Gear, 1971). The Runge-Kutta routine solves the equations by stepping forward
in time, using four internal steps for each whole time step. Concentrations of biomass, electron donor,
electron acceptor, and the chlorinated aliphatic are updated at each internal time step. Thus, a simultane-
ous solution of the equations is achieved. The numerical code was written in FORTRAN, and the simula-
tions were performed on an IBM-AT.
The solution can become unstable if too large a time step is taken. For the model simulations
reported here, however, stability problems were encountered only when the biostimulated biomass
accumulated to high concentrations in the node nearest the input boundary, resulting in very fast
transformation rates. Decreasing the time step overcame this instability. For most simulations, stability was
achieved when the time step, At, satisfied the criteria given in Bear (1979):
At <
Ax
\r\
(13-17)
To verify the numerical code, the numerical model was compared with analytical solutions to
equation (13-8). A comparison was made between numerical and analytical solutions for example A1-2 in
van Genuchten and Alves (1982) for the breakthrough of a short injected pulse at the outlet of a semi-
infinite column. Agreement was excellent, with little numerical dispersion. Numerical simulations were
also in excellent agreement with analytical results given in example C8-2 of van Genuchten and Alves
(1982), which includes a first-order decay term and zeroth-order source term added to equation (13-8).
176
-------
The numerical solution of equations (13-1), (13-12), (13-13), and (13-14) was checked by
comparing: 1) the total steady-state biomass predicted by the model, and 2) the total biomass deter-
mined by kinetics and mass balance in a plug flow system. The plug flow analytical expression is given by
(13-18)
where X^ represents the total steady-state biomass, CAO is the input substrate concentration, and Q is the
volumetric flow rate. Comparison at several different flow rates and substrate concentrations indicated
deviations of less that 1%.
Simulations of the transport portion of the non-equilibrium sorption model were found to be in
excellent agreement with the analytical solutions of Valocchi (1986) for the case of converging radial flow.
These equations demonstrated that the numerical code solves correctly the set of coupled nonlinear
partial differential equations.
MODEL SIMULATIONS OF BIOSTIMULATION EXPERIMENTS
Model simulations were compared with the results of biostimulation and biotransformation
experiments presented in Section 6. The ability of the model to simulate the transient uptake of methane
and DO observed in the field experiments was tested. Since the biotransformation of the chlorinated
organics depends on the biosimulation of the methanotrophic biomass, simulations of the biotransforma-
tion of the chlorinated organics were attempted only after good matches were obtained to the biostimula-
tion portions of the experiments.
Model Inputs
Model simulations of transient DO and methane concentration responses at the S1 and S2
observation wells assumed that the flow between the injection and monitoring wells can be represented
by 1-D uniform flow. An analysis of the induced flow field created by the operating conditions was per-
formed using RESSQ, a 2-D semi-analytical transport model (Javandel et al., 1984). Figure 13.1 illustrates
the comparative distance traveled by injected fluid with time, along the flow path between injection and
observation wells for the 2-D model (squares) and the 1-D model (line). While some deviation occurs close
to the injection well, the velocity is relatively constant over much of the distance. Thus, the 1-D model
appears adequate for these initial simulations.
Breakthroughs of injected bromide, methane, and DO during the early stages of the biostimulation
experiments were used to estimate the average interstitial fluid velocity. Shown in Figure 13.2 is the
model match of field observations at the S1 well. Consistent with the uniform flow assumption, the 1-D
transport equation fits these data and the S2 data well using the same groundwater velocity of 2.9 m/d.
The results confirm that these three constituents are not retarded with respect to fluid flow; thus, retarda-
tion factors for them were set to 1 in the model simulations. Hydrodynamic dispersion coefficients ranged
from 0.6 m2/d for well S1 to 1.0 m2/d for well S2.
Model simulations are presented for two sets of field data, obtained from the first and second years'
tests, termed Biostiml and Biostim2, respectively (Section 6). The second year's test was started
approximately eight months after methane and oxygen addition had been terminated in the first year's
tests.
177
-------
2.4
I
Figure 13.1. Arrival of fluid at distances along a direct path from the injection to observation wells under
hydraulic conditions of the field experiment based on simulations using RESSQ.
1.1
i -
0.9 -
0.8 -
0.7-
0.6 -
0.5-
0.4-
0.3-
0A -
0.1 •
0
—TTI
O 00
D Bromide
20
TIME (MRS)
O DO
40
Methane
Figure 13.2. Breakthrough of bromide, methane, and DO at the S1 observation well and the fit to
equation (13-4).
178
-------
Table 13.2 lists the model input parameters, which were obtained by independent means of
estimation to the extent possible, including: measurement in the field or laboratory, 2) estimation based
on literature values, or 3) adjustment within a range of literature values to obtain good model fit. A
heuristic fitting procedure was used. Adjusted values were constrained within a reasonable range based
on literature or theoretically derived values. As indicated in Table 13.2, dispersion coefficients lower than
those derived from the fit to the complete breakthrough (Figure 13.2) were required to match the field
response to alternate pulsing of DO and methane.
Table 13.3 contains operational data used in the model. The use of 24 nodes over an interval of 2.4
m was found to provide sufficient numerical accuracy. Time steps between 0.0025 d and 0.01 d were
sufficient to maintain stability for simulations of 20 to 60 days.
Tables 13.2 and 13.3 show different input values of the dispersion coefficient, and initial microbial
concentrations, were required to match responses at the S1 and S2 wells. Also, in the Biostiml experi-
ment, the injected fluid accounted for 90% and 80% of the fluid sampled at wells S1 and S2, respectively.
The 10 to 20% dilution by native groundwater was incorporated into the 1-D simulation by adjusting the
injection concentrations of methane and DO. Thus, separate simulations were required for the response
at the S1 and S2 wells.
TABLE 13.2. INPUT PARAMETERS USED IN THE BIOSTIMULATION MODEL SIMULATIONS
Parameter
V(m/d)
Dh (m2/d)
k(g/g-d)
KSD(mg/l)
KSA(mg/l)
Y (mg/mg)
b(d'1)
F (mg/mg)
fd
dc (mg/mg)
Rd
Ra
Biostiml
2.9
(S1) 0.165
(S2) 0.25
2.30
5.5
1.0
0.5
0.15
2.4
0.8
1.42
1.00
1.00
Biostim2
(S1) 2.9
(S2) 3.3
(S1) 0.165
(S2) 0.25
2.0
1.0
1.0
0.5
0.10
2.4
0.8
1.42
1.00
1.00
Value
Basis
M
F
F
F
L
L
F
F
L
L
M
M
Literature
Values
-
3.5-5 (32°C)
0.2-0.3
0.01-0.1
0.35-1.1
0.15-0.4
2.2-4
0.8
1.42
-
-
Refs.
C
A,B
C,D
C, E, F,G
E,F,G
C.E.F
H
H
M = measured; F = fitted; L = literature (held constant).
References: A, Ferenci et al., 1975; B, Harrison, 1973; C, Wilkinson et al., 1974; D, Morinaga et al.,
1979a; E, Morinaga et al., 1979b; F, Whittenbury et al., 1970; G, Heijnen and Roels, 1981; H, McCarty,
1975.
179
-------
TABLE 13.3. MODEL SETUP PARAMETERS
Parameter Biostiml Biostim2
Total Simulation Length (m) 2.4 2.4
Num. Nodes 24 24
Ax(m) 0.1 0.1
At (d) 0.01 0.0025
Initial Conditions
Xaj (mg/l) (81)0.015 1.13a
(S2) 0.035 1.7a
CDJ (mg/l) 0.0 0.0
CAJ (mg/l) 0.0 0.0
Injection Cone.
Pulse Interval
(mg/l) (S1)18.0 (S1.S2) 20
(S2) 16.5b
CAO (mg/l) (31)31.0 (S1.S2) 36
(S2) 26.3b
0.01 (0-19.1 d)C 0.01 (0-1 d)c
0.17 (19.1-25 d) 0.02 (1-10 d;
tA(d) 0.02 (0-19.1 d) 0.04 (0-1 d)
tA(d) 0.34 (19.1-25 d) 0.03 (1-10d
^average value for the distributed concentration over the distance of 2.2 m.
°lower injection concentration to account for dilution by native groundwater.
Experimental interval over which the pulse length was used.
Simulation of the First Season's Test (Biostiml)
Figure 13.3 illustrates the simulation match for methane and DO concentration response at the S2
well for Biostiml. Here, dissolved methane and DO were injected continuously in short alternating pulses.
The good simulation match demonstrates that the field response results from biostimulation of indigenous
methane-utilizing bacteria. Initially the concentration of methane-utilizing bacteria was sufficiently low so
that little uptake was observed. With biostimulation, an increase in the microbial concentration occurred,
resulting in a marked decrease in methane and DO concentrations beginning at about 200 hrs. The
simulated biomass increase with time is illustrated in Figure 13.4. A fairly uniform spatial increase in
180
-------
1
^x
o
Figure 13.3. Model simulation and observed methane and DO response at the S2 observation well in the
Biostiml experiment.
60
o
3
50 -
30 -
20 -
10 -
DAYS = 10
—I 1 1 1 1 1
0.4 0.8 1.2 1.6
DISTANCE (METERS)
-B—B—B—B—B—B—O
2.4
Figure 13.4. Snapshots of the predicted distribution of methane-utilizing biomass during Biostiml
experiment.
181
-------
biomass is predicted during the first 10 days, during which methane was uniformly available through out
the test zone. However, with time, the methane-utilizing biomass increases to a greater extent in closer
proximity to the injection well as a result of rapid microbial growth.
The initial biomass concentration was unknown, and thus had to be treated as a major fitting
parameter for the model. Other parameters affecting the lag time are k, b, KSD- and F. These are basic
rate coefficients for which average literature values were used initially, but slight adjustments were made
subsequently to improve model fit. The adjusted values indicated in Table 13.2 are consistent with values
measured in the laboratory or determined from theoretical considerations. Values for dc and fd were not
adjusted. The good fit between model simulations and field results, using coefficients largely derived
from basic studies, is encouraging.
Even the initial concentration of methane-utilizing biomass selected to obtain a reasonable
simulation for data for S2 is reasonable, based upon acridine orange direct counts of bacterial numbers
(Ghiorse and Balkwill, 1983) measured on aquifer solids from the test zone prior to biostimulation (Sec-
tion 4). Assuming a methanotroph bacterial cell has a mass of 0.5x10'13 g, the best-fit initial biomass
concentration of 0.035 mg/l is about one thousandth of the estimated enumerated concentration. This
appears reasonable, since methanotrophs are expected to represent only a small portion of the total
aquifer biomass.
Biomass estimates were also made to match the response at the S1 well (not shown). The best-fit
initial biomass of 0.015 mg/l was quite similar to that used for S2 data, considering the heterogeneities in
the system. All other parameters were left unchanged in achieving this value.
The fact that S1 and S2 simulations yielded good matches with field results using the same kinetic
parameters is not unexpected since the types of methane-utilizers is not expected to vary greatly in the
test zone. This is important since it suggests that the model, when calibrated with one set of field data,
can be applied to other locations within the same aquifer system. The only parameter that then needs to
be adjusted is the initial biomass concentration which is expected to vary spatially.
Pulsing of Electron Donor and Acceptor
The Biostiml experiment was started with short alternating pulse cycles to simulate continuous
injection of methane and DO. Based on model simulations (Figure 13.4), microbial growth would
eventually become concentrated near the injection well; this would be undesirable, as it would increase
the clogging potential and greatly reduce the desired uniform distribution of methanotrophic biomass in
the stimulated zone. In order to reduce these potential problems, the methane and oxygen pulse cycle
was increased, as discussed in Section 6.
After 458 hrs of injection, the pulse times of methane- and oxygen-containing injection water were
increased from 0.01 and 0.02 d to 0.17 and 0.34 d, respectively. The responses at the S2 and S1 obser-
vation wells are illustrated in Figures 13.5 and 13.6, respectively. The simulation results are in good
agreement with the field observations.
The DO data are more reliable for evaluations of system response to pulsing because methane
concentrations were near the detection limit. The mean DO concentration at the S2 well increased from
3.5 mg/l before pulsing to 4.5 mg/l after pulsing. Both simulations and field data indicated a continued
decrease in concentration with time. This predictably resulted from an increased oxygen demand from cell
decay as biomass continued to increase in the test zone.
Simulation for the S1 well (Figure 13.6) shows less attenuation in the pulse heights than at S2,
which is consistent with field observations and the analytical solutions of Valocchi and Roberts (1983).
Greater microbial uptake with longer distance traveled also acts to attenuate the pulse heights at the S2
well.
182
-------
400
Figure 13.5. Model and field response at the S2 well resulting from the change from short to bng
alternating pulse cycles.
PUUSMO (WELL SI)
900 520 540 560 900 800
400
Figure 13.6. Model and field response at the S1 well resulting from the change from short to long
alternating pulse cycles.
183
-------
In order to obtain correct peak attenuations in simulations shown in Figures!3.5 and 13.6, the
dispersivity had to be decreased by a factor of 4 below that obtained from a fit of the complete
breakthrough data (Figure 13.2). However, this decrease in the dispersivity resulted in a somewhat
sharper simulated initial breakthrough of methane and DO compared to the field results (Figure 13.3).
This phenomenon was observed when simulating DO pulse experiments before biostimulation (Figure
5.16). The higher values of dispersivity required to fit the complete breakthrough curves perhaps result
from fitting the 1-D macroscopic advection-dispersion equation with uniform flow to an aquifer that
undoubtedly has vertical variations in hydraulic conductivity. On the other hand, pulsing in the test zone
may reflect the response of a highly conductive zone of lower dispersivity, which conveys a larger portion
of the sampled groundwater. Tracer and modeling studies by Molz et al. (1986a) support this hypothesis.
Figure 13.7 shows the growth with time of the simulated biomass concentration at a node in the
region of the S2 observation well. Biomass concentration reaches a maximum concentration after
injecting for 400 hrs (with short pulse cycles). The concentration then decreases due to cell decay as the
decreasing amount of methane reaching that node can no longer support the resulting biomass. The
initiation of longer pulses at 458 hrs results in a reestablishment of a higher biomass concentration since
more methane then reaches this area.
Pulsing Effect on Biomass Distribution
Figure 13.8 presents a snapshot of the predicted spatial distributions of methane and DO during
pulsing. Growth of biomass at that instant occurs primarily in the region 0.8 to 1.3 m and 1.5 to 2.2 m from
the injection well, where the concentrations of both methane and oxygen overlap. In the next instant in
time, the methane peak moves to the right, as does the location of the highest biomass growth rate.
For a given pulsing strategy, biomass would reach some steady-state level after a sufficiently long
time. The near steady-state model distributions predicted to result from two different pulse cycle lengths
after 120 days are shown in Figure 13.9. With alternate pulses of 4 and 8 hrs, most of the microbial mass
would be located within the first meter of the test zone. Increasing the pulse length to 8 and 16 hrs is
predicted to help distribute the biomass more uniformly over the test zone. The total biomass present
(area under the curves) for this case is less, however, since more methane and oxygen reach the outer
boundary and are removed from this particular system.
Simulation of Second Season's Test (Biostim2^
Biostimulation was achieved in the second season of field testing using the same methodology as
in the previous year (Section 6). In the second year, methane and DO uptake was observed almost imme-
diately after commencing biostimulation, which implies that a significant population of methanotrophs
remained from the previous season of stimulation. Thus, the initial biomass in the simulation model had to
be adjusted to fit this response. The initial biomass distribution was assumed to be that illustrated in Figure
13.9 (4 and 8 hr pulses), but reduced in concentration by first-order decay over the eight months of
starvation. In addition, the use of rate parameters from the Biostiml experiment (Table 13.2) did not yield
as good matches, predicting more gradual decreases in methane and oxygen concentrations than were
observed in the field.
Several rate parameters were adjusted to obtain a good match with second-season field observa-
tions, as summarized in Table 13.4. In the resulting comparison for the response at the S1 well shown in
Figure 13.10, the more rapid uptake of methane and DO compared to the first year's results(Figure 13.3)
is apparent. Both the simulation and the field observations show two peak methane concentrations within
the first 60 hr, resulting from a planned increase in the methane injection pulse length after the first 24 hrs
in order to supply more methane to the test zone. The model is able to simulate these transient
conditions.
184
-------
200
400
•00
TIME (MRS)
Figure 13.7. Predicted biomass concentration at the node 2.2 meters from the injection well due to
biostimulation with short and long pulse cycles.
(CO HR SIMUIATION - 4 HR CH4/ • HR 00)
0.4
1.2
DISTANCE (METERS)
1.*
Figure 13.8. Snapshot of the predicted spatial distribution of DO and methane during a DO pulse cycle
after a steady-state biomass is achieved.
185
-------
\s
6
8
0.4
1.2
DISTANCE (METERS)
1.6
Figure 13.9. Predicted steady-state biomass distributions achieved at two different pulse cycle lengths.
o
z
o
180
200
Figure 13.10. Simulated and observed DO and methane response at the S1 observation well during the
Biostim2 experiment.
186
-------
TABLE 13.4. COMPARISON OF ADJUSTED MODEL PARAMETERS FOR THE
BIOSTIMULATION EXPERIMENTS (WELL S2)
Initial Biomass (mg/l)
k (mg/mg-d)
KSD (mg/i)
k/Kso (l-mg/d)
b(d'1) (methane and DO)
b1 (d'1) (no methane or DO)
Biostiml
(First Year)
0.035
2.3
5.5
0.42
0.15
0.01
Biostim2
(Second Year)
1.75a
2.0
1.0
2.0
0.10
Biostim3
(Third Year)
5.75a
2.0
1.0
2.0
0.10
a Average concentration over the 2.2-m simulated interval.
Simulations were also performed on the response at the S2 well (not shown). An initial biomass
concentration 50% higher than that used in the S1 simulation was required, which is consistent with simu-
lations of the first year's experiment. This consistency further supports the hypothesis that the S1 and S2
wells do not sample identical zones in the aquifer.
As indicated in Table 13.4, the initial average biomass was increased in the second-year simulation
by a factor of 50 over the first year simulation. This represents approximately 6% of the predicted steady-
state biomass at the end of the first year (Figure 13.9). The b' value listed in Table 13.4 was required to
achieve this estimated 17-fold decrease in biomass over the 8 month resting period, and is substantially
lower than the b value used under active biostimulation. During most of the 8 month period, oxygen was
not present and, thus, a lower decay rate would be expected.
Table 13.4 also lists the rate coefficients for the two periods. In order to achieve the rapid decrease
in methane to low concentrations as observed in the second season, KSD was lowered by a factor of 5,
while k was lowered only slightly from the first-year value. These changes resulted in an increase in the
ratio of k to KSD by a factor of 5. A lower b was found appropriate as well. These changes are more
favorable for reducing methane rapidly to low concentrations, and perhaps reflect an evolution of
methanotrophic capability towards a population with greater ecological advantage. A similar change in
laboratory column study with acetate as primary substrate was noted by Bouwer and McCarty (1985).
Simulations of the BiostimS experiment of the third season of field testing also were performed.
Uptake of methane was observed to occur even more rapidly than in the second season. The simulations
of the BiostimS experiment match the field observations using the same kinetic parameters as in the
second season. The initial microbial mass, however, had to be increased by a factor of 5 compared to that
used in the second season simulation.
Summary of Biostimulation Simulations
Model simulations provided good matches to the observed transient uptake of methane and DO in
the test zone. The matches indicate a mathematical model combining transport, utilization of the electron
donor and acceptor, and microbial growth provide a good representation of the behavior observed in the
field tests. This good agreement justifies adopting the simplifying assumption of uniform one-dimensional
transport.
Simulations indicated that initial biomass of methane utilizing bacteria was a key fitting parameter.
The initial biomass estimates were reasonable when compared with prestimulation microbial
187
-------
enumerations. Increases in estimated initial bipmass over the field seasons indicate that methanotrophic
biomass stimulated in the previous year's experiments degraded only very slowly.
Kinetic parameters obtained from model matches are quite reasonable based on laboratory
measured values and theoretical considerations. Thus, the model can be used with some confidence in
simulating the biostimulation of methanotrophs in the field test.
MODEL SIMULATIONS OF BIOTRANSFORMATION EXPERIMENTS
Model simulations of the biotransformation of chlorinated aliphatics were compared with the results
of the second and third seasons of field testing. The simulations were performed after the match to the
DO and methane uptake was achieved in the biostimulation simulations previously discussed. The
different kinetic models for biotransformation and sorption were evaluated to determine which models
best described the behavior observed in the field.
Model Input Parameters
Tables 13.2 and 13.3 contain the model parameters used in the biostimulations. The additional
parameters needed for the chlorinated aliphatic decomposition simulations are those for the sorption and
biotransformation of the chlorinated aliphatics. The sorption parameters include the retardation factor, for
the equilibrium sorption model, and the partition coefficient, Kd, and mass transfer coefficient a, for the
first-order kinetic-model. Sorption parameters, lying within the range of values used to fit the break-
through response of the organics before biostimulation, were used in the model simulations.
Both the Monod and competitive inhibition models for secondary substrate utilization required as
inputs the V.2 and Ks2 values for each compound. Literature values for the Ks2 for each compound were
not available. Laboratory-determined values for KS, presented in Section 9, show a range for TCE from
0.1 to 0.9 mg/l. Ksa values used in the simulations ranged from 1 to 2 mg/l. Since the concentrations of
the chlorinated aliphatics in the field studies ranged from 50 to 100 u.g/1, these KS values essentially
reduce equation (13-4) to a second-order rate expression, where the ratio of k/Ks2. is a second-order rate
parameter. It should be noted that in the competitive inhibition model (equation 13-5), the Ks2 and KSD
values are used to determine the extent of methane inhibition. The Ks2 values were adjusted to yield
simulation responses to competitive inhibition similar to those observed in the field.
The maximum utilization rate, k2, for each compound was adjusted to obtain matches to the field
observations. Use of different V.% values was necessary depending on whether the Monod or competitive
inhibition model was being used.
Results of Biotransformation Simulations
The initial simulations of the second season's biotransformation experiment (Biostim2) were per-
formed using the Monod kinetic model (equation 13-4) and the equilibrium sorption transport model.
Figure 13.11 shows simulations and field observations for the response of TCE, cis-DCE, and trans-DCE
at the S1 observation well. The simulation corresponds to that for methane and oxygen shown in Figure
13.10. The model simulations are generally in good agreement with the observed transient decreases of
the chlorinated organics in response to biostimulation of the methanotrophs. In obtaining a fit to the later
time trans-DCE concentrations, the model underestimated transformation at early time.
Table 13.5 presents retardation and kinetic factors used in the simulations. The retardation factors
are in the range of values determined in the transport experiments. The main parameter that was varied in
these simulations was the maximum utilization rate (k2), ranging from 0.007 d'1 for TCE, the most slowly
degrading compound, to 0.15 d'1 for trans-DCE, the most rapidly degrading compound. The ratio of
k2/Ks2 values of the chlorinated aliphatics to that of methane range from 0.0035 for the slowly degraded
TCE, to 0.075 for the rapidly degraded trans-DCE.
188
-------
(RESULTS WELL SI)
300
400
TIME (MRS)
Figure 13.11. Simulation and field response of trans-DCE, cis-DCE, and TCE at the S1 well during the
second season's biostimulation experiment (Biostim2). Simulations performed with the
Monod kinetic model (equation 13-4) and the equilibrium sorption transport model.
TABLE 13.5. MODEL PARAMETERS FOR SIMULATION OF CHLORINATED
ORGANICS IN BIOSTIM2 (Figure 13.11)
Compound
Methane
trans-DCE
cis-DCE
TCE
R
1
8
8
12
k
(a"1)
2.0
0.15
0.018
0.007
k/KS
(l/mg-d)
2.0
0.15
0.018
0.007
Relative to
Methane
1.0
0.075
0.009
0.0035
Ks value of 1 mg/l used for all the compounds.
These initial simulations strongly support the hypothesis of biotransformation due to biostimulation
of methanotrophs. The different responses of the chlorinated organics to biostimulation results mainly
from the compounds having different transformation rates. The similar responses of the model and the
field results indicate that the processes incorporated into the model are representative of those occurring
in the field. However, some salient differences remain.
189
-------
The lack of fit of the trans-DCE simulation to the early time response suggests that the kinetically
limited sorption model or competitive inhibition biotransformation model may more accurately represent
the field behavior. Another possible explanation for the lack of fit is that the concentrations of trans-DCE
in this experiment were improperly corrected for a large background concentration of 1,1-DCA, that was
subsequently found to co-elute with t-DCE during GC analysis. If the actual concentration of 1,1 -DCA was
higher than assumed, then the actual extent of biotransformation of trans-DCE would have been
underestimated. Model parameters that would better simulate the early time response also would predict
greater removals at a later time than shown. This is consistent with the third season's field results. In the
latter, the 1,1-DCA problem was eliminated.
Competitive inhibition of methane on transformation of the chlorinated organics, as well as rate-
limited sorption-desorption kinetics, were strongly indicated in the third season's results. These data,
therefore, provide a basis for evaluating the more complicated biotransformation models.
Figure 13.12 shows model simulations and the response of vinyl chloride, trans-DCE, and cis-DCE
at the S2 well, due to biostimulation (BiostimS). In order to activate the large mass of methanotrophs
initially present to degrade the chlorinated aliphatics, a lag period of 6 hrs was incorporated into the
simulations. The simulations match the observed response quite well, especially for vinyl chloride and
trans-DCE, the most rapidly degraded compounds. The simulations show that a rapid reduction of the
aqueous-phase concentration occurs once the large initial methanotrophic population becomes acti-
vated. As time proceeds, nondegraded compounds from the sorbed phase are slowly released to the
aqueous phase, resulting in a slow decrease in the aqueous-phase concentration. Thus, rate-limited
desorption appears to be an important process.
For comparison purposes Figure 13.13 presents simulations for vinyl chloride, where sorption is
modeled as an equilibrium process. All other parameters are those used in Figure 13.12 simulation. The
simulations do not match either the initial rapid decrease in concentration nor the gradual decrease at later
time. For the trans-DCE simulations (not shown), a similar response is obtained. However, for the more
slowly transformed cis-DCE, it is difficult to determine which model provides a better fit to the data. Thus, it
appears that the first-order rate model for sorption-desorption is a better choice for simulating field
response, especially when the transformation rate is rapid.
Parameter inputs for the simulations shown in Figure 13.12 are presented in Table 13.6. Parame-
ters include the sorption parameters of Kd and a, as well as biotransformation rate parameters of k2 and
Ks2 for the chlorinated aliphatics.
Sorption parameters are in the range of values used to model transport experiments before bio-
stimulation. Transformation rate parameters (k2/Ks2) for vinyl chloride and trans-DCE were found to be in
the range of those for methane. While vinyl chloride and trans-DCE had similar biotransformation rate and
parameters, simulated and measured trans-DCE concentrations decreased more slowly. The main
parameter causing the slower decrease in the simulations is the higher Kd value for trans-DCE. Thus, the
more gradual decrease of trans-DCE appears to result from sorption interactions, rather than from a slower
rate of transformation. The higher K§2 value for vinyl chloride compared to trans-DCE was used to match
the competitive inhibition response, as will be discussed. The transformation rates used for cis-DCE and
TCE are significantly lower than for vinyl chloride and trans-DCE.
Model Simulations of Competitive Inhibition Response
Transformation rates for trans-DCE, cis-DCE, and TCE (k2/Ks2) used to simulate third-season field
results are significantly higher than those from the the second season (Table 13.5). The higher rates are
partly associated with the use of a competitive-inhibition biotransformation model and rate-limited
desorption in the third year
190
-------
o
o
o
a:
o
1 1 1 1 1 1 1 1 1 1 1 1 I I I
140 160 180 200
D VINYL CHLORIDE
TIME (HRS)
O TRANS-DCE
X CIS-DCE
Figure 13.12. Simulations and field response of vinyl chloride, trans-DCE, and cis-DCE at the S2 well
during the third season's biostimulation experiment (BiostimS). Simulations performed
with the multiple substrate (inhibition) kinetic model (equation 13-5), and the non-
equilibrium sorption model.
20 40 60 80 100 120 140 160 180 200
IMC (MRS)
Figure 13.13. Simulation of vinyl chloride response at the S2 well using the equilibrium sorption model.
191
-------
TABLE 13.6. MODEL PARAMETERS FOR SIMULATION OF CHLORINATED ORGANICS
IN BIOSTIM3 (Figure 13.12)
Compound
Methane
VC
trans-DCE
cis-DCE
TCE
Kd
(ml/g)
0.0
0.40
1.60
1.90
2.25
a
(d-1)
0.00
0.33
0.33
0.33
0.33
k
(d-1)
2.0
2.0
2.0
0.10
0.015
KS
(mg/l)
1.0
2.0
1.0
1.0
1.0
k/Ks
(l/mg-d)
2.0
1.0
2.0
0.1
0.025
The effects of competitive inhibition on the observed concentrations were most pronounced at the
S1 well. Figure 13.14 shows both the simulated aqueous- and sorbed-phase concentrations of vinyl
chloride at the S1 well. The rapid response of the system to methane pulses is apparent. The simulated
aqueous-phase concentration response shows amplitude heights and a general concentration decrease
(Figure 6.15), similar to those observed in the field experiments. Oscillations of the calculated concen-
tration of VC in the sorbed phase in response to methane pulsing are greatly attenuated compared to
those of the aqueous phase. This results from the slow transfer of the vinyl chloride from the sorbed
phase to the aqueous phase, where biotransformation is assumed to occur. Simulations for trans-DCE
were similar. By contrast, simulations using the equilibrium sorption model, Figure 13.15, show oscilla-
tions in aqueous vinyl chloride concentration were greatly attenuated, as instantaneous desorption
greatly damped the aqueous-phase response to competitive inhibition. Simulations for the more strongly
sorbed trans-DCE showed oscillations being even more attenuated when the equilibrium sorption model
was used. The simulations, therefore, indicate that both the processes of competitive inhibition and rate-
limited sorption-desorption strongly contribute to the responses observed in the field test.
The concentration oscillations, in response to competitive inhibition, become attenuated with
transport through the test zone, as indicated by simulations of the S2 well (Figure 13.12). This results
partly because methane concentrations are lower and more attenuated in that region. The combined
processes of dispersion, sorption, and biotransformation also act to attenuate oscillations in the chlori-
nated organic concentrations.
The effects of reduction in competitive inhibition were simulated also for the transient experiments
in which formate and methanol were substituted for methane. Model simulations, used in the design of
these experiments, indicated that enhanced transformation would result if the compound that was
substituted for methane did not inhibit the rate of transformation, but supplied reducing power to keep the
MMO enzymes activated. The simulations assumed that biotransformations would proceed at a maximum
rate (no inhibition) in the absence of methane. The simulation also assumed that methanotrophs would
not grow when formate was substituted for methane.
The simulation of the response for trans-DCE at the S1 well to substitution of formate for methane is
presented in Figure 13.16. The simulated response is similar to that observed in the field (Figure 6.18).
Upon adding formate, oscillations in trans-DCE concentration stopped, and the concentration decreased
to levels slightly lower than the best achieved when minimum methane conditions were present. The
model assumption that maximum utilization rates would be achieved when formate was substituted for
methane, appears to be correct, at least in the early stages of formate addition.
192
-------
AQUEOUS AND SORBEO PHASE CONCENTRATIONS
100 120 140 160 180 200
O AQUEOUS
Figure 13.14. Simulation of the aqueous- and sorbed-phase vinyl chloride concentrations at the S1 well
using the non-equilibrium sorption model. The field results are shown in Figure 6.15.
(EQUILIBRIUM SORPTION MODEL)
o
8
80 100 120
TIME (MRS)
140 160 180
200
Figure 13.15. Simulation of the aqueous vinyl chloride response at the S1 well using the equilibrium
sorption model.
193
-------
o
8
o
a:
O
0.9 -
O.B -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0
SHORT CH4 PULSE
D LONG OU PULSE
200
TIME (MRS)
a TRANS-DCE
400
CH4
Figure 13.16. Simulation of the trans-DCE response at the S1 well to methane pulsing and formate
substitution for methane. The field results are shown in Figure 6.18.
With prolonged formate addition, the field results and the model simulations differ, with a greater
increase in concentration with time observed in the field. Since the simulations do include microbial
decay, this process does not account for the greater decrease observed in the field. Processes not
incorporated in the model may be occurring, such as the gradual deactivation of the MMO enzyme with
prolonged formate addition or growth of a separate formate-using non-chlorinated aliphatic-degrading
culture.
DISCUSSION
The agreement found between model simulations and field-test results demonstrates the useful-
ness of a model that incorporates fundamentals of microbial and transport processes. Since knowledge of
these processes is incomplete, and basic rate and growth coefficients differ from one species of methan-
otrophs to another, some adjustment of these coefficients was necessary for a good match to field data.
Also, some adjustment can be expected because of the simplifying assumptions made in model devel-
opment, as well as due to the fact that aquifer heterogeneities make development of a perfect simulation
model difficult. Experiments are currently underway in order to determine how well coefficients for rate
and growth, as well as initial population levels, as determined in the laboratory using Moffett Field aquifer
material, match the coefficients used to fit the field results. It is hoped this will remove some of the uncer-
tainties and will provide an approach for better predicting field behavior.
Of particular interest are the parameter adjustments required in order to fit the first and second year's
field results. The ratio of k/Kso was increased by a factor of 5 in the second year's simulations. Bpuwer
and McCarty (1985) found ratios of k/Kso to increase with time when modeling the aerobic utilization of
acetate in a continuous flow laboratory column. Ratios increased by factors of 15 to 30 over a three-year
period. They suggest either the development of a separate faster-degrading population or the slow
194
-------
adaptation of the initial population as possible reasons for the changes in this system. Such changes are
possible for the population of methanotrophs at Moffett Field. This may limit the usefulness of rate-
coefficients derived from short-term laboratory studies, although the changes noted tend to lead to more
efficient removals with time. Thus, short-term laboratory studies may tend to provide conservative esti-
mates of treatment potential.
The 1-D representation for fluid flow from the injection to the observation wells appears to be ade-
quate for estimating responses at the observation wells under the experimental conditions presented.
Two-dimensional simulations would, however, be required for adequate modeling of the production well's
response to biostimulation, or for determining the 2-D biomass distribution in the test zone. For the 1-D
simulations that were performed, the velocity varies most near the injection well. Simulations also were
performed using a variable volume element form of the 1-D model that fits the actual times of arrival more
closely. This model indicates a distribution similar to the uniform-flow case of the microbial mass near the
injection location, despite the higher predicted fluid velocities in that region. Matches to methane and DO
response, and those of the chlorinated organics, were similar.
The different dispersivities required to fit the conditions of pulsing compared with near-continuous
injection suggest significant aquifer heterogeneities were present, most likely in the vertical direction.
This is an important limitation of any model, as it requires information about an aquifer that is difficult to
obtain. More tracer studies here would be very useful, and would be required as inputs into 2-D or 3-D
simulations as well. Advanced 2-D models that consider vertical heterogeneities may be useful, as
demonstrated by Molz and Widdowson (1988).
The simulated responses of the chlorinated organics to biostimulation indicate that the organics are
transformed at different rates depending on their chemical structure. The observation that the
methanotrophic transformation rates of chlorinated alkenes increase with decreasing chlorine substitution
agrees with the laboratory findings of Henson et al. (1987, 1988). Sorption of organics also plays an
important role in the responses observed. Nevertheless, the observed responses can be closely
matched by using a competitive inhibition model with rate-limited sorption and desorption. In this
exercise, a simple model that incorporates relatively few input parameters was used to incorporate these
processes. More information on the basic microbial processes is required to select the most appropriate
inhibition model, and additional sorption studies are needed to determine if a more complex sorption
model is required.
Model simulations also were performed to investigate the response of the system to hydrogen
peroxide addition as a substitute for oxygen. This would permit addition of a higher methane concentra-
tion. However, simulations using the competitive inhibition model showed that little additional trans-
formation would result if a higher methane concentration were added, even though a greater biomass
would develop in the test zone. The higher methane concentration input this requires, inhibits more
strongly transformation rates for the chlorinated aliphatic compounds. This simulated result is consistent
with the field results from peroxide addition, discussed in Section 6, where no additional TCE transforma-
tion was observed when greater amounts of methane were added.
Simulations of field responses from switching the electron donor from methane to formate were
quite good initially, but did not mimic the gradual increase in chlorinated aliphatic concentration observed
in the field. More basic information on the microbial kinetics and ecology is required to model such
transient changes.
The use of the 1 -D model greatly simplified the simulations made. However, model parameters
used to obtain fits to the field response probably reflect this 1-D simplification. For instance, the first-order
mass transfer coefficient used in the sorption model may also include the effects of aquifer
heterogeneities. The slow breakthrough of solute moving in a low conductivity zone would produce a
similar response as slow sorptive release from the aquifer solids. Microbial process parameters, such as
the appropriate saturation coefficient, Ks, for simulation would also be affected by such mass-transfer
effect. Thus such parameters may be site- as well as model- and compound-specific.
195
-------
The fact that the relatively simple 1-D non-steady-state model fit the field results so well when using
typical coefficients for bacterial growth and decay, secondary utilization rates, and measured values of
transport, dispersion, and sorption indicates that the understanding of the basic process underlying
movement and fate are fairly well understood. The value of a basic model of this type is that it can be used
with confidence to obtain reasonable predictions for other systems. The knowns and unknowns are better
understood, thus making it easier to determine what information about a given system is most critical for
evaluation. The 1-D model used here does have limitations, and 2-D and 3-D models can be applied
where this is necessary to obtain the information required, assuming this is justified by the availability of
sufficient knowledge of aquifer characteristics.
196
-------
SECTION 14
IN-SITU BIOTRANSFORMATION METHODOLOGIES
Perry McCarty, Lewis Semprini, and Paul Roberts
In order to carry out successful bioremediation, a good understanding of the processes involved,
and the development of appropriate methodologies for design and control, are necessary. This requires
knowledge of the physical, chemical, and biological characteristics of the contaminated subsurface envi-
ronment, and of the nature and distribution of contaminants. In addition, methods are needed for in-situ
stimulation of the required population of microorganisms, for effecting their distribution to areas where
needed, and for maintaining their transforming abilities.
PROTOTYPE SCENARIOS
For illustration of the requisite information, Figure 14.1 shows two possible remediation systems
that might be used to degrade contaminants in the saturated zone of an aquifer. The system depicted in
Figure 14.1 a is designed to permit the injection and extraction of fluid in a zone that runs transverse to the
natural flow of the groundwater. Methane, oxygen, and other nutrients that may be required by a native
population of methanotrophs are dissolved in the injection water, which then is passed through the
aquifer towards the extraction well. An above ground treatment system, such as air stripping, is used to
remove contaminants contained in the extracted water. The treated water is used here as the injection
water. This system would be capable of removing contaminants from the aquifer, even in the absence of
bioremediation. The advantages of bioremediation are 1) more rapid removal of contaminants, and
2) destruction of that portion removed biologically.
In the absence of bioremediation, a single pass of injected water through the aquifer depicted in
Figure 14.1 a would displace the existing groundwater in its path and effect the removal of the dissolved
portion of the contaminants there contained for above ground treatment. Thus, the originally contami-
nated aquifer water would be replaced with non-contaminated injection water. Then, the contaminants
sorbed to the aquifer solids would begin to desorb into the injected water, in turn contaminating it. Injec-
tion of another pass of cleaned injection water, along with extraction, would displace the contaminated
injection water as previously, and again in turn, the newly injected water would become contaminated
through further desorption of contaminants from the solids. Through repeated injection and extraction,
the contaminant eventually would be purged from the pore water and leached from the solids, and the
aquifer would slowly become restored. The greater the degree to which the contaminants are sorbed, the
larger the number of passes of injection water required, and the longer the time for cleanup.
With bioremediation, each pass of injected water might contain primary substrate and nutrients to
stimulate the methanotrophic population, which in turn would oxidize a portion of the desorbed contami-
nants. Since the organisms essentially remain in place in the aquifer, contaminants could be decomposed
as they desorb from the aquifer solids. The degradation would reduce the solution concentration, thus
enhancing the rate of desorption. In the scheme shown, once the aquifer in the path of the extraction
system is adequately cleaned, then the bioremediation system could be discontinued until more contami-
nated water moves into the treatment zone as a result of natural groundwater movement, and the treat-
ment could be reinitiated. In this manner, down-gradient water users would be protected.
197
-------
ABOVE GROUND TREATMENT
/I
GROUND
WATER
FLOW
BIOSTIMULATED ZONE
B
ABOVE GROUND TREATMENT
INJECTION
EXTRACTION
BIOSTIMULATED ZONES
Figure 14.1. Two possible bioremediation systems.
198
-------
An alternative scheme is illustrated in Figure 14.1b. Here, a series of biostimulated zones are
developed that run transverse to the direction of groundwater movement. Once the biologically active
zone is developed, then an injection and extraction system is operated in the direction of groundwater
flow. This system is more efficient than the one illustrated in Figure 14.1 a, but depends upon the ability of
the methanotrophs to maintain activity during a period when they are not supplied with methane and
oxygen.
Several other treatment schemes are possible, depending upon the given situation. However, the
approaches illustrated in Figure 14.1 can be used to outline the characteristics of the system that need to
be understood in order to carry out successful remediation as related to 1) the properties of the aquifer,
2) the distribution of contaminants, and 3) the design features of the treatment system.
Important physical characteristics of the aquifer include the composition of the aquifer materials,
aquifer heterogeneity, transmissivity, dispersivity, and hydraulic gradient. These properties will provide
information on direction and speed of groundwater flow, pumping requirements for injection and extrac-
tion, and the degree of certainty with which aquifer response to injection and extraction can be predicted
and controlled.
Important chemical properties of the groundwater include pH, temperature, major inorganic ions,
and presence of nutrients (N,P). These factors will indicate whether the environment is suitable for the
growth of methanotrophs, and will allow some prediction to be made of reaction rates. Redox characteris-
tics of importance are perhaps best characterized through analysis for dissolved oxygen, nitrate, nitrite,
ammonia, Fe(ll), Mn(ll), and sulfide concentrations. The potential presence of organics that are readily
biodegraded aerobically also needs to be known. Aquifers that are highly reduced, or contaminated with
degradable organics may have a high oxygen demand that would affect the ability to maintain an aerobic
system.
The nature, distribution, and concentrations of contaminants in the aquifer also must be known.
The rates of transformation of halogenated alkanes and alkenes vary widely, depending on the degree of
halogenation, the degree of sorption, and the concentration. Transformation is more efficient with high
concentrations of contaminants; that is, a greater mass can be transformed per unit time per unit mass of
organisms. However, if the concentration is too high, then inhibition may result. There is ample evidence
that concentrations in the low mg/l range are not inhibitory, but just what concentrations need be of con-
cern is not yet well established. Microorganisms that can degrade other contaminants present that are
readily oxidized under aerobic conditions may also compete with methanotrophs for oxygen, and the
possible presence of this condition needs to be established.
An important property of contaminants is their distribution in the aquifer, and in particular their parti-
tioning between the aqueous and solid phases. Compounds that do not partition onto aquifer solids are
not readily susceptible to biotransformation in an aquifer by the methanotrophic process if they are
displaced by the injection of water containing the primary substrates for organism growth. It is the sorbed
fraction for which this in-situ bioremediation process has the greatest potential, as this is the fraction that is
most difficult to remove by the pump and treat method. Biotransformation lowers the aqueous concentra-
tion, thus increasing the driving force and hence rate of desorption. On the other hand, the greater the
degree of sorption, the less will be the rate of biotransformation, because strong sorption causes low
aqueous concentrations. The impact of sorption on treatment rate is thus complex; it appears that moder-
ately sorbed contaminants are the most amenable to treatment by this bioremediation approach.
In-situ biotransformation is most attractive when indigenous bacteria are used, as this avoids the
significant problem of injecting and distributing a population of contaminant-degrading bacteria. Indige-
nous organisms that can be stimulated by primary substrate injection tend to be hardy and competitive
with other organisms in the environment where they are found. This might not be true of injected organ-
isms. In addition, injection of non-indigenous organisms may raise questions about the potential envi-
ronmental consequences. In order to use indigenous organisms it must be established that they 1) are
present and distributed throughout the subsurface environment of interest, 2) can be stimulated to grow
199
-------
through injection of primary substrates and required nutrients, and can survive in that environment, and
3) have the ability to transform the contaminants of concern once stimulated.
Given the above information, the feasibility of a bioremediation system can be evaluated for situa-
tions where the subsurface environment is reasonably homogeneous. This requires a suitable numerical
model that considers 1) pumping and extraction efficiencies, power requirements, and associated
advection and dispersion of water in the aquifer, 2) a strategy for injecting the primary substrate or elec-
tron donor for energy and growth (methane) and electron acceptor (oxygen) in a manner to achieve
growth of methanotrophic bacteria throughout the treatment zone as uniformly as possible, and 3) the
rates of bacterial growth, decay, primary substrate (electron donor and acceptor) utilization, contaminant
oxidation, and contaminant desorption. It would be difficult to predict the effectiveness of a given treat-
ment scheme without such a model because of the complex interactions between the many physical,
chemical, and biological processes involved. A properly calibrated and verified model is thus an essential
tool for evaluating possible bioremediation schemes.
CONTAMINATION CHARACTERIZATION
At any site it is important initially to identify the contaminants present and to quantify their spatial
distribution. This is especially important when in-situ biorestoration is considered, since the rates of trans-
formation can be compound-specific. The laboratory and field studies reported here demonstrate that the
methanotrophic oxidation process is best suited for less chlorinated compounds such as vinyl chloride,
trans-DCE, and cis-DCE, compared to more highly substituted compounds such TCE and PCE. Currently
the processes does not seem well suited for more chlorinated compounds such as PCE.
At some sites, however, anaerobic transformations may convert the more highly chlorinated com-
pounds to less chlorinated products (Section 1). This may occur when there is co-contamination with
hydrocarbons that can act as primary substrates for anaerobic growth. The remediation effort could take
advantage of these transformations, especially if the less chlorinated transformation products have
become spatially separated from the more chlorinated parent compounds. Due to sorption processes,
spatial separation would be anticipated since the less chlorinated products tend to be less retarded
(Sections 5 and 8) and, therefore, would move down-gradient faster.
Core samples for laboratory studies should be obtained from the contaminated zone where the in-
situ treatment process is most likely to be applied, using aseptic techniques to prevent contamination by
surface microorganisms. In the laboratory, aseptic procedures should be used to obtain uncontaminated
center core material from microbiological analysis, construction of laboratory cores, and sorption studies,
as described in Section 10.
Evaluating the feasibility of in-situ remediation of contamination at a specific site requires controlled
laboratory studies. The laboratory work presented in Sections 8, 9,10, and 11 represents different types
of methods that can be applied.
COMPARISON OF LABORATORY AND FIELD RESULTS
As summarized in Section 2, the laboratory and field results of this work agreed quite well. The
mixed culture studies (Section 9), the batch column study (Section 10), and the continuous column study
(Section 11) all demonstrated that indigenous methanotrophs were present in the test zone, and that
they could be easily biostimulated. The batch and continuous flow columns, and the field study showed
agreement between the lag times before methane consumption was observed.
The three types of microbial studies also demonstrated that TCE was degraded by the stimulated
methanotrophs, consistent with the field results. The degree of transformation achieved in the batch and
continuous flow columns also agreed quite well with the field observations. These studies also demon-
strated further that less chlorinated compounds-including vinyl chloride, 1,2-DCA, and trans-DCE-would
200
-------
be degraded more rapidly than TCE. The degree of transformation of trans-DC E observed in the continu-
ous flow column was essentially identical to that observed in the field experiment.
These studies also provided qualitative information that agreed with field observations. The studies
indicated that methane probably inhibited transformation rates, as increased methane concentration did
not enhance the rates of transformation of the target compounds.
The batch and continuous soil columns, like the field, are complex systems, in which numerous
processes occur simultaneously. Thus, the determination of rate parameters from these systems requires
the application of models, as was performed for the field results (Section 13). Limited modeling has been
performed on these laboratory data. A preliminary model simulation of the response of the continuous
flow column to biostimulation (Figure 11.3) was made. The simulation and the laboratory results agreed
quite well using the same parameters as the field test simulations (Table 13.2). Thus, for this one case,
parameters derived from the model fit to laboratory column data would be expected to give reasonable
results when used in the reverse direction to predict field behavior.
It is also important to compare parameter estimates that are derived from simple, quantitative mea-
surements with well-defined systems with the field results. Comparisons between field results and the
sorption study (Section 8) and the mixed and pure culture study microbial study (Section 9) are summa-
rized below.
Sorption Studies
Sorption studies determined the extent and rates of partitioning of contaminants onto the aquifer
solids. Both sets of information are important in determining the response of a system to bioremediation.
The Kd values for the synthesized bulk sample shown in Table 8.3 are in qualitative agreement with the
response in the field based on retardation estimates given in Tables 5.4 and 5.5. The laboratory and field
data both show the following rank order in retardation: TCE > trans-DCE and cis-DCE > VC.
Retardation estimates based on the laboratory K
-------
VINYL CHLORIDE ADDITION (FOK SORPTION)
0.08
RESPONSE AT W01 S2
•nME(MYS)
Figure 14.2. Simulation of the aqueous phase (boxes) and sorted phase (crosses) concentrations of
vinyl chloride at the S2 well resulting from a two concentration step addition of vinyl chloride.
concentration; due to slow sorption, the solid phase concentration shows a gradual approach to equilib-
rium. The rapid approach to steady-state aqueous concentration was later demonstrated in the field
experiment (Figure 5.12).
Transformation Rates
The mixed and pure culture studies provide quantitative data on the rates of TCE transformation
which can be compared with values obtained independently from the model fit to the field observations.
As discussed in Section 9, the rates were dependent on growth and incubation conditions. Rates of TCE
transformation by mixed cultures based on k/Ks values (Table 9.1) ranged from 0.01 to 0.046 l/mg-d. The
model fit to the field response yielded a rate of 0.015 l/mg-d, which is within the range of laboratory esti-
mates. It is important to note that the laboratory estimates are based on the total mixed culture biomass,
while the model-fitted rate is based only on the estimated methanotrophic biomass. The model-fitted
estimate therefore agrees more closely with the lower laboratory values. The model-fitted rate is also in
the range of low values for the pure culture, 0.01 l/mg-d, where EDTA was not added (Table 9.4).
The agreement between the laboratory rates and model-fitted rate values to the field results is,
however, quite remarkable considering the differences between the laboratory and the field systems, and
the multiple parameters involved. Using the laboratory derived rate parameters, the model would have
yielded predictions for TCE transformation that would be in the range of those observed in the field.
These results indicate there is real promise of success using laboratory derived rate parameters in models
used for estimating the response of a system to bioremediation.
202
-------
MODEL SIMULATIONS OF RESTORATION SCENARIOS
Models such as the one described in Section 13 can be used as a tool in evaluating different
restoration scenarios. Basic rate parameters derived from the laboratory studies can be used as inputs for
the simulations. A modeling exercise would then permit evaluation of different restoration scenarios for a
given situation.
Modeling exercises were performed for the two treatment scenarios shown in Figure 14.1. The first
exercise represents the scheme shown in Figure 14.1 a, where treatment is applied over the complete
section of the aquifer. The first case simulates the restoration of an aquifer contaminated with a com-
pound that can be rapidly degraded by methanotrophs and is sorbed to a moderate extent (R = 12). The
simulations assumed that methane inhibits the rate of transformation, and that sorption and desorption are
rate-limited processes. Model parameters used were those for trans-DCE given in Table 13.6 and those
for methanotrophs given in Table 13.2. The simulation compares the in-situ bioremediation method to the
pump-and-treat remediation method. For the pump-and-treat case, groundwater was extracted and
treated at the surface, and reinjected. For the in-situ bioremediation, groundwater was extracted, satu-
rated with methane and oxygen at the surface, and reinjected without removing the contaminants at the
surface.
Figure 14.3 shows the results of the model simulations, where the extraction well concentration
history is shown. Cases for bioremediation using both long (10 and 15 days) and short (5 and 10 days)
alternating pulses of methane and oxygen are shown. The most effective treatment results with long
alternating pulses that distribute methanotrophic growth throughout the treatment zone. The time for
aquifer restoration is reduced by a factor of three compared to the pump-and-treat case. An additional
benefit of the bioremediation is that all the contaminants are degraded in the subsurface.
The results of another modeling exercise is illustrated in Figure 14.4. The simulation represents a
multiple-zone scheme similar to that shown in Figure 14.1b. The scheme is proposed for the treatment of
more slowly degrading compounds, such as TCE. The simulation was performed using a biotransforma-
tion rate coefficient for TCE derived from the field and laboratory experiments. A comparison is illustrated
between the pump-and-treat system with and without biostimulation, but each employing reinjection of
the extracted water. Similar hydraulic conditions were used in both cases, and sorption-desorption was
assumed to be rate-limited, as indicated by the field and laboratory studies. For the case of pump-and-
treat, groundwater was assumed to be surface treated and reinjected. Then biostimulation was added,
three zones are stimulated, each having a methanotrophic biomass equivalent to that achieved in our
studies.
Biostimulation is estimated to significantly decrease both the time for aquifer restoration and the
amount of contaminant that must be treated at the surface. For the case illustrated, about 3/4 of the TCE
would be biodegraded in-situ. For more rapidly degraded compounds, such as vinyl chloride and trans-
DCE, restoration should occur even faster.
These simulated scenarios are intended as illustrative exercises only. Any real applications would of
course necessitate careful site characterization of the kinds undertaken in this research and summarized
above in this section.
203
-------
600
D LONG PULSE
PUMP-AND-TREAT
Figure 14.3. Comparison of biorestoration versus pump-and-treat of rapidly degrading, moderately
sorbed compounds, such as trans-DCE, based upon the scheme illustrated in Figure
14.13.
(Theiitondf)
TIME (DAYS)
Figure 14.4. Comparison of biorestoration versus pump-and-treat remediation of TCE contamination with
and without biostimulation based upon the scheme illustrated in Figure 14.1b.
204
-------
REFERENCES
Anthony, C. 1975. The Biochemistry of Methylotrophic Bacteria. Sci. Prog. (London) 62:167-206.
Anthony, C. 1979. The Prediction of Growth Yield of Methylotrophs. J. Gen. Microbiol. 104:91-104.
Anthony, C. 1982. The Biochemistry of Methylotrophs. Academic Press, Inc., London.
Atwater, B. F., C. W. Hedel, and E. J. Helley. 1977. Late Quaternary Depositional History, Holocene Sea-
Level Changes, and Vertical Crustal Movement, Southern San Francisco Bay, California. U.S.G.S
Prof. Paper 1014, Washington, D.C.
Bailey, J. E., and D. F. Olis. 1986. Biochemical Engineering Fundamentals, 2nd ed. McGraw-Hill, Inc.,
New York.
Ball, W.P., and P.V. Roberts. 1985. Rate-Limited Sorption of Halogenated Aliphatics onto Sandy Aquifer
Material. Experimental Results and Implications for Solute Transport. Presented at Fall Meeting of
the American Geophysical Union, AGU, San Francisco, December.
Ball, W.P., and P.V. Roberts. 1987. Sorption Kinetics of Synthetic Organic Chemicals in Sandy Aquifer
Material. Presented at 1987 Annual Meeting, Geological Society of America, Phoenix, AZ, October.
Bail, W. P., C. Buehler, T. C. Harmon, D. M. Mackay, and P. V. Roberts. 1989. Characterization of a Sandy
Aquifer Material at the Grain Scale. Submitted to J. Contaminant Hydrol.
Barrio-Lage, G., F. Z. Parsons, R. S. Nassar, and P. A. Lorenzo. 1986. Sequential Dehalogenation of
Chlorinated Ethenes. Env. Sci. Techn. 20:96-99.
Bear, J. 1972. Dynamics of Fluids in Porous Media. American Elsevier, New York.
Bear, J. 1979. Hydraulics of Groundwater. McGraw-Hill, New York.
Beebe, R. A., J. B. Beckwith, and J. M. Honig. 1945. The Determination of Small Surface Areas by Kryp-
ton Adsorption at Low Temperature. J. Am. Chem. Soc. 67:1554-1558.
Belay, N., and L. Daniels. 1987. Production of Ethane, Ethylene, and Acetylene from Halogenated
Hydrocarbons by Methanogenic Bacteria. Appl. Environ. Microbiol. 53:1604-1610.
Borden, R. C., and P. B. Bedient. 1986. Transport of Dissolved Hydrocarbons Influenced by Oxygen-
Limited Biodegradation. 1. Theoretical Development. Water Resour. Res. 2(13):1973-1982.
Bouwer, H. 1978. Groundwater Hydrology. McGraw-Hill Book Company, New York.
Bouwer, E. J., and P. L. McCarty. 1983. Transformation of 1- and 2-Carbon Halogenated Aliphatic
Organic Compounds under Methanogenic Conditions. Appl. Environ. Microbiol. 45:1286-1294.
205
-------
Bouwer, E. J., and P. L. McCarty. 1985. Utilization Rates of Trace Halogenated Organic Compounds in
Acetate-Grown Biofilms. Biotechnol. Bioeng. 27:1564-1571.
Bouwer, E. J., B. E. Rittmann, and P. L. McCarty. 1981. Anaerobic Degradation of Halogenated 1- and 2-
Carbon Organic Compounds. Env. Sci. Technol. 15:596-599.
Bowker, A. H., and G. J. Lieberman. 1972. Engineering Statistics. Prentice-Hall, Englewood Cliffs, NJ.
Brigham, W. E. 1974. Mixing Equations in Short Laboratory Cores. Soc. Pet. Eng. J. 14:91-99.
Brock, T. D., D. W. Smith, and M. T. Madigan. 1984. Biology of Microorganisms, 4th ed. Prentice-Hall,
Englewood Cliffs, NJ.
Brunauer, S., P. H. Emmett, and E. Teller. 1938. Adsorption of Gases in Multimolecular Layers. J. Am.
Chem. Soc. 60:309-319.
Burlinson, N. E., L. A. Lee, and D. H. Rosenblatt. 1982. Kinetics and Products of Hydrolysis of 1,2-
Dibromo-3-Chloropropane. Env. Sci. Technol. 16(9):627-632.
Canonie Engineers. 1983. Subsurface Hydrogeologic Investigation, Mountain View Facility. Project
CES 82-023, June.
Carter, D. L., M. D. Heilman, and C. L. Gonzalez. 1965. Ethylene Glycol Monoethyl Ether for Determining
Surface Area of Silicate Minerals. Soil Sci. 100:356-360.
Chrysikopoulos, C. V., P. V. Roberts, and P. K. Kitanidis. 1989. One-Dimensional Solute Transport in
Porous Media with Well-to-Well Recirculation: Application to Reid Experiments. Submitted to Water
Resour. Res.
Coats, K. H., and B. D. Smith. 1964. Dead-End Pore Volume and Dispersion in Porous Media. Soc. Pet.
Eng. J. 4(1):73-84.
Colby, J., D. I. Stirling, and H. Dalton. 1977. The Soluble Methane Mono-Oxygenase of Methylococcus
capsulatus (bath). Its Ability to Oxygenate n-Alkanes, n-Alkenes, Ethers, and Alicyclic, Aromatic,
and Heterocyclic Compounds. Biochem. J. 165:395-402.
Colby, J. H., H. Dalton, and R. Whittenbury. 1979. Biological and Biochemical Aspects of Microbial
Growth on Cl Compounds. Annu. Rev. Microbiol. 33:481-517.
Crank, J. 1975. The Mathematics of Diffusion, 2nd ed. Oxford University Press, Oxford.
Griddle, C. S., J. Dewitt, and P. L. McCarty. 1989. Biotransformations of Carbon Tetrachloride by Station-
ary Phase Escherichia coli K-12 under Different Electron Acceptor Conditions. Submitted to Appl.
Environ. Microbiol.
Curtis, G.P. 1984. Sorption of Hydrophobic Organic Solutes onto Aquifer Materials: Comparisons
Between Laboratory Results and Field Observations. M.S. Thesis, Stanford University, Stanford,
CA.
Curtis, G. P., P. V. Roberts, and M. Reinhard. 1986. A Natural Gradient Experiment on Solute Transport
in a Sand Aquifer: IV. Sorption of Organic Solutes and Its Influence on Mobility. Water Resour.
Res. 22(13):2059-2067.
Dennis, J. E., D. M. Gay, and R. E. Welsch. 1981. An Adaptive Nonlinear Least-Squares Algorithm. ACM
Trans. Math. Software 7(3)-.369-383.
206
-------
Donaldson, J. R., and R. V. Tryon. 1983. Nonlinear Least-Squares Regression Using STARPAC: The
Standards Times Series and Regression Package. National Bureau of Standards Tech. Note
1068-2, Boulder, CO.
Donaldson, J. R., and R. B. Schnabel. 1987. Computational Experience with Confidence Regions and
Confidence Intervals for Nonlinear Least Squares. Technometrics 29(1):67-93.
Ferenci, T., T. Strom, and J. R. Quayle. 1975. Oxidation of Carbon Monoxide and Methane by Pseu-
domonas methanica. J. Gen. Microbiol. 91:79-91.
Fogel, M. M., A. R. Taddeo, and S. Fogel. 1986. Biodegradation of Chlorinated Ethenes by a Methane-
Utilizing Mixed Culture. Appl. Environ. Microbiol. 51:720-724.
Freeze, R. A., and J. A. Cherry. 1979. Groundwater. Prentice-Hall, Inc., New York.
Fried, J. J., and M. A. Combarnous. 1971. Dispersion in Porous Media. Adv. Hydrosci. 7:169-281.
Gaines, G.L., Jr., and P. Cannon. 1960. On the Energetics of Physically Adsorbed Films, with Particular
Reference to the Use of Krypton for Surface Area Measurement. J. Phys. Chem. 62:997-1000.
Gear, C. W. 1971. Numerical Initial Value Problems in Ordinary Differential Equations. Prentice-Hall,
Englewood Cliffs, NJ.
Ghiorse, W. C., and D. L. Balkwill. 1983. Enumeration and Morphological Characterization of Bacteria
Indigenous to Subsurface Environments. Dev. Indust. Microbiol. 24:213-224.
Goltz, M. N. 1986. Three-Dimensional Analytical Modeling of Diffusion-Limited Solute Transport. Ph.D.
Dissertation, Stanford University, Stanford, CA.
Goltz, M. N., and P. V. Roberts. 1986a. Three-Dimensional Solutions for Solute Transport in an Infinite
Medium with Mobile and Immobile Zones. Water Resour. Res. 22(7):1139-1148.
Goltz, M. N., and P. V. Roberts. 1986b. Interpreting Organic Transport Data from a Field Experiment
Using Physical Nonequilibrium Models. J. Contaminant Hydrol. 1:77-93.
Golub, G. H., and V. Pereya. 1973. The Differential of Pseudo-Inverses and Non-Linear Squares Prob-
lems Whose Variables Separate. SIAM J. Numer. Anal. 10:413-432.
Gossett, J.M. 1985. Anaerobic Degradation of C1 and C2 Chlorinated Hydrocarbons. U.S. Air Force
Report ESL-TR-85-38. National Technical Information Service, Springfield, VA.
Graydon, J. W., K. Grab, F. Zuercher, and W. Giger. 1983. Determination of Highly Volatile Organic Water
Contaminants by the Closed-Loop Gaseous Stripping Technique Followed by Thermal Desorption
of the Activated Carbon Filters. J. Chromatogr. 285:307-318.
Gregg, S. J., and K. S. W. Sing. 1982. Adsorption, Surface Area and Porosity. Academic Press, New
York.
Griesbaum, K., R. Kibar, and B. Pfeffer. 1975. Synthese und Stabilitat von 2,3-Dichlorooxiranen. Liebigs
Ann. Chem. 214-224.
Hanson, R.S. 1980. Ecology and Diversity of Methylotrophic Organisms. Adv. Appl. Microbiol. 26:3-39.
Hantush, M. S., and C. E. Jacob. 1955. Nonsteady Radial Flow in an Infinite Leaky Aquifer. Am.
Geophys. Un. Trans. 36:95-100.
207
-------
Harding Lawson Associates (HLA). 1986. Technical Memorandum: Short-and Long-Term Aquifer Tests.
Remedial Investigation/Feasibility Study, Middlefield-Ellis-Whisman Area, Mountain View, CA.
Report for U.S. EPA, Region 9, HLA Job No. 17,580,012.02.
Harrison, D. E. F. 1973. Studies on the Affinity of Methanol- and Methane-Utilizing Bacteria for Their Car-
bon Substrates. J. Appl. Bacteriol. 36:301-308.
Harrocks, D. L. 1974. Applications of Liquid Scintillation Counting. Academic Press, New York.
Harvey, R. W., R. L. Smith, and L. George. 1984. Effect of Organic Contamination upon Microbial Distri-
butions and Heterotrophic Uptake in a Cape Cod, Mass. Aquifer. Appl. Environ. Microbiol. 48:1197-
1202.
Hashimoto, I., K. B. Deshpande, and H. C. Thomas. 1964. Peclet Numbers and Retardation Factors for
Ion Exchange Columns. Ind. Eng. Chem. Fund. 3(3):213-218.
Hayat, M. A. 1981. Fixation for Electron Microscopy. Academic Press, Inc., New York.
Heijnen, J. J., and J. A. Roels. 1981. A Macroscopic Model Describing Yield and Maintenance Relation-
ships in Aerobic Fermentation Processes. Biotechnol. Bioengin. 23:739-763.
Henderson, J. E., G. R. Peyton, and W. H. Glaze. 1986. A Convenient Liquid-Liquid Extraction Method
for the Determination of Halomethanes in Water at he ppb Level. In: Identification and Analysis of
Organic Pollutants in Water, L. H. Keith, ed. Ann Arbor Science Publishers, Ann Arbor, Ml.
pp. 105-112.
Henry, S. M., and D. Grbic-Gali6. 1986. Aerobic Degradation of Trichloroethylene (TCE) by Methylotrophs
Isolated from a Contaminated Aquifer. Abstr. Q-64, Annu. Meet. Am. Soc. Microbiol. p. 294.
Henry, S. M., and D. Grbic-Gali6. 1987. Variables Affecting Aerobic TCE Transformation by Methane-
Degrading Mixed Cultures. Abstr. #401, Soc. Environ. Toxicol. Chem. Eighth Annual Meeting, p.
207.
Henry, S. M., F. Thomas, and D. Grbi6-Gali6. 1988. Electron Microscopy Studies of TCE-Degrading
Groundwater Bacteria. Abstr. #2261, Soc. Environ. Toxicol.Chem. Ninth Annual Meeting, p. 74.
Henschler, D., W. R. Hoos, H. Fetz, E. Dallmeier, and M. Metzler. 1979. Reactions of Trichloroethylene
Epoxide in Aqueous Systems. Biochem. Pharmacol. 28:543-548.
Henson, J. M., M. V. Yates, and J. W. Cochran. 1987. Metabolism of Chlorinated Aliphatic Hydrocarbons
by a Mixed Bacteria Culture Growing on Methane. Abstr. Q-97, Annu. Meet. Am. Soc. Microbiol.
p. 298.
Henson, J. M., M. V. Yates, J. W. Cochran, and D. L. Shackleford. 1988. Microbial Removal of
Halogenated Methanes, Ethanes, and Ethylenes in an Aerobic Soil Exposed to Methane. FEMS
Microbiol. Ecol. 53:193-201.
Higgins, I. J., R. C. Hammond, F. S. Sariaslani, D. J. Best, M. M. Davies, S. E. Tryhorn, and F. Taylor. 1979.
Biotransformation of Hydrocarbons and Related Compounds by Whole Organism Suspensions of
Methane-Grown Methylosinus trichosporium OB3b. Biochem. Biophys. Res. Comm. 84(2):671-
677.
Higgins, I. J., D. J. Best, R. C. Hammond, and D. Scott. 1981. Methane-Oxidizing Microorganisms.
Microbiol. Rev. 45:556.
208
-------
Hopkins, G. D., L. Semprini, P. V. Roberts, and D. M. Mackay. 1988. An Automated Data Acquisition
System for Assessing in-situ Biodegradation of Chlorinated Aliphatic Compounds. Proc. of the
Second Outdoor Conference on Groundwater Monitoring and Aquifer Restoration, NWWA, Las
Vegas, Nevada, May.
Horvath, A. L. 1982. Halogenated Hydrocarbons. M. Dekker, New York.
Horvath, R. S. 1972. Microbial Co-Metabolism and the Degradation of Organic Compounds in Nature.
Bact. Rev. 36:146-155.
Hou, C. T. 1984a. Microbiology and Biochemistry of Methylotrophic Bacteria. In: Methylotrophs:
Microbiology, Biochemistry, and Genetics, C. T. Hou, ed. CRC Press, Inc., Boca Raton, FL.
pp. 1-53.
Hou, C. T. 1984b. Other Applied Aspects of Methylotrophs. In: Methylotrophs: Microbiology, Biochem-
istry, and Genetics, C. T. Hou, ed. CRC Press, Inc., Boca Raton, FL. pp. 145-180.
Hou, C. T., R. N. Patel, A. I. Laskin, and N. Barnabe. 1979a. Microbial Oxidation of Gaseous Hydrocar-
bons: Epoxidation of C2 to C4 n-Alkenes by Methylotrophic Bacteria. Appl. Environ. Microbiol.
38:127-134.
Hou, C. T., R. Patel, A. I. Laskin, N. Barnabe, and I. Marczak. 1979b. Microbial Oxidation of Gaseous
Hydrocarbons: Production of Methyl Ketones from Their Corresponding Secondary Alcohols by
Methane- and Methanol-Grown Microbes. Appl. Environ. Microbiol. 38:135-142.
Hutzler, N. J., J. C. Crittenden, J. S. Gierke, and A. J. Johnson. 1986. Transport of Organic Compounds
with Saturated Groundwater Flow: Experimental Results. Water Resour. Res. 22:285-295.
Janssen, D. B., G. Grobben, and B. Witholt. 1987. Toxicity of Chlorinated Aliphatic Hydrocarbons and
Degradation by Methanotrophic Consortia. Proc. 4th European Congress on Biotechnology, 1987.
Vol.3, pp. 515-519.
Javandel, I., C. Doughty, and C. Tsang. 1984. Groundwater Transport: Handbook of Mathematical
Models. Water Resources Monograph SerieslO, Amer. Geophys. Union, Washington, D.C.
Johns, R. A., L. Semprini, and P. V. Roberts. 1989. Estimating Aquifer Properties by Regression Analy-
sis of Pump Test Response Data. Submitted to Ground Water.
Kennedy, S. I. T., and C. A. Fewson. 1968. Enzymes of the Mandelate Pathway in Bacterium NCIB 8250.
Biochem. J. 107:497-506.
Kissel, J. C., P. L McCarty, and R. L. Street. 1984. Numerical Simulation of Mixed-Culture Biofilm. J. Env.
Eng. (ASCE) 110(2):393-411.
Kreft, A., and A. Zuber. 1978. On the Physical Meaning of the Dispersion Equation and Its Solutions for
Different Initial and Boundary Conditions. Chem. Eng. Sci. 33:1471-1480.
Lapidus, L., and N. R. Amundson. 1952. Mathematics of Adsorption in Beds: The Effect of Longitudinal
Diffusion in Ion Exchange and Chromatographic Columns. J. Phys. Chem. 56:984-988.
Large, P. J., and J. R. Quayle. 1963. Enzyme Activities in Extracts of Pseudomonas AM 1. Biochem J.
87:386-396.
209
-------
Lee, M. D., J. M. Thomas, R. C. Borden, P. B. Bedient, C. H. Ward, and J. T. Wilson. 1988. Biorestoration
of Aquifers Contaminated with Organic Compounds. CRC Grit. Rev. in Environ. Control
18(1):29-89.
Leonard, B. P. 1979. A Stable and Accurate Modeling Procedure Based on Quadratic Upstream Model-
ing. Comput. Methods Appl. Mech. Eng. 19:59-98.
Liebler, D. C., and F. P. Guengerich. 1983. Olefin Oxidation by Cytochrome P-450: Evidence for Group
Migration in Catalytic Intermediates Formed with Vinylidene Chloride and t-1-Phenyl-1-Butene.
Biochemistry 22:5482-5489.
Lindstrom, F. T., R. Haque, V. H. Freed, and L. Boersma. 1967. Theory on the Movement of Some Her-
bicides in Soils, Linear Diffusion and Convection of Chemicals in Soils. Env. Sci. Techn. 1(7):561-
565.
Little, C. D., A. V. Palumbo, S. E. Herbes, and D. M. Genung. 1987a. Stimulation of Trichloroethylene
Biodegradation in Ground Water Samples. Abstr. Q-105, Annu. Meet. Am. Soc. Microbiol., p 299.
Little, C. D., A. V. Palumbo, S. E. Herbes, M. E. Lidstrom, R. L. Tyndall, and P. J. Gilmer. 1987b.
Trichloroethylene Biodegradation by a Methane-Oxidizing Bacterium. Abstr. 402, Annu. Meet.
Soc. Environ. Toxicol. Chem. p. 207.
Little, C. D., A. V. Palumbo, S. E. Herbes, M. E. Lidstrom, R. L. Tyndall, and P. J. Gilmer. 1988.
Trichloroethylene Biodegradation by a Methane-Oxidizing Bacterium. Appl. Environ. Microbiol.
54:951-956.
Mackay, D., and W. Y. Shiu. 1981. A Critical Review of Henry's Constants for Chemicals of Environmental
Interest. J. Phys. Chem. Ref. Data 10:1175-1199.
Mackay, D. M., W. P. Ball, and M. G. Durant. 1986. Variability of Aquifer Sorption Properties in a Field
Experiment on Groundwater Transport of Organic Solutes: Methods and Preliminary Results.
J. Contaminant Hydrol. 1:119-132.
March, J. 1985. Advanced Organic Chemistry. John Wiley & Sons, New York. pp. 804-805.
McCarty, P. L 1965. Thermodynamics of Biological Synthesis and Growth. In: Proceedings of Second
International Water Pollution Research Conference, Tokyo. Pergammon Press, New York.
McCarty, P. L. 1975. Stoichiometry of Biological Reactions. Progress in Water Technology 7(1):
157-172.
McCarty, P. L. 1984. Application of Biological Transformation in Groundwater. Proc. Second Int. Conf. on
Ground Water Quality, Tulsa, OK, March 27.
McCarty, P.L. 1988. Bioengineering Issues Related to in-situ Remediation of Contaminated Soils and
Groundwater. In: Environmental Biotechnology, G. S. Omenn, ed. Plenum Publishing Corp., New
York.
McCarty, P. L., M. Reinhard, and B. E. Rittmann. 1981. Tracer Organics in Groundwater. Environ. Sci.
Techn. 15:40-51.
McKinney, L. L., E. H. Uhing, J. L. White, and J. C. Picken. 1955. Autoxidation Products of
Trichloroethylene. J. Agric. Food Chem. 3:413-419.
210
-------
Miller, C. T., and W. J. Weber. 1986. Sorption of Hydrophobic Organic Pollutants in Saturated Soil Sys-
tems. J. Contaminant Hydrol. 1243-261.
Miller, R. E., and F. P. Guengerich. 1982. Oxidation of Trichloroethylene by Liver Microsomal
Cytochrome P-450: Evidence for Chlorine Migration in a Transition State Not Involving
Trichloroethylene Oxide. Biochemistry 21:1090-1097.
Molz, F. J., O. Guven, J. G. Melville, R. D. Crocker, and K. T. Matteson. 1986a. Peformance, Analysis, and
Simulation of a Two-Well Tracer Test at the Mobile Site. Water Resour. Res. 22(7):1031-1037.
Molz, F. J., M. A. Widdowson, and L. D. Benefield. 1986b. Simulation of Microbial Growth Dynamics
Coupled to Nutrient and Oxygen Transport in Porous Media. Water Resour. Res. 22(8):1207-1216.
Molz, F.J., and M. A. Widdowson. 1988. Internal Inconsistencies in Dispersion-Dominated Models That
Incorporate Chemical and Microbial Kinetics. Water Resour. Res. 24(4):615-619.
Morinaga, Y., S. Yamanaka, K. Takinami, and Y. Hirose. 1979a. Optimum Feeding Proportion of Methane
and Oxygen in Cultivation of the Obligate Methane-Utilizing Bacterium, Methylomonas flagellata, in
Batch Culture. Agricult. Biol. Chem. 43:2447-2451.
Morinaga, Y., S. Yamanaka, K. Takinami, and Y. Hirose. 1979b. Methane Metabolism of the Obligate
Methane-Utilizing Bacterium, Methylomonas flagellata, in Methane-Limited and Oxygen-Limited
Chemostat Culture. Agricult. Biol. Chem. 43:2453-2458.
Nation, J. L. 1983. A new Method Using Hexamethyldisilazane for preparation of soft insect tissues for
scanning electron microscopy. Stain Tech. 58:347-351.
Nelson, M. J. K., S. O. Montgomery, E. J. O'Neill, and P. H. Pritchard. 1986. Aerobic Metabolism of
Trichloroethylene by a Bacterial Isolate. Appl. Environ. Microbiol. 52:383-384.
Nelson, M. J. K., S. O. Montgomery, W. R. Mahaffey, and P. H. Pritchard. 1987. Biodegradation of
Trichloroethylene and the Involvement of an Aromatic Biodegradative Pathway. Appl. Environ.
Microbiol. 53:949-954.
Nelson, M. J. K., S. O. Montgomery, and P. H. Pritchard. 1988. Trichloroethylene Metabolism by
Microorganisms That Degrade Aromatic Compounds. Appl. Environ. Microbiol. 54:604-606.
Nkedi-Kizza, P., J. W. Biggar, M. Th. van Genuchten, P. J. Wierenga, H. M. Selim, J. M. Davidson, and
D. R. Nielsen. 1983. Modeling Tritium and Chloride 36 Transport Through an Aggregated Oxisol.
Water Resour. Res. 19(3):691-700.
Norris, J. R., and H. Swain. 1971. In: Methods in Microbiology, Vol 5A, J. R. Norris and D. W. Ribbons,
eds. Academic Press, London, pp. 105-133.
Oliveira, L., A. Burns, T. Bisalputra, and K. Yang. 1983. The Use of Ultra-Low Viscosity Medium
(VCD/HXSA) in the Rapid Embedding of Plant Cells for Electron Microscopy. J. Microscopy
132(2):195-202.
Parsons, F. and G. B. Lage. 1985. Chlorinated Organics in Simulated Groundwater Environments. J. Am.
Water Works Assoc. 77(5):52-59.
Patel, R. N. 1984. Methane Monooxygenase from Methylobacterium sp. Strain CRL-26. In: Microbial
Growth on C1 Compounds: Proceedings of the 4th International Symposium, R. L. Crawford and
R. S. Hanson, eds. American Society for Microbiology, Washington, D.C. pp. 83-90.
211
-------
Patel, R. N., C. T. Hou, A. I. Laskin, A. Felix, and P. Derelanko. 1979. Microbial Oxidation of Gaseous
Hydrocarbons. II. Hydroxylation of Alkanes and Epoxidation of Alkenes by Cell-Free Paniculate
Fractions of Methane-Utilizing Bacteria. J. Bacteriol. 139:675-679.
Patel, R. N., C. T. Hou, A. I. Laskin, and A. Felix. 1982. Microbial Oxidation of Hydrocarbons: Properties
of a Soluble Methane Monooxygenase from a Facultative Methane-Utilizing Organism Methylobac-
terium sp. Strain CRL-26. Appl. Environ. Microbiol. 44:1130-1137.
Press, F., and R. Siever. 1974. Earth. Freeman & Co., San Francisco.
Prior, S. D., and H. Dalton. 1985. The Effect of Copper Ions on Membrane Content and Methane
Monooxygenase Activity in Methanol-Grown Cells of Methylococcus capsulatus (bath). J. Gen.
Microbiol.131:155-163.
Quayle, J. R. 1972. The Metabolism of One-Carbon Compounds by Microorganisms. Adv. Microb.
Physiol. 7:119-203.
Rao, P. S. C., D. E. Rolston, R. E. Jessup, and J. M. Davidson. 1980. Solute Transport in Aggregated
Porous Media: Theoretical and Experimental Evaluation. Soil Sci. Soc. Am. J. 44:1139-1146.
Rasmuson, A., and I. Neretnieks. 1980. Exact Solution of a Model for Diffusion in Particles and Longitu-
dinal Dispersion in Packed Beds. AlChE J. 26(4):686-690.
Raymond, R. L. 1974. Reclamation of Hydrocarbon Contaminated Groundwaters. United States Patent
Office, Pat. No. 3,846,290, Nov. 5, 1974.
Raymond, R. L., V. W. Jamison, and J. O. Hudson. 1976. Beneficial Stimulation of Bacterial Activity in
Groundwaters Containing Petroleum Products. AlChE Symp. Series 73:390.
Rittmann, B. E., and P. L. McCarty. 1980. Model of Steady-State Biofilm Kinetics. Biotechnol. Bioeng.
22:2343-2357.
Roberts, P. V., and D. M. Mackay, eds. 1986. A Natural Gradient Experiment on Solute Transport in a
Sand Aquifer. Technical Report No. 292, Department of Civil Engineering, Stanford University,
Stanford, CA.
Roberts, P. V., P. L. McCarty, M. Reinhard, and J. Schreiner. 1980. Organic Contaminant Behavior Dur-
ing Groundwater Recharge. J. WPCF 52:161-172.
Roberts, P. V., M. N. Goltz, and D. M. Mackay. 1986. A Natural Gradient Experiment on Solute Transport
in a Sand Aquifer. III. Retardation Estimates and Mass Balances for Organic Solutes. Water
Resour. Res. 22(13):2047-2058.
Roels, J. A. 1983. Energetics and Kinetics in Biotechnology. Elservier Biomed Press, New York.
Schwarzenbach, R. P., and J. Westall. 1981. Transport of Non-Polar Organic Compounds from Surface
Water to Groundwater. Env. Sci. Technol. 15(11):1360-1367.
Semprini, L., P. V. Roberts, G. D. Hopkins, and D. M. Mackay. 1988. A Field Evaluation of in-situ
Biodegradation for Aquifer Restoration. EPA/600/S2-87/096. NTIS No. PB 88-130 257/AS.
Springfield, VA.
Semprini, L., P. V. Roberts, G. D. Hopkins, and P.L. McCarty. 1989. A Field Evaluation of in-situ
Biodegradation Methodologies for Aquifer Contaminated with Chlorinated Aliphatic Compounds:
212
-------
Part 2: The Results of Biostimulation and Biotransformation Experiments. Submitted to Ground
Water.
Siegrist, H., and P. L. McCarty. 1987. Column Methodologies for Determining Sorption and Biotransfor-
mation Potential for Chlorinated Aliphatic Compounds in Aquifers. J. Contaminant Hydro). 2:31-50.
Sing, K. S. W., and D. Swallow. 1960. Krypton Adsorption and Surface Area of Silica. J. Appl. Chem.
10:171-175.
Speitel, G. E., Jr., K. Dovantzis, and F. A. DiGiano. 1987. Mathematical Modeling of Bioregeneration of
GAC Columns. J. Env. Eng. (ASCE) 113(1):32-48.
Stehfest, H. 1970. Algorithm 368 Numerical Inversion of Laplace Transforms. J. ACM 13(1):47-49.
Stirling, D. I., and H. Dalton. 1979. The Fortuitous Oxidation and Cometabolism of Various Carbon Com-
pounds by Whole-Cell Suspensions of Methylococcus capsulatus (bath). FEMS Microbiol. Lett.
5:315-318.
Stirling, D. I., J. Colby, and H. Dalton. 1979. A Comparison of the Substrate and Electron-Donor
Specificities of the Methane Monooxygenase from Three Strains of Methane-Oxidizing Bacteria.
Biochem. J. 177:361-364.
Suidan, M. T., B. E. Rittmann, and U. K. Traegner. 1987. Criteria Establishing Biofilm Kinetic Types.
Water Resour. Res. 21 (4)-.491-498.
Tonge, G. M., D. E. F. Harrison, and I. J. Higgins. 1975. Properties and Partial Purification of the Methane-
Oxidizing Enzyme System from Methylosinus trichosporium. FEBS Lett. 58:293-299.
U.S. Environmental Protection Agency. 1984. National Revised Primary Drinking Water Regulations,
Volatile Synthetic Organic Chemicals in Drinking Water; Proposed Rulemaking. Fed. Reg.
49:24329-24355.
U.S. Environmental Protection Agency. 1985. National Primary Drinking Water Regulations; Volatile
Synthetic Organic Chemicals. Final Rule 40 CFR 141. Fed. Reg. 50:46880-46901.
Valocchi, A. J. 1986. Effect of Radial Flow on Deviation from Local Equilibrium During Sorting Solute
Transport through Homogeneous Soils. Water Resour. Res. 22:1693-1701.
Valocchi, A. J., and P. V. Roberts. 1983. Attenuation of Groundwater Contaminant Pulses. J. Hydrol.
Eng. (ASCE) 109(12):1665-1682.
van Genuchten, M. Th. 1981. Non-Equilibrium Transport Parameters from Miscible Displacement
Experiments. Research Report 119, U.S. Salinity Lab., Riverside, CA.
van Genuchten, M.Th. 1985. A General Approach for Modeling Solute Transport in Structured Soils. In:
Hydrologeology of Rocks of Low Permeability, Proc. 17th Int. Congress, Int. Association of Hydro-
geologists, Tucson, AZ, January.
van Genuchten, M. Th., and W. J. Alves. 1982. Tech. Bull. No. 1661, U.S. Department of Agriculture.
p. 151.
van Genuchten, M. Th., and P. J. Wierenga. 1976. Mass Transfer Studies in Sorbing Porous Media:
I. Analytical Solutions. Soil Sci. Soc. Am. J. 40(4):473-480.
213
-------
van Genuchten, M. Th., J. M. Davidson, and P. J. Wierenga. 1974. An Evaluation of Kinetic and Equilib-
rium Equations for the Prediction of Pesticide Movement through Porous Media. Soil Sci. Soc. Am.
Proc. 38:29-35.
Vogel, T. M., and P. L. McCarty. 1985. Biotransformation of Tetrachloroethylene to Trichloroethylene,
Dichloroethylene, Vinyl Chloride, and Carbon Dioxide under Methanogenic Conditions. Appl.
Environ. Microbiol. 49:1080-1083.
Vogel, T. M. and P. L. McCarty. 1987. Abiotic and Biotic Transformations of 1,1,1 -Trichtoroethane under
Methanogenic Conditions. Env. Sci. Techn. 21:1208-1213.
Vogel, T. M., C. S. Griddle, and P. L. McCarty. 1987. Transformations of Halogenated Aliphatic Com-
pounds. Env. Sci. Techn. 21(8):722-736.
Webster, J. J., G. J. Hampton, J. T. Wilson, W. C. Ghiorse, and F. R. Leach. 1985. Determination of
Microbial Cell Numbers in Subsurface Samples. Ground Water 23(1 ):17-25.
Westrick, J. J., J. W. Mello, and R. F. Thomas. 1984. The Groundwater Supply Survey. J. Am. Water
Works Assoc. 76(5):52-59.
Whittenbury, R., K. C. Phillips, and J. F. Wilkinson. 1970. Enrichment, Isolation, and Some Properties of
Methane-Utilizing Bacteria. J. Gen. Microbiol. 24:225-233.
Wilkinson, T. G., H. H. Topiwala, and G. Hamer. 1974. Interactions in a-Mlxed Bacterial Population Growing
on Methane in Continuous Culture. Biotechnol. Bioeng. 16:41-59.
Wilson, J. T., and B. H. Wilson. 1985. Biotransformation of Trichloroethylene in Soil. Appl. Environ.
Microbiol. 29:242-243.
Wilson, J. T., J. F. McNabb, D. L. Balkwill, and W. C. Ghiorse. 1983. Enumeration and Characterization of
Bacteria Indigenous to a Shallow Water Table Aquifer. Ground Water 21 (2) :134-141.
Wilson, J. T., L. F. Leach, M. J. Henson, and J. N. Jones. 1986. In-situ Biorestoration as a Ground-Water
Remediation Technique. Ground Water Monit. Rev., Fall, p 56.
Wilson, J. T., S. Fogel, and P. V. Roberts. 1987. Biological Treatment of Trichloroethylene in situ. In:
Proceedings: Symposium on Groundwater Contamination, ASCE National Convention, Atlantic
City, NJ, April 27-30.
Wolfe, R. S., and I. J. Higgins. 1979. Microbial Biochemistry of Methane--a Study in Contrasts. Int. Rev.
Biochem. 21:267-300.
Wu, S-C., and P. M. Gschwend. 1986. Sorption Kinetics of Hydrophobic Organic Compounds to Natural
Sediments and Soils. Env. Sci. Techn. 20:717-725.
Wu, S-C., and P. M. Gschwend. 1988. Numerical Modeling of Sorption Kinetics of Organic Compounds
to Soil and Sediment Particles. Water Resour. Res. 24:1373-1383.
*U.S. GOVERNMENT HUNTING OFFICE: 1989 -6i»8- 16700306
214
------- |