APPLICATION OF A GEOGRAPHIC
INFORMATION SYSTEM FOR CONTAINMENT
SYSTEM LEAK DETECTION
By
Randall R. Ross1, Milovan S. Beljin2 and Baxter E. Vieux3
ABSTRACT
The use of physical and hydraulic containment systems for the isolation of
contaminated ground water associated with hazardous waste sites has increased during the
last decade. Existing methodologies for monitoring and evaluating leakage from hazardous
waste containment systems rely primarily on limited hydraulic head data. The number of
hydraulic head monitoring points available at most sites employing physical containment
systems may be insufficient to identify significant leakage. A general approach for evaluating
the performance of containment systems based on estimations of apparent leakage rates is
used to introduce a methodology for determining the number of monitoring points necessary
to identify the hydraulic signature of leakage from a containment system. The probabilistic
method is based on the principles of geometric probability. A raster-based GIS (TDRISI) was
''Hydrologist, U.S. EPA Nat. Risk Mgmt. Res. Lab, Subsurface Protection and Remediation Div.,
P.O. Box 1198, Ada, OK 74820
:Consulting Hydrologist, M.S. Beljin & Associates, 9416 Shadyoak Court, Cincinnati, OH 45231
3Assoc. Prof., School of Civil Eng. and Env. ScL, Univ. of Oklahoma, 202 West Boyd Street,
Norman, OK 73019
-------
used to determine the critical diminsions of the hydraulic signature of leakage from a
containment system, as simulated under a variety of hydrageologie conditions using a three-
dimensional ground-water flow model MQDR1ST, a set of computer programs was used to
integrate ground-water flow modeling results into the hydraulic signature assessment method
DTTRODUCnOIV
Subsurface vertical barriers have been used to control ground-water seepage in the
construction industry for many years. Recently, the industrial and regulatory communities
have applied vertical barrier containment technologies as supplemental or stand-alone
remedial alternatives at hazardous waste sites to prevent or reduce the impact of contaminants
on ground-water resources (Rurner and Ryan, 1995) While subsurface barriers appear to be
useful for isolating long-term sources of ground-water contamination at many sites, the
potential exists for leakage of contaminants through relatively high hydraulic conductivity
zones ("windows") within the barriers
This paper describes the application of a Geographic Information System (G[S) as a
tool to help identity leakage through discrete zones within & subsurface vertical barrier The
proposed techniques could be useful for evaluating existing containment systems by providing
insight as to how many monitoring points are necessary to determme the approximate
locations of discrete leaks, given specified confidence and constraints.
-------
Containment Systems
Subsurface containment systems may be active (e.g., ground-water extraction to
manage hydraulic gradient), or passive (e.g., physical barriers only) depending on the remedial
objectives and complexity of the hydrogeologic setting (Canter and Knox, 1986), Frequently,
containment systems employ a combination of active and passive components, which
commonly incorporate vertical barriers keyed into underlying low-permeability units. Many
containment systems also include a low permeability cover to reduce the rainfall infiltration,
extraction and injection wells, and trenches for ground-water management.
Soil-bentonite slurry cutoff walls (slurry walls) are the most common type of
subsurface vertical barriers used at hazardous waste sites and are generally installed around
suspected source areas (U.S. EPA, 1984). Construction defects or post-construction
property changes are potential failure mechanisms of subsurface vertical barriers (Evans,
1991). Construction defects may result in the formation of relatively high hydraulic
conductivity "windows" in a barrier. Some of the mechanisms responsible for the formation
of such windows include emplacement of improperly mixed backfill materials, sloughing or
spalling of in situ soils from trench walls, and failure to excavate all in situ material when
keying wall to the underlying low permeability unit (U.S.EPA, 1987). Post-construction
property changes may result from wet-dry cycles due to water table fluctuations, freeze-thaw
degradation, or chemical incompatibility between the slurry wall material and ground-water
contaminants.
-------
Monitoring of Containment Systems
The performance of hazardous waste containment systems has generally been
evaluated based on construction specifications. Most subsurface vertical barriers are required
to maintain a hydraulic conductivity of IxlO"7 cm/s, or less. The use of appropriate
construction quality assurance (QA) and quality control (QC) testing during installation is
essential to ensure that the design performance specifications are achieved. The regulatory
community recognized the need to develop procedures to verify post-construction
performance and identify unsatisfactory zones in containment systems (U.S.EPA, 1987).
While construction dewatering systems are deemed successful if the barriers limit ground-
water leakage to reasonably extracted quantities, there are no uniform methods to reliably
measure and document the hydrologic performance of existing and proposed hazardous waste
containment systems (Grube, 1992).
The minimum number of monitoring points necessary to determine whether a
containment system is functioning as designed depends on site-specific conditions. For
example, in some cases it may be possible to determine whether leakage has occurred by
analyzing the water level trends in monitoring wells (Ross and Beljin, 1998). Subtle
variations in the hydraulic head distribution associated with leakage through a subsurface
barrier may be identifiable if sufficient hydraulic head data are available for analysis. Such an
undertaking would generally be considered prohibitively expensive due to the high cost of
installing a piezometer network capable of adequately defining the hydraulic head distribution.
However, the recent development of relatively inexpensive installation techniques may make
-------
it feasible to install a sufficient number of small diameter piezometers to identify the hydraulic
signatures associated with containment system leakage.
A New Monitoring Method
The process of locating a leak in a hazardous waste containment system can be
analogous to mineralogical prospecting where a compromise is sought between the cost of
exploration and the thoroughness of the search. For mineral exploration applications, the
expected benefit of a search is the sum of the value of each target multiplied by the probability
of finding h, assuming that the target exists in the search area (Singer, 1972). For containment
system leak detection, the expected benefit of a search is the potential reduction in risk to
human health and the environment associated with the detection and abatement of significant
leaks.
Gilbert (1987) presents a methodology based on the work of Savinskii (1965), Singer
and Wickman (1969), and Singer (1972) that can be used to determine the grid spacing
required to detect highly contaminated local areas or hot spots at a given level of confidence,
or estimate the probability of finding a hot spot of specified dimensions, given a specified grid
spacing. Given a specific grid spacing, the probability of detecting a target is determined by
the method of geometric probability, which is a function of the ratio of the area of the target
to the area of the grid cell. The method assumes that the highly contaminated areas are
circular or elliptical in shape, the boundaries of the hot spot are clearly identifiable based on
contamination levels, hot spot orientation is random with respect to the sampling grid, and
the distance between grid points is much larger than the area sampled. In order to address
-------
variations in the distribution of hydraulic head, rather than contaminant concentrations, the
assumptions were modified for the methodology presented in the paper.
METHODOLOGY
The hydraulic signature associated with leakage from a containment system is
simulated using a numerical model for a variety of hydrogeological settings. The modeling
results provide the data on which the hydraulic signature assessment method is demonstrated.
A set of computer programs was developed (Ross and Beljin, 1995) to import modeling data
into a raster-based GIS, for further processing. The GIS was used to generate the input data
for the ground-water model.
Ground-Water Modeling
A model may be defined as a simplified version of a real system that approximates the
stimulus-response relationships of that system (Bear and others, 1992). By definition, the use
of a model requires the application of simplifying assumptions to describe the pertinent
features, conditions, and significant processes that control how the system reacts to stimuli.
In this study, one of the primary objectives of the modeling was to predict the hydraulic head
distribution associated with leakage through discrete leaks in a vertical barrier under different
hydrogeologic conditions.
The conceptual model presented in this paper is based on characteristics of several
specific hazardous waste sites that incorporate physical containment as a major component
of the remedy. The sites which influenced the development of the model used in this study
6
-------
include the Gilson Road Superfund site (Nashua, New Hampshire), the G.E, Superfund site
(Moreau, New York), and the Velsicol/Michigan Chemical Company Superfund site (St.
Louis, Michigan). The conceptual model for the containment system consists of a slurry wall
fully penetrating an unconsolidated surficial aquifer, keyed in to an underlying low
permeability aquitard (Fig. 1).
Hydraulic head values are assumed to be higher in the interior of the containment
system, simulating a "worst-case" scenario for potential contaminant losses from the system
(Fig. 1). The elevated water levels within the conceptual containment system are assumed
to be derived from deficiencies in the the system (i.e., leakage under or through the
upgradient wall and infiltration through the cap), and water levels are assumed to be relatively
stable over time. Ground-water flow is assumed to be horizontal, except in the immediate
vicinity of the vertical barrier. Given the long-term nature of most hazardous waste
containment systems, the hydraulic heads are averaged over long time periods. Consequently,
steady-state flow conditions are assumed for all simulations used in this study.
The hydraulic head distribution associated with a linear segment of a conceptual
vertical barrier was simulated using Visual MODFLOW* (Guiger and Franz, 1995), a
commercial version of the three-dimensional, finite difference ground-water flow model
MODFLOW, developed by the U.S. Geological Survey (McDonald and Harbuagh, 1988).
Data Processing with a GIS
The hydraulic head data generated by the numerical simulations are extracted,
visualized, sampled, analyzed, and appropriately manipulated using several software packages.
-------
Idealized ConHanmerrt System
Undsrabte Conditions
n , — 5 1 n » n
1
Aquifer A
S7
=
=
—
\
Cap 1
\
,- Slurry Wall
r
7
1
•<
)
i
^-
Aquifer B
H= Hydraulic Head In
HM = Hydraulic Head h
H, = Hydraulic Head In
Q,,«U*qeOuaf<
Q^ = Leakage Into Com
side Containment System
i Adjacent Aquifer A
Underlying Aquifer B
jontatnment System
tainmem system via up
i
=
Fig. 1. Major components of an idealized hazardous waste containment system exhibitin|
unfavorable conditions (e.g., outward hydraulic gradient).
-------
Hydraulic head data from a vertical cross-section parallel to, and immediately down gradient
from the simulated vertical barrier are used throughout this study. The data are extracted
from MODFLOW output files and reformatted as image files for analysis using MODRJSI
(Ross and Beljin, 1995). The GIS software used in this study is IDRISI (Eastman, 1995), a
raster GIS that provides numerous analytical capabilities that are directly applicable to this,
and other hydrogeologic studies. The uniform grid spacing facilitates the transfer of data
from one software package to another. The raster format allows import and export of
uniform grid model data and also provides a robust platform for the analysis, visualization and
data manipulation,
Model Setup
The model domain consists of 51 rows, 51 columns, and 25 layers (Fig. 2) and is
discretized into uniform 1 m3 blocks. This configuration is sufficiently large to reduce
boundary effects and provides sufficient resolution to allow identification of subtle variations
in hydraulic heads associated with leakage through a vertical barrier. The uniform grid size
allows consistent precision over the entire model domain and simplifies data management and
transfer between software packages.
The slurry wall is simulated as a one-meter thick barrier with uniform properties,
except for the window. The hydraulic conductivity values for the aquifer and window are
scenario dependent. Leakage through the wall is simulated as a window with dimensions of
2x3 cells (6 m2), located in the approximate center of the vertical barrier (row 25, columns
24-26, layers 12 and 13).
-------
No-Row Boundary
Constant-Head Boundary
Vertical Barrier (row 25)
Columns (J) = 51
Ground-Water flow
Direction
X
I Layers (K) = 25
1 i v, ~ \ m
= 1 m
fiows (i) = 51
No-flow Boundary
r,= 1 m
Constant-Head Mjndary
Fig, 2. Conceptual model domain and boundary conditions.
10
-------
Boundary conditions are depicted in Fig. 2. The upgradient and downgradient sides of the
model are constant-head boundaries, resulting in a horizontal hydraulic gradient across the
model domain of 0.0196 m/m. This value falls within the range of hydraulic gradients
commonly observed in the field. The sides and lower surface of the model oriented parallel
to ground-water flow are simulated as no-flow boundaries.
The applicability of the numerical model for simulating the hydraulic head distribution
associated with leakage from a containment system was demonstrated by comparing model
results to data generated from a laboratory bench scale model of a cutoff wall (Ling, 1995).
Simulation results agreed favorably with the physical model results, indicating that the
approach described in this study is appropriate for simulating the hydraulic head distribution
associated with leaking vertical barriers.
General Simulation Scenarios
Several hypothetical hydrogeologic conditions are evaluated in this study. Different
scenarios are used to better understand the potential variability of the hydraulic signatures
associated with different subsurface conditions and to account for potential uncertainties
associated with predictive modeling.
A range of homogeneous and isotropic conditions were simulated in an effort to
provide a reference case for evaluating the effects of varying average aquifer hydraulic
conductivity values on the hydraulic signature of a simulated leak. The scenarios spanned a
wide range of hydraulic conductivity values with respect to the aquifer material and zone of
leakage. The hydraulic conductivity values for the aquifer range from 1 x 10"2 cm/s to 1 x
11
-------
10"5 cm/s. The hydraulic conductivity of the vertical barrier is maintained throughout the
study at 1 x 1CP7 cm/s. The hydraulic conductivity values for the window ranged from 1 x
10"2 cm/s to 1 x 10"5 cm/s. The hydraulic conductivity value for the window is assumed to
be less than or equal to that of the adjacent aquifer materials. The scenarios simulate the
general effects of layering by varying the horizontal to vertical hydraulic conductivity ratios
of aquifer materials.
One of the primary limitations of using ground-water flow models as a predictive tool
results from the uncertainty associated with input parameters. This uncertainty is directly
related to the spatial variability of hydrogeologic properties of the porous medium (i.e.,
aquifer material). To account for some of the spatial variability and uncertainties associated
with three-dimensional predictive flow modeling, several scenarios utilizing heterogeneous
distributions of hydraulic conductivity were assessed. The assumption of lognormally
distributed hydraulic conductivity is used for the heterogeneous, isotropic and heterogeneous,
anisotropic simulations. Unique lognormal hydraulic conductivity distributions were
generated for each of the 25 layers using built-in functions of the GIS software. This
approach resulted in the generation of approximately 63,000 hydraulic conductivity values
within the model domain.
Hydraulic Signature Assessment Method
The methodology used to address the hydraulic head distribution associated with
leakage from a containment system was developed based on the work of Singer and Wickman
(1969) and Gilbert (1987), The proposed method is directly applicable to determining the
12
-------
grid spacing necessary to detect the hydraulic signature associated with a discrete leak in a
subsurface vertical barrier. The methodology requires the following assumptions:
_ the hydraulic signature of the leak is circular or elliptical;
_ hydraulic head data are acquired on a square grid;
_ the criteria delineating the hydraulic signature are defined, and
_ there are no measurement misclassification errors.
The model results indicate that the hydraulic signatures associated with the simulated
leaks range in shape from approximately circular to elliptical when viewed in vertical cross-
section. An increase in the anisotropy results in the elongation of the signatures in the
horizontal directions. As expected, the greater the anisotropy, the more elliptical the hydraulic
signature of the leak.
The criteria for delineating the hydraulic signature of a leak from background noise
are based on the average hydraulic head value (xV) of the model cross-sectional surface. For
this study, hydraulic head values of Xh+0.05 m and XK+O, 1 m were identified as critical values
(CX), indicating the presence of a hydraulic anomaly associated with containment system
leakage. This follows the assumption that any background noise associated with the hydraulic
head measurements is significantly less than 0.05 m. The dimensions of the hydraulic
anomalies are determined using GIS software by image reclassification to delineate nodes
exceeding the average hydraulic head by the specified critical values. The dimensions of the
hydraulic signatures delineated by the two values for Cv are expressed as shape factors (S),
defined as the ratio of the length short axis to the length of the long axis of the hydraulic
13
-------
signature. The shape factor for a circular feature is 1. An increase in anisotropy results in the
elongation of the feature and a decrease in S, where 0 < S < 1.
The probability tables of Singer and Wickman (1969), were used to determine the
probability of not detecting a leak when a leak is present (P) to the ratio of the semi-major
axis to grid size (L/G). The semi-major axis is defined as one half the length of the long axis
of an elliptical feature. The general procedure for determining monitoring point spacing
necessary to detect a hydraulic anomaly of given dimensions and specified confidence is
outlined in Table 1, and in the following example.
Table 1. General steps for determining monitoring point grid spacing.
Specify the radius or one half the length of the long semi-major axis (L)
of the .hydraulic signature (mound) associated with the leak;
2.
3.
4.
Assuming a circular hydraulic signature, let the shape factor (S) equal
one; for elliptical features, S may be calculated using equation (9);
Specify the maximum acceptable probability (|3) of not detecting the
hydraulic feature (=0.1);
Knowing L, S and assuming a value for J3, determine L/G from Fig. 4,
and solve for G (minimum grid spacing required to detect the hydraulic
anomaly associated with the leak, given the specified constraints).
In order to determine the minimum grid spacing necessary to identify a hydraulic
feature of specified dimensions, an acceptable probability of not detecting the feature must
be established. For this example, a value of p= 0.1 is assumed for a leak signature with
14
-------
dimensions of 5 m by 4 m, as delineated by Cv = 0,1 in Fig. 3 a. From Fig. 4, a value of
approximately 0,64 is indicated for the ratio of the length of the semi-major axis to grid size
(L/G), given = 0.1 and S = 0.8. Therefore, solving for G using L = 2.5, it is determined that
a minimum grid spacing of approximately 3.9 m is necessary to identify the specified feature
with a 90% probability of success. The resulting grid spacing (G) may be used to determine
the minimum number of block-centered monitoring points required to detect the feature for
a specified area by dividing the total area by the area of one square grid (G2).
The probability tables were also used to generate nomographs relating the probability
of not detecting a leak (P) of specified dimensions (L), for different grid dimensions (G)
Figure 5 illustrates this relationship for circular hydraulic signature (S = 1.0). The
nomographs may be used to estimate the dimensions of the smallest hydraulic signature
capable of being identified by a monitoring network of known dimensions within an
acceptable level of confidence (P), For example, given a monitoring point spacing of 20 m,
what is the smallest circular hydraulic anomaly that can be detected with 80% probability of
success (p= 0.2). From Fig. 5 it is noted that a circular feature with a radius of approximately
10.1 m can be detected with the specified probability and grid spacing. The probability of not
detecting the anomaly will increase as the radius of the hydraulic signature decreases.
RESULTS AND DISCUSSION
The dimensions of the hydraulic signatures associated with leakage through a
subsurface vertical barrier are a function of the hydrogeologic properties of the aquifer,
vertical barrier, and zone of leakage. Assuming all other variables remain constant, the
15
-------
00
o. w
C ff.
(P O
i-3
I ?
n> (S
3 O
o »
§ o
3 >-•»
s
c g-
O 21
09.3
o S-
M Ej'
STtS
o
W5
f
B
Homogeneous
Conditions
Kwin=lB-2cnVs
Heterogeneous
Conditions
Kwin=lE-2cnVs
Homogeneous
Conditions
KwinplB-3cnVs
Homogeneous
Conditions
Kwin= 1B-4 cm/s
2.40000E+01
2.40143E+01
2.40286E+01
2.40429E+01
2.40571E+01
2.40714E+01
2.40857E+01
2.41000E+01
2.41143E+01
2.41286E+01
2.41429E+01
2.41571E+01
2.41714E+01
2.41857E+01
d] >2.42000E+01
Hydraulic Head Values
OH
g-
t
X
Isotropic Conditions
Kh=Kv
Anisotropic Conditions
Kh:Kv=10
Anisotropic Conditions
Kh:Kv=100
D-
-------
0,5 1 1.5
Ratio of Semi-Major Axis to Grid Size (L/G)
Fig. 4. Nomograph relating ratio of semi-major axis of elliptical target and grid size to the
probability of missing the target (Beta) for different shape factors using a square grid
pattern.
G=10 G=15 \G=20 \G=25
6 8 10 12
L = Radius of Circular Target (m)
16
18
Fig. 5. Nomograph relating radius of circular hydraulic signature to probability of not
detecting leak (Beta) for different grid spacings.
17
-------
magnitude of the hydraulic signature diminishes significantly as the hydraulic conductivity
of the window decreases (Fig. 3), The hydraulic signature of leakage through the hydraulic
conductivity window becomes less prominent as its value is reduced by one order of
magnitute (Fig. 3g). As the value is further reduced, the hydraulic signature becomes
discemable only immediately adjacent to the window (Fig. 3j), The decrease in hydraulic
signature corresponds to a decrease in flux through the window, as the window hydraulic
conductivity is reduced (Table 2),
Table 2. Simulated flux through windows of varying hydraulic conductivity.
Window
Hydraulic
Conductivit
y (cm/s)
lxlO'2
IxlO"3
lxlO'4
IxlO"5
Minimum
Head
Value (m)
24.0293
24.0117
24.0071
24.0063
Maximum
Head
Value (m)
24,2627
24.0826
24.0165
24.008
Range
(m)
0.2334
0.0709
0.0094
0.0017
Flux
Through
Window
(m3/d)
1.31101
3.98
4.961Q-1
5.0910'2
The effect of varying the horizontal to vertical hydraulic conductivity values is
illustrated in Fig. 3. For example, the hydraulic signature from leakage through a window
under homogeneous and isotropic conditions forms an approximately circular feature (Fig,
3a). However, as the horizontal to vertical hydraulic conductivity ratio increases, the
hydraulic signature of the leak becomes more elliptical (Fig. 3b,c). Similar trends are
18
-------
observed with respect to increasing the horizontal to vertical hydraulic conductivity ratio for
the heterogeneous simulations (Fig. 3d,e,f) and other homogeneous simulations with smaller
hydraulic conductivity values for the windows (Fig, 3g-l).
The method was applied to different hydraulic signatures developed from ground-
water flow simulations of leakage through a vertical barrier. The criteria used to differentiate
the hydraulic signature of leakage from background noise are Cv = xh+0.05 m and xh+0.1 m,
Figure 6a depicts the head distribution associated with hydraulic signature of leakage through
a window located in the approximate center of a vertical barrier in a homogeneous, isotropic
aquifer. The approximate dimensions of the vertical hydraulic mound as defined by
Cv=xh+0.05 and xh+0.1 are 7 m by 6 m, and 5 m by 4 m, respectively.
An increase in the anisotropy of the simulated aquifer by one order of magnitude
produces a vertically compressed and horizontally elongated hydraulic signature (Fig 6b).
Similarly, increasing the anisotropy of the simulated aquifer by two orders of magnitude
results in even greater compression and elongation of the hydraulic signature in the vertical
and horizontal directions, respectively (Fig. 6c).
Hydraulic signatures for leakage through a window with a hydraulic conductivity
value of IxlO"3 cm/s exhibit similar trends in response to increases in anisotropy (Fig. 7a,b,c).
However, the overall hydraulic signature of the window is decreased significantly relative to
that of the base case. This results in a lack of head values greater than the elevation threshold
for Cv=>*+0,l for the homogeneous, isotropic simulations. The hydraulic head values
associated with leakage through windows with hydraulic conductivities < 1x10° cm/s were
all less than Cv^xh+0.05, and therefore, could not be evaluated as described above.
19
-------
a. Homogeneous and feotropic Simulation Results
b. Homogeneous and Amsotropic Results (KhK¥=10)
CD
a
2.4QQOOE+Q1
2.40143E+01
Z40286E-KJ1
2.40429E+Q1
2.40571 E-H31
2.40714E+Q1
2.40857E+01
2.41000E+01
2.41143E+01
2.41286E+01
Z41429E+01
2.41571E-KJ1
2.41714E-MJ1
2.41857E+01
Hydraulic Head Values
X
c. Homogeneous and Anisotropic Results (KhKv=100)
Fig. 6. Vertical cross-section of model results illustrating variations in hydraulic head
values due to changes in anisotropy (Kaq=l x 10"2 cm/s, Kwin=l x 10'2 cm/s). The ellipses
define the approximate boundaries of the hydraulic features defined by specified critical
values(Cv^x+O.l and x+0,05).
20
-------
a. Homogeneous and fcotropic Simulation Results
b. Homogeneous and Amsotropic Results (KhKv=10)
2.40000E+Q1
2,40143E+01
2.40286E+D1
2.4Q429E+01
2.40571 E+01
2,40714E+Q1
240857E+01
2.41COOE+01
2.41143E+01
2.41286E+Q1
2.41429E+-01
2.41571E+Q1
Z41714E*01
I I 2.41857E^D1
I I >2.4200DE+01
a
Hydraulic Head Values
c. Homogeneous and Amsotropic Results (Kh"Kv=100)
Fig. 7. Vertical cross-section of model results illustrating variations in hydraulic head
values due to changes in anisotropy (Kaq=l v. 10"2 cm/s, Kwin^l x 10"3 cm/s). The ellipses
define the approximate boundaries of the hydraulic features defined by specified critical
values (Cv=x+0.1 and x+0.05).
21
-------
The grid sizes necessary to identify the hydraulic features described above with a 90%
probability of success ((3=0.1) were obtained using the nomograph in Fig. 4. The number of
sampling points (Ns) necessary to identify the hydraulic features within the domain of the
model cross-section is determined by dividing the cross-sectional area of the model (1,275
m2) by the area of one square grid spacing (G2). The results are listed in Table 3.
The number of monitoring points required to identify the hydraulic signatures of the
simulated leaks using the prescribed constraints and confidence ranges from approximately
40 to over 300. The wide range of values is a function of the variability in the size and shape
of the hydraulic features. This variability results from the use of different critical values to
define the hydraulic signatures of the leaks and the wide range of shape factors resulting from
the three orders of magnitude range of the anisotropy values.
CONCLUSIONS
Numerical modeling of ground-water flow through high hydraulic conductivity
windows in subsurface vertical barriers was conducted to provide data sets for use with a
probabilistic method for determining the grid spacing necessary to identify the hydraulic
signature associated with the leaks. The proposed method of combined ground-water
modeling and GIS represents a potential tool that may be used by the regulatory community
and others to evaluate the adequacy of existing and proposed hazardous waste containment
systems for identifying containment system leakage. The utility of the proposed method is
22
-------
Table 3. Parameters and Results Obtained from Hydraulic Assessment Method.
Kwin
(cm/s)
IxlO-2
IxlO'2
IxlO*2
lxlO'2
IxlO-2
IxlO"2
lxl(T3
IxlO-3
IxlO"3
IxlO'3
IxlO'3
IxlO'3
IxlO'2*
IxlO'2*
lx!0"2*
IxlO"2*
IxlO'2*
IxlO'2*
Kh:Kv
1
1
10
10
100
100
1
1
10
10
100
100
1
1
10
10
100
100
Cv
0,1
0.05
0,1
0.05
0.1
0.05
0,1
0.05
0.1
0.05
0.1
0.05
0.1
0.05
0.1
0.01
0.1
0.05
S
0.8
0.85
0.28
0.31
0.13
0.16
BCL
0.67
0.67
0.4
0.4
0.15
0.8
0.85
0.28
0.31
0.13
0.16
L
2.5
3.5
3.5
6.5
7,5
12.5
-
1.5
1.5
2.5
2.5
6.5
2.5
3.5
3.5
6,5
7.5
12.5
L/G
0.64
0.62
1.64
1.51
3.5
2.9
_
0.74
0.74
1.17
1,17
3.05
0.64
0.62
1.64
1,51
3.5
2.9
G
3.91
5.65
2.13
4.3
2.14
4.3
_
2.03
2.03
2.14
2.14
2.13
3.91
5.65
2.13
4.3
2.14
4.31
N.
84
40
280
69
278
69
_
311
311
280
280
281
84
40
280
69
278
69
BCL = All head values below critical value threshold.
""Heterogeneous simulations; all other simulations homogeneous
23
-------
demonstrated using simulated data. Based on the application of the method presented, the
following conclusions were made:
• The number of points necessary to identify the hydraulic signature of a discrete leak
within prescribed constraints is a function of the criteria used to delineate the feature;
• The hydraulic signature associated with a minor leak in a vertical barrier may be
difficult to detect with a realistic number of monitoring points;
• By using the nomographs described above, the probability of failing to detect the
hydraulic signature of a leak can be estimated for a given monitoring well spacing and
specified confidence;
• The dimensions of the smallest hydraulic signature detectable with a given monitoring
point spacing can be estimated, given the appropriate constraints and specified
confidence;
• The monitoring point spacing used at many hazardous waste sites is likely inadequate
to detect the hydraulic signatures of all but the largest leaks, and
• The method for delineating the hydraulic signature of a leak using the average
hydraulic head plus specified values does not appear to be as sensitive to the
heterogeneity of the aquifer as it is to anisotropy.
24
-------
Disclaimer
The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded and managed the research described here through Inhouse efforts. This information
has not been subjected to the Agency's peer or administrative review and therefore does not
necessarily reflect the views of the Agency; no official endorsement should be inferred.
Mention of trade names or commercial products does not constitute endorsement of
recommendation for use.
25
-------
References
Bear, J., Beljin, M.S., and Ross, R., (1992). Fundamental of Ground-Water Modeling,
Ground Water Issue, U.S. Environmental Protection Agency, Office of Research and
Development, R.S. Kerr Environmental Research Laboratory, Ada, OK, EPA/540/S-92/005.
Canter, L.W., and Knox, R.C, (1986). Ground Water Pollution Control, Lewis Publishers,
Boca Raton, FL.
D'Appolonia, DJ. (1980). "Soil-bentonite slurry trench cutoff," Journal of Geotechnical
Engineering. ASCE. Vol. 106, no. 4, pp.399-417.
Eastman, J.R. (1995). IDRISIfor Windows: User's Guide, Version 1.0, Clark University,
Worchester, MA,
Evans, J.C, (1991).,"Geotechnics of Hazardous Waste Control Systems," Chapter 20,
Foundation Engineering Handbook, 2nd ed,, H.Y. Fang, ed., Van Nostrad-Reinhold
Company, New York.
Gilbert, R.O. (1987). Statistical Methods for Environmental Pollution Monitoring, Van
Nostrand Reinhold, New York, N.Y.
26
-------
Grube, W.E., Jr. (1992). "Slurry Trench Cut-Off Walls for Environmental Pollution Control,"
Slurry Walls: Design. Construction and Quality Control. ASTM STP 1129, David B. Paul,
Richard R. Davidson, and Nicholas J. Cavalli, Eds,, American Society for Testing and
Materials, Philadelphia.
Guiger, N., and Franz, T. (1995), "VISUAL MODFLOW, The Integrated Modeling
Environment for MODFLOW and MODPATH, Version 1.1," Waterloo Hydrogeologic,
Ontario, Canada.
Ling, K., (1995b). Windows Development and Detection in Soil-Bentonite Cutoff Walls,
Ph.D. Dissertation, University of Cincinnati.
McDonald, M.G., and Harbuagh, A.W. (1988). "A Modular Three-Dimensional Finite-
Difference Ground-Water Flow Model (MODFLOW)," U.S. Geological Survey Techniques
of Water-Resources Investigation, Book 6, Chapter Al.
Ross, R.R., and Beljin, M.S. (1995). "MODRISI: A PC Approach to GIS and Ground-Water
Modeling," Proceedings. National Conference on Environmental Problem-Solving with
Geographic Information Systems. Cincinnati, OH, September 21-23, 1994. EPA/625/R-
95/004.
27
-------
Ross, R.R., and Beljin, M.S. (1998), "Evaluation of Containment Systems Using Hydraulic
Head Data," J. Envir. Eng.. 124(6), 575-578.
Rumer, R.R., and Ryan, M.E., eds, (1995). Barrier Containment Technologies For
Environmental Remediation Applications^ John Wiley & Sons, Inc., New York.
•«j
Savinskii, J.D. (1965). "Probability Tables for Locating Elliptical Underground Masses with
a Rectangular Grid," Consultants Bureau, New York, pp. 110
Singer, D.A., (1972). "ELIPGRID, a Fortran IV program for calculating the probability of
success in locating elliptical targets with square, rectangular and hexagonal grids," Geocon
Programs, Vol. 4, No. 1, p. 1-16.
Singer, D.A., and Wickman, F.E. (1969), Probability Tables for Locating Elliptical Targets
with Square, Rectangular and Hexagonal Point-Nets," Mineral Sciences Experiment Station
Special Publication 1-69, Penn. State University, University Park Pennsylvania, 100 p
U.S. EPA (1987). "Construction Quality Control and Post-Construction Performance
Verification for the Gilson Road Hazardous Waste Site Cutoff Wall," Hazardous Waste
Engineering Research Laboratory, Office of Research and Development, Cincinnati, OH,
EPA/600/2-87/065
28
-------
MBMHi AnA aa-n-ia TECHNICAL RgPQflT DATA
flWM*Ai-~fUW 3* J o fpfiest rtwl InicntctiBrti on fftt fwfnf tftfaf* C ffmpk t jnf
i, Be?tsrtT N-SS, Ta
KPA/ftOO/A-Ol /O29 1
Application of a Gaographic infomation Sysnem Cor
Corstaureieiiit System Leak Detection
fiatra3ali R. Ross1, Miiovan S. Beljia2, and Baxter E. Viaux3
'U.S. EPA., ORD, WRMRL, SPFJ3, P.O. Bex 1199, Ada, OK 74S21
^Beljin S Associatda, 9416 Shadyoak Ctjurt, Cincinnati,
Ohio 45331
3 School of Civil Engineering & Enviromnentai Scisnce,
Iftliversit^ pf pfol^fiopra . Norman- rtK TIOTiJ
13, SrONSOntNG ACCNCV MAMlf. ANQ AOO«6SS
U.S. Eivircntrasntal Prot action toeciqy
Office of pe search anci Developnint
National Risk MarageniraTt lesearcri Labaratoiy
SulDBurface Protection S Rfimediatiori Division '
P.O. Box 1193, Ada, Oklahoma 74321
*.*«
s,^i?oflTOArE
*. ^n^RM[NC<»A<;ANt2ArroN«i»0£
a. *fiH?c«M«Nq. {jn^AMiZA'iQf BifORT MC,
TEKY1A
In-House
ia, TYPS OP fl£po«T AWO ^qflico cove^ec
Confersnce Parser
™A««5
Project Officsr: Randall R. Ross 580-436-3511
The use of physical and hydraulic containment systems for the isolation of contaminated
ground water associated with hazardous waste sites has increased during the last decade.
Existing methodologies for monitoring and evaluating leakage from hazardous waste
containment systems rely primarily on limited hydraulic head data. The number of hydraulic
head monitoring points available a! most sites employing physical containment systems may be
insufficient to identify significant leakage. A general approach for evaluating the performance of
containment systems baaed1 on estimations of apparent leakage rates is used to introduce a
tnethodotogy for determining the number of monitoring points necessary to identify the
hydraulic signature of leakage from a containment system. The probabilistic method ia based on
the principles of geometric probability, A raster-based CIS (JDRJSI) was used Co determine the •
critics! dimensions of the hydraulic signature of leakage from a containment system, as simulated
under a variety of hydrogeologic conditions osing a three-dimensional ground-water flow model
MODRISI, a set of computer programs was used to integrate ground-water flow modeling results
into the hydraulic signature assessmem method.
IT.
S!*MS
S-TATSJ,
RKLEASS TD THE PUBLIC
31 WO.
CWCLASSUIH)
i ry fit,ASS fritd
------- |