United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/R-92/247
December 1992
c/EPA
LNAPL Distribution and
Hydrocarbon Vapor
Transport in the
Capillary Fringe
-------
EPA/600/R-92/247
December 1992
LNAPL DISTRIBUTION AND HYDROCARBON VAPOR TRANSPORT
IN THE CAPILLARY FRINGE
c by
\
(\i David W. Ostendorf, Ellen E. Moyer, Robin J. Richards
Erich S. Hinlein, Yuefeng Xie, and R.V. Rajan
\ Environmental Engineering Program
Civil Engineering Department
University of Massachusetts
V Amherst, MA 01003
Project No. CR-816821
Project Officer
Jong Soo Cho
Processes and Systems Research Division
Robert S. Kerr Environmental Research Laboratory
Ada, OK 74820
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ADA, OK 74820
Printed on Recycled Paper
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Hoof
Chicago, IL 60604-3590
-------
DISCLAIMER NOTICE AND QUALITY ASSURANCE STATEMENT
The Information in this document has been funded wholly or in part by the
United States Environmental Protection Agency under Cooperative Agreement CR-
816821 to the University of Massachusetts at Amherst. It has been subject 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.
All research projects making conclusions or recommendations based on
environmentally related measurements and funded by the Environmental
Protection Agency are required to participate in the Agency Quality Assurance
Project Plan. Information on the plan and documentation of the quality
assurance activities and results are available from the Principal
Investigator, Dr. David Ostendorf.
ii
-------
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 zone, and the saturated zones of the
subsurface environment; (b) define the processes 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 field sampling, laboratory experiments, and
mathematical models designed to determine the distribution, fate, and
transport of aviation gasoline liquid and vapor constituents through the
residually contaminated capillary fringe.
Clinton W. Hall
Director
Robert S. Kerr Environmental
Research Laboratory
iii
-------
ABSTRACT
We measured and modeled the vertical distribution of water and light
nonaqueous phase liquid (LNAPL) from a well documented aviation gasoline spill
at the US Coast Guard Air Station in Traverse City, Michigan. Two field
sampling methods for the determination of LNAPL content were presented.
Existing models of the free and residual LNAPL profiles described the data
with calibrated error standard deviations ranging from 14 to 23% in magnitude.
A field trapping protocol was also developed to document the distribution of
LNAPL constituent vapors in the contaminated capillary fringe over a 0.03 m
spacing.
The evaporation of 2,2,4 trimethylpentane and 2,2,5 trimethylhexane
vapors from the LNAPL was measured in the laboratory under diffusive and
advective conditions to simulate soil venting and sparging. A common source
term was used in simple and accurate models of each process: calibration
error standard deviations for the diffusive and advective profiles varied from
4 to 9% and 6 to 16%, respectively. The data and theory suggested that the
source strength varied with the LNAPL remaining in the core but was
independent of the air flow rate. This finding indicated that lower soil
venting or sparging flow rates were in principle as effective as higher rates
in stripping gasoline vapors from contaminated soils.
We also generated data documenting the biodegradation of hydrocarbon
vapors in soil microcosms obtained aseptically from the site. The microcosms
featured an exterior water seal design that proved successful in the difficult
task of vapor retention, and Michaelis-Menten type kinetics were fit to the
data under substrate limiting conditions with a calibrated error standard
deviation that varied from 7 to 20%. The microcosm based kinetics were input
to an existing model of coupled hydrocarbon and oxygen transport in an
accurate prediction of independent field observations of the vapor
concentrations.
This report was submitted in fulfillment of project no. CR-816821 by the
University of Massachusetts under the sponsorship of the U.S. Environmental
Protection Agency. This work covers a period from June 1990 to June 1992, and
work was completed as of June 1992.
iv
-------
CONTENTS
Disclaimer Notice and Quality Assurance Statement ii
Foreword iii
Abstract iv
Figures viii
Tables x
Abbreviations and Symbols xi
Acknowledgments xvii
1 Introduction 1
Scope and engineering relevance 1
Report organization 2
2 Conclusions and Recommendations 3
Conclusions 3
Recommendations 6
3 Materials , Methods , and Experimental Procedures 8
General site and plume characteristics 8
LNAPL profiles 10
Core barrel extrusion 11
Core sleeve partitioning 13
Extraction and gas chromatography 14
LNAPL profile stations 19
Field trapping of hydrocarbon vapors 21
Vapor sampling technology 21
Intact core sleeve acquisition 23
Field trapping protocol 24
Sorbent tube desorption and gc analysis 25
Core sleeve diffusion 26
Vapor transport in the unsaturated zone 26
Gas chromatography 28
Transport test apparatus 30
Core sleeve advection 31
Transport test apparatus 31
Experiments and displaced pore volumes 32
-------
Soil microcosms 34
Overview of soil microcosms 34
Sampling protocol and laboratory microcosm design 35
Dosage and gas chromatography 37
Supporting measurements 39
Gas chromatography/mass spectrometry 39
Moisture content 40
Grain size distribution 43
4 Mathematical Analysis 44
LNAPL profiles 44
Overview of LNAPL distribution in the capillary fringe.44
Free moisture retention 45
LNAPL entrapment 48
Typical profiles of water and LNAPL 49
Core sleeve diffusion 51
Hydrocarbon concentration profile model 52
Saturated vapor concentration iteration and
diffusive flux model 55
Diffusion in uniform and nonuniform soils 59
Core sleeve advection and Michaelis-Menten type
kinetic calibration 59
Core sleeve advection 60
Advection in uniform and nonuniform soils 62
Michaelis-Menten type kinetic calibration 65
5 Results and Discussion 68
LNAPL field sampling precision 68
Online headspace sampling 68
Comparative LNAPL field sampling precision 70
Discussion 74
LNAPL profile model calibration 76
Model calibration 77
Discussion 79
Field trapping of hydrocarbon vapors 84
Chromatography 84
Method evaluation 85
Comparision with diffusion model 87
Core sleeve diffusion 89
vi
-------
Moisture content and saturated vapor concentration 90
Efflux measurements and source strength 93
Hydrocarbon vapor concentration profiles 94
Soil venting considerations 96
Core sleeve advection 97
Concentration profiles 98
Source strengths and advective fluxes 103
Sparging implications 105
Soil microcosms 106
Abiotic controls 106
Calibrated Michaelis-Menten type kinetics 108
Field test of Michaelis-Menten type kinetics Ill
References 114
Appendix-Publications and Presentations 121
vii
-------
FIGURES
3.1 Plan view of US Coast Guard Air Station LNAPL plume 9
3.2 Cumulative density function (CDF) of grain size (d) at representative
sampling location, Traverse City research site 10
3.3 Field sampling devices for LNAPL determination 11
3.4 Gas chromatography of methylene chloride extract solutions 17
3.5 Location of sampling stations at Traverse City, Michigan 20
3.6 Intact core sleeve and transport test stand for diffusion experiments.22
3.7 Method detection limit runs for 2,2,4 TMP and 2,2,5 TMH 29
3.8 Transport test stand for core sleeve advection experiments 32
3.9 Soil microcosm for hydrocarbon vapor degradation studies 36
3.10 Typical chromatograms for vapor degradation runs,
standard and microcosm 38
3.11 Normalized El spectra from gas chromatograph/mass spectrometer runs...41
3.12 Moisture content (by mass) as a function of drying time at 104 deg C..42
3.13 Moisture content (by volume) as a function of elevation
in a typical borehole 42
4.1 Definition sketch for three phase fluid distribution in soil 46
4.2 Typical profiles of fluid saturation 50
4.3 Water "aggregate" schematization, showing local mass transport
vapor concentration gradient 53
4.4 Integral function I 56
4.5 Typical vapor concentration profiles for advective stripping 62
4.6 Simulated performance of air stripping of LNAPL constituent
from uniform soil under high and low air flows 64
4. 7 Michaelis -Menten type kinetics 66
4.8 Calibrated degradation data from soil microcosm 67
5.1 Barrel extruded LNAPL data and combustible
hydrocarbon meter readings for Stations SOBS and 50BT 71
5.2 Total saturation (S) profiles, with data (circles)
and predictions (curves) 80
5.3 LNAPL saturation (SL) profiles, with data (circles)
and predictions (curves) 81
viii
-------
5.4 Vertical distribution of LNAPL volatility fractions
in a segmented core sleeve 83
5.5 Typical chromatograms for trapped vapor standard and sample 86
5.6 Moisture and hydrocarbon vapor profiles at Station 50CL 88
5.7 Moisture and 2,2,4 IMP efflux at Station 50CE 92
5.8 Observed (symbols) and predicted (curves) vapor concentrations
(2,2,4 TMP and 2,2,5 TMH) for intact core diffusion experiment 95
5.9 Observed (symbols) and predicted (curves) vapor concentrations
for core sleeve advection at 3 ml/min 99
5.10 Observed (symbols) and predicted (curves) vapor concentrations
for core sleeve advection at 5 ml/min 100
5.11 Observed (symbols) and predicted (curves) vapor concentrations
for core sleeve advection at 10 ml/min 101
5.12 Calibrated (curves) and observed (symbols) advective stripping of
2,2,4 TMP from core sleeve 50CL 104
5 .13 Abiotic control performance 107
5.14 fiiodegradation of total hydrocarbon vapor concentrations in soil
microcosms 109
5.15 Field test of hydrocarbon vapor degradation model 112
ix
-------
TABLES
3 .1 Laboratory standard for LNAPL 15
3.2 Empirical constants for calculation of pure saturated
vapor pressure 16
3.3 Standard constituents of typical LNAPL sample 18
3 .4 Laboratory standard for aviation gasoline vapor 25
3 . 5 Test conditions sleeve diffusion runs 31
3 . 6 Test conditions sleeve advection runs 33
3.7 Summary of gas chromatograph/mass spectrometer instrument parameters..40
4.1 Profile parameter values for uniform and nonuniform soils 51
4 . 2 Saturated vapor concentration iteration 57
4.3 Advective stripping parameter values for uniform
and nonuniform soils 63
5.1 Observed headspace concentrations, moisture and LNAPL content
Core Barrel SOBS Jars 69
5.2 Observed headspace concentrations, moisture and LNAPL content
Core Barrel 50BT Jars 70
5.3 Observed moisture and LNAPL content-Core Sleeve SOBS Profile 1 72
5.4 Observed moisture and LNAPL content-Core Sleeve SOBS Profile 2 73
5.5 Observed moisture and LNAPL content-Core Sleeve 50BT Profile 1 74
5.6 Observed moisture and LNAPL content-Core Sleeve 50BT Profile 2 75
5.7 LNAPL sample precision 76
5.8 Common parameter values for LNAPL calibration and testing 77
5 . 9 Total saturation calibration results 79
5.10 Standard and sample chromatograms for trapped vapors 85
5.11 Observed vapor concentrations, 2,2,4 trimethylpentane 91
5 .12 Observed vapor concentrations , 2,2,5 trimethylhexane 91
5 .13 Results of sleeve diffusion runs 93
5 .14 Intact core sleeve advection data 98
5 .15 Results of sleeve advection runs 102
5.16 Results of microcosm analyses at Station 50CL 110
-------
ABBREVIATIONS AND SYMBOLS
A FID area, area units
o
A, Core sleeve gross cross sectional area, 1
2
B Depth integrated LNAPL content (Equation 4.24a), m/1
2
BO Depth integrated LNAPL content at start of stripping, m/1
b Depth below ground surface, 1
bL Depth to LNAPL table (Figure A.I), 1
^LMAX Location of maximum free LNAPL saturation, 1
bj. Depth to top of free LNAPL (Figure 4.1), 1
k TN Annual minimum depth to top of free LNAPL, 1
by Depth to water table (Figure 4.1), 1
KjMAX Annual maximum water table depth, 1
1>WM Annual minimum water table depth, 1
C FID response factor, m/1 -area units
C-r Empirical constants
CDF Cumulative density function
CME Central Mine and Equipment Company
2
D Gaseous diffusivity, 1/t
2
DA Free air diffusivity, 1/t
A
2
D, Aggregate diffusivity, 1/t
o
DR Reference diffusivity, 1/t
D Dimensionless grain size (Equation 3.12b)
d Grain size, 1
dj. Mean grain size, 1
2
F. Vapor flux from he-adspace, m/1 -t
xi
-------
Vapor flux at sleeve exit, m/1 -t
rL
FID
GC/MS
g
H
H*
HE
HL
H^
He
"SP
H
ST
Hrp
HP
h
hT
KH*)
ID
K
KD
k
L
LNAPL
M
Local vapor flux from LNAPL source, m/1 -t
Flame ionization detector
Gas chromatograph/mass spectrometer
2
Gravitational acceleration, 1/t
3
Vapor concentration, m/1
Dimensionless vapor concentration (Equation 5.23b)
3
Vapor concentration in effluent port, m/1
Vapor concentration at sleeve exit, m/1
Initial vapor concentration, m/1
Dimensionless vapor concentration at bottom of unsaturated zone
3
Saturated vapor concentration, m/1
3
Pure saturated vapor concentration, m/1
Total saturated vapor concentration, all constituents, m/1
Total vapor concention, all constituents, m/1
Hewlett - Packard
3
Relative vapor concentration (Equation 4.19b), m/1
Relative vapor concentration at column exit (Equation 4.20b), m/1"
Integral function (Equation 4.21b)
Integral function (Equation 5.23c)
Internal diameter, 1
Half saturation constant (Equation 4.38), m/1
Distribution coefficient, 1 /m
Source strength, t
Sleeve length, 1
Light nonaqueous phase liquid
Moisture content by mass (Equation 3.la)
xii
-------
MM
MS
MDL
m
mR
N
n
0
PR
PSP
PW
PWM
Q
R
RD
RD'
Ru
RSKERL
r
rL
rw
S
SL
SLF
SLMAX
SLFMAX
SLR
Methylene chloride mass, m
Vet soil mass, m
Method detection limit (Equation 3.5), m/1
Molar mass, m/mole
Reference molar mass, m/mole
Number of sample points
Porosity (Equation 3.6c)
3
Oxygen concentration, m/1
Oxygen concentration at the base of the root zone, m/1"
2
Reference vapor pressure, m/l-t
2
Pure saturated vapor pressure, m/l-t
2
Water pressure, m/l-t
2
Water pressure at depth b^, m/l-t
Air flow rate, 1 /t
Aggregate spacing (Figure 4.3), 1
Retardation factor (Equation 4.39)
Apparent retardation factor (Equation 4.40b)
Universal gas constant (Equation 3.3b)
Robert S. Kerr Environmental Research Laboratory
Mean pore radius (Equation 4.4a), 1
LNAPL droplet radius (Figure 4.3), 1
Water aggregate radius (Figure 4.3), 1
Total saturation (Equation 4.2a)
Total LNAPL saturation (Equation 4.13)
Free LNAPL saturation (Equation 4.2b)
Maximum LNAPL saturation
Maximum free LNAPL saturation
Residual (trapped) LNAPL saturation
xiii
-------
*LRMAX
W
JWMAX
3WMIN
T
t
t.
VGA
w
WI
X
z
Q
•v
H
Maximum residual LNAPL saturation
(Apparent effective) water saturation (Equation 4.la)
Annual maximum water saturation
Annual minimum water saturation
LNAPL content by mass (Equation 3.1b)
Time, t
Reaction time (Equation 5.22), t
Diffusive time scale (Equation 4.37a), t
Duration of advective flow, t
Acclimation time, t
3
Maximum reaction rate (Equation 4.38), m/1 -t
o
Volume of empty voids in intact core sleeve (Equation 3.7b), 1
Volatile organic analysis
Specific discharge (Equation 3.7c), 1/t
Specific discharge during experiment, 1/t
Dimensionless relative temperature (Equation 3.2b)
Elevation above water table, 1
Brooks and Corey (1966) pore size uniformity exponent (Equation 3.8a)
Grain size uniformity exponent (Equation 3.11a)
van Genuchten (1980) scaling factor for LNAPL/air interface
(Equation 4.3), 1
van Genuchten (1980) scaling factor for water/air interface
Equation 4.5), I"1
Stoichiometric ratio
van Genuchten (1980) pore size uniformity exponent (Equation 4.3)
Annual fluctuation amplitude (Equation 4.10), 1
H error (Equation 5.15)
Microcosm error (Equation 5.20)
0W error (Equation 3.11a)
Mean saturation error (Equation 5.5a)
xiv
-------
5W Mean 8,, error (Equation 3.lie)
t Contaminated soil interval, 1 (Equation 4.7c)
f. Headspace length scale, 1
A
fs Soil layer thickness, 1
f,, Unsaturated zone thickness, 1
r) Hysteretical trapping factor (Equation 4.11)
6 Air porosity (Equation 3.7d)
6, LNAFL content by volume (Equation 4.1b)
6^ Residual LNAPL content (Equation 4.9b)
0y Water content by volume (Equation 3.8b)
flyr. Irreducible moisture content
K Efflux coefficient for advection (Equation 4.33a), m0'33/l°'67-t
A Maximum elevation of LNAPL above water table, m
£ Number of displaced pore volumes (Equation 3.7a)
p Water density, m/1
pfi Bulk density (Equation 5.3b), m/1
PL LNAPL density, m/13
Ps Solid grain density, m/1
£7U H error standard deviation
n
o
a,. LNAPL/air surface tension, m/t
a,. Microcosm error standard deviation
aMDL Method detection limit standard deviation (Equation 3.6b), m/1
Of, Saturation error standard deviation (Equation 5.5b)
aw 0y error standard deviation (Equation 3.7c)
o
<7y. Water/air surface tension, m/t
o
aWL Water/LNAPL surface tension, m/t
T Temperature, deg
xv
-------
Tn Reference temperature, deg
X Mole fraction
^B Bubbling pressure head, 1
3
0 Source strength, m/1 -t
u Efflux coefficient for diffusion (Equation 4.24c), m°'33/l°'67-t
2,2,4 THP 2,2,4 trimethylpentane
2,2,5 TMH 2,2,5 trimethylhexane
xvi
-------
ACKNOWLEDGMENTS
This project was a cooperative effort that relied on the assistance of
numerous researchers in the field and laboratory. We appreciate the site
logistical and field support provided by the U.S. Coast Guard through its
engineers with the Traverse Group, Inc. and Solar Universal Technologies, Inc.
The field drilling and extrusion were performed with the capable assistance of
the Robert S. Kerr Environmental Research Laboratory drill rig crew, and the
efforts of Montie Fraser, Frank Beck, Lowell Leach, and Alton Tweedy were
critical to the successful acquisition of samples. Thanks, too, go to Don
Kampbell and Michael Cook of the Laboratory for their analytical support, and
to Dorothy Bertino of ManTech Environmental Technology Inc. for her quality
assurance audit and report review. The Authors acknowledge the efforts of the
other report reviewers and of course, the Project Officer, Jong Soo Cho, who
coordinated all the efforts on this interdisciplinary project.
xvii
-------
SECTION 1
UTRODDCTim
SCOPE AND ENGINEERING RELEVANCE
This final report describes work done under the Cooperative Research
Agreement "Air Stripping of Hydrocarbons from Residually Contaminated Soil
Cores" (CR 816821) between the University of Massachusetts at Amherst and the
Robert S. Kerr Environmental Research Laboratory (RSKERL) of the US
Environmental Protection Agency. Two complementary field sampling methods
were studied for the determination of the vertical distribution of separate
phase liquid aviation gasoline content in a 23 year old spill at the US Coast
Guard Air Station in Traverse City, Michigan. We measured the concentration
of hydrocarbon vapors in the field as well, using a field trapping protocol on
core sleeve samples removed intact from the contaminated capillary fringe.
The diffusive and advective release characteristics of the gasoline vapors
from the contaminated soil were measured and modeled in the intact core
sleeves, and the degradation of the vapors was analyzed in a series of soil
microcosms. These latter studies elucidated the mechanics of soil venting,
sparging, and gaseous bioremediation, respectively.
The observed distribution, evaporative release, and gaseous degradation
of light nonaqueous phase liquids (LNAPLs) in the capillary fringe and
overlying unsaturated zone bore upon the larger problem of organic
contamination of the subsurface environment, a phenomenon of documented
importance due to the widespread use of organic fluids and solutions in Post
World War II America [National Academy of Sciences (1984), Wise and
Fahrenthold (1981)]. Pesticides, herbicides, solvents, gasoline, heating oil,
creosote, and transmission fluid are typical examples, and their eventual
distribution in waste lagoons, agricultural runoff, landfills, spills, and
buried drum sites commonly leads to subsurface pollution [Schwille (1967),
Jury et al. (1986), Reinhard et al. (1984), Kitunen et al. (1987)].
-------
REPORT ORGANIZATION
The report was prepared in accordance with US Environmental Protection
Agency guidelines. Thus, this introductory chapter is followed by a brief
summary of conclusions and recommendations. Chapter 3 describes the
materials, methods, and protocol leading to the project data base for the
following experiments:
LNAPL profiles
Field trapping of hydrocarbon vapors
Intact core sleeve diffusion
Intact core sleeve advection
Soil microcosms
Mathematical models are put forward for these data in Chapter 4. We cite
relevant literature from other experimental and modeling investigations in the
methods and mathematical chapters, respectively, and the results are discussed
in Chapter 5. The Abstract, which provides an overview of the research,
appears at the beginning of the document.
-------
SECTIOM 2
OOHCLUSIOHS AMD RBOOHMKHDATIOHS
We measured and modeled the vertical distribution of water and light
nonaqueous phase liquid (LNAPL) in the capillary fringe of a fine uniform sand
in this Cooperative Agreement. Hydrocarbon vapors from the LNAPL were
considered in detail as well, through separate experimental and mathematical
studies of field trapping, diffusive and advective stripping, and microcosm
degradation. A well documented aviation gasoline spill at the US Coast Guard
Air Station provided the field samples used for our investigation through
cooperative drilling with the US Environmental Protection Agency. Gas
chromatography (with compound identification and elution order confirmation by
gas chromatography/mass spectrometry) was used on methylene chloride extracts,
vapor injections, and thermal trap desorptions to analyze the hydrocarbon
content of the soil and vapor samples that comprised the project data base.
CONCLUSIONS
Two field sampling methods for the determination of LNAPL content were
developed. The first method featured field extrusion of core barrels into
Mason jars, while the second consisted of laboratory partitioning of intact
stainless steel core sleeves inserted into the core barrels. The methods
yielded consistent estimates of the LNAPL profile and were complementary,
since rapid core barrel extrusion provided depth integrated data, while time
intensive segmenting of sleeves resolved the detailed vertical structure of
the separate phase content. Online headspace sampling of extruded samples
identified contaminated soil intervals in an effective and timely fashion.
Existing models of the free [Lenhard and Parker (1990)] and residual
[Parker and Lenhard (1987)] LNAPL profiles successfully described the data.
The total saturation profile was governed by van Genuchten's (1980) moisture
retention characteristic, with the matric pressure head scaled by LNAPL/air
surface tension in the presence of free LNAPL and water/air surface tension in
the absence of free LNAPL, as suggested by Lenhard and Parker (1990). The
residual LNAPL was described by a simple application of Parker and Lenhard's
-------
(1987) empirical equation reflecting hysteretical trapping by a fluctuating
water table. A single, physically plausible uniformity parameter and scaling
coefficient was used at five separate locations, and the total moisture and
LNAPL saturation were jointly described by the theory and its calibration.
The calibrated error standard deviations for the free and residual LNAPL
profiles ranged from 14 to 23% in magnitude.
A field trapping protocol for hydrocarbon vapors was also developed to
document the distribution of gaseous LNAPL constituents in the contaminated
capillary fringe over a 0.03 m spacing. Sample ports were fabricated in the
intact core sleeves as part of the trapping method, which yielded total
hydrocarbon concentration data that tested a model of vapor diffusion with an
error mean of 8% and an error standard deviation of 76%. The considerable
scatter of the trapped data reflected difficulties in vapor sampling in nearly
saturated soil under field conditions. The low mean error did not indicate a
systematic variation of of the data from the theory however, and the
observations were also consistent with independently obtained tubing cluster
observations reported by Ostendorf and Kampbell (1991).
We measured the diffusion of 2,2,4 trimethylpentane (2,2,4 TMP) and 2,2,5
trimethylhexane (2,2,5 TMH) vapors evaporating from the LNAPL in the
laboratory. The headspace of a strongly contaminated intact core sleeve
sample was swept with nitrogen gas to simulate the diffusive release of
hydrocarbon vapors from the aviation gasoline to a soil venting air flow above
the contaminated capillary fringe. The resulting steady state profile was
modeled using existing diffusivity and air porosity estimates in a balance of
diffusive flux and a first order source term. The source strength, which was
calibrated with the observed flux of 2,2,4 TMP leaving the sleeve, varied with
the LNAPL remaining in the core, but was independent of the headspace sweep
flow rate. This finding suggested that lower soil venting air flow rates were
in principle as effective as higher rates in venting gasoline vapors from
-6 -5 -1
contaminated soils. The source strength varied between 1.5x10 to 10 s
in the intact core sleeve during the diffusion runs. The saturated vapor
concentration ratio of 2,2,4 TMP to 2,2,5 TMH decreased from 6.6 to 3.5 over
the five month duration of the experiments in an expression of distillation
effects. The vertical profile model was compared with sample port data in
four separate experiments for both species; the calibration error standard
deviation varied from 4 to 9%.
-------
The source term schematization used successfully in the diffusion model
was balanced against uniform advection in a model of air sparging as the next
phase of the Cooperative Agreement. The resulting quasi steady equation was
calibrated with a series of advective experiments featuring a flow of nitrogen
gas through a weakly contaminated intact core sleeve from a separate borehole
at Traverse City. The 2,2,4 TMP source strength ranged from 2x10 to 4x10
s for the seven advective stripping experiments. The data fit the theory
with an error standard deviation from 6 to 16%, so that a common source term
accurately described vapor transport in strongly contaminated soil subjected
to soil venting diffusion and weakly contaminated soil stripped by advection.
The rate of hydrocarbon removal was essentially independent of air flow for
both advection and diffusion at our site, but varied with time due to
weakening of the source strength by LNAPL evaporation. Thus both remedial
measures were controlled locally by pore scale diffusion processes and were
equally efficient in removing contamination from the capillary fringe.
Sparging was expected to be more efficient that soil venting for NAPL sources
below the water table due to the anticipated establishment of bulk air streams
in previously saturated soil. We did not study this latter phenomenon,
however.
Data were also presented documenting the biodegradation of hydrocarbon
vapors in soil microcosms obtained aseptically from the site. The microcosms
featured an exterior water seal design that proved successful in the difficult
task of vapor retention, and the calibrated kinetics were consistent with
field estimates as well. In the latter regard, existing observations of total
hydrocarbon and oxygen vapor concentrations were correctly predicted by a
coupled model [Ostendorf and Kampbell (1991)] of aerobic biodegradation, with
the independently obtained, microcosm based Michaelis-Menten type kinetics
used to quantify the reactive term. The error standard deviations ranged from
7 to 20% for the eight degradation runs in this regard. The maximum reaction
rate was quite uniform from 0.6 to 3.5 m below the ground surface, averaging
—8 3
at a value of 3.3x10 kg/m -s. The half saturation constant was set equal to
a literature value of 10 kg/m for model testing and microcosm calibrations.
-------
RECOMMENDATIONS
Some of these findings can be cast as recommendations to guide sampling,
laboratory analysis, and mathematical modeling of LNAPL and gaseous
contamination in the subsurface environment:
1. Core barrel extrusion into quart size Mason jars with headspace
sniffing in a nitrogen filled glove box was a reliable protocol for
delineation of LNAPL contamination on an online basis. The quart size
interval of 0.20 m yielded accurate depth integrated LNAPL data, suitable for
lateral and longitudinal surveys of separate phase contamination.
2. Half pint Mason jars resolved vertical profiles to 0.05 m accuracy,
sufficient to characterize the LNAPL profile. Intact core sleeves can provide
even better resolution, at the expense of time and labor.
3. The existing theory of Parker and Lenhard (1987) and Lenhard and
Parker (1990), which had already been tested against laboratory data,
accurately described field observations of the three phase distribution of
LNAPL, air, and water in the capillary fringe. Conventional single phase
matrix characteristics [e.g Brooks and Corey (1966)] provided a simple and
empirical description of moisture profiles that was appropriate for models of
gaseous transport. The latter process was predicated on a known distribution
of empty pores through the capillary fringe.
4. Intact core sleeves, deployed inside a standard core barrel, can be
simply machined to permit field vapor sampling with an accurate (0.03 m)
vertical resolution, necessary to quantify the diffusive release
characteristics of the contaminated capillary fringe. The use of intact
contaminated soil from the field was particularly important in this regard,
since the release rate was governed by the pore scale distribution of LNAPL,
and artificially introduced separate phase contaminant would not necessarily
replicate the microscale distribution.
5. A simple balance of diffusion and a first order source term
accurately described the core sleeve diffusion data and yielded consistent
estimates of the evaporation rate. A comparably simple balance of advection
and the same source term correctly modeled core sleeve advection data. The
theory and data for the diffusive and advective investigations suggested that
stripping efficiency was essentially independent of the air flow rate used to
vent or sparge the contaminated soil. Thus low flows, which delivered higher
exhaust concentrations with pumping economy, were recommended.
-------
6. Sparging and soil venting acted with comparable (pore scale
controlled) removal efficiency for LNAPL contamination in the capillary
fringe, as was sampled in this Cooperative Agreement. Other spill
configurations would differentiate the methods: a vertically dispersed LNAPL
spill through an arid unsaturated zone would best be treated with conventional
soil venting, while a submerged NAPL of large lateral extent could well
require sparging. We recommended the obvious: the mass, global distribution,
and local configuration of NAPL should be accurately determined before the
selection of an in situ remedial alternative.
7. An exterior water seal was needed for proper abiotic control of soil
microcosms designed to assess the degradation of hydrocarbon vapors.
Michaelis-Menten type kinetics accurately described the reaction at Traverse
City. The soil microcosm kinetics, when inserted in an existing vapor
degradation transport model, accurately predicted independently observed field
measurements of total hydrocarbon and oxygen concentrations. Thus laboratory
microcosms, if sampled and analyzed carefully, can successfully scale up to
verify field reactions.
8. The aerated mid depth region of the unsaturated zone has potential
for effective remediation of volatile hydrocarbon contamination. This natural
degradation effectiveness should be considered as part of an integrated
remediation plan. Aseptic sampling and soil microcosm analysis were valuable
tools in the quantitative assessment of the removal kinetics.
-------
SBCTIOH 3
MATERIALS, METHODS, AMD EXPERIMENTAL PROCEDURES
GENERAL SITE AND PLUME CHARACTERISTICS
The study area was a US Environmental Protection Agency field site at the
US Coast Guard Air Station in Traverse City, Michigan, as sketched in Figure
3.1. The site was underlain by a uniform, fine, sandy, unconfined aquifer
about 25 m thick [Twenter et al. (1985)], with a permeability of 4.75xlO-11 m2
and a water table positioned 5 to 6 m below the ground surface, sloping 0.004
to the northeast. The mean grain size in the capillary fringe [Ostendorf
(1990)] and the unsaturated zone [Ostendorf and Kampbell (1991)] was 3.8xlO~4
m, and the uniform character of the grain size distribution is suggested by
Figure 3.2. The cumulative density function CDF of the dimensional grain size
d is sketched in this figure. We found the total porosities in the fringe and
unsaturated zone to be 0.367 and 0.354 respectively, with an average air
porosity of 0.274 in the latter region.
A 1969 release of over 100,000 kg of aviation fuel at the northwest
corner of the Hangar Administration Building [Ostendorf (1990)] existed at the
time of study primarily as a separate phase liquid emulsion in the capillary
fringe covering an area about 80 m wide and 250 m long to the northeast, as
documented by soil gas [Kampbell et al. (1990)] and solid core [Ostendorf et
al. (1989), Ostendorf (1990)] sampling programs. Both surveys suggested that
the LNAPL contamination migrated along the water table far downgradient from
its original release point, so that residual liquid product was confined to
the 0.3 m thick capillary fringe, subject to historical fluctuations of the
water table. Thus, an appreciable fraction of the vertical LNAPL
contamination profile was recoverable in a single core barrel from a
moderately shallow depth at locations easily identified by elevated
hydrocarbon vapor concentrations in the soil gas [Kampbell et al. (1990),
Ostendorf et al. (1991)]. The residual product had been sampled and analyzed
extensively, so that its composition, extraction protocol, and chromatography
were well established [Vandegrift and Kampbell (1988), Ostendorf et al.
-------
Interdiction
Well Line
N
\
\
>$K Direction
/ Flow T '
Building
Boundary
606 / / ' • Plurae
Hangar
Administration
Building
Spill
Origin
Figure 3.1. Plan view of the US Coast Guard Air Station LNAPL plume.
(1989)].
The geologic simplicity of the Traverse City site, well documented
history of pollution, and relative ease with which a volatile LNAPL may be
tracked in the subsurface rendered the Traverse City gasoline plume an ideal
candidate for fundamental and continuing fate and transport studies of
hazardous waste site remediation. Indeed, the Cooperative Agreement was part
of a wider US Environmental Protection Agency study of the Traverse City site,
which included mathematical models of dissolved [Rifai and Bedient (1990)],
depth integrated separate phase [Ostendorf et al. (1989), Ostendorf (1990)],
and gaseous [Ostendorf and Kampbell (1991)] contamination, along with
laboratory investigations of sorptive characteristics [Bouchard et al. (1989)]
and biodegradation potential of the dissolved [Hutchins et al. (1991a)] and
gaseous [Ostendorf and Kampbell (1990)] plumes. Field [Kampbell et al.
-------
1.0
L_
Q
o
0.5
0.0
-B-
JCL
0.01
0.1
0
d, mm
Figure 3.2. Cumulative density function (CDF) of grain size (d) at
representative sampling location, Traverse City research site.
(1990)] and laboratory [Vandegrift and Kampbell (1988)] methods for sampling
and analysis emerged from the Michigan project as well, and major in situ
experiments were conducted to demonstrate oxygen [Rifai et al. (1988)] and
nitrate [Hutchins et al. (1991b)] based respiration of the aromatic fractions
of spills on the site. A pilot scale study of soil venting was nearing
completion at the Air Station as well at the time of this report [Kampbell et
al. (1992)].
LNAPL PROFILES
Conventional core barrel extrusion and new intact core sleeve
partitioning methods were employed at Traverse City in the sampling of LNAPL
from the capillary fringe. Both methods are described here, along with the
methylene chloride extraction and gas chromatographic analysis used to
determine the liquid aviation gasoline content of the contaminated soil
samples.
10
-------
(a)
(b)
(c)
1.5m
ro
0.0891 m
ID
-------
barrel was percussion or hydraulically driven through the hollow stem auger
annulus into the underlying undisturbed soil. The augers were equipped with a
clamshell cap covering the hollow stem to prevent blockage by heaving soil.
The core barrel included a wireline piston to maintain a vacuum above the
0.916 m soil sample, along with a pressure relief ball valve in the drive head
and a cutting shoe fitted with a core retainer basket as indicated by Figure
3.3. The barrel sampler was disassembled by removing the drive head and
piston. The core was then hydraulically extruded into autoclaved, wide mouth
Mason jars inside a (dry grade) nitrogen filled glove box equipped with an
iris to reduce site air contamination along the barrel. A flat spatula was
used to pack the jars; typically 0.050 or 0.089 m of barrel sample was fed
into half pint or pint size jars to analyze soil in the immediate vicinty of
the capillary fringe. Quart size Mason jars were used for regions beyond the
fringe, where vertical resolution was not as critical to the study; 0.204 m of
barrel material was loaded into the larger jars. The Mason jars were sealed
with bands and autoclaved lids as they were filled, and Kimwipes were used to
clean the spatula between jars.
This extrusion protocol was augmented in our study by incorporating
online headspace sampling of the Mason jar samples. The glove box was
equipped with copper gas lines welded to a Mason jar lid inside the box and
attached to a portable gas meter outside, so that the headspace of the
extruded sample could be analyzed. This procedure was run online with a
Bacharach TLV total organic vapor analyzer. This combustible hydrocarbon
meter was accurate to 10 ppm and had a range of 10 to 10,000 ppm on three
scale settings. The Mason jar was unscrewed inside the glove box, attached to
the welded lid, and headspace sampled by the meter. The maximum vapor
concentration was recorded to provide a rapid and qualitative measure of the
degree of volatile hydrocarbon contamination in the soil sample. The observed
meter readings varied from a background value of 10 ppm to off scale levels
(>10,000 ppm) associated with strong LNAPL contamination in adjacent soil.
Typically, the headspace analysis for the half pint jars associated with a
given core barrel was run in 10 min, allowing timely feedback to the rig crew
on the placement of the next sampled interval.
Upon completion of the core barrel extrusion process, the Mason jars were
subsampled in the glove box with autoclaved, disposable, Becton Dickinson 10
ml syringe barrels or a curved spatula rinsed with methylene chloride between
uses. Roughly 0.025 and 0.01 kg of wet soil were placed into 20 ml volatile
12
-------
organic analysis (VOA) bottles equipped with screw caps and Teflon faced
silicone closures for moisture and aviation gasoline determination,
respectively. The gasoline VOA bottles were pretared with 5 ml deionized
water to disaggregate the soil and 3 ml methylene chloride to dissolve the
hydrocarbons. The field subsampling procedure was intended to stabilize the
samples, thus minimizing evaporative losses and cross contamination during
transport and storage. In this regard, the heavy methylene chloride LNAPL
solution was overlain by an immiscible water layer in the field vials, which
were packed in ice until arrival at the University of Massachusetts
Environmental Engineering Laboratory in Amherst, MA, where they were stored at
4 deg C in an explosion proof refrigerator.
Core Sleeve Partitioning
The 1.5 m CME core barrel was modified to accept the insertion of a 0.90
m long, 0.0763 m ID, 0.00318 m thick stainless steel intact core sleeve,
prescored on 0.03 m intervals so it could be easily segmented by a Rigid 0.05-
0.1 m adjustable pipe cutter. The CME barrel was cut at mid-section and fit
with a coupling for easy removal of the core sleeve after sample collection.
The sleeve contained a wireline piston and a modified core basket similar to
the barrel components, and also included a 0.6 m long steel bar used to
maintain its relative position in the barrel (Figure 3.3). The sleeve piston
was 0.15 m long and featured four double Neoprene seals separated by brass
bushings, and compressed against the inner wall of the sleeve with eight Allen
screws in the bottom of the piston. Once the compression was properly set, a
Teflon wiper disk and stainless steel plate were screwed onto the end of the
piston to protect the soil sample from organics contained in the seals.
After the piston sleeve sampler was assembled, it was steam cleaned,
acetone washed, and air dried before deployment in the hollow stem auger
annulus. The sampler was lowered inside the auger column with center rods
while maintaining minimum tension on the wireline to preserve the position of
the piston. When the sleeve sampler contacted the clamshell, the augers were
lifted with the rods held fixed, thus opening the clamshell and bringing the
sleeve sampler to bear upon the upper surface of the undisturbed soil
interval. The wireline was then pulled taut as the sleeve sampler was
percussion or hydraulically driven into the soil, so that the piston was held
stationary, creating a vacuum above the soil sample and preventing its escape
13
-------
as the device was lifted from the borehole. The core sleeves were stored
upright in ice packed vertical coolers until delivery to the explosion proof
refrigerator.
We partitioned the core sleeves in a vertical jig at the laboratory using
the Rigid pipe cutter along their scores, which were rinsed with methylene
chloride before separation. A methylene chloride rinsed, wide, flat spatula
was slid through the completed cut, forming a sample base in the uppermost
sleeve segment. The top end cap was removed and 0.025 and 0.010 kg wet soil
samples were withdrawn with a disposable syringe barrel and injected into 20
ml VGA bottles for moisture and LNAPL content determination as was done for
the extruded Bubsamples in the field. The partitioned sleeve segment was
removed with the spatula after subsamples were taken and the end cap was
replaced while the next lower segment was cut. The sleeve segment VOA bottles
were stored at 4 deg C alongside their barrel extruded counterparts, and both
sets of samples were extracted and analyzed identically from this point on.
Gravimetric and gas chromatographic analyses resulted in moisture and LNAPL
mass contents M and T, respectively, as defined by
mass moisture
M = wet soil mass (3'la)
_ mass LNAPL
T ~ wet soil mass (3'lb)
Extraction and Gas Chromatoaraohy
The sealed 20 ml VOA bottles containing the soil samples and pretared
solutions were shaken in a Tyler sieve shaker for 15 min in the laboratory to
further break down the soil structure and dissolve the aviation gasoline into
the methylene chloride. He removed about 2 ml of the methylene chloride phase
from the VOA bottle by syringe and injected it into a 0.15 m long Pasteur
pipette filled with reagent grade sodium sulfate to strip dissolved water and
particulates out of the organic solution, which passed into 2 ml vials. The
vials were sealed with Teflon faced silicone closures and screw caps and
refrigerated at 4 deg C until their analysis in the gas chromatograph.
We withdrew a 1 fil volume of the water free, gasoline rich, methylene
chloride solution from the sample vial and injected it into a Varian 3500 gas
chromatograph through a split/splitless injector using a 10 jil Hamilton
Gastight syringe equipped with a 26 gage bevel tip needle and a plunger guide.
A hot needle injection was adopted, featuring about 3 pi of preceding air and
14
-------
TABLE 3.1. LABORATORY STANDARD FOR LNAPL (FIGURE 3.4a)
Compound
Methylene Chloride
2,3 dimethylbutane
2,4 dimethylpentane
2,3 dimethylpentane
2,2,4 trimethylpentane
2,4 dimethylhexane
2,3,4 trimethylpentane
2,3,3 trimethylpentane
2,3 dimethylhexane
Toluene
2,2,5 trimethvlhexane
Molar Mass
ka
0.086
0.100
0.100
0.114
0.114
0.114
0.114
0.114
0.092
0.128
HSP*
ka/m
% Mass
% Area Retention
Time. B
(Solvent)
0.666
0.303
0.207
0.167
0.0975
0.0870
0.0879
0.0647
0.0732
0.0576
4.24
5.21
12.23
22.98
3.39
15.26
13.44
7.39
5.57
10.29
4.04
5.00
12.56
23.51
3.42
15.20
14.00
5.88
5.85
10.54
59
62
73
84
90
112
122
125
131
136
150
*Based on site average temperature of 12 deg C and estimate of Reid et al.
(1987), Equation 3.2 and Table 3.2.
a 5 s exposure in the injector before syringe activation. The gas
chromatograph was equipped with a Hewlett Packard HP-5 25 m capillary column
of 0.32 mm ID fused silica with a film thickness of 0.17 /im. The injector
temperature was 325 deg C and a split ratio of about 1:80 was employed with
zero grade nitrogen serving as the carrier gas at a rate of 2 ml/min. The
oven initial temperature of 33 deg C was held for 3 min and then increased at
a rate of 10 deg C/min for about 10 min, sufficient for the arrival of the
slowest aviation gasoline constituent. A flame ionization detector (FID) at
300 deg C sensed the separated constituents using an attenuation and range of
32 and 12, respectively. The results were tabulated on a Spectra Physics
Chromjet integrator, then stored on an automatic data logging system.
An LNAPL laboratory standard consisting of 10 primary constituents
previously identified by the US Environmental Protection Agency in an
unpublished study was prepared in accordance with the composition cited in
Table 3.1. The 10 compounds accounted for more than 85% of the weathered
aviation gasoline mass at the site, with retention times of less than 180 s
and pure saturated vapor concentrations H ranging from 0.06 to 0.66 kg/m in
SP
magnitude (at the site average ground temperature r of 285 deg K). The pure
saturated vapor concentrations were computed in accordance with the estimate
15
-------
TABLE 3.2. EMPIRICAL CONSTANTS FOR CALCULATION
PURE SATURATED VAPOR PRESSURE*
OF
Compound
2,3 dimethylbutane
2,4 dimethylpentane
2,3 dimethylpentane
2,2,4 trimethylpentane
2,4 dimethylhexane
2,3,4 trimethylpentane
2,3,3 trimethylpentane
2,3 dimethylhexane
Toluene
2,2,5 trimethvlhexane
PR
PaxlO5
31.
27.
29.
26.
25.
27.
28.
26.
41.
23.
6
7
4
0
9
6
5
6
4
5
TR
dea
500.
519.
537.
544.
553.
566.
573.
563.
591.
568.
K
0
8
4
0
5
4
6
5
8
0
Cl
-7
-7
-7
-7
-7
-7
-7
-7
-7
-7
.279
.464
.461
.389
.652
.620
.417
.752
.286
.806
C2
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
563
432
478
253
414
603
428
586
381
680
C3
-3
-3
-3
-3
-3
-3
-3
-3
-2
-4
.054
.424
.371
.166
.628
.578
.192
.808
.834
.509
C4
-1.578
-2.202
-1.890
-2.220
-3.065
-2.044
-1.814
-2.585
-2.792
-0.788
*Reid et al. (1987).
of Reid et al. (1987)
X = 1-—
1-X
(3.2a)
(3.2b)
Table 3.2 lists the empirical constants C , reference pressure p , and
X R
reference temperature r for the 10 constituents. The ideal gas law converted
R
the pure saturated vapor pressure p to the corresponding concentration in
SP
accordance with
H
SP
mpe
8.31 kg-m2/mole-°K-s2
(3.3a)
(3.3b)
with universal gas constant R. and constituent molar mass m.
A range (0.05% to 3%) of standard solutions in methylene chloride was run
to cover the variation of sample strengths encountered in the analyses.
Figure 3.4a and Table 3.1 summarize a 1.2% standard solution run: the
correspondence of mass fraction and chromatographic area fraction for all ten
16
-------
100000
c
o
a.
cc
0
50000
(a)
1 2
Retention Time, min
o
o
o
o
o
o
o
QJ
0.00 0.01 0.02 0.03
Concentration
0 0-
100000
aj
in
c
o
Q.
£ 50000
at
Q
0
-
(o)
Jbl
-
I
JM_
JiUUUU
-------
TABLE 3.3. STANDARD CONSTITUENTS OF TYPICAL LNAPL SAMPLE (FIGURE 3.4O
Compound
2,3 d imet hy Ibut ane
2,4 d imet hylpent ane
2,3 d imet hylpent ane
2,2,4 trimethylpentane
2 , 4 dimethy Ihexane
2,3,4 trimethylpentane
2,3,3 trimethylpentane
2,3 dimethylhexane
Toluene
2,2,5 trimethylhexane
% Total Area
1.30
2.40
8.81
24.52
6.99
15.01
10.10
4.03
5.35
7.61
% of Standard
Compounds
1.51
2.78
10.23
28.47
8.12
17.43
11.73
4.68
6.21
8.84
Retention
Time , s
62
73
85
91
113
124
127
133
138
152
components was striking, and suggested a similar instrument response for every
constituent of the standard. A calibration curve was prepared for each batch
of analyses, with each batch run by a single operator over a two day period.
A typical calibration curve is sketched in Figure 3.2b; its slope was used to
compute the liquid aviation gasoline mass content defined by Equation 3.1b,
based on the total chromatographic area A, methylene chloride mass M , and
H
observed wet soil mass M in the 20 ml VOA bottle
M (3.4a)
Ms
M = 4.02xlO~3 kg (3.4b)
The calibration coefficient C related chromatographic area units to the
concentration of standard solutions in methylene chloride. The observed C
values for the batch plotted in Figure 3.4b all were within 15% of their
predicted value, indicative of the accurate linear regression through the
calibration data.
Figures 3.4c, 3.4d, and Table 3.3 display a typical analysis of an
aviation gasoline sample (SOBS Profile 1-35): the retention times were within
1.5% of the standard values, indicating excellent instrument precision. The
10 standard compounds comprised about 86% of the total chromatographic sample
area, in good agreement with the unpublished US Environmental Protection
18
-------
Agency gas chromatograph/mas8 spectrometer (GC/HS) analysis. The relative
distribution of peak areas for the standard compounds, cited in the third
column of Table 3.3, compared favorably with the laboratory standard cited in
Table 3.1. The drop in relative amounts of the more volatile components (2,3
dimethylbutane, 2,4 dimethylpentane) in the field sample may have indicated a
minor evaporative loss during handling and storage. The presence of heavier
constituents among the minor chromatographic peaks is shown on Figure 3.4d,
with lower vapor pressures and later retention times for these relatively
nonvolatile compounds. These lesser components would attain a greater
importance in the latter phases of soil venting remediation due to their lower
vapor pressures and greater resistance to the stripping process. Nonetheless,
the LNAPL standard was judged to be representative of the sampled product,
particularly with the focus on total liquid hydrocarbon content for this phase
of the research. The calibration coefficient for the Figure 3.4c run was
-9 3
2.56x10 kg/m -area unit, leading to an estimated extract solution
concentration of 0.00983 kg aviation gasoline/kg methylene chloride from
sample SOBS Profile 1-35.
Replicate injections run in the course of the laboratory work
consistently returned instrument precisions well within the data quality
objective of 5% set for the research. Uncontaminated soil samples from near
surface depths in the Mason jar boreholes were spiked with 100 /il of weathered
aviation gasoline during the field sampling trip. The spiked samples were
stored in sealed VOA bottles pretared with deionized water and methylene
chloride and subjected to the same handling, storage, extraction, and
analytical procedures as the barrel extruded aviation gasoline samples. An
average of 90% of the spiked aviation gasoline was recovered from the spiked
samples, a reasonable endorsement of the sampling protocol. Methylene
chloride blanks run at least every ten analyses indicated no carryover between
subsequent samples, and numerous sample replicates were prepared to test for
variation of aviation gasoline content within a given sleeve segment or Mason
jar.
LNAPL Profile Stations
Intact core sleeves and barrel extruded samples were obtained from
19
-------
\
Building s
606//
^^
Well Line
^"" X
/ \
/ \
/ A \
! Flow f }
Direction / /
/ ' /
/ /
1^. «50BT
Smith /VX50™/
HallV^/y /
>/ /Vi
/ SOBS / ]
/SOCL*50^ / J
X
Plume
' Hangar
Administration
| 50m | / /
-------
samples were obtained at all five stations for use in calibration of the LNAPL
profile model.
FIELD TRAPPING OF HYDROCARBON VAPORS
Vapor Sampling Technology
A suite of increasingly sophisticated methodologies was available for the
determination of gaseous hydrocarbon concentrations in soil gas samples from
the unsaturated zone [Devitt et al. (1987)]. Passive [Kerfoot and Mayer
(1986)] and active [Glaccum et al. (1983)] total organic vapor samplers
provided a measure of the gaseous hydrocarbon content in an air stream. The
active meters were particularly effective in the delineation of the horizontal
[Kampbell et al. (1990)] and vertical [Ostendorf and Kampbell (1991)] extent
of LNAPL contamination when used with stainless steel tubing clusters or
driven probes. They also detected the presence of separate phase liquids in
contaminated soil with extreme sensitivity when used to sample extruded cores
in a nitrogen filled glove box, as was demonstrated in the LNAPL sampling
protocol described earlier [Ostendorf et al. (1991)]. Robbins et al. (1990)
suggested that caution be exercised when the instrument response, usually
flame or photoionization, was correlated to vapor concentration, since
relative humidity and nonlinearity can affect the detection characteristics of
the total organic vapor analyzers. These instruments, while offering rapid
deployment, suffered the disadvantage of a relatively large sample volume that
in turn required a sparse vertical spacing between sample points. Individual
compounds were not distinguished by these mapping devices either. Thus
closely spaced profiles of individual hydrocarbon constituents in the
immediate vicinity of the (moist) separate phase soil material were not
feasible with this technology.
Gas chromatography has emerged as a useful tool in the identification of
individual constituents of contaminated soil gas, however. Swallow and
Gschwend (1983) trapped hydrocarbon vapors from a stainless steel sampling
tube onto Tenax resin, for subsequent thermal desorption and gas
chromatographic analysis. Wallingford et al. (1988) used an activated carbon
trap in a similar fashion, using a carbon disulfide extraction. Hinchee and
Reisinger (1985) pulled headspace vapors from a capped monitoring well and
trapped hydrocarbons onto carbon adsorber tubes. The trapped vapors were
21
-------
Detail A
Aluminum end cap (Bottom)
Threaded hole 0.64 cm diameter
(sealed for diffusion experiments)
Glass beads 0.1 cm diameter
Neoprene seal
DetailB
Pilot hole 0.33 cm diameter
through solid plug
Coarse threads to fit (7 ml)
plastic screw cap
Pierce Teflon-faced Mininert
valve to fit screw cap and bear
against plug end
Fine
Valve
Coarse
Valve
Regulator
Prepurified
Nitrogen
Influent
Port
Effluent
Port
It (
)v s
A
Carbon
Trap
3cm
- - Detail "B"
Detail "A"
Pinch Clamp (typ.)
Tygon Special
Tubing
Copper
Tubing
Figure 3.6. Intact core sleeve and transport test stand for diffusion
experiments.
22
-------
extracted with methanol, for analysis by gas chromatograph. Bednas and
Russell (1967) performed a headspace analysis on soil cores in the laboratory
without intermediate trapping, using a helium carrier gas to flush the
headspace of the cores directly into a gas chromatograph. Holbrook (1988)
analyzed hydrocarbon samples without trapping in the field by a direct
connection between the sampling probe and a portable gas chromatograph. We
applied elements of this vapor trapping and gas chromatographic protocol to
the intact core sleeves in the field as part of the Cooperative Agreement.
Intact Core Sleeve Acquisition
Intact core sleeves from Stations 50CL and 50CM were sampled in'the field
for their gaseous hydrocarbon content during the September 1991 field trip.
Ambient air temperature on September 23 and 24 was about 10 deg C. Three
closely spaced boreholes were drilled at each of the stations, with the first
devoted to Mason jar sampling for contaminated interval identification and
subsequent LNAPL and moisture content determination. The RSKERL rig crew
withdrew intact core sleeves at overlapping intervals through the contaminated
capillary fringe out of the second and third boreholes. The core sleeves used
for vapor sampling were drilled with alternating, threaded 8.7 mm diameter
holes on approximate 0.03 m centers, as sketched in Figure 3.6. Stainless
steel threaded flush plugs (not shown in the figure) were Teflon taped and
screwed into the holes to preserve a flush circumference and ensure soil
retention during field deployment. Upon removal from the ground, the ends of
the core sleeve were sealed with paraffin, capped with stainless steel plates
and taped. Bednas and Russell (1967) observed excellent retention of natural
gas in similarly sealed steel sample tubes.
After the paraffin seal in the core had set for an hour, the core was
carried to a nearby building for sampling and trapping. The temperature in
the building was about 20 deg C. Stainless steel solid plugs with coarse
threads for Teflon faced Mininert valves were inserted in place of their flush
equivalents in the field prior to trapping. The solid plugs, sketched in
Figure 3.6, featured a 3.3 mm diameter, 120 /il drillspace used for vapor
sampling. Moisture retention due to capillary forces prevented liquid and
grains from entering the drillspace, thus reducing the potential for clogging
of the syringe. A Mininert valve with a stainless steel needle guide was then
23
-------
screwed onto the plug to provide a septum over the sampling drillspace.
Teflon tape was used to seal both connections.
Field Trapping Protocol
A 50 pi Hamilton gas-tight syringe with 26 gauge bevelled tip needle was
used to evacuate 150 pi of air from the drillspace and then collect a 20 /il
air sample from the approximate center of the cavity over a 10 second period.
Twenty /jl of ambient air were then added to the syringe to reduce sample
diffusion leakage from the needle area. Vapor samples were trapped on 0.2 m
long, 0.006 m ID Envirochem glass sorbent tubes packed with glass beads,
Tenax, Ambersorb and charcoal using ultra high purity nitrogen as a carrier
gas. We preconditioned the sorbent tubes with an Envirochem Model 85 Tube
Conditioner for 10 min at 320 deg C using zero grade nitrogen.
Carrier gas was regulated by a two stage regulator, a NUPRO fine metering
valve, and a Whitey shutoff valve. The injection port was a Supelco septum
nut with a Teflon faced silicone septum installed on a Swagelok union tee.
The septum was replaced after every seventh injection. The components were
connected by 6.36 mm copper tubing and Swagelok fittings. Each sample was
observed for visible liquid before injection, and those few deep samples
containing free liquid were not trapped. The vapor sample was injected into a
39 ml/min carrier gas stream which was run for 4 minutes. The flow rate was
checked periodically with a Humonics Optiflow Model 520 digital flow meter.
The syringe was cleaned prior to each use by removing the plunger and passing
20 psi ultra high purity nitrogen through the syringe barrel and needle for 30
seconds.
Sorbent tubes were stored and transported in Envirochem glass containers
with screw caps, which were packed in groups in bubble wrap and plastic bags
and placed in plastic containers. Granular activated charcoal was added to
the plastic containers to protect the tubes from air contaminants such as auto
exhaust. A total of twelve 50CL and eleven 50CM vapor samples were trapped.
We checked for breakthrough of hydrocarbon vapors by trapping a sample onto a
pair of sorbent tubes attached in series by a Swagelok union with Vespel
ferrules. In this regard, complete retention on the upstream tube would
result in a clean downstream tube. One sorbent tube was spiked with vapor
standard each day, and a final sorbent tube served as a trip blank.
Liquid hydrocarbon standards from ChemService were combined to form a
24
-------
TABLE 3.4. LABORATORY STANDARD FOR AVIATION GASOLINE VAPOR*
Compound Elution
Order
2,3 dimethylbutane
2,4 dimethylpentane
2,3 dimethylpentane
2,2,4 trimethylpentane
2,4 dimethylhexane
2,3,4 trimethylpentane
2,3,3 trimethylpentane
Toluene
2,2.5 t r imet hvl hex ane
1
2
3
4
5
6
6
7
8
Mass Fraction used Observed Hydrocarbon
in Standard. % Vaoor Fraction. %**
22.4
10.9
17.3
23.0
2.0
7.9
10.4
3.0
3.1
2.4
4.4
11.7
37.5
6.5
23.2
***
0.8
5.9
*Nitrogen and oxygen content was excluded from the fractions.
**As measured (FID area fractions) in port midway along the core sleeve during
the 7/22/91 experiment.
***2,3,4 TMP and 2,3,3 TMH coeluted.
vapor standard of the composition cited in column 3 of Table 3.4. The vapor
standard varied from the LNAPL standard of Table 3.1 due to the unavailability
of 2,3 dimethylhexane during the latter phases of the study period. Raoult's
law was applied to the original fractionation of the liquid gasoline
[Ostendorf et al. (1989)] in deriving the mass fractions used in the vapor
standard (Table 3.4, column 3). Since the nine standard compounds comprised
over 80% of the original LNAPL, the mixture was thought to represent the
original vapor composition of the soil gas (excluding oxygen and nitrogen).
The nine standard compounds represented over 90% of the hydrocarbons found in
the vapor samples for the study, as suggested by the typical fractionation in
column 4 of Table 3.4. Daily vapor standards were made by injecting 20 fii
(liquid) volumes of the standard into a 155 ml serum bottle having a butyl
rubber stopper and aluminum crimp cap. The serum bottle was subsequently set
aside for 45 min to homogenize. Field standards were trapped using the same
procedure as was used for samples.
Sorbent Tube Desorption and GC Analysis
The sorbent tubes were desorbed and analyzed at the University of
Massachusetts Environmental Engineering Laboratory on October 4 to 18, 1991.
25
-------
An Envirochem Unacon Model 810 thermal tube desorber was used. It desorbed
each tube, focused the sample, and transfered it to the GC column.
Temperatures were regulated by an Envirochem Model 815 temperature control
module, and zero grade nitrogen was used as the carrier gas. The Unacon
desorber had two internal glass traps packed with glass beads, Tenax,
Ambersorb and charcoal. The field trapped sorbent tube was desorbed in the
tube chamber at 215 deg C for 7 minutes, during which time the sample
collected on internal Trap 1. A one minute carrier gas flow followed to
remove water vapor from trap and transfer lines. The nitrogen flow through
Trap 1 was then reversed and the trap was rapidly heated to 230 deg C for 2
minutes to desorb the sample for transfer to Trap 2. The flow through Trap 2
was then reversed and the trap rapidly desorbed at 230 deg C and carried to
the GC column at a carrier gas rate of 2 ml/min. The Unacon desorber was
operated in the splitless mode.
A Varian 3500 GC equipped with a 30 m long, 0.53 mm ID Supelco VOCOL
capillary column with a 3.0 pm thick stationary phase was operated in the
splitless mode. The stationary phase was bonded diphenyl dimethyl
polysiloxane, a material of low to medium polarity. The column oven
temperature was programmed at 50 deg C for 1 minute, then ramped at 10 deg
C/min to 250 deg C and held for 5 minutes. An FID at 300 deg C detected
compounds using an attenuation and range of 32 and 12, respectively. A
Spectra Physics ChromJet integrator and an automated data logger were used to
process the data. Sample runs were terminated after the nine compounds in
Table 3.4 had generated peaks on the chromatograms. Daily vapor standards
were transferred onto conditioned sorbent tubes from which field samples had
been previously desorbed and analyzed; these standards were prepared similarly
to the field standards described in the previous section. The laboratory
trapped standards generated calibration curves used to convert sample peak
areas to concentrations, and blanks were analyzed daily to evaluate background
noise and desorption efficiency. A total of 27 field tubes, 24 vapor
standards, and 20 blanks were analyzed.
CORE SLEEVE DIFFUSION
Vapor Transport in the Unsaturated Zone
26
-------
Evaporation was shown to be an important transport mechanism for LNAPLs
of high volatility due to preferential gaseous phase partitioning [Baehr
(1987)] and the proximity of the separate phase contaminant to the unsaturated
zone as it rested above the water table [Hinchee and Reisinger (1985)].
Indeed, Ostendorf and Kampbell (1991) estimated that over 35% of the spilled
product at the field site considered in this study volatilized over a 22 year
period along a 260 m plume trajectory. Falta et al. (1989), Sleep and Sykes
(1989), Mendoza and McAlary (1990), and Jury et al. (1990) all derived
mathematical models of the vapor transport process. Bruelle and Hoag (1986),
recognizing the importance of gaseous diffusion in this context, measured the
diffusion of gasoline components artificially introduced into laboratory sand
columns through observations of influent and effluent vapor concentration.
Farmer et al. (1980) observed the diffusion of hexachlorobenzene through a
test section under laboratory conditions too, while various field
investigations of gaseous contamination have been described earlier in this
report. The laboratory data bases featured concentration profiles from
artificially contaminated soil columns, while the field data bases were
spatially averaged measurements of gaseous contamination from actual LNAPL
spill sources. We thus noted a lack of vertically resolved gaseous profiles
for soils contaminated under realistic field conditions. The lack was
particularly acute due to the importance of pore scale distribution to the
mass transport of hydrocarbons to bulk air or water flows passing through the
contaminated capillary fringe. This microscale configuration of water, air,
and LNAPL adjacent to the water table governed the source strength of soil gas
and groundwater pollution and, by the same token, the rate of LNAPL stripping
from the residually contaminated soil.
Station 50CE (as shown on Figure 3.5) was selected for the investigation
of core sleeve diffusion. An intact core sleeve was obtained on November 4,
1990 after identification of the contaminated interval by barrel extrusion and
online headspace sampling of Mason jar samples. Two target compounds were
selected for study of the mass transport mechanism: 2,2,4 trimethylpentane
(2,2,4 IMP) and 2,2,5 trimethylhexane (2,2,5 TMH). The former hydrocarbon
was the primary constituent of the gasoline in liguid and vapor phases, while
the latter was the heaviest alkane of appreciable content, and so elucidated
LNAPL distillation effects.
27
-------
Gaa Chromatographv
The core sleeve was sampled in the laboratory using the solid plug and
Mininert sample port protocol described for field trapping. Each 20 /il vapor
sample was withdrawn from the drillspace of the intact core sleeve over a 10 s
period after passing prepurified nitrogen gas through the syringe barrel and
needle for 30 s at 20 psig. A 20 jil air volume was immediately drawn into the
syringe after the vapor sample to reduce volatile losses from the needle. The
vapor sample was then injected directly into a Varian 3500 gas chromatograph
through a split/splitless injector. The GC was equipped with a Hewlett-
Packard HP-5 25 m capillary column of 0.32 mm ID fused silica, 0.17 /«n film
thickness, and crosslinked 5% phenyl methyl silicone stationary phase. The
injector temperature was 250 deg C and a split ratio of 10:1 was employed with
zero grade nitrogen serving as the carrier gas at a rate of 2 ml/min. The
initial oven temperature of 35 deg C was held for 3 min and then increased at
a rate of 10 deg C/min for about 10 min, sufficient for arrival of the slowest
aviation gasoline constituent. An FID at 300 deg C sensed the separated
compounds using an attenuation and range of 32 and 12, respectively. The
results were tabulated on a Spectra Physics ChromJet integrator and stored on
an automated data logger.
The vapor standard is given by Table 3.4, and standards of varying
strengths were prepared by liquid injection into 155 ml serum bottles. The
standards were analyzed concurrently with the daily sleeve samples to
establish separate calibration curves relating the FID response to 2,2,4 TMP
and 2,2,5 TMH concentrations. Thus, separate calibration coefficients C were
used for each compound in the core sleeve diffusion study.
We determined the method detection limit (MDL) for chromatographic
analysis of 2,2,4 TMP and 2,2,5 TMH following standard methods [APHA (1989),
Taylor (1987)]. Seven 20 /il injections of a low (1.47 mg/m total
hydrocarbons) vapor standard concentration were analyzed in the gas
chromatograph. The sample standard deviation OKDL of the seven runs was used
to estimate the MDL at the 99% confidence level in accordance with
MDL = 3-14aMDL 10<7MDL) (3<5)
with sample mean and standard deviation defined by
mean = SA (N samples) (3.6a)
28
-------
o
0)
CL
CN
CN
7000
a. 6800
6600
0
(J
O
U
Mean+MDL
Mean
0 0
Mean-MDL
345
GC Run
0
8
/ uu
D
O
(1)
°~ 600
X
in
CN
son
c
i 1 1 1 1 1 1
Mean+MDL
0 0
Mean ^
0 ° ° 0
Mean-MDL
i i i i i i i
) 1 2 3 4 5 6 7 g
GC Run
Figure 3.7. Method detection limit runs for 2,2,4 TMP and 2,2,5 TMH.
MDL
S(A2)-mean2]1/2
(3.6b)
The MDL of the Varian 3500 gas chromatograph for 2,2,4 TMP and 2,2,5 TMH,
subject to the vapor sampling method described above, was found to be 168 and
76 FID area units, corresponding to respective concentrations of 8.3 and 5.6
The intact core sleeve vapor concentrations greatly exceeded the MDL
value. Figure 3.7 summarizes the MDL runs.
29
-------
Transport Test Apparatus
The transport test apparatus consisted of the intact core sleeve, end
caps, and a gas supply system, as sketched in Figure 3.6. The ends of the
core sleeve were trimmed with a Rigid 0.05 to 0.1 m adjustable pipe cutter in
a vertical jig, then packed with glass beads of 1 mm diameter and set into
aluminum end caps with Neoprene seals. The end caps were tightened with
threaded steel rods and wing nuts, and the top cap was fit with a Teflon tee
connected to hydrocarbon resistant Tygon Special tubing with pinch clamps.
The apparatus was leak tested with a brief exposure to 6 psi nitrogen gas, and
the experimental runs consisted of headspace swept flows of prepurified grade
nitrogen gas regulated at low values of flow and gage pressure. The use of
the nitrogen was predicated by a desire to replicate the field conditions
observed by Ostendorf and Kampbell (1991) in the capillary fringe, where
oxygen occured at low concentrations, having been consumed by aerobic
biodegradation in the overlying unsaturated zone. The sweep flow was intended
to simulate soil venting in the unsaturated zone above the contaminated
capillary fringe. We neglected degradation in the core diffusion runs, due to
the presence of separate phase LNAPL and lack of oxygen in the core sleeve
soil.
The influent tubing was run through a distilled water saturator to
prevent moisture desorption in the soil column, and the effluent tubing was
connected to a carbon trap, with flow checked using a Humonics Optiflow 520
digital flow meter. Influent and effluent tubing were equipped with sample
ports, consisting of brass tees and Teflon faced Mininert valves. The inlet
port was checked to ensure a zero influent concentration. Temperature was
monitored in the experimental apparatus, and steady state conditions were
verified by successive profile measurements after a typical equilibration
period of two to three weeks.
Table 3.5 lists the test conditions for the 4 experiments conducted to
elucidate the diffusive release of hydrocarbon vapor from the contaminated
capillary fringe. Sweep flows Q ranging from 1 to 5 ml/min were selected for
study, with one flow rate repeated at two different times to assess how the
relationship between release rate and headspace sweep flow rate changed with
the remaining LNAPL content in the sleeve. The range of flows was low enough
to ensure detectable concentrations of hydrocarbons in the effluent port, yet
high enough to induce diffusion out of the soil. The time t cited in the
30
-------
TABLE 3.5. TEST CONDITIONS SLEEVE DIFFUSION RUNS
Date
7/22/91
8/26/91
11/25/91
1/3/92
t
8
7
1.26x10
1.56xl07
7
2.35x10
2.69xl07
r
deer K
296.3
293.5
293.0
296.8
Q
ml/min
1.08
2.94
4.89
3.32
table refers to the initiation of sweeping on February 26, 1991. The
experiment temperature was r.
CORE SLEEVE ADVECTION
We chose Station 50CL (Figure 3.5} for the study of intact core sleeve
advection. An intact core sleeve was obtained on September 23, 1991 in
concert with core barrel extrusion and online headspace sampling for
identification of the contaminated capillary fringe. The hydrocarbon vapor
sampling and gas chromatographic protocol used for advection was identical to
the diffusion study mentioned above. The same two compounds (2,2,4 TMP and
2,2,5 TMH) were analyzed as well.
Transport Test Apparatus
The transport test apparatus used for diffusion was slightly modified to
accomodate the advective experimental program. The Tygon Special tubing,
pinch clamps, and Teflon tees were replaced with 6 mm copper tubing, Whitey
shutoff valves, and brass Swagelok unions, respectively, with connections
maintained using Vespel ferrules. As suggested by Figure 3.8, the influent
prepurified nitrogen gas used as a carrier gas was routed to the bottom end
cap, which was fitted with a Swagelok tee. Thus advection proceeded in an
upflow mode, to simulate sparging of gas below the water table. Temperature
control was provided by running copper coils from a constant temperature water
bath around the intact core sleeve. Insulation was wrapped around the test
apparatus as well.
31
-------
Brass
Tee (typ.)
Effluent
Port
Constant
Temperature
Water Bath
Fine
Valve
— X
1
Cooling A"
Coil A
.-/
Intact
Core /"
Sleeve ^
D
Shutoff Valve (typ.)
- - ) Sample Port
Influent
Port
Coarse
Valve
Regulator
Prepurified
Nitrogen
| Copper
Tubing
Water
Saturator
Figure 3.8. Transport test stand for core sleeve advection experiments.
Experiments and Displaced Pore Volumes
h series of seven experiments was run to document the stripping of
hydrocarbons from the intact core sleeve by advection, as summarized in Table
3.6. Nominal flow rates ranging from 3 to 10 ml/min were considered in
ascending and then descending order to assess the influence of specific
discharge and remaining LNAPL content on the source strength. The runs were
classified by both the date and the (cumulative) number of pore volumes £
displaced by the nitrogen gas. The latter variable was defined by
32
-------
TABLE 3.6. TEST CONDITIONS SLEEVE ADVECTIQN RUNS
Date
7/13/92
7/15/92
7/16/92
7/17/92
7/20/92
7/22/92
7/23/92
£ s
<
105
120
180
200
245
274
283
t
A Jwdt '
V.
9.
1.
1.
1.
1.
1.
1.
t
s
48xl05
24xl06
32xl06
40xl06
66xl06
83xl06
92xl06
r
dea
285.
285.
285.
285.
285.
285.
285.
K
7
8
7
6
8
5
9
w
m/s
1.07x10
4.07x10
3.87x10
3.86x10
2.03x10
2.03x10
9.73x10
-5
-5
-5
-5
-5
-5
-6
(3.7a)
(3.7b)
w
volume air
total volume
(3.7c)
(3.7d)
with vertical distance z above the water table, specific discharge w, and air
porosity 8. Also appearing in Equation 3.7 are the gross cross sectional area
A_ and empty pore volume V. of the contaminated interval.
Li 0
A Brooks and Corey (1966) moisture characteristic was fit to the water
content 6 for the advective transport analyses
w
W
volume water
total volume
(3.8a)
(3.8b)
(3.8c)
33
-------
with irreducible moisture content 8 , uniformity exponent a, total porosity
WR
n, and bubbling pressure head \l> . The empty pore volume was recovered by
O
integrating the Brooks and Corey (1966) expression from ^> to the top of the
o
LNAPL contaminated interval at z equals A. We accordingly invoked Equations
3.7 and 3.8 and expressed £ as
(a-l)Sw t
« U r-I-r (3.9)
rg rg
Equation 3.9 then related the number of displaced pore volumes to a series of
specific discharge w values for varying durations t over the course of the
experiments.
SOIL MICROCOSMS
Overview of Soil Microcosms
A substantial and continuing interdisciplinary literature emerged
regarding the biodegradation of light hydrocarbons in dissolved [Lee et al.
(1988)], separate phase [Song et al. (1990)], and gaseous [Kampbell et al.
(1987)] form as they existed in the saturated zone, capillary fringe, and
unsaturated zone, respectively. Ascending study length scales were used, from
soil microcosms [Aelion and Bradley (1991)] to laboratory columns [Song and
Bartha (1990)] and field installations [Ostendorf and Kampbell (1991)]. Soil
microcosms contained natural or artificial soil, water, nutrients, substrate,
microbes, and an electron acceptor to facilitate the study of microbial
degradation under controled conditions. Wilson and Noonan (1984) suggested
that microcosms may be classified according to their focus on the population
and function of the biological community or the kinetics and products of the
pollutant substrate reactions. Our interest was primarily in the latter
function, since the microcosmic work was geared to corroboration of
independently derived hydrocarbon vapor degradation kinetics at Traverse City.
In the latter regard, Ostendorf and Kampbell (1991) analyzed hydrocarbon and
oxygen soil gas cluster data from the site and inferred half saturation
constants and maximum reaction rates for the aerobic degradation of aviation
gasoline fumes in the unsaturated zone.
34
-------
Sampling Protocol and Laboratory Microcosm Design
Station 50CL was selected for the analysis of the kinetics of hydrocarbon
vapor degradation by soil microcosms. We obtained microcosm samples under
relatively aseptic conditions during the core barrel extrusion process within
the nitrogen filled glove box. As cores were hydraulically extruded from the
core barrel sampler, they were sliced at the desired depths for sampling,
using a Kimwiped flat spatula. A washed and autoclaved Becton Dickinson 10 ml
syringe barrel was inserted into the inner portion of the exposed soil face
and removed, at which point it contained the desired soil sample. This sample
was in turn extruded from the syringe into a washed and autoclaved 20 ml VOA
vial which was capped with a Teflon faced silicone septum and plastic screw
cap. A separate syringe barrel was used for each depth in the borehole, and a
total of six depths were analyzed, with duplicate microcosms taken from each
depth. The field vials were packed in ice and transported to the
Environmental Engineering laboratory at the University of Massachusetts at
Amherst. Field microcosms were stored in the laboratory at 4 deg C prior to
being transferred to laboratory microcosm vials for analysis.
Figure 3.9 shows a schematic representation of a typical laboratory
microcosm. The devices consisted of 12 ml screw cap vials containing enough
soil so that soil depth was approximately 5/8 of the total vial depth. Each
vial was equipped with a Teflon faced Mininert valve which had been fitted
with a guide enabling the use of a 26 gauge Hamilton Gastight syringe needle.
The Mininert valve was a push-button closure which, when the valve is open,
allowed for the insertion of a syringe needle for sampling purposes. A rubber
gasket above the Teflon stem provided a seal for the needle and helped to
prevent leakage and contamination of the vial contents. An exterior water
seal for the microcosms was effected by placing each inverted laboratory
microcosm vial into a 100 ml centrifuge tube, and adding enough distilled
water to cover the cap/vial juncture. A glass wool plug and glass rod held
the soil in place within the vial and provided a clear headspace for
subsequent vapor extraction. Each vial was equipped with a handle to
facilitate easy removal of the vial from the centrifuge tube. The handle also
supported the vial in an inverted position during vapor sampling for GC
analysis. Handles were fabricated from silicone tubing (3/32" OD and 1/32"
ID) into which copper wire had been inserted. Centrifuge tubes were capped
35
-------
I. D. 1/32", O. D. 3/32"
Silicone Tubing
(Cole-Farmer)
with Copper Wire
Inserted
Glass Rod
Mininert Valve
SC-15
For 7 ml
Screw Cap Vials
(Pierce)
Friction-fit Closure
(Nalgene)
100 ml High-Speed
Polypropylene
Centrifuge Tube
(Nalgene)
12 ml Sample Vials
(Wheaton)
Soil
Silane Treated
Glass Wool
(SUPELCO)
Headspace
Distilled
Water
Stainless Steel
Needle Guide
Figure 3.9. Soil microcosm for hydrocarbon vapor degradation studies.
with friction-fit closures, and the laboratory microcosms were stored in a
Fisher Isotemp incubator at 15 deg C.
We transferred the soil from the field VOA bottles to the laboratory
microcosm vials in a laminar flow hood. The latter provided a non-turbulent
air flow pattern which decreased the possibility of contamination by airborne
microbial spores. The use of aseptic techniques in laboratory microcosm
preparation further increased the probability that laboratory microcosms
contained only those microorganisms indigenous to the site. Prior to
microcosm preparation, vials and valves were washed, rinsed with distilled
water, and autoclaved for 20 minutes at 121 deg C. Also autoclaved were 0.2 g
plugs of glass wool and 1/32" diameter glass rods, 3.5 cm in length, the
number of glass wool plugs and glass rods equalling the number of microcosms
to be prepared. All microcosm preparation in the laminar flow hood was
performed next to a flame, and tools were alcohol flamed prior to each use.
36
-------
Each laboratory microcosm was prepared by transferring with a small
spatula approximately 7 g soil from a field-acquired microcosm vial to a
laboratory microcosm vial. The soil was topped with a plug of glass wool
which was held in place with a glass rod. We added 0.3 ml of sterilized
distilled water to each microcosm vial, which was then capped with a Mininert
valve.
Dosage and Gas Chromatooraphv
The samples were dosed with the vapor standard cited in Table 3.4, for
subsequent analysis by gas chromatograph. The dose injected into the
microcosm vials was prepared as follows: a liquid volume of 30 /*! of the
liquid standard was injected into a 155 ml serum bottle through a crimp-capped
butyl rubber septum. The liquid was allowed to evaporate and diffuse within
the serum bottle for 45 minutes in order to ensure complete homogeneity. On
the first day of the sampling period each microcosm vial was injected with 285
(Ml of the vapor standard, resulting in a total aviation gasoline vapor
concentration in the head and porespace of about 4.7 g/m (1000 ppm by
volume).
The dosage concentration was based on a liquid standard density of 695
kg/m , and was calculated stoichiometrically so that oxygen was not a limiting
factor. In this regard, the complete aerobic mineralization of the
hydrocarbon vapors was represented by [Ostendorf and Kampbell (1991)]
C_H,0+12.50,-*8CO_+9H,0 (3.10)
8 lo 2 22
Equation 3.10 suggested that 3.51 ^g of oxygen were required to degrade 1 /ig
of aviation gasoline vapors. Assuming 50% porosity of the soil that filled
5/8 of the 12 ml laboratory microcosm, an air space of 8.3 ml occupied the
vial. The air contained 39 nq of hydrocarbons and 2400 /jg of oxygen at the
onset of the experiments, and the oxygen was abundant.
The dosed vapors were allowed to equilibrate within the microcosm vial
over a period of at least 10 minutes, after which a 30 jil volume from the
headspace was extracted for injection into the gas chromatograph and analysis
for initial aviation gasoline content. The 30 /*1 injection volume depleted
the vapor content of the head and porespace of the microcosm by less than
0.4%. Subsequent testing involved periodic headspace vapor extractions and
GC analyses while waiting for the soil bacteria to acclimate to the
37
-------
Q
1 2
Retention Time, min
-------
those described for the core diffusion experiments, and a range of 3 vapor
standards bracketing the expected microcosm vapor concentration was prepared
at the beginning of each sampling day. These standards (6.60 g/m , 4.40
g/m , and 2.20 g/m ) were prepared through serial dilution of the vapor
standard described previously. Thirty fjtl injections of the standards were
used to convert total chromatogram peak areas to total concentrations, thereby
normalizing the sampling results over the testing period. Typical gas
chromatograms showing the relative amounts of aviation gasoline vapor
compounds within a microcosm at a given sampling time are presented in Figure
3.10, along with a standard chromatogram. The summation of peak areas was the
source of aviation gasoline vapor content values used in the kinetics
analysis. The elution order is cited in Table 3.4.
Two sets of abiotic controls were prepared for the microcosm runs to
document the successful retention of hydrocarbon vapors by the exterior water
seal design of Figure 3.9. The performance of the abiotic controls is
discussed further in the results section of this report.
SUPPORTING MEASUREMENTS
Gas Chromatoqraphv/Mass Spectroroetry
We confirmed the identity and elution order of primary hydrocarbon
constituents in the vapor samples (as cited in Table 3.4) using gas
chromatography/mass spectrometry on a Hewlett-Packard (HP) Model 5988A mass
spectrometer interfaced to an HP Model 5890 capillary gas chromatograph.
Methylene chloride solutions containing standard compounds were introduced by
split/splitless injection onto an HP-5 capillary column of 30 m length and
0.25 mm ID, with a cross-linked 5% phenyl silicone stationary phase and a 0.32
pm film thickness. All relevant gas chromatograph/mass spectrometer
instrument conditions are presented in Table 3.7. Mass spectra from the
samples were stored on an online data system and analyzed using the HP 59970
MS Chemstation software. Constituent compounds were identified using the
probability based matching reverse search of a reference spectral library
established by the analysis of pure hydrocarbon standards.
The initial column temperature was reduced to 10 deg C using liquid
carbon dioxide to reconcentrate the sample at the head of the column via the
39
-------
TABLE 3.7. SUMMARY OF GAS CHROMATOGRAPH/MASS SPECTROMETER INSTRUMENT
PARAMETERS
Gaa Chromatooraph Conditions Mass Spectrometer Conditions
Injector Mode: Splitless Mode: Electron Impact Scan
Injector Temperature: 200 deg C Electron Energy: 70 electron volts
Inlet Purge Delay: 30 s Solvent Delay: 120 B with manual
override
Carrier Gas: He at 30 cm/s Mass Range: 10 to 400 atomic mass
Column Temperature: 10 to 225 deg C units
at 10 deg C/min
Transfer Line Temperature: 280 deg C Scans per Second: 1.1
Samples per Peak: 4
Peak Threshold: 1000
solvent effect [Grob and Grob (1974)]. This improved the resolution of 2,3,4
trimethylpentane and 2,3,3 trimethylpentane, which coeluted at higher initial
temperatures. The mass spectrometer did not analyze 2,3 dimethylbutane since
it eluted very close to the methylene chloride peak, but all other
constituents were positively identified in the reference library, using the
probability based matching search. The electron impact spectrum of the most
abundant peak in the vapor sample (2,2,4 TMP) is presented in Figure 3.11,
along with the standard spectrum from the library. The sample and standard
spectra of the only aromatic component (toluene) are presented in the figure
as well. The relative abundances were normalized to the most abundant mass
fragment, and the agreement between library and sample spectra was excellent.
Moisture Content
Moisture content was determined gravimetrically. Roughly 0.020 kg moist
samples were placed in preweighed aluminum dishes, which were air dried
overnight, then dried in a Fisher Isotemp Model 500 convection oven held at
104 deg C for 60 minutes. The 60 minute drying time was verified
experimentally (Figure 3.12) as sufficient to drive off nearly all moisture.
After drying, the dishes were removed from the oven and cooled in a dessicant
chamber, then weighed on a Sartorius Basic Series toploading balance, accurate
to 10 mg. Moisture content was reported on a mass basis in accordance with
40
-------
LJ
o
Q
Z
ID
CD
LJ
1
LJ
I UU
75
50
25
n
(a)
_
41
-
1527
, Ll .
,
i ,i
57
-
_
99 ~
I
.„. , i,
20 40 60 80 100
MASS TO CHARGE RATIO
o
z
<
O
Z
^)
CD
LJ
Q:
1 UU
75
50
25
n
(b)
-
-
1527
i I.I
41
ll al
57
-
-
99 "
i.i,, . .
20 40 60 80 100
MASS TO CHARGE RATIO
LJ
O
O
Z
r>
CD
<
LJ
LJ
a:
1 UU
75
50
25
n
(c)
—
-
65
39
. .1 ii ll . i,
91
—
-
-
U —
20 40 60 80 100
MASS TO CHARGE RATIO
LJ
Q
Z
ID
CD
<
LJ
LJ
1 UU
75
50
25
n
(d)
-
-
65
39
i n ill _i,
91
-
-
_
U —
20 40 60 80 100
MASS TO CHARGE RATIO
Figure 3.11.
Normalized El spectra from gas chromatograph/mass spectrometer
runs: (a) 2,2,4 trimethylpentane standard; (b) 2,2,4 trimethyl-
pentane in vapor sample; (c) toluene standard; (d) toluene in
vapor sample.
Equation 3.la. A subsequent 180 deg C exposure indicated no appreciable
additional moisture loss, so that higher boiling point residuals contributed
negligible error to the moisture determination. This was not surprising, in
view of the relatively small contribution of gasoline to the total moisture
mass at the site. Replicate moisture content samples taken at the same time,
station, and depth exhibited a typical precision of 5%.
41
-------
s
M
w
•
!-•
Ul
O rt
P» C
sr
n
. o
a> n
3- rt
O ID
ID
rt
S 2-
^ fD
I O
i ^
• «
PI '-"
01 ^_
HI
c
o
rt
O
M>
(D
O
3
PI
0_
c
3
CD
Depth, m
ID
OJ
o o
a rt
5 S
o
o
rt
a
rt
tr
•<
a
a>
ID
PI
O
ft
O
Ml
rt
I
Rl
ft
Moisture Mass Fraction
D
UD
3
(D
3
D
0
0
O
c
c
o(
K>
0
0
en
3 0 0 C
3 -» NJ C
•^
' 1 1
O
O
O
O
O
o
o
o -
o
0
o
i I r\
-------
The vertical distribution of water was fit to a Brooks and Corey (1966)
moisture characteristic (Equation 3.8a) for empirical input to our analyses of
vapor transport. Figure 3.13 displays a typical result of the two parameter
Fibonacci search [Beveridge and Schechter (1970)] minimizing the water content
error mean <5 and standard deviation a defined by [Benjamin and Cornell
w w
(1970)]
* measured 6 -predicted 9 , , ,
*W -- measured 8 - <3'lla)
(N samples) (3. lib)
(3. lie)
Grain Size Distribution
Grain sizes were measured by mechanical sieving in accordance with
conventional [USACOE (1970)] methodology. After overnight drying, about 400
grams of soil was weighed on the OHaus Model B5000 balance and placed in a
stack of Fisher Scientific standard sieves, then shaken for about 15 minutes
in a Tyler Model RX24 sieve shaker. The mechanically graded soil fractions
were swept from the sieves and weighed. The individual masses were added and
compared to the original total mass and checked against a 1% accuracy
constraint.
The grain size data were fit to an empirical cumulative density function
*
F(D ) proposed by van Genuchten (1980)
*
F(D) = [1+(D) (/3>1) (3.12a)
* d
D = — (3.12b)
M
with dimensional grain size d, mean grain size d , and dimensionless grain
M
*
size D . The mean diameter and uniformity exponent ft were used to minimize
the mean error and error standard deviation through a nested Fibonacci search
similar to that used for the water content fitting. Figure 3.2 displays a
typical result.
43
-------
SECTION 4
MATHEMATICAL ANALYSIS
LNAPL PROFILES
Overview of LNAPL Distribution in the Capillary Fringe
The vertical distribution of free and residual aviation gasoline through
the capillary fringe of a fine, uniform, sandy soil was measured and modeled
in this investigation. The gasoline was immiscible with and less dense than
water, so that infiltrating product from a surface source spread out over the
water table, subject to retention in the overlying capillary fringe under the
action of surface tension [Schwille (1967), Finder and Abriola (1986)]. The
initial spreading of immiscible contaminants through the subsurface as a
continuous phase was considered in numerous mathematical [Abriola and Finder
(1985), Corapcioglu and Baehr (1987), Sleep and Sykes (1989)] and laboratory
[Schwille (1988), Reibel et al. (1990)] investigations. The dynamic models
were complemented by static matrix pressure-saturation characteristic curves
for continuous LNAPL-air-water systems, and substantial advances were achieved
in the prediction [Lenhard and Parker (1987,1990)] and laboratory measurement
[Lenhard and Parker (1988)] of the necessary constitutive relations. The
occurrence and distribution of a discontinuous residual nonaqueous phase
liquid had received a degree of theoretical [Parker and Lenhard (1987)] and
laboratory study [Wilson and Conrad (1984), Hoag and Marley (1986), Kia and
Abdul (1990), Conrad et al. (1992)] comparable to its continuous counterpart.
We used a simple, empirical Brooks and Corey (1966) profile to model
moisture (water) content for analysis of vapor movement, since the focus of
gaseous transport was on air porosity, and water was the dominant fluid in the
soil at Traverse City. The actual distribution of moisture was more complex
when two immiscible phases were present, however, and the observed LNAPL and
water data base were available to test an existing model of liquid
distribution in the contaminated capillary fringe.
44
-------
Free Moisture Retention
A particular example of three phase saturation was measured at the field
site. The usual wetting progression of water-LNAPL-air was in force when free
LNAPL was present, and the subsurface was comprised of initially water wet
soil, so that water filled all the small pores to an irreducible volumetric
water content 6 . The water was assumed to be continuous over the entire
WR
spatial domain with an apparent [Parker and Lenhard (1987)] effective
saturation S__ that included the residual LNAPL content 6 trapped inside the
W 1*K
water
6 +8 -6
' * *
. volume LNAPL ...
6, - . ' , - : - (4
L total volume
The total saturation S included the LNAPL content, free and residual
g +g +g -g
S = W LFfl LR ™ (4.2a,
n WR
SLF = S-SW (4.2b)
The free LNAPL saturation S followed from Equations 4. la and 4.2a.
JJF
A van Genuchten (1980) pore size distribution described all free liquid
phases, with scaling factors reflecting capillary pressures and surface
tensions across controlling interfaces in accordance with Parker and Lenhard
(1987). The total saturation was governed by the LNAPL/air interfacial
tension a_ and the LNAPL matric pressure in the region where free LNAPL was
I*A
present [Lenhard and Parker (1990)]
i-1
S = {l+[0(b-b)]7}7 (b>b>b) (4.3a)
S = 1 (b>b_ ) (4.3b)
Jbi
with depth b below the ground surface, and free LNAPL extending from b to the
H
"LNAPL table" at depth b below the ground surface, as sketched in Figure 4.1.
jj
The latter parameter would mark the surface of free product in a monitoring
well.
The uniformity exponent appearing in Equation 4.3 was 7 and the LNAPL
A5
-------
1, 11
Water + Trapped LNAPL
0
Volumetric Liquid Content
Ground
Surface
Air
Water
v
b
Figure 4.1.
Monitoring
Well
Definition sketch for three phase fluid distribution in soil.
scaling factor f)T was related to the mean pore radius r by the capillary
L
equation [Streeter and Wylie (1979)] for presumedly circular pores and a small
contact angle
(4.4a)
(4.4b)
46
-------
with aviation gasoline density p and gravitational acceleration g. The
L
water/air surface tension a was equal to the sum of the water /LNAPL tension
WA
CT._ and a. so long as dissolved hydrocarbons did not 'appreciably alter the
WL LA
interfacial chemistry of the water [Lenhard and Parker (1990)].
The apparent effective water saturation was scaled by the LNAPL/water
surface tension in the presence of free oil, with a capillary pressure
differential across the interface of the two fluids
(bw-b)] (bw>b>bM) (4.5a)
SM - 1 (b>V (4.5b)
with distance to the water table b_, delineating the thickness (b -b_ ) of free
w w if
product in a monitoring well. The scaling factor /? was given by Lenhard and
W
Parker (1990) as
with water density p. The free LNAPL persisted until the depth b , determined
H
by setting S and S equal in Equations 4.3a and 4.5a with the result
W
b
* = Vw (
PWM = WP^VV (4.7b)
« = bw-bM (4.7c)
Lenhard and Parker (1990) demonstrated that the water matrix pressure p,_. at
WM
this depth was equal to the value cited in Equation 4.7b. The distance b -b
W M
corresponded to the free LNAPL contaminated soil interval 6 .
The total saturation and water saturation were coincident above b as
H
suggested by Figure 4.1, and were governed by the water/air tension and the
water matrix pressure p . By imposing a continuous p constraint at b , the
W WM rl
apparent effective water saturation in the shallower region followed from
Equations 4.4a and 4.7b
Pw = PwM+P9(bM~b> (4.8a)
47
-------
--1
sw= <1+{V[bL"V^~(Vb)]}7)7 (Vb) (4-8b)
WA L
Equations 4.5a and 4.8b matched at b by virtue of Equations 4.6 and 4.7a,
M
ensuring a continuous water saturation profile as well.
The free water and free LNAPL profiles reflected the positions of the
LNAPL and water tables at the time of sampling, in contrast to the trapped
LNAPL, which was a function of historical fluctuations of these parameters.
LNAPL Entrapment
The residual LNAPL was thought to be the result of hysteretical trapping
of gasoline as it fell and rose through the water wet soil over a fluctuating
water table. The residual LNAPL saturation S reflected the fraction of the
LK
drainable pore space that was filled by discontinuous masses of aviation
gasoline left behind by the process
SLR
WR
volume trapped LNAPL .. g, .
LR ~ total volume
Parker and Lenhard (1987) suggested that the residual product distribution was
governed by the position of the LNAPL/water interface, as tracked by the water
saturation and LNAPL/water capillary pressure differential.
Available sites at a given depth b drained as the water table fell to its
maximum depth b in the mid-winter at Traverse City. The minimum water
WMAX
saturation S at this reversal point [Parker and Lenhard (1987)] was
WMXN
specified by Equation 4.10a, since free oil was present
SWMAX=
b = b +A (4.10c)
WMAX WMIN
Some of the free LNAPL was trapped during spring imbibition as the water table
rose along a drier moisture characteristic towards a summer minimum depth
b , defined by the fluctuation amplitude A. The minimum water table depth
WMIN
48
-------
yielded a maximum water saturation S (Equation 4.10b) at the given b
WMAX
value. The hysteretical wetting process yielded a maximum trapped LNAPL
saturation S if the water table rose past the depth b under
LRMAX
consideration, as described by Parker and Lenhard (1987)
1~SWMIN
SLRMAX -
with the hysteretical factor rj geometrically defined by the primary drainage
and imbibition curves. In the absence of detailed field data describing the
primary moisture characteristics, r\ was calibrated in this study from observed
maximum residual saturation data at the site.
Those elevations higher than the minimum water table depth attained
during trapping experienced occluded masses less than the maximum, in
accordance with Parker and Lenhard's (1987) empirical relationship
SLR ' 8LM«X(14lffllM) (4
LR LRMAX 1 SmIJt WMIN
SLR = SLRMAX (bWMINbWMAX> «
SLR = ° <4
Equation 4.12c reflected our concern with an LNAPL, which floated on the
heavier water, and so was not present below the water table. Free LNAPL must
exist at a given location in order for trapping to occur, so that occluded
product could not exist above the minimum free product depth b .
MMXN
Typical Profiles of Water and LNAPL
The total LNAPL saturation S was comprised of the free and residual
partitions, as described by Equations 4.2b and 4.12, respectively
S. - STR+S (4.13)
Lt Ltt\ LiC
Equations 4.3 and 4.8b comprised a profile model for the total moisture
saturation S, while Equations 4.2b, 4.3, 4.5, 4.11, and 4.12 specified the
vertical distribution of free [Lenhard and Parker (1990)] and residual [Parker
and Lenhard (1987)] LNAPL saturation.
Figure 4.2 displays water and LNAPL profiles for two typical cases, with
properties summarized in Table 4.1. Case 1 represented free and residual
49
-------
Q.
-------
TABLE 4.1. PROFILE PARAMETER VALUES FOR UNIFORM AND NONUNIFORM SOILS
Parameter Case 1 Case 2
7 4.00 (uniform) 1.80 (nonuniform)
r) 40 (weak hysteresis) 5 (strong hysteresis)
b. 5.9 m (free LNAPL) 6.5 m (no free LNAPL)
L
b 6.5m 6,5m
w
r 40 /«n (sand) 40 pm (silty sand)
a 0.074 N-m 0.074 N-m
WA
a 0.023 N-m (gasoline) 0.023 N-m (gasoline)
LA
p 700 kg/m (gasoline) 700 kg/m (gasoline)
L
VAX 6'8 m 6'8 m
A 0.7m 0.7m
were the historical extremes.
The total saturation curves in Figure 4.2 illustrated the well known
influence of pore size uniformity on moisture retention: liquid was confined
to a thinner capillary fringe for uniform sands, while nonuniform soils spread
moisture with gentler vertical gradients over more of the unsaturated zone.
The free LNAPL displayed in Figure 4.2a and 4.2b exhibited a sharp peak in the
vicinity of the steepest vertical moisture gradient, as was noted
experimentally by Ostendorf et al. (1991). Trapped LNAPL on the other hand,
was distributed more uniformly over a larger vertical interval, as noted in
Figure 4.2c and 4.2d. It should be remarked that Case 2 would not generate an
appreciable separate phase layer in a monitoring well, since the LNAPL was
trapped in the capillary fringe, while the free hydrocarbon of Case 1 would
generate a lens of product 0.6 m thick.
CORE SLEEVE DIFFUSION
The free and residual LNAPL evaporated in the soil, generating a source
term for hydrocarbon vapors. These vapors were carried away from the
capillary fringe by diffusion into static soil gas or advection into a moving
air stream. The diffusive process was induced on a macroscale basis by soil
51
-------
venting, which featured a horizontal air flow sweeping over the top of the
capillary fringe, thus inducing a vapor concentration gradient through the
contaminated soil. We simulated this mechanism in the intact core sleeves by
sweeping the headspace while maintaining static air in the sleeve. The
experiments were analyzed on a quasi steady basis in order to elucidate the
behavior of the vapor source term in the core sleeve.
Hydrocarbon Concentration Profile Model
The conservation of mass equation reduced to the following balance of
hydrocarbon vapor diffusion and LNAPL evaporation in the assumed absence of
advection, transience, and reactions:
with vapor concentration H, source term 0, and vertical distance z above the
water table. Density driven advection, which would induce a downward gas flow
[Falta et al. (1989)] since the LNAPL vapors were heavier than air, was
ignored because the bottom of the intact core sleeve was sealed during the
diffusion experiments. The effective gaseous diffusion coefficient D was
approximated with a Millington (1959) estimate
,2.33
<4.15a)
DA = D (-)')' (mR=0.114 kg) <4.15b)
R
D_ = 6.5xlO~6 m2/s (r -293 deg K) (4.15c)
R R
with free air diffusion coefficient D based upon Bruelle and Hoag's (1986)
A
isooctane reference D value, and the 7/4 temperature correction exponent
R
cited by Reid et al. (1987). The molar mass factor reflected the decrease of
diffusivity for heavier compounds [Baehr (1987)], and the reference mass mR of
isooctane was adopted. A Brooks and Corey (1966) characteristic (Equation
3.8c) was fit to the air porosity in this simple account of gaseous diffusion.
The source term was due to the local flux F_ of hydrocarbon vapor
LI
diffusing away from the LNAPL droplets of typical radius r., as sketched in
L
Figure 4.3. The liquid hydrocarbon was assumed to be trapped hysteretically
52
-------
rw--
R
Air (Large pores), D
H
-h
Hs
J Centerline
Water + Air (Small pores), DL
LNAPL (Small pores)
Figure 4.3. Water "aggregate" schematization, showing local mass transport
vapor concentration gradient.
[Parker and Lenhard (1987)] by free water, and a saturated vapor concentration
H existed at the LNAPL surface. The hydrocarbon vapor concentration was
s
assumed to fall to a "bulk" value H across a free water "aggregate" of radius
r. which limited mass transfer to the air stream beyond, as indicated by
Figure 4.3. The relatively low gaseous diffusivity D. of the aggregate was
L
due to the locally large free water content, and gave rise to a steep
concentration gradient between the LNAPL and the air stream boundary. The
vapor diffusivity (D) was high in the air stream by contrast, so that the
gradient was flat in this locale, as suggested by Figure 4.3. This local
53
-------
(microscale) schematization was analogous to the source term configuration put
forward by Crittenden et al. (1986) in the context of dissolved contaminant
transport.
A distance 2R separated adjacent aggregates, so that the source flux per
gross volume of subsurface soil was given by
3r 2F
0 - L3 ** <4.16a)
R
D (H -H)
FL - -*-*— (4.18a)
4
rw
(4.18b)
with source strength k. It should be noted that rw and DL were constants in
this conceptual model, while R and r varied directly with the moisture and
LNAPL contents, respectively. Thus, a completely saturated soil would have no
air stream, and R would approach r in magnitude. As the liquid aviation
W
gasoline evaporated, r decreased in size.
The gradient of gaseous diffusivity appearing in Equation 4.14 was much
steeper than the gradient of vapor concentration in the moist capillary
fringe, leading (in view of Equation 4.18) to an order reduction of the
diffusion analysis
54
-------
h = H -H (4.19b)
O
^ » 2
with relative vapor concentration h. The simplifying Equation 4.19c reflected
the sensitivity of D to the air porosity, which varied strongly with depth in
the intact core sleeve. We combined Equations 3.8, 4.15, and 4.19, leading to
the differential formulation
'
h - h_ (z=L) (4.20b)
Li
with known vapor concentration at the top of the column. Integration of
Equation 4.20 yielded the diffusive profile
H = H -(H -H )exp[- - r-rr] (4.21a)
R n 0 ^^
/ U-y~ ) ydy <4.2ib)
The integral function I was evaluated numerically using Gauss quadrature
[Abramowitz and Stegun (1972)] with the results sketched in Figure 4.4 to
assist in model usage.
With presumedly known values for the exit concentration, moisture
retention characteristics, and gaseous diffusivity, the remaining unknowns in
the profile model (Equation 4.21) were the saturated vapor concentration and
source strength for a given hydrocarbon constituent. We accordingly derived
an approximate H iteration predicated on the observed composition of a vapor
5
sample in the intact core sleeve. A simple and consistent diffusive flux
model yielded an estimated value for k based on a measured effluent port
concentration .
Saturated Vapor Concentration Iteration and Diffusive Flux Model
Table 4.2 presents an iterative and approximate estimate of the saturated
55
-------
Figure 4.4. Integral function I (Equation 4.21b).
vapor concentration H of each constituent of the blend of LNAPLs evaporating
s
from the capillary fringe. The iteration was predicated on the equivalence of
the observed constituent to total concentration ratio H/H in the vapor sample
to the corresponding saturated vapor concentration ratio HS/HST adjacent to
the LNAPL droplets in the core. Raoult's law [Reid et al. (1987)] described
the saturated vapor concentration for given liquid mole fractions x *n<* pure
constituent vapor concentrations H
56
-------
TABLE 4.2. SATURATED VAPOR CONCENTRATION ITERATION
Step 1 - Set
Step 2 - Compute
Step 3 - Compute HS=HST(|-).
H
Step 4 - Compute X™™-^ •
SP
Step 5 - Compare XM»M to Xnm anc* iterate until closure.
M£in OLD
Step 6 - Xg - XpINALHsp.
Hs = XHsp (4.22)
The chromatographic area fractions from the vapor sample served as the initial
estimate of the liquid mole fractions, as suggested by Step 1. The
temperature dependent Equation 3.3 was used to compute H for each compound,
SP
which was then multiplied by the appropriate x value. The sum of the *H
SP
products was equal to H and this total was proportioned among the various
sx
constituents in accordance with the observed H/H fractions, yielding H
T S
estimates in Step 3. We divided these H values by H__ in a new estimate of
o SP
the liquid fractionation. The process was iterated until the x values
converged, and the final source concentration estimate H followed as the
S
product of x and H .
SP
The flux F leaving the core sleeve at z equals L was computed
£i
analytically by taking the gradient of Equation 4.21 while invoking Equation
3.8 and the Millington (1959) diffusivity model with the result
F = ~D* (Z=L) (4.23a)
E
-
- <4-23b>
57
-------
r 2D L(n-0 ) V0
F* = \i» ^r^VV (4.23C)
rw (l.lla)
Equations 4.18b and 4.23b were combined and approximated to derive Equation
4.23c, a relationship between the efflux and the radius of the LNAPL droplets.
This radius was related to the depth integrated mass B of LNAPL on the
assumption of spherical masses of hydrocarbon
00
B - p-fTdz (4.24a)
V
*B
B « r_3 (4.24b)
Li
F_ « WB°*67 (H.»H.) (4.24c)
E 5 Li
We noted that the definition of B included the bulk (moist) soil density p ,
o
and the proportionality factor w was taken as essentially constant on the
assumption of a small H. and a constant H .
Li O
Equation 4.24c was substituted into the unsteady depth integrated
(macroscale) conservation of mass equation for sleeve diffusion
£ = -a*0'67 (4.25a>
at
B = BQ (t=0) (4.25b)
Equations 4.25b and 4.25c yielded a macroscale estimate of the diffusive flux
leaving the core sleeve which was compared with Equation 4.23b (for H »H ) in
O LJ
the derivation of an estimate of the source strength
-33- )2 (4.26m)
k = - ~ - - (4.26b)
Thus the source strength decreased with time due to drying up of the aviation
gasoline. It was stressed that the time scale of the LNAPL source term
evaporation was slow compared to sleeve diffusion, so that the earlier profile
analysis was quasi steady in nature.
-------
Diffusion in Uniform and Nonuniform Soils
We gained insight into the role of grain size uniformity in the
determination of capillary zone diffusion by simplifying the vapor
concentration profile. Equation 4.21a reduced to an explicit formula for the
typical case of a small exponential argument, which led by Taylor expansion to
the approximation
H - H [1- - - — ] (4.27)
3 -
The vapor concentration gradient approached a linear slope as we neared the
exit elevation L due to a small I for all cases. Equation 4.27 also suggested
that the integral function I, which varied with z, provided a rough measure of
the shape of the concentration profile since all other factors were constants
in the relation.
Figure 4.4 was thus used to assess the effects of soil uniformity on H.
Uniform soils had higher a's and lower V> values that resulted in linear
B
concentration gradients, exhibited by the the a=3 and L/V" =4 contour in Figure
B
4.4. Nonuniform soils, with lower uniformities and bigger bubbling pressure
heads, generated a nonlinear gradient, embodied by the a=1.2 and L/V>0=2 curve
B
in the figure. Physically speaking, we encountered less empty porespace in
moist, nonuniform capillary fringes, so that the gaseous diffusivity was
weaker by virtue of Equation 4.15. A steeper, nonlinear slope of vapor
concentration was needed to drive the evaporating hydrocarbons out of the soil
in this instance.
CORE SLEEVE ADVECTION AND MICHAELIS-MENTEN TYPE KINETIC CALIBRATION
Core sleeve advection and soil microcosms were the final two sets of
experiments run on the Cooperative Agreement. Nitrogen gas was blown directly
through the intact core sleeve in the advective work, in a simulation of air
sparging. Oxygen was intentionally deleted from the laboratory air stream to
preclude aerobic biodegradation, so that the mass transfer mechanics of
evaporation were isolated for study. The aerobic degradation potential of the
unsaturated zone (without the complications of mass transport) was the focus
of the soil microcosms.
59
-------
Core Sleeve Advection
Core sleeve advection in the absence of transience and reactions
proceeded on a comparatively simple basis as the balance of advection and a
source term
0 (zA) (4.28b)
dz
H = 0
M^V (4.29)
where the Brooks and Corey (1966) moisture characteristic (Equation 3.8a) was
invoked. The solution was readily obtained by separation of variables
B WR
H » H_ ( l-exp{ , , ,
S w(a-l)
H = H
L
n^V-1-!]))
(4.30a)
(A . Equation 4.31a was solved for the simple case
B
of a relatively small bulk air concentration by substituting Equation 3.8b for
the free water content with the result
60
-------
FE " ^S^'^WR (T]dz (Hg»H) (4.32a)
*B
wH (a-1) V
~ r (4-32b)
The source strength could thus be estimated from the measured efflux of
hydrocarbons from the core sleeve.
Equation 4.32a suggested that the advective efflux was proportional to
the source strength, as was the case with diffusion. A similar line of
reasoning then led to an analog of Equations 4.24c and 4.25a
PE - KB0'67 (4.33a)
dB 0.67
5£ " -*B (4.33b)
with coefficient K taken as a constant. Anticipating the use of pore volumes
in place of time for the advective experiments, Equations 3.7a and 4.33b were
combined and restated as
ALW dB _0.67
v- JS = -*B (4.34a)
o
B • B0 (C-0) (4.34b)
We integrated Equation 4.34 for the experimental condition of constant
specific discharges w for segments Af of displaced pore volumes with the
result
.. V.K
0 33 V
FE = K[B0 -5A^
)
-1 I-(a-l)-^()l (4.35C)
B rB
Equation 4.35b followed from Equations 4.33a and 4.35a. In view of Equation
4.32b, the source strength declined with advectively displaced pore volumes.
As with its diffusive counterpart (Equation 4.26b), air sparging evaporated
the LNAPL, reducing r and weakening k as a consequence.
61
-------
en
0.
3
(a)
0
Case 2
(Low w)
10
e
H, mg/m1
to
CL
(b)
0
Case 4
(Low w
20
r
H, mg/rrT
40
Figure 4.5. Typical vapor concentrations profiles for advective stripping:
(a) uniform soil; (b) nonuniform soil.
Advection in Uniform and Nonuniform Soils
Figure 4.5 displays vapor concentration profiles expected from the
interaction of evaporation and an advective stripping flow in uniform and
nonuniform soils. The soil and hydrocarbon characteristics are listed in
Table 4.3, and the k and H values corresponded to a strong LNAPL source term
s
of a volatile compound. Cases 1 and 2 described the profile in a uniform soil
subjected to relatively high and low air flows, respectively. The interaction
of advection and evaporation was apparent in the relative scales of the two
curves sketched in Figure 4.5a: since the mass transfer was limited
essentially by the local source term, the air stream could only carry off the
local flux delivered by diffusion through the water aggregate of Figure 4.3.
Thus higher w's must have induced lower concentrations to maintain an
essentially constant product wH leaving the capillary fringe.
Lt
Figure 4.5b simulated the effect of a nonuniform soil on the vapor
concentration profile. He recalled Equation 4.18a and noted that the higher
moisture content of the nonuniform soil gave rise to a shorter distance R
between adjacent aggregates and a stronger 0 source term for the same k value.
Thus the concentrations were stronger in Cases 3 and 4 than their uniform
counterparts. The increased slope of the low flow profile over the high flow
62
-------
TABLE 4.3. ADVECTIVE STRIPPING PARAMETER VALUES FOR UNIFORM
AND NONUNIFORM SOILS
Parameter
V m
A, m
k, a'1
n-0 _
WR
a
Hg, kg/m3
w, m/s
Case 1
0.30
1.00
io-5
0.30
2.50
0.040
io-5
Case 2
0.30
1.00
ID"5
0.30
2.50
0.040
io-6
Case 3
0.80
2.00
io-5
0.30
1.20
0.040
io-5
Case 4
0.80
2.00
io-5
0.30
1.20
0.040
ID'6
gradient reflected the conservation of mass equation (4.28a): since w
decreased, dH/dz must have risen to dispose of the same evaporative fi term.
Figure 4.6 illustrated the simulated stripping of LNAPL from the uniform
2
soil, assuming an initial mass content B of 0.5 kg/m for Cases 1 and 2. The
contaminated interval of 0.7 m implied by Table 4.3 yielded an initial mass
content of 380 mg LNAPL constituent/kg wet soil, averaged over the
contaminated capillary fringe. The H concentration of Table 4.3 reflected a
D
constituent liquid mole fraction of about 0.2 for a typical gasoline mixture,
so that the adopted B value was consistent with an average T of 1900 mg
LNAPL/kg wet soil. This represented heavily contaminated soil.
Both the remaining LNAPL and effluent vapor concentration are displayed
in the figure. We noted that sparging was a slow process, since over two
years were required to completely strip the LNAPL constituent out of the soil,
although half of the material was gone after about 4 months of air flow. The
insensitivity of removal efficiency to specific discharge was striking: Cases
1 and 2 both required roughly the same remediation period, although the low w
case required fewer pore volumes. The insensitivity of advective efflux to
air flow was another expression of the dominant role of local diffusion in
controling all larger scale transport mechanisms.
63
-------
0.0
Time, yrs
2000
Pore Volumes
4000
Time, yrs
0.4 -
C\2
E
en
CD"
200 400
Pore Volumes
0
600
Figure 4.6. Simulated performance of air stripping of LNAPL constituent from
uniform soil under high and low air flows.
64
-------
Michaelia-Menten Type Kinetic Calibration
The soil microcosms were treated as fully mixed chemostats in the
analysis of headspace concentration data. We remarked that this headspace
comprised an appreciable fraction of the VOA bottle volume, so that the
assumption of spatially uniform vapor concentrations throughout the device
needed to be justified as the first step in the mathematical analysis of the
degradation observations. The diffusive time scale t for full mixing in the
headspace was estimated to this end in accordance with
*D • r~ <4-36a)
A
t « 10 s (4.36b)
-5 2
with free air diffusivity and headspace length scale C of order 10 m /s and
A
10 m, respectively. The small thickness of the soil layer in the microcosm
r yielded a comparably small time scale in the soil as well, so that gaseous
S
diffusion acted swiftly to homogenize vapor concentrations in all the pore and
headspace of the laboratory vial.
The hydrocarbon concentrations accordingly decayed in time in the soil as
a depth integrated balance of storage, linear sorption, headspace influx F ,
A
and reaction [Ostendorf and Kampbell (1990)]
-F = -^0 <4'37a)
FA ' -fA d? (4'37b)
H = HQ (4. 37e)
with the substrate constrained sink fi defined by Michaelis-Menten type
kinetics [Alexander and Skow (1989)]
0=0 (tt0) (4.38b)
The maximum reaction rate V and half saturation constant K characterized the
kinetics in accordance with Figure 4.7, and were expressed in terms of gaseous
concentrations. An acclimation time t was adopted to permit the
reestablishment of a steady state active biomass in the microcosm vials. In
65
-------
V/2
K
H
Figure 4.7. Michaelis-Menten type kinetics.
this regard, the soil samples were continuously exposed to hydrocarbon vapors
for a prolonged period in the field, and the initial concentration dosage H
was representative of these field conditions. The form of Equation 4.38
assumed an abundance of oxygen and nutrients in the microcosms [Molz et al.
(1986)] in that a single half saturation constant was used.
The true retardation factor IL. was the conventionally defined parameter
[Freeze and Cherry (1979)] reflecting linear sorption onto grain surfaces and
into bound moisture in accordance with the soil bulk density and a
distribution coefficient K
(4.39)
.n
Due to the low organic carbon content of the sandy soil at Traverse City, the
true retardation factor was set equal to unity in the analysis of the
microcosms. The influx F of vapors into the soil surface was created by a
/\
storage change in the overlying headspace, embodied by Equation 4.37b:
physically speaking, gasoline fumes were degraded by bacteria in the soil and
replenished by vapor from the headspace.
Equations 4.37 and 4.38 were combined and solved by separation of
66
-------
Data o v
Theory
0.00
Figure 4.8. Calibrated degradation data from soil microcosm.
varibles with the resulting solution
-7"
(4.40a)
(4.40b)
with apparent retardation factor R ' comprised of true sorption to the soil
and pseudo sorption in the headspace [Ostendorf and Kampbell (1990)]. The
microcosm data were used to calibrate the half saturation constant and maximum
reaction rate by a nested Fibonacci search, with typical results (for a zero
acclimation time) sketched in Figure 4.8.
67
-------
SKCTIOH 5
RESULTS AMD DISCUSSION
LNAPL FIELD SAMPLING PRECISION
We compared the relative precision of the barrel extrusion and intact
core sleeve partitioning methodologies for the Cooperative Agreement. The
utility of online headspace sampling in delineating residually contaminated
soil was successfully documented as well. These studies were done on samples
collected at Stations SOBS and 50BT (Figure 3.5) on June 21 and 22, 1990, with
laboratory extractions and gas chromatographic analysis completed by September
1990.
Online Headspace Sampling
A single barrel extrusion (50xx Jars) and two adjacent (<2 m distant)
core sleeve boreholes (50xx Profile 1, 50xx Profile 2) were used to
characterize each station, so that a total of six boreholes were drilled at
the site during the June 1990 sampling trip. Headspace readings over the
Mason jar samples were used to identify the contaminated soil region, as
summarized by Table 5.1, Table 5.2, and Figure 5.1. The open circles in the
figure denote the meter readings, which registered high (including some off
scale) values over a fairly well defined interval roughly 5 m below the ground
surface. The meter readings approached their maximum levels gradually with
increasing depth through the unsaturated zone, reflecting a vertical gradient
and an upward diffusive transport of hydrocarbon vapor concentration from the
upper regions of the capillary fringe. The lower limit of separate phase
contamination in the saturated soil was much more sharply defined by the
headspace readings of Figure 5.1, due presumably to the buoyancy of LNAPL in
the ground water.
The open triangles in the figure represented the corresponding LNAPL
content in the extruded Mason jar samples. A positive correlation existed
between the headspace concentration and the aviation gasoline content, thus
confirming the utility of the Bacharach total organic vapor analyzer as an
68
-------
TABLE 5.1. OBSERVED HEADSPACE CONCENTRATIONS, MOISTURE AND LNAPL CONTENT
CORE BARREL SOBS JARS
ID Depth Interval Mid Interval
inches Depth, m
Grains
Grain6
Grainll
Grain9
Grain28
Grain26
Grain24
Avgasl
Avgas2
Avgas3, 4
AvgasS
Avgas6, 7
AvgasS
Avgas9
AvgaslO
Avgasll
Avgasl2,13
Avgasl4
Avgasl 5
Avgasl 6
Avgasl?
AvaaslS
63.0-72.0
117.0-126.0
138.0-146.0
154.0-162.0
168.0-171.4
174.5-178.0
181.4-184.5
188.0-191.4
191.4-194.5
194.5-198.0
198.0-204.0
205.0-207.5
207.5-210.0
210.0-212.5
212.5-215.0
215.0-217.5
217.5-220.0
220.0-222.5
222.5-225.0
225.0-234.0
244.0-250.5
263.5-270.0
1.72
3.09
3.61
4.02
4.31
4.48
4.66
4.83
4.91
4.99
5.11
5.25
5.31
5.38
5.44
5.50
5.57
5.63
5.69
5.84
6.29
6.79
M
ka moisture
ka wet soil
0.031
0.023
0.047
0.035
0.038
0.040
0.039
0.033
0.034
0.032
0.029
0.128
0.122
0.113
0.105
0.129
0.122
0.133
0.153
0.200
0.126
0.143
T 1
ka avaas
ka wet soil
<0. 00001
<0. 00001
<0. 00001
<0. 00001
<0. 00001
<0. 00001
<0. 00001
0.00002
0.00029
0.00025(32)*
0.00032
0.00022(41)
0.00031
0.00043
0.00049
0.00130
0.00704(22)
0.00232
0.00098
<0. 00001
<0. 00001
<0. 00001
•ieadspace
ppm
60
50
50
100
720
400
550
3000
4200
>10000
4700
>10000
3200
3600
3000
>10000
>10000
>10000
2200
2500
220
80
*Precision in percent.
online indicator of vertical contamination. Indeed, the headspace readings
were used in the field (without the benefit of the subsequently determined
Mason jar extrusion data) to specify the intact core sleeve sampling
intervals. The raw data at the two locations for the latter method are
summarized in Tables 5.3-5.6.
69
-------
TABLE 5.2. OBSERVED HEADSPACE CONCENTRATIONS, MOISTURE AND LNAPL CONTENT
CORE BARREL 50BT JARS
ID
6rain9
Grain?
Grainll
GrainlS
Grain23
Grain22
Avgasl
Avgas2
AvgasS
Avgas4, 18
AvgasS, 6
Avgas7
AvgasS, 17
Avgas9, 10
Avgasl 1,12
Avgasl 3
Avgasl4
Avgasl 6
Grain32
GrainlO
GrainS
Depth Interval
inches
57.0-66.0
93.0-102.0
129.0-138.0
153.0-162.0
162.0-165.0
165.0-168.0
168.0-171.0
171.0-174.0
180.0-183.7
183.7-187.5
187.5-191.2
191.2-195.0
195.0-198.7
198.7-202.5
202.5-206.2
206.2-210.0
216.0-223.0
227.4-229.6
231.8-234.0
234.0-240.0
240.0-246.0
Mid Interval
~ ^,- kg
Depth, m ^~
1.56
2.48
3.40
4.01
4.16
4.24
4.31
4.39
4.63
4.73
4.82
4.91
5.01
5.10
5.20
5.30
5.59
5.81
5.93
6.03
6.18
M
moisture
wet soil
0.041
0.036
0.040
0.044
0.086
0.050
0.051
0.038
0.070
0.053
0.075
0.140
0.121
0.136
0.136
0.190
0.104
0.129
0.163
0.180
0.188
T
ka avaas
ka wet soil
<0. 00001
<0. 00001
<0. 00001
<0. 00001
<0. 00001
<0. 00001
0.00029
0.00058
<0. 00001
0.00074(16)
0.00058(24)
0.00191
0.00137(7)
0.00119(34)
0.00883(25)
0.00015
<0. 00001
<0. 00001
Not Done
Not Done
Not Done
Headspace
ppm
10
10
60
200
750
750
4100
5200
1800
* 5400
3500
2600
1800
>10000
>10000
2000
240
90
80
20
10
*Precision in percent.
Comparative LNAPL Field Sampling Precision
(5.1)
Aviation gasoline sample replicate precision, defined by
. . I replicate value-replicate mean I
precision = ••— ~—~ *"
* replicate mean
is cited for the sleeve and barrel data in Tables 5.1-5.6. The precision
characterized the product uniformity in soil samples taken from the same
segment or jar and so reflected sample variation across lateral scales of
about 0.08 m and respective vertical scales of 0.03 m and 0.1 m. Table 5.7
70
-------
0
Headspace ppm
5000 10000 0
Heodspoce ppm
5000 10000
-2'
r- ~3*
E
\
£ -4T
0. 5
v 3
a >
-5
^
\
i
5 (a)
J
3
7 —
^P _.
F^^00 v 1
D
* i
-i
5
-2
5
-3
T
\
~4S
^
) -5
o f
7
i
5 (b)
3
_ _
>
3
^ O o
- ^^O r
7 O " ^
0.000 0.005 0.010
Avgos Moss Content
0.000 0.005 0.010
Avgos Moss Content
Figure 5.1.
Barrel extruded LNAPL data and combustible hydrocarbon meter
readings for Stations SOBS (a) and 50BT (b). Triangles and
circles represent LNAPL content (T) and jar headspace
concentrations, respectively.
summarizes the resulting statistics for the two sampling locations. The
sleeve partitions were more precise than the barrels, with segment average
precision varying from 13 to 23%, while the barrel range was 21 to 32%. The
precision of the sleeve segment replicates suggested a relatively uniform
horizontal distribution of the separate phase gasoline, in keeping with the
constancy of the capillary tension and resulting moisture content at a given
elevation. By the same token, a strong vertical variation of tension and
LNAPL content was expected, so that the coarser vertical sampling interval of
the jars gave rise to a decreased precision for the sampling method. This
expectation was borne out by the statistics of Table 5.7, which lists the
average precision for all the replicates run in a set of sleeve or core barrel
extractions.
71
-------
TABLE 5.3. OBSERVED MOISTURE AND LNAPL CONTENT
CORE SLEEVE SOBS PROFILE 1
ID
7
8,9
10
11,12
13
14,15
1,16
17,18
2,19
20,21
3,22
23,24
4,25
26,27
5,28
29,30
6,31
32,33
34
35.36
Depth Interval
inches
200.0-201.2
201.2-202.4
202.4-203.5
203.5-204.7
204.7-205.9
205.9-207.1
207.1-208.3
208.3-209.5
209.5-210.6
210.6-211.8
211.8-213.0
213.0-214.2
214.2-215.4
215.4-216.5
216.5-217.7
217.7-218.9
218.9-220.1
220.1-221.3
221.3-222.4
222.4-225.0
Mid Interval
Depth, m
5.10
5.13
5.16
5.19
5.22
5.25
5.28
5.31
5.34
5.37
5.40
5.43
5.46
5.49
5.52
5.55
5.58
5.61
5.64
5.69
M
kg moisture
ka wet soil
0.033
0.040
0.032
0.029
0.035
0.032
0.036
0.032
0.038
0.036
0.055
0.068
0.103
0.117
0.130
0.132
0.116
0.119
0.118
0.136
T
ka avaas
ka wet soil
<0. 00001
<0. 00001
0.00002
0.00021(43)*
0.00052
0.00034(18)
0.00049(32)
0.00046(11)
0.00033(32)
0.00064(34)
0.00046(7)
0.00087(11)
0.00062(8)
0.00047(8)
0.00085(9)
0.00159(29)
0.00084(5)
0.00142(0)
0.00091
0.00354(19)
*Precision in percent.
Estimates of the mass of aviation gasoline per horizontal area defined
exactly by Equation 4.24b were simply compared by summation of Tables 5.1-5.6
over the depth increments Az
B - 2(pBTAz) (5.2)
with wet soil bulk density computed in accordance with
wet soil mass
total volume
B
(n-0.367) (5.3a)
(ps=2650 kg/in ) (5.3b)
72
-------
TABLE 5.4. OBSERVED MOISTURE AND LNAPL CONTENT
CORE SLEEVE SOBS PROFILE 2
ID
7
8,9
10
11,12
13
14,15
16
17,18
1,19
20,21
2,22
23,24
3,25
26,27
4,28
29,30
5,31
32,33
6,34
35,36
Depth Interval
inches
185.0-186.2
186.2-187.4
187.4-188.6
188.6-189.7
189.7-190.9
190.9-192.1
192.1-193.3
193.3-194.5
194.5-195.6
195.6-196.8
196.8-198.0
198.0-199.2
199.2-200.4
200.4-201.5
201.5-202.7
202.7-203.9
203.9-205.1
205.1-206.3
206.3-207.4
207.4-210.0
Mid Interval
Depth, m
4.72
4.75
4.78
4.81
4.84
4.87
4.90
4.93
4.96
4.99
5.02
5.05
5.08
5.11
5.14
5.17
5.20
5.23
5.26
5.31
M
kg moisture
ka wet soil
0.027
0.032
0.036
0.058
0.141
0.074
0.050
0.069
0.039
0.031
0.042
0.039
0.038
0.044
0.040
0.048
0.057
0.062
0.072
0.077
T
ka avaas
ka wet soil
<0. 00001
<0. 00001
<0. 00001
0.00001
0.00139
0.00016(31)*
0.00001
0.00028(30)
0.00011(36)
0.00050(18)
0.00059(27)
0.00013(12)
0.00005
0.00024(8)
0.00035(3)
0.00065(23)
0.00071(7)
0.00104(2)
0.00221(24)
0.00112<12>
•Precision in percent.
Ostendorf (1990) estimated the porosity based on earlier coring work at the
site, and the solid grain density pg was appropriate for quartz sands. The
sleeve segments and Mason jars yielded respective B estimates of 1.40 and 1.77
2 2
kg/m at Station SOBS and 3.71 and 2.82 kg/m at 50BT. The replicate
precision of 12% and 14% for these estimates was excellent, and indicated that
the continuous pint size Mason jar profile resolution of 0.1 m was sufficient
to accurately determine the vertically integrated mass of LNAPL (B) at a given
location.
73
-------
TABLE 5.5. OBSERVED MOISTURE AND LNAPL CONTENT
ID
4
5
6
7
8,9
10
11,12
13
14,15
16
17,18
19
20,21
22
23,24
25
26,27
1,28
29,30
2,31
32.33
Depth Interval
inches
190.2-191.4
191.4-192.6
192.6-193.7
193.7-194.9
194.9-196.1
196.1-197.3
197.3-198.5
198.5-199.6
199.6-200.8
200.8-202.0
202.0-203.2
203.2-204.4
204.4-205.5
205.5-206.7
206.7-207.9
207.9-209.1
209.1-210.3
210.3-211.4
211.4-212.6
212.6-213.8
213.8-214.9
Mid Interval
Depth, m
4.85
4.88
4.91
4.94
4.97
5.00
5.03
5.06
5.09
5.12
5.15
5.18
5.21
5.24
5.27
5.30
5.33
5.36
5.39
5.42
5.45
M
kg moisture
ka wet soil
0.056
0.053
0.049
0.049
0.123
0.095
0.077
0.067
0.078
0.109
0.092
0.096
0.136
0.155
0.164
0.161
0.155
0.142
0.148
0.144
0.138
T
ka avaas
ka wet soil
0.00009
0.00435
0.00158
0.00154
0.00416(2)*
0.00296
0.00330(6)
0.00242
0.00820(8)
Not Done
0.00188(32)
0.00374
0.00596(33)
0.00002
0.00001
<0. 00001
<0. 00001
0.00203(24)
0.00093(54)
<0. 00001
<0. 00001
*Precision in percent.
Discussion
The appropriate field sampling protocol for LNAPL contamination depended
on the intended use of the data. Core barrel extrusion into pint size Mason
jars was relatively rapid, and was the method of choice for vertically
integrated data, with the headspace sampling protocol included for online
specification of maximum contamination intervals. The efficient determination
of vertically integrated LNAPL mass estimates by this protocol would
74
-------
TABLE 5.6. OBSERVED MOISTURE AND LNAPL CONTENT
CORE SLEEVE 50BT PROFILE 2
ID
5
6
7
8,9
10
12
13
14,15
16
17,18
19
20,21
22
23,24
1,25
26,27
2,28
29,30
3,31
32,33
4,34
35.36
Depth Interval
inches
178.6-179.8
179.8-181.0
181.0-182.1
182.1-183.3
183.3-184.5
184.5-185.7
185.7-186.9
186.9-188.0
188.0-189.2
189.2-190.4
190.4-191.6
191.6-192.8
192.8-194.0
194.0-195.1
195.1-196.3
196.3-197.5
197.5-198.7
198.7-199.9
199.9-201.0
201.0-202.2
202.2-203.4
203.4-205.4
Mid Interval
Depth, m
4.56
4.59
4.62
4.65
4.68
4.71
4.74
4.77
4.80
4.83
4.86
4.89
4.92
4.95
4.98
5.01
5.04
5.07
5.10
5.13
5.16
5.20
M
kg moisture
ka wet soil
0.051
0.050
0.033
0.034
0.036
0.050
0.051
0.057
0.051
0.110
0.103
0.113
0.096
0.104
0.159
0.120
0.127
0.138
0.123
0.120
0.127
0.154
T
ka avaas
ka wet soil
0.00008
<0. 00001
<0. 00001
<0. 00001
<0. 00001
0.00066
0.00100
0.00161(24)*
0.00163
0.00169(6)
0.00163
0.00092(3)
0.00107
0.00122(1)
0.00115(25)
0.00176(30)
0.00225(6)
0.00564(2)
0.00996(7)
0.01465(19)
0.01705(8)
0.00535(25)
*Precision in percent.
facilitate sampling at numerous horizontal locations for studies detailing the
lateral extent of separate phase pollution.
Vertically defined profiles for volatilization or dissolution model data
bases should be obtained with the complementary use of both sampling methods.
The intact core sleeves were labor intensive in the field and laboratory, and
should be used judiciously for a vertically resolved data base (0.03 m
intervals) at a few key locations. The barrel extrusion data defined much of
75
-------
TABLE 5.7. LNAPL SAMPLE PRECISION
ID
SOBS Profile
SOBS Profile
SOBS Jars
50BT Profile
50BT Profile
50BT Jars
1
2
1
2
Type
Sleeve
Sleeve
Barrel
Sleeve
Sleeve
Barrel
Number of Replicates
15
13
3
7
12
5
Precision
18
18
32
23
13
21
the vertical structure of the aviation gasoline distribution as well, although
the maximum contamination thickness of 0.2 m approached the method resolution
of 0.1 m at Traverse City. Thus the details of peak concentration were
somewhat obscured by the pint size jars, and the complementary use of the core
sleeves was well advised. In the absence of sleeve partitions, half pint
Mason jar sizes should be considered in zones of peak concentration, perhaps
with replicate barrel extruded boreholes for vertically defined data bases.
In this regard, half pint Mason jars would retain about 0.05 m of the barrel
and accordingly would increase the resolution and precision of barrel extruded
data, where permitted by constraints of field sampling and laboratory
analytical loads.
LNAPL PROFILE MODEL CALIBRATION
The existing models of free [Lenhard and Parker (1990)] and residual
[Parker and Lenhard (1987)] LNPAL were calibrated and tested with five sets of
total and LNAPL saturation data from Stations SOBS, 50BT, 50CE, 50CL, and
50CM, as located in Figure 3.5. Intact core sleeve segments were averaged
with barrel extruded Mason jar data for the first two cores, while barrel
extruded (1/2 pint) Mason jar data solely characterized the latter stations.
Equations 3.1 and 4.2 related the mass based observations to the volume based
measures used in our model
p T ,
ST - „ ln „ : L
with liquid aviation gasoline density reported by Ostendorf et al. (1989).
76
-------
TABLE 5.8. COMMON PARAMETER VALUES FOR LNAPL CALIBRATION AND TESTING
Symbol
Parameter
Value
Source
n
7
A
€
WA
LA
WR
-1
Porosity 0.367
Uniformity exponent 2.97
Scaling coefficient 8.15 m
Trapping coefficient 40
Hysteretical amplitude 0.354 m
Thickness of LNAPL 0.40 m
LNAPL density 707 kg/m3
Water/air surface tension 0.074 N/m
LNAPL/air surface tension 0.023 N/m
Irreducible moisture 0.059
Ostendorf et al. (1989)
50CE Calibration
50CE Calibration
50CL Calibration
NOAA (1992)
50CE Calibration
Ostendorf et al. (1989)
Streeter and Wylie (1979)
Lenhard and Parker (1990)
50CE Calibration
Model Calibration
Table 5.8 lists common parameter values adopted for all five stations.
Equations 4.3-4.13 suggested that the presence of free LNAPL influenced the
total saturation as well, so that S and S_ were jointly calibrated. A series
L
of nested Fibonacci searches was run through both S and S data sets, to
L
establish minimal statistics of the saturation error, with mean 5_ and
S
standard deviation a defined by
6 = —2[S(observed)-S(predicted)+
o ^N
[S (observed) -S (predicted)] +
SL( observed )-SL( predicted)
S (observed)-S_(predicted)
J-i J^
SLMAX(°bSerVed)
(5.5a)
SLMAX(°b8erVed)
}"*
(5.5b)
The joint calibration required that both total and LNAPL saturation data and
theory determine the error statistics. The use of the maximum observed LNAPL
saturation S to normalize the LNAPL error contribution placed equal weight
77
-------
on the total and LNAPL saturation in determining the error statistics, since S
had a maximum value of unity.
The hysteretical trapping factor was specified first, using the weakly
contaminated Station 50CL. Free product was neglected in the LNAPL profile at
this station, and the maximum observed concentration was used to estimate rj
through a simplification of Equation 4.11
SLRMAX * J <">>:L> (5'6)
The Table 5.8 value for r? reflected a maximum LNAPL saturation of 0.025 at
Station 50CL. The cited value for the fluctuation amplitude A was based upon
historical fluctuations of the water level of Lake Michigan, as measured at a
gaging station in Holland, Michigan [NOAA (1992)]. The difference between the
maximum and minimum monthly average lake level observed at the gage in a given
year was taken as a measure of the annual water table fluctuation at the Air
Station, due to its proximity to Lake Michigan (within 1300 m of Grand
Traverse Bay). The figure cited in Table 5.8 was the average of 22 such
amplitudes, taken from 1970-1991 records. The minimum residual LNAPL depth
was related to the minimum water table depth by the thickness of free product
in the soil at the time of entrapment
b = b -e (5.7)
MMIN WMIN
The representative value for e listed in Table 5.8 was observed at 50CE.
We used an approximate peak concentration matching relation to enforce a
relation between the van Genuchten (1980) pore size parameters (7,0. ) and the
LI
oil and water table depths (b. ,b__) for the free LNAPL profiles. The maximum
L W
observed free LNAPL saturation S. _„,._. and its depth b were assumed known.
Lir MAA U1AA
The matching relations stemmed from Equations 4.2b, 4.3a, and 4.5a, which
specified the free LNAPL concentration between b. and b , where the maximum
Li M
value resided
SLF = <1+[0L
Equation 5.8b was a Baylor expansion of 5.8a, valid for saturations near
unity. Approximate equations for the location of SLPMAX followed by setting
the derivative of 5.8b with respect to b equal to zero at depth
78
-------
TABLE 5.9. TOTAL SATURATION CALIBRATION RESULTS
Station
SOBS
50BT
50CE
50CL
50CM
Sampling Date
6/21/90
6/22/90
11/4/90
9/23/91
9/24/91
m
5.60
5.16
5.70
5.56
5.46
bw
m
5.77
5.50
6.10
5.56
5.75
WMAX
m
6.00
5.50
6.10
6.06
5.88
'•
11
9
0
0
2
'•
16
14
17
23
18
L
b = (b -b
DL ( W
LMAX
(5.9b)
V '
Equation 5.9 was imposed on Station 50CE in a nested Fibonacci search for 7,
0 , and b with the results summarized in Tables 5.8 and 5.9 and Figures
L WMAX
5.2 and 5.3. The uniformity parameter and scaling coefficient from the 50CE
calibration were adopted for all five stations due to the homogeneity of the
sand at the site.
The observed value and depth of peak LNAPL concentration were input into
the free LNAPL station (SOBS, 50BT, and 50CM) calibrations, which were one
parameter searches for optimal b values. The trapped Station 50CL
WHAX
prediction was a two parameter fit to b
WHAX
and the water table depth b
(which was equal to b in the assumed absence of free LNAPL). Tables 5.8 and
W
5.9 and Figures 5.2 and 5.3 summarized the calibration results for all five
stations.
Discussion
The range of the error mean (0 to 11%) and standard deviation (14 to 23%)
for the five stations indicated a reasonably accurate calibration of a complex
three phase system under field conditions. The matching of data and theory
79
-------
-2 -
a
a>
a
-4
-6 -
-2 -
a.
(D
a
-4
-6
O
50BT
O
0.0 0.5 1.0
Total Saturation S
0.0 0.5 1.0
Total Saturation S
CL
01
o
-2
-4
-6
50CM
0.0 0.5 1.0
Total Saturation S
a
-------
Q_
(D
Q
Q.
0)
Q
0.05 0.10 0.15
LNAPL Saturation
LNAPL Saturation
LNAPL Saturation
0.00 0.02 0.04
LNAPL Saturation
0.1 0.2
LNAPL Saturation
Figure 5.3. LNAPL saturation (S ) profiles, with data (circles) and predictions (curves)
81
-------
was particularly gratifying since total (Figure 5.2) and LNAPL (Figure 5.3)
saturations were interdependent and hence had to be jointly calibrated. The
use of a single set of van Genuchten (1980) retention parameters across the
five stations was a strong endorsement of the Parker and Lenhard (1987) and
Lenhard and Parker (1990) model application as well. The uniformity value of
2.97 was quite reasonable for a uniform fine sand, while the scaling factor of
8.15 m~ implied a mean pore radius r of 5.4x10 m, in view of Equation 4.4a.
This mean pore radius agreed remarkably well with the grain size based
estimate put forth by Ranjitkar (1989) on theoretical and experimental grounds
d (n-0 )
- _« - HE — (5.10)
-4
The mean grain size diameter d of 3.8x10 m thus implied a mean pore size of
5.6x10 m.
The LNAPL profiles at the heavily contaminated Stations 50BT, 50CE, and
50CM were very well described, both free and residual. There was a poorer fit
in the cleaner Stations (SOBS and 50CL) however, notably in the shallower
depths, which were constrained theoretically by an abrupt plateau of residual
contamination at the representative value chosen for b- A probabilistic
superposition of annual contributions to S would have in all likelihood
improved the residual model by simulating a gradual gradient of contamination
due to random variations of b,^_M about its deterministic "mean" value. This
HMXN
stochastic approach would have required a detailed record of site water table
fluctuations over the life of the plume though, and so was not pursued further
in the present analysis.
Cores SOBS, 50CE, and 50CL were within 5 m of each other at a location
roughly 80 m downgradient from the source plane. Thus, two weakly polluted
soil profiles 15 months apart bracketed a strongly contaminated free LNAPL
profile taken at the same location. The plume migration of 260 m over a 23
year period implied an apparent LNAPL average linear velocity of 3.5x10 m/s
which, when multiplied by the 15 month interval, yielded a lateral length
scale of 10 m for the "raft" of free LNAPL. The vertical scale of the mobile
separate phase structure was 0.2 m (e/2), while its saturation was of order
0.05. Thus gasoline masses of 200 kg were thought to travel as coherent
structures through the upgradient region of the plume at Traverse City. The
82
-------
Figure 5.4.
Distribution of Constituent Groups
by Vapor Density
•J.W
*— s
£
DW Surface
Ul
•
en
"5
CD
+j
a
V
o
6.0
1 1 1 1
50BSP1
0 f • V
n ° J*
O •
O 9
0 rs •t.
O 0v
%]pn
v » o
Vapor Density V • O
O High
• Medium
V Low
1 L. 1
10 20 30 40
Percent of Total LNAPL Content by Mass
Vertical distribution of LNAPL volatility fractions in a
segmented core sleeve.
speed of these mobile partitions was retarded by the "pseudosorption" of LNAPL
to the stationary residual phase [Ostendorf (1990)] as it fluctuated through
the capillary fringe.
Lenhard and Parker (1990) noted that the difference between the oil and
water table depths was identically equal to the depth of free LNAPL in a
monitoring well, so that one would expect to see the episodic appearance of
0.3 m thick lenses of product as the mobile rafts passed through monitoring
well screens. The phenomenon had in fact been reported by the US Coast Guard
engineers at the site.
The gas chromatographic analysis of the methylene chloride extracts
provided speciated data for every sample. In a heuristic exercise, the
constituents of the gasoline were grouped into volatile, semi-volatile, and
nonvolatile fractions defined by (Table 3.1)
83
-------
volatile-(H__>0.15 kg/m3)
Of
semivolatile-(0.15>HC_>0.08 kg/m3)
OF
nonvolatile-(H<0.08 kg/m )
SP
Table 3.1 indicated that the four lightest standard compounds comprised the
volatile set, while the next three eluting constituents formed the
semivolatile fraction. The remaining three standard compounds and all the
nonstandard species (which arrived after the standards, as indicated by Figure
3.4) were all lumped into the nonvolatile category.
Figure 5.4 displays a typical vertical structure associated with this
volatility. Light compounds with strong volatility were less pronounced
towards the top of the contaminated capillary fringe due to the abundance of
void space in this region. Heavy compounds that did not evaporate dominated
the upper soil, while the moderately volatile fraction remained essentially
uniform throughout the capillary fringe. The research has generated a
substantial data base that will support future quantitative models of
speciated transport of LNAPLs in intact cores. Some aspects of speciated
transport were pursued in the intact core sleeve studies.
FIELD TRAPPING OF HYDROCARBON VAPORS
The performance of the the field trapping protocol was assessed through
chromatographic comparisons, desorption efficiency studies, breakthrough
traps, and field spikes. We fitted the hydrocarbon diffusion model to trapped
data at Station 50CL as well.
Chromatoaraphv
Example chromatograms for a trapped sample and a 0.0449 kg/m laboratory
standard are shown on Figure 5.5. Retention times and relative peak areas for
the chromatograms of Figure 5.5 are listed in Table 5.10. The retention times
for the sample constituent peaks were within 1.5% of the standard, an
excellent precision in view of the large number of internal traps and
desorptions in the protocol.
The agreement between the observed hydrocarbon vapor fractionation of the
trapped standard (Table 5.10) and the vapor standard mass fractions cited in
84
-------
TABLE 5.10. STANDARD AND SAMPLE
Compound Elution
Order
2,3 dimethyl but ane
2,4 dimethylpentane
2,3 dimethylpentane
2,2,4 trimethylpentane
2,4 dimethylhexane
2,3,4 trimethylpentane
2,3,3 trimethylpentane
Toluene
2,2,5 trimethylhexane
1
2
3
4
5
6
7
7
8
CHROMATOGRAMS FOR TRAPPED VAPORS ( FIGURE 5.5^
Standard
Retention Mass
Time, min Fraction**
9.08
9.89
10.71
10.93
11.71
12.24
12.47
12.47
13.60
22.2
12.0
18.0
22.9
1.9
6.2
12.0
*
2.4
Sample
Retention Mass
Time, min Fraction**
9.22
9.99
10.80
11.03
11.80
12.35
12.57
12.57
13.70
2.3
3.1
9.4
33.9
3.0
16.1
17.6
*
0.5
*2,3,3 trimethylpentane and toluene coeluted.
**FID area fractions.
Table 3.4 also endorsed the desorption method chromatography. As with the
LNAPL analyses, we noted a fairly close correspondence between the FID
response and the mass composition of the mixture of hydrocarbons, suggestive
of similar combustion among standard constituents. As indicated by Table
5.10, toluene and 2,3,3 trimethylpentane coeluted during the trap desorption
analysis, although 2,3,4 trimethylpentane arrived with a distinct peak.
Refinement of the method might overcome this problem in future studies:
preliminary work with direct splitless vapor injection using the VOCOL column
separated these compounds, although the peaks were closely spaced.
Table 5.10 also suggested that the standard compounds comprised over 85%
of the ambient soil gas hydrocarbons in the field, a number in agreement with
LNAPL analyses of Table 3.3. The dominant compound in the trapped vapor was
2,2,4 TMP.
Method Evaluation
There were a number of potential sources of method error:
1. Incomplete desorption of external or internal traps
2. Background noise
85
-------
v
en
c
o
o:
o
20000
10000
Standard
8 10 12
Retention Time, min
14
en
c
O
Q_
u
o
10000 -
5000 -
0
8
1 4
10 12
Retention Time, min
Figure 5.5. Typical chromatograms for trapped vapor standard and sample.
3. Incomplete trapping or breakthrough
4. Contamination or loss during transport and storage
We compared peak areas from vapor standards trapped onto sorbent tubes
with direct injections onto internal Trap 2 of the Unacon desorber in an
assessment of incomplete trapping or desorption of the external sorbent tubes.
The resulting chromatogram areas had a 0.3% precision, implying complete
86
-------
trapping and desorption efficiency of the external sorbent tube protocol. The
internal traps were not desorbed as effectively: eleven samples were
reanalyzed with a second internal desorption cycle yielding an average
concentration of 3.0 g/ro , an order of magnitude less than the first cycle
response. A third cycle reduced the concentrations to a background noise
value of about 1 g/m , distributed among nonstandard compounds. These results
indicated that about 2.0 g/m of hydrocabon vapors remained on the Unacon
traps after typical sample analysis, with an irreducible background noise
level of about 1 g/m . Since the trapped concentrations varied from about 30
to 100 g/m in the capillary fringe of Station 50CL, the internal desorption
efficiency ranged from 90 to 97% for the observations, and no correction
factor was introduced for blank or background concentration.
A pair of sorbent tubes was connected in series during the field sampling
trip to check for breakthrough of contamination. Analysis of the downstream
sorbent tube indicated little or no breakthrough of sample: the total
downstream concentration of 3.03 g/m was similar to that of the trip blank
(3.37 g/m ) and much less than 32.5 g/m obtained from the upstream trap. In
view of the high efficiency of the external trapping process, we concluded
that the breakthrough trap was contaminated not by the upstream sample, but
rather by storage and transport sources. In this regard, the chromatograms of
the breakthrough and trip blank traps were similar, with large fractions
comprised of 2,3 dimethylbutane and a lighter nonstandard compound. Field
spiked traps did exhibit a minor loss of the former constituent, so that the
sample traps themselves (along with other types of LNAPL samples collected on
the field trip) could have been sources of minor contamination. The field
spiked traps contained 89.8 g/m of vapor standard. Analytical results were
compared with a standard trap prepared similarly in the laboratory on the day
of analysis. The chromatograms indicated recoveries of 97 and 93% for the two
field spikes, so that sample losses were small for our study.
Comparison with Diffusion Model
The trapped hydrocarbon concentrations offered a resolved profile of
vapor levels through the capillary fringe under ambient conditions at Traverse
87
-------
Q.
-------
and Table 5.9. The water table depth of 5.56 m associated with the van
Genuchten (1980) fit corresponded closely to the bubbling pressure depth (b -
w
V> ) of 5.45 m implied by the Brooks and Corey (1966) moisture calibration.
D
Independent values of vapor transport parameters supported a test of the
trapped concentration data, also summarized in Figure 5.6. The contaminated
interval A was set equal to 0.47 m in Equation 4.21, based upon the core
advection calibration discussed below, while the free air diffusivity and
temperature were 2.41x10 m /s and 285 deg K [Ostendorf and Kampbell (1991)].
-6 -1
The source strength k of 9.44x10 s followed from the observed natural flux
entering the unsaturated zone and Equation 4.22b, with a saturated vapor
concentration of 120 g/m inferred from the measured composition of a typical
trap sample and the iteration of Table 4.2. An 8% mean error and a 74% error
standard deviation are displayed on Figure 5.6, based on a hydrocarbon error
6 defined by
H
- H measured-H predicted _ ,,
5H ' H measured (5'11)
The low mean error was encouraging, since all the test variables were
independently specified. The broad standard deviation reflected the wide
scatter of the field data pictured on the figure, particularly in the wetter
regions of the core. This random variation was in part attributed to the
difficulties associated with vapor sampling from nearly saturated soil under
field conditions. The sample ports nearest the bottom of the intact core
sleeve posed the greatest challenge, since liquids were held with the least
tension, empty porespace was at a minimum, and the LNAPL source was most
concentrated in this region. It was accordingly not surprising to note the
widest scatter at the base of Figure 5.6, including one point exhibiting near
saturated vapor concentration. The sampling was simpler (and more precise as
a consequence) in the higher and drier part of the capillary fringe, and it
was reassuring to note the close correspondence of the trapped data and the
Ostendorf and Kampbell (1991) asymptote of 28.7 g/m at the top of the sleeve.
CORE SLEEVE DIFFUSION
The foregoing natural diffusion profile differed from the intact core
sleeve diffusion experimental program on a number of counts. The headspace of
89
-------
the core sleeve was swept in the laboratory, thus eliminating additional mass
transfer limitations of the unsaturated zone, and the room temperature was
considerably hotter than the site subsurface temperature. Both effects
accelerated the flux of contamination from the intact core sleeve. Individual
constituents were analyzed in the intact core sleeve diffusion experiments and
samples were withdrawn under controlled laboratory conditions. Thus better
sampling and analytical precision existed for the intact core sleeve data
base, as was reflected in the comparatively well behaved nature of the
observed concentrations for 2,2,4 TMP (Table 5.11) and 2,2,5 TMH (Table 5.12).
Moisture Content and Saturated Vapor Concentrations
As with the trapping experiments, a Brooks and Corey (1966) equation was
fit through the water content data at Station 50CE to quantify the profile of
air through the intact core sleeve used for diffusion under laboratory
conditions. Figure 5.7a showed the results of the Fibonacci search through 6
w
data, an exercise yielding a uniformity exponent a of 1.63 and a bubbling
pressure head of 0.257 m. These calibrated values, which were consistent with
their 50CL counterparts, were based on an irreducible moisture content of
0.059 and Brooks and Corey (1966) water table b of 6.00 m.
W
Table 5.13 summarized the results of the model testing and calibration
for 2,2,4 TMP and 2,2,5 TMH. The saturated vapor concentrations implied by
the vapor composition observed at a mid depth port (as computed by the
iteration of Table 4.2) are listed in the table, along with the run
temperatures. The 2,2,4 TMP H_ values ranged from 63.0 to 87.5 g/m over the
5
course of study, a limited variation that manifested its moderate volatility
and large, stable fraction in the liquid gasoline composition. The invariance
of the 2,2,4 TMP saturated vapor concentration supported its use in
calibrating source strength, since w (a function of H ) was assumed constant
s
in Equation 4.24.
The heavier 2,2,5 TMH saturated vapor concentration rose uniformly over
the four runs by contrast. In this regard, the ratio of saturated vapor
concentrations (2,2,4 TMP to 2,2,5 TMH) decreased from 6.6 to 3.5 over the
five month duration of the experiments, reflecting distillation of more
volatile compounds out of the liquid gasoline. The distillation phenomenon
90
-------
Port No.
1
2
3
4
5
6
7
8
9
10
Effluent
Depth, tn
5.48
5.51
5.53
5.56
5.62
5.65
5.68
5.71
5.75
5.77
7/22/91
0.0244
0.0250
0.0251
0.0258
0.0279
0.0283
0.0300
0.0322
0.0367
0.0394
0.00867
*Concentrations expressed in kg/m
TABLE 5.12.
Port No.
1
2
3
4
5
6
7
8
9
10
Effluent
Decth. m
5.48
5.51
5.53
5.56
5.62
5.65
5.68
5.71
5.75
5.77
OBSERVED VAPOR
7/22/91
0.00461
0.00473
0.00470
0.00490
0.00538
0.00548
0.00597
0.00642
0.00750
0.00785
0.00152
8/26/91
0.0123
0.0127
0.0128
0.0133
0.0144
0.0152
0.0162
0.0179
0.0198
0.0215
0.000835
3
•
CONCENTRATIONS .
8/26/91
0.00274
0.00284
0.00294
0.00308
0.00330
0.00350
0.00369
0.00428
0.00475
0.00494
0.000179
11/25/91
0.00579
0.00588
0.00591
0.00619
0.00635
0.00751
0.00799
0.00890
0.01000
0.01100
0.000259
1/3/92
0.00437
0.00440
0.00470
0.00467
0.00512
0.00556
0.00583
0.00657
0.00767
0.00839
0.000309
2,2,5 TRIMETHYLHEXANE*
11/25/91
0.00174
0.00165
0.00175
0.00184
0.00192
0.00226
0.00246
0.00281
0.00304
0.00326
0.0000751
1/3/92
0.00154
0.00152
0.00170
0.00165
0.00178
0.00196
0.00206
0.00236
0.00274
0.00304
0.000105
Concentrations expressed in kg/m .
91
-------
Ct
0)
Q
00
X
w
Water Content by Volume
Time, sxlO
Figure 5.7. Moisture and 2,2,4 TMP efflux at Station 50CE: (a) Brooks and
Corey (1966) curve fit through 0 data; (b) calibrated curve and
rf
observed (circles) efflux of 2,2,4 TMP from intact core sleeve.
was further evidenced by the dramatic decrease in the lighter constituent
vapor mass fractions cited in Table 3.4: the standard 2,3 dimethylbutane
fraction of 22.4% was based on an estimate of the LNAPL composition at the
outset of US Environmental Protection Agency research at Traverse City in
92
-------
TABLE 5.13. RESULTS OF
Date
HE
ka/m
k
-1
SLEEVE DIFFUSION RUNS
Hs
ka/m
HT ^H
ij n
ka/m %
°"
2,2,4 trimethylpentane
7/22/91
8/26/91
11/15/91
1/3/92
8.67xlO~3
8.35xlO~4
2.59xlO~4
-4
3.09x10
l.OSxlo"5
9.87xlO~6
3.73xlO~6
-6
1.54x10
0.0875
0.0694
0.0630
0.0735
0.0244 -11
0.0123 -24
0.00579 -15
0.00437 -2
4
9
5
5
2,2,5 trimethylhexane
7/22/91
8/26/91
11/15/91
1/3/92
1.52xlO~3
1.79xlO"4
_c
7.71x10
-4
1.05x10
9.88xlO~6
9.31xlO~6
_6
3.52x10
-6
1.46x10
0.0132
0.0139
0.0156
0.0209
0.00461 0
0.00274 -14
0.00174 -8
0.00154 4
6
6
6
8
1988. We noted a tenfold decrease in 2,3 dimethylbutane vapor concentration
four years later in the laboratory. Heavier constituents like 2,2,5 TMH
doubled in relative importance over the same period.
Efflux Measurements and Source Strength
The observed effluent port concentration H cited in Table 5.13 was used
to compute the flux of 2,2,4 TMP vapors out of the intact core sleeve in
accordance with
. 00457 m ) (5.12)
A nested Fibonacci search was used to calibrate Equation 4.26a to the four
observed fluxes, resulting in optimal parameter values for 2,2,4 TMP (Figure
5.7b)
2
BQ = 0.674 kg/m
« = 7.60X10'8 kg°'33/m0-67-s
(2,2,4 TMP)
(5.13a)
<5.13b)
93
-------
The error standard deviation of the flux calibration was 38%, reflecting the
experimental difficulties in regulation and measurement of low gas flows and
the simplifying assumptions in the derivation of Equation 4.26. The decrease
of efflux with time was clearly evident in Figure 5.7b, however. We also
noticed that Equation 5.13a was generally consistent with the methylene
chloride extract profiles from Stations SOBS, 50CE, and 50CL, which yielded
2
respective B values of 0.38, 1.85, and 0.77 kg/m for 2,2,4 TMP. The average
2
of these three values was 1.00 kg/m .
The 2,2,4 TMP source strengths implied by Equations 5.13 are listed in
Table 5.13. These latter k estimates yielded source strengths for 2,2,5 TMH
as well, in view of Equation 4.18b
"R 1/2
k(2,2,5 TMH) = k(2,2,4 TMP) (^) ' (m=0.128) (5.14)
with reference molar mass cited in Equation 4.15b. The molar mass factor
reflected liquid diffusion rates for the two compounds [Reid at al. (1987)]
and the resulting 2,2,5 TMH source strengths also appear in Table 5.13. The
initial Station 50CE source strengths were higher than the k value for 50CL
implied by the trapping results; a plausible finding since Station 50CE had
far more LNAPL (so that the r values were larger).
L>
Equation 4.18b led to a rough r. estimate of 10 m, based on values of
LI
0.001 m, 10~8 m2/s, and 3xlO~ s~ adopted for rr,, D., and k. The implication
W Li
was that discrete pore size droplets of LNAPL served as the source term,
rather than large continuous ganglia. These droplets were one fifth the size
of the mean pore radius at Traverse City computed in accordance with Equation
5.10.
Hydrocarbon Vapor Concentration Profiles
The observed vapor concentration for the four sample profiles are
sketched as symbols in Figure 5.8. The gradient decreased uniformly with
increasing time and elevation in all cases. The observed vapor concentrations
were 3 to 17 times smaller than their saturated counterparts, suggesting
strong diffusion gradients from the LNAPL droplets to the bulk air stream
(Figure 4.3). Figure 5.6 and 5.8 taken together showed the flattening of the
94
-------
-C
"o.
o
-5.5
-5.6
-5.7
0.00
0.02
f
2,2,4TMP, kg/mv
0.04
_C
•+-*
Q.
O
-5.5
-5.6
-5.7
0.000
0.005
f
2,2,5TMH, kg/m"
O 7/22
V 8/26
D 11/25
A 1/3
0.010
Figure 5.8. Observed (symbols) and predicted (curves) vapor concentrations
(2,2,4 TMP and 2,2,5 TMH) for intact core diffusion experiment.
concentration gradient with increasing time and evaporation of the LNAPL
source. The progressively weaker source term required a flatter vapor
gradient to drive the balancing diffusion term.
The independently calibrated source strengths of Figure 5.7b and
saturated vapor concentrations were input to the diffusive vapor profile model
95
-------
(Equation 4.21a) for comparison with the port data with the predictions
sketched as curves in Figure 5.8. The observed concentrations at the highest
port (depth of 5.48 m) were taken as H values, as cited in Table 5.13. Due
L
to the relatively heavy contamination in Station 50CE and short length of the
sleeve, we assumed that LNAPL persisted to the sleeve exit, and set L equal to
A in the analysis of core diffusion.
The Brooks and Corey (1966) water table depth was used to minimize the
vapor error standard deviation, with the vapor error 6 defined in accordance
H
with
t measured H-predicted H ._ ...
*H " measured H (5>15)
A single b value of 6.08 m was obtained for the eight runs so that the length
L was set equal to 0.604 m for all tests, with the results summarized in Table
5.13 and Figure 5.8. The error mean 3 varied from 0 to -24% with an overall
H
average value of -9%, while the error standard deviation o ranged from 4 to
9% with an overall value of 6% for the eight profiles. The reasonably close
correspondence of the moisture and vapor calibration water table depths for
Station 50CE (6.00 and 6.08 m, respectively) was within the accuracy and
terrain variation associated with drilling rig repositioning at adjacent
boreholes. It should be stressed that the scale of the profiles sketched in
Figure 5.8 was not tested by this data base, since observed H values were
used in the calculation of H. The form of the profiles was tested however,
since all other parameters in the model were fixed by independent data.
Soil Venting Considerations
We found no correlation between Q and F over the range of tested air
E*
flows, suggesting that local gradients controlled the diffusive efflux rather
than the headspace sweep flow. The finding was important, since it indicated
that low soil venting air flows in principle should work as well as higher
flow rates in spills limited by the microscale distribution of LNAPL. The low
flow would require lower energy costs and induce higher vapor concentrations
for better treatment efficiency. A minimum low flow value would be needed to
overcome mass transfer limitations imposed by the gaseous diffusion through
96
-------
the overlying unsaturated zone, however. Soil venting advection (QH ) was
L
compared qualitatively to this unsaturated zone diffusion mechanism in order
to establish a lower limit for the air flow
H
QHL > ^00^ (5.16a)
w
A D0
Q > -J— (5.16b)
W
2
Thus a treatment area A of 1000 m in a 5 m thick unsaturated zone with air
porosity and diffusivity values similar to that of Traverse City would require
-4 3
air flows in excess of 1.8x10 m /s, or 0.4 cfm, to overcome diffusion in the
unsaturated zone.
The results of the intact core sleeve study indicated no dramatic
increases in removal efficiency for air flows greatly in excess of the minimum
value (since mass transfer within the fringe governed the release rate for
soil venting). This modest air flow would need to be applied immediately
above the contaminated capillary fringe to ensure the removal efficiency.
Finally, we noted that the gaseous release from a NAPL source far above
the water table would be augmented considerably by advection down vertical and
lateral vapor density gradients [Mendoza and Frind (1990)]. This density
driven advection would be particulary important in the immediate vicinity of
the separate phase source [Sleep and Sykes (1989)]. The purely diffusive
release considered here would be operative only for laterally extensive LNAPL
sources resting on the water table, which was just the case at Traverse City.
In this regard, lateral density gradients were small over the wide expanse of
the plume, and the water table was an impermeable barrier to downward
advection.
CORE SLEEVE ADVECTION
Advection experiments were conducted on an intact core sleeve obtained
from Station CL, so that the Brooks and Corey (1966) moisture characteristic
of Figure 5.6 was appropriate. Thus a uniformity coefficient of 1.87 and a
bubbling pressure head of 0.30 m were used in the assessment of the 2,2,4 TMP
and 2,2,5 TMH vapor concentration data, which are listed in Table 5.14 and
sketched in Figures 5.9-5.11. The site porosity of 0.367 and irreducible
97
-------
TABLE 5.14. INTACT CORE SLEEVE ADVECTION DATA - mq/m"
Depth, m
7/13
7/15
7/16
7/17
7/20
7/22
7/23
2,2,4 trimethylpentane
5.10
5.13
5.22
5.25
5.28
5.34
5.40
5.43
5.46
5.49
5.55
0.402
0.424
0.391
0.382
0.352
0.371
0.366
0.343
0.336
0.296
0.133
0.110
0.101
0.105
0.0919
0.108
0.0977
0.0950
0.0974
0.0926
0.0879
0.0281
0.103
0.0968
0.113
0.0945
0.102
0.0972
0.101
0.102
0.0855
0.0945
0.0275
0.103
0.0888
0.0848
0.0905
0.119
0.0878
0.103
0.0938
0.104
0.0542
0.0346
0.0968
0.0937
0.0916
0.0902
0.0950
0.0863
0.0926
0.0779
0.106
0.0577
0.0303
0.0578
0.0710
0.0634
0.0660
0.0680
0.0670
0.0729
0.0618
0.0818
0.0421
0.0276
0.216
0.239
0.228
0.215
0.224
0.226
0.224
0.218
0.268
0.184
0.0803
2,2,5 trimethylhexane
5.10
5.13
5.22
5.25
5.28
5.34
5.40
5.43
5.46
5.49
5.55
0.389
0.433
0.373
0.376
0.319
0.345
0.336
0.286
0.292
0.227
0.102
0.0804
0.0685
0.0648
0.0644
0.0734
0.0869
0.0709
0.0582
0.0718
0.0376
0.0206
0.0632
0.0603
0.0693
0.0730
0.0653
0.0653
0.0664
0.0628
0.0639
0.0461
0.0213
0.0536
0.0523
0.0465
0.0527
0.0537
0.0530
0.0606
0.0473
0.0588
0.0326
0.0194
0.0730
0.0786
0.0739
0.0769
0.0747
0.0709
0.0807
0.0412
0.0887
0.0363
0.0303
0.0396
0.0503
0.0448
0.0523
0.0535
0.0503
0.0606
0.0503
0.0689
0.0330
0.0287
0.0966
0.101
0.105
0.0962
0.105
0.103
0.110
0.107
0.129
0.0844
0.0539
moisture content of 0.059 were assumed, along with a 50CL water table depth of
5.89 m that corresponded closely to the van Genuchten (1980) fit for the three
fluid saturation study.
Concentration Profiles
Generally speaking, the vapor profiles for both constituents exhibited
similar behavior: concentrations rose abruptly with elevation to an
asymptotic value marking the limit of the LNAPL source at z equals A. As was
98
-------
CL
-------
JC
"o.
o
-5.0
-5.2
-5.4
O 7/20
V 7/22
V i
V;
V
V!
O
0.00
0.05
i
2,2,4 IMP, mg/m'
O
D
O
0.10
Q.
0)
Q
-5.0
-5.2
-5.4
O 7/20
V 7/22
V
V
V
V
v i
O
O ...V
c
V
O
O
p
O
0.00
0.05
t
2,2,5 TMH, mg/ml
O
0.10
Figure 5.10. Observed (symbols) and predicted (curves) vapor concentrations
for core sleeve advection at 5 ml/min.
the simple advective stripping model of Equation 4.30 with the 14 measured
profiles. A nested Fibonacci search was run through the sets, with a common
contaminated interval A value used to minimize the vapor concentration error
(Equation 5.11) standard deviation for all the sets. Individual source
strengths comprised the second internal search variable, so that the error
100
-------
CL
Q
-5.0
-5.2
-5.4
0.00
O 7/15
V 7/16
D 7/17
D
D V ,
D V O
D
D
O
V
O
O
0.05
r_
2,2,5 TMH, mg/m"
0.10
Figure 5.11. Observed (symbols) and predicted (curves) vapor concentrations
for core sleeve advection at 10 ml/min.
mean for each run was separately minimized, and 14 different k's were
determined. Table 5.15 and Figures 5.9-5.11 summarized the results of the
calibration. The A value of 0.474 m yielded a total standard deviation of
14%, indicating an accurate calibration that was apparent in the figures as
well.
101
-------
TABLE 5.15. RESULTS OF SLEEVE ADVECTION RUNS
Date
7/13/92
7/15/92
7/16/92
7/17/92
7/20/92
7/22/92
7/23/92
7/13/92
7/15/92
7/16/92
7/17/92
7/20/92
7/22/92
7/23/92
Hs
ka/m3
0.0225
0.0253
0.0245
0.0260
0.0239
0.0255
0.0360
0.0167
0.0145
0.0145
0.0145
0.0169
0.0168
0.0146
FE
ka/m -s
2,2,4 trimethylpentane
4.15xlO~9
4.15xlO"9
3.95xlO~9
3.74xlO~9
1.88xlo"9
1.40xlO~9
2.30xlO~9
2,2,5 trimethylhexane
_g
3.71x10
2.80xlO~9
-Q
2.59x10
2.08x!0"9
1.43xlO~9
1.09xlO~9
_Q
1.09x10
k
-1
5.36xlO~6
4.74xlO~6
4.66xlO~6
4.16xlO~6
2.27xlO~6
1.59xlO~6
1.85xlO~6
-6
6.49x10
5.60xlO~6
-6
5.16x10
4.14xlO~6
2.46xlO~6
1.87x!0"6
-6
2.16x10
°«
6
10
11
13
10
13
10
10
14
13
11
26
21
16
The saturated vapor pressures cited in Table 5.15 were computed from
observed mid-depth chromatographic compositions, and suggested an essentially
constant liquid fractionation for the compounds studied. This constancy
contrasted somewhat with the diffusive experimental values cited in Table
5.13. We noted that the diffusion runs occurred at higher temperatures (about
10 deg K hotter than the advective experiments), giving rise to the higher
2,2,4 TUP saturated vapor pressures. The 2,2,5 TMH initially was a relatively
minor fraction of the LNAPL in the Station 50CE core, and was sensitive to
distillation effects as a consequence. At the advective Station 50CL however,
2,2,5 TMH was a much larger and more stable proportion of the LNAPL.
102
-------
Source Strengths and Advective Fluxes
Table 5 . 15 suggested that the source strengths used to zero the
individual run error means in the search varied in a physically plausible
fashion. The k's at a given time for the 2,2,4 TMP and 2,2,5 TMH profiles
were comparable to each other (as they should have been), and decreased with
time from 6x10 to 2x10 s over the course of study. These numbers
compared favorably with an "initial" source strength in excess of 9xlO~ s
inferred from the trapped profile test. This temporal decrease of k was
consistent with a shrinking LNAPL source mass droplet, as was mentioned in the
diffusive release discussion. The source strengths obtained in the diffusive
release runs, as cited in Table 5.13, approached the magnitude of their
advective counterparts in the latter phases of the stripping process. The
earlier k's at Station 50CE were considerably higher than the Table 5.15
values, due to the stronger degree of LNAPL contamination.
The calibrated 2,2,4 TMP hydrocarbon fluxes appearing in Table 5.15 were
used to evaluate the depth integrated parameters of Equation 4.35, which were
be expressed in temporal fashion as
A Fibonacci search through the 2,2,4 TMP runs yielded optimal values for the
advective proportionality factor and initial depth integrated mass in the
advective core sleeve
BQ = 0.00971 kg/m2 (2,2,4 TMP) (5.18a)
K = 1.20xlO~7 kg°'33/m°-67-s <5.18b)
The calibration is sketched in Figure 5.12a. We noted that the initial source
mass was far less than the B value found at Station 50CE through the
diffusion experiments (Equation 5.13a), again reflecting the much stronger
contamination found at the latter station. The observed total hydrocarbon
_o 2
flux of 1.15x10 kg/m -s reported by Ostendorf and Kampbell (1991) implied a
2
BQ value of 0.030 kg/m , when inserted into Equation 4.26 at time equals zero.
This total hydrocarbon estimate was consistent with Equation 5.18a.
The depth integrated mass of 2,2,4 TMP implied by Equations 4.35 and 5.18
(with time as the independent variable) is plotted in Figure 5.12b as a
103
-------
CD
I
o
1-H
X
I
6
DO
. Cd
Time, sxlO
0.0
Time, sxlO
2 64
Time, sxlO
Figure 5.12.
Calibrated (curves) and observed (symbols) advective stripping
of 2,2,4 TMP from core sleeve 50CL: (a) efflux vs time; depth
integrated mass vs time; (c) exit concentration for low
(3ml/min), medium (5 ml/min), and high (10 ml/min) air flows.
104
-------
function of time; the hydrocarbon compound was predicted to disappear after
about 60 days of sparging. This brief duration was a measure of the
relatively weak starting LNAPL contamination, and contrasted with the longer
simulation period of the heavily contaminated Figure 4.6. The expected exit
concentration appears in Figure 5.12c. The symbols represented the seven
calibrated exit concentrations of this study, appearing as asymptotes on
Figures 5.9-5.11. The low and high flow rates were well described by the
curves, although the 5 ml/min runs were a bit overpredicted. Figure 5.12c
underscored the insensitivity of removal to flow rate; higher flow rates
engendered lower exhaust concentrations but did not change the removal period.
Sparging Implications
Table 5.15 and Figures 5.9-5.11 had potentially important implications on
advective stripping, or sparging. The data bore out the conclusion reached in
the simulations of Figure 4.5: the uniform decline of F regardless of the
E
air flow rate implied that the stripping efficiency was independent of w.
Thus economical, low flow rates should be as effective as higher rates in
removing hydrocarbons from the capillary fringe. This effect was particularly
important for sparging, since it was difficult to inject large amounts of gas
into the saturated zone. We also anticipated that an essentially uniform flux
of hydrocarbon vapors would rise up through the zone of influence around a
sparging injection point, even though w diminished with radial distance from
the well. The findings here suggested that the vapor concentrations would
simply increase with radial distance so that the flux remained at a constant
value controlled by local mass transport effects of Figure 4.3.
These findings were all similar to those found for diffusive stripping,
which was also limited by the identical local transport process. Indeed, the
same source term was used successfully in both advective (Equation 4.30) and
diffusive (Equation 4.21) profile models, providing confidence in the validity
of the pore scale Bchematization. The depth integrated equations (Equations
4.25 and 5.17) offered proof as well in the rough equivalence of the
proportionalities relating F_ to the depth integrated mass. In this regard, w
JS
and K were within 40% of each other (Equations 5.13b and 5.18b), regardless of
their radically different source strengths and bulk air flow conditions. The
105
-------
configuration of the water aggregate between the LNAPL and the bulk air stream
dictated k, hence removal efficiency.
SOIL MICROCOSMS
Duplicate soil microcosms were taken from six depths through the
unsaturated zone at Station 50CL and subjected to a 1000 ppm dose of the
hydrocarbon vapor standard of Table 3.4 under aerobic conditions. Abiotic
control performance documented complete retention of vapors, so that the
observed decay of concentrations in the active vials was taken to represent
biodegradation. The data were fit to a Michaelis-Menten type reaction, and
the resulting kinetics were input to a coupled hydrocarbon oxygen degradation
model to be tested against existing field data with encouraging accuracy.
Abiotic Controls
It was difficult to hold hydrocarbon vapors in conventional microcosm
vials due to their relatively high volatility. Indeed, earlier studies showed
that containment of aqueous phase hydrocarbons posed a distinct problem in
hydrocarbon biodegradation analyses due to evaporation and subsequent vapor
leakage (Wilson et al. (1986), Hinchee et al. (1989)]. Previous
experimentation in our work showed abiotic vapor losses in excess of 10% per
day; the resulting decrease in vapor content was such that by the time the
soil microorganisms had acclimated to the administered vapor dosages, the
vapors had largely disappeared, and further measurement of vapor
biodegradation was rendered infeasible.
Figure 5.13a shows the results of preliminary abiotic control testing,
using various methods of closure. A conventional Teflon faced Mininert valve
with screw cap lost over 60% of the 1000 ppm dose in less than 250 hours, and
even poorer retention was exhibited by the addition of teflon tape to the vial
threads. The 155 ml serum bottles with aluminum crimp capped seals and thick
butyl rubber septa used to hold daily vapor standards were tried as long term
microcosms with little success. The substitution of a special "Tuf" septa for
Teflon in the Mininert valve resulted in rapid leakage as well.
We solved this experimental conundrum by applying an exterior water seal
to the laboratory microcosm vials; the lack of affinity between hydrocarbons
and water decreased the tendency of the hydrocarbon vapors to equilibrate with
106
-------
•.wU
-.25
" P"1 1
I'll
(a)
O C
o 8 °
° °<£° o °°° °o o °
•°° to Of g* Si g°o °8e c
c.
c
c
3
5
C.5C
L '
0.25 |-
C.OC >-
• Mininert
A Tape
• Butyl
* "Tuf"
o Water Seal
500
1000 1500 2000
TIME (hours)
1.50
1.25 -
1.00
»_ 0.75
H0
0.50
0.25
0.00
o o
o o
°
*«..o'
°
O MIABA
A MI ABB
v 50CLMIA6
• 50CLMA10
i i i i
0 500 1000 1500 2000 2500 3000 3500 4000 «500
TIME (hours)
Figure 5.13. Abiotic control performance: (a) conventional microcosms: (b)
long term retention of hydrocarbon vapors in water seal
microcosms.
107
-------
the atmosphere surrounding the vial, thus reducing leakage. Figure 5.13b
documented the remarkable effectiveness of the design: 100% of the vapors
were retained over the study period, and no abiotic correction factor was
necessary in the data analysis. Two sets of abiotic controls were prepared as
for the laboratory microcosms, with soil excluded from one set. Sterilization
of soil-containing abiotic controls was achieved by autoclaving previously
prepared soil microcosms at 121 deg C for one hour on each of three
consecutive days [Liu et al. (1990)].
Calibrated Michaelis-Menten Type Kinetics
The microcosm data were used to calibrate Michaelis-Menten type
degradation, as given by Equation 4.40, with the true retardation factor set
equal to unity in view of the low ambient organic carbon content of the soil
at the site [Twenter et al. (1985)] and nonarid condition of the unsaturated
zone. The half saturation constant was set at a literature value for
hydrocarbon vapor degradation [Ostendorf and Kampbell (1990,1991)]
K = 0.001 kg/m3 (5.19)
The apparent retardation factor R ' was equal to 2.5, based on packing of the
laboratory microcosm vials with a 50% air porosity and an approximately 5/8
full serum bottle. Table 5.16 and Figure 5.14 summarize the nested Fibonacci
search for the optimal maximum reaction rate and acclimation time associated
with the replicated data. The mean and standard deviation a of the microcosm
M
error 6 defined by
M
, _ H measured - H predicted 5 2Q
M = H '
O
were minimized by the two parameter search. We used the dose concentration to
normalize the error to prevent small values from skewing the calibration. The
Michaelis-Menten type reaction accurately modeled the observed decay of
contamination, as seen in Figure 5.14a and the relatively low standard errors
of the table.
The optimum V was quite uniform in soil from 0.6 to 3.5 m below the
ground surface, with an average value given by
V - 3.32xlO~8 kg/m3-s (mid depth) (5.21)
-6
The mid depth value lay between the acclimated surface soil rates (6x10
108
-------
1.2
C.8
o 50CLMI5A
» 50CLMI5B
Prediction
tQ - 33 hours
5
H
0.2
0.0
(a)
0 25 50 75 IOC '25 '.50 175 200
TIME (hours)
1.50
1.25
°'75
0.50
0.25
0.00
1 \ 1 1 —i r— 1 1
o 50CLM37A
• * 50CLM37B
• • •• " •• 50CLM38A '
1 " • Pred. - 37A
" B • • • Pred. - 378
•~ ~°I~ "• o •• • . * - —Pred - 38A'
80 0^»
(b)
0 250 500 750 1000 1250 1500 1750 2000 2250
TIME (hours)
Figure 5.14. Biodegradation of total hydrocarbon vapor concentrations in soil
microcosms: (a) 1.62 m depth at Station 50CL; (b) 4.58 m depth
at Station 50CL.
109
-------
Sample ID
50CLMI1A,1B,3B
50CLM5A,5B
50CLM29A,29B
50CLMI9A,9B
50CLM33A,33B
50CLM37A
50CLM37B
50CLM38A
Depth
m
0.64
1.63
2.26
2.54
3.46
4.58
4.58
4.58
Ho
ka/m
0.0040
0.0044
0.0037
0.0041
0.0035
0.0027
0.0031
0.0038
fco
8
4
5.40x10
5
1.19x10
5
2.81x10
5
2.52x10
4
2.88x10
4
3.24x10
5
4.93x10
g
1.58x10
V
ka/m -B
-B
3.14x10
-8
3.48x10
-8
3.52x10
-8
3.07x10
-8
3.38x10
-9
1.20x10
-g
2.80x10
-9
1.30x10
aM
%
7
6
11
15
9
9
9
20
kg/m -s) observed at the site by Ostendorf and Kampbell (1990) and the much
-9 3
slower rate of 2 xlO kg/m -s found at the 4.48 m level. We noted the
presence of abundant organics in the topsoil at the site, which fostered the
commonly observed phenomenon of accelerated microbiological activity in this
region. Kampbell et al. (1990) measured abundant oxygen and hydrocarbon vapor
substrate in the mid depth region at Traverse City, which perhaps accounted
for the uniformity of V exhibited by our microcosms there. The oxygen died
out with depth due to aerobic demand, while the hydrocarbon vapors rose to
relatively high (perhaps inhibitory) concentrations near the LNAPL
contaminated capillary fringe. Thus biodegradation was strongly reduced in
the vicinity of the water table.
The behavior of the calibrated acclimation time was inconclusive,
although a general increase of t with depth may possibly be inferred from
Table 5.16. The mid depth microcosms were all active within 3 days of the
dose, while the 4.48 m bacteria took as long as 3 weeks to acclimate to the
hydrocarbons.
In view of Eguation 4.38b, the Michaelis-Menten kinetics were
characterized by a reaction time t given by
c
t - * (5.22a)
110
-------
t = 3.11x10 s (raid depth) (5.22b)
G
This time was thought of at the period required for the microorganisms to
appreciably reduce the Vapor concentration. The uniform mid depth kinetics of
Equations 5.19 and 5.21 yielded the mid depth reaction time of Traverse City
in Equation 5.22b: hydrocarbon vapor concentrations in the unsaturated zone
decreased appreciably at the site over a 9 hour period.
Field Test ef Mieha^ti8^M6fl£«fl^Tvoe Kinetics
The uniform Michael is-Miiltefl type kinetics of Equations 5.19 and 5.21 and
soil gas cluster data cited in the field trapping discussion were appropriate
for testing the aerobic degradation model derived by Ostendorf and Kampbell
(1991). the model was a steady state balance of gaseous diffusion and
hydrocarbon limitid Michael is-Ment*fl type reaction with stoichiometric
coupling of oxygen O and hydrocar&ofl concentrations. The solution for the
hydrocarbon concentration was implicit
z - (^)1/2I(H*) (5.23a)
H* = J (5.23b)
1\
<5-23c>
.
with dimensionless concentration H and integral function I(H ) sketched in
Figure 5.15a. The hydrocarbon corteent ration had a known value H of 0.0287
kg/m at the top of the capillary fringe, where z was set equal to zero. The
hydrocarbon concentration decreased with increasing elevation due to diffusion
ind biodegradation mechanisms.
Oxygen was taken as a known value O equal to 0.112 kg/m at the base of
the root zone, where z equalled a value f of 4 m at Traverse City. The data
Were Observed in a stainless steel tubing cluster adjacent to Station 50CL and
set at 0.5 m depth increments through the unsaturated zone at Traverse City.
Portable tetal organic and oxygen meters were used to sample the soil gas over
a prolonged period of time [Ostendorf and Kampbell (1991)], and averaged data
at the 0.5 and 4.5 m depths were used for the H and O values. Oxygen
111
-------
KH*)
0.0
H /H
Figure 5.15.
H/HO,O/OC
Field test of hydrocarbon vapor degradation model [Ostendorf and
Kampbell (1991)]: (a) integral function I(H); (b) oxygen
(triangles) and hydrocarbon (circles) concentrations with
predicted curves.
112
-------
increased with increasing elevation due to downward diffusion and degradation,
and the profile was computed from its coupled hydrocarbon counterpart
O z
O = -p»T[H-H0(l-*)] (5.24)
In view of Equation 3.10, the Btoichiometric constant F was 3.51 for aviation
gasoline. Ostendorf and Kampbell (1991) estimated a gaseous diffusivity 0 of
2.41xlO~6 m /s at Traverse City.
The Michaelis-Menten type kinetic parameters were the remaining unknowns
in Equations 5.23 and 5.24, and the independently obtained values of the
present study were used to generate the predicted curves sketched in Figure
5.15b. The good fit between measured and predicted values in the figure
endorses this application of microcosm based kinetics to field scale
conditions.
113
-------
REFERENCES
Abramowitz, M. and I.A. Stegun. 1972. Handbook of Mathematical Functions.
National Bureau of Standards, Washington, DC, 1046 pp.
Abriola, L.M. and G.F. Finder. 1985. A multiphase approach to the modeling of
porous media contamination by organic compounds 1. Equation development. Water
Resources Research, v. 21, pp. 11-18.
Aelion, C.M. and P.M. Bradley. 1991. Aerobic biodegradation potential of
subsurface microorganisms from a jet fuel contaminated aquifer. Applied and
Environmental Microbiology, v. 57, pp. 57-63.
Alexander, M. and K.M. Scow. 1989. Kinetics of biodegradation in soil.
Reactions and Movement of Organic Chemicals in Soils, edited by B«L. Sawhhey
and K. Brown, SSSA, Madison, WI, pp. 243-269.
American Public Health Association. 1989. Standard Methods for the Examination
of Water and Waste. Port City Press, Baltimore, MD, pp. 1-18 to 1-20.
Baehr, A.L. 1987. Selective transport of hydrocarbons in the unsaturated none
due to aqueous and vapor phase partitioning. Water Resources Research, v. 23t
pp. 1926-1938.
Bednas, M.E. and D.S. Russell. 1967. Determination of natural gas leakage via
gas chromatography of drill core samples. Journal of Qas Chromatooraphv, v. 5,
pp. 592-594.
Benjamin, J.R. and C.A. Cornell. 1970. Probability. Statistics, and Decision
for Civil Engineers. McGraw-Hill, New York, NY, 684 pp.
Beveridge, G.S.G. and R.S. Schechter. 1970. Optimization; Theory and
Practice. McGraw-Hill, New York, NY, 773 pp.
Bouchard, D.C., C.G. Enfield, and M.D. Piwoni. 1989. Transport processes
involving organic chemicals. Reactions and Movement of Organic Chemicals in
Soils, edited by B.L. Sawhney and K. Brown, SSSA, Madison, WI, pp. 349-371.
Brooks, R.H. and A.T. Corey. 1966. Properties of porous media affecting fluid
flow. Journal Irrigation and Drainage Division ASCE. v. 92, pp. 61-88.
Bruelle, C.J. and G.E. Hoag. 1986. The diffusion of gasoline range hydrocarbon
vapors in porous media. Proceedings Petroleum Hydrocarbons and Organic
Chemicals in Ground Water. NWWA/API, Dublin, OH, pp. 420-443.
114
-------
Conrad, S.H., J.L. Wilson, W.R. Mason, and W.J. Peplinski. 1992. Visualization
of residual organic liquid trapped in aquifers. Water Resources Research, v.
28, pp. 467-478.
Corapcioglu, M.Y and A.L. Baehr. 1987. A compositional multiphase model for
groundwater contamination by petroleum products 1. Theoretical considerations.
Water Resources Research, v. 23, pp. 191-200.
Crittenden, J.C., N.J. Hutzler, D.G. Geyer, J.L. Oravitz, and G. Friedman.
1986. Transport of organic compounds with saturated groundwater flow: model
development and parameter sensitivity. Water Resources Research, v. 22, pp.
271-284.
Devitt, D.A., Evans, R.B., Jury, W.A., Starks, T.R., Eklund, B., and Gnolson,
A. 1987. Soil Gas Sensing for Detection and Mapping of Volatile Organics.
NWWA, Dublin, OH, 270 pp.
Falta, R.W., I. Javandel, K. Pruess, and P.A. Witherspoon. 1989. Density
driven flow of gas in the unsaturated zone due to the evaporation of volatile
organic compounds. Water Resources Research, v. 25, pp. 2159-2169.
Farmer, W.J., M.S. Yang, J. Letey, and W.F. Spencer. 1980. Hexachlorobenzene:
Its vapor pressure and vapor phase diffusion in soil, Soil Science Society of
America Journal, v. 44, pp. 676-680.
Freeze, R.A. and J.A. Cherry. 1979. Groundwater. Prentice-Hall, Englewood
Cliffs, NJ, 604 pp.
4-Glaccum, R., Noel, M., Evans, R., and McMillion. L. 1983. Correlation of
geophysical and organic vapor analyzer data over a conductive plume containing
volatile organics. Proceedings Third National Symposium on Aquifer Restoration
and Groundwater Monitoring. NWWA, Columbus, OH, pp. 421-427.
Grob, K. and K. Grob. 1974. Isothermal analysis on capillary columns without
stream splitting: The role of the solvent. Journal of Chromatoaraphv. v. 94,
pp. 53-64.
Hinchee, R.E. and H.J. Reisinger. 1985. Multiphase transport of petroleum
hydrocarbons in the subsurface environment. Proceedings Petroleum Hydrocarbons
and Organic Chemicals in Ground Water. NWWA/API, Dublin, OH, pp. 58-76.
Hinchee, R.E., D.C. Downey, J.K. Slaughter, D.A. Selby, M.S. Westray, and G.M.
Long. 1989. Enhanced bioreclamation of jet fuels-a full scale test at Eglin
AFB. ESL/TR/88/78. US Air Force, Tyndall AFB, FL.
Hoag, G.E. and M.C. Marley. 1986. Gasoline residual saturation in unsaturated
uniform aquifer materials. Journal of Environmental Engineering, v. 112, pp.
586-604.
115
-------
Holbrook, T. 1988. Combining two field methods of soil gas analysis to define
the horizontal and vertical extent of soil contamination. Proceedings
Petroleum Hydrocarbons and Organic Chemicals in Ground Water, NWWA/API,
Dublin, OH, pp. 91-104.
Hutchins, S.R., G.W. Sewell, D.A. Kovacs, and G.A. Smith. 1991a.
Biodegradation of aromatic hydrocarbons by aquifer microorganisms under
denitrifying conditions. Environmental Science and Technology, v. 25, pp. 68-
76.
Hutchins, S.R., W.C. Downs, J.T. Wilson, G.B. Smith, D.A. Kovacs, D.D. Fine,
R.H. Douglass, and D.J. Hendrix. 1991b. Effect of nitrate addition on
biorestoration of fuel contaminated aquifer: Field demonstration.
Groundwater. v. 29, pp. 571-580.
Jury, W.A., H. El Abd, and M. Resketo. 1986. Field study of napropamide
movement through unsaturated soil. Water Resources Research, v. 22, pp. 749-
755.
Jury, W.A., D. Russo, G. Streile, and H. El Abd. 1990. Evaluation of
volatilization by organic chemicals residing below the soil surface. Water
Resources Research, v. 26, pp. 13-20.
Kampbell, D.H., J.T. Wilson, H.W. Read, and T.T. Stocksdale. 1987. Removal of
volatile aliphatic hydrocarbons in a soil bioreactor. Journal Air Pollution
Control Federation, v. 37, pp. 1236-1240.
-Kampbell, D.H., J.T. Wilson, and D.W. Ostendorf. 1990. Simplified soil gas
sensing techniques for plume mapping and remediation monitoring. Proceedings
Fourth National Conference on Petroleum Contaminated Soils, Lewis Publishers,
Chelsea, MI, pp. 125-139.
Kampbell, D.H., Wilson, J.T., Griffin, C.J., and Ostendorf, D.W. 1992.
Bioventing reclamation project-aviation gasoline spill. Proceedings Subsurface
Restoration Conference. USEPA, Dallas, TX, pp. 297-299.
Kerfoot, H.B. and Mayer, C.L. 1986. The use of industrial hygiene samplers for
soil gas surveying. Groundwater Monitoring Review, v. 6, pp. 74-78.
Kia, S.F. and A.S. Abdul. 1990. Retention of diesel fuel in aquifer material.
Journal of Hydraulic Engineering, v. 116, pp. 881-894.
Kitunen, V.H., R.J. Valo, and M.S. Salkinoja-Salonen. 1987. Contamination of
soil around wood preserving facilities by polychlorinated aromatic compounds.
Environmental Science and Technology, v. 21, pp. 96-101.
Leach, L.E., F.P. Beck, J.T. Wilson, and D.H. Kampbell. 1988. Aseptic
subsurface sampling techniques for hollow stem auger drilling. Proceedings
116
-------
Second National Outdoor Action Conference on Aquifer Restoration. NWWA, Las
Vegas, NV, pp. 31-51.
Lee, M.D., J.M. Thomas, R.C. Borden, P.B. Bedient, C.H. Hard, and J.T. Wilson.
1988. Biorestoration of aquifers contaminated with organic compounds. CRC
Critical Reviews in Environmental Control, v. 18, pp. 29-89.
i Lenhard, R.J. and J.C. Parker. 1987. Measurement and prediction of saturation-
pressure relationships in three phase porous media systems. Journal of
Contaminant Hydrology, v. 1, pp. 407-424.
Lenhard, R.J. and J.C. Parker. 1988. Experimental validation of the theory of
extending two phase saturation-pressure relations to three fluid phase systems
for monotonic drainage paths. Water Resources Research, v. 24, pp. 373-380.
.Lenhard, R.J. and J.C. Parker. 1990. Estimation of free hydrocarbon volume
from fluid levels in monitoring wells. Ground Water, v. 28, pp. 57-67.
Liu, S.Y., M.H. Lu, and J.M. Bollag. 1990. Transformation of metolachlor in
soil inoculated with a Streetomvces sp. Biodearadation. v.l, pp.9-17.
Mendoza, C.A. and T.A. McAlary. 1990. Modeling of groundwater contamination
caused by organic solvent vapors. Groundwater. v. 28, pp. 199-206.
Mendoza, C.A. and E.O. Frind. 1990. Advective dispersive transport of dense
organic vapors in the unsaturated zone 2. Sensitivity analysis. Water
Resources Research, v. 26, pp. 388-398.
Millington, R.J. 1959. Gas diffusion in porous media. Science, v. 130, pp.
100-102.
Molz, F.J., M.A. Widdowson, and L.D. Benefield. 1986. Simulation of microbial
growth dynamics coupled to nutrient and oxygen transport in porous media.
Water Resources Research, v. 22, pp. 1207-1216.
National Academy of Sciences. 1984. Groundwater Contamination. National
Academy Press, Washington, DC, 179 pp.
National Oceanic and Atmospheric Administration. 1992. Daily Mean Water
Levels. Lake Michigan at Holland. Michigan. Rockville, MD.
Ostendorf, D.W. 1990. Long term fate and transport of immiscible aviation
gasoline in the subsurface environment. Water Science and Technology, v. 22,
pp. 37-44.
Ostendorf, D.W. and D.H. Kampbell. 1990. Bioremediated soil venting of light
hydrocarbons. Hazardous Waste and Hazardous Materials, v. 7, pp. 319-334.
..Ostendorf, D.W. and D.H. Kampbell. 1991. Biodegradation of hydrocarbon vapors
in the unsaturated zone. Water Resources Research, v. 27, pp. 453-462.
117
-------
'Ostendorf, D.W., O.K. Kampbell, J.T. Wilson, and J.H. Sammons. 1989.
Mobilization of aviation gasoline from a residual source. Research Journal
Water Pollution Control Federation, v. 61, pp. 1684-1690.
Ostendorf, D.W., L.E. Leach, E.S. Hinlein, and Y.F. Xie. 1991. Field sampling
of residual aviation gasoline in sandy soil. Ground Water Monitoring Review.
v. 11, pp. 107-120.
Parker, J.C. and R.J. Lenhard. 1987. A model for hysteretic constitutive
relations governing multiphase flow 1. Saturation pressure relations. Water
Resources Research, v. 23, pp. 2187-2196.
Finder, G.F. and L.M. Abriola. 1986. On the simulation of nonaqueous phase
organic compounds in the subsurface. Water Resources Research, v. 22, pp.
109S-119S.
Ranjitkar, S.B. 1989. Prediction of hydraulic properties of unsaturated
granular soils based on grain size data. Doctoral Thesis. CE Department,
University of Massachusetts, Amherst, MA, 205 pp.
Reible, D.D., T.H. Illangasekare, D.V. Doshi, and M.E. Malhiet. 1990.
Infiltration of immiscible contaminants in the unsaturated zone. Groundwater.
v. 28, pp. 685-692.
Reid, R.C., J.M. Prausnitz, and B.E. Poling. 1987. The Properties of Gases and
Liquids. McGraw-Hill, New York, NY, 741 pp.
Reinhard, M., N.L. Goodman, and J.F. Barker. 1984. Occurrence and distribution
of organic chemicals in two landfill leachate plumes. Environmental Science
and Technology, v. 18, pp. 953-961.
Rifai, H.S. and P.B. Bedient. 1990. Comparison of biodegradation kinetics with
an instantaneous reaction model for groundwater. Water Resources Research, v.
26, pp. 637-645.
Rifai, H.S., P.B. Bedient, J.T. Wilson, K.M. Miller, and J.M. Armstrong. 1988.
Biodegradation modeling at aviation fuel spill site. Journal of Environmental
Engineerinc. v. 114, pp. 1007-1029.
Robbins, G.A., B.G. Deyo, M.R. Temple, J.D. Stuart, and M.J. Lacy. 1990. Soil
gas surveying for subsurface gasoline contamination using total organic vapor
detection instruments. Groundwater Monitoring Review, v. 10, pp. 122-131.
Schwille, F. 1967. Petroleum contamination of the subsoil—a hydrological
problem. Joint Problems of the Oil and Water Industries. Institute of
Petroleum Engineers, New York, NY, pp. 23-54.
Schwille, F. 1988. Dense Chlorinated Solvents in Porous and Fractured Media.
Lewis Publishers, Chelsea, MI, 146 pp.
118
-------
Sleep, B.E. and J.F. Sykes. 1989. Modeling the transport of volatile organics
in variably saturated media. Water Resources Research, v. 25, pp. 81-92.
Song, H.G. and R. Bartha. 1990. Effects of jet fuel spills on the microbial
community of soil. Applied and Environmental Microbiology, v. 56, pp. 646-651.
Song, H.G., X. Wang, and R. Bartha. 1990. Bioremediation potential of
terrestrial fuel spills. Applied and Environmental Microbiology, v. 56, pp.
652-656.
Streeter, V.L. and E.B. Wylie. 1979. Fluid Mechanics. McGraw-Hill, New York,
NY, 562 pp.
Swallow, J.A. and Gschwend, P.M. 1983. Volatilization of organic compounds
from unconfined aquifers. Proceedings Third National Symposium on Aguifer
Restoration and Groundwater Monitoring. NWWA, Columbus, OH, pp. 327-333.
Taylor, J.K. 1987. Quality Assurance of Chemical Measurements. Lewis
Publishers, Chelsea, MI, 328 pp.
Twenter, F.R., T.R. Cummings, and N.G. Grannemann. 1985. Ground water
contamination in East Bay Township, Michigan. WRIR 85-4064. USGS, Lansing, MI,
63 pp.
US Army Corps of Engineers. 1970. Laboratory soils testing. EM 1110/2/1906.
Washington, DC, Appendix V.
Vandegrift, S.A. and D.H. Kampbell. 1988. Gas chromatographic determination of
aviation gasoline and JP-4 jet fuel in subsurface core samples. Journal
Chromatographic Science, v. 26, pp. 566-569.
Van Genuchten, M.T. 1980. A closed form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Science Society of America Journal, v.
44, pp. 892-898.
Wallingford, E.D., F.A. DiGiano, and C.T. Miller. 1988. Evaluation of a carbon
adsorption method for sampling gasoline vapors in the subsurface. Groundwater
Monitoring Review, v. 8, pp. 85-92.
Wilson, B.H., B.E. Bledsoe, D.H. Kampbell, J.T. Wilson, J.M. Armstrong, and
J.H. Sammons. 1986. Biological fate of hydrocarbons at an aviation gasoline
spill site. Proceedings Petroleum Hydrocarbons and Organic Chemicals in
Groundwater. NWWA/API, Houston, TX, pp. 78-90.
Wilson, J.L. and S.H. Conrad. 1984. Is physical displacement of residual
hydrocarbons a realistic possibility in aquifer restoration? Proceedings
Petroleum Hydrocarbons and Organic Chemicals in Groundwater. NWWA/API,
Houston, TX, pp. 274-298.
119
-------
Wilson, J.T. and M.J. Noonan. 1984. Microbial activity in model aquifer
systems. Groundwater Pollution Microbiology. Wiley Interscience, New York, NY,
pp. 117-133.
Wise, H.E., and P.D. Fahrenthold. 1981. Predicting priority pollutants from
petrochemical processes. Environmental Science and Technology, v. 15, pp.
1292-1304.
Zapico, M.M., S. Vales, and J.A. Cherry. 1987. A wireline piston core barrel
for sampling cohesionless sand and gravel below the water table. Groundwater
Monitoring Review, v. 7, pp. 74-82.
120
-------
APPKHDIZ
PUBLICATIONS AMD PRESKKTATIOHS
Here is a listing of the publications and presentations made under this
project, as of October 1992:
JOURNAL ARTICLES
Ostendorf, D.W., L.E. Leach, E.S. Hinlein, and Y.F. Xie. 1991. Field sampling
of residual aviation gasoline in sandy soil. Groundwater Monitoring Review, v.
11, pp. 107-120.
Richards, R.J., D.W. Ostendorf, and M.S. Switzenbaum. 1992. Aerobic soil
microcosms for long term biodegradation of hydrocarbon vapors. Hazardous Waste
and Hazardous Materials, in press.
Ostendorf, D.W., E.E. Moyer, Y.F. Xie, and R.V. Rajan. 1993. Hydrocarbon vapor
diffusion in intact core sleeves. Groundwater Monitoring and Remediation, in
press.
Ostendorf, D.W., R.J. Richards, and F.P. Beck, F.P. 1993. LNAPL retention in
sandy soil. Groundwater. in press.
PROCEEDINGS
Moyer, E.E. and D.W. Ostendorf. 1992. Field trapping of subsurface hydrocarbon
gasoline vapors. Proceedings Subsurface Restoration Conference. US
Environmental Protection Agency, Dallas, TX, pp. 171-173.
ABSTRACTS
Ostendorf, D.W., R.J. Richards, and F.P. Beck. 1991. Retention of residual
aviation gasoline in sandy soil. International Symposium on In Situ and On
Site Bioreclamation. Battelle, San Diego, CA.
Ostendorf, D.W., E.S. Hinlein, and L.E. Leach. 1991. LNAPL fate, transport,
and distribution at Traverse City, Michigan. 1991 Groundwater Research
Seminar. RS Kerr Environmental Research Laboratory, Oklahoma City, OK.
121
-------
Oatendorf, D.W., D.H. Kampbell, and E.E. Moyer. 1991. Laboratory and field
studies of the kinetics of bioventing. Symposium on Bioremediation of
Hazardous Wastes. US Environmental Protection Agency, Falls Church, VA.
Moyer, E.E. and D.W. Ostendorf. 1992. Aerobic biodegradation of hydrocarbons
in an unsaturated zone soil core sample. Focus Conference on Eastern Regional
Groundwater Issues. NGWA, Newton, MA.
Richards, R.J. and D.W. Ostendorf. 1992. Aerobic soil microcosms for long term
biodegradation kinetics of hydrocarbon vapors. Focus Conference on Eastern
Regional Groundwater Issues. NG -•- •
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard loth n
Chicago, IL 60604 3590 l00
122
*US GOVERNMENT PRINTING OFFICE 1992—750-002/60,123
------- |