United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati, OH 45268
Research and Development
EPA/600/S2-91/008 Aug. 1991
i&EPA Project Summary
Factors Controlling Minimum
Soil Liner Thickness
David C. Anderson, Mark J. Lupo, James A. Rehage, Joseph O. Sai,
Ronald L. Shiver, Robert C. Speake, K.W. Brown, and D. Daniel
This report describes a three-part
study to gather Information on liquid
flow through soil liners that are incor-
porated into double-liner systems used
in hazardous waste disposal facilities.
In the first part of the study, a model
was developed to simulate flow occur-
ring through discreet channels in lifts
(a layer of compacted soil) and in the
horizontal layer between lifts. The
model indicated that high overall field
hydraulic conductivity values may re-
sult from excessive horizontal flow be-
tween lifts. In contrast, the model
showed that even relatively high hy-
draulic conductivity lifts can be used
to construct low conductivity soil liners
if horizontal flow between lifts can be
sufficiently reduced.
In the second part of the study,
laboratory tests using large 60-cm-di-
ameter permeameters showed that the
conductivity to water typically increased
by one order of magnitude with depth
in a 23-cm-thick lift of compacted clay.
Clod sizes ranging from 2.5 to 7.5 cm
had little influence on the hydraulic
conductivity. In addition, it was shown
that exposure of the compacted soil to
the atmosphere for as little as 24 hr
resulted in severe cracking and asso-
ciated high conductivities resulting
from flow through the desiccation
cracks. The data show that bulk density
was a poor predictor of the conductiv-
ity of a compacted soil. Dye patterns in
the permeameters also indicated flow
through preferential channels and
interclod spaces.
In the third part, field studies of a 3-
lift liner revealed that horizontal flow
does indeed occur at the interface be-
tween the lifts when channels penetrate
the overlying lift.
This Project Summary was developed
by EPA's Risk Reduction Engineering
Laboratory, Cincinnati, OH, to announce
key findings of the research project
that Is fully documented in a separate
report of the same title (see Project
Report ordering information at back).
Introduction
Compacted soil liners are a required
component of a double-liner system in
hazardous waste disposal facilities. The
compacted soil can be incorporated into
double-liner designs as the sole compo-
nent of a secondary soil liner or as the
lower component in a secondary compos-
ite liner.
The primary purpose of this project was
to initiate the data base required to deter-
mine the minimum thickness necessary
for a soil liner to meet the following per-
formance objectives:
1. Maintaining an in-place hydraulic
conductivity of less than or equal to 1 X
10'7 cm sec'1;
2. Retaining sufficient strength to sup-
port all potential overlying loads; and
3. Preventing the breakthrough of any
contaminant before the end of the post-
closure care period.
A literature review concentrated on the
factors that influence hydraulic conductiv-
ity and strength in compacted soils. A
computer model was developed that ex-
Printed on Recycled Paper
-------
aminod the influence of both the defects
that penetrate a lift and those that cause
horizontal flow between lifts. Laboratory
and field studies of liquid flow patterns in
a compacted soil were designed to exam-
ine both vertical and horizontal flow paths.
Computer Models
A useful model of liquid flow within and
through a soil liner must include flow within
each lift and flow between lifts (interlift
flow). Flow within a lift must encompass
both matrix flow and flow through channels
that short-circuit the liner matrix. Channel
flow may be caused by construction prac-
tices that result in liner defects such as
cracks, clods, channels, continuous
macropores, and incomplete bonding be-
tween lifts. Ignoring defects, the total per-
meability of a lift can be modeled as
that of a slab with a population of vertical
cylinders or pores, all of which have the
same radius and all of which are equally
spaced.
Actually, the pores are not equally
spaced, they do not have the same ra-
dius, and they are not perfectly oriented in
the vertical direction. In addition, they do
not have the same radius throughout their
length. This model is still a useful approxi-
mation to reality. Only through-going pores,
those with outlets to both the top and
bottom of a lift, are contributing to the
permeability in this model. Through-going
pores with radii significantly larger than
those of the bulk of the pores are referred
to as channels or defects.
Basse' on these assumptions, the con-
tribution to the hydraulic conductivity to a
lift made by the channels would be:
R4P9
1000 i,
8uD2
where R is the radius of the channels, D
is the spacing, p is the density of and u
the viscosity of ordinary water, and g is
the acceleration of gravity. The total per-
meability is caused by a combination of
the channels and small pores in the ma-
trix. The channel permeability is vastly
greater than the matrix permeability if the
channels have radii significantly greater
than those of the pores contributing to the
matrix permeability. Consequently, the
smaller pores can be ignored in a bi-
modal distribution of grain sizes.
Figure 1 shows hydraulic conductivity
(K) as a function of channel radius (R)
and spacing (I), the two most important
parameters in determining hydraulic con-
ductivity. If all of the conductivity of a lift is
due to cylindrical channels 10 u,m in ra-
dius, the channels would have to be an
100
•
700 750 200 250 300
: Channel Radius (Microns)
Figure 1. Spacing versus channel radius for KsHt5,Iff', Iff* and Iff* cm seer'.
350
average of 1.64 X 10-2 cm apart to main-
tain a conductivity of 10'5 cm sec-1. If the
channels were 2.5 cm in radius, they would
have to be 4.23 km apart. Figure 1 dem-
onstrates that only a few channels or de-
fects can lead to large conductivities in
compacted soil lifts. It is interesting to
note that if all of the conductivity in a liner
resulted from channels 850 |im in radius,
the channels would have to be 47.5 m
apart for the liner to meet the EPA stan-
dard of 10'7 cm sec-1. This would corre-
spond to less than one such flaw per 0.2
ha.
Channel density is important in inter-
preting laboratory and field hydraulic con-
ductivity measurements. If the con-
ductivity of a lift is controlled by a few
large channels and a small area is
sampled, it is likely that these channels
would be missed and the actual conduc-
tivity would be underestimated.
Suppose a lift has two populations of
through-going pores. One population has
a diameter of 1 mm (500 \im radius)
spaced 164.2 cm apart. The other popu-
lation has a radius of 1 urn spaced 0.203
mm apart. The first population will result
in a conductivity of 1 X 10-5 cm sec'1,
whereas the second population will result
in a conductivity of 1 X 10~8 cm see'1. The
resulting actual conductivity would be
1.001 X 10'5 cm sec'1. Suppose a sample
of the soil was taken with a standard 7.62-
cm-diameter shelby tube for laboratory
measurement. It would probably contain
106,000 through-going pores each of
whose radius is 1 u.m. But since there is
-------
only one channel of the millimeter popula-
tion per 2.7 M2, there is only a 0.17%
chance of such a flaw being present in
the 7.62-cm-diameter sample. Thus, the
analyst will measure a conductivity of 1 X
10-8 cm sec'1, having totally missed the
channels that contribute 99.9% of the con-
ductivity. The chance of detecting one of
the larger pores would nof be greatly im-
proved by taking several samples. With a
1.5 X 1.5 m infiltrometer, however, there
would be an 86% chance of finding one of
the larger pores.
Thus, if the conductivity of a lift is the
result of only closely spaced small pores,
a small sample will suffice. If it is caused
by a few large pores, however, a very
large sample is needed. Consequently, it
is easy to understand the magnitude of
discrepancies often found between labo-
ratory and field measurements of hydrau-
lic conductivity.
Compacted soil liners typically consist
of several lifts. Since a lift is generally
compacted to a 15-cm-thickness and since
a liner is typically 61 to 122 cm thick,
there should be four to eight lifts. If most
of the conductivity in each lift is caused by
a few widely spaced channels, then the
interlift flow becomes important. Interlift
flow is defined as the horizontal flow in
the plane between two lifts. If interlift flow
is restricted so that the continuous pores
in adjacent lifts cannot communicate, the
conductivity of a soil liner can be substan-
tially less than that of the individual lifts.
Two models for interlift flow follow. The
first is the "channel-centered model" (Fig-
ure 2). In this model, it is assumed that
there is a circular disk area in the interlift
plane at the bottom of every channel, cen-
tered on that channel. Inside the disk,
Figure 3. The pocket model for interlift flow.
flow is considered to be unrestricted. The
lateral distance from a channel at the bot-
tom of a lift where liquid can flow freely is
designated the interlift flow radius (IR).
This radius defines the disk. In this model,
it is assumed that if any top of a channel
in the lower lift is within the IR of a chan-
nel in the upper lift, liquid will flow from
the channel in the upper lift into the chan-
nel in the lower lift. Further, if a channel in
the lower lift is not within the IR of channel
in the upper lift, then the channels will not
constitute a flow path (through-going chan-
nel) (Figure 2).
The second model is the "pocket model"
(Figure 3). This model assumes an
inhomogeneous interlift plane. The plane
is divided into disks with radii of IR. In this
case, the disks are not necessarily cen-
tered on channels. These disks are pock-
ets of favorable horizontal flow. If any
channels from an upper lift end in the
same pocket that a channel begins in a
lower lift, then liquid can flow from the
upper channel into the lower one. If a
channel does not have its opening in a
pocket or if it does not have its opening in
(Not a Path)
Flow Path
Figure 2. The channel-cen tered model for interlift flow
a pocket with another channel from an-
other lift, then that channel is not a flow
path.
A program was written for each of the
models. Variables for the models included:
1) the desired effective hydraulic conduc-
tivity, 2) the channel radius, 3) the IR, and
4) the number of lifts. The program then
computes the number of channels there
would have to be in a 1.5 X 1.5 m section
of each lift to attain the selected, conduc-
tivity. The channels were assigned posi-
tions randomly. In the channel-centered
model, the program then measures the
distance between defects in adjacent lifts.
If a defect was not matched up with the
interlift flow from a channel above it, it
was considered "dead" and did not figure
in computations in the next interlift plane.
In this way, a determination can be
made of how many effective flow paths
there are through a certain number of lifts
as a function of IR. Each computation is a
stochastic experiment, so the program
must be run many times and the results
tabulated. Furthermore, each run must be
carefully examined in the form of a tree
diagram to find the number of through-
going flow paths. Figures 4 and 5 show
examples in which the through-going flow
passes through one channel, but it may
exit at several points.
The "channel-centered model" was run
for two through eight lifts. The assumptions
for each lift were a hydraulic conductivity
of 1 X 10-5 cm sea' and a channel radius
of 250 u,m. Thus, the number of channels
required in the 1.5 X 1.5 m lift section is
13. The model also assumed the exist-
ence of one circular interlift flow area (IR).
For each set of lifts, many computer runs
were made, and the results averaged.
The results of the computer runs of the
channel-centered model are summarized
in Figure 6 for two to eight lifts. It is
apparent in Figure 6 that IR and the num-
ber of lifts have a great effect on the total
hydraulic conductivity of a soil liner. The
-------
First Lift
Second Lift
Third Lift
Fourth Lift
Fifth Lift
Sixth Lift
Seventh Lift
Eighth Lift
Figure 4. The difference' between the number of effective flow paths and the number of active
channels In the last lift Circles represent channels and linos represent Interconnec-
tion. Notice the constriction at the fourth lift Although there are four active defects In
tho last lift, the number of effective passageways Is one.
results, of course, would have been differ-
ent if the assumptions were changed as
to the average diameter of a channel. But
the implications of this experiment are
clear. Suppose that a lift with a matrix
conductivity of 10~* cm sec-t contains 13
channels (with diameters of 1/2 mm) in
each 2.3 m2 area. The lift would have an
effective hydraulic conductivity of 10"s cm
seer1. Even then, however, the EPA stan-
dards of 1 X 10-7 cm sec'1 would be satis-
fled with eight lifts if interlift flow radii could
be limited to 20 cm. If IR is limited to 10
cm, just four lifts will result in a conductiv-
ity of less than 1 X 10-7 cm seer1.
The "pocket model" was run for two
through eight lifts with the same assump-
tions, i.e., hydraulic conductivity (1 X 10-5
cm see1) and channel radius (250 um).
The results of these experiments are given
}n Figure 7. Again, increasing the number
of lifts and decreasing the IR decreased
the total conductivity of the liner.
Pockets with IR - 15 cm resulted in a
total conductivity of less than 1 X 10'7 cm
sec*1 if there were at least five lifts. If there
were seven lifts, 20 cm of interlift flow
yielded a conductivity of 4.4 X 10'* cm
see1.
In both models, there was a substantial
decrease in conductivity as the number of
lifts increased and as IR decreased. As IR
approached 100 cm, the conductivity be-
came that of individual lifts, and the num-
ber of lifts became unimportant. Except
for two cases (IR = 10 cm with four lifts
and IR -15 cm with eight lifts), the pocket
model gave lower conductivities than the
channel-centered model.
the moment that liquid first reaches a chan-
nel in a new lift. When the bottom lift has
liquid in a channel, it is assumed thai
breakthrough has occurred.
To operate the program, the same as-
sumptions (vertical hydraulic conductivity
and channel radius) are supplied as for
the previously described models. The
program also assumes 30 cm of liquid on
top of the first lift. Additional variables
include horizontal conductivity a\. the bot-
tom of a lift, starting head, liquid density,
and liquid viscosity.
The program generates the connections
between channels in neighboring lifts as a
function of time, and it tells when a new
lift has been penetrated. If the same pa-
rameters are used as in models presented
above (i.e., a vertical hydraulic conductiv-
ity of 10-5 cm sec"1 caused by channels
250 |j.m in radius in a eight-lift liner) with a
horizontal hydraulic conductivity in the
interlift plane of 10-s cm sec-1, the result
would be very rapid breakthrough. To
break through a liner such as this would
take slightly more than 2 wk. If the con-
ductivity between flaws is 10-8 cm sec'1,
and increases to 10~5 cm sec'1 at the bot-
tom of each lift because of density varia-
tion caused by improper compaction, the
liner would also be broken through in 2
wk. A 7.62 cm shelby tube sample taken
at the surface would likely give a labora-
tory conductivity of 10-B cm sec-1.
When using 10'7 cm sec-1 as the hori-
zontal conductivity between lifts, the aver-
age time of 50 model runs to breakthrough
was 4.7 yr. When the horizontal conduc-
tivity between the lifts was modeled as
10-* cm sec'1, the average breakthrough
time was in decades (71 yr).
Interlift Flow Breakthrough
Model
A unifying model would be one that
combined the channel-centered model, for
which there has been supporting evidence
in field studied, to its physical basis. In
this model, IR would not be constant, but
rather a function of time. IR begins to
grow from the !time liquid reaches a par-
ticular throughj-going channel and grows
as a function of head, conductivity, and
time. In the program, each channel has
its own IR. Thp program gives an evolv-
ing picture of the flow in a liner and marks
First Lift
Second Lift
Third Lift
Fourth Lift
Fifth'Lift
Sixth Lift
Seventh Lift
Eighth Lift
Figure 5. The difference between the number of effective flow paths and the fewest number of
active channels In a lift. The fewest number of interconnections between lifts Is three1,
yet the number of flow paths Is one.
-------
o
I-
I 4-\
2-
1 -
•6
•5
•4
10 20 30
Interlift Flow Radius (cm.)
.
'S
4
40
Figure 6. Interlift flow radius vs. average number of flow paths for two to eight lifts.
Laboratory Study
The laboratory portion of this research
was conducted to evaluate the effects of
clod size on the hydraulic conductivity of
compacted soils and the uniformity of
conductivity with depth within a lift of com-
pacted soil. The study used two soils: 1)
Beaumont clay (primarily smectite) and 2)
Kosse clay (primarily kaolinite). The hy-
draulic conductivity as measured in the
laboratory with water as the permeant av-
eraged 1.5 X 10'9 and 0.8 X 10'7 cm sec'1
for the Beaumont and Kosse soils, re-
spectively. The soils were sieved to create
maximum clod sizes of <2.5, <5.0, and
<7.5 cm. The sieved samples were brought
to optimum moisture content, equilibrated
for 1 wk, and then compacted in 60-cm-
diameter permeameters using a specially
constructed compaction foot. Soils were
compacted to form a single lift averaging
23 cm in thickness. After compaction, the
permeameters were equipped with head
tanks to allow the addition of a 1 m head
of water above the soil. A fluorescent dye
was added to the permeant to aid in
identifying flow paths through soil. For all
clod sizes of the Beaumont soil, the hy-
draulic conductivity of the complete 23 cm
lift was much greater than that measured
in the laboratory using small-diameter
permeameters. However, the hydraulic
conductivity of the complete compacted
lift of the Kosse soil approximated that
measured in the small laboratory
permeameters. Under the carefully con-
trolled conditions imposed in this study,
clod size did not have a significant effect
on the hydraulic conductivity of the Beau-
mont clay.
To evaluate the uniformity of conductiv-
ity with depth, 7.5 cm of soil was removed
from each permeameter and the conduc-
tivity was again measured in the remaining
soil. Results indicated that after removing
the upper 7.5 cm of soil, the conductivity
ranged from 0.9 to 2.5 times that of the
complete lift. Removing an additional 7.5
cm of soil resulted in conductivities ranging
from 0.9 to 13 times that of the entire lift.
It appears that for each soil, the lower
portion of the 23 cm-thick-lift of compacted
soil is about 10 times more permeable
than the top.
Each 7.5 cm layer of soil was removed
in 2.5 cm increments. After each 2.5 cm
increment was removed, the soil surface
was viewed under ultraviolet light to ob-
serve dye patterns. Drawings of dyed ar-
eas with depth generally showed that a
large fraction of the cross section was
dyed at the 2.5 cm depth and that the
dyed fraction decreased with depth.
From the data collected in the labora-
tory study, it appears that the compactive
effort applied during construction destroys
the original clods present in the upper
portion of a soil lift and results in a more
platy structure. The amount of platy struc-
ture decreases and the number of highly
conductive pores increases with depth
within a single lift of compacted soil.
Therefore, water flow through a lift of com-
pacted clay is initially through a tortuous
path between soil structural units a'nd at
deeper depths occurs through highly con-
ductive interclod macropores.
Field Study
A field study of a compacted liner, using
soil with properties considered to be ideal,
was conducted to investigate the possible
existence of zones of high, horizontal, hy-
draulic conductivity in interlift areas. The
test was run on a compacted soil liner
consisting of three approximately 20 cm
lifts. A tracer solution was allowed to infil-
trate through auger holes for 10 days, at
which time excavations were made to ex-
pose a vertical plane extending outward
from the auger boreholes. Three auger
boreholes completely drained the solution
within a matter of minutes. Two boreholes
were excavated, including one in which
rapid draining had occurred. Observations
revealed that pockets of dye had accu-
mulated in the interlift zone around that
borehole. These pockets were completely
saturated with dye. Another interesting
observation was the occurrence of debris-
enveloping voids. A dye-stained root was
unearthed in the infiltrated area, which
suggested that fluids flow preferentially
along the periphery of such features.
Evidence gathered during the field study
indicated there was preferential interlift fluid
flow because of the existence of high
permeability zones lying between liner lifts.
Pockets of dye 1 to 4 cm wide in the
vertical dimension were found in the bot-
tom portion of lifts in some places. In
other places, only a thin horizontal band
of dye, 1 to 5 mm thick, was observed.
The leading edge of the dye (IR) was
located at an average distance from the
borehole of 36 cm after 10 days of flow.
This indicated that the rate of horizontal
-------
flow at intertills far exceeded that of verti-
cal flow within a lift.
The results of the field study indicated
that the concept of the channel-centered
interlift flow model was valid for compacted
soil liners. Breaches will occur in a soil
liner if the IR of a channel overlaps chan-
nels in underlying lifts. Consequently, the
IR is a critical factor in the analysis of liner
integrity.
Conclusions/Recommendations
1. Hydraulic conductivity is the most
important property in determining the thick-
ness required in a soil liner. Conductivity
achieved in the liner is affected by several
factors including liner thickness, soil ma-
trix permeability, lift thickness, horizontal
flow between lifts, and the presence of
channels or defects.
2. Hydraulic conductivity of the soil
liner is also the most important factor af-
fecting the time it takes for hazardous
constituents to break through the liner.
The computer model presented in this
report simulates flow occurring through
discrete channels (defects) in each lift and
describes how these channels affect the
overall conductivity of the liner. The chan-
nels can be interconnected by horizontal
flow between lifts to form a continuous
flow route through the liner. Minimizing
horizontal flow will reduce the conductivity
of the overall liner by reducing the num-
ber of interconnected channels. The com-
puter model can also be used to simulate
the effects of different horizontal hydraulic
conductivities [between lifts on break-
through of leachate. Breakthrough time
was found to be primarily a function of
horizontal conductivity.
The following studies should aid in ex-
panding the daia base necessary to de-
termine the required thickness for a soil
liner: !
1. Field and laboratory studies to in-
vestigate the horizontal flow of liquids be-
tween lifts and [aid in determining factors
that affect the rate and extent of interlift
flow;
2. Studies of the depth functions for
density, effective porosity, tracer move-
ment, strength, [and hydraulic conductivity
within individual; compacted lifts of varying
thickness;
3. Investigation into whether construct-
ing soil liners wijh a greater number of lifts
effectively reduces flow between lifts and
total flow through the liner, and
4. Field morphological studies of dye
tracer movement through a liner and indi-
vidual lifts to develop relationships be-
tween dye-filled porosity and liner perfor-
mance variables including hydraulic con-
ductivity and strength.
Ideally, studies such as these would be
conducted on a variety of soils compacted
over a range of moisture contents and
with the use of a selection of commonly
available equipment.
Studies conducted during this project
suggest that flow through soil liners is
controlled by a combination of vertical
conductivity within lifts and horizontal con-
ductivity between lifts. Consequently, the
construction quality assurance program for
a hazardous waste disposal facility should
determine if such preferential flow paths
exist in a given liner by conducting dye
tracer studies on the soil liner of the test
fill. It is suggested that dye be placed in
the field infiltrometers and augured bore-
holes used on the test fill so that subse-
quent dissection will reveal the extent and
routes of preferential liquid flow.
The full report was submitted in fulfill-
ment of Contract No. 68-03-1816, Work
Assignment No. 2-10, by K. W. Brown
and Associates, Inc., and Cooperative
Agreement No. CR813444 with Texas
A&M University, and the University of
Texas, under the sponsorship of the U.S.
Environmental Protection Agency.
10
20
Interlift Flow Radius (cm)
40
Figure 7. Pocket radius vs. average number of flow paths for two >to eight lifts.
6
•&U.S. GOVERNMENT PRINTING OFFICE: 1991 - S48-OZ8/40M8
-------
-------
David C. Anderson, Mark J. Lupo, James A. Rehage, Josepli O. Sai, Ronald L. Shiver,
and Robert C. Speake are with K.W. Brown and Associates, Inc., College
Station, TX 77840; K. W. Brown is with the Soil and Crop Sciences Dept, Texas
A&M University, College Station, TX 77843-2474; and D. Daniel is with the Civil
Engineering Dept., University of Texas, Austin, TX 78713-^726
G. Kenneth Dotson is the EPA Project Officer (see below).
The complete report, entitled "Factors Controlling Minimum Soil Liner Thickness,"
(Order No. PB91-191346AS; Cost: $31.00, subject to change) will be available only
from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information <
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA PERMIT NO. G-35
Official Business
Penalty for Private Use $300
EPA/600/S2-91/008
------- |