Professional Documents
Culture Documents
by
EXECUTIVE SUMMARY
Past regulations required that excess foundry sands be disposed in landfills. However,
new legislation permits the reuse of foundry sands in various applications, such as earthwork
projects, which often require large amounts of earthen materials to be used as structural fill. These
projects often use geosynthetics for reinforcement. Thus, interaction between foundry sands and
geosynthetics must be characterized.
A testing program was conducted in this study to characterize the strength of foundry
sands and their interaction with geosynthetics. Small-scale direct shear, large-scale multistage
interface shear, and pull-out tests were conducted. Multistage interface direct shear tests were
used to facilitate the test schedule. Preliminary testing showed that the multistage tests gave
residual shear strengths similar to those obtained from single specimen test set-ups.
The undrained friction angle of as-compacted foundry sands was found to be the same as
that of the base silica sand. The friction angles ranged between 39 and 50". The undrained
cohesion increased with increasing fines content and ranged between 17 and 28 kPa. lnterface
direct shear tests showed that foundry sands develop good interaction with geosynthetics, and that
the interface strength is primarily due to friction (i.e., adhesion was small). Typical interface friction
angles ranged between 25 and 3S0, with the highest friction angles being obtained for a
geomembrane interface with foundry sand having low bentonite content. Frictional efficiencies
ranged between 51 and 86%. Soaking had little effect on interface friction angle but caused a
decrease in adhesion to values corresponding to machine friction. lnteraction coefficients
calculated from the pull-out tests ranged between 0.30 and 1.7 in the normal stress range of 10 to
50 kPa.
Recommendations for design were developed based on the test data. For mechanically
stabilized earthen structures, interaction coefficients between 0.5 to 0.9 are recommended for
geotextiles and geogrids. Interaction coefficients recommended for embankments range between
0.5 and 0.7. Interface friction angles between 28 and 31" are recommended for embankments and
landfill covers.
ACKNOWLEDGEMENT
Financial support for this study was provided by the Wisconsin Solid
University of Wisconsin System. This report has not been reviewed by SWRP,
excess foundry sand for use in this project, as well as other research efforts on
2 . BACKGROUND.....................................................................................................
3
4 . RESULTS ...........................................................................................................
43
4.1 SMALL-SCALE DIRECT SHEAR TESTS ..................................................... 43
4.1.1 Testing Program ..................... . .........................................................43
4.1.2 Mohr-Coulomb Failure Envelopes ......................................................... 43
4.1.3 Effect of Fines..................................................................................... 49
6 . REFERENCES .................................................................................................. 86
APPENDICES ...........................................................................................
89
APPENDIX A ................:................................................................
-90
APPENDIX B.................................................................................92
APPENDIX C ................................................................................
104
APPENDIX D................................................................................
108
LIST OF .FIGURES
Fig. 3.2. Standard Proctor Compaction Curves for the Foundry Sands (Adapted
from Kleven (1998) and Abichou et al. (1998a)).
Fig. 3.3. Grain Size Distribution for the Foundry Sands (Adapted from
Kleven 1998).
Fig. 3.5. Variations of lnterface Shear Testing (Adapted from Takasumi et al.
1991).
Fig. 3.6. lndentation Effects of Soil Substrate (Sand B-Geotextile lnterface Direct
Shear Test).
Fig. 3.9. Shear Stress Displacement Curves from Multistage and New Specimen
Test Set-Ups.
Fig. 4.3. Fines Content versus Bentonite Content for Data from Kleven (1998).
C
Fig. 4.7. Effect of Fines Content on (a) lnterface Frictional Angle and (b) Adhesion
Fig. 4.8. Typical Plots of Pullout Force per Unit Width versus displacement for (a)
Sand A-Geogrid and (b) Sand C-Geotextile
Fig. 4.9. Estimation of Mobilized Length of Geosynthetic from Telltail LVDT Data.
Fig. 4.10. Relationship between Ci and Normalized Mobilized Length (LmIL). For all
Tests L = 1.31 m.
Fig. 4.1 1. lnteraction Coefficient versus Normal Stress for the Geogrid
Fig. 4.1 2. lnteraction Coefficient without Cohesion versus Normal Stress for the
Geogrid.
Fig. 4.1 3. lnteraction Coefficient versus Normal Stress for the Geotextile
Fig. 4.14. lnteraction Coefficient without Cohesion versus Normal Stress for the
Geotextile
Fig. 4.15. lnteraction Coefficient versus Fines for Geogrid: (a) Ci with Cohesion (b)
Ci without Cohesion.
Fig. 4.16. lnteraction Coefficient versus Fines for Geotextile: (a) Ci with Cohesion
(b) Ci without Cohesion.
LIST OF TABLES
Table 3.2. Foundry Sand Properties (adapted from Kleven 1998 and Abichou et al.
1998a).
Table 3.3. Data for Miragrid 12XT Geogrid (1997 GFR SpecifierlsGuide).
Table 3.4. Data for Belton 113 Geotextile (1997 GFR Specifier's Guide).
Table 3.5. Data for GSE HDT Textured Geomerr~brane(1997 GFR Specifier's
Guide).
-
Table 4.9. Summary of Pullout Mobilized Lengths.
INTRODUCTION
landfills. New legislation, however, permits the reuse of foundry sands as well as
which often require large amounts of earthen materials to be used as structural fill
(Wisconsin DNR, 1998). Granular soils are most desirable for these projects,
because of their higher hydraulic conductivity, low compressibility, and high shear
strength. Previous research has shown foundry sands to have these properties
1998b). Also, foundry sands with sufficient bentonite content have been shown to
have low hydraulic conductivity, a property that allows for the possibility of foundry
structural fill, the interaction between foundry sands and geogrids or geotextiles
different than interaction with traditional granular fill materials because foundry
geomembranes come into direct contact with compacted cover soils having low
hydraulic conductivity, such as foundry sand. Thus, design of covers with foundry
sand requires knowledge of the interface strength between the geomembrane and
The objective of this study was to characterize the strength of foundry sands
and their interaction with geosynthetics. To meet this objective, a testing program
was implemented which included direct shear tests, interface shear tests, and
pullout tests. Results of these tests were used to develop recommendations for
This report describes the results of this study. Section 2 describes the
process that generates excess foundry sand, the constituents of foundry sand, and
equipment, and materials used in the laboratory portion of this study are described
in Section 3. Section 4 presents the results of the laboratory tests and design
BACKGROUND
This section describes the process by which excess foundry system sand is
generated, the constituents of the sand, and several previous research projects on
using excess foundry system sand in structural fill. For brevity, the phrase "foundry
Foundry sands are generated as a by-product of the cast metal process, which
involves the making of molds into which molten metal is poured to create castings.
These molds are made of system sand, a combination of clean, uniformly graded
sand (usually mostly silica), a binding agent (either clay or chemical), water, and
additives such as seacoal, cereal, fuel oil and wood flour (Javed 1994). Waste
resulting in a larger volume of sand than is needed for the foundry process.
System sand is hammered around a pattern to form a void into which molten
metal will be poured. These molds consist of an upper and lower box and
sometimes have cores in the middle to create voids within the metal. Cores are rigid
pieces and are usually made of chemically bonded sand. After the molten metal
has been poured and has cooled, the casting is removed in a process called
shakeout. During shakeout, system sand is collected and becomes mixed with
some core sand. This mixture of sands is then put back into the sand circuit for
reuse.
of core sand into system sand during shakeout dilutes the original mixture. Second,
plasticity, and gas permeability deteriorate due to thermal degradation of the clay
The silica sand portion of foundry sand usually makes up 85% to 95% of the
total mix by weight. The clay'binder, around 5% to 15% of the mix, is usually
added to "activate" the clay portion, giving the mixture cohesion and plasticity. The
organic portion usually makes up 2% to 8% of the mix and is added to improve the
surface finish of the cast metal (Kleven 1998). Foundry sand is often
heterogeneous due to the presence of core butts, metal, slag, and additives. This
sand.
sands. Javed (1994) performed a variety of laboratory tests on seven clay bound
conductivity. Javed found that the plasticity index ranged from non-plastic to 12.
Friction angles of 37" and 41" were reported for two dry sands tested in direct shear
at relative densities of 90% and 98%. Javed concluded that foundry sands have
,that foundry sands are more compressible than clean sands, with constrained
moduli in the range of 1500 kPa to 2000 kPa in a vertical stress range of 0 to 1000
sand with 5.3% fines (material passing the No. 200 sieve, 0.075 mm mesh), and 2.6
x 1o ' cmlsec
~ for a foundry sand with 35.2O/ f'~nes.
high embankment consisted of three subsections made of clay, foundry sand, and
clean sand. The foundry sand Mast used had 9% clay sized particles (c 0.005
mm). The plasticity index was 6 and the friction angle measured in direct shear was
38" at the as-received water content in a dense state. The hydraulic conductivity
that settlement of the foundry sand was less than 4 mm, which was comparable to
that of the clean sand. Lateral movements were less than 6 mm. The in situ
6
tests (SPT) gave N-values ranging from 33 to 54, showing that the foundry sand had
relatively high strength. Mast (1997) concluded that the foundry sand he used had
removal of large objects (e.g., core butts, metal, slag), dust control, and erosion
control. These issues were addressed in his field study without much difficulty.
Abichou et al. (1998b) describe two other highway embankments that were
constructed with foundry sand near Waupaca, Wisconsin, and New London,
(WDNR). 'The first embankment was an overpass approach 213 m long with a
maximum foundry sand thickness of 4.6 m. No data are available regarding the
performance of this embankment. The second embankment was for a bypass and
used 19,000,000 kg of core sand and mold sand. Very little information is available
on this embankment.
Abichou et al. (1998a) and Kleven (1998) conducted tests on several foundry
limits, particle size distribution, "activeIP clay content (i.e. bentonite content),
modulus. Table 2.1 summarizes his results. One notable aspect of Table 2.1 is the
Table 2.1. Summary of Data from Kleven (1998).
Soil # I Binder (
I I
PPm I p2.,, I Active
I I
1 1 1 1 I NP I NP 1 I 1 1 1
p 1
5
Clay
Raw
Clay
10.7
14.3
6.7
9.2
5.1
7.0
NP SP- A-3(0) 2.62 18.4 9.6
variation in active clay content. The unconfined compression strength ranged from
71 to 190 kPa for foundry sands with plasticity. Unconfined compressive strength
desiccation, and chemical permeation. Table 2.2 summarizes the data in Abichou et
al. (1998a). The hydraulic conductivity of foundry sands was found to decrease with
increasing bentonite content and plasticity index. At bentonite contents > 6%,
of concern. Lovejoy et al. (1989) conducted laboratory and field tests to determine
.
the leaching characteristics of excess foundry system sands. Toxicity Characteristic
Leaching Procedure (TCLP) tests were performed on three foundry sands and none
beneath a foundry sand pile and a natural soil pile showed no evidence of
C
15 NT 27 20 7 13 9.0x lo-'
16 2.73 NP NP NP 4.7 5.3x I O * ~
Notes: (i) NP = Non Plastic
(ii)
1
NT = Non Tested
Hydraulic Conductivity at Optimum Water Content Based on Standard
Proctor
SECTION 3
3.1 MATERIALS
Two sands were used as reference materials for comparison with the foundry
sands: Portage sand and base sand. Portage sand, which was previously used by
Foose (1993) and Tatlisoz (1996), is a medium uniformly graded quartz sand with a
0.77. Base sand is fine, uniformly graded quartz sand (SP) used by several
foundries as their base for system sand and has a roundness of 0.63. The base
sand was obtained from Badger Mining Corporation and is designated as GFN 65
(grain fineness number) silica sand. GFN is a term used by the foundry industry to
designate grain size. Particle size distribution curves for ,these sands are given in
Four foundry sands were tested. The sands had different percentages of
bentonite that span the range of bentonite4ontents typically found in foundry sands.
They were obtained from four different foundries located in Wisconsin and Illinois.
Compaction curves (from ASTM 0 698) and grain size distribution curves for these
sands are in Figs. 3.2- 3.3. Table. 3.2 provides the characteristics of ,the sands
- Base Sand - -m - - Portage Sand
1 0.1
Fig. 3.2. Standard Proctor Compaction Curves for the Foundry Sands
(Adapted from Kleven (1998) and Abichou et al. (1998a)).
-A- Sand B
+ - Sand C
--o--SandD
methylene blue titration). Foundry Sand A has the lowest bentonite content (2.3%)
and Foundry Sand D the highest (13%). For pullout tests, Foundry Sand A was
used as the low bentonite foundry sand, while Foundry Sand B was used as the low
bentonite content sand in the interface direct shear tests. Foundry Sand A had a
larger fraction of coarse material, as shown in Fig. 3.3. This coarse material
consisted of fragments of core remnants and slag. The as-received sand had large
core butts and slag pieces that were removed with a No. 4 sieve.
3.1.3 Geosynthetics
geosynthetics are in Tables 3.3 - 3.5. Belton 113 is a woven, slit-film polypropylene
geotextile. Miragrid 12XT is a woven polyester geogrid with PVC coating. GSE
.
that' would be likely be used with foundry sands in reinforced earthen structures or in
landfill covers.
shear machine, and an interface direct shear machine. A detailed description of the
Table 3.3. Data for Miragrid 12XT Geogrid (1997 GFR Specifier'sGuide).
Table 3.4. Data for Belton 113 Geotextile (1997 GFR Specifier's Guide).
Table 3.5. Data for GSE HDT Textured Geomembrane (1997 GFR Specifier's
Guide).
-
pullout box can be found in Tatlisoz (1996). The interface direct shear machine is
soil and measuring the force required to displace the geosynthetic. The soil is
friction (all geosynthetics) and rib-bearing (geogrids only). Friction develops between
the upper and lower surfaces of the geosynthetic and the surrounding soil when the
geosynthetic is displaced. Rib bearing is the passive resistance put forth against the
transverse members of a geogrid by the soil. The soil provides this resistance by
"st rike-through," which means that soil particles protrude through the apertures in
end, but the magnitude of the displacement diminishes with distance from the
geosynthetic creating friction with the soil is constantly changing and the friction that
is mobilized varies with displacement. For this reason, many pullout boxes are
19
The extensibility and progressive failure vary with normal stress. At low
normal stresses, the geosynthetic may fail progressively until the entire length of the
become anchored at a given distance from the front of the box. Thus, only a portion
parameter called the interaction coefficient, Ci. For cohesionless soils, the
the length of ,the embedded portion of the reinforcement, on is the normal total
stress applied to the soil, and $ is the total stress friction angle of the soil. For
c.' = -I
2WL(on tan 9 + c)
where c is the total stress cohesion of the soil. An interaction coefficient between
0.5 and 1.0 indicates a good bond between the soil and reinforcement. Poor bonds
20
reinforcement results in the shear strength of the interface being mobilized along a
3.2.1.2Experimental Set-up
GeoSyntec Consultants. The machine, shown in Fig. 3.4, consists of a box 152 cm
hydraulic pistons that are driven by an electrically powered hydraulic pump. Normal
stress is applied by a rubber air bladder that is confined by a reinforced steel cover
geosynthetic are measured at various distances from the front of the box by four
LVDTs mounted on a rack at the back of the box. These LVDTs are attached to
"telltale" wires that enter the box through a slot in the back and are protected by
.
aluminum tubes. During pullout, the wires are pulled along with the moving
Pullout load is measured by a load cell mounted on the head (Fig 3.4). A rod
C
threaded into the load cell pushes against a plate that is attached to the
To reduce the amount of soil necessary for testing, a false bottom was placed
was placed on top of the plywood to prevent soil moisture from moving into the
wood. To reduce sidewall friction, plastic was attached to the sidewalls of the box
The pullout test procedures used were conducted in general accordance with
GRI test methods GG5 and GT6. For tests using dry sand, the soil was weighed,
shoveled into the box, and tamped to create the desired density. Foundry sands
were first passed through the No. 4 sieve and then wetted to optimum water content
based on standard Proctor. The moist soil was then sealed with plastic and allowed
to sit one week to allow for hydration and equilibration. Kleven (1998) showed that
The foundry sands were compacted to about 92% of the standard Proctor
maximum dry density, except for Foundry Sand D, for which only 85% could be
achieved. Compaction was performed with two compactors, one large (15.2-cm by
15.2-cm plate, 10 kg), and one small (described in Sec. 3.2.3.2). Two passes were
made with the large compactor (40 - delivered from height of 30 + 4 cm)
+ 5 blowsllift
and one pass was made with the small compactor (100 f 20 blowsllift delivered from
30 + 4 cm). Soil below the geosynthetic was compacted in three lifts (total height =
18 cm) and leveled to the appropriate height for the geosynthetic. The geosynthetic
23
(131 cm by 40.6 cm) was then placed and the telltale wires were attached at four
locations (i.e., 6, 36, 72 and 130 cm from the front of the geosynthetic). The upper
soil was compacted on top of the geosynthetic in three lifts (total height = 26 cm),
leaving a gap of about 5 cm between the top of the soil and the top edge of the box
A nonwoven geotextile was placed on top of the soil to protect the bladder.
Then the bladder was placed and the reinforced steel cover was bolted to the box.
After pressurizing the bladder, the pullout test was performed by setting up the data
collection system and turning on the hydraulic pump. All tests were continued to 10
and in applying the desired normal stress. The displacement rate is controlled by an
electric-powered hydraulic pump fitted with a rate control arm. This pump circulates
fluid through the two pistons, which provide the pullout force (Fig. 3.4). In addition to
the rate control arm, each of the pistons has a valve that controls the displacement
rate of the piston. A constant displacement rate is difficult to maintain because the
valves on the pistons are extremely sensitive at the rate used in this study (1
mmlmin). Also, it is difficult to keep the pistons moving at the same rate. This
probleni was overcome by constantly monitoring the pistons during the test and
Another problem encountered in the initial tests was that the desired normal
stress was not always achieved. In some tests, an apparent increase in normal
24
stress resulted in lower pullout resistance. To check if the air bag pressure was
transmitted as normal stress to the soil, the pressure in the air bag was measured
by a transducer attached to a tee in the air pressure line and the normal stress
applied to the soil was measured with an earth pressure cell placed between the
inflatable bladder and the soil. This test showed that the air pressure applied to the
bag was much higher than the normal stress being applied to the soil. Examination
showed that the low normal stress was caused by leaks in the rubber bladder. In
subsequent tests, the normal stress applied to the soil was verified by the earth
Engineering Ltd. direct shear machine, as described in Ligler (1998). The tests
consists of a 6.4 by 6.4 by 3.2-cm shear box. Normal load is applied to an upper
loading platen through a lever arm. Shear force is measured by a load cell and
Direct shear tesis were also performed in the large-scale machine described
-
in Sec. 3.2.3, and while the results for the reference sands were comparable to the
small-scale machine, the results for the compacted foundry sands were anomalous.
The foundry sands tested in the large-scale machine failed along a concave
inteface, which is inconsistent with the failure plane assumed in direct shear.
3.2.3 lnterface Direct Shear Tests
between soil and reinforcing materials, particularly geosynthetics. The test consists
the resistance. There are three different variations of this test, as described by
Takasumi et al. (1991): fixed shear, partially fixed shear, and free shear, shown in
Fig. 3.5. The difference between these variations is the manner in which the
geosynthetic is attached.
In the fixed shear test, the geosynthetic is glued to a rigid substrate, usually
wood. 'The advantages of this test are that the rigid substrate ensures a horizontal
failure plane and that less soil is required. 'The disadvantage of the fixed shear test
is that the geosynthetic is not allowed to elongate, which can affect the interface
strength. In partially ,fixed shear, the geosynthetic is clamped at one end and free at
the other, with soil both above and below the geosynthetic. The advantage of this
test is that it models the field situation most accurately. The disadvantage of this
test is that the geosynthetic may deform into the lower shear box. Free shear
leaves the geosynthetic unattached at either end, with soil above and below the
geosynthetic, as in the partially fixed test. -The advantage of this test is that a
special box is not required. Disadvantages of the free shear test are that the
geosynthetic may deform into the lower ;hear box and the geosynthetic can slide
osynthetic
\ Clamp
(b) Partially Fixed Shear
,Geosynthetic
Fig. 3.5. Variations of Interface Shear Testing (Adapted from Takasurr~iet al.
1991).
27
Partially fixed shear testing was initially attempted in this study. Data from a
test with Foundry Sand B and the geotextile, shown in Fig. 3.6, indicated that the
constant peak or a residual value. Fig. 3.6 shows that this increase is more
because of indentation of the geosynthetic into the lower soil. This problem is
illustrated schematically in Fig. 3.7, which shows.soil from the upper box being
forced down into the lower soil due to the normal stress. This caused the
resistance with increasing displacement at the leading edge of the upper soil.
was used as the substrate (without gluing the geosynthetic to the wood) and soil
was placed in the upper portion of the box. Testing with a wooden substrate
in Fig. 3.8. Thus, for the remainder of the tests, a rigid substrate was used in the
lower box. The substrate consisted of a porous stone prepared with a mixture of
Portage sand and epoxy. The stone provided a rigid base, like the wood, and also
permitted drainage and had frictional characteristics similar to the soils being tested.
Data from a multistage test with the porous stone is also plotted in Fig 3.8 for
comparison with the wooden substrate. Another variation used in this project was
"multistage shearing.'' Multistage shearing means only one test set-up is used
instead of a new soil sample for each normal stress. The advantages of multistage
Displacement (mm)
Fig. 3.6. Indentation Effects of Soil Substrate (Sand B-Geotextile Interface Direct
Shear Test).
lndentation of geosynthetic
causes passive resistance
at leading edge of upper soil
0 10 20 30 40 50 60
Displacement (mm)
testing are time savings and elimination of variability in test specimens (Bemben and
Shulze 1993). However, the friction angle obtained from multistage tests is smaller
than the peak friction angle obtained from a traditional test. Head (1982) suggests
that when the multistage set-up is used in traditional, soil-only direct shear tests, the
failure plane is forced down into the lower box during each consolidation step which
causes a new failure plane in each shearing step. This did not occur in this study,
however, because the location of the failure plane was maintained by the rigid
substrate. Bemben and Shulze (1993, 1995) tested Ottawa sand with smooth and
textured geomembranes and found that the multistage shear stresses coincided
with the residual shear stresses found from new test set-ups.
To assess how results from multistage tests differ from those obtained with
new test setups, interface shear tests were performed for Sands B, C, and D with
the geotextile using both multistage and new specimen set-ups. Stress-
displacement curves for Foundry Sand B and the geotextile are shown in Fig. 3.9.
Displacements for the multistage shearing tests in Fig. 3.9 are offset slightly from
zero so that the curves for the same normal stresses can be compared. The shear
stress from multistage shearing is essentially the same as that from the new
kPa. 'The percent reduction in tan6 from the new specimen (peak) set-ups to
18% for Foundry Sand D. Because a residual friction angle is obtained in multistage
testing, a consistent method was needed to define the residual shear stress. This
0 10 kPa, new specimen 30 kPa, step 2
I 30 kPa, new specimen 50 kPa, step 2 1
x 50 kPa, new specimen 0 50 kPa, step 3
Displacement (mm)
Fig. 3.9. Shear Stress Displacement Curves from Multistage and New Specimen
Test Set-Ups for Foundry Sand B.
33
was done by graphing mobilized friction angle versus displacement on a log scale
for a multistage set-up, as shown in Fig. 3.10 for interface tests with base sand and
Foundry Sand B with the geotextile. The friction angle becomes stable after about 8
Thus, residual friction angles from interface tests were found by choosing the shear
likely to occur during construction and thus, the residual friction angle obtained for
Interface shear testing was performed using the interface direct shear
lower box, 35.6 cm long by 30 cm wide by 7.6 cm deep. The longer lower box
allows for 5 cm of travel without requiring an area correction for the stress
calculations.
Normal stress is applied to the soil by a square loading platen that fits into the
upper box (Fig. 3.1 1). This platen is loaded by an air bladder that rests on top of the
platen and is confined above by a circular plate connected by a rod to the outer
C
frame. The vertical load is measured by a load cell, into which this rod is threaded.
Both shearing boxes are contained by an outer box, which allows soaked tests to be
performed.
onstant after 8 mm
of displacement
a
U a
H a
I I
0
0.1 1 10 100
Displacement (mm)
During testing, the outer containment box, with the lower shear box clarr~ped
inside, is screw driven horizontally on two greased tracks by a stepper motor. The
rate and distance of displacement are controlled by a computer that is wired to the
motor and three limit switches. 'The upper box is fixed horizontally by a reaction bar
that is connected to a load cell to measure shear force. Thus, the outer
containment box, along with the lower shear box that is clamped to it, are the only
variable differential transducers (LVDTs). All data collection was done using a PC
For soaked tests, the outer box was sealed and filled with water. A
tensiometer was placed in the soil specimen to verify saturation. The tensiometer
consists of a plastic tube (diameter = 5 mm) with a porous ceramic tip (diameter =
10 mm) attached to the end of the tube and sealed with epoxy. This ceramic tip was
placed in the center of the upper shear box, 2.5 centimeters above the interface,
dissipated after the specimen was submerged in water. A soaking time of 4 hours
was required for Foundry Sand B, 72 hours for Foundry Sand C, and 96 hours for
Foundry Sand D.
-
3.2.3.2 Specimen Preparation
Foundry sand specimens were prepared by mixing dry sand with tap water in
Proctor). Prior to n-lixing, the sands were passed through the No. 4 (4.8 mm) sieve
to remove large metal pieces and core remnants. Each sand was allowed to
equilibrate in a sealed bucket for at least one week to allow for hydration of the
bentonite.
first cut and clamped to the lower box on top of the porous stone. The upper box
was then placed on top of the geosynthetic. To create a gap of 1 mm between the
upper and lower boxes, a piece of geomembrane cut to the shape of the box edges
was placed between the boxes. The upper box was then filled with three lifts of
sand.
The reference sands (Portage Sand and base sand) were placed dry by
pouring a known mass of soil into the upper box and tamping it to the correct volume
optimum water content. Compaction was performed in three 2-cm-thick lifts using
25 f 5 blows per lift delivered from a height of 10 f 3 cm. 'The compactor weighed
C
1.9-cm diameter by 122-cm long rod. Maximum standard Proctor unit weight was
achieved by corrlpacting a measured mass of soil at optimum water content into the
38
known volume of the upper box. A gap of 1.3 cm was left between the top of the
soil and the top of the box for loading platen. The upper surface of the soil was
flattened with a scraper to create an even gap below the top of the box.
with deaired water was placed on top of the first lift. The tube for the tensiometer
was routed through a hole in the back of the upper box. Sand was then compacted
Tests with dry sand or foundry sand compacted at optimum water content
were conducted using the following procedure. First, the normal stress was applied
by pressurizing the air bladder until the vertical load cell read the desired force.
Next, the machine was set to displace the lower box at a rate of 1 mmlmin, as
recommended by ASTM D5321 for tests in which no excess pore pressures are
anticipated. Data acquisition was then initiated. Shearing was performed in four
"multistage" steps. The first step was run at the lowest normal stress to a
displacement of 10 mm to remove the "peak behavior of the soil, so that all shear
strengths would correspond to a similar displacement. The next three steps were
-
The position of the box was not reset between steps. Thus, the total displacement
was 52 mm.
3.2.3.4 Tests with Soaked Foundry Sands
For soaked tests, the specimen was first allowed to soak long enough to
expedite testing, soaking was done without application of a normal stress. In a field
situation, soaking would likely occur under a normal stress, since it would probably
happen after construction. The time for soaking ranged from about 4 hours for
Foundry Sand B to 4 days for Foundry Sand D. After soaking was complete, the
platen was placed and a normal load was applied. The specimen was allowed to
rate that would prevent generation of excess pressure. 'The time for 90'
consolidation (tgO)was found using a Taylor construction and the time to 50% (t50)
was calculated as described in ASTM D 3080 (i.e., t50 = tgd4.28). Some typical
displacement at failure (df), the displacement rate (d,) was calculated from the
equation:
Table 3.6 summarizes the data obtained from consolidation steps performed
in multistage shear tests. The displacement rates used were the same for a given
Table 3.6. Displacement Rate Data.
Sand D 1
I
13 1
I
600 1
I
5.0 1
I
0.01 1
I
0.013
' ~ r o mtests at optimum water content
2
Calculated from consolidation data
3 . .
From Sabatini et al. (1998)
41
soil, irrespective of the geosynthetic used. The displacement rate could not be
determined for Foundry Sand C using the ASTM D3080 method because the
vertical consolidation data were lost. Thus, the shear rate reported in Sabatini et al.
(1998) for several clayey soils tested with textured geomembranes was used. 'This
rate (0.01 mmlmin) was the same as was calculated from a later test with Foundry
Sand D. The higher displacement rates used for Sands B and C were believed to be
The effect of displacement rate was investigated prior to starting the testing
program by testing Foundry Sand B with a geotextile. Results of these tests are
presented in Fig. 3.12. One multistage set-up was used with four shearing steps.
The first two steps were performed at displacement rates of 0.03 and 0.1 mmlmin at
a normal stress of 30 kPa. Fig. 3.12 shows that variation in the shear stress-
displacement curves for the different rates is small. This is also true for shearing
steps at 50 kPa, with displacement rates of 0.03 and 0.02 rnmlmin. Thus,
displacement rates in the range of 0.02 to 0.1 mmlmin do not have a great effect on
0 1 2 3 4 5 6 7
Displacement (mm)
RESULTS
4.1.ITesting Program
Small-scale direct shear tests were conducted on the two reference sands
and the four foundry sands. Table 4.1 summarizes the testing program for small-
scale direct shear testing. Three normal stresses were used for each soil: 11, 34,
and 56 kPa. Reference soils were placed dry in the shear box at 90% relative
density, while foundry sands were compacted at optimum water content to 100% of
the maximum dry unit weight as determined by standard Proctor. A shearing rate of
1.2 mmlmin was used in all tests. None of the specimens were soaked prior to
testing.
Failure envelopes for Portage sand and base sand are shown in Fig. 4.1.
They were obtained by least-squares regression on the three points for each sand.
The peak friction angles from these envelopes are 42" for both sands. Das (1994)
reports friction angles ranging from 35 to 38" for dense rounded sand.
Table 4.1. Testing Program for Small-Scale Direct Shear.
Base
ii,34,57 1.2 16.9 0
Sand
Portage 1.2 17.5 0
11, 34 ,57
Sand
Foundry 18.4 9.6
. 11,34, 56 1.2
Sand A
Foundry 17.3 13.0
10, 33, 55 1.2
Sand B
Foundry 18.2 12.3
11,34, 55 1.2
Sand C
Sand D
1,
1
11. 33, 55 I
I
1.2
I
16.9
I
15.0
Fig. 4.1 shows a cohesion intercept of 6 to 8 kPa for the reference sands, which
friction." Reference sands tested in the large-scale direct shear machine (Appendix
A) showed slightly lower machine friction (2 to 5 kPa) to those tested in the small-
scale machine. The large-scale test, however, gave lower friction angles (38" for
Portage and base) than the small-scale machine (Appendix A). Tt- is discrepancy
may be due to boundary effects of the small-scale machine, which would be less of
a factor in the large-scale machine. The small-scale values will be used, however,
because of the anomalous results obtained for the as-compacted foundry sands
4.1.2.2Foundry Sands
optimum water content are given Fig. 4.2. Table 4.2 summarizes the results of the
small-scale direct shear tests. The friction angles for Foundry Sands 6 ,C, and D
range from 39" to 43", which is similar to the values of 37" and 41" reported by
Javed (1994) for two foundry sands at relative densities of 90 and 98%. The
foundry sand friction angles were also similar to that of the reference sands (42").
Foundry Sand A, however, had a friction angle of 50". An explanation for this
Cohesion in the as-compacted foundry sands (16.7 to 28.4 kPa) was higher
than the reference sands (2 to 5 kPa) and Javed (1994) (< 2 kPa). This is probably
a result of suction and non-planar failure surfaces observed in these tests, which
may have increased the apparent cohesion. Fredlund and Rahardjo (1993) suggest
cohesion and suction. Thus, it is likely that the effective cohesion measured in the
direct shear tests on compacted foundry sands had a suction component, since the
foundry sands were unsaturated. Also, the failure surfaces in these direct shear
tests were not planar, but undulated. This may have been due to non-uniform
compaction and heterogeneity of the foundry sands, which could cause the failure
sands than clean sands because the plastic fines in foundry sands are not
distributed evenly throughout the sand matrix, but instead can form small clods.
The undulated failure surface may have created an "interlocking rough surface,"
Lupini et al. (1981) suggest that shear strength of a soil can be correlated to
the clay fraction if the mineralogy of the clay fraction is constant. Thus, the shear
strength of the foundry sands as a function of fines content was examined, since
C
industrial bentonite, which has reasonably constant clay mineralogy, usually makes
up the clay fraction in foundry sands. Fines content (O/O particles by weight < 0.075
Also, the bentonite contents obtained from methylene blue tests conducted for this
study were not in agreement with those in Kleven (1998) for the same sands.
Moreover, fines content and bentonite content are correlated, as shown in Fig. 4.3.
Thus, fines content should be a good index test to use for comparing foundry sands
since it is easy to perform and correlates well with bentonite content and shear
strength.
friction angles for the base sand and four foundry sands against fines content as
shown in Fig. 4.4. The fines in the base sand had no effect on the friction angle of
the foundry sands. The friction angle, however, for Foundry Sand A (50") is higher
than the other foundry sands. Foundry Sand A contained large, angular pieces
(-2O0l0 larger than 1 mm) of core remnants and slag. Friction angles for the other
foundry sands were similar (39 to 43"). Overall, the friction angle for foundry sands
Sand A. This is consistent with the findings of Lupini et al. (1981), who measured
fairly constant residual friction coefficients for sand-bentonite mixtures with 0 to 15%
bentonite.
Cohesion generally increases with fines content as shown in Fig. 4.5. This
C
result is expected because suction in the compacted sand should increase as the
bentonite content increases. The effective pore size decreases as bentonite fills the
0 2 4 6 8 10 12
Bentonite Content (%)
Fig. 4.3. Fines Content versus Bentonite Content for Data from Kleven (1998).
/ I I I l l / I I I I I l l 1
- -
- -
-
A
-
- -
- A -
- C -
- D --
- Base
AkA B A A -
-
A -
-
\ -
- \ -
- Portage -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
- -
I l l / l l l l l I I I l I l I I
% Fines
Fig. 4.4. Effect'of Fines Content on Friction Angle
- -
-
"\
-
Portage
C
-
-
-
-
I
I ' I I I I I I I I I I I I I I I I
0
0 5 10 15 20
OO
/ Fines
voids of the base sand, causing higher capillary forces, and thus increased suction.
Foundry Sand A, however, has a higher cohesion (28 kPa) than expected for sand
with low bentonite content (2.3%). This could have occurred if the testing water
content was dry of the actual optimum water content (i.e. the optimum water content
measured was not correct), since suction increases as the degree of saturation
Overall the friction angle of the foundry sands is reasonably constant and the
increased cohesion exhibited by the foundry sands (as compared to the reference
sands) are due to higher suctions that result form the inclusion of fines and water.
Table 4.3 summarizes the testing program for the interface direct shear tests.
Multistage interface tests were run for three foundry sands (B, C, and D) and the two
reference sands with the geogrid, geomembrane, and geotextile. Foundry Sand A
was not used in interface testing because it had not yet been acquired. As in the
small-scale direct shear tests, reference soils were placed dry in the shear box at
approximately 90% relative density and foundry sands were compacted at optimum
water content to 100% of the maximum d_ry unit weight. The foundry sands were
tested as-compacted and soaked. Shear stress versus displacement curves are
Shear rate
Normal Stresses (kPa)
(mmlmin) Dry Unit Water
Soil Dry or Weight content1
Geomem- (kN/m3)
Geogrid Geotextile Com- Soaked ("/o)
brane
pacted
Base 30,50, 30,50, 30,50, I - 1 16.9 0
Sand 100 100 100
Portage - 30, 50, - -
1 16.7 0
Sand 100
Foundry 30,50,
S a n d B 100 ( 30, 50,
100 1 30, 50,
100 1 0.05
1 17.3
1 12.0
SandC
1
~ o u n d r y 30, 50,
100
1 309 5 0
100
309 5 0
100
1 1 0.05 18.2 12.3
' ~ o r n ~ a c t i owater
n content
4.2.2 Reference Sands
Results of interface direct shear tests are summarized Table 4.4. The base
sand-geogrid interface had the highest friction angle (32"). Some of this friction was
due to soil dragging across the porous stone substrate through the apertures of the
geogrid. 'The geomembrane (30") had the next highest friction angle with the base
sand, and then the geotextile (28"). The Portage sand-geotextile interface (with
Portage sand substrate) had a friction angle of 32". All of the interface friction
angles are smaller than the friction angles for the sands. Frictional efficiency
(tantiltan$) ranged from 0.59 to 0.69 for the three geosynthetics, which shows that
the base sand has reasonably good frictional interaction with all of the
geosynthetics. The small adhesion intercepts obtained for reference sands are
Interface friction angles ranged from 25" to 35" for the three foundry sands
ranged from 0.51 to 0.86. Adhesion ranged from 4.4 to 22 kPa, with adhesional
The geomembrane gave the highest interface friction angle for Sands B and
D. This was expected since the geomernbrane had the roughest surface of the
Soil Data
Geomembrane Geotextile
C
Geogrid Efficiency Efficiency Efficiency
Soil Fines (kPa) (kpa)
(kPa)
Base Sand 1.1 8.3 2.9 0.35 7.6 0.92 8.2 0.99
Foundry
9.2 17 17 1.OO 4.4 0.26 7.8 0.46
Sand B
Foundry
11 19 14 0.73 10.4 0.53 8.0 0.40
Sand C
Foundry
19 28 22 0.76 8.3 0.27 11.9 0.40
Sand D
,
I
57
the lowest friction angle of the three geosynthetic interfaces tested with Foundry
Sand C.
The highest adhesions were obtained with geogrid interfaces. This was
probably caused by adhesion between the foundry sand and the porous stone
caused by soil suction. Adhesion is higher for the geogrid than the other
geosynthetics because the foundry sand protrudes through the large apertures,
resulting in more contact with the porous stone. In a field situation (under
Table 4.5 surr~marizesresults of the soaked interface tests and includes the
as-compacted test results for comparison. Friction angles ranged from 24" to 35"
for the nine foundry sand-geosyn.thetic combinations tested. Adhesion ranged from
1.6 to 8.8 kPa. The soaked failure envelopes for Foundry Sands C and D gave
representative soaked failure envelope of these sands is presented in Fig. 4.6 for
Foundry Sand D and the geotextile. The soaked failure envelope is parallel to the
Foundry Sand B gave the same friction angles and adhesion intercepts as in the as-
compacted tests. This was expected since Foundry Sand B is has low bentonite
content (2.4%) and has low cohesion. Overall, soaking the foundry sands and
testing them under fully drained conditions caused a decrease in interface adhesion
Table 4.5. Results of Soaked Interface Direct Shear Tests.
Sand C 11 43 32 31 24 28 29 30
Sand D f 19 42 26 25 32 33 29 29
Sand D
-- - -- -
1 --
19
1
- - -- - -28--
- I_- 1
4.4
--- - -
22.0
-- -
1 --
-
4.3
- -- -
] --
7.8
-- 1 -- - 1 ---
11.9
-
- -
0 20 40 60 80 100 120
Fig. 4.6. Failure Envelopes for Sand D-Geotextile Interface Direct Shear Tests
(as-compacted and soaked).
to near the value corresponding to machine friction reported for the base sand
In Sec. 4.2.3, the geogrid interface was reported to have greater adhesion
than the other geosynthetics because of suctions between the foundry sand and the
porous stone. The low adhesions obtained in the soaked geogrid interface tests
4.7, which shows interface friction angle (4.7 (a)) and adhesion (4.7 (b)) versus fines
content. Fig. 4.7 -(a) shows that friction angle is reasonably constant with increasing
fines content. This means that in as-compacted undrained conditions, fines have no
effect on the friction angle of foundry sands. This trend is consistent with that found
for soil-only direct shear tests, also plotted in Fig. 4.7 (a).
and the geosynthetic, as shown in Fig. 4.7 (b). The data in Fig. 4.7 (b) are grouped
into soaked tests and as-compacted tests (and not by geosynthetic type) to
.
emphasize the contrast between interface adhesion in the soaked and as-
increasing fines and is comparable to the -"machine friction." In contrast, for the as-
adhesion with increasing fines content that is above the machine friction for fines
A Soil Geosynthetic Interfaces
~ O ~ I I I I I I I I ~ I I I I ~ ~ ~ I I ~
% Fines
As-Compacted GG Interface
As-Compacted lnterfaces
Soaked lnterfaces
0 5 10 15 20
C
% Fines
Fig. 4.7. Effect of Fines Content on (a) Interface Frictional Angle and (b) Adhesion
62
content > 6.5. The as-compacted geogrid interface showed the greatest increase in
adhesion with increasing fines, near that of the soil alone. 'These trends are
interfaces resulting from suction in the soil and porous stone between the geogrid
apertures, and the lack of adhesion for the other geosynthetic interfaces. That is,
suction can only cause an increase in adhesion at the interface between two porous
Table 4.6 summarizes the pullout testing program. This program included
three foundry sands and the base sand with the geogrid and geotextile. Foundry
Sands A and C were compacted to about 92% of the maximum standard Proctor dry
unit weight at optimum water content. Foundry Sand D was compacted to about
850h of the maximum dry unit weight. These were the maximum dry unit weights
that could be reasonably achieved with the available compaction methods. All tests
-
Typical graphs of pullout force per unit width versus displacement are
presented for Foundry Sand A-geogrid in Fig. 4.8 (a) and Foundry Sand C-geotextile
in Fig. 4.8 (b) for all three normal stresses. Graphs of pullout force per unit width
versus displacement for the remainder of the tests are given in Appendix D and
Table 4.6. Pullout Testing Program.
Dry Water
Normal Stresses Shear rate
Soil Density content'
(kpa) (mmlmin)
(kN/m ) (O/o)
Base
Sand 1
Foundry
Sand A
1 10,30, 50 1
Foundry
Sand C
Foundry
Sand D
1 I I I
Displacement (mrn)
Fig. 4.8 Typical Plots of Pullout Force per Unit Width versus Displacement for
(a) Sand A-Geogrid and (b) Sand C-Geotextile
65
Table D-1 summarizes 'the maximum pullout force unit width for each test. In
general, the pullout force increases rapidly at low displacements (< 20 mm) and
normal stresses, the pullout force becomes constant because the entire
geosynthetic is displaced and thus shear strength is completely mobilized along the
entire length of the geosynthetic. At higher normal stresses, the pullout force
continually increases due to progressive failure, meaning that only a portion of the
non-uniform. At a normal stress of 50 kPa, both the geogrid and geotextile broke.
Geosynthetic failure at a normal stress of 50 kPa occurred in the geogrid for all
sands and in the geotextile for all sands except Foundry Sand A.
Results of the pullout tests are summarized in Tables 4.7 - 4.8. The
maximum pullout force was used in all calculations of interaction coefficients. The
interaction coefficients (Ci) were calculated with and without soil cohesion (Eqs. (3.1)
and (3.2)) for each foundry sand. All calculations of Ci for base sand were made
without cohesion using Eq. (3.1). In all cases, the interaction coefficients for the
The higher interaction coefficients obtained for the geogrid are believed to be
C
due to shear strength being mobilized along a greater length of the geosynthetic
than occurred for the geotextiles. The mobilized length (L,) was estimated by
plotting normalized nodal displacement versus distance from the front of the pullout
Table 4.7. Pullout-Test lnteraction Coefficients (Calculated with Cohesion).
Geogrid Geotextile
Soil % Fines
10kPa 130kPa150kPa 10kPa 1 3 0 k P a / 50kPa
1
I
Foundry
6.5 0.33 0.31' 0.28' 0.24 0.23 0.20
Sand A
Foundry
11.3 0.48 0.45 0.40~ 0.38 0.33 0.26'
Sand C
Foundry
19.2 0.40 0.48 0.43' 0.28 0.28 0.22'
Sand D
'~eosyntheticbroke
2
Estimated for 50 kPa by extrapolating data
Geogrid Geotextile
Soil % Fines
10kPa
10k~a130k~al50k~ a 1 3 0 k ~ a l 50kPa
Base Sand 1.1 0.98 0.72 0.80 0.98 0.39 0.27'
Foundry
6.5 1.I 0.56' 0.42' 0.80 0.41 0.30
Sand A
Foundry
11.3 1.5 0.98 0.71~ 1.2 0.56 0.36'
Sand C
Foundry
19.2 1.7 0.98 0.70' 1.2 0.58 0.35'
Sand D
'Geosynthetic broke
2~stimatedfor 50 kPa by extrapolating data
67
box as shown in Fig. 4.9. IVormalized nodal displacements (at the end of the test)
LVDTs by those monitored by the front end LVDT. The estimated L, for each
pullout test are summarized in Table 4.9. In every test, the geogrid had a greater L,
and thus a higher Ci. Fig. 4.10 also shows that Ci is strongly correlated with L
,, with
Figs. 4.1 1 - 4.14 show Ci (calculated with and without soil cohesion) versus
normal stress for the geogrid and geotextile. In general, Ci decreases with
coefficients for the foundry sands calculated with soil cohesion are < 0.5, whereas
the interaction coefficients are often > 1 when soil cohesion is ignored. Also, the
typical drop in Ci with increasing normal stress is not apparent when cohesion is
are obtained without cohesion because the cohesion was large relative to the
frictional component of the soil shear strength, particularly at low normal stresses.
interface in a pullout test, since the suction in the foundry sand that is responsible
for soil cohesion is not transferred to the geosynthetic interface. This concept is
consistent with the low adhesional efficiency measured for the foundry sand-
geosynthetic interfaces in direct shear able 4.4). Thus, calculating Ci without soil
cohesion is probably more realistic for most foundry sands with clay binder.
Estimated
I . ,
I l l J I r ! J d ' l
Fig. 4.9. Estimation of Mobilized Length of Geosynthetic from Telltail LVDT Data.
Table 4.9 Summary of Pullout Mobilized Lengths.
Foundry
2.2 131 80 60 100 60 40
Sand A
Foundry
11.3 131 131 90 80 50 35
Sand C
Foundry
19 131 131 90 80 50 40
Sand D
Geogrid 0 Geotextile
0.0
0.0 0.2 0.4 0.6 0.8 1.O 1.2
Fig. 4.1 1. lnteraction Coefficient versus Normal Stress for the Geogrid.
2.0 I l r l l l l l l , l , l l l l l l l I l l l l , l l l l
-
-
-
- Q -w- Sand A
Fig. 4.12. lnteraction Coefficient without Cohesion versus Normal Stress for the
Geogrid.
l I I . , \ l l l ~ l l l ~l l l l l ~ l l l l ~ l l l l
-
-
-
-It-- Sand A
-
- - - 0-Sand
- C
-
-
-
C -
Fig. 4.14. lnteraction Coefficient without Cohesion versus Normal Stress for the
Geotextile.
4.3.4 Effect of Fines Content on Pull-Out Strength
lnteraction coefficient versus fines content is shown in Fig. 4.15 for the
geogrid and in Fig. 4.1 6 for the geotextile. lnteraction coefficients calculated with
and without cohesion are shown. Figs. 4.15 (a) and 4.16 (a) show that the Ci for the
base sand (1.1OO/ fines) is significantly larger than Ci for the foundry sands (> 6.5%
fines) when cohesion is included, and that Ci is similar for all three foundry sands.
Figs. 4.1 5 (b) and 4.16 (b) both show in contrast, that when Ci is calculated without
cohesion, it is essentially the same for all sands at a given normal stress. 'Thus, if
be used for foundry sands having a fines content in the range of fines contents in
4.4DESIGN RECOMMENDATIONS
These inputs are based on the test results from this project and are intended to be
used as general guidelines and for preliminary design. They are not specific design
inputs for final design. Tests on the specific soil and geosynthetic being considered
C
should be conducted during final design. Furthermore, these guidelines are based
general acceptance.
% Fines ( c 0.075 mm)
Fig. 4.15. Interaction Coefficient versus Fines for Geogrid: (a) Ci with Cohesion
Ci without Cohesion.
% Fines ( c 0.075 rnm)
Fig. 4.1 6. Interaction Coefficient versus Fines for Geotextile: (a) Ci with Cohesion
(b) Ci witliout Cohesion.
4.4.1 Mechanically Stabilized Earth (MSE) Walls
the geosynthetic, which must be sufficient to provide resistance to the active earth
pressures exerted on the wall facing attached to the geosynthetics. The anchorage
where Lan is the anchorage length, d, is the depth of overburden over the
geosynthetic, yr is the unit weight of the reinforced soil, q is the surcharge and @, is
recommended for foundry sands. The design interaction coefficients for foundry
sand (Cid) were determined by taking an average Ci from the three normal stresses.
computing Cid. The normal stresses used in this study (10, 30, 50 kPa) represent
The foundry sands in Table 4.10 were grouped according to fines contents
since Foundry Sand A (fines content = 6.5%) gave lower interaction coefficients
than Foundry Sands C and Dl which contain more fines. The geogrid interaction
Table 4.1 0. Interaction Coefficients for Reinforced Retaining Wall.
Geogrid Geotextile
Soil Type
Ci Source Ci Source
Foundry Sand A
0.7 Test Data 0.5 Test Data
(SP-SM)
Foundry Sands C
and D 0.9 Test Data 0.6 Test Data
(SW-SM, SC)
coefficients recommended for the base sand (0.8), Foundry Sand A (0.7), and
geogrid mar~ufacturerfor similar soils. The calculated Ci for the geotextile (0.5 - 0.6)
may be due to lower stiffness of the geotextile (Belton 113), which did not permit
Leshchinsky and Perry (1987) suggest that proper drainage requires that the
soil is less permeable. Leshchinsky and Perry also recommend that the reinforced
soil should have less than 12OI0fines to avoid migration of fines into the geotextile by
seeping water. Retention of fines can also be assessed by conducting gradient ratio
tests (ASTM 05101) or by checking particle size criteria (Christopher and Holtz
1989) for the foundry sand and geotextile being considered. When designing
-
should be considered, since foundry sands often have greater than 12O/0 fines
failure) or pullout of the geosynthetic. Thus, both interface shear strength and
interaction coefficients are necessary for design. The design recommendations are
summarized in Tables 4.1 1 - 4.12. For interface strength, only friction angles are
given (Table 4.1 1). Adhesion should be ignored since it is generally small, and is
The recommended interaction coefficients in Table 4.12 are lower than those
only one geosynthetic layer which is likely to be at a greater depth. 'Thus, the design
Ci should be lower since the normal stress will be higher. Recorr~mendedCi for a 3-
m high ernbankment are shown in Table 4.12. In three of these cases the Ci would
be less than 0.5 for a 3-m embankment. Thus, the maximum embankment height
design of a landfill cover or liner requires interface friction angles between the cover
soil and the geosynthetics for use in stability analysis. An average 6 = 31" and
frictional efficiency of 0.65 were obtained for the textured geomembrane interface
with the two foundry sands having sufficiently low hydraulic conductivity to be used
as landfill cover soils. This friction angle is higher than that typically found for clay-
Table 4.1 1. Interface Friction Angles for Reinforced Embankments.
Geotextile
' 1
(")
Efficiency / Source
Silty Sand (
I
- 1
I I
23 0.87
Koerner
(1998)
Rounded I - Koerner
Sand I - I I (1998)
Angular
Koerner
Sand 35 - 42' 0.72 - 0.93
(1998)
ISW)
Sand
1 30 1 0.68 1 Test Data Test Data
Foundry
sand2
I
I
29 1
I
0.62 ( Test Data.
I
Test Data
Soil
.
Geogrid Ci Geotextile Ci
Adhesion should be ignored in design since the adhesion is typically low, and will
CONCLUSIONS
The objective of this study was to characterize the strength of foundry sands
and their interaction with geosynthetics to facilitate the reuse of foundry sands in
structural fill and in landfill covers. Direct shear tests, interface shear tests, and
then made for design parameters based on the results of these tests.
Results of the direct shear tests showed that as-compacted foundry sands
have friction angles ranging from 39 to 50" under undrained shear. These friction
angles similar to the friction angle for dry base sand (42"). An exception is the
friction angle for Foundry Sand A, which had a larger coarse fraction.
The cohesion ranged between 16.7 and 28.4 kPa and generally increased
with increasing fines content. Larger suctions obtained at higher bentonite contents
are probably responsible for the greater cohesion. However, under fl-lily drained
direct shear box. As-compacted undrained and soaked drained tests were
83
performed. The tests were conducted using a multistage set-up, meaning that one
Interface direct shear tests yielded interface friction angles ranging from 25"
to 35" for the three foundry sands tested as-compacted with three geosynthetics.
Soaking the foundry sands and testing them under fully drained conditions did not
affect the friction angle. Frictional efficiency was reasonably constant with
increasing fines.
Adhesion was low for the geotextile and geomembrane interfaces, but was
near the cohesion of the soil for the geogrid. Higher adhesion was obtained with the
geogrid because the foundry sand protrudes through the large apertures, resulting in
more contact with the porous stone. Soaking caused a decrease in interface
adhesion to near the value corresponding to machine friction reported for the base
drainage condition and the geosynthetic. For the soaked interfaces, adhesion
interfaces, there is a slight increase in adhesion for fines contents > 6.5. The as-
sands with a woven geotextile and a geogrid. The pullout tests were conducted
showed that the pullout force increased rapidly at low displacements (< 20 mm) and
normal stresses, the pullout force became constant because the entire geosynthetic
was displaced and thus the shear strength was mobilized along the entire length of
the geosynthetic. At higher normal stresses, 'the pullout force continually increased
Interac.tion coefficients for foundry sands ranged from 0.20 to 0.48 when
calculated with cohesion and from 0.30 to 1.7 when calculated without cohesion.
Calculating Ci without soil cohesion is more realistic for most foundry sands with clay
interface in a pullout test. Higher interaction coefficients were obtained for the
geogrid than the geotextile, probably due to shear strength being mobilized along a
greater length of the geogrids than occurred for the geotextiles. For both the
geotextile and geogrid, Ci generally decreases with increasing normal stress due to
5.4DESIGN RECOMMENDATIONS
walls are 0.8 for base sand, 0.7 for Foundry Sand A, and 0.9 for Foundry Sands C
the geogrid manufacturer for similar soils. For the geotextile, the recommended Ci
85
was lower than the manufacturers' suggestion, possibly due to the stiffness of the
geotextile (Belton 113), which did not permit enough mobilization of shear strength.
5.4.2 Embankments
For lateral spreading analysis an interface friction angle of 30" can be used
for foundry sands with geotextiles and geogrids. Adhesion should be ignored since
Interaction coefficients ranging from 0.5 to 0.7 are recommended for geosynthetic
pullout analysis of embankments. These recommendations are lower than for MSE
walls because of the higher normal stresses that will likely be exerted on the
geosynthetics.
The interface friction angle between the two foundry sands with low hydraulic
conductivity (C and D) and the textured geomembrane was 31". This friction angle
is higher than that typically found for clay-textured geomembrane interfaces, but is
suitable for use in preliminary design. Adhesion can be ignored in design for
REFERENCES
Abichou, T., Benson, C., and Edil, T. (1998a), "Beneficial Reuse of Foundry Sands
in Construction of Hydraulic Barrier Layers." Environmental Geotechnics
Report 98-2, Dept. of Civil and Environmental Engineering, University of
Wisconsin-Madison.
Abichou, T., Benson, C., and Edil, T. (1998b), "Database on Beneficial Reuse of
Foundry By-Products." Environmental Geotechnics Report 98-2, Dept. of Civil
and Environmental Engineering, University of Wisconsin-Madison.
Bergado, D. T., Werner, G., Tien, M. H., and Zou, X. H., (1995), "Interaction
Between Geotextiles and Silty Sand by Large Direct Shear and Triaxial
Tests," Proceedings of the Geosynthetics '95 Conference, Nashville, TN , pp.
1097-1 109.
Farrag, K., (1993), "Evaluation of the Effect of Moisture Content on the Interface
properties of Geosynthetics," Proceedings of the Geosynthetics '93
Conference, Vancouver, Canada, pp. 1031-1 041.
Fredlund, D. G., and Rahardjo, H. (1993), Soil Mechanics for Unsaturated Soils,
John Wiley & Sons, Inc., New York.
Lee, K. L., and Seed, H. K., (1967), "Drained Strength Characteristics of Sands,"
Journal of the Soil Mechanics and Foundations Division, ASCE, Vol. 93, No.
SM6, pp. 1 1 7 - 141.
Leshchinsky, D., and Perry, E. B. (1987), "A Design Procedure for Geotextile-
Reinforced Walls," Geosynthetics '87, IFAI, St. Paul, MN, pp. 95 - 107.
Lopes, M. L., and Ladeira, M., (1996), "Role of Specimen geometry. Soil Height, and
Sleeve Length on the Pull-Out Behaviour of Geogrids," Geotextiles and
Geomembranes, Vol. 3, , No. 6, pp. 701-719.
Lupini, J. F., Skinner, A. E., and Vaughan, P. R. (1981), "The Drained Residual
-
Strength of Cohesive Soils," Gebtechnique, Vol. 31, No. 2, pp. 181 - 213.
Mallick, S. B., Shai, H., Adanur, S., and Elton, D. J., (1996), "Pullout and Direct
Shear Testing of Geosynthetic Reinforcement: State-of-the-Art Report,"
Transportation Research Record, No. 1534, pp. 80-90.
Mast, D. G. (1997), "Field Demonstration of a Highway Embankment Using Waste
Foundry Sand," Master's Thesis, Purdue University, West Lafayette, IN,
1997.
Mirafi (1990) "Design Methodology for Miragrid Reinforced Retaining Walls," Mirafi,
Inc. Charlotte, NC.
Mitchell, J. K. (1993) Fundamentals of Soil Behavior, John Wiley & Sons, Inc., New
York.
Nataraj, M. S., Maganti, R. S., and McManis, K. L., (1995) "lnterface Frictional
Characteristics of Geosynthetics," Proceedings of the Geosynthetics '95
Conference, Nashville, 'TN, pp. 1057-1069.
Takasumi, D. L., Green, K.R., and Holtz, R.D., (1991) "Soil-Geosynthetics lnterface
Strength Characteristics: A Review of State-of-the-Art Testing Procedures"
Proceedings of the Geosynthetics '9 1 Conference, Atlanta, Georgia, pp. 87-
100.
C
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
L
1 l 1 1 l 1 l l l l ! l l I l i l l l ! l l l l l l I -
0 10 20 30 40 50 60
Displacement (rnrn)
Fig. B-1. Shear Stresr-Displacement Curves for (a) Base Sand-Geogrid and (b)
Base Sand-Geomembrane.
0 10 20 30 40 50 60
Displacement (mm)
Displacement (mm)
Fig. B-3. Shear Stress-Displacement Curves for Foundry Sand B-Geogrid: (a) As-
compacted and (b) Soaked.
. Displacement (mm)
on= 30 kPa 0 o n =50 kPa
Displacement (mm)
Fig. 8-4. Shear Stress-Displacement Curves for Foundry Sand 6-Geomembrane:
(a) As-compacted and (b) Soaked.
Displacement (mm)
03 on= 30 kPa o on= 50 kPa
Displacement (mm)
Fig B-5. Shear Stress-Displacement Curves for Foundry Sand B-Geotextile: (a) As-
compacted and (b) Soaked.
0 10 20 30 40 50 60
Displacement (mm)
'Displacement (mm)
Fig. B-6. Shear Stress-Displacement Curves for Foundry Sand C-Geogrid: (a) As-
compacted and (b) Soaked.
0 10 20 30 40 50 60
Displacement (mm)
Displacement (mm)
Fig. B-7. Shear Stress-Displacement Curves for Foundry Sand C-Geomembrane:
(a) As-compacted and (b) Soaked.
Displacement (mm)
Displacement (mm)
Fig. 6-8. Shear Stress-Displacement Curves for Foundry Sand C-Geotextile: (a) As-
compacted and (b) Soaked.
I i i l l 1 1 1 1 1 l 1 1 i I ~ 1 1 1 , 1 1 1 1
0 10 20 30 40 50 60
Displacement (mm)
Displacement (mm)
Fig. B-9. Shear Stress-Displacement Curves for Foundry Sand D-Geogrid: (a) As-
compacted and (b) Soaked.
0 10 20 30 40
Displacement (mm)
Displacement (mm)
Fig. B-10. Shear Stress-Displacement Curves for Foundry Sand D-Geomembrane:
(a) As-compacted and (b) Soaked.
Displacement (mm)
Displacement (mm)
Fig. B-11. Shear Stress-Displacement Curves for Foundry Sand D-Geotextile: (a)
As-compacted and (b) Soaked.
APPENDIX C
Geogrid Geotextile
Soil % Fines
10k~a130k~a150k~
10kPa
a 1 3 0 k P a 50kPa
Foundry
6.5 34 522 642 20 38 43
Sand A
Foundry
11.3 36 50 61 29 41 442
Sand C
Foundry
19.2 39 70 822 27 41 41
Sand D
Geosynthetic broke
3~stimatedfor 50 kPa by extrapolating data
Displacement (mm)
Displacement (mm)
Fig. D-1. Pullout Force per Unit Width versus displacement for (a) Base Sand-
Geogrid and (b) Base Sand-Geotextile.
Displacement (mm)
Displacement (mm)
Fig. D-2. Pullout Force per Unit Width versus displacement for (a) Foundry Sand A-
Geogrid and (b) Foundry Sand A-Geotextile.
l l I l I l l I l l l l i l l l l l l l l 1 1
0
0 20 40 60 80 100 120
Displacement (mmj
Displacement (mm)
Fig. D-3. Pullout Force per Unit Width versus displacement for (a) Foundry Sand C-
Geogrid and (b) Foundry Sand C-Geotextile.
Displacement (mm)
Displacement (mm)
Fig. D-4. Pullout Force per Unit Width versus displacement for (a) Foundry Sand D-
Geogrid and (b) Foundry Sand D-Geotextile.