;an investigation of the effectiveness of
TENSILE REINFORCEMENT IN STRENGTHENING
AN EMBANKMENT OVER SOFT FOUNDATION
B
B.S.
by
Barry Duane Siel
.S. Geology, Oregon State University, 1978
C.E., University of Colorado at Denver, 1984
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Department of Civil Engineering
1986
This thesis for the Master of Science degree by
Barry Duane Siel
has been approved for the
Department of
Civil Engineering
by
Date.
Siel, Barry Duane (M.S., Civil Engineering)
An Investigation of the Effectiveness of Tensile
Reinforcement in Strengthening an Embankment
Over Soft Foundation
Thesis directed by Assistant Professor Tzong H. Wu
The purpose of this study was to investigate
the effectiveness of placing a single layer of tensile
reinforcement at the base of an embankment constructed
on a soft foundation in order to strengthen an embankment
and decrease detrimental settlement. The use of two
types of reinforcement was investigated; Tensar SS2
geogrid and Hoechst Trevira 1127 geotextile. The
Tensar product was a polypropylene geogrid and the
Trevira was a nonwoven, polyester, needle punched
geotextile.
Because it was not practical to construct
multiple embankments, a study comparing the results
when using no reinforcement to using geogrid to using
geotextile was done using finite element analysis.
The Colorado Department of Highways constructed an
embankment on a soft foundation that was reinforced
with a layer of Tensar SS2 geogrid placed at the
base of the embankment. This embankment was instrumented
to measure movements in the foundation, both vertical
and horizontal. The field measurements and lab results
IV
from this project were used to determine and refine
the soil parameters used in the finite element analysis
The use of a single layer of Tensar SS2 geogrid did
not significantly reduce the settlement of the embankment,
compared to using a single layer of Trevira 1127 geotextile
or using no reinforcement. This was due in part to
insufficient stiffness and probably due to the fact
that the deformation of the embankment was predominantly
in the vertical direction.
CONTENTS
CHAPTER
1. INTRODUCTION.......................... 1
. Problem Statement..................... 1
Objective of the Study................. 2
Method of Approach..................... 2
2. PERFORMANCE OF THE GEOGRID
REINFORCED EMBANKMENT.................. 4
The Foundation......................... 4
Role of Tensile Reinforcements......... 5
The Geogrid............................ 7
Instrumentation and Field
Measurements....................... 10
3. FINITE ELEMENT ANALYSIS OF THE
GEOGRID REINFORCED EMBANKMENT......... 14
The CANDE Code....................... 14
The DuncanChang Model................ 16
Finite Element Discretization......... 20
Geogrid Parameters.........,........ 24
Soil Parameters....................... 24
SoilGeogrid Composite Parameters... 25
Results and Discussion of Results... 29
4. PREDICTED PERFORMANCE OF
THE EMBANKMENT...................... 3 3
Background.......................... 3 3
vi
The Geotextile. ................ 33
The Test Apparatus.................... 34
InSoil Geotextile
StressStrain Test................ 3 6
SoilGeotextile Composite
Parameters.......................... 40
Results and Discussion of Results... 41
5. SUMMARY AND CONCLUSIONS................ 58
Summary............................... 58
Conclusions........................... 59
REFERENCES.................................... 61
VI1
TABLES
Table
1. ' Laboratory Results for Cement Wash
Material............................... 6
2. Specifications for ABC Material......... 10
3. Initial DuncanChang Soil Parameters... 26
4. DuncanChang Parameters for
SoilGeogrid Composite................ 27
5. Final DuncanChang Soil Parameters
for the Cement Wash and
SoilTrash Materials.................. 29
6. Secant Moduli at 2% Strain for
InSoil Geotextile................... 40
7. DuncanChang Parameters for
SoilGeotextile Composite.............. 41
FIGURES
Figure
1. Tensar SS2 Geogrid........................ 9
2. Instrumentation at Embankment Site...... 11
3. Stratigraphic CrossSection............... 21
4. Finite Element Mesh....................... 23
5. Load vs. Displacement Relationship
for SoilGeogrid Composite.............. 28
6. CANDE Generated and Field
Settlement Curves....................... 31
7. Composite and Discrete Settlement
Curves of the Geogrid
Reinforced Embankment. ................ 32
8. InSoil Geotextile StressStrain
Test Box................................ 35
9. Grain Size Distribution for
#30 Ottawa Sand........................ 37
10. Load vs. Displacement Relationship
for InSoil Geotextile.................. 39
11. Load vs. Displacement Relationship
for SoilGeotextile Composite........... 42
12. Settlement Curves of the Embankment
Reinforced with Geogrid
and Geotextile...,...................... 44
13. Settlement Curves of the Embankment
Reinforced with Geogrid and
Without Reinforcment.................. 45
14. Composite and Discrete Settlement
Curves of the Geotextile
Reinforced Embankment.................. 46
ix
15. Settlement Curves of the Embankment
Reinforced with Geogrid and Steel.... 47
16. Principal Stresses with Geogrid
Reinforced Embankment.................. 49
17. Principal Stresses with Unreinforced
Embankment............................ 50
18. Principal Stresses with Geotextile
Reinforced Embankment.................. 51
19. Principal Stresses with Steel
Reinforced Embankment.................. 52
20. Elements with Tensile Stresses with
Geogrid Reinforced Embankment.......... 53
21. Elements with Tensile Stresses with
Unreinforced Embankment................ 54
22. Elements with Tensile Stresses with
Geotextile Reinforced Embankment..... 55
23. Elements with Tensile Stresses with
Steel Reinforced Embankment............ 56
24. Thrust Distribution in
Reinforcements......................... 57
CHAPTER 1
INTRODUCTION
Problem Statement
When a new highway is planned for an urban
or suburban area, a route with the least amount of
development is preferred for the rightofway. Because
of this undesirable foundation conditions for embankment
construction often result. Effective and economical
solutions to instability and excessive settlement
of embankments constructed on soft foundations are
common geotechnical engineering problems.
A number of methods have been developed and
used to improve stability and decrease detrimental
settlement of embankments constructed on soft foundations.
Most of the methods involve modification of the foundation
material. Examples of these methods are preloading
with or without drains, overexcavation and backfill
with structural fills, compaction grouting, vibroflotation,
compaction piles, electroosmosis, chemical additives,
and thermal treatment.
In recent years, however, the concept of
strengthening the embankment by including tensile
reinforcement near its base has begun to gain popularity
2
(Barsvary et al. 1982; Hutchins 1982 and Petrik
et al. 1982) .
Objective of Study
The Colorado Department of Highways had const
ructed a highway embankment on a soft foundation that
was reinforced with Tensar SS2 geogrid. This embankment
was well instrumented to measure vertical and horizontal
movements in the foundation.
The objective of this study was to investigate
the effectiveness of using a single layer tensile
reinforcement to stengthen the embankment. The
effectiveness of two different types of tensile
reinforcements, Tensar SS2 geogrid and Trevira 1127
geotextile, was investigated.
Method of Approach
The best way to conduct this investigation
would have been to construct three identical embankments,
one using geogrid, one using geotextile and the third
using no tensile reinforcement, and compare the performance
of each against the others. Since this was both
impractical, because it would have been difficult
to assure identical foundation conditions for each
embankment, and prohibitively expensive, another approach
was taken. A numerical method which is capable of
I
I 3
analyzing the problem in the most realistic manner,
finite element analysis, was used in this investigation.
The finite element program used for this investigation
was CANDE.
In this study, the field measurements, were
compared to the finite element analysis results and
the soil parameters input into the program were refined
so that the computer generated settlement curve was
in reasonable agreement with the field measurement
data. All the settlement curves in the figures were
drafted in reference to a line coinciding with the
geogrid layer. The computer program, with the refined
soil parameters, was used to predict the performance
of the embankment with the geotextile tensile reinforcement
and with no reinforcement to evaluate the effectiveness
of using the reinforcements.
Two methods of simulating the inclusion of
the tensile reinforcements in the analysis were used.
The first method was to use insoil tensile reinforcement
properties and input the reinforcement as discrete
beam elements. The .other method was to determine
soilreinforcement composite properties and'incorporate
them in the analysis as strengthened soil elements,
Herrmann and AlYassin (1978).
CHAPTER 2
PERFORMANCE OF THE GEOGRID REINFORCED EMBANKMENT
The Foundation
The location where the embankment was built
was in the flood plain of Clear Creek, west of Tennyson
Street in Denver, Colorado. The site had been used
as a source of gravel for the Owens Brothers' concrete
operation. The abandoned gravel pit, which had filled
with water from nearby Clear Creek, was later used
as a dump for the cement wash waste from the cleaning
of cement trucks. This resulted in a thick, stratified
layer of soft, high void ratio, saturated cement muck.
As the pond was filled with cement wash, the upper
part of the cement wash material hardened to form
a crust which became strong enough to drive vehicles
over. Later, the pit was covered over with fill dirt
and organic and inorganic trash.
The surrounding area was a flat to gently
rolling flood plain deposit of Clear Creek. Subsurface
materials in the unmined areas consisted of up to
20 feet of sandy cobbly gravel underlain by grey shale
bedrock. The water table was found to be approximately
8 to 10 feet below the present topographic surface.
5
Laboratory tests were performed on samples
of the cement wash material including CD, CU and UU
triaxial tests and a onedimensional, consolidation
test. Results from these tests, shown in Table 1,
indicated that the cement wash material could pose
stability and differential settlement problems.
Because of the high coefficient of consolida
tion the cement wash material would gain strength
quickly upon loading. It was decided by the Colorado
Department of Highways that by using staged construction
and tensile reinforcement at the base of the embankment
the problem could be solved.
Role of Tensile Reinforcement
Embankments constructed on soft foundations
often require reinforcing to maintain stability.
Many materials have been used to reinforce such embankments
including woven and nonwoven fabrics, plastic and
steel nets and grids, used automobile tire casings,
steel landing mats, fasine mats and reinforced plastic
and rubber membranes (Boutrup and Holtz 1983). For
a reinforcement to be effective it must satisfy two
primary requirements. It must have sufficiently high
deformation modulus in tension and it must be able
to develop sufficient frictional resistance with the
surrounding soils.
6
TABLE 1
Laboratory Results for Cement Wash Material
1. Physical Properties
(a) Description: Sandy silt, very soft, white
(b) AASHTO Classification: A4(0)
(c) Specific Gravity = 3.11
(d) Density: Wet = 7380 pcf; Dry = 1933 pcf
(e) Void Ratio = 4.79.3
(f) Plastic Index = 0
(g) Moisture Content = 39% to 93%
(h) pH = 8.1
(i) Chemical Composition: Silicon 10%, Calcium
10%, Aluminum 5%, Iron 2%, Sodium 1%, Sulfate
.148%, Chlorides .015%
2. Triaxial Tests
(a) CD test: C = 0 psf, 0 = 41
(b) CU test: C = 0 psf, 0 = 41
C = 0 psf, 0 = 22
(c) UU test: C = 130 to 460 psf, 0=0
3. Consolidation Test
(a) Compression Index = 3.6
(b) Recompression Index =0.1
(c) Coef. of Consolidation = .0152.0235 in2/min.
(d) Max. Preconsolidation Pressure = 0.8 tsf
(e) Immediate Settlement = 50% of Total Settlement
(f) Primary Consolidation = 40% of Total Settlement
(g) Secondary Compression = 10% of Total Settlement
7
When the reinforcement is placed horizontally
between the embankment and the weak foundation, the
primary roles of the reinforcement are: 1. to increase
the factor of safety against slope stability failure
through the foundation, 2. to increase the bearing
capacity of the foundation, and 3. to reduce the lateral
spreading of the embankment. In many cases the foundation
is not laterally homogeneous and weak lenses or soft
pockets will exist. When this is the case, the
reinforcement will serve to 1. bridge the weak spots
and redistribute the loads thereby reducing differential
settlements, and 2. increase the shear strength near
the embankmentfoundation interface and thereby reduce
the risk of localized failure.
In view of the preceeding requirements for
a tensile reinforement, the Colorado Department of
Highways chose to use Tensar SS2 geogrid to reinforce
the highway embankment constructed at 176 and Tennyson
Street. On a weight basis, geogrid is as strong as
steel but costs less and has much higher resistance
to corrosion and bacterial action than most conventional
materials including metals.
The Geoqrid
Tensar SS2 geogrid is an oriented, polypropylene,
high strength grid manufactured for use in heavy
8
construction applications such as reinforcement, support,
containment and enclosure, Figure 1. It is manufactured
by a process that aligns long chain hydrocarbon molecules
into continuous geometric patterns. A sheet of
polypropylene is punched with a pattern of holes and
then stretched in two orthogonal directions, the result
being a product with high tensile strength and modulus
in two perpendicular directions.
The geogrid used in the embankment at 176
and .Tennyson Street was placed with the machine direction
(see Figure 1) parallel to the axis of the embankment.
Therefore, the cross machine direction strength parameters
were used in this analysis. SS2 geogrid is available
in sections that are 164 feet long and 9.8 or 13.1
feet wide. When a large area is covered, such as
the embankment foundation in question, the geogrid
sections are overlapped and sandwiched between two
layers of coarse, granular material. The effect of
the rectangular apertures of the geogrid and the granular
material is an interlocking that results in the geogrid
performing as a single unit. The granular material
used for this project was an ABC material with
specifications shown in Table 2.
ROLL LENGTH
Figure 1. Tensar SS2 Geogrid
VO
10
TABLE 2
Specifications for ABC Material
Sieve
% Passing
2 inch
3/4 inch
#4
85100
4590
2050
530
010
#30
#200
The geogrid was sandwiched between two 8 inch
thick layers of ABC material. The overlap of the
geogrid was two feet on the sides and ends. The section
geogrid size used was 164 feet by 13.1 feet.
Instrumentation and Field Measurements
Prior to the construction of the embankment
at 176 and Tennyson Street, instrumentation was installed
to measure vertical and horizontal movements and porewater
presures in the foundation. The instrumentation,
(see Figure 2) consisted of six piezometers, four
vertical inclinometers, one horizontal inclinometer
and three liquid settlement transducers. The piezometers
were installed at various depths and locations so
as to monitor the porewater pressure dissipation in
both the cement wash and soiltrash materials. This
allowed for the monitoring of the consolidation process
during and after the construction of the embankment..
12
The vertical inclinometers were installed one at either
end of the horizontal inclinometer and along the south
side of the embankment. The construction area was
bordered on the south side by Clear Creek and there
was* some concern about possible lateral movement of
the foundation along the south side. The horizontal
inclinometer was placed on a four inch bed of squeegee
material in a three foot by two foot deep trench that
was dug across the entire width of the embankment
site at station 194+60. The trench was then backfilled
with squeegee material. Two by two foot pits, two
feet deep, were dug for the placement of the liquid
settlement transducers at various locations including
one next to the horizontal inclinometer so the settlement
measurements could be compared. The pits were backfilled
with squeegee material.
Initial field measurements were taken and
then, once construction of the embankment began, field
measurements were taken about once per week for all
instruments except the horizontal inclinometer from
which data was collected about once a month.
Data from the piezometers were used to determine
when primary consolidation had been completed for
each stage of construction. Results from the horizontal
inclinometer and the one liquid settlement transducer
were in reasonable agreement.
13
All instrumentation was manufactured and supplied
by Slope Indicator Company of Seattle, Washington.
CHAPTER 3
FINITE ELEMENT ANALYSIS OF THE GEOGRID
REINFORCED EMBANKMENT
The CANDE Code
CANDE, (Culvert ANalysis and DEsign) was
developed for the Federal Highway Administration in
an effort to present a unified computer methodology
for the structural design, analysis and evaluation
of buried culverts made of corrugated steel, aluminum,
reinforced concrete and a class of plastic pipe by
Katona et al. (1976). The program has three solution
levels corresponding to successive increases in analytical
sophistication. Level one is a modified elasticity
solution applicable for round pipes deeply buried
in homogeneous soil. Level two provides a canned
finite element mesh applicable for most symmetric
culvert installations, including trench and embankment
conditions. Level three permits consideration of
any arbitrary culvert configuration, however, the
finite element mesh must be defined by the user (Katona
and Smith 1976). Since solution level three requires
the user to define the finite element mesh, it allows
for a great deal of versatility in the configuration
15
of the soil and the structure. This means CANDE can
be used in the design and analysis of soilstructure
problems other than culverts, such as retaining walls
and reinforced embankments.
Other finite element programs are available
that can handle the soilstructure problem presented
in this study, such as SSTIP, but CANDE is one of
the best available finite element programs for soil
structure problems (Leonards et al. 1982). The solution
of nonlinear problems by the finite element method
is usually attempted by one of three basic techniques,
incremental, iterative and mixed or incrementaliterative.
The mixed method, which is the method used by CANDE,
generally provides the best solution in geotechnical
applications. This is because incremental technique
is required to simulate sequential construction of
earth structures and iterative procedure is needed
to ensure covergence of the solution. CANDE allows
the user to chose from between five different models
for the modeling of the structure. CANDE also allows
the user to select the model to be associated with
each soil type. The soil models available include
linear elastic (isotropic), linear elastic (orthotropic),
overburdan dependent the ExtendedHardin model and
the DuncanChang model. For this study, the DuncanChang
Hyperbolic model was used.
16
Another option that is provided with the CANDE
program is the use of interface elements that can
be used to model relative movements between structure
and soil or soil and soil as fully fixed, partially
fixed (slip along a common boundary), or free (tensile
separation along a boundary).
The most valuable aspect of CANDE is its
expandibility. The modular nature of the program
permits new structure and soil models to be added
with relative ease. CANDE could be used, potentially,
in the design and analysis of many geotechnical projects,
such as underground storage facilities and tunnels.
Interface elements permit modeling of jointed rocks
and other contact surfaces.
The DuncanChanq Model
The finite element method is a powerful tool
for analysis of the stresses and movements of soil
masses and soilstructure problems mainly because
the nonlinear constitutive relationships of the soil
can be represented in a realistic manner.
The DuncanChang model uses a hyperbolic
stressstrain relationship in an attempt to provide
a simple model that characterizes soil stressstrain
behavior and uses data from conventional laboratory
tests (Duncan and Chang 1970). The development of
17
this soil model is an extension of the work done by
Kondner (1963) and Konder and Zelasko (1963), who
showed that the stressstrain curves for many soils
tested in the triaxial condition could be approximated
by hyperbolas as
(1)
E1 + V'Vult
This representation has two important characteristics
that make it very convenient to use. First, the hyperbolic,
parameters have physical significance. Ej_ is the
initial tangent modulus of the stressstrain curve
and the ultimate stress difference, (USD), is the
asymptotic value of the deviatoric stress related
to the soil strength. Second, the values of Ej_ and
USD can be easily determined for any given stressstrain
curve by transforming the stressstrain curve to a
linear relationship between strain/USD and strain.
The values of strain/USD calculated from the test
data can be plotted against the strain and the best
fit straight line used to determine the parameters
that represent best fit hyperbola for the stressstrain
plot.
After plotting several hundred stressstrain
curves Wong and Duncan (1974), found that a straight
18
line that passed through the points, where 75% and
95% of the maximum stress is mobilized usually produced
a good match.
unconsolidated^undrained conditions, all soils show
an increase in initial modulus and ultimate strength
with increasing confining pressure. Equation ('2) ,
suggesting the variation of E^ with the confining
pressure was proposed by Janbu (1963):
where K is the modulus number, n is the modulus exponent
and a3 is the confining pressure. pa is the atmospheric
pressure expressed in the same units as the confining
pressure. This makes K and n dimensionless and
independent of the units used for the confining pressure.
Ej^ is in the same units as the confining pressure.
failure, determined by the MohrCoulomb criterion,
and USD is Rf and is called the failure ratio.
By differentiating equation (l)\with respect
to the strain and making some substitutions, an equation
Except for fully saturated soils tested under
n
o
The ratio between the stress difference at
19
is obtained that relates the instantaneous slope of
the stressstrain curve to the friction angle, cohesion
and DuncanChang parameters.
Using a similar process and representing non
linear volume change as hyperbolas, stress dependent
volume change can be represented by varying the Poisson's
ratio with the confining pressure. For unsaturated
or drained conditions the initial Poisson's ratio
decreases with increasing confining pressure. Three
DuncanChang parameters, G, F and d, were developed
to express this relationship.
The DuncanChang hyperbolic parameters can
be determined from the results of conventional triaxial
compression tests and can be used for total (UU test)
or effective (CD test) stress analyses. The values
of the hyperbolic parameters have been calculated
and tabulated for many soils so that they can be used
in the absence of sufficient data on all the soils
involved in a given problem (Wong and Duncan 1974) .
They can also be used to check the reliability of
parameters obtained from laboratory tests.
There are three significant limitations to
the simple hyperbolic relationships. First, since
the relationships are based on the generalized Hooke's
law, they should not be used for analysis of stresses
20
and movements close to or beyond failure. Second,
the relationships cannot simulate dilative volume
changes due to changes in shear stress, or shear
dilatancy. Third, the parameters are not fundamental
soil properties, but are empirical coefficients which
represent the behavior of a given soil under the test
conditions.
It is to be noted that values for K and n
can also be determined from onedimensional consolidation
test results:
Ap(l+e )
______o
Ae
E.
(3) i
2JC
1
(1+K )
o
1 
PUKo)Rf
Kop(tan^(45Hf>'/2)l) + 2c' tan (45+'/2)
This procedure can be used when onedimensional
consolidation conditions are approximated in a field
situation.
Finite Element Discretization
Seven borings were obtained along the alignment
of the horizontal inclinometer. From these drill
logs a stratigraphic crosssection was constructed,
Figure 3. The foundation had three major layers above
the shale bedrock. Just above the shale was a dense
sandy gravel, probably a remnant from the gravel mining
operation. Above the gravel was the cement wash material
10 ft
Figure 3= Stratigraphic CroBsSection
to
22
and above that was a layer of mixed soil and trash.
A three foot layer of squeegee material was placed
just below ground level during the installation of
the horizontal inclinometer. Each of these layers
formed one layer of elements in the finite element
mesh, Figure 4. The embankment was formed of three
layers of elements so that the program could be run
using four construction increments, one for the foundation
and three for the embankment. This was done because
multiple construction increments, with finite element
analysis, best represented the actual field construction
sequence. Preliminary finite element analyses and
early field settlement measurements showed that the
settlement of the embankment was reasonably symmetrical
about the axis, therefore, the mesh represented one
half of the crosssection of the embankment. Data
from the vertical inclinometers showed no significant
horizontal movement of the foundation at the margins
of the embankment. Because of the relatively large
stiffness of the shale bedrock, it was assumed that
it would not displace vertically as the embankment
was constructed. Consequently, boundary conditions
were imposed such that the base of the mesh was not
allowed to move vertically and the margins of the
mesh were restrained from moving in a horizontal direction.
Figure 4. Finite Element Mesh
to
w
24
Geogrid Parameters
Tensile strengths for Tensar SS2 geogrid
were provided by the manufacturer in units of pounds
per foot. The cross machine direction tensile strength
at 2% strain was reported to be 600 lb/ft. The tensile
tests were done on single ribs, (see Figure 1) which
are .050 inches by .100 inches in crosssection and
are separated by 1.1 inches. The stress at 2% strain
per inch of geogrid was calculated to be 11,000 psi.
The secant modulus at 2% strain was calculated to
be 550,000 psi.
The equivalent crosssectional area of the
geogrid mesh was found to be 15 square inches per
inch of geogrid, therefore, the average thickness
was .15 inches. The Poisson's ratio was assumed to
be 0.3 .
Soil Parameters
A number of laboratory tests were performed
on the cement wash material by the Colorado Department
of Highways, including onedimensional consolidation
tests and CD, CU and UU triaxial tests. Since the
embankment was very wide in relation to the depth
of the soft foundation material and the vertical
inclinometers showed no significant lateral movement
at the margins of the embankment, the onedimensional
25
consolidation test results were used to deduce the
DuncanChang soil parameters. The Poisson's ratio
parameters were obtained from the CD triaxial test
results as were the friction angle and cohesion.
No tests were performed on embankment or
foundation soils so the DuncanChang parameters used
in the analysis were selected from the table of parameters
that were calculated and compiled for a wide variety
of soil types tested under drained conditions (Wong
and Duncan 1974). The initial DuncanChang parameters
used in this analysis are shown in Table 3.
SoilGeoqrid Composite Parameters
Two large CD triaxial tests were performed
on soilgeogrid samples. The samples were six inches
in diameter by twelve inches in length. The reason
for using the large samples was to include a large
enough piece of geogrid to allow interlocking between
the coarse soil and the geogrid. The soil material
was the ABC material used in the field to sandwich
the geogrid. During sample preparation, the ABC material
was compacted to 90% + 1% of standard proctor or about
128.0 pcf. A layer of geogrid was placed horizontally
at the center of the sample so that there were six
inches of ABC material above and below the geogrid.
The two tests were performed at two confining pressures,
TABLE 3
Initial DuncanChang Soil Parameters
Soil Laver Unit Wt. (ncfrc (psil0 (dea.1 K n P r G F d
Sandy Gravel 150 0 53 1300 0.4'0 f 0.72 0.33 0.07 7.1
Cement Wash 75 0 41 10. 0.40 0.50 0.49 0.00 0.0
TrashSoil 1.00 0 47 330 0.62 0.61 0.3 3 0.02 7.3
Squeegee 135 0 58 2500 0.21 0.75 0.35 0.17 14.6
Embankment 130 .52 34 430 0.58 0.70 0.26 0.04 6.5
to
c\
27
15 and 60 psi, Figure 5. Each sample was sheared
at a constant deformation rate of 0.05 inches per
minute. In each case the sample bulged above and
below the geogrid but not at the location of the geogrid,
as the sample was sheared. The results from these
tests were used to determine the DuncanChang parameters
for the soilgeogrid composite as shown in Table 4.
TABLE 4
DuncanChang Parameters for
SoilGeogrid Composite
Parameter Value
c 11 psi
0 35
K  454
n 0.4
*f 0.73
G 0.33
F 0.07
d 7.1
28
60 psi
Displacement (in.) 2*
2.5
Figure 5. Load vs. Displacement Relationship for
SoilGeogrid Composite
29
Results and Discussion of Results
The finite element analysis was performed
for one half the embankment using the mesh shown in
Figure 4. Field settlement measurements from the
horizontal inclinometer, for both sides of the embankment,
were compared to the settlement curve generated by
the CANDE program. The DuncanChang soil parameters
were refined for the cement wash and soiltrash materials,
Table 5.
TABLE 5
Final DuncanChang Soil Parameters for the
Cement Wash and SoilTrash Materials
Parameter Cement Wash SoilTrash
C 0 0
0 41 47
K 6.4 700
n 0.40 0.62
*f 0.50 0.61
G 0.40 0.40
F 0.0 0.02
d s 0.0 7.3
These two layers contributed the most to the deformation
of the embankment so the amount of deformation was
30
most sensitive to the DuncanChang soil parameters
of these two materials.
The settlement curve reproduced by CANDE did
not exactly match the field settlement curves from
either side of the embankment, but generally fell
somewhere between the two field curves and represents
what might be expected for settlement on the average,
Figure 6.
The CANDE generated curves using discrete
elements to represent the geogrid and using the
soilgeogrid composite properties were very nearly
the same, Figure 7. The settlement curve using the
soilgeogrid composite properties showed slightly
better agreement with the field settlement curves
at the margins of the embankment. Both computer runs
used the same mesh configuration and soil parameters,
only the method of accommodating the geogrid properties
differed. This indicates that either method to account
for the geogrid properties will work as well as the
other.
Figure 6 CANDE Generated and Field Settlement Curves
Figure 7. Composite and Discrete Settlement Curves of the
Geogrid Reinforced Embankment
to
CHAPTER' 4
PREDICTED PERFORMANCE OF THE EMBANKMENT
Background
The selection of Tensar SS2 geogrid as a
tensile reinforcement for the embankment at 176 and
Tennyson, by the Colorado Department of Highways,
was a conservative decision because the consequence
of slope failure would have been high in both.time
and money. In order to investigate the effectiveness
of geogrid to reduce differential settlement, analyses
were made using the same embankment but with a different
type of tensile reinforcement and with no tensile
reinforcement. The other tensile reinforcement
investigated was a nonwoven geotextile.
The Geotextile
The geotextile used for this investigation
was a nonwoven, needlepunched, polyester geotextile
called Trevira 1127, manufactured by Hoechst Fibers
Industries.
The manufacturer provided inair strength
parameters for the geotextile, but these could not
be used for.this investigation. ElFermaoui and Nowatzki
34
(1982) and McGowen et al. (1982) showed that the confining
pressure had a significant effect on the strength
characteristics of geotextiles. These studies showed
that the stiffnesss and the strength of the geotextile
increased with increasing confining pressure.
The Test Apparatus
To determine the insoil strength properties
of Trevera 1127, insoil stressstrain tests at various
confining pressures needed to be conducted. To perform
these tests a testing box was designed and manufactured.
The test box was designed to be used with a Wykeham
Farrance direct shear loading frame. The test box
was constructed of aluminum with a steel top loading
cap. The test box was designed with a fixed clamp
inside the box to hold the geotextile secure and a
slot on the opposite side through which a movable
clamp extended to securely grab the other side of
the geotextile sample. The test was designed so that
the geotextile sample was entirely within the test
box and subject to the confining pressure throughout
the test. The interior dimensions of the test box
were 6.5 inches by 4.7 5 inches by 2.5 inches deep.
The top loading cap was 6.5 inches by 4.7 5 inches
to distribute the normal load uniformily over the
soil material contained in the test box, Figure 8.
iH
H
p
Movable X Q) Fixed
Clamp P o Clamp
Q)
U
Top View
InSoil Geotextile StressStrain Test Box
Figure 8.
36
InSoil Geotextile StressStrain Test
The insoil stressstrain tests on the Trevira
1127 geotextile were performed using the testing apparatus
described in the last section. Dry #30 Ottawa sand
was used for the support and cover material. The
gradation of this material is shown in Figure 9.
For the tests, the Ottawa Sand was placed at density
of 107.0 pcf (Dr = 68%). The exposed geotextile was
three inches by 1.5 inches. The geotextile sample
was chosen to be this size for two reasons. First,
to place it as far away from the sides of the box
as possible, to minimize and side friction effects.
Second, to maintain a plane strain configuration and
avoid necking effects at small strain. In the field,
geotextiles are rarely strained more than 10%. In
the case of this embankment the strain did not exceed
2%. This was the range of strain of greatest interest
in this analysis, but the tests were continued to
failure or 100% strain, whichever occured first.
Inair stressstrain tests are usually run at 2% to
10% strain per minute. ElFermaoui and Nowatzki (1982)
ran insoil tests at 2.5% strain per minute and McGowen
et al. (1982) ran insoil tests at 2% strain per minute.
The strain rate used for these tests was about 3.15%
per minute or a deformation rate of 1.2 mm per minute.
37
Grain diameter, mm
Figure 9. Grain Size Distribution for #30 Ottawa Sand
38
ElFermaoui and Nowatzki (1982) reported that the
shear stresses developed along the galvanized steel
movable clamp due to the confining pressures were
assumed to be very small compared to the shear stresses
developed on the soilgeotextile interface and were
neglected. However, pullout tests with the test box
and the galvanized steel movable clamp with no geotextile
showed that the shear stresses were significant, especially
during the early part of the tests. The insoil
stressstrain test results were corrected for these
shear stresses. The tests were run at two confining
pressures, 8.36 psi and 16.53 psi. The smaller of
these confining pressures corresponded to the field
condition for the 176 and Tennyson embankment after
the first stage of construction. An inair test,
using the test box, was also performed for comparison,
Figure 10.
39
Figure 10. Load vs. Displacement Relationship for
InSoil Geotextile
40
The secant moduli at 2% strain for the tested
confining pressures are shown in Table 6.
TABLE 6
Secant Moduli at 2% Strain for
InSoil Geotextile
Confining Secant
pressures Moduli
(psi) (psi)
inair 646
8.36 1131
16.53 1697
The insoil thickness of Trevira 1127 was
found to be approximately 0.1 inches and the Poisson's
ratio was assumed to be 0.3.
SoilGeotextile Composite Parameters
Three CD triaxial tests were performed on
soilgeotextile composite samples. The samples were
two inches in diameter by four inches in length.
The soil material was the dry #3 0 Ottawa sand that
was used for the insoil geotextile stressstrain
tests. The Ottawa sand was placed at 107.0 pcf (Dr
= 68%). A layer of Trevira 1127 geotextile was placed,
horizontally, at the center of the sample so that
there was two inches of Ottawa sand above and below
the geotextile. The three tests were performed at
41
three different confining pressures, 15, 30 and 60
psi, Figure 11. Each sample was sheared at a constant
deformation rate of 0.05 inches per minute. In each
case the sample bulged above or below the geotextile,
but not at the location of the geotextile, as the
sample was sheared. The results from these tests
were used to determine the DuncanChang soil parameters
as shown in Table 7.
TABLE 7
DuncanChang Parameters for
SoilGeotextile Composite
Parameter Value
C 0
jor 40
K 700
n 0.62
*f 0.79
G 0.46
F 0.22
d 8.9
Results and Discussion of Results
Computer runs, using the CANDE program were
made using the same mesh configuration and soil parameters
but substituting, for the geogrid properties, the
42
43
geotextile properties or the the soilgeotextile composite
parameters. A run was also made with the tensile
reinforcement eliminated.
The settlement curves showed no significant
difference in the settlement curve when using the
single layer of geogrid or geotextile tensile
reinforcement, Figure 12, or using no reinforcement
near the base of the embankment, Figure 13. The settlement
curves using discrete geotextile element properties
and using soilgeotextile composite parameters were,
again, very nearly the same with the settlement curve
using the soilgeotextile composite showing slightly
better agreement with the field settlement curves
at the margins of the embankment, Figure 14.
Chirapuntu and Duncan (1976) did a study of
s
the effectiveness of reinforcing in preventing cracking
in embankments. The reinforcement was steel tierods
near the base of the embankment. Finite element analysis
showed that this type of reinforcement reduced differential
settlement significantly.
For comparison, a CANDE run was made using
the stiffness of steel for the reinforcement. In
this case the settlement was not greatly affected
by using steel reinforcement, Figure 15.
Figure 12. Settlement Curves of the Embankmeht Reinforced
with Geogrid and Geote^tile
Figure 13. Settlement Curves of the.Embankment Reinforced
with Geogrid and Without Reinforcement
tn
Figure 14. Composite and Discrete Settlement Curves of
the Geotextile Reinforqed Embankment
cn
Figure 15. Settlement Curves of the Embankment Reinforced
with Geogrid and Steel ^
vj
48
The CANDE results showed that the magnitude
and direction of principal stresses, in the embankment
and foundation, were about the same with geogrid
reinforcement, Figure 16, as with no reinforcement,
Figure 17. This was also the case with geotextile
reinforcement, Figure 18. With steel reinforcement
the principal stresses in the embankment were signifi
cantly reduced, Figure 19.
Critical areas in the embankment, from a
stability standpoint, were determined by plotting
those elements in the mesh that the CANDE results
showed were subject to tensile stress. These plots
for the embankment unreinforced, Figure 20, and rein
forced with geogrid, Figure 21, showed that the geogrid
reduced and redistributed the tensile forces in the
embankment. The results using geotextile reinforcement
showed no reduction in tensile stresses, Figure 22.
With steel reinforcement the tensile stresses were
greatly reduced, Figure 23.
The magnitude of tensile stress that the
reinforcement is subjected to is indicative of its
relative effectiveness. The thrust distribution profiles
for geotextile, geogrid and steel reinforcement were
plotted in Figure 24. As the stiffness of the reinforcement
increases, the stress it picks up also increases.
y 4W+
ir
25 .psi
Figure 16. Principal stresses with Geogrid Reinforced
Embankment
VO
is:
'A
/
S
X
\
x
X
v
$
i1
25 psi
Figure 17. principal stresses with Unreinforced Embankment
i
01
o
s
t*
A
/
s
X
\
V.
/
22
t
h
''Ai.
**N
I
X
5
\
\
ft
H: <
25 psi
Figure 18. Principal Stresses with Geotextile Reinforced
Embankment
Ul
25 psi
Figure 19. Principal stresses with Steel Reinforced
Embankment
ui
to
Figure 20. Elements with Tensile Stresses with Geogrid
Reinforced Embankment
Figure 21. Elements with Tensile Stresses with Unreinforced
Embankment
U1
/
Figure 22. Elememts with Tensile Stresses with Geotextile
Reinforced Embankment
Figure 23. Elements with Tensile Stresses With Steel
Reinforced Embankment
56
57
lOOOi
750
500
250
250
500
750
10001
Figure 24. Thrust Distribution in Reinforcements.
CHAPTER 5
SUMMARY AND CONCLUSIONS
Summary
The Colorado Department of Highways constructed
an embankment on a soft foundation. To increase the
stability and decrease detrimental settlement, the
embankment was reinforced by placing a single layer
of Tensar SS2 geogrid between the foundation and
the embankment. This study investigated the effectiveness
of using Tensar SS2 geogrid to strengthen the embankment
by comparing the settlements resulting when using
Tensar SS2 geogrid, Trevira 1127 geotextile and using
no ;tensile reinforcement. To make this comparison,
a finite element program, CANDE, was used. The
DuhcanChang hyperbolic model was used to model the
soil behavior. Field settlement data were used to
refine the soil parameters used in the program. To
determine the insoil stressstrain properties of
the Trevira 1127 geotextile, a test device was designed
and fabricated and a series of tests, at various confining
pressures were performed. The tensile reinforcement
j
properties were incorporated into the program in two
different manners. The first was to use the insoil
59
stressstrain properties and discrete beam elements
for the tensile reinforcement. The second was to
use soilreinforcement composite parameters. These
parameters were determined from the results of CD
triaxial tests using a single layer of reinforcement
in the center of the soil sample.
Conclusions
Based on the results of the study, the
following conclusions are advanced for the use of
tensile reinforcements to strengthen the embankment.
1. The use of a single layer of Tensar SS2
geogrid did not significantly reduce the
differential settlement of the embankment,
compared to using a single layer of Trevira
1127 geotextile or using no tensile reinforce
ment. This is due in part to insufficient
stiffness and probably due in part to the
fact that the deformation of the embankment
is predominantly in the vertical direction.
2. The stressstrain relationship of geotextile
has to be determined in the confinement of
the backfill prepared at the field density.
Correct interpretation of the test results
must account for the frictional resistance
between the metal clamp and the soil.
60
3. For simulation of tensile reinforcements in
finite element analysis, using soilreinforce
ment composite parameters with quadrilateral
elements gives essentially the same deformation
pattern as using insoil reinforcement propert
ies with discrete beam elements.
REFERENCES
Barsvary, A. K. M. D. Maclean, and C. B. H. Cragg,
"Instrumented Case Histories of Fabric Rein
forced Embankments Over Peat Deposits," Second
International Conference on Geo'textiles, Las
Vegas, Nevada, Vol. 3, 1982, pp. 647652.
Boutrup, E., and R. D. Holtz, "Fabric Reinforced Emban
kments. Constructed on Weak Foundations," Report
No. FHWA/IN/JHRP82/21, Purdue University,
West Lafayette, Indiana, 1983.
Chirapuntu, S., and J. M. Duncan, "The Role of Strength,
in the Stability of Embankments on Soft Clay
Foundations," Report No. TE 753, University
of' California, Berkeley, California, 1976.
Duncan, J. M., and C. Y. Chang, "Nonlinear Analysis
of Stress and Strain in Soils," Journal of
the Soil Mechanics and Foundation Division.
ASCE, Vol. 96, No. SM9, Proceedings Paper
7513, 1970, pp. 16351653.
ElRermaoui, A., and E. Nowatzki, "Effect of Confining
, Pressure on Performance of Geotextiles in
Soils," Second International Conference on
Geotextiles, Las Vegas, Nevada, Vol. 3, 1982,
pp. 799804.
Herrmann, L. R., and z. AlYassin, "Numerical Analysis
of Reinforced Soil Systems," Proc. ASCE Sympos
ium on Earth Reinforcement, Pittsburg, Pennsyl
vania, 1978, pp. 428457.
Hutchins, R. D., "Behavior of Geotextiles in Embankment
Reinforcement," Second International Conference
on Geotextiles, Las Vegas, Nevada, 1982, pp.
617619.
Janbu, N., "Soil Compressibility as Determined by
Oedometer and Triaxial Tests,1.1 European Con
ference. .on Soil Mechanics and Foundation Eng
ineering, Wiesbaden, Germany, Vol. 1, 1963,
62
Katona, M. G., and J. M. Smith, "CANDE User Manual,"
Report No. FHWARD776, Washington, D. C.,
1976.
Katona, M. G., j. M. Smith, R. s. Odell, and J. R.
Allgood, "CANDE A Modern Approach for the
Structural Design and Analysis of Buried
Culverts," Report No. FHWARD775, Washington,
D. C., 1976.
Kondner, R. L., "Hyperbolic StressStrain Response:
Cohesive Soils," Journal of the Soil Mechanics
and Foundation Division. ASCE, Vol. 89, No.
SMI, Proc. Paper 3429, 1963, pp. 115143.
Kondner, R. L., and J. S. Zelasko, "A Hyperbolic
StressStrain Formulation for Sands," Proceeds,
ings, 2nd PanAmerican Conference on Soil
Mechanics and Foundation Engineering, Brazil,
Vol. 1, 1963, pp. 289324.
Leonards, G. A., T. H. Wu, and C. H. Juang, "Predicting
Performance of Buried Conduits," Report No.
FHWA/IN/JHRP813, Purdue University, West
Lafayette, Indiana, 1982.
McGowen, A., K. Z. Andrawes, and M. H. Kabir, "Load
, Extension Testing of Geotextiles Confined
InSoil," Second International Conference
on Geotextiles, Las Vegas, Nevada, Vol.3,
1982, pp. 783798.
Petrik, P. M., R. Baslik, and F. Leitner, "The Behavior
of Reinforced Embankments," Second Internat
ional Conference on Geotextiles, Las Vegas,
Nevada, 1982, pp. 631634.
Wong, K. S., and J. M. Duncan, "Hyperbolic StressStrain
Parameters for Nonlinear Finite Element Analysis
of Stresses and Movements in Soil Masses,"
Report No. TE 743, University of California,
Berkeley, California, 1974.
