UST CORRECTIVE ACTION TECHNOLOGIES:
Engineering Design of
Free Product Recovery Systems
i .
i by
i
Jack C. Parker,
Dap W. Waddill,
! and
Jeffrey A. Johnson
Environmental Systems & Technologies, Inc.
Blacksburg, Virginia 24060
Contract No. 68-C2-0108
Project Officer
i
Chi-Yuan Fan
Land Remediation & Pollution Control Division
National Risk Management Research Laboratory
Edison, N^ew Jersey 08837
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
Disclaimer
i
I
The information in this document has been partially funded by the U.S. Environmental
Protection Agency (EPA) under Contract No.; 68-C2-0108 to International Technology
Corporation and its subcontractor Environmental Systems and Technologies, Inc. It has been
subjected to the Agency's peer and administrative review, and has been approved for publication.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. i
11
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the
Nation's land, air, and water resources. Undejr a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future. :
|
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks from threats to
human health and the environment. The focus of the Laboratory's research program is on
methods for the prevention and control of pollution to air, land, water, and subsurface resources;
protection of water quality in public water systems; remediation of contaminated sites and ground
water; and prevention and control of indoor air pollution. The goal of this research effort is to
catalyze development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to support regulatory
and policy decisions; and provide technical support and information transfer to ensure effective
implementation of environmental regulations and strategies.
I ' ;
This publication has been produced as part of the Laboratory's strategic long-term
research plan. It is published and made available by EPA's Office of Research and Development
to assist the user community and to link researchers with their clients.
E. Timothy Oppelt, Director
Irrational Risk Management Research Laboratory
: 111
-------
Abstract
The objective of this project was to develop a technical assistance document for
assessment of subsurface hydrocarbon spills ahd for evaluating effects of well placement and
pumping rates on separate phase plume control and on free product recovery. Procedures
developed for estimation of hydrocarbon spill volume include interpolation and spatial integration
of measurements from soil cores, and fluid level data from monitoring wells. The first method
involves vertical integration of soil concentration measurements to yield oil volume or species
mass per unit area followed by kriging and areal integration to estimate the total mass or volume
within the measurement zone. This method is especially well suited to determine the amount of
residual product in the unsaturated zone. The second method involves kriging of well fluid levels,
calculation of free oil volume per area using a physically based model for vertically hydrostatic
three phase fluid distributions that converts wpll product thickness to soil product thickness,
followed by areal integration to estimate the vjolume of free product floating on the water table.
A procedure is presented to evaluate effects of steady-state water pumping from multiple point
sources on the oil flow gradients to evaluate if hydraulic control of plume spreading will be
obtained for a selected system of pumping wells and/or trenches. Estimates of residual oil in the
unsaturated and saturated zones are made from three phase capillary pressure-saturation relations
and from the initial oil thickness distributions and computed water table drawdown, which enable
determination of the recoverable spill volume for alternative well configurations. A variety of
practical examples and case studies are presented to illustrate the methodology and to
demonstrate how various factors interact to affect free product recovery system effectiveness. The
applicability of trenches and vacuum enhanced product recovery to hydrocarbon spills is also
discussed. |
I . - - -- -
This report was submitted in fulfillment of Contract No. 68-C2-0108 by International
Technology Corporation and its subcontracted Environmental Systems & Technologies, Inc.,
under the sponsorship of the U.S. Environmental Protection Agency. This report covers a period
from June 1993 to April 1995, and work was completed as of April 30, 1995.
IV
-------
Contents
Foreword j iii
Abstract ; iv
Tables |. vi
Figures • vii
Acknowledgment j be
1 Introduction j 1
2 Basics of Hydrocarbon Behavior . . . I 4
2.1 Water, Air andNAPLFlow . \ 4
2.2 Capillary Retention Relations | 5
2.3 Vertical Equilibrium Fluid Distributions 10
2.4 Residual Oil in Saturated and Unsaturated Zones 14
2.5 Oil Relative Permeability and Transmissivity 16
2.6 Estimation of Fluid Properties' 18
2.7 Estimation of Soil Properties i 21
3 Spill Assessment Methods 27
3.1 Interpretation of Soil Concentration Data " 27
3.2 Free Oil Volume from Monitoring Well Data 32
3.3 Estimation of Dissolved and Free Phase Travel 39
4 Product Recovery System Design . . . j. 43
4.1 Specification of Design Criteria 43
4.2 Effects Of Well Placement And Operation 45
4.3 Plume Capture and Travel Time Analysis 52
4.4 Estimation of Recoverable Product 56
4.5 Considerations in Using Trenches 63
4.6 Vacuum Enhanced Free Produjct Recovery 65
5 Case Study of Spill Site [ 71
5.1 Introduction j. 71
5.2 Model Application for Site Assessment 71
5.3 Model Results • , 74
5.4 Summary and Conclusions . . 1 78
j - •
References '. 79
v
-------
Tables
2-1 Soil and fluid properties for three phase flow
2-2 Typical specific gravity, oil-water viscosity ratio, and capillary
scaling factors for various hydrocarbon mixtures
2-3 Representative soil properties for various soils
3-1 Example spreadsheet calculations from soil boring data
3-2 Soil and fluid properties for example problem
3-3 Spreadsheet for free oil specific volume from well product thickness ....
4-1 Summary of recovery system results for Case Study I
4-2 Soil and fluid parameters for Case Study II
4-3 Pumping rates, time to reach asymptotic recovery and total water
pumped for Case Study II scenarios
4-4 Spreadsheet for calculation of residual and recoverable oil specific volume
5-1 Soil and fluid property values for niodel simulations
5-2 Results of SPILLCAD modeling simulations
5-3 Results of SPILLCAD regional water table fluctuations simulations .....
5-4 Optimal recovery characteristics for scenarios 1, 2, and 3
19
20
24
31
37
38
46
47
49
59
74
75
76
78
VI
-------
Figures
2-1 Water retention in a collection of capillary tubes 5
2-2 Water distribution in pores during imbibition and drainage 7
2-3 Hysteresis in capillary pressure function during water drainage and imbibition 7
2-4 Drainage air-water capillary pressure functions for different soil types 8
2-5 Distribution of air, oil and water in a porous medium 9
2-6 Three phase fluid distributions in equilibrium with a screened well 12
2-7 Free oil specific volume versus well product thickness
for gasoline in different soils . . . . j 13
2-8 Typical oil transmissivity functions for gasoline in sandy and silty soils 17
3-1 Schematic of baildown test over time 34
3-2 Baildown test hydrograph for a) sandy soil with diesel, and b) silty soil with diesel 35
3-3 Oil saturation distribution for van Genuchten model and "oil pancake" model for
a) sandy soil with diesel,. and b) silty soil with diesel 36
3-4 Relative oil mobility versus apparent oil thickness for two soils 40
3-5 Apparent thickness 100 and 700 days after a gasoline leak predicted by ARMOS . . 42
4-1 Apparent product thickness prior t0 recovery and well locations for Case Study ... 45
4-2 Initial well oil thickness distribution and location of recovery wells for Case Study II 48
4-3 Final product recovery vs. water pumping rate for Case Study II scenarios 50
4-4 Final unsaturated zone residual product versus water pumping rate for Case Studyll 50
4-5 Final saturated zone residual product versus water pumping rate for Case Studyll . 51
4-6 Flow net for travel time analysis. Solid lines are steady state Zaw contours 55
4-7 Comparison of recoverable oil volume versus water pumping rate 62
4-8 Comparison of unsaturated zone residual oil volume versus water pumping rate ... 62
4-9 Comparison of saturated zone residual oil volume versus pumping rate 63
4-10 Schematic of vacuum enhanced product recovery system 66
Vll
-------
Figures (continued)
4-11 Effects of VER on product recovery for system with a water pump rate of 52 gpm 67
4-12 Zones of influence for water and air phases in VER system with cover 68
4-13 Zones of influence for water and air phases in VER system without cover 68
4-14 Asymptotic product recovery versus liquid drawdown with and without VER .... 69
5-1 Plan map of xylene spill site : 72
5-2 Contour map of apparent free xylene thickness in feet 72
5-3 Groundwater contours at spill site in feet 73
5-4 Results of particle-tracking analysis for two well scenario 77
vm
-------
: s
Acknowledgment
This project was undertaken by Environmental Systems and Technologies, Inc., a
subcontractor to International Technology Corporation for the U.S. Environmental Protection
Agency (EPA) National Risk Management Research Laboratory (NRMRL). Support was also
provided through a cooperative undertaking by the American Petroleum Institute(API) and U.S.
EPA as part of a workshop entitled "Assessment, Control and Remediation of LNAPL
Contaminated Sites". Bruce Bauman and Bob Hockman with the API Soil and Groundwater
Task Force, Chi-Yuan Fan and Anthony N. Tafuri of the NRMRL, and Roy Chaudet of
International Technology Corporation were instrumental in enabling the latter project to be
undertaken. Technical review was provided by Milovan Beljin of International Technology
Corporation. Junlin Zhu, Ravi Narasimhan, Ram Pemmireddy and Caroline Bennett with ES&T
contributed significantly to the effort. \
IX
-------
Section 1
Introduction
The characterization and remediation of groundwater contamination is one of the most
challenging problems facing the environmental field today. In particular, the problems caused by
hydrocarbon spillage and disposal are both widespread and complex. For example, it has been
estimated that over 75,000 underground storage tanks annually release 11 million gallons of
gasoline to the subsurface (Hall and Johnson^ 1992). Given that from 100 to 150 compounds can
be identified in a typical gasoline, each having distinct physical and chemical characteristics,
significant scientific and technical knowledge is required to successfully manage the potential
environmental impacts and health risks associated with hydrocarbon releases in the subsurface.
I
Hydrocarbons are fluids that are immiscible with water and as such are considered
nonaqueous phase liquids (NAPLs). In general, most hydrocarbon compounds are less dense than
water and are termed light nonaqueous phase liquids (LNAPLs). When released in the subsurface,
LNAPLs remain distinct fluids and flow separately from the water phase. The downward
migration in the vadose zone is generally rapid, and depending upon the complexity of the
heterogeneities in the soil, may form an intricate network of pathways. Once in the vicinity of the
capillary fringe, hydrocarbons will spread horizontally with minimal penetration below the water
table due to buoyancy. Contact with groundwater as well as infiltrating recharge causes chemical
constituents to dissolve from the hydrocarbon into the groundwater resulting in contamination of
the aquifer. Further, volatile constituents may partition into and move in the soil vapor. Through
this complex array of physical and chemical processes, the hydrocarbon continually changes. It is
due to these processes that the characterization, containment, and remediation of hydrocarbons
pose unique and difficult problems for the environmental professional.
The first step in assessing a hydrocarbon spill generally involves delineating the vertical
and horizontal extent of soil and groundwater'contamination. Characterization may include visual
observations of soil borings, in situ vapor readings, laboratory analysis of soil concentrations,
measurements of fluid levels and dissolved and vapor concentrations in monitoring wells or
probes, and surface or subsurface geophysical methods.
Measurements of soil concentrations (e.g., total petroleum hydrocarbons or individual
species) provide the most reliable quantitativejinformation on the actual volume or mass of
hydrocarbon in the subsurface. However, since laboratory analyses of soil samples are costly and
are not practically amenable in monitoring temporal changes (that occur over time), the estimation
of spill volume from fluid level measurements in monitoring wells has a much greater
-------
practical value. Unfortunately, estimation of hydrocarbon volume from well fluid level data is less
straightforward than estimation from soil concentration data. A general lack of understanding in
this area, compounded by promulgation of methods of dubious validity and poor accuracy, have
resulted in widespread misunderstanding. |
It is well known that the hydrocarbon volume in the soil per unit area ("oil specific
volume") is significantly less than well produqt thickness (a.k.a."apparent product thickness")
(e.g., van Dem, 1967). Using a very simplifiecl theoretical approach, de Pastrovich etal (1979)
suggested that well product thickness will typically be about four times greater than the soil zone
thickness within which free product occurs (a;k.a., "soil product thickness"). Hall et al. (1984)
investigated the relationship between soil product thickness and well product thickness in the
laboratory and proposed an empirical relationjto correct for discrepancies in the method of de
Pastrovich. Laboratory investigations by Hampton and Miller (1988) found the methods of both
de Pastrovich and Hall lacked accuracy, and they questioned the relevance of estimating soil
product thickness, since it does not directly relate to oil specific volume, which is of more
fundamental interest. i
A theoretically based method for estimating oil specific volume from well product
thickness was developed and reported independently by Lenhard and Parker (1990) and Farr et
al. (1990). The method is based on the assumption of vertical equilibrium pressure distributions
near the water table, which can be inferred fro;m well fluid levels. From the fluid pressure
distributions and a general model for three ph^se capillary pressure relations, vertical oil
saturation distributions are computed and integrated to yield oil specific volume.
In addition to "free" product that is sufficiently mobile to enter a monitoring (or recovery)
well, a significant portion of the total spill volume may occur as "residual product," confined as
hydraulically isolated blobs or thin films of oiljthat are effectively prevented from moving by
capillary forces. Changes in water table elevation will generally result in increased residual
volumes over time. These fluctuations may result from natural recharge variations, drawdown or
injection as well as air-oil and oil-water fluid interface elevation changes due to plume spreading
or recovery operations. The key to maximizing product recovery from spill sites involves
minimizing the volume of residual product that is induced as a result of recovery system
operations. •
i
i
Product recovery systems are often implemented based solely on containment
considerations. That is, trenches and/or wells are located to prevent further plume migration.
While such an approach may be effective in a limited sense, it disregards an evaluation of
efficiency, as plume containment can be achieved using many different well/trench configurations
and operating conditions. Depending on the regulatory requirements, risk characteristics, and
hence the cleanup objectives, "efficiency" may have different meanings: total volume of product
recovered, ratio of product recovered per gallon of water pumped, time to reach asymptotic
recovery, capital and operating costs, etc. Once specific objectives have been defined, various
design strategies may be evaluated to obtain the desired "efficiency."
-------
Recent advances in numerical models for multiphase flow (ES&T, 1994 abc; Parker et al,
1990, 1991) along with increases in microcomputer speed and capability have made it possible to
perform sophisticated analyses to assess the effects of various design options and natural events
on spill migration and recovery system performance. Although such analyses require significant
computational effort and personnel commitment, which can limit their applicability to large or
high-risk sites, their use is essential to fully evaluate the potential complexities of hydrocarbon
assessment and remediation. !
l
| --
The purpose of this report is to outline a set of accurate yet computationally simple
protocols for spill site assessment and remedial design for hydrocarbon spills. The protocols
provide a practical means to improve the quality and to reduce the costs of site investigations and
remedial actions at hydrocarbon spill sites. The methods are particularly suited to small spills, for
which more sophisticated analyses may not be warranted, and as a preliminary modeling tool for
larger sites. ;
I
This report discusses the physical processes that control hydrocarbon retention, movement
and recovery. It describes algorithms for estimating free and residual hydrocarbon volumes from
monitoring well and soil boring data, as well as for evaluating plume migration and containment,
and product recovery volume and time as affected by well and/or trench placement and operation.
The methodologies are simple, and although somewhat laborious for hand calculations, require
minimal computational effort for readily available desktop computers. The algorithms in this
report have been implemented in the programl SpillCAD (ES&T, 1994c).
-------
Section 2
i
Basics of Hydrocarbon Behavior
I
2.1 Water, Air, and NAPL Mow I
A good place to start to develop an understanding of how hydrocarbons move in the subsurface
is Darcy's law. This well known relationship, which is the cornerstone of groundwater
hydrology, may be generalized to describe the movement of nonaqueous phase liquids
(NAPLs) and air in soil. The general form o|f Darcy's law may be written as
(2.1)
where qp is the Darcy velocity of fluid p (elg., p = air, NAPL or water), x is distance, h
is the water height equivalent pressure of phase p, Ksw is the saturated hydraulic conductivity
of the soil to water, k^ is the p -phase relative permeability, r)^ is the ratio of p -phase to
water viscosity, prp is the p -phase specific gravity, and u is a gravitational vector that is one
in the vertical direction and zero in the horizontal direction.
Darcy's law says that fluid flows in response to a pressure gradient and to gravity at a
rate inversely proportional to the fluid viscosity, and directly proportional to the relative
permeability. Relative permeability is a factor that reflects the ability of fluid to move through
the pore space when it is partially occupied by other fluids. When p -phase fluid completely
fills the pore space, the relative permeability' for the phase is one, and when no mobile p-
phase is present the relative permeability is zero.
I
Relative permeability depends on the ifraction of the pore space filled with p -phase
(i.e., p -phase "saturation"), but more importantly on the hydraulic radius of the flow channels
created by the pore geometry and by the interfaces with other fluids. Thus, narrow flow
channels exhibit a smaller relative permeability than wide channels, even if the saturation is the
same. '
i
! 4
-------
Understanding how multiple fluids interact within a porous medium and compete for
the available pore space is the key to understanding relative permeability and, more
fundamentally, to the relationship between pjhase pressures and phase saturations. This will be
considered in the next section. i
Air
Water
Water Saturation
Figure 2-1. Water retention in a collection of capillary tubes.
2.2 Capillary Retention Relations j
i
Two Phase Capillaty Pressure Relations I
In the absence of NAPL, water and air may poexist in the pore space and the saturation of
each phase will depend on the pressure difference between the fluid phases, which is referred
to as the capillary pressure. When two immiscible fluids are in contact, the pressure
difference between the phases will induce a curvature to the interface. Since pore geometry
ultimately controls interface curvature in a porous medium, as the capillary pressure changes,
the interfaces recede or expand into different; pores and the fluid saturations change. This may
be more clearly understood by consideration |of an idealized representation of a porous medium
consisting of a bundle of parallel capillary tubes (Figure 2-1). If the bundle is placed in
contact with a free water surface (i.e., plane of zero capillary pressure, a.k.a. the "water
table"), and allowed to equilibrate, water will rise in the tubes to a height at which capillary
-------
forces are balanced by gravitational forces. I The Laplace capillary equation relates interface
curvature to capillary pressure as -,
i
i
, 2o i
hc=-S- i (2.2)
where hc is the capillary pressure head (the(difference between air and water pressure heads),
a is the interfacial tension between the fluids (here, air and water), and R is the radius of
curvature of the air-water interface. At equilibrium, hc corresponds to the height of capillary
rise above the free water surface, and if the Isolid surface is readily wettable by water, the
radius of curvature corresponds to the tube radius . As we view the system from bottom to top,
we notice that as the capillary pressure increases, the fraction of the pore space occupied by
water (i.e., water saturation) decreases (Figure 2-1).
This provides a simple illustration of the concept of a "capillary fringe" above a water
table. It should be emphasized that the transition from a water-saturated condition is gradual,
due to the distribution of pores of different sizes. The example also demonstrates the
relationship between capillary pressure and phase saturation and its relationship with the pore
size distribution of the porous medium. i
i
Of course, pores in real soils are not; really straight tubes, but rather channels of
complex shapes that exhibit cross sections of varying size and shape along their length. The
capillary pressure required to remove water from a pore of variable cross section diameter will
be controlled by the smallest pore throat, while drainage will be controlled by the largest pore
diameter (Figure 2-2). This results in the dependence of water saturation in the pore space on
the history of capillary pressure changes ~ a phenomenon referred to as hysteresis. As a
result, lower water saturations occur at a giyen capillary pressure when capillary pressure is
decreasing ("imbibition) than when capillary pressure is increasing ("drainage"). Within the
limits of the primary imbibition and drainagfe curves, scanning curves define wetting and
drying paths for less extreme water content changes (Figure 2-3). Futhermore, during
imbibition displacement of the nonwetting phase (e.g., air or NAPL) is incomplete due to pore
bypassing by the wetting phase, resulting iniresidual nonwetting phase in the soil. The
maximum residual nonwetting phase will occur for the primary imbibition path (Figure 2-3).
-------
Drainage
Figure 2-2. Water distribution in pores during imbibition and drainage.
Main Wetting Curve
Main prainage Curve
Scanning Curve
WATER SATURATION
Figure 2-3. Hysteresis in capillary pressure function during drainage and imbibition.
i 7
I
-------
The shape of the curve relating water saturation to air-water capillary pressure can be
described parametrically using an empirical [formula developed by van Genuchten (1980) given
by i
(2.3)
where Sw is the water saturation corresponding to a given air-water capillary pressure haw, Sm
is an apparent "irreducible" water saturation, a and n are van Genuchten parameters for the
soil, and m = 1 - 1/n. Kool and Parker (1988) have shown that for a wide range of soils,
a for imbibition is approximately two times jthe value of a for drainage.
i
Because soil grain size distribution, in conjunction with grain packing geometry,
controls the soil pore size distribution, soils jof different types will exhibit different saturation-
capillary pressure relations. Typical capillary pressure functions for two soil types are shown
in Figure 2-4. Finer materials require larger capillary pressures before air can occupy a
significant fraction of the pore space. Finer; materials also tend to hold more water at very
high capillary pressures, when the water content tends to approach an apparently "irreducible"
value by water held tightly due to short range attractive forces between water molecules aid
solid surfaces. I
25
15
l«
8
sand
0.2 0.4 0.6 0.8
Water Saturation
Figure 2-4. Drainage air-water capillary pressure functions for different soils
-------
When a third fluid enters the picture {i.e., NAPL or "oil" for brevity), the distribution
of fluids in the pore space gets more complicated, but the same physical processes still control
the system. The first important issue to understand is the relative "wettability" of the different
fluids in the porous media. The fluid of highest wettability by definition has the greatest
affinity for the solid grains that make up the; porous medium, while the fluid of lowest
wettability will occupy pores farthest from the solid phase. Most geologic media are "water-
wet," meaning that water is the phase of highest wettability. Air is usually the phase of lowest
wettability, and NAPLs exhibit intermediate; wettability. Under such circumstances, the
distribution of fluids in the pore space will appear something like that shown in the illustration
of Figure 2-5. |
'«- ',
When three fluids (air, oil and water) jointly occupy the pore space, fluid-fluid
interfaces will occur on the pore scale between the fluids of highest and intermediate
wettability (i.e., water and oil) and betweenjthe fluids of intermediate and low wettability
(i.e., oil and air). Capillary pressures may be defined for each of these interface pairs. The
radius of curvature of oil-water interfaces will be controlled by the oil-water capillary pressure
(pressure difference between oil and water), jand air-oil interface curvatures will be controlled
by the air-oil capillary pressure (difference between air and oil pressure). The Laplace
capillary equation (eq. 2.2) applies to both interfaces, if the appropriate capillary pressure and
interfacial tension are used. i
oil
Figure 2-5. Distribution of air, oil and water in a porous medium
-------
If wettability follows the usual order; the radius of curvature of the oil-water interfaces will
be controlled by the fraction of pore space occupied by water, while the radius of curvature of air-
oil interfaces will depend on the fraction of pores filled with oil and water, since both phases are
less wettable than air. Therefore, water saturation is expected to be a function of oil-water capillary
pressure, and total liquid saturation is expected to be a function of air-oil capillary pressure. If the
pore size distribution is time invariant, the liaplace capillary equation indicates that capillary
pressure function for two and three fluid phase systems should be related by
where Sw (haw) is the water saturation versus| air-water capillary pressure function in the two
phase air-water system, S™(how) is the watet saturation function in the three fluid phase
system, St (hao) is the total liquid saturation function for the three phase system, and Pao and
POM, are scaling factors that depend on the interfacial tensions for the various fluid pairs (see
Section 2.8). Combining (2.3) and (2.4) provides a general parametric form to describe three
phase capillary pressure relations in terms of air-water capillary parameters and the two fluid-
dependent scaling factors (Parker et al., 198J7).
2.3 Vertical Equilibrium Fluid Distributions
After a NAPL that has a specific gravity less, than one reaches the water table, further vertical
movement becomes limited by buoyancy forces. If vertical hydraulic gradients are small near
the water table, vertical pressure distributions will approximate hydrostatic conditions.
Deviations from true vertical equilibrium may be expected, especially in the unsaturated zone.
However, these deviations may be accommodated by suitable definition of "quasi-static"
capillary pressure relations, as discussed in Section 2.7.
i .. • _ . :
Vertical equilibrium pressure distributions can be defined in terms of various fluid
"table" elevations. In a monitoring well screened over an interval with free oil (Figure 2-6),
one will observe an oil lens in the well that may be characterized by an air-oil table elevation,
zao (where the, air-oil capillary pressure is zero), and an oil-water table elevation, Zow (where
the oil-water capillary pressure is zero). One may also define an air-water table elevation, Zav/
(where air and water pressures are equal), which is related to the observed table elevations by
!••-'-.. = •• - - --.- •
aw ow r*ro o
10
-------
where H0=Zao-Zow is the apparent oil thickness or well hydrocarbon thickness. The vertical
equilibrium capillary pressure distributions will be given by
h =p (Z-Z ) (2-6a)
ao rroV ao-f
where Z is any elevation above the datum used to define table elevations. Stipulation of any
two of the three table elevations completely jdefines the static three phase capillary pressure
distributions. From the capillary pressure distributions and known or assumed saturation versus
capillary pressure relations, vertical fluid distributions may be computed. Water saturation will
be a function of oil-water capillary pressure^ hence of height above Z^. Total liquid saturation
will be a function of air-oil capillary pressure, hence of height above Zao. Using the three
phase van Genuchten model of Parker et al.\ (1987), water and free oil saturations may be
computed as '
- (2'7a)
Vertical equilibrium water and oil saturation; distributions for a representative case are shown in
Figure 2-6. Note that oil saturation varies continuously with elevation and is not reasonably
represented by a pancake-shaped distribution1, contrary to common misconception (see Section
3.2). Field studies have indicated that the three phase van Genuchten model provides a good
representation for both coarse and fine grained soils (Huntley and Hawk, 1992; Ostendorfet
al, 1993.)
11
-------
Soil Profile
Well
6.00
Z
0
o
i)
o
3.00
~2.00
0)
_j
-Jh.00
0.00
H,
0.00 0.20 0.40 0.60 0.80- 1.00
So tural Ion
Oil
Water
Figure 2-6. Three phase fluid distributions in equilibrium with a screened well.
Oil Specific Volume or True Product Thickness"
The volume of free oil per unit area in the soil ("oil specific volume") may be computed by
integrating the oil content over depth as '
(2.8)
i
where the limits of integration represent the upper and lower elevations where free oil occurs.
The lower limit of integration is Zow and thet upper limit may be computed as
I (2.9)
12
-------
Oil specific volume may be thought of as the "true oil thickness," in the sense that it is
the thickness of product that would occur if jail of the free oil over an area were extracted from
the soil and placed in a container of same area. Employing (2.6), (2.7) and (2.8) will yield a
relationship between Vof and Ho. Free oil specific volume versus well product thickness for
gasoline in a sandy loam and typical silt loam soil (Table 2-3) is shown in Figure 2-7.
Hysteresis in the capillary pressure relations; will be manifested as hysteresis in the VJfl^
function. Hysteresis in the curves is not shown, but some variations will occur during wetting
and drainage paths. The functions illustrate the effects of a "capillary fringe" above the oil-
water table, which results in a threshold apparent oil thickness before significant free oil occurs.
The threshold for the sandy soil occurs between 0.5 to 1.0 feet, and for the silty soil it is about
twice as great. '
4 : 6
Apparent oil thickness (ft)
8
10
Figure 2-7. Free oil specific volume versus weU oil thickness for gasoline in different soils.
i 13
-------
2.4 Residual Oil in Saturated and Unsaturated Zones
i
i •.
Saturated Zone Residual Oil \
When the oil-water table rises, oil-water capillary pressures decrease, resulting in increases in
water saturation. Due to pore scale heterogeneities, water will displace oil from some pores
faster than others, leaving islands of strandejl oil, cut off from the continuous oil phase. This
will result in "smearing" of the oil distribution below the zone of mobile phase hydrocarbon. At
a given elevation, the trapped oil saturation* Sot, may be calculated using the empirical model
of Land (1968) as a function of the current water saturation, the historical minimum water
saturation and the maximum residual oil saturation. The trapped oil specific volume may be
computed by integrating the trapped oil specific volume as
, I (2.10)
vot=
where Sot is the trapped oil saturation that occurs as hydraulically discontinuous blobs occluded
by the water phase. Since oil-water capillary; head controls water saturation, it is evident that
changes in the oil-water capillary head will govern oil entrapment, and that these changes will
govern the trapped oil specific volume. Integration of (2.10) using Land's (1968) model
indicates that trapped oil specific volume is of the form
where dVot is the change in trapped oil specific volume, dZow is the increase in the oil-water
table elevation above an elevation Z^ over k time interval, and Qot is the average incremental
trapped oil content in the soil, which may be! estimated as
eof=min ((jj^, 8o/)
i
where Sor is the maximum residual oil saturation, and 6 of is the average free oil content in the
soil, approximated by i
[ : : - - " -
i (2.Hc)
6.,.=——
14
-------
where Vof and H0 are the free oil specific vblume and apparent product thickness, respectively.
The elevation Zm^i is given by I
|
! aw (2-lld)
2. =Z +Z
"^ min °
Z =mi
H
where Z ow is the historical minimum oil-Water table elevation, and Zc is the change in the
oil-water capillary fringe thickness due to a reversal in wetting history from water drainage to
water imbibition. i
I
Unsaturated Zone Residual Oil
In addition to residual oil caused by fluid enitrapment during water imbibition, another source of
residual oil arises due to retention in the unsaturated zone. During periods of falling Zao,
downward oil redistribution eventually becomes negligible under gravitational forces, as oil
saturation reaches a critical value. We refer -to this value as the unsaturated zone residual
saturation. The increase in residual oil specific volume in the unsaturated zone due to an
incremental drop in the air-oil table, AZ00, may be described by
(2.12)
where S - Min (Sog,S0), in which Sog is the maximum unsaturated zone residual saturation
after drainage from a high oil content, and S>0 is the average free oil saturation at a given area!
location. i
i •
Water Table Fluctuations |
i
When water table fluctuations occur, the volume of residual product may increase and decrease
over time as product becomes trapped or released. If fluctuations occur within a historical band,
when the water table falls, the saturated zone residual decreases and the unsaturated zone
residual increases. Usually the volume of product released from the saturated zone exceeds the
volume tied up in the unsaturated zone, so the free oil specific volume increases. This results in
an increase in apparent oil thickness that will be observed in monitoring wells (Kemblowski and
\
1 15
-------
Chiang, 1990). When the water table rises, the converse will occur. Additional factors that tend
to magnify the effect of water table fluctuations on apparent thickness include hysteresis in
capillary pressure relations and nonequilibrium vertical pressure gradients.
2.5 Oil Relative Permeability and Transmissivity
Horizontal movement of free product at the jwater table is controlled by the oil piezometric
gradient and the oil transmissivity given by the vertically integrated form of Darcy's law as
i ,, " H
; (2.13)
i $¥0
n =-T 2.
^ ° dx
\ . _ . __..-.
i
where Q0 is the vertically integrated oil flux [L2r~1], Yo=Zao+/?ypra is the oil piezometric
head in which Zao is the air-oil table and /zai is the air pressure in units of water height, pro is
oil specific gravity, and T0 is the oil transmissivity defined by
o
•r
(2.14)
where pro is the oil specific gravity, T\ro is the oil-water viscosity ratio, kro is the oil relative
permeability, Ksw is the saturated hydraulic jconductivity, and the limits of integration are the
upper and lower elevation where free oil occurs.
Note that in the absence of a gas pressure gradient, oil flow occurs in response to a.
gradient in the air-oil table. If a gas pressur^ gradient occurs, its effect on oil flow will be
additive with the air-oil table gradient. j
i
Oil relative permeability increases with increasing water and oil contents and may be
computed from the van Genuchten (1980) model with a refinement to correct for residual oil
after Kaluarachchi and Parker. (1992) as i
! (2.15)
16
-------
Where S=(Sw+Sof-Smy(l -Sm) is effective total liquid saturation and Sw=(Sw-Smy(l -SJ is
effective water saturation. Assuming vertical iequilibrium fluid distributions, (2.14) may be
integrated using (2.15) to obtain oil transmissivity as a function of free oil specific volume or
apparent oil thickness for a given soil, and for a given hydrocarbon. Oil transmissivity is nearly
a linear function of free oil specific volume (Figure 2-8).
4 | 6
Apparent oil thickness (ft)
10
0.5 , 1 1.5
Free oil specific volume (ft)
Figure 2-8. Typical oil transmissivity functions for gasoline in sandy and silty soils.
! 17
-------
2.6 Estimation of Fluid Proterties
The following two sections summarize methods for estimating soil and fluid properties that
govern NAPL retention and movement in the subsurface listed in Table 2-1.
Product Specific Gravity \
Oil specific gravity, pro, varies significantly for different petroleum hydrocarbons depending
on their chemical composition (Table 2-2). We recommend direct measurement of specific
gravity, since such determinations are very sample and inexpensive to perform. Measurements
may be performed in the laboratory on product samples using standard methods for fluid
density determination. Since hydrocarbon density varies with temperature, measurements
should be made at a temperature close to that expected in the field.
I
A simple field procedure to determine product density in wells with free product is to
measure the water piezometric elevation (Z^) using a tube inserted through the oil layer in the
monitoring well, and to measure the air-oil and oil-water table elevations under static
conditions. We compute the product density by
I
= Zaw-zow (2.16)
Pro
i ao aw
Since (2.16) assumes that equilibrium conditions exist within the well bore, it is advisable to
wait until fluid levels are stable after inserting piezimeter tubes before taking readings.
Product Viscosity \
Dynamic (also called "intrinsic") viscosity can be measured using standard methods (e.g.,
ASTM D-88, D-4243, D-871, D-1795). For refined petroleum hydrocarbons, we have found
the following approximate correlation between specific gravity and viscosity for various
hydrocarbons '<
where r\m is the ratio of dynamic viscosity oif product to that of water.
18
-------
Typical values for various products are given in Table 2-2 at 15 ° C (API, 1989).. Viscosity
increases with decreasing temperature by about 1-2 percent per degree Celsius.
|
Table 2-1. Soil and Fluid Properties for Three Phase Flow
Fluid properties:
r •'< " '
pro Ratio of oil to water density [L °] I
r\ro Ratio of oil to water viscosity [L°] \
Pao Ratio of water surface tension to oil surface tension [X°]
Pow Ratio of water surface tension to oil-water interfacial tension [L °]
Soil properties: !
i *
i ' ,
i
K^ Saturated hydraulic conductivity [LT'r1]
4> Total porosity [i°] j
Sm Water saturation at field capacity [L 6]
i
Sog Maximum unsaturated zone residual oil saturation [L ° ]
Sor Maximum saturated zone residual oil [saturation JX°]
a VG pore size parameter [L ~l ] j
n VG pore size distriution exponent [Z°]
: 19
-------
Table 2-2. Typical Specific Gravity (Rpro), Oil-Water Viscosity Ratio (Hr\ro), and
Capillary Scaling Factors ($Bao and 5pow) j for Various Hydrocarbon Mixtures
Product Pra r\m |3ao Pw
I
Gasoline 0.73 0.45 3.5 1.4
Diesel fuel 0.83 2.7 2.2 1.9
Fuel oil #2 0.87 5.3 1.9 2.1
Fuel oil #5 0.92 215 1.6 2.8
Fluid Scaling Factors
Air-oil and oil-water scaling factors (Pao and pow) are necessary to describe three phase
saturation-capillary pressure relations. One may estimate the scaling factors from oil surface
tension and oil-water interfacial tension date (Lenhard and Parker, 1987) as
I (2.18a)
! <2.18b)
B =a' /a
~ow w ow
I
where aw is the surface tension of water (ca.j 68 dynes/cm), a0 is the surface tension of the
organic liquid, and aow is the oil-water interfacial tension. An alternative protocol for
determining Pow is to measure the surface tension of water saturated with dissolved
hydrocarbon (in other words, water which has been shaken with hydrocarbon and decanted to
remove all traces of free liquid) and to estimate the interfacial tension via
20
-------
(2.19)
where a'wis the surface tension of water saturated with dissolved hydrocarbon. In the absence
of measurements of either aow or a'w, we may obtain an estimated ofpovt;, assuming aw«ow,
which indicates that ; -
l-l/Pao (2.20)
Surface and interfacial tensions may be determined in the laboratory using standard
methods (e.g., ASTM D-971). For unrefined petroleum hydrocarbons (that is, crude oil),
scaling factors may be estimated using a correlation between oil surface tension and specific
gravity given by Baker and Swerdloff (1956) as
B =-lj (2.21a)
* GO t i i /*».
which provides a simple procedure for estimating scaling factors for unrefined hydrocarbons in
the absence of additional information. Lymdn et al. (1982) reviewed procedures for estimation
of surface tension and interfacial tensions of'fluid mixtures.
2.7 Estimation of Soil Properties i
Soil properties required to describe oil and vsfater retention and movement include parameters
defining the fluid retention properties and soil permeability. If properties exhibit variations in
the vertical direction, parameters relevant to jthe capillary fringe zone (where more oil occurs)
should be used to predict oil recovery with maximum accuracy. If this results in an under- (or
over-) estimate of the water transmissivity, one may correct it by adjusting the effective and
21
-------
f
actual aquifer lower boundarier deeper (or shallower) in proportion to the error in the aquifer
conductivity. I
i
Pump tests or slug tests are the perferred method to obtain saturated hydraulic
conductivity, although laboratory tests may t>e sufficiently accurate if sample disturbance is
minimal and enough samples are obtained to compute a representative average. When
averaging multiple determinations of hydraulic conductivity or other soil parameters, we
recommend employing a geometric average jas
| (2.22)
i
^ i ISlnXi
G= exp 1 '-I
I ' N
where X. are the measurements (i = 1,...,N) and Gis the geometric average.
i
Total Porosity, Effective Porosity and Field Capacity
Total porosity, 4> may be determined directly from soil cores, or indirectly from neutron
logging or other in situ methods. The parameter Sm represents the minimum water saturation
that will occur in the soil under field conditibns. Note that the minimum saturation determined
by fitting to laboratory moisture retention data is usually smaller than the minimum field water
content, because equilibrium conditions do not occur in the field. We may estimate S from
; J tn
direct measurements of the degree of saturation on soil cores taken from the field at elevations
above the "capillary fringe" where water saturation drops more or less sharply.
If the specific yield of the unconfined aquifer is known, this may be used to estimate Sm
as j m
1 (2.23)
where (J> is the total porosity of the soil, and! e is the specific yield or "effective porosity."
Measured specific yields often increase withjthe duration of pump tests due to "delayed yield"
effects. Since long term drainage is of concern here (mostly weeks to months), specific yields
from short term pump tests may lead to overfestimation of S .
22
-------
I '
If laboratory moisture retention data are available, we may make an estimate of Sm by
evaluating the water saturation at an air-water capillary pressure head of ca. 100 to 300 cm,
which is commonly regarded as an approximation of "field capacity."
I
Residual Oil Saturations \
Maximum residual oil saturations in the unsaturated and saturated zones are needed tro estimate
recoverable product. One may determine the maximum unsaturated zone residual oil
saturation, Sog, in the laboratory by measuring the oil saturation in a soil core taken at a
location where oil has been able to drain from a previous oil imbibition event for at least
several days. Note that water and oil will not drain from a short column in the laboratory as it
does in the field becuase the capillary pressure at the column boundary is zero. Typical values
of Sog for field soils are in the range given by
! (2.24)
where fog may range from 0.2 to 0.5 with ajmedian of around 0.3. Fluids with higher viscosities
and soils that are more heterogeneous with tend to have larger f values. Theoretical analyses
indicate that residual saturation increases approximately proportional to the fourth root of product
viscosity (i.e., fog^(\ro>, where r\ro is the oil-water viscosity ratio).
One may determine the saturated zone residual oil saturation, Sor, by measuring the final oil
saturation in an initially water saturated soil j:ore subjected to oil flooding followed by water
flooding. Typical values of Sor are given by
j
S^fjd-SJ (2-25)
where for ranges from 0.2 to 0.5 with a med'ian of about 0.3. Fluids with higher viscosities, and
soils that are more heterogeneous, tend to have larger for values.
I
Capillary Pressure Parameters from Soil Cores
Air-water capillary pressure curves are often!characterized by fitting model parameters (i.e., a, n,
and Sm) to water content versus capillary pressure data obtained in the laboratory on soil cores
(which yield true equilibrium parameters). However, in the field, equilibrium is never truly
attained since low relative permeabilities impede fluid drainage as wetting phase saturations
|
i 23
-------
I-
diminish. To correct for the deviation from lequilibrium conditions, we use quasi-static model
parameters, which yield the correct water saturation distribution under field conditions when
assuming a hydrostatic water pressure distribution. Lenhard and Parker (1990) describe
procedures to estimate quasi-static retention parameters from laboratory data. The simplest
approach is to fix Sm at a value corresponding to the minimum field saturation, discard moisture
data below about Sw*l.l x Sm, then fit the parameters a and n to the reduced data set using a
nonlinear regression method. Table 2-3 gives typical quasi-static van Genuchten (VG) model '••
parameters for various soil types computed from equilibrium values reported by Carsel and. Parish
(1988). • I
Table 2-3. Representative Soil Properties ifor Various Soils
Soil type*
Sand
Loamy sand
Sandy loam
Sandy clay loam
Loam
Silty loam
Clay loam
Sandy clay
Silty clay loam
Silt clay
(ft/d)
23.1
11.5
3.48
1.02
0.82
0.36
0.20
0.095
0.056
0.016
a n
(I/ft)
C1 O ft
AJM O tj
m or og
4.5 2.7 0.13
3.8 2.4 0.21
2.3 2.0 0.24
1.8 1.!
0.28
1.1 1.7. 0.35
0.67 1.7 0.43
0.64 1.7, 0.55
0.97 1.8 0.66
0.37 1.9 0.68
0.26 2.8 0.84
0.26
0.24
.0.23
0.22
0.19
0.17
0.13
0.10
0.10
0.05
0.03
0.05
0.05
0.06
0.07
0.07
0.07
0.07
0.06
0.04
Size classes in the USD A! classification system.
Capillary Pressure Parameters from Grain Size Data
Arya and Paris (1981) derived a theoretical procedure to estimate air-water capillary pressure
parameters (that is, a, n, Sm) based on the proposition that capillary pressure relations relate to the
pore size distribution of the soil, which we may in turn infer from the grain size distribution.
Mishra et al (1988) calibrated and implemented the method in the program SOILPROP (ES&T,
1990), which is also incorporated into the program SpillCAD (ES&T, 1994).
1 :
! 24
-------
Capillary Pressure Parameters from Saturated Conductivity
Another method to estimate the VG parameter a is to employ a correlation with saturated hydraulic
conductivity as j
i
i
'*" (2.26)
Based on laboratory analyses of vertical conductivity .4 * 1 . 5dV:ft ~3/2(±50%) . Since field-
measured horizontal conductivities (for example, from slug or pump tests) are generally much
higher than vertical laboratory values, estimates of a using the foregoing value of A and field
measured conductivities may be higher than yalues estimated from grain size distribution data. The
true field parameter values probably lie betwieen these estimates.
!
Capillary Pressure Parameters from Soil TPHData
The most critical parameter in estimating oil; saturation distributions and spill volume is generally
the capillary curve parameter a . If independent data are available on oil saturation at points in the
field, under certain conditions it may be possible to use these to calibrate the value of a. Since oil
saturation and soil total petroleum hydrocarbon (TPH) are related, methods may be developed to
calibrate a from TPH and monitoring well data. The method is based on the premise that oil
saturation distributions computed for given well fluid levels, from the three phase saturation
relations discussed in Section 2.3, should agree with TPH data, if fluid levels at the time soil
samples are taken are known and if one properly calibrates the capillary model. Required data to
employ this strategy include well product thickness (H0) and oil-water table elevations (Z0J in
monitoring wells at specified coordinates (x jj'J , TPH measurements from a given depth interval
for specified coordinates (x^y^ , as well as estimates of total porosity (cj>) , irreducible water
saturation (Sm) , the van Genuchten parameter (n), oil specific gravity (pro) , and fluid scaling
factors (Pao and
The major steps of the algorithm are as follows: (i) interpolate Zow and H0 at locations
, where TPH is measured onto a regular computational grid, (ii) at each soil bore location,
calculate an average oil saturation S0 from the interpolated fluid levels over the interval of TPH
measurements, and use this S0 to calculate a Corresponding TPH value, (iii) compare measured and
calculated TPH values and iteratively adjust the value of a to minimize the sums of squared
deviations. \ . '
25
-------
Free oil saturation Sof at a given location (Xp yj) and elevation Z is computed as described in
Section 2.3, and So is determined by averaging S0 computed at midpoints, lower and upper limits for
TPH measurement intervals. TPH (in mg/kg);is calculated from S0as
(2.27)
p~S 6
° ° x 10
' ' '
where p0 is the oil density and pb is the soil bulk density.
i • •
i •
The applicability of this method is critically dependent on accurately defining the fluid levels at
the time TPH samples are taken. The method will be most reliable if well product thickness is large
(several feet), because uncertainty in fluid levels will have less effect on TPH predictions. The
difficulty of making accurate measurements of TPH in the zone of free product should be carefully
considered in employing this method.
26
-------
Section 3
i
Spill Assessment Methods
3.1 Interpretation of Soil Concentration Data
If measurements of soil total petroleum hydrocarbon (TPH) are available at a sufficient number of
points in space to accurately define a three-dimensional distribution of NAPL in the subsurface,
hydrocarbon volume estimation is straightforward. We simply interpolate the hydrocarbon content
in space and perform the volume integration. However, three-dimensional interpolation is
computationally intensive and requires a high (density of sampling points to justify its use. To
minimize these difficulties, a two-step integration procedure may be used, which is more efficient
and robust with sparse data sets. The procedure involves (I) vertical integration of oil content
using linear interpolation between measurement depths, followed by (ii) areal interpolation and
integration using a two-dimensional kriging algorithm.
Determination of soil TPH involves extraction of soil samples and analytical quantification
of hydrocarbons in terms of mass of hydrocarbon per mass of dry soil. TPH extractions yield
hydrocarbon mass present as a nonaqueous phase liquid, as dissolved components in the aqueous
phase, as adsorbed components, and as components in the gas phase (to the extent they are not
lost during sample processing). Calculations based on equilibrium partitioning indicate that when a
nonaqueous phase exists in a sample, the majority of the hydrocarbon mass is in the separate
phase, unless the natural organic content of tiie soil is high. Therefore, estimating nonaqueous
phase volume directly from TPH incurs a very small error. In practice, interpolation and
measurement errors are much greater. |
The volumetric oil content, Q0, which is the volume of oil per volume of soil, is related to
TPH by i
! (3.1)
_.77W
*o=-
where p6 is the soil bulk density [ML ~3], pj is the oil density [ML ~3], and TPH is expressed in
mg/kg. SpillCAD estimates bulk density as p^(l-({>), where is soil porosity, and p^ is particle
27
-------
I
density, which is generally between 2.6 and 3 JO g cm ~3 (the density of quartz which is 2.65 g
cm "3 is often assumed). :
For a given bore hole, in which TPH measurements are available at various depths, we
may compute the oil volume per unit horizontal area, or oil specific volume, V0, by integrating
the volumetric oil content over the vertical dimension as
(3.2)
where zl and zu are the lower and upper elevations where oil occurs (or the range over which we
are interested in estimating product volume). The integration may be readily carried out
numerically by linearly interpolating oil contents between measurement points along the vertical
dimension. i
i
i
Oil specific volume computed from (3-2) for each soil boring may then be interpolated
over a specified areal domain to obtain values: of oil specific volume at Allocations on a regular
grid (for example, 20 x 20 nodes), with block jdimensions AxxAy using a kriging algorithm,
yielding lvalues of specific volume, Voi, where I = 1,2,.. .,N. The total hydrocarbon volume, £0,
within the horizontal limits of the areal domaiti and within the vertical limits of the TPH sampling
elevations is computed as i
(3-3)
where ^4=AxAy is the grid block area. If TPH! data are restricted to the unsaturated zone, then the
estimated volume represents only the product in the unsaturated zone. Likewise, if the
computational domain is limited to a subregion or TPH data are used for a limited depth range,
the computed volume represents an estimate of the volume within a subregion.
Note that the hydrocarbon volume estimated from soil TPH data represents both free (i.e.,
mobile) and residual (i.e., immobile) hydrocarbon to the extent that both occur in soil samples.
Samples taken in the unsaturated zone may represent primarily residual oil, whereas samples at
the water table may be mostly free product-- although sampling within the zone of free product is
not very reliable. i
28
-------
Estimating the Mass of a Component j
If soil concentrations of specific components |(for example, BTEX) are available, we may use the
same procedure described above for total hydrocarbon to estimate the total mass of a measured
component in the soil. If the soil concentration of species a, expressed in mg/kg, is 7a, then we
may compute the mass of a per area in the soil from the vertical distribution of 7a in a bore hole
as i
J (3-4)
m =
where zt and zu are the lower and upper elevations where the contamination occurs (or the range
over which we want to estimate species mass). Integration limits are the maximum and minimum
elevations at which soil concentration measurements are available for a given bore hole.
Integration of (3.4) may be performed numerically using piecewise linear interpolation between
measurement elevations. |
i
Values of the contaminant mass per afea, ma, computed at soil boring locations from
(3.4) may be interpolated over a specified areal domain to obtain values at locations on a regular
grid (for example, 20 x 20 nodes), using a killing algorithm. We may then compute the total mass
of species within the indicated domain as I
j (3-5)
N
where ^4=AjcAy is the grid block area, Nis the number of nodes, and ma is the mass of
contaminant per area at node /. j
i
Estimating the Volume of Contaminated Soil
When soil excavation or ex situ treatment of contaminated soil is being considered, the volume of
contaminated soil must be estimated. The first problem then is to define operationally what is
meant by "contaminated." This is typically either mandated by regulations or subject to
negotiation with regulatory agencies based ori risk assessment studies. We assume that the
definition of "contaminated" is that the soil concentration for TPH, or for a given species (for
example, benzene or total BTEX), exceeds a Specified threshold value and that soil is considered
to be "contaminated" if the soil concentration of interest, 7, is greater than Ycrit.
Using values of 7 that are linearly inte;rpolated between sampled elevations in the vertical
direction as described above, we may define an indicator function as
29
-------
(3.6)
The volume of contaminated soil per area at soil boring locations is computed as
I (3-7)
where z, and zu are the lower and upper elevations we specify. The L values are computed over a
specified areal domain to obtain values at N nodes on a regular grid (for example, 20 x 20 nodes)
using a kriging algorithm. We may then commute the volume of soil with concentration exceeding
the threshold by \
' • (3-8)
; t=i
I .... •. . •. ; .•
where A=kxky is the grid block area, Nis the number of nodes, and Lf is the nodal value of L.
i ;
Example Calculations '
To illustrate the calculations discussed above, soil concentration data for samples taken from a
soil boring to a depth of 45 feet are used to compute total hydrocarbon specific volume, benzene
mass per area, and volume of soil per area with TPH greater than 1000 mg/kg in a spreadsheet
format given in Table 3-1. Given multiple soil borings, similar calculations may be repeated for
other locations and the results interpolated over a spatial domain to determine total hydrocarbon
volume, benzene mass and contaminated soil volume within the sampled region using (3.8).
InteqDolation may be carried out using commercial or public domain software (e.g., GEO-EAS or
on a regular grid Surfer). The soil under consideration is assumed to have a porosity of 0.35 and a
bulk density of 1.72 g/cm3 . The hydrocarbon density is assumed to be 0.80 g/cm3 . Values in
Table 3-1 were computed as follows: j
!
Column A. Sample depth is the distance from the ground surface to the center of the
core sample. '
Column B. The sample interval, dZ, is half the distance from the current sample to the
next shallower sample, plus half the distance from the current sample to the next
deeper sample, except for the shallowest and deepest samples, in which case, the
interval is half the distance from the sample center to the next sample, plus 0.5 ft
to account for one half of the actual length of the sample core.
| .-.--.. .
i 30
-------
Column C. Measured TPH in the soil core given as mg hydrocarbon per kg dry soil.
Column D. The average volumetric oil content in the sample interval computed from
(3-1). !
i - . '• •
Column E. Each entry in the column is calculated as Q0dZ, and the entire column is
summed to obtain the oil specific volume, V0=0.206 ft3 per ft2.
Column F. Measured soil benzene concentration expressed as mg benzene per kg dry
soil. :
i
Column G. The average benzene mass; per area in g/ft 2 in the sample interval is
computed from eq. (3.4), and the factoir/= 0.0283 is inserted to make the proper unit
conversions. The column is summed to1 obtain the mass of benzene per area over
the boring depth, mienz=39.8 g/ft2. |
i •• '
Column H. An indicator variable that is 1 if TPH<; 1000 mg/kg and 0 if TPH is smaller.
Column I. The volume of contaminated soil per area in each sample interval is calculated
as 6(Z)dZ. The column is summed to obtain the contaminated soil volume of 22.5 ft3
per ft2, '!
i
Table 3-1. Example Spreadsheet Calculations From Soil Boring Data
A
Depth
(ft)
5
15
21
28
37
45
B
dZ
(ft)
5.5
8.0
6.5
8.0
8.5
4.5
C
TPH
(mg/kg)
542
1180
3937
6836
678
27
D
90
0
0.0012
0.0025
0.0085
0.0147
0.0015
0.0001
E
\vo
(ft3/ft)
0.006
0.020
,0.055
0.118
'0.012
0.000
0.206
F
r.
(mg/kg)
5
9
37
53
6
0
G
fpj«
(g/ft2)
1.3
3.5
11.7
20.7
2.5
1.2
39.8
H
8(2)
(-)
0
1
1
1
0
0
I
8(Z)dZ
(ft3/ft2)
0.0
8.0
6.5
8.0
0.0
0.000
22.5
: 31
-------
3.2 Free Oil Volume From Monitoring Well Data
i
Description of Method \
As discussed in Section 2.3, free oil specific volume may be computed from well product
thickness data, if vertical equilibrium pressure distributions are assumed and soil capillary pressure
relations can be estimated. Areal integration of oil specific volume values will then provide an
estimate of free oil volume over the area! domain.
I
The procedure that is recommended for performing these calculations begins with a set of
well product thickness data for a given point in time from a monitoring well network. Well
product thickness values (H0) are interpolated aerially using a 2-D kriging algorithm to obtain
estimates of H0 on a regular grid overlaying the site area. After interpolating well product
thickness, the free oil specific volume, Vof, niay be computed as described in Section 2.3 for each
node. The total free product volume is then determined by summing over the area as
where ^4=AxAy is the grid block area, -N is th'e number of nodes in the grid, and Vof is the free oil
specific volume at node /. \ °'
t
r
In addition to wells within the liquid Hydrocarbon plume, enough wells should be available
beyond the plume to define its perimeter. If sufficient wells do not exist to adequately delineate
the plume, it may be necessary to define "control points" with zero oil thickness at locations
dictated by professional judgment, rather than to rely on an interpolating algorithm to extrapolate
the data. j
An important consideration in computing product volume from well product thickness
data is the validity of the assumption of equilibrium between the well and the surrounding soil as
well as the assumption of vertical equilibrium'in the soil. Under conditions in which an upward
hydraulic gradient exists, oil specific volume computed assuming vertical equilibrium conditions
may underestimate the actual free oil specific jvolume. With a downward gradient, oil specific
volume may be overestimated. j
If a well was recently installed or bailed, fluid levels will take some time to equilibrate
between the soil and the well. How long this will actually take depends on the oil transmissivity,
which will in turn depend on the soil and fluid properties as discussed in Section 2.5. The
equilibration time may be determined in the field by monitoring well recovery versus time after
bailing a well. A rough estimate of the equilibration time in days may be taken as T^ where T0
is the oil transmissivity in square feet per day computed per Section 2.5 as a function of soil and
fluid properties and equilibrium well product thickness. For commonly encountered conditions,
the well equilibration time may range from minutes to weeks. When wells have been bailed and
32
-------
deviations from vertical equilibrium conditions are suspected, it may be better to employ the
maximum product thickness at each well to compute spill volume rather than a time-synchronous
data set. j
i
Since well product thickness only reflects free oil under the existing hydraulic conditions,
the computed spill volume is only an estimate; of the free oil volume corresponding to the
specified product level data. Historical changes in air-oil and oil-water table elevations can induce
some oil to become hydraulically discontinuous and hence non-detectable by monitoring wells.
This may cause computed free oil volumes tojfluctuate over time with the occlusion or release of
residual oil.
Evaluation of the "Oil Pancake" Approximation
A common conceptualization of the vertical distribution of free product at the water table is based
on the idea that oil occurs as a distinct lens in which the drainable pore space is saturated with oil.
This is often referred to as the "oil pancake." According to the general theory of capillary
retention discussed in Section 2, such a step function fluid distribution in the soil will be
approached only if the soil pores are very large so capillarity is negligible (i.e., high value of van
Genuchten a) or if the pore size distribution is very narrow (/'. e., high value of van Genuchten »).
The oil pancake model is often employed to interpret baildown tests (GruszcensJd, 1987). In a
baildown test, accumulated product in a monitoring well is bailed out over a short period of time
and the response of the air-oil and oil-water interfaces in the well are recorded over time (Figure
3-1). The rate of oil flow into the well is controlled by the oil transmissivity and the air-oil head
difference between the soil and the well. Water flow is controlled by the water transmissivity and
the air-water (i.e., "corrected water table") head difference. During the initial stage of a baildown
test, water and oil flow into the well in response to gradients in the air-water and air-oil tables,
respectively. Since water transmissivity is usually much greater than oil transmissivity, water
reaches an equilibrium condition earlier than the oil. As further oil flows into the well, the
air-water table gradient is reversed and water is displaced from the well. According to the oil
pancake model, the product thickness in the well when water flow reverses (oil-water table
reaches a maximum) corresponds to the "formation product thickness" which is multiplied by the
effective porosity to determine oil specific volume, often referred to as "true product thickness" in
this context. i
j
i -
An analysis of baildown test results with the numerical model ARMOS was reported by
Zhu et al. (1993). Well hydrographs for baildown tests with 2 feet of diesel fuel initially
present in a monitoring well are shown in Figure 3-2 for two soils. For the sandy aquifer, the
oil-water table peaks within one minute, whileiwater flow into the well is still occurring after 30
minutes for the silty soil. A comparison of the toil saturation distribution inferred by the oil
pancake model and the profile computed from! the van Genuchten capillary model is shown in
Figure 3-3 for the two tests. It is evident that the pancake model is not a very good approximation
of the oil saturation distribution. Inspection of the area under the oil pancake and van Genuchten
curves, which relate directly to the oil specific Volume, indicates that the oil pancake model yields
an estimate of oil specific volume that is withiii 10 percent of the actual volume for the sandy soil.
I 33
-------
For the silty soil, the oil pancake model overestimates oil specific volume by a factor of 2.5 to 4.0,
depending on the coarseness of the filter pack material. Oil drainage from the filter pack after
bailing is misinterpreted as recovery from the;formation by the oil pancake model.
i
The results indicate that the oil pancake model provides a poor approximation of the free
oil distribution and leads to erroneous estimates of free oil volume. Overestimation of free oil
specific volume will be increasingly severe for finer grained materials. Quantitative determination
of oil transmissivity and oil specific volume from baildown tests requires consideration of transient
two phase oil-water flow using an appropriate numerical model.
-z
Figure 3-1. Schematic of baildown test over time.
34
-------
92.S
92.0
90.0
89.5
air-oil
oil-water
10
|15 20
i Time (min)
25
30
35
92.0 r
89.5
15 20
, Tune (min)
Figure 3-2. Baildown test hydrograph for a) sandy soil and b) silty soil with diesel.
. 35
-------
2.5
"Pancake1
/model
1.0
O.S
VG;
model
0.2
0.4 0.6
Effective oil saturation
0.8
"Pancake" model
VG model
0.2
0.4 0.6
Effective oil saturation
0.8
Figure 3-3. Oil saturation distribution for yan Genuchten model and "oil pancake" model
for a) sandy soil with diesel and b) silty soil with diesel.
Example Calculations
An example calculation is shown in spreadsheet form to compute free oil specific volume, Vof,
from well product thickness, H0, in a monitoring well. The well product thickness for the
example is 3 feet and the air-oil and oil-water table elevations are 93 and 90 feet, respectively.
Assumed soil and fluid properties for the problem are given in Table 3-2. Vertical integration is
performed from a lower elevation of Zow=9Q feet to an upper elevation of ZM=93:4 feet,
computed from (2.9). Calculations are performed at 35 equal depth intervals of 0.1 feet to
numerically integrate for oil specific volume. To compute total free oil volume, calculations of oil
specific volume must be repeated at various areal locations. Since well product thickness must be
36
-------
a smooth function in space (discontinuities in the piezometric gradients cannot occur), while oil
specific volume may be discontinuous due to soil heterogeneity, it is preferable to interpolate H
from monitoring wells onto a computational grid and to compute Vof on the grid from
interpolated apparent thickness values. Values! in Table 6 were computed as follows:
Column A. The first value of Z is 2U from (2.9), the final value is Zm , and intermediate
values are incremented in intervals of dZ=Q. I ft.
Column B. The oil-water capillary pressure h^, is calculated as .
Column C. Air-oil capillary pressure is calculated using eq. (2.6a) above Zao. At lower
elevations hao=0 is employed to compute total liquid saturation^
Column D. Water saturation is calculated using eq. (2.7a) given how at the specific
depth. j
]
Column E. Free oil saturation is calculated for each depth as S f=S -S
s Oj t W
Column F. The free oil volume per unit area for each depth interval is computed as
§SdZ. A sum of all values in column F gives the free oil specific volume of
0.309 ft3/ft2. !
Table 3-2. Soil and Fluid Properties for Example Problem.
P«r°-* ^S.Oftday-1 =0.15
= 0.35 ^=0.06
or
Pao=3-2 a=2.5 ft-1 i £..=0.20
Pow=l-5 »=2.0 !
'37
-------
Table 3-3. Spreadsheet for Free Oil Specific Volume from Well Product Thickness
A
Z
(ft)
93.4
93.3
93.2
93.1
93.0
92.9
92.8
92.7
92.6
92.5
92.4
92.3
92.2
92.1
92.0
91.9
91.8
91.7
91.6
91.5
91.4
91.3
91.2
91.1
91.0
90.9
90.8
90.7
90.6
90.5
90.4
90.3
90.2
90.1
90.0
B
h
(ft)
0.68
0.66
0.64
0.62
0.60
0.58
0.56
0.54
0.52
0.50
0.48
0.46
0.44
0.42
0.40
0.38
0.36
0.34
0.32
0.30
0.28
0.26
0.24
0.22
0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
C i
hao i
(ft) i
0.32 \
0.24 i
0.16
0.08 !
0.00 1
0.00 |
0.00 |
0.00 |
0.00 i
0.00 '
0.00 j
0.00 i
0.00
0.00 i
0.00 |
0.00 j
o.oo !
0.00 i
0.00
o.oo 1
0.00 |
0.00 |
0.00 1
0.00 !
o.oo i
o.oo !
o.oo ;
o.oo I
0.00 i
0.00 j
o.oo i
0.00 !
0.00 \
0.00 ,
o.oo ;
i
i
1
D
s*
(-)
0.46
0.47
0.48
0.49
0.50
0.51
0.52
0.53
0.54
0.55
0.56
0.58
0.59
0.61
0.62
0.64
0.66
0.67
0.69
0.72
0.74
0.76
0.78
0.81
0.83
0.85
0.88
0.90
0.93
0.95
0.96
0.98
0.99
1.00
1.00
E
sof
(-)
0.00
0.08
0.20
0.38
0.50
0.49
0.48
0.47
0.46
0.45
0.44
0.42
0.41
0.39
0.38
0.36
0.34
0.33
0.31
0.28
0.26
0.24
0.22
0.19
0.17
0.15
0.12
0.10
0.07
0.05
0.04
0.02
0.01
0.00
0.00
F
$SdZ
(ft3/ft2)
0.000
0.003
0.007
0.013
0.018
0.017
0.017
0.017
0.016
0.016
0.015
0.015
0.014
0.014
0.013
0.013
0.012
0.011
0.011
0.010
0.009
0.008
0.008
0.007
0.006
0.005
0.004
0.003
0.003
0.002
0.001
0.001
0.000
0.000
0.000
0.309
38
-------
3.3 Estimation of Dissolved and Free Phase Travel
Description of Method \
In order to assess the potential for adverse impacts of a hydrocarbon spill, it is often useful to
estimate the migration rates of dissolved and free phase plumes. These can be used in turn to
estimate travel times to property boundaries qr potential receptor locations (e.g., wells, streams,
etc.). The average velocity of a dissolved phase plume, vw, is given by
v -
dx
where qw is the Darcy velocity for water, 4> is porosity, R is a retardation factor accounting for
adsorption, K^ is the saturated hydraulic conductivity, and dZaJdx is the corrected water table
gradient. Use of total porosity in (3.10) assumes that all pore space is accessible to dissolved
contaminant. A retardation factor of 1.0, which corresponds to no adsorption, represents the
worst case with regard to dissolved plume migration ~ i.e., maximum migration rate.
... I.-....-.
The average velocity of the free phase i hydrocarbon plume, v , is given by
1 (3.11)
v
0
dx
where q0 is the Darcy velocity for oil, (f> is porosity, S0 is the oil saturation, T0 is the oil
transmissivity, Vof is the free oil specific volume, and dZJdx. is the air-oil table gradient. The
relationship between the mobility factor, M0=TJVof, and apparent oil thickness, H0, may be
determined from the functions discussed in Septions 2.3 and 2.5. The results indicate that the
mobility factor increases with H0 up to a maximum value given by
| (3.12)
'
0
where pro is the oil specific gravity, f\ro is the; oil-water viscosity ratio, Kswis the soil hydraulic
conductivity, is total porosity and Sm is the residual water saturation.
Variations in the mobitity factor with H0 are illustrated in Figure 3-4, which shows the
relative mobility factor, M0/Ad , for a sandy and a silty soil. For the sandy soil, the mobility
reaches a maximum between 5 to 10 feet. Forjthe silty soil, a maximum value is reached at an
apparent oil thickness of about 15 feet. IfM0/A/"ax is used to estimate oil plume velocity, the
results will tend to overestimate the movement on the perimeter of the plume where the mobility
factor diminishes. Figure 2-7 indicates that thfe minimum apparent thickness for the same sandy
! 39
-------
and silty soils due to capillary exclusion are approximately 1 and 2 feet, respectively. Inspection of
Figure 3-4 indicates that the oil relative mobility for both soils is approximately 0.5 at their
respective minimum apparent thickness values. Therefore, the oil mobility factor on the plume
perimeter should be approximately half of the maximum value given by (3.12).
1 .0-1
0.0
0.0
5.0 : 10.0
Ho (ft)
15.0
Figure 3-4. Relative oil mobility versus apparent oil thickness for two soils.
The time for a dissolved or free phase'plume to travel a distance along a streamline may be
computed as j
where v is the velocity of phase/? (i.e., oil or water). The distance must be measured along a
path perpendicular to the potential contours, which correspond to the air-water table for water
flow or the air-oil table for oil flow in the absence of air pressure gradients. If the gradients
change significantly along a streamline, the travel time calculations are performed along intervals
i 40
-------
for which the gradient can be approximated as constant, and travel times for the increments are
then added. |
I -,...- ;
Example Calculations I
The foregoing algorithm for estimating free phase plume migration rate was evaluated by
comparing analytical results with those of the numerical model ARMOS for transient water and
NAPL flow (ES&T, 1994). The problem involves a 54,000 gallon gasoline leak in an aquifer with
a 0.5 percent regional water table gradient. Relevant soil and fluid properties are shown in Table
3-2. I ., , , ;
The distribution of the apparent oil thickness 100 days after the gasoline leak is repaired
and 700 days after the leak is repaired, as predicted by the numerical model, is shown in Figure
3-5. Over the 600 day period, the outer edge of the free phase plume (taken operationally as the
0.25 ft apparent thickness contour) has migrated about 50 feet, indicating an average oil plume
velocity v of 0.083 feet per day.
To estimate the free phase plume velocity
mobility factor for the gasoline plume as
o
2TJ
The average oil piezometric gradient (air-oil
gradient (air-water table), which is 0.5 percent,
estimated as
table) will be roughly equal to the water piezometric
:. The maximum free phase plume velocity may be
dx
=0.09 ft/day
ity with the analytical model, we first compute the
(3.14)
-=17.9 ft/day
(3.15)
The estimated oil velocity of 0.09 feet is approximately equal to the value of 0.083 feet per day
indicated by the numerical model.
41
-------
600
560
520
480
-J 400
^ 360
0> 320
" 280
«200 -
'—' 160 -
120 -
80 -
40 F
0
0 40 81 121 161 201 242282 322 363 403 443 483 524 564 604 644 6B5 725
I I I I I I I I I I I I I I
) I I I I I I I I I I I I I I I I I I
Ho distribution at 100 days
600
560
520
480
440
400
360
320
280
240
200
160
120
80
40
0 40 81 121161201242282322363403443483524564604644685725
. Distance C ft)
0
600
560
520
480
440
-> 400
q-
^ 360
(D 320
c 280
O
0 40 81 121 161 201 242282322363403443483524564604644685725
CO
Q
> 240
200
160
120
80
40 -
0
I I I I I I I I I I I I I I j I I
I I I I I I I I I I I I I I I I I I
Ho distribution at 700 days ;
i i
......... i i i i i i i'
600
560
520
480
440
400
360
320
280
240
200
160
120
80
40
0 40 81 121 161 201 242 282 J322 363 403 443 483 524 564 604 644 685 725
Dls tiance C ft)
Figure 3-5. Apparent thickness 100 and 700 j days after a gasoline leak predicted by AKMOS.
42
-------
Section 4
I
Product Recovery System Design
4.1 Specification of Design Criteria j
Successful remedial design for subsurface hydrocarbon spills inevitably requires compromises
among a variety of technical, economical and socio-political issues. Among the factors that may
be taken into consideration are: [
• Control environmental and human health risks to acceptable levels,
• Meet regulatory requirements for closure,
• Minimize disruption to normal activities caused by construction/remedial measures,
• Mitigate negative publicity and potential lawsuits associated with spill, and
• Minimize total cost for site investigation and remediation.
Ideally, the first two points are consistent, but this is often not the case. In some cases,
exposure risk may be low enough that no action may be warranted or free product removal in
conjunction with subsequent monitoring may be an acceptable endpoint. In such cases design
optimization is fairly straightforward. The objective is simply to reach asymptotic free product
recovery with minimal cost, subject to constraints imposed by needs to minimize disruption, or
assuage socio-political concerns. For a given design option, it is only necessary to determine if the
design constraints are met and then assess the Capital and operating costs amortized to present
value.
-| • . ., -
Factors that will affect the cost and effectiveness of a product recovery system will include
the number and location of wells and/or trenches, water and product pumping rates, treatment
costs, and period of operation required to reach asymptotic product recovery. In some cases, a
surrogate for cost may be employed to optimize the design. For example, if treatment costs aVe
high, minimizing total water pumping may be an effective surrogate for design optimization. If
costs associated with site disruption during remediation are high, minimizing the duration of the
recovery system operation may be a reasonable surrogate for design optimization.
For most sites, regulatory requirements necessitate remediation beyond removal of free
product to meet soil, groundwater or air quality criteria. In such cases, free product recovery is no
longer the endpoint, but simply a first step in the remedial process. This significantly complicates
the estimation of cost-effectiveness because total project cost, not simply the cost of free product
recovery, must be considered. If the free product recovery design can achieve cost savings in
\43
-------
secondary remediation that exceed the additional costs of free product recovery, higher costs for
free product recovery will be justified. Accurate estimation of the time and cost that will be
required to reach specified soil or groundwater concentration levels is difficult. However, for a
given system design, the duration of the remedial operation and the operating costs may be
expected to be approximately proportionate to the mass of contaminant remaining after free
product recovery is complete. Since it is generally less costly to remove a unit mass of
contaminant as liquid hydrocarbon than as a dissolved or vapor phase constituent, a reasonable
design criteria, in cases where soil and groundwater criteria must be met, may be to maximize the
volume of free product recovered. i
In summary, the most exacting method of evaluating various remedial alternatives is to
perform a detailed cost analysis of all options that will meet imposed criteria (e.g., limit further
plume migration, avoid construction in certain areas, meet regulatory criteria, etc.). For cases
where the effort of performing such analyses is not warranted, or for preliminary screening of
alternatives in any case, cost may be replaced by a surrogate variable that can be more readily
evaluated. j
i
To design a cost-effective free produci recovery system, various surrogates for cost may
be applicable in different circumstances. Some practical guidelines are given below.
Controlling Factor For Design | Design Surrogate
i
• Cleanup required to meet low soil or j • Maximize product recovery
groundwater standards j
• High costs associated with disruption, i • Minimize time to asymptotic recovery
maintenance, monitoring or leasing ;
• High pumping or water treatment costs j • Minimize total water pumped
1
For either the full cost analysis or the cost surrogate approach, the first step is to
determine whether a proposed design strategy !is expected to meet pre-conditions for a valid
design. Plume containment will often be specified as a prerequisite. If two pumps with a 10 gpm
capacity are readily available, a prerequisite may be to limit the number of pumps and the rates to
suit the equipment (if other conditions can also be met -- e.g. plume control). Site conditions may
dictate other limitations, such as constraints on well placement, feasibility of constructing
trenches, etc. i
i ' . • - ' • " • -
In order to determine if pre-conditions lean be met for a given design alternative, and then
to compare the optimization criteria (i.e., cost br cost surrogate), models must be used. This
chapter describes methods that can be utilized ito design free product recovery systems at
hydrocarbon spill sites. Secondary remedial measures to reduce soil and groundwater
concentrations will not be directly addressed. However, the final objectives for site closure must
always be taken into account from the outset of a remedial design, beginning with the manner in
44
-------
I
which recovery systems utilized for free product recovery can be optimally utilized in the final
system configuration. | v
4.2 Effects of Well Placement and Operation
. i .. .
i
Case Study I \
A numerical study of product recovery in an aiiisotropic fractured rock aquifer was reported by
Parker etal. (1992). The problem involves a 27,000 gallon diesel spill investigated using the areal
oil-water flow model ARMOS (ES&T, 1994).! The distribution of the NAPL plume at the time
recovery was commenced and locations of potential recovery wells are shown in Figure 4-1. Well
RW-7 is about 50 feet downgradient of the spill source, wells RW-4, -5 and -6 are 250 feet
downgradient, and wells RW-1, -2 and -3 are about 450 feet downgradient. The following
recovery well configurations were considered in the study:
Case A - Wells RW-1 through RW-3 operating at 0.5 gpm each,
Case B - Wells RW-1 through RW-6 operating at 0.5 gpm each,
Case C - Wells RW-1 through RW-7 operating at 0.5 gpm each.
1000
900
800
700
600
500
400
300
200
100
0 100 200 300 4001 500 600 700 800 900 1000
M I II II I II II II |1;I I I I I I I 1 Mil I I J I | II j I L.
I
N
o rt I i 1111111! m 11 m 1111 n i n n 1111 11111-
1000
900
800
700
600
500
400
300
200
100
0
0 100 200 300 400 ! 500 600 700 800 900 1000
Figure 4-1. Apparent product thickness prior to recovery and well locations for Case Study.
j
J45
-------
Preliminary simulations indicated a primping rate of 0.5 gpm yielded capture for wells
RW-1, -2 and -3 when operating. Simulations of product recovery were performed for each case
until asymptotic recovery was achieved. The results are summarized in Table 4-1.
Table 4-1. Summary of Recovery System Results for Case Study I.
Case
A
B
C
Product Recovery
Gallons Percent
5,700 21
2,900 11
2,700 10
Water Pumped
(Gallons x 10s)
1 1.94
i
! 1.51
1.01
i
Time
(days)
900
350
200
Oil/Water Cut
xlOOO
2.9
1.9
2.7
The maximum recovery was obtained for the downgradient well system (Case A), which
recovered 21 percent of the original spill volume. Asymptotic recovery was achieved in 900 days,
after pumping 1.94 million gallons of water, giving an oil-water cut of 2.9 gallons of oil per 100
gallons water. !
Adding an additional bank of wells in base B reduced the recovery time to 350 days, and
the recovery volume was cut nearly in half. The reduced recovery is a result of increased smearing
in the drawdown depression associated with the additional upgradient wells. The recovery time is
decreased because the average travel distance| to the wells decreases. Oil-water cut dropped to
only 1.9 gallons of oil per 1000 gallons of water.
Adding well RW-7 in Case C resulted'in a small additional decrease in recovery volume,
and reduced the recovery time to only 200 days. Oil-water cut increased to 2.7 gallons of oil per
1000 gallons of water due to the decrease in recovery time.
It is notable that increasing the number of wells decreases the volume of recovered
product for this problem. This is due to the geometry of the plume and to the configuration of the
recovery wells. Since the plume is rather elongated, plume capture can be achieved most
efficiently with a line of wells near the toe of the plume. Additional upgradient wells do not
improve containment. Increased "smearing" from the upgradient wells reduces product recovery
but reduces the travel distance, thus decreasing recovery time. An option that was not
investigated, but which may be advantageous,^ would be to include upgradient wells to reduce the
travel time, but with lower pumping rates than the downgradient wells to minimize smearing.
46
-------
Defining the optimal design depends on the objectives of the system. Case A provides the
maximum product recovery and yields the highest efficiency, as measured by oil-water cut. It also
has the lowest capital cost. Case C reaches asymptotic recovery within a much shorter time period
with significantly less total water pumping. This may result in lower total operating costs,
although capital costs will be greater. If residual product must be remediated, Case C becomes
less attractive due to higher long term pumping needed to remove the additional mass from the
soil. The results clearly illustrate the fact that recovery system design generally requires trade-offs
to be made among various cost and efficiency factors.
Case Study H ' ' \
An example problem is presented to investigate effects of water pumping rate, well placement and
number of wells on free product recovery system effectiveness. Data for the problem was taken
from a field investigation of a pipeline leak involving a mixture of gasoline, diesel and fuel oil. The
estimated spill volume is 290,000 gallons. Numerical simulations were carried out using the
numerical model ARMOS (ES&T, 1994). Tolsimplify the analysis and to focus on the effects of
design variables, effects of seasonal water table fluctuations observed in the field were not
considered in the simulations. \
The model domain is a 43 acre area, discretized by a mesh with 804 elements (Figure 4-2).
The water table at the site occurs at a depth of 10 feet, and slightly to the west with a gradient of
about 0.002. The apparent hydrocarbon thickness prior to initiation of product recovery is shown
in Figure 4-2. The unconfined aquifer consists primarily of coarse sand and gravel. Soil and fluid
parameters for the problem are given in Table 4-2.
Table 4-2. SoU and Fluid Parameters for Case Study
id"1 S=0.10
• w u in ffl
1ro=2.0 (|>=0.35j ^og=0.05
Jao=3.4 a=3.5 fr-1 5^=0.25
5^=1.42 « = 2.0i
\ 47
-------
Pumping wells were located as shown|in Figure 4-2. RW-1 was placed downgradient of
the center of the plume in order to take advantage of the natural movement of the groundwater.
RW-2 was placed at the center of the plume, and RW-3 was placed near the location of the
highest apparent oil thickness. Three scenarios involving different wells operating were
considered: j
• Case A- Pumping from well RW-1| only,
• Case B - Pumping from well RW-3 [only,
• Case C ng from wells RW-d, 2 and 3.
5900
5800
5700
B600
5508
>
5400
4000 4104 4208 4312 4415 451^9 4623 4727 4831 4935 5038 5142 5246 53S0
O
c B200
O
^5100
Q B000
4900
4800
4700
4600
J—I—1 I I 1 I I 1 l' l i i l l l l i i
I i i i i
5900
5800
5700
5600
5500
5400
5300
5200
5100
5000
4900
4800
4700
4000 4104 4208 4312 4415 4519 4623 4727 4831 4935 5038 5142 5246 5350
Distance (ft)
4600
Figure 4-2. Initial well oil thickness distribution and location of recovery wells for Case
Study IL |
For each case, simulations were performed for several water pumping rates to determine
the optimum rate for each well configuration. Water pumping rates for the different scenarios are
given hi Table 4-3. The rates shown for Case [C are the totals for all wells, and the individual
wells each were assumed to pump at one third| of the total rate. Time to reach asymptotic free
product recovery and the cumulative water pumping for each scenario are also summarized in
Table 4-3. i
48
-------
Asymptotic product recovery versus water pumping rate for each case is shown in Figure
4-3. The results clearly indicate that for a given well configuration, an optimum pumping rate
exists at which maximum product recovery is achieved. This optimum represents a trade-off
between two factors. If the pumping rate is tob low, the capture zone of the well field is
inadequate to encompass most of the free phase plume and recovery is below optimum because
plume spreading occurs. If the pumping rate is higher than necessary to achieve plume capture,
recovery diminishes due to increased smearing of product within the zone of drawdown. As
pumping rate increases, the final residual oil volume hi the unsaturated zone increases due to
smearing over the region of drawdown (Figure 4-4). However, the volume of residual in the
saturated zone decreases with increasing pumping as more free product is removed and as
increased smearing occurs (Figure 4-5). Maximum recovery will occur when the total residual
volume (sum of saturated and unsaturated residual) is at a minimum.
Table 4-3. Pumping Rates, Times to Reach Asymptotic Recovery and Total Water Pumped
for Case II Scenarios (optimum jfor each case in bold)
Pumping
Rate
(gpm)
25
50
75
100
125
150
200
300
Pumping Duration
Case A
( days )
8,000
5,000
3,600
3,000
2,300
2,000
1,500
1,000
CaseB
( days )
9,000
9,000
7,800
5,800
3,800
3,300
2,400
1,600
CaseC
( days )
10,000
i6,000
3,200
' '2,700
J2,400
i,ioo
1,400
1,200
Water Pumped
Case A
(Mgal)
288
360
389
432
414
432
432
432
CaseB
(Mgal)
324
648
842
835
684
713
691
691
Case C
(Mgal)
360
432
346
389
432
454
403
518
49
-------
175-r
"150-
Q
125-
o
CD
-Q
0
C-
100-
75-
o
o
CD
50-
25-
11111111111111111111111111
50 100 ' 150
e C
I I 1J I M I I I I I I | I I 1 111 I I If
200 250 300
Rumpling Rate ( g p m)
Figure 4-3. Final product recovery versus water pumping rate for Case Study II scenarios.
•160-
o
D)
Q
.120-
to
J? 80
I '' 1 I I ' I I I 1,1 1 1 1 I I M I I I I | I I 1 M I I I I | I I 1 1 1 1 M I | I I I 1 1 1 I I I | 1 I I
50 100 I 150 200 250 300
Pumping Rate (gpm)
Figure 4-4. Final unsaturated zone residual product versus water pumping rate
for Case Study n scenarios. ;
! 50
-------
160-
-------
Case C, with three wells operating, yielded somewhat less recovery than Case B but
reached asymtotic recovery in the shortest time period. The total water pumping was much less
than that for Case B and only slightly more than that for Case A. Although the capital cost for
Case C is somewhat higher and the total recovery somewhat lower than for Case B, the system is
significantly more effective in terms of total product removed, operating time, and oil pumped per
volume of water operating time and operating costs.
In contrast to the results of Case Study I in the previous section, the present results
indicate that upgradient wells may improve system effectiveness and even efficiency. Clearly,
many factors affect the outcome of various design options and a priori inferences concerning
system behavior are not always possible or reliable. Plume geometry, hydraulic gradient,-soil and
fluid properties, well placement and distribution of pumping among pumps, and many other
factors must be taken into consideration. Quantitative models that can take the complex
interactions of these factors into consideration are critical to making good engineering decisions.
i
4.3 Flume Capture and Travel Time Analysis
i
Evaluation of Plume Capture i
As suggested in the preceding section, free phase plume control is generally a prerequisite to
achieving maximum free product recovery for a given well configuration. In most cases, plume
control will be a design prerequisite regardless of whether maximum recovery is the primary
objective, because plume migration results in increased potential groundwater contamination and
increased potential liability. ;
Plume containment is generally achieved by installing a system of wells and/or trenches
from which water and product are pumped. For a given configuration of wells or trenches, the
maximum product recovery is achieved by pumping in a manner that just controls lateral plume
migration. Increasing water pumping further generally diminishes recovery. This occurs due to
the increasing volume of residual product that becomes smeared over the cone of depression of
the water table drawdown. Thus, for a given ^vvell or trench configuration, water pumping rates
generally should be adjusted to just control product spreading.
In the absence of air pressure gradients, water flow occurs in response to gradients in the
air-water table, Zavf, often referred to as the corrected water table (i.e., Zaw=Zow+prJ3o ). For a
single well, the steady state water table drawdown distribution may be computed approximately
by solving the groundwater flow equation, disregarding effects of hydrocarbon on water flow.
Since the saturated aquifer thickness is usually much greater than the zone with hydrocarbon, the
assumption that groundwater flow is independent of the presence of hydrocarbon is usually a
reasonable assumption. j
The groundwater flow equations may be solved using commonly available numerical or
analytical models. The simplest approach is to use an analytical solution for steady state flow to a
point source or sink (i.e., an injection or pump'ing well), and to superpose the resulting drawdown
j "*
i
I 52
-------
to approximate a system of wells or a line sink (i.e., a trench). Superposition is strictly valid for
confined aquifers, but may be used to approximate unconfined flow if drawdown is small
compared to the aquifer thickness. From the principle of superposition, the effect of multiple wells
on the water table distribution (assuming no air pressure gradients) is computed by
! \ (4-1)
1 j
7 -7 - \^ \7
~^
where Z ° is the air-water table elevation at a given location without pumping, AZ^, is the
drawdown at the location attributable to well j, and J is the total number of wells. The
pre-pumping water table configuration, Z (x.y), may be determined by kriging air-water table
elevations at monitoring wells onto a grid and the post-pumping water table may be computed
from (4.1) and drawdowns at grid points attributable to each well.
j
i
Horizontal flow of separate phase hydrocarbon occurs in response to gradients in the
air-oil table elevation, Zao, in the absence of air pressure gradients. Once we determine the
air-water table, the air-oil table may be estimated for each point in the computational grid as
i (4.2)
where Zaw is the post-pumping water table, pra is the oil specific gravity, and H0 is the apparent
product thickness at the location prior to commencing pumping interpolated from monitoring well
data. Equation (4.2) yields an approximation of the air-oil table after water flow approaches
steady state conditions, but before significant oil redistribution has occurred. Since oil tends to
accumulate at the wells and gradually diminishes in thickness near the plume perimeter, the oil
gradient around the edge of the plume, computed from (4.2), will generally be an upper estimate.
Thus, if we can show that the gradient of ZaJ is flat or towards the recovery wells at all locations
around the plume perimeter, control of plume migration should be achieved for the assumed
pumping conditions. I
To evaluate whether or not oil plume 'control will be achieved, we first delineate the plume
perimeter by a small apparent product thickness contour (for example, 0.1 ft). Inspection of the
direction of the air-oil gradient on the plume perimeter indicates whether plume control has been
achieved. Alternatively, oil streamlines perpendicular to Zao contours may be drawn to assess
capture by the recovery well system. ;
Estimation of Recovery Time !
Once flow rates have been determined for a given well or trench configuration that achieves free
phase plume control, an estimate of the time to reach asymptotic recovery may be made. An
approximate method for doing this is based on the travel time analysis discussed in Section 3.3.
53
-------
The first step is to draw streamlines perpendicular to the estimated steady state Zao
contours. The longest travel path between a recovery well or trench and the perimeter of the
plume should be identified, as this will control the time required to reach asymptotic free product
recovery. The travel path may be divided intb a number of intervals. The air-oil gradient for an
interval times the mobility coefficient (Section 3.3) gives the travel time for the interval. Summing
the interval travel times will yield the total travel time.
Example Problem - Section 3.3 showed that the mobility factor Mo can be multiplied by the
water table gradient to estimate the average oil velocity (v0) during redistribution without
pumping. The same principle can be applied to the case of oil moving to a well or trench, with the
added complication that the pathline may rtot be straight nor the gradient constant. In these
situations, the pathline can be broken into smaller segments that are approximately straight with
uniform gradient. ;
Returning to the example of Section 4.2, we analyze travel time using the optimum case of
three recovery wells (Case C) pumping at a combined rate of 150 gpm. Figure 20 shows the initial
distribution of the free phase plume overlain with pathlines. Each pathline depicts the movement
of an oil "particle" from the edge of the plume to a well. The travel time along each pathline is
computed using the steady state water flow field. The average velocity of the oil phase can be
computed from the mobility factor M0 and the gradient in Zao, via eq. (3.11). During oil
recovery, Zao will drop continuously until it reaches the water table. Thus, for a steady state
approximation, we may use the water table drawdown around a recovery well as an estimate of
\ 54
-------
Figure 4-6. Flow net for travel time analysis. Solid lines are steady state
dashed lines are streamlines, area within dark line is plume,
arrow shows longest path.
contours,
The longest pathline in Figure 4-6 depicts the maximum travel distance for oil during free
product recovery. Thus the travel time along this pathline should correspond to the time needed
to reach asymptotic recovery. For the longest pathline, an oil travel time of 2360 days from the
edge of the plume to the well is predicted using the maximum mobility factor M0max (eq. 3. 12).
This is slightly higher than the recovery time predicted by the transient oil-water°flow model
ARMOS of 2100 days (Table 4-3). !
: 55
-------
4.4 Estimation of Recoverable Product !
i ' "
Description of Method \
Methods for estimating the volume of free oz/iwere discussed in Section 3.2. Unfortunately, only
a fraction of this volume will be recoverable as a separate phase, due to various processes that
lead to the occurrence of residual oil. We distinguish between residual oil in the liquid-saturated
zone, which occurs as hydraulically discontinuous blobs trapped within a continuous water phase,
and residual oil hi the unsaturated zone, which occurs as thin films and as pendular rings of oil at
particle contacts. The recoverable volume is given by
i =1-1 -y (4J)
Z_roo i—iof f—IOt i-lOg
I :
where £0/ is the current free oil volume computed as described in Section 3.2, £ofis the volume
of current free oil that becomes trapped in the [liquid saturated zone, and £0 is the volume of
current free oil that is retained as residual oil in the unsaturated zone.
The total saturated and unsaturated zone residual oil volumes may be computed from
specific volumes on a grid as \
! (4.4)
JV
— A X""^ W
>t Z-*1 Of/
N
E
/=!
(4-5)
— A X~^ T7"
J= -^f !«,
where Vot/ is the residual oil specific volume that is trapped in the liquid saturated zone at
location /, Vogf is the residual oil specific volume held against gravitational drainage in the
unsaturated zone, and A is the element area. !
Residual oil in the unsaturated zone arises during periods of falling Zao, when downward
oil redistribution eventually becomes negligible under gravitational forces, as^oil saturation
reaches a critical value referred to as the unsaturated zone residual saturation (see Section 2.4).
The increase in residual oil specific volume in the unsaturated zone due to a drop in the air-oil
table, AZao, is described by equation (2.12) which is applied in a series of pseudo "time-steps"
until the cumulative change in Zao is equal to A.Zaw+(l-pr()H0+kZR where AZflw is the water
table drawdown, Ho is the initial well oil thickness at the location, and AZ^ is the regional change
in Zao associated with seasonal water table fluctuations.
Saturated zone residual oil specific volume may be computed as described in Section 2.4
for the anticipated change in the oil-water table elevation AZow. With removal of product from
; 56
-------
the formation during steady water pumping, the oil-water table will increase until well oil
thickness (H0) approaches zero at which point 2^=2^. Also considering possible water table
fluctuations, &Zow=prcfIo+&ZR, where AZow is the cumulative change in Zow, AZ^is the regional
change in Zow due to seasonal water table fluctuations, and H0 is the initial well product
thickness. The latter is corrected for residual oil in the unsaturated zone by computing the value
of H0 at each grid point corresponding to the' initial Vof minus 2xFog , where the factor 2 takes
into consideration the fact that gradual oil drainage from the unsaturated zone reduces saturated
zone trapped oil. A correction for the change! in capillary fringe thickness due to hysteresis may
be applied to AZOW as described in Section 2.4.
The estimates of asymptotic recoverable product volumes described above are based on
projected estimates of residual oil in the saturated and unsaturated zones, associated with changes
in fluid table elevations due to water pumping; and product removal. This assumes that lateral
plume spreading is controlled. If this condition is not met, the actual recoverable product may be
substantially less than the computed values, since residual volumes will increase as the plume
spreads over a greater area. Therefore, it is very important to verify plume control by
determining that oil gradients are inwards on the plume perimeter, before performing
recoverable product calculations. ;
Example Problem - The following example demonstrates the calculation of residual and
recoverable oil specific volumes. To compute total recovery, similar calculations must be repeated
for each location where fluid levels are known or interpolated, so that actual volumes can be
computed by summation over the area. The example problem continues with the problem
introduced in Section 3.2, which has an initial Apparent product thickness of 3 feet, with the air-oil
table at 93 feet and the oil-water table at 90 feet. Pumping is assumed to cause a 1.5 foot
drawdown in Zao at the location. Soil and fluid properties for the problem are given in Table 5. A
step-by-step description of the methodology for computing residual and recoverable product
specific volumes is given below for the example problem.
1. The initial free oil specific volume prior to beginning pumping is computed as described in
Section 3.2., which indicates F0/=0.309fee|t.
2. The average initial free oil saturation is estimated as
C -
.
which gives a value of 0.294. '
3. An initial estimate of the unsaturated zone residual oil specific volume, V™li ', is computed
as
Vinit=Min(V V V
yg jviw^Y gl,v g2,y g
57
-------
where
= 0-1544 feet
V =
83
of
1 - exp
H
= 0.1217 feet
/f
which gives Fgr=0.0315 feet. j
4. A revised estimate of the free oil specific volume corrected for oil retention in the unsaturated
zone is computed as '
of ~v of
init
where the factor 2 is assumed to account for the finite time for oil to drain to a new vertical
equilibrium after the onset of pumping, during which time product recovery will proceed. The
corrected free oil specific volume for the example is V"™ '=0.246 feet.
i
5. A revised value of apparent product thickness corresponding to the updated free oil specific
volume of 0.246 feet is determined by interpolation from a table of H0 versus Vof generated
using the method described in Section 3.2, yielding H"ew=2.65 feet.
6. To simulate oil recovery H0 is reduced to zero in N "time- steps", causing Zao to drop and
ow
imulate oil recovery H0 is reduced to zero in N "time- steps", causing Zao to drop an
to rise in incremental steps. The incremental changes in Zao and Zow are calculated as
where Zc is the change in the capillary fringe due to reversal from oil imbibition to oil
drainage given by (2. lie). For the example problem, Zc=0.67feet and employing ^=30
"tune-steps" gives AZao=0.0177 feet and AZOW=0.0484 feet. Calculations of residual oil in
the saturated and unsaturated zones for 30 • 'time-steps" are shown in spreadsheet form in
Table 4-4 with columns in the spreadsheet computed as follows:
58
-------
Table 4-4. Spreadsheet for Calculation of Residual and Recoverable Oil Specific Volume
B
D
Step
0
1
• 2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
H0
(feet)
2.65
2.56
2.47
2.39
2.30
2.21
2.12
2.03
1.94
1.85
1.77 .
1.68
1.59
1.50
1.41
1.32
1.24
1.15
1.06
0.97
0.88
0.79
0.71
0.62
0.53
0.44
0.35
0.26
0.18
0.09
0.00
V
-------
Column A:
Column B:
Column C:
Column D:
Column E:
This column is simply an integer step number, 1,2, N.
The value H"*w is input in row 1. At each new step, H0/30 is subtracted from the
previous H0 to reduce H0 to zero in 30 steps.
i
Vof is computed from the corresponding H0 in column A using the method
described in Section 3.2.
; _ y
The average free oil saturationiis calculated as S f= — — .
of
The residual oil saturation in the saturated zone for the step is computed as
^On (Sor,S0)
This column is summed to obtain 215^=3.473 which is multiplied by
AZow=0.0169 to yield F;=OJ0587 feet.
Column F: The residual oil saturation in the unsaturated zone for the step is computed as
Sog=Mn(Sog,Sof)
This column is summed to obtain *LSot=1.453 which is multiplied by
-------
Verification of Recoverable Oil Algorithm
The residual and recoverable oil calculations described above, as implemented in the program
SPILLCAD (ES&T, 1994) which also performs areal interpolation and integration of specific
volumes, were verified by comparison with a model for transient oil and water flow. The test
problem is the gasoline spill described in Section 3.3 with pumping from a single recovery well
placed down-gradient of the center of the plume initiated 100 days after the leak stopped.
Water is pumped from the recovery well at rates varying from 0 to 18 gpm, while oil is
recovered at a rate to maintain product thickness in the well near zero. Transient oil and water
pumping was simulated using the numerical model ARMOS (ES&T, 1994). Both models
predicted plume containment at a pumping rate of about 4 gpm. Comparisons of recoverable oil,
unsaturated zone residual oil, and saturated zone residual oil as functions of water pumping rate
predicted by the two models are shown in Figures 4-7, 4-8, and 4-9.
SPILLCAD results for pumping rates less than 4 gpm are shown as dashed lines because
residual and recoverable estimates using this approach are not meaningful unless containment is
achieved. The pseudo-transient calculations in SPILLCAD somewhat under-predict product
recovery, mostly because it traps more oil thaiji the more rigorous numerical model in the
saturated zone. This effect can be explained by movement of oil to the well as drawdown occurs.
In the early stage of recovery, a certain amount of oil will be recovered before water flow reaches
steady state. I
During this period, Zow drops with Z^, even though oil is being recovered. SPILLCAD
cannot capture this transient effect, as the pseudo time-steps always cause Zow to rise and trap oil
(see previous example problem). Considering this limitation, the agreement between the two
models is quite good. In both models, residual oil in the unsaturated zone increases with pumping
rate as oil smears over a larger volume of soil. This causes residual oil in the saturated zone to
decrease with pumping rate as less free oil redistributes below the water table.
61
-------
O i
JBi
ARMOS
SPILLCAD
Pumping Rate (gpm)
IB
Figure 4-7. Comparison of recoverable oil volume versus water pumping rate for ARMOS
and SFILLCAD. Dashed line indicates pumping rate too low to provide plume
capture. ;
a.
i
p
ARMOS
: 6 9 12 15
Pumping Kate (gpm)
Figure 4-8. Comparison of unsaturated zone residual oil volume versus water pumping
rate for ARMOS and SPILLCAD. Dashed line indicates pumping rate too low
to provide plume capture.
' 62
-------
SPILLCAD
2008©
£3-
C3
-*9
S3
3
-------
A major advantage of using trenches for free product plume control is that definitive
plume containment can be achieved with little or even no water pumping. If the trench intercepts
the oil plume pathline, skimming product from the trench without water pumping will prevent
further free phase plume migration past the trench. Even if product thickness in the trench is
nonzero due to placement or operating conditions for skimming pumps, positive plume
containment can be achieved due to a capillary barrier effect which prevents oil infiltration
through the downgradient trench wall. As indicated by the Laplace capillary equations (eq. 2.2), a
finite nonwetting phase pressure (hence, trench oil thickness) is needed before nonwetting fluid
(oil) can displace a wetting phase (water). The phenomenon is evident in the form of the
relationship between free oil specific volume or oil transmissivity and apparent oil thickness
discussed in Section 2.5. This finite oil thickness, which is greater for finer soils, is required for oil
entry into the soil. This thickness may range from a few inches hi a sandy soil to several feet in a
fine grained soil. The capillary barrier effect provides a safety factor for free phase plume control
using trenches. ;
i
As a product recovery measure, trenches are generally both more efficient and more
effective than a line of wells. Because the flow field to a well is radially converging, while flow to
a trench is planar, a larger drawdown will be needed for wells to obtain the same oil recovery rate
achieved by a trench. Furthermore, because residual oil will be increased by the greater
drawdown, the ultimate product recovery will jtypically be lower for the well system.
The recovery rate for free product in a'trench may be estimated directly from Darcy's law
i ;
Lz (4-6)
as
,dx
where R0 is the oil recovery rate (e.g., cubic feet per day), T0 is the oil transmissivity at the
trench (see Section 2.5), L is the trench length, and dZJdx is the air-oil table gradient at the
trench wall (assuming no air pressure gradient). For a given trench configuration and water
pumping rate (hence air-water table), the maximum oil recovery rate will be achieved by
maintaining the oil thickness in the trench near zero, although oil recovery rate will be relatively
insensitive to the oil thickness in the trench. This is because the product thickness in the soil
adjacent to the trench will adjust naturally to form an oil seepage face with an upper elevation that
maximizes the oil flow rate into the trench. The maximum flow rate corresponds to the maximum
of the product of oil transmissivity, which increases with oil thickness and hence oil seepage face
elevation, and decreases with oil gradient, which decreases with increasing oil seepage face
elevation. Unless the oil level in the trench is raised above the natural seepage face elevation,
negligible reduction in flow rate will be observed. Similar seepage face phenomena are observed
in wells.
164
-------
4.6 Vacuum Enhanced Free Product Recovery
Basic Concepts \
Vacuum enhanced recovery (VER) is a modification of conventional free product recovery system
design, in which a vacuum is applied to the well bore to increase the hydraulic gradient. In
conventional product recovery systems, free phase hydrocarbon is removed by pumping
hydrocarbon from recovery wells, usually in conjunction with water pumping to increase the
gradient and the radius of influence. The efficiency of such systems depends on soil and fluid
properties, environmental variables, as well as design variables such as the number of wells and
the rates of water pumping. Since light hydrocarbons float on the water table, the radius of
influence for hydrocarbon flow is directly related to the cone of depression produced by pumping
groundwater. When the pumping rate is too low, hydraulic control of the free product plume is
not achieved and spreading of the plume leads to an increase in residual product. Increasing the
water pumping rate generally increases the rate of product recovery initially and may increase the
total recovery up to the point where plume control is achieved. However, for large drawdowns
associated with high pumping rates, the mobile hydrocarbon is smeared at residual saturation in
the unsaturated zone. Therefore, an optimal pumping rate exists for maximum product recovery
for a given well configuration at which the pumping rate is high enough to yield plume capture.
Applying a vacuum to a recovery well enables the piezometric gradient for oil and water
flow to a well to be increased, without a corresponding reduction in the air-oil and oil-water
tables. When a vacuum is applied in a well, liquids in the well and the soil will rise due to the
reduced pressure above the fluid interfaces. This phenomenon called upwelling can be reversed
by lowering the water pressure in the bore hole and increasing the water pumping rate. The
locations of the piezometric surfaces near a recovery well before and after the vacuum is applied
are shown in Figure 4-10. The application of vacuum (ha) lowers both the oil and the water
piezometric surfaces (*f0 and Yw), which results in an increased gradient for flow. If the oil and
water pump intake (or control switch) elevations are not altered, increased water and oil flow
rates may be achieved without lowering the liquid levels and thus without increasing the amount
of residual product in the unsaturated smear zone. The radius of influence of the air vacuum will
depend to a significant degree on the magnitude of vertical "leakage" of air that occurs from the
ground surface to vacuum wells. i
65
-------
Initial oil piezometric surface
Initial water piezometric surface
Final water piezometric surface
Figure 4-10. Schematic of vacuum enhanced product recovery system.
Example Problem Description - An example problem is presented to investigate the effects of
various soil and design variables on VER system effectiveness. The problem is the same as Case
Study II discussed in Section 4.2 involving a pipeline leak of gasoline, diesel and fuel oil with
three recovery wells (Case C). ARMOS was used to simulate water, NAPL and air flow. Water
and NAPL recovery was simulated assuming water pumping was performed at a specified rate,
while NAPL was removed at a rate sufficient to maintain zero well product thickness. Air flow
was controlled by specifying the air vacuum at the well, assumed to be 3.5 feet of water head.
The ground surface was assumed to be either covered or uncovered. For the covered case, no air
leakage from the ground surface was permitted, while vertical flow from the ground surface was
considered for the uncovered case. The following three cases were considered:
I
1. Base case involving conventional product recovery without vacuum,
2. VER system without a surface cover, and
3. VER system with a surface cover. i
For each of these cases, several simulations were performed with fluid levels specified at
different drawdowns relative to the initial water table elevation. Water pumping rates varied from
13 to 103 gpm for the various subcases. For the assumed vacuum and well placement, the
optimum drawdown (or water pumping rate) tja maximize product recovery may be determined.
A comparison of the predicted cumulative product recovery versus time for the three cases
at a water pumping rate of 52 gpm, which is close to the optimum rate for maximum recovery for
all cases, is shown in Figure 4-11. The results indicate that the recovery rate for the VER system
with a covered surface is initially nearly 70 percent greater than the conventional recovery system
(no VER). However, the difference in recovery rates narrows over time, such that the VER
system yields only about 7 percent more cumulative recovery after 3000 days of operation. The
VER system with an uncovered ground surface exhibits a recovery curve that is about halfway
! 66
-------
between the covered VER system and the no VER system.
The differences between the VER systems with and without a cover are due to the effects
of vertical air leakage on the radius of influence of the vacuum wells. As seen in Figures 4-12 and
4-13, the radius of influence of the vacuum wells (operationally defined as a vacuum of 0.01 feet
of water) is much smaller for the uncovered than for the covered case. Due to the limited areal
extent of the vacuum, effects on hydrocarbon jrecovery are greatly reduced. It is important to note
that the "covered" case assumed that the ground surface is perfectly sealed to air flow.
Pavements and buildings seldom closely approximate such an ideal condition, and vertical leakage
must always be considered a real possibility, j
itst ! net itoi
Tune (davst
Figure 4-11. Effects of VER on product recovery for systems at a water pumping rate near
optimum for recovery. ;
! 67
-------
Figure 4. Zones of Influence for Water and Air Phases
in VER System with Cover.
Vacuum Zone
of Influence
Vlfater Plezometric
Surface
Extent of Free
Product Plume
Figure 4-12. Zones of influence for water and air phases in VER system with cover.
Figure 5. Zones of Influence for Water and Air Phases
in VER System without Cover.
Extort of Froo
Figure 4-13. Zones of influence for water and air phases in VER system without cover.
! 68
-------
I
5
naict —
fltlD* -
f
HUH •
_.,.....
Liquid Drawdown (feet)
Figure 4-14. Asymptotic product recovery [versus liquid drawdown with and without VER.
Discussions of conventional product recovery system design in Section 4.2 indicated that
an optimum water pumping rate will exist that; yields the maximum asymptotic product recovery
for a given well configuration. To determine now vacuum affects recovery optimization, we turn
to results of simulations of product recovery for varying piezometric drawdown. Piezometric
drawdown is the difference between the initial'static water level and the well piezometric head
(Yw = Zaw + ha). Asymptotic product recovery versus drawdown for all three cases indicate an
optimum piezometric drawdown of about 2.0 feet for the no vacuum and uncovered VER cases
and about 2.5 feet for the covered VER case (Figure 4-14). These correspond to pumping rates
of about 50 and 60 gpm, respectively. i
Evidently, the vacuum area of influence for the uncovered case is so small that the
optimum recovery is controlled by the water pumping needed to control the plume perimeter,
which is beyond the influence of vacuum wells: For the covered case, the vacuum exerts some
influence near the plume perimeter and a small increase in pumping is needed to overcome the
mounding effect. It is interesting that the optimum scenarios for the uncovered and covered
cases correspond to well bore liquid levels thai: are about 1.0 and 1.5 feet above the initial static
levels. i -
Additional simulations were performed; with different vacuums applied to the wells. The
results indicate that higher vacuum increases the initial rate of hydrocarbon recovery substantially,
but results in only small increases in final recovery. The results indicate that vacuum enhanced
product recovery may enable slightly higher asymptotic recovery to be achieved. However, the
primary utility of VER is to increase the recovery rate and hence reduce the time to complete free
product recovery. The extent to which this objective can be achieved will depend on the area of
influence for the vacuum that can be achieved, which will in turn depend strongly on the extent of
69
-------
vertical air leakage that occurs. '<
I -•. •
Some general conclusions concerning VER follow:
• For specified water and oil pump levels in recovery wells, water and oil recovery
rates will increase as the vacuum is increased,
To achieve maximum product recovery, well placement and water levels in wells
must be maintained at a level tnat will yield plume containment,
i
If the vacuum area of influence is smaller than the oil capture area, vacuum will
have a minor effect on the asymptotic recovery or on the optimal pumping rate at
which this occurs,
I
i
The effectiveness of VER decreases as the vacuum area of influence decreases,
which occurs as vertical air leakage increases and air flow rate decreases.
Effects of VER on residual soil concentrations in the unsaturated zone, due to extraction
of volatile components and to enhanced biodecay associated with increased oxygen fluxes, have
not been considered in the above discussion. In practice, these factors will provide additional
benefits of VER systems. If a soil vapor extraction or bioventing system is under consideration for
treatment of residual contamination, and hardware requirements are compatible with the operation
of a VER system, consideration should be given to co-design of the vapor extraction/bioventing
system to function initially as a VER system to maximize efficiency and reduce overall system
cost.
70
-------
Section 5
Case Study of Spill Site
5.1 Introduction
This case study presents an application of the methodology discussed in this report for assessment
and remedial design of a hydrocarbon spill site. The study is based on a release documented by
Dakin etal. (1992), which occurred in British; Columbia. The study was conducted to illustrate
how modeling can be employed to provide information to facilitate decision-making when
characterization data is limited and time is a significant constraint. Specifically, modeling was
performed (1) to assess the contamination at the site, and (2) to design a recovery system to
remediate the phase separated hydrocarbon lens existing at the water table.
The spill of concern involved the accidental release of xylene from an elevated platform into
underlying soils (see Figure 5-1). Xylene infiltrated the soils and ponded on the water table.
Response activities were constrained at the site due to existing structures and railroad operations.
In particular, the property owner required unlimited access to the site along the adjacent rail lines.
Given these constraints, modeling was considered an essential tool in the assessment and remedial
operations. :
i .
5.2 Model Application for Site Assessment
Site characterization activities performed shortly after the release determined the approximate
extent of contaminated soil, contaminated groundwater, and a separate phase lens of xylene
(Figure 5-2). The contamination occurred in fill material consisting of medium to fine sand. The
fill overlies natural sediments composed of low permeability silts. At the time of the release, the
water table was located approximately 4.5 feel! below the ground surface. In the area of the spill,
the water table slopes to the south and west with a gradient of approximately 0.02 upgradient of
the spill but flattens to a gradient of 0.005 downgradient of the release (Figure 5-3). Regional
hydrologic information suggested that the water table fluctuates between 0.75 to 1.5 feet annually
due to seasonal variations in precipitation and evaporation.
71
-------
Elevated,
Platform
00 0
I
Storage Tanks
Figure 5-1. Plan map of xylene spill site.
•N
Property Line
Rail Lines
JV
\ °
1 0
0
o
o= Monitoring Wells
'25ft.'
Figure 5-2. Contour map of apparent free
xylene thickness in feet.
72
-------
Figure 5-3. Groundwater contours at spill site in feet.
The initial assessment activities focused on: (1) definition of project objectives, (2)
development of a site conceptual model, and (3) selection of an appropriate model. In this study
the objectives were to assess the contamination from the site characterization data and to evaluate
remedial system designs to control the phase separated hydrocarbon source. Time was
considered a major factor because it was recognized that a rapid remedial response would
increase the volume of recoverable xylene and minimize the development residual NAPL.
Specifically, it was speculated, based on normal weather patterns, that a significant rise in the
water table would probably begin within approximately 4 months (120 days) of the release date.
The site characterization investigation had taken 40 days, which left about 80 days to design,
install, and implement the recovery system as well as recover the mobile xylene.
The site conceptual model derived from the characterization data consisted of a shallow,
thin, unconfined aquifer that was underlain by an aquiclude. The unconfined portion of the
aquifer was between 3.5 and 4 feet thick. This aquifer was relatively permeable and was
considered isotropic and homogeneous. Dissolved, residual, and mobile free phase xylene was
contained in the unsaturated and saturated portion of the aquifer.
; 73
-------
An independent survey of available computer codes indicates that 6 models, ARMOS,
MAGNAS, MOFAT, MOTRANS, SPILLCAD and SWANFLOW, could be utilized to model
multiphase flow conditions at the site (Weaver^ and Johnson, 1993). However, due to limitations
in site data and time constraints, it was considered necessary to utilize a practical, user-friendly
code that could generate reasonably accurate results and would not require significant data input
and setup time. Furthermore, the project objectives required the application of a multiphase flow
model that could simulate remedial activities such that recoverable product volumes, residual
product volumes, and the time of recovery could be estimated. Based on these needs,
SPILLCAD (ES&T, 1994), a screening level model for evaluating the feasibility and effectiveness
of recovery designs, was selected. SPILLCAD is an areal two-dimensional model that calculates
mobile hydrocarbon volume, contaminated soil volume, recoverable hydrocarbon volume and
residual hydrocarbon volume, and enables the;flow field and time of recovery to be estimated
using particle-tracking (pathline) analysis. Soil and fluid property values utilized in the modeling
effort are presented in Table 5-1. j
Table 5-1. Soil and Fluid Property Values for Model Simulations
P™=0-86 Ksw=23.1ftd~l Sm=0.l3
r)ra=0.81 (j>=0.30 Sog=Q.Q3
Pao=2.30 a=4.50feet ,$^=0.26
\
P =2.00. n=2.70!
5.3 Model Results •
The initial phase of the modeling study involved data entry of water and hydrocarbon elevations
as measured in the installed wells. Using this data the volume of free xylene was estimated to be
172 gallons. The NAPL lens was over a 1400 square feet in area with the thickest portion of the
lens occurring about 5 feet downgradient of the platform. Of particular concern was the fact that
a considerable portion of the lens occurred underneath the railroad lines, thus, potentially
complicating recovery. Using the initial information generated by the model, the next phase of the
modeling effort addressed two aspects: (1) the design of an optimal recovery system that would
minimize smearing of xylene and that would incorporate the physical constraints of the site and
(2) evaluate the impact of water table fluctuations on the effectiveness of the recovery operations.
i
Three recovery systems were investigated. The first scenario involved locating the
recovery well placed upgradient of the rail lines in the thickest zone of free xylene, the second
scenario involved locating the recovery well downgradient of the rail lines, and the third scenario
I
: 74
-------
entailed both of these well locations. Using these well locations, a variety of pumping rates were
evaluated for each of the wells. As shown in Table 12 capture was obtained when pumping rates
were at least 0.5 gpm for Scenarios 1 and 3 arid 1.0 gpm for Scenario 2. Pumping rates greater
than 1.25 gpm resulted in excessive drawdown corresponding to water levels below the pump
elevation. Model results indicate that scenarips with pumping rates between 0.5 and 1.25 gpm
yielded similar recovery (Table 5-2). i
i
Table 5-2. Results of SPILLCAD Modeling Simulations
Pump-
ing
Rate
Per
Well
(gpm)
0.1
0.5
1.0
1.25
1.5
Single Downgradient Well
Recov-
ery
(gal)
-
75
76
74
Residual (gal)
Unsat-
urated
-
21
37
44
Satu-
rated
-
73
58
53
Cap-
ture
N
Y
Y
Y
Single Upgradient Well
Recov-
ery
(gal)
-
-
77
75
' Residual (gal)
jUnsat-
1 urated
i
s
! 41
' 49
Satu-
rated
-
-
54
48
Cap-
ture
N
-N
Y
Y
Both Wells
Recov-
ery
(gal)
-
77
71
69
Residual (gal)
Unsat-
,urated
-
36
66
77
Satu-
rated
-
58
35
26
Cap-
ture
N
Y
Y
Y
Aquifer pumped past screen bottom
Given these results, and in particular, the potential of smearing the hydrocarbon vertically
within the soil profile, concern was then focused on how water table fluctuations might impact the
recovery operations. The model enables the user to simulate regional water table fluctuations to
evaluate the impact on recoverable hydrocarbon volume. This feature was found to be extremely
useful, as the only other approach to obtain this information would have required the construction
of a complex transient finite difference or finite-element model, which not only would have
consumed valuable resources but also valuable time. Water table fluctuations of 1.5 and 0.75 feet
were simulated for both well configurations at their optimal water pumping rates (Table 5-3).
The results indicated that the seasonal water table fluctuations would decrease recoverable xyllene
volumes by 30 and 60 percent. Given that only half of the xylene could be recovered under even
optimal conditions, it was considered imperative to complete recovery operations prior to the
seasonal rise in groundwater levels to reduce long term cost and minimize dissolved impacts to
the groundwater. i
75
-------
Table 5-3. Results of SPILLCAD Regional Water Table Fluctuations Simulations
Well Location &
Pumping Rate (gal)
Upgradient (1.0 gpm)
Downgradient (0.5 gpm)
Two Wells (1.0 gpm total)
Fluctuation !
(ft) ;
i
0
0.75 :
1.5 I
0
0.75 !
1.5
0 !
0.75 :
1.5 ;
Product
Recovery
(gal)
77
55
36
78
49
30
77
55
36
Residual (gal)
Unsaturated
41
62
82
21
50
71
36
58
78
Saturated
54
54
54
73
72
70
58
58
58
The final aspect of the modeling effort involved determining the time required to reach
asymptotic recovery. More specifically, could the optimal recovery of 80 gallons of xylene be
achieved within the eighty day time interval remaining prior to the advent of the seasonal rising
water table using the rates and well locations previously defined. To address this issue, backward
particle-tracking analysis was conducted for each of the recovery wells. Using the NAPL travel
time analysis, it was determined that optimal recovery could occur within 60 days at each individ-
ual well. The downgradient recovery well could recover the lens at a pumping rate of 0.5 gpm
while the upgradient well would require a pumping rate of 1.0 gpm. As shown in Figure 5-2, the
distribution and length of the pathlines indicate that most of the plume north, south and east of the
well is recovered prior to 60 days. However, areas southwest of the well do not become captured
until 60 days (Figure 5-4). A third scenario was executed to determine the time of recovery if
both wells were operational. A recovery rate of 0.5 gpm was defined for each well. This
simulation resulted in a similar recoverable volume, but the time of recovery was decreased by 15
days. !
76
-------
Figure 5-4. Results of particle-tracking analysis for two well scenario. Pathlines are based
on a 60 day time of travel. Arrow indicates streamline that limits recovery
time.
Information from the SPILLCAD modeling effort was used to determine the efficiency of
the recovery system, as measured by volume of product recovered, relative to total fluids
recovered. As shown in Table 5-4, the calculated efficiency ratios were determined to be.2.0, 1.1,
and 0.9 gallons per 1000 gallons for the downgradient single well system, the two well system,
and the upgradient single well system, respectively. Since the recoverable volume was similar for
all three scenarios, the rate and recovery time were the primary variables affecting the efficiency.
Hence, the most effective system was the downgradient single well as it produced the lowest
volume of water. Hence, less water is pumped per volume of recovered xylene. The downgradient
system was 1.8 times more efficient at product recovery than the two well system.
77
-------
Table 5-4. Optimal Recovery Characteristics for Scenarios 1,2, and 3
Well Location & Total
Pumping Rate (gpm)
UpgradientWell(l gpm)
Downgradient Well (0.5 gpm)
Both Wells (1.0 gpm)
Recovery
(gal)
77
78
77
Residual (gal)
Unsaturated
41 1
21!
36!
Saturated
54
73
58
Asymp-
totic
Recovery
Time
60
55
45
Water
Pumped
(X1000
gal)
86.4
39.6
64.8
Oil-
Water Cut
(gal/1000
gal)
0.9
2.0
1.1
5.4 Summary and Conclusions
The results of the SPILLCAD modeling effort provided critical information regarding possible
remedial efforts at the release. This information included:
1) Optimal xylene recovery was approximately 80 gallons,
2) Optimal rates ranged from 0.5 ito 1 gpm,
3) Seasonal fluctuations decreased the volume of recoverable xylene by 30 and 60
percent, :
4) Optimal recovery could be achieved within 45 and 60 days, and
5) The efficiency of the recovery systems ranged from 0.9 to 2.0 gallons recovered
per 1000 gallons of water pumped.
Based on this modeling effort, the downgradient single well system appears to provide the
needed results with the highest efficiency and the lowest capital costs. The other systems,
although capable of producing a similar recovery volume, are not as efficient, and hence, will
result in higher remedial costs. j
Information from simple models can provide valuable information to evaluate remedial
options. In this case study, the answers provided by modeling could not be obtained within the
needed time frame by any other means. As a result, application of the models discussed in this
report is proven to be an important tool for assessment of xylene contamination at the site as well
as in the design of a recovery system. Although the constraints illustrated in this case study are
possibly more severe than at most sites, the methods are broadly applicable and can be used at all
sites to better assess, contain, and remediate hydrocarbon contaminants.
: 78
-------
References
American Petroleum Institute. 1989, A Guide to the Assessment and Remediation of
Underground Petroleum Releases, 2nd edition.
Arya, L. M. and J. F. Paris. 1981. A physicoempirical model to predict soil moisture
characteristics from particle size distribution and bulk density data, Soil Sci. Soc. Am.
I, 45, 1023-1030. !
Baker, O. and W. Swerdloff. 1956. Finding surface tension of hydrocarbon liquids, oil and gas
J., 125. :
Carsel, R. F. and R. S. Parrish. 1988. Developing joint probability distributions of soil water
retention characteristics, Water Resour. Res., 24, 755-769.
I
Conrad, S. H., J. L. Wilson, W. R. Mason and W. Peplinski. 1992. Visualization of residual
organic liquid trapped in aquifers, Water Resource Res. 28: 467-478.
Dakin, R. A., Holmes, A. T. and Tiplady, D. J. 1992. Practical aspects of hydrogeology studies
for hydrocarbon migration in British Columbia, Subsurface Contamination by Immiscible Fluids,
K.U. Weyer (editor), A.A. Balkema, Rotterdam.
de Pastrovich, T. L., Y. Baradat, R. Barthel, A. Chiarelli and D. R. Fussell. 1979. Protection of
groundwater from oil pollution, CONCAWE, Report 3/79, The Hague, 61 p.
Environmental Systems and Technologies, Incl 1994c. SPILLCAD 3.2: Data Management and
Decision Support for Hydrocarbon Spills, Blacksburg, VA.
Environmental Systems and Technologies, Inc. 1994b. SOILPROP: A Program to Estimate
Unsaturated Soil Hydraulic Properties from Particle Size Distribution Data, Version 2.0,
Blacksburg, VA. ;
Environmental Systems and Technologies, Inci 1994a . ARMOS: Areal Multiphase Organic
Simulator for Free Phase Hydrocarbon Migration and Recovery, Vers. 5.0, Blacksburg, VA.
Environmental Systems & Technologies, Inc. ;1994. SPILLCAD 3.12: Data Management and
Decision Support for Hydrocarbon Spills, Blacksburg, VA.
i
: 79
-------
Fair, A. M., R. J. Houghtalen and D. B. McWorter. 1990. Volume estimation of light
nonaqueous phase liquids in porous media, Ground Water, 28, 48-56.
i
Gruszcenski, T. S. 1987. Determination of a,realsistic estimate of the actual formation product
thickness using monitoring wells: a field bailout test, Proc. Petroleum Hydrocarbon in
Ground-water: Prevention, Detection and Restoration, NWWA, Houston.
Hall, C. W. and J. A. Johnson. 1992. Limiting factors in groundwater remediation, Journal of
Hazardous Materials, 32, 215-225, Elvesier Science Publishers, Amsterdam.
Hall, R. A., S. B. Blake and S. C. Champlin Jr. 1984. Determination of hydrocarbon thicknesses
in sediments using borehole data, Proc. 4th Nat. Symp. on Aquifer Restoration and Ground
Water Monitoring, NWWA, 300-304.
Hampton, D. R. and P. D. G. Miller. 1988. Laboratory investigation of the relationship between
actual and apparent product thickness in sands, Proc. Petroleum Hydrocarbons and Organic
Chemicals in Ground Water, NWWA. :
i
Huntley, D. and R. N. Hawk. 1992. Non-aqueous phase hydrocarbon saturations and mobility in
a fine-grained, poorly consolidated sandstone, Proc. Petrol. Hydrocarbons and Organic
Chemicals in Groundwater, NGWA, Houston.
Kaluarachchi, J. J. and J. C. Parker. 1992. Multiphase flow in porous media with a simplified
model for oil entrapment, Transport in Porous Media, 7, 1-14.
Kemblowski, M. W. and C. Y. Chiang. 1990.J Hydrocarbon thickness fluctuations in monitoring
wells, Ground Water, 28, 244-252. j
i
i
Kool, J. B. and J. C. Parker. 1988. Analysis of the inverse problem for transient unsaturated
flow, Water Resour. Res., 24, 817-830. i
Lenhard, R. J. and J. C. Parker. 1987. Measurement and prediction of saturation-pressure
relationships in three-phase porous media systems, J. Contain. Hydrol., 1, 407-424.
Land, C. S. 1968. Calculation of imbibition relative permeability for two- and three-phase flow
from rock properties, Trans. Am. Inst. Min. Metall. Pet. Eng., 243, 149-156.
Lenhard, R. J., J. C. Parker and J. J. Kaluarachchi. 1988b. A model for hysteretic constitutive
relations governing multiphase flow, 3. Refinements and numerical simulations, Water Resour.
Res., 25, 1727-1736. ;
Lenhard, R. J. and J. C. Parker. 1990. Estimation of free hydrocarbon volume from fluid levels
in observation wells, Ground Water, 28, 57-67.
i 80
-------
Lyman, W. J., W. F. Reehl and D. H. Rosenblatt. 1982. Handbook of Chemical Property
Estimation Methods, McGraw-Hill, New York.
Misbra, S., J. C. Parker and N. Singhal. 1989;. Estimation of soil hydraulic properties and their
uncertainty from particle size distribution data, J. Hydrol., 108, 1-18.
Osteridorf, D. W., R. J. Richards and F. P. Peck. 1993. LNAPL Retention in Sandy Soil,
Ground Water, 31, 285-292. j
Parker, J. C., R. J. Lenhard and T. Ruppusamy. 1987. A parametric model for constitutive
properties governing multiphase flow in porous media, Water Resour. Res., 23, 618-624.
Parker, J. C., J. J. Kaluarachchi, V. J. Kremesec and E. L. Hocktnan. 1990. Modeling free
product recovery at hydrocarbon spill sites, Proc. Petrol. Hydrocarbons and Organic Chemicals
in Groundwater, NWWA. -I
Parker, J. C., J. J. Katyal, J. L. Zhu, V. J. Kremesec and E. L. Hockman. 1991. Free product
recovery at spill sites with fluctuating waterjtables, Proc. 5th Nat. Outdoor Action Conf,
NWWA. j
Parker, J. C., J. Zhu and H. White. 1992. Modeling phase seperated hydrocarbon migration and
recovery hi anisotropic fractured media, Proc. Petroleum Hydrocarbons in Groundwater
Conference, NGWA. j
I
Parker, Jack C., A. K. Katyal, J. L. Zhu, V. J. 'Kremesec and E. L. Hockmann. 1994. Modeling
free product migration and recovery at hydrocarbon spill sites, Ground Water, 32, 119-128.
van Dam, J. 1967. The migration of hydrocarbons in a water-bearing stratum, In: The Joint
Problems of the Oil and Water Industries, P.: Hepple, ed. Inst. Petrol., London, 55-96.
van Genuchten, M. Th. 1980. A closed-form equation for predicting the hydraulic conductivity of
unsaturated soils, Soil Sci. Soc. Amer. J., 44, 892-898.
F
Weaver, J.W. and J.A. Johnson. 1993. Modeling for assessment of remediation of hydrocarbon
spills, 3rd Annual Symposium on Groundwater and Soil Remediation, Quebec City, Quebec.
Wilson, J. L. 1992. Pore scale behavior of spreading and non-spreading organic liquids in the
vadose zone, Subsurface Contamination By Immiscible Fluids. U. Weye,. Ed.: 107-114,
Rotterdam: Balkema. :
Zhu, J. L., J. C. Parker, A. K. Katyal, V. J. Kremesec, E. L. Hockmann and M. N. Gallagher.
1991. Effects of delays in recovery system startup on product recovery at hydrocarbon spill
sites, Proc. Conference on Hydrocarbons in Groundwater, NWWA.
81
-------
-------
------- |