Download as pdf or txt
Download as pdf or txt
You are on page 1of 56

Technical Paper by S.W. Perkins and M.

Ismeik

A SYNTHESIS AND EVALUATION OF


GEOSYNTHETIC-REINFORCED BASE LAYERS IN
FLEXIBLE PAVEMENTS: PART I

ABSTRACT: In recent years, geosynthetics have been proposed and used as rein-
forcement in the base course layer of flexible pavements for the purpose of improving
performance and/or to allow for the reduction of base course thickness. Much of the pio-
neering work with geosynthetic-reinforced unpaved roads on very soft subgrades has
indicated that appreciable deformation of the roadway surface is necessary before the
reinforcement qualities of the geosynthetic can be realized. It may be expected that this
same condition is necessary in a paved road, thereby obviating the practical use of geo-
synthetics as reinforcement. It appears that this view has gained acceptance in the re-
search and practice oriented engineering communities. The purpose of this paper is to
provide a synthesis and evaluation of the literature focusing on this application. The
paper focuses on studies involving laboratory-scale experiments using stationary cyclic
loads or moving wheel loads and field studies using controlled vehicle loads or random
traffic loads. The majority of the studies reviewed indicate that appreciable improve-
ment can be realized by proper placement of a geosynthetic in the base course of a flex-
ible pavement and that improvement is seen over the entire service life of the pavement
and not only for conditions of excessive surface deformation.

KEYWORDS: Geogrid, Geotextile, Reinforcement, Base course, Flexible pavement,


Road, Highway, Laboratory experiment, Field experiment.

AUTHORS: S.W. Perkins, Assistant Professor of Civil Engineering, and M. Ismeik,


Postdoctoral Research Associate, Department of Civil Engineering, 205 Cobleigh Hall,
Montana State University, Bozeman, Montana 59717, USA, Telephone:
1/406-994-6119, Telefax: 1/406-994-6105, E-mail: stevep@ce.montana.edu.

PUBLICATION: Geosynthetics International is published by the Industrial Fabrics


Association International, 1801 County Road B West, Roseville, Minnesota
55113-4061 USA, Telephone: 1/612-222-2508, Telefax: 1/612-631-9334.
Geosynthetics International is registered under ISSN 1072-6349.

DATES: Original manuscript received 3 June 1997, revised version received 11


November 1997 and accepted 17 November 1997. Discussion open until 1 July 1998.

REFERENCE: Perkins, S.W. and Ismeik, M., 1997, “A Synthesis and Evaluation of
Geosynthetic-Reinforced Base Layers in Flexible Pavements: Part I”, Geosynthetics
International, Vol. 4, No. 6, pp. 549-604.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 549


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

1 INTRODUCTION

Geogrids and geotextiles have been studied and used for more than 15 years as rein-
forcement in the base course layer of flexible pavements. The attraction of this applica-
tion lies in the possibility of reducing the thickness of the base course layer such that
a roadway of equal service life results or in extending the service life of the roadway.
The first alternative is beneficial if the cost of the geosynthetic is less than the combined
cost of the replaced base course material and any construction related costs associated
with a reduced base thickness (such as excavation, relocation of utilities, and purchase
of right-of-way). Benefits are seen with the second alternative when maintenance and
replacement costs are offset by the cost of the geosynthetic. Both alternatives are partic-
ularly attractive in areas where quality gravel sources are scarce, in urban areas where
these sources have become depleted, or in environmentally sensitive areas where the
siting of gravel quarries is not permitted.
Early uses of geotextiles in roadways included separation, filtration, and drainage in
paved and unpaved roads and reinforcement in unpaved roads (Steward et al. 1977;
Bender and Barenberg 1978). Geotextiles were first examined for use as reinforcement
in paved roads in the early 1980s (Brown et al. 1982; Ruddock et al. 1982), while geo-
grids were first studied in the late 1980s (Barker 1987; Haas et al. 1988; Barksdale et
al. 1989). The focus of this paper is reinforcement applications. A limited discussion
of separation and filtration functions is provided only for the purpose of evaluating the
impact of subgrade-base course mixing on subsequent reinforcement benefits. Separa-
tion and filtration functions are briefly defined in this section, while a discussion of the
impact of these functions on reinforcement functions is presented in Section 4.
The function of separation refers to the ability of the geosynthetic to maintain a physi-
cal separation between the base course layer and the underlying subgrade, where the
potential for mixing is strongest primarily during construction and secondarily during
the operation of design traffic. While separation appears to be a viable function for geo-
grids, this function is typically associated, and is more effective, with geotextiles. Sepa-
ration is required for conditions where the subgrade is relatively weak; i.e. California
Bearing Ratio (CBR) less than 3 (Holtz et al. 1995).
Filtration refers to the ability of the geosynthetic to filter the fine subgrade soil par-
ticles as water passes from the subgrade to the base course layer. Water flow is typically
induced by the generation of excess pore water pressure in the subgrade as a result of
repetitive traffic loads and is a necessary function only for saturated or near saturated
subgrades. The need for filtration is strongly dependent on the fines content of the base
course and subgrade soils and the plasticity limits of the subgrade. Separation and filtra-
tion functions are viable for both paved and unpaved roads and can play an important
role in interpreting results from studies examining the reinforcement effect of geosyn-
thetics in paved roads.
Studies involving geosynthetic reinforcement of roadways have identified three po-
tential reinforcement functions: lateral restraint, increased bearing capacity, and ten-
sioned membrane. Lateral restraint (Figure 1a) is caused by frictional interaction and
interlocking between the base course soil and the geosynthetic. Repetitive wheel loads
create a spreading motion of the base layer, which can be reduced if the geosynthetic
is properly positioned at a location of maximum lateral strain. Interaction between the
base aggregate layer and the geosynthetic transfers shear load from the base layer to a

550 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

tensile load in the geosynthetic. The geosynthetic, being stiff in tension, limits the ex-
tensional lateral strains in the base layer. In addition, the geosynthetic confines the base
layer increasing the mean stress and subsequently the stiffness and shear strength of the
base layer. This mechanism does not imply the need for significant rut depths nor does
it imply an ensuing failure state in the base or subgrade layers. Adequate frictional and
interlocking characteristics between the soil and the geosynthetic are necessary to real-
ize this mechanism. For geogrids, this implies that the geogrid apertures and base soil
particles must be sized properly and that more interaction is possible if the geogrid is
placed within the base such that particle strike-through occurs.
Figure 1b illustrates the function of increased bearing capacity. The geosynthetic
forces the potential failure surface to develop along an alternate surface that has a great-
er total resistance, which implies that failure is incipient. The tensioned membrane ef-
fect (Figure 1c) develops as a result of vertical deformations creating a concave shape
in the geosynthetic. The tension developed in the geosynthetic helps support the wheel
load and reduce the vertical stress on the subgrade. Significant rut depths are necessary
to realize this effect. Each of these functions is inappropriate for permanent paved roads
where significant rut depths cannot be tolerated, and where bearing capacity failure is
not a permissible failure mechanism.
Much of the early work on geosynthetic reinforcement of base course layers focused
on unpaved roads where rut depths in excess of 25 mm were tolerable. Design ap-

(a) Wheel load (b) Wheel load

Base AC
Base AC

Subgrade
Subgrade

Hypothetical shear surface


Geosynthetic without geosynthetic
Lateral restraint of Hypothetical shear surface
the geosynthetic with geosynthetic Geosynthetic

(c) Wheel load


Base AC
Subgrade

Membrane tension

Vertical support Geosynthetic


component of membrane

Figure 1. The reinforcement functions attributed to geosynthetics in roadways:


(a) lateral spread; (b) bearing capacity; (c) tensioned membrane.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 551


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

proaches were developed for these conditions where functions such as increased bear-
ing capacity and tensioned membrane support were included (Giroud and Noiray 1981;
Christopher and Holtz 1985). More recently, design solutions have been proposed for
unpaved roads for conditions of moderate rut depth (Houlsby and Jewell 1990; Douglas
1993). This earlier work on unpaved roads appears to have influenced the thinking of
designers of paved, permanent roadways where the notion of reinforcement has been
dismissed because of the perceived requirement for the development of large rut depths.
The purpose of the current paper is to review, critique, synthesize, and evaluate test
results from studies that span the past 15 years and involve geosynthetic reinforcement
of base course layers in flexible pavements. The review is limited to paved roads and
conditions pertinent to permanent roads (i.e. rut depths less than 25 mm). The studies
reviewed discuss the conditions under which the reinforcement was both effective and
ineffective for tolerable rut depth levels. The lack of agreement in the studies examined
and the absence of an accepted design technique suggests why this topic is still being
researched after its introduction over 15 years ago.
The experimental studies included in this review have been restricted to those involv-
ing flexible pavements where geogrids and geotextiles have been placed in the base
course layer or at the top of the subgrade. The experimental studies include laboratory
work using cyclic loads applied to the pavement surface through stationary circular
plates, indoor test tracks where moving wheel loads have been applied, and field work
where either controlled or random traffic has been applied. Section 2 describes the perti-
nent experimental features of each study and has been subdivided into categories of
common experimental variables.
Section 3 provides a brief summary of the major experimental findings reported by
the authors of each study. The information contained in Section 3 corresponds to state-
ments and conclusions made directly by the authors of these studies. Section 4 provides
a critical synthesis and evaluation of the data contained in the reviewed studies; inter-
pretations and conclusions are made by the authors of the current paper. Whenever pos-
sible, these interpretations and conclusions are compared to those of other studies.
Analytical studies using finite element techniques to predict roadway response and to
illustrate reinforcement mechanisms have been summarized in a companion paper
(Perkins and Ismeik 1997) as have existing design techniques and a description of on-
going published and unpublished studies.

2 EXPERIMENTAL TEST SECTIONS

Twelve studies that involve experiments where geogrids and geotextiles have been
used as reinforcement contained in the base course layer of flexible pavements have
been identified. This section summarizes details associated with the experimental setup
of each study. Tables 1 to 5 at the end of this paper have been created to aid in this sum-
mary. Test results from selected studies are summarized in Figures 2 to 8. Tables 6 to
9, also at the end of this paper, provide information on the test section variables for Fig-
ures 2, 3, 6, and 7 and are useful for understanding the setup of the experimental test
sections corresponding to these studies. The data contained in Figures 2 to 8 is analyzed
more extensively in Section 4.

552 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

2.1 Introduction and Overview of Previous Studies

Al-Qadi et al. (1994) performed laboratory-scale repeated load tests in a concrete test
tank to compare separation and reinforcement functions of geotextiles and geogrids and
to prepare for the construction of field test sections. The test results from four sections
are presented in Al-Qadi et al. (1994). The test results from an additional 14 sections
using the same experimental setup are reported in Smith et al. (1995) and Al-Qadi et
al. (1997). Test results presented in this paper that are derived from Smith et al. (1995),
Al-Qadi et al. (1997), and Al-Qadi et al. (1994) are, in general, cited as Al-Qadi et al.
Selected test results from Al-Qadi et al. are presented in Figure 2 with accompanying
information on the test sections given in Table 6.
Anderson and Killeavy (1989) described a field study where a geogrid was used in
combination with a weak nonwoven geotextile separator to reinforce the base layer of
a truck staging yard. The access road leading into the yard was designed and constructed
as a conventional nonreinforced section without a geotextile separator between the base
and subgrade. A third study area, consisting of an extension of the staging yard, was
designed and constructed as a nonreinforced section with a geotextile separator. The
thickness of the base layer was adjusted between each section to provide for equivalent
performance. The test results presented correspond to a one-year monitoring period,
while the section was designed for a 15 year service life. Recent personal communica-

(a) 25
20
Rut depth (mm)

15 Section 3
Section 4
10 Section 6
Section 10
5 Section 11
Section 12
0
100 101 102 103 104
Load cycles
25
(b)
20
Rut depth (mm)

15
Section 15
10 Section 16
Section 17
5
Section 18
0
100 101 102 103 104
Load cycles
Figure 2. The experimental test results from Al-Qadi et al. (1994): (a) subgrade CBR ≅
4; (b) subgrade CBR ≅ 2.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 553


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

tion with the second author has provided additional information on the performance of
the sections (Killeavy 1997).
Barker (1987) focused on the application of open-graded base course material for use
in airport runways. Barker (1987) constructed the open-graded base layer on top of a
strong subbase and subgrade system because the stability of the base layer was of inter-
est. Using this construction method, pavement failure would be forced to occur by de-
formations in the base layer. One test section was constructed with geogrid
reinforcement and compared to one control section of identical geometry and properties.
Barksdale et al. (1989) performed a comprehensive experimental and finite element
modeling study using one type of geogrid and one type of geotextile. Pavement test sec-
tions were constructed in an indoor test track and loaded with one moving wheel. The
pavement surfacing and base course layer were relatively thin, while the applied load
was relatively light. Selected test results are shown in Figure 3 with accompanying test
section information given in Table 7.
Brown et al. (1982) examined the use of two nonwoven geotextile products as reinfor-
cement. This is one of the first reported studies of reinforcement applications for flex-
ible pavements. Pavement sections were constructed in the facility later used by
Barksdale et al. (1989). Discussion of the study by Brown et al. (1982) is given by
Brown et al. (1983).

25
(a) Section 1
20
Rut depth (mm)

Section 2
Section 3
15 Section 4

10
5
0
100 101 102 103 104 105 106
Load cycles
25
(b) Section 5
Rut depth (mm)

20 Section 6
Section 7
15 Section 8

10
5
0
100 101 102 103 104 105 106
Load cycles
Figure 3. The experimental test results from Barksdale et al. (1989): (a) geosynthetics
in the middle of the base; (b) geosynthetics in the bottom of the base.

554 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Cancelli et al. (1996) performed cyclic load plate tests in an indoor metal test tank.
Five geogrids and one geotextile were compared. The strength of the fine sand subgrade
ranged from a CBR value of 1 to 18 and was prepared by varying the material density.
Selected test results from this study are presented in Figure 4, and are also reported in
the paper by Montanelli et al. (1997).
Collin et al. (1996) performed moving wheel load tests in an indoor test track for the
purpose of supplementing the test results of Haas et al. (1988) and Webster (1993). Two
geogrid types and sections with various base course thicknesses (Figure 5) were ex-
amined in the study.

(a) 25
CBR = 1
20 75 mm AC
Rut depth (mm)

300 mm base
15 Geogrid A
Geogrid H
10 Geogrid I
Geogrid J
5 2 Geogrid H

0
100 101 102 103 104 105 106
Load cycles
25
(b) CBR = 3
20 75 mm AC
Rut depth (mm)

300 mm base
15 Control
Geogrid D
10
Geotextile F
5 Geogrid H
Geogrid J
0
100 101 102 103 104 105 106
Load cycles
25
Control, 300 mm base
(c) 20 Control, 350 mm base
Rut depth (mm)

Control, 400 mm base


15 Geogrid H, 300 mm base
Control, 500 mm base
10
5 CBR = 3
75 mm AC
0
100 101 102 103 104 105 106
Load cycles
Figure 4. The experimental test results from Cancelli et al. (1996): (a) subgrade CBR
= 1; (b) subgrade CBR = 3; (c) equivalent base course thickness of Geogrid H.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 555


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Halliday and Potter (1984) constructed paved test sections where one geotextile was
used as reinforcement to validate the test results obtained from earlier work on unsur-
faced roads, which indicated that the permanent rut depth and vertical strain in the sub-
grade could be reduced by the presence of a geotextile. The sections contained a
relatively thick asphalt concrete (AC) layer. A partial summary of the test results from
this study have been reported by Ruddock et al. (1982).
Haas et al. (1988) performed laboratory experiments using a stationary plate to which
a cyclic load was applied. The experimental program was devised to examine the influ-
ence of AC and base thickness, reinforcement location, subgrade strength, and geogrid

(a) 25
CBR = 1.9
20
Rut depth (mm)

50 mm AC
180 mm base
15
10 Control
Geogrid A
5 Geogrid B

0
100 101 102 103 104 105
Load cycles
25
(b) CBR = 1.9
20 50 mm AC
Rut depth (mm)

235 mm base
15
10
Control
5 Geogrid A
Geogrid B
0
100 101 102 103 104 105
Load cycles
25
CBR = 1.9
(c) 20
Rut depth (mm)

50 mm AC
290 mm base
15
10 Control
Geogrid A
5 Geogrid B
0
100 101 102 103 104 105
Load cycles
Figure 5. The experimental test results from Collin et al. (1996): (a) 180 mm base;
(b) 235 mm base; (c) 290 mm base.

556 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

pretensioning. One layer of geogrid was used in the laboratory experiments. Several
references are available for this study including Penner (1985), Penner et al. (1985),
Kennepohl et al. (1985), and Carroll et al. (1987). The test results and section details
are presented in Figure 6 and Table 8.
Miura et al. (1990) conducted laboratory and field experiments and modeled the test
sections using a finite element program. Laboratory tests were conducted to investigate
the structural influence of different design variables, while finite element analyses were
performed to illustrate reinforcement mechanisms. The field tests were performed to
demonstrate the reinforcement application. The test results and section details are pre-
sented in Figure 7 and Table 9.
Moghaddas-Nejad and Small (1996) conducted relatively small-scale experiments
using an indoor test track. The loading and pavement section thickness were reduced
by a ¼ scale from that commonly encountered in practice.
Webster (1993) constructed full-scale, outdoor test track sections designed for use in
airports for light aircraft. The test variables included subgrade strength, base thickness,
geogrid modulus, and geogrid type and position. The load used in the study was rela-
tively heavy. A literature review for this study was reported by White (1991). The test
results and information on the test sections are shown in Figure 8.

2.2 Test Section Location and Load Type

The studies reviewed employed one of two types of laboratory facilities, or were
constructed in the field using test tracks or real roadways, or were a combination of the
two. Table 1 provides a summary of the type of test sections constructed and details con-
cerning the type of load that was applied. The most common laboratory facility and test
configuration was a pavement test section constructed in a large box. The AC layer was
typically loaded with a stationary rigid circular plate, where the load was cycled using
a constant amplitude until a certain rut depth was achieved. Studies utilizing this type of
load facility have been reported by Al-Qadi et al. (1994), Cancelli et al. (1996), Haas et
al. (1988), and Miura et al. (1990).
The response of test sections have also been evaluated in laboratory facilities using
moving wheel loads operated on indoor test tracks (Barksdale et al. 1989; Brown et al.
1982; Collin et al. 1996; Moghaddas-Nejad and Small 1996). These test tracks are typi-
cally constructed in a long, rectangular box with a wheel attached to a moving load
frame that rides along rails attached to the sides of the box. Barksdale et al. (1989) and
Brown et al. (1982) used the same facility at the University of Nottingham in the United
Kingdom. Test tracks with prepared and natural subgrades have also been constructed
in the field (Barker 1987; Halliday and Potter 1984; Webster 1993). These sections were
typically loaded by trucks of a known weight.
Test sections have also been constructed in the field along new or reconstructed road-
ways, parking areas, or operating yards (Anderson and Killeavy 1989; Miura et al.
1990). Field test sections were typically loaded by daily operating traffic and, therefore,
it was necessary to use traffic monitoring and weighing devices.
The indoor test track used by Brown et al. (1982) and Barksdale et al. (1989) utilized
one wheel for loading that measured 560 mm in diameter and 150 mm in width. For both
studies, the full wheel load was applied in both the forward and return motion of the
track carriage (bi-directional loading).

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 557


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

(a) 25
Section 4
20

Rut depth (mm)


Section 1
Section 2
15 Section 3

10
5
0
100 101 102 103 104 105 106
Load cycles
(b) 25
20
Rut depth (mm)

15
10
Section 5
5 Section 6

0
100 101 102 103 104 105 106
Load cycles
(c) 25
20
Rut depth (mm)

15
Section 8
10 Section 7
Section 9
5 Section 10

0
100 101 102 103 104 105 106
Load cycles
(d) 25
20
Rut depth (mm)

15
Section 11
10
Section 14
5 Section 13
Section 12
0
100 101 102 103 104 105 106
Load cycles
(e) 25
Rut depth (mm)

20
15
10 Section 17
Section 15
5
Section 16
0
100 101 102 103 104 105 106
Load cycles
Figure 6. The experimental test results from Haas et al. (1988): see Table 8 for a
description of each test section in Figures 6a to 6e.

558 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

(a) 25
Section 1

Rut depth (mm)


20 Section 3
Section 2
15 Section 4

10
5
0
100 101 102 103 104 105 106
Load cycles
(b) 25
Section 5
20
Rut depth (mm)

Section 7
Section 6
15
10
5
0
100 101 102 103 104 105 106
Load cycles
Figure 7. The experimental test results from Miura et al. (1990): (a) stronger
subgrade; (b) weaker subgrade.

Barksdale et al. (1989) constructed three 1.6 m long test sections at one time in the
test track. As seen in Table 1, these sections are short in comparison to test sections used
in other studies involving test tracks. Two types of tests were performed with respect
to the channelization of traffic. One test series involved multiple-track loading where
a certain number of wheel passes were made in one of nine lateral locations or tracks.
The maximum lateral distance of the wheel wander for the nine tracks was 610 mm.
The lateral offset between each track was 75 mm. The distribution of wheel passes with
track location was arranged in the form of a bell curve with the maximum number of
passes occurring at the center track location. The sequence of wheel passes along the
nine tracks was carried out as a traffic cycle, with the cycle repeated until the section
exhibited excessive rutting. The second test series involved single-track wheel passes
where the wheel remained in one track for the duration of the test. These tests were car-
ried out on the same sections for which multiple-track tests were performed and where
the single-track location was off to one or both sides of the multiple-track locations. The
offset distance of the two test locations was not directly stated; however, the informa-
tion provided indicates that the outside-to-outside distance between the wheels for the
multiple- and single-track tests was 300 mm.
Brown et al. (1982) used only multiple-track loading under conditions similar to those
described by Barksdale et al. (1989). Collin et al. (1996) also used an indoor test track
where loading was provided by one moving wheel measuring 1.05 m in diameter having

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 559


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

(a) 25
CBR = 3 Control
20

Rut depth (mm)


50 mm AC
Geogrid E
300 mm base
Geogrid F
15
Geogrid D
10 Geogrid G
Geogrid A
5 Geogrid B

0
100 101 102 103 104 105
Load cycles
(b) 25
CBR = 3
20
Rut depth (mm)

50 mm AC
350 mm base
15 Geogrid A

10
Control
5 Middle of base
Bottom of base
0
100 101 102 103 104 105
Load cycles
(c) 25 Control, 150 mm base
Control, 250 mm base
20
Rut depth (mm)

Geogrid B, 150 mm base


15 Geogrid B, 250 mm base

10
5 CBR = 8
50 mm AC
0
100 101 102 103 104 105
Load cycles
(d) 25
CBR = 3
20
Rut depth (mm)

50 mm AC

15
10
Control, 300 mm base
Control, 450 mm base
5
Geogrid B, 300 mm base
0
100 101 102 103 104 105
Load cycles

Figure 8. The experimental test results from Webster (1993): (a) 300 mm base,
subgrade CBR = 3; (b) 350 mm base, subgrade CBR = 3; (c) variable base, subgrade
CBR = 8; (d) variable base, subgrade CBR = 3.

560 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

a tread width of 230 mm. The wheel was operated in one direction under a load of 20
kN and was returned to its starting position only under the load of the equipment (9 kN).
For the load cycle test results presented, only the wheel passes under the 20 kN load
were counted.
Moghaddas-Nejad and Small (1996) used a relatively small wheel (230 mm in diame-
ter) under a light load. The wheel-pavement contact area was 2450 mm2. The wheel
traveled in only one direction across the rectangular test section. Similar to the study
of Barksdale et al. (1989), single-track and multiple-track loading was applied.
Barker (1987) applied traffic loads on an outdoor test track with one tire simulating
an F-4 aircraft. Halliday and Potter (1984) performed loading over the course of two
spring seasons, with a rear axle load of 97.5 kN in the first spring and 136 kN for the
second spring. Construction of the sections began in the early spring, while the site was
slightly flooded. Webster (1993) constructed outdoor test lanes beneath a covered struc-
ture. A moving traffic load was applied by one wheel of a C-130 aircraft, delivering a
130 kN load over a tire width of 340 mm (the tire was inflated to provide a contact pres-
sure of 470 kPa). In two of the lanes, the wheel load was applied in a random manner
over five paths contained in a 2.2 m wide area. The remaining two lanes utilized chan-
nelized loading over a 600 mm wide area.
Anderson and Killeavy (1989) constructed three test sections within an access road
and truck staging area. Liquid tanker and multi-axle tractor trailer trucks with axle loads
up to 130 kN trafficked the site. Test sections in the staging yard experienced more se-
vere damage caused by truck turning maneuvers; the trucks did not perform turning ma-
neuvers in the one section within the access road.
Miura et al. (1990) constructed six field sections along a public road. The details on
the in situ site conditions and construction quality control were not provided, although
Miura et al. (1990) state that problems with site variability and quality control may have
influenced the test results. The authors report field test results up to six months after the
road was opened to traffic and do not provide details of the traffic history.

2.3 Pavement Layer Materials and Thickness

In the majority of the studies examined, pavement sections consisting of an AC and


base course layer over a subgrade were constructed. Several studies used a subbase in
combination with a base course layer. The thickness of the subbase and/or base course
layers used in each of the studies examined are listed in Table 2. The Unified Soil Classi-
fication System (USCS) and the American Association of State Highway and Trans-
portation Officials (AASHTO) type and classification of the base, subbase, and
subgrade materials used are given in Table 2. If reported, the CBR of the subgrade is
also listed in Table 2. The structural number (SN) of the experimental test sections was
computed by assuming a layer coefficient for the AC and base layers of 0.4 and 0.15,
respectively, for layer thicknesses given in inches. Due to the general lack of consistent
information, no attempt was made to tailor these numbers to the specific materials used
in the various studies. These values should be considered approximate and are only an
indication of the composite thickness of the structural layers of the pavement. Addition-
al details not provided in Table 2 are provided in the remainder of this section.
Al-Qadi et al. (1994) constructed 18 test sections where the target base thickness was
either 150 or 200 mm. Smith et al. (1995) reported the measured base thicknesses, sev-

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 561


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

eral of which are given in Table 6. The subgrade contained 40 to 47% fines and was
prepared to a degree of saturation ranging from 74 to 83%. The base material contained
a maximum particle size of 51 mm.
Anderson and Killeavy (1989) constructed the nonreinforced section without the geo-
textile separator with a 450 mm thick base layer. The section in the truck yard expansion
area that contained only the geotextile separator had a base thickness of 350 mm. The
section in the truck staging yard that contained the geogrid and geotextile separator had
a base thickness of 200 mm. The geogrid was placed directly on top of the geotextile
separator at the bottom of the base layer.
Barker (1987) constructed sections which were strong in comparison to the majority
of the other studies. The subgrade and subbase were constructed to provide a strong
and stable platform for the open-graded base course layer. The subbase was cement-
treated, meaning that most deformations were forced to occur in the open-graded base
course layer.
Barksdale et al. (1989) constructed test sections that were thin in comparison to other
studies. For one test series, the AC thickness was nominally 25 mm, while the base con-
sisted of a 150 mm thick layer of a rounded gravel and sand mixture. This base material
was considered a poor material in comparison to that used in the remaining three test
series. The three remaining test series used a 38 mm thick AC layer over a 200 mm thick
base that consisted of crushed dolomitic limestone. These pavement layer dimensions
varied slightly from section to section and are presented in Table 7 for certain sections.
The top 450 mm of the subgrade had a CBR value that ranged from 2.6 to 3.2. The CBR
value of the underlying subgrade ranged from 8 to 10. The subgrade was reused for each
of the four test series and resulted in slightly increasing CBR values from one test series
to the next. Only for the second test series was 64 mm of the top of the subgrade removed
to allow for the increased base and AC thickness used in the subsequent three test series.
The subgrade was placed at a degree of saturation of approximately 95%.
Brown et al. (1982) constructed seven pavement section sets, with each set being
constructed at the same time. One section in the set served as the control section without
reinforcement while the other contained a geotextile. The thickness of the AC and base
layers and the CBR strength of the subgrade varied between the sets with the range of
values given in Table 2.
Cancelli et al. (1996) used a constant thickness of AC (75 mm) and base (300 mm).
Additional unreinforced sections used a base layer thickness greater than 300 mm to
examine the base layer equivalency that could be attributed to the geosynthetic. The
crushed aggregate base was compacted to a density that is lower than that which may
be used in practice due to rutting that would have taken place in the loose subgrade. The
fine sand subgrade was similar to that used by Haas et al. (1988). The moisture content
of the sand was kept constant while the density was varied to achieve CBR values rang-
ing from 1 to 18.
Collin et al. (1996) constructed indoor test sections with constant AC thicknesses and
base and subgrade thicknesses that varied linearly within a given test section. This con-
figuration was adopted to compare the performance of different base course thicknes-
ses. The base layer thickness ranged from 150 to 460 mm. The bottom of the subgrade
was level and, therefore, the subgrade thickness ranged from 200 to 450 mm, with the
lower subgrade thickness corresponding to the greater base course layer thickness. The
level bottom of the subgrade was underlain by a 450 mm thick layer of sand and gravel.

562 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

The thickness of the pavement layer materials reported in Table 2 for the study by
Halliday and Potter (1984) are nominal values. The measured thickness varied consid-
erably across the sections due to the soft conditions at the field site. Magnitudes of varia-
tion were not reported for the base layer, while the thickness of the AC layer varied
between 100 and 220 mm. The subgrade CBR values increased from 0.7 at the subgrade
surface to 4.3 at a depth of 300 mm.
For approximately half of the tests performed by Haas et al. (1988), a fine, poorly-
graded sand was used for the subgrade, and subgrade CBR values of 8 and 3.5 were mea-
sured. For weaker subgrade conditions, a peat was blended into the sand to achieve CBR
values of 1 and 0.5.
For the laboratory tests, Miura et al. (1990) used a combination of a base and subbase
over a soft, saturated subgrade. The subgrade was a reconstituted sensitive marine clay
with a natural water content of 129% and a liquid limit of 117%. The clay was placed
as a slurry in the test box and consolidated under a vertical pressure of either 5 or 10
kPa, depending on the test series. This technique probably produced a highly consistent
subgrade, although no information on subgrade mechanical properties from section-to-
section were provided. The test section layer thicknesses listed in Table 2 were common
to all of the laboratory tests.
For the field tests, Miura et al. (1990) varied the thickness of the base and subbase
between the reinforced and unreinforced sections. The reinforced sections contained
either 50 or 100 mm less base and/or subbase than the unreinforced section. The upper
600 to 800 mm of the subgrade consisted of mine tailings compacted to a CBR value
ranging from 4 to 6. Miura et al. (1990) imply that the use of this material introduced
variability in the subgrade condition from one test section to the next. This appears to
be confirmed by the measurement of subgrade stiffness using Benkelman beam tests
performed immediately after construction.
Moghaddas-Nejad and Small (1996) used sand for both the base and subgrade soils
that had mean particle diameters of 4.8 and 0.35 mm, respectively. No information was
provided on the water content, dry density, or degree of saturation to which the subgrade
was compacted. The resulting CBR value for the subgrade was also not provided.
Webster (1993) used a subgrade consisting of a highly plastic clay with CBR values
of 3 and 8, which were achieved by compacting the clay at different water contents. The
base course material was of marginal quality due to the amount of fines. The test sec-
tions were constructed in a covered hanger.

2.4 Geosynthetic Reinforcement Type and Location(s)

The studies reviewed have examined the role of geosynthetic type (geogrid or geotex-
tile) and geosynthetic stiffness, location of placement, and layering on reinforcement
benefit. Table 3 summarizes the pertinent properties of the geosynthetics used in the
studies reviewed. The properties listed in Table 3 were provided by the manufacturer.
Table 4 lists the geosynthetics associated with each study and the placement location
of the geosynthetic with respect to the base course layer (i.e. “bottom” refers to the bot-
tom of the base course layer). For situations where two layers of geosynthetic were used,
both locations are denoted (i.e. “bottom and middle”).
One of the geotextiles selected by Al-Qadi et al. (1994) had a stiffness at 5% strain
that was equal to the stiffness of the geogrid. The geotextile separator used by Anderson

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 563


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

and Killeavy (1989) was a weak, nonwoven, needle-punched material with a grab
strength of 556 N and was believed to offer no reinforcement. Barksdale et al. (1989)
used Geogrid A and Geotextile D, where Geotextile D had a stiffness of approximately
three times that of Geogrid A.
Cancelli et al. (1996) used five different types of geogrids and one geotextile. The
stiffness and the manufacturing process of the geogrids differed. Due to the relatively
small size of the test box, the geosynthetics were wrapped upwards around the sides of
the box to simulate anchorage. Haas et al. (1988), however, showed for a similar testing
arrangement that the concept of anchorage for this type of application was not relevant
because the geogrid experienced tensile strains only within a 250 mm radius from the
load center.
The only geotextile used by Halliday and Potter (1984) was strong, with an ultimate
tensile strength of 83 kN/m and an extension strain at failure of 14.7%. In the laboratory
tests performed by Miura et al. (1990), the effect of geogrid stiffness was examined by
using three geogrids of differing stiffness from the same manufacturer. Webster (1993)
concentrated on the use of Geogrid B; this geogrid was used exclusively in test Lanes
1 and 2, and the remaining geogrids were used in Lanes 3 and 4.

2.5 Instrumentation

Several authors have noted the importance of an instrumentation program to aid in


the evaluation of test section performance. Instrumentation can aid in determining why
the section performed as it did, in formulating an analytical solution to the problem, and
in devising a design methodology. While instrumentation can be useful, it is often times
cost prohibitive to place a sufficient amount of sensors to meet the objectives listed
above. The instruments are expensive and the man hours necessary to properly install,
monitor, and interpret the data are often times prohibitive.
Table 5 provides a list of the instrumentation and/or measurement parameters used
in the studies reviewed. When an instrument has been used to measure a basic parame-
ter, such as stress, strain, water content, or temperature, the parameter itself has been
listed as opposed to the type of sensor that was used for the measurement. On the other
hand, when a parameter, which is more or less specific to the type of instrument used
for its determination, is reported, such as the layer modulus measured using a falling
weight deflectometer (FWD) test, the test or instrument is listed.

2.6 Performance (Failure) Criteria

The manner in which performance or failure is defined can significantly influence the
conclusions that are made regarding the level of improvement obtained by adding geo-
synthetics to pavement systems. Performance can be defined in terms of the response
of the pavement section to nondestructive tests, such as a FWD test, or by the observa-
tion and measurement of pavement surface features such as rutting and crack develop-
ment. The instrumentation included in the various studies noted in Section 2.5 can also
be used to evaluate and compare pavement response. The observation of rut develop-
ment with load cycle number appears to be the most common method used. The second
column of Table 10 summarizes the methods used in the studies reviewed. Additional

564 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

details are also provided in Section 4.1 where test results from these studies are more
closely examined.

2.7 Other Variables

Anderson and Killeavy (1989) noted that traffic on the site prior to the placement of
the AC layer prestressed the geogrid. The magnitude and distribution of this prestress-
ing was not evaluated. To examine the influence of construction traffic on a partially
completed road, Barksdale et al. (1989) carried out prerutting operations on the aggre-
gate base layer prior to the placement of the final AC layer. Prerutting consisted of al-
lowing the loaded wheel to travel back and forth along the centerline of the section until
a rut depth of approximately 50 mm was achieved. The rut was filled and compacted
prior to the placement of the AC layer. Prerutting was thought to induce tensile strain
in the geosynthetics, thereby improving the subsequent reinforcement potential of the
geosynthetic when the section was surfaced. Prerutting was also performed on unrein-
forced sections where it was thought to have induced local densification of the aggre-
gate base soils.
For one test section, Barksdale et al. (1989) carried out a pretensioning operation of
the geogrid. This was accomplished by applying a uniform lateral tensile load equal to
approximately 35% of the ultimate strength of the material. The load was applied prior
to the placement of the remaining base material and was released only after the base
layer was brought up to its final lift height and compacted. Haas et al. (1988) also ex-
amined the effect of pretensioning the geogrid.

3 HISTORICAL EXPERIMENTAL TEST RESULTS

The purpose of this section is to summarize the major experimental findings of all of
the studies described in Section 2. In this section, the authors of the current paper did
not attempt to synthesize or evaluate the test results. A critical review of the major ex-
perimental findings (i.e. a thorough synthesis and interpretation of the test results) is
presented in Section 4.

3.1 Al-Qadi et al. (1994, 1997) and Smith et al. (1995)

These three studies conclude that the improvement mechanisms associated with geo-
grids and geotextiles were different. Mixing of the subgrade and base course material
was observed in the control and geogrid-reinforced sections, while contamination was
not observed in the geotextile-reinforced sections.
Separation was identified as the predominant function of the geotextile, which proved
to be more important than the reinforcement function of the geogrid. As a result, the
geotextile-reinforced sections performed better than the geogrid-reinforced sections
with regard to rut development. The geotextile sections carried two to three times the
number of load repetitions than similar control sections, while the geogrid sections typi-
cally performed the same as control sections. In terms of service life, similar results
were obtained.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 565


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

3.2 Anderson and Killeavy (1989) and Barker (1987)

Anderson and Killeavy (1989) performed FWD tests on field test sections that were
designed to yield equivalent performance. The test results indicated that similar sub-
grade deformations were observed in each section, and reduced asphalt strain and better
load distribution occurred in the geogrid-reinforced sections. Back-calculation tech-
niques showed that the geogrid section had the largest base layer modulus (560 MPa),
followed by the geotextile section (400 MPa) and the control section (170 MPa). Visual
observation of the sections after a one year and a ten year monitoring period indicated
little difference between the surface rut characteristics of the three sections.
Barker (1987) compared one test section reinforced with Geogrid B, where the rein-
forcement was placed in the middle of the 150 mm thick base, to an identical unrein-
forced test section. In terms of rut depth development, the reinforced section performed
worse than the control section for rut depths less than 18 mm. For instance, for 10 traffic
passes, the reinforced section showed 4 mm of displacement while the control section
showed less than 1 mm. For rut depths greater than 18 mm, the performance of the rein-
forced section, as compared to the control section, continued to increase.

3.3 Barksdale et al. (1989)

Barksdale et al. (1989) noted that a geogrid with less stiffness than a geotextile gener-
ally led to better performance. This improved performance was attributed to the inter-
locking ability of the geogrid and its role in preventing lateral spreading of the base
layer soil. The test results suggested that the geotextile required significantly higher de-
formation in order to mobilize the same reinforcing potential as the geogrid. The geo-
textile was superior to the geogrid in preventing mixing of the subgrade with the base
course soil.
The importance of geosynthetic location was noted, with the proper location being
dependent on the quality and thickness of the base layer soil. For the loading used in
this study, the most effective location appeared to be in the middle of the base layer. It
was suggested that higher locations in the base layer may have produced even better
results. The process of prerutting had a significant effect on performance: prerutting a
control section was more beneficial than the inclusion of a geotextile. Geosynthetic pre-
stressing yielded positive improvements, although the practical consequences of such
a construction activity were questioned.

3.4 Brown et al. (1982) and Cancelli et al. (1996)

Brown et al. (1982) demonstrated that the two nonwoven geotextiles tested were inef-
fective in improving performance, and in most cases resulted in larger deformations as
compared to the control sections. Observed slippage of the base soil and the geotextile
was noted as the cause of this occurrence. Brown et al. (1982) noted that geotextiles
were beneficial for construction on wet soils where the geotextile functions of separa-
tion and filtration were required.
Cancelli et al. (1996) showed that a woven, flexible geogrid and a high-modulus geo-
textile-reinforced section performed marginally better than the control section, while
stiff geogrids and a multilayer geogrid provided significant improvement. For very

566 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

weak subgrades, two layers of a relatively weak geogrid were found to be more effective
than one layer of a very stiff geogrid. Geogrids were effective in allowing for reduced
base course layer thicknesses and in extending the service life of a section of a given
thickness. Geogrid benefits generally increased with decreasing subgrade strength and
increasing allowable rut depth.

3.5 Collin et al. (1996) and Halliday and Potter (1984)

Collin et al. (1996) showed that test section performance improved due to geogrid
reinforcement and increased with increasing base thickness up to a base thickness of
255 mm, and then decreased with continued increasing base thickness when the geogrid
was placed at the bottom of the base. The number of wheel loads carried by the rein-
forced sections exceeded that of the control section by a factor as great as 10 for a 25
mm rut depth. Geogrid reinforcement decreased the initial pavement deformations that
occur during the first several hundred load cycles before the section stiffens. The rein-
forcement caused the deflection versus load cycle curves to flatten and become approx-
imately linear.
Similar to the study by Brown et al. (1982), Halliday and Potter (1984) demonstrated
no improvement in the rate of development of rut depth or vertical strain in the subgrade
when a strong woven geotextile was used for reinforcement. The presence of the geo-
textile appeared to have no influence on the structural quality of the roadway but facili-
tated the construction process through the mechanism of separation. It was estimated
that the subbase soil penetrated into the soft subgrade by a depth of 70 mm in the control
section where a geotextile was not present.

3.6 Haas et al. (1988)

The laboratory experiments performed by Haas et al. (1988) demonstrated the impor-
tance of variables such as geogrid placement position, base course thickness, and sub-
grade strength. In general, it was shown that reinforced sections carried three times the
number of load cycles as compared to a similar control section, and that reinforcement
allowed up to a 50% reduction in base course thickness.
The optimum geogrid location was at the bottom of thin bases and at the midpoint
for base thicknesses in excess of 250 mm. For very weak subgrades, optimum perfor-
mance occurred when two layers of geogrid were used, where in excess of three times
the number of load cycles could be carried. Strain measurements on the geogrid showed
that tensile strains in excess of 1% were observed under the load center and that these
strains diminish to zero at a radius of 1.5 times the radius of the load plate, indicating
that long anchorage lengths were not required in such an application.

3.7 Miura et al. (1990), Moghaddas-Nejad and Small (1996), and Webster (1993)

The laboratory test results reported by Miura et al. (1990) demonstrated that geogrids
of increased stiffness provide for increased levels of improvement. Improvement was
closely related to the magnitude of strain measured in the geogrid during loading. The
performance of the field sections also improved with the use of more stiff geogrids, but
not to the same level as that demonstrated in the laboratory tests.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 567


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Moghaddas-Nejad and Small (1996) demonstrated that placing Geogrid B at the bot-
tom of a thin base results in a 40% decrease in rut depth for single-track wheel passes
and a relatively light load, whereas geogrid placement in the middle of the base results
in a 70% decrease. These values correspond to 5000 wheel passes. For multiple-track
loading, both placement positions resulted in a 50% decrease in rut depth. The deflec-
tion basins beneath the reinforced sections were wider and shallower than those of the
unreinforced section, suggesting that the reinforcement improved the pressure distribu-
tion on the subgrade. Measurements of vertical deformation at the level of the geogrid
showed that there was substantial improvement when load-induced vertical deforma-
tions of the geogrid were less than several millimeters. This suggests that improved per-
formance was not due to a tensioned membrane effect, but was due to lateral restraint
of the base soil.
Webster (1993) showed that flexible geogrids were not as effective as stiff geogrids
for base layer reinforcement in flexible pavements. The torsional rigidity of the geo-
grid, which is governed in part by the characteristics of the junctions, was the most im-
portant factor in determining the suitability of the geogrid product. Sections reinforced
with a stiff geogrid carried as much as 21 times the number of traffic loads as an unrein-
forced section. The value of this factor depended on the placement position of the geo-
grid, the thickness of the base layer, and the strength of the subgrade soil.

4 INTERPRETATION OF EXPERIMENTAL TEST RESULTS

The purpose of this section is to summarize, interpret, and synthesize the major find-
ings from the 12 studies described above. The majority of the interpretations made in
this section are those of the authors of the current study and are not necessarily derived
from direct statements and conclusions made by the authors of the reviewed studies.
Major findings have been summarized in the sections below for each study according
to variables or conditions believed to control test section performance. Table 10 at the
end of this paper summarizes these findings.

4.1 Overview

This section provides details required to understand the methods used to interpret data
and are common to Sections 4.2 to 4.8. The majority of the comparisons made in Sec-
tions 4.2 to 4.8 are made with reference to a traffic benefit ratio (TBR) which is defined
as the ratio of the number of cycles to achieve a particular rut depth in a reinforced sec-
tion to that of an unreinforced section of identical section thickness, material properties,
and loading characteristics. A TBR value is used because most of the studies reviewed
defined performance in terms of rut depth. The rut depth that is used as the basis for this
comparison is provided when necessary.
In the papers by Al-Qadi et al. (1994, 1997) and Smith et al. (1995), three approaches
were used to present the measured data. One approach was to plot the raw permanent
displacement of the loading plate against the load cycle number. Recognizing that rela-
tively large permanent displacements occurred in the early stages of loading, a second
series of plots was created where the displacement occurring during the first 25 load
cycles was subtracted from the remaining measurements. The surface displacement

568 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

that occurred during these first 25 load cycles was said to be comparable to that which
takes place during the first several months of normal traffic loading. Figure 2 presents
the raw rut depth values versus load cycle number for 10 test sections (Smith et al.
1995). The raw rut depth values are given in Figure 2 because they are consistent with
the results reported in the other studies. The numbers assigned to the different sections
correspond to those used by Smith et al. (1995). Table 6 provides details on each of the
constructed test sections.
Smith et al. (1995) recognized that small differences between the thickness of the
pavement layers, the CBR strengths of the subgrade, and the load magnitude achieved
by the loading system existed from one test section to the next. Therefore, the authors
proposed a method to compare sections whereby the allowable equivalent single-axle
loads (ESAL) to failure were calculated based on the geometry and properties of the
test section and were compared to the applied ESAL for the section.
The number of ESALs applied to each section was determined for a 25 mm rut depth
(corrected rut depth measurements were used). This was accomplished by using the av-
erage load applied to the section computed over the duration of the test, and then cor-
recting this load for load arrangement, contact area, and load duration. These
corrections resulted in a constant factor being applied to the average load for each sec-
tion. Use of the average load accounts for load variation from section-to-section. The
allowable ESALs for each section was computed using the AASHTO (1993) method
and the computer program KENLAYER (Huang 1993).
The resilient moduli values for the various pavement layers are required by both meth-
ods to calculate the allowable ESAL values. Modulus values for the subgrade were gen-
erated from estimates of the CBR value for each section. The CBR value was estimated
by comparing the water content and dry density of the subgrade to laboratory-soaked
CBR test results for various pre-soaked combinations of these two variables. The com-
puted allowable ESAL values accounted for variations in layer thickness and strength.
For most sections, the computed allowable ESAL values exceeded the applied va-
lues. For allowable ESAL values computed using the KENLAYER program, a loading
ratio (the ratio of the allowable ESAL to the applied ESAL) was also calculated for each
section. The calculated loading ratio values for each section ranged between 0.14 and
1.18. For the control sections, the loading ratio values ranged between 0.31 and 0.49
with an average value of 0.4. This average value was used as a measure of the degree
in which KENLAYER over or under predicted the actual applied loads. Values in excess
of 0.4 for sections containing geosynthetics were attributed to the reinforcing or stabi-
lization function of the geosynthetic. Values less than 0.4 indicated sections which per-
formed worse than similar control sections.
Benefit ratios for the sections were then computed by dividing the load ratio for a par-
ticular section by the average value (0.4) for the control sections. The benefit ratios that
were calculated for the control sections ranged from 0.78 to 1.2. These benefit ratios
are equivalent to the TBR values defined previously because they represent the ratio
of load cycles of a reinforced section to an unreinforced section for a given rut depth.
The test results of rut depth versus load cycle number for the studies performed by
Barksdale et al. (1989), Cancelli et al. (1996), Collin et al. (1996), Haas et al. (1988), Miu-
ra et al. (1990), and Webster (1993) are shown in Figures 3 to 8, respectively. Any small
changes in the data presented in Figures 3 to 8 as compared to the original data are due
to scaling errors. Figures 3 to 8 and Tables 6 to 9 provide descriptions of each test section.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 569


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

4.2 Geosynthetic Type

Of interest in this section is whether geogrids or geotextiles provide more reinforce-


ment and whether factors such as the geosynthetic axial and torsional stiffness and the
process used to manufacture the geosynthetic influence reinforcement potential and
performance. As in all of the test sections discussed in Section 4, no mention was made
of the studies where the particular factor being discussed has not been included.

4.2.1 Al-Qadi et al. (1994, 1997) and Smith et al. (1995)

Al-Qadi et al. (1994) demonstrated that when separation and filtration were the main
problems associated with a pavement design, geotextiles offered superior improvement
as compared to geogrids. The amount of improvement for the different geosynthetics
depended on which of the three data interpretation techniques were used for comparison
(Section 4.1). For stronger subgrades with a nominal CBR value of 4, the amount of
improvement depended on which sections were used for comparison and whether the
raw data, or the data corrected for deformation after the first 25 cycles, was used. Table
11 is a summary of calculated TBR values for a 25 mm rut depth from the data reported
by Smith et al. (1995), and is shown in Figure 2a.
Two test sections were reinforced with Geogrid B (Section 4 and 12, Figure 2a) and
two with Geotextile A (Section 3 and 11, Figure 2a). Section 6 and 10 (Figure 2a) were
used as control sections for comparison purposes, which allowed two TBR values to be
computed for each reinforced section. Greater TBR values were calculated when con-
trol Section 6 was used in combination with the raw data.
The differences in TBR values when using the different control sections was signifi-
cant, and was the result of different pavement thicknesses, strength properties, and ap-
plied load magnitudes of each test section. A summary of this data is presented in
Table 6. Despite these differences, the data in Table 11 indicates that the geotextiles
provided superior performance when compared to the geogrid. The test results re-
ported by Smith et al. (1995) also indicate no discernible difference between the two
geotextiles used (i.e. Geotextile A and B), indicating that the tensile modulus of the
reinforcement was not important.
Test sections were also constructed on a weaker subgrade with a nominal CBR value
of 2. At a 25 mm rut depth, TBR = 3.0 for two sections containing Geotextile A (Section
16 and 17, Figure 2b), and TBR = 1.4 for a section containing Geogrid B (Section 18,
Figure 2b). These TBR values were generated from raw data. For data that was cor-
rected for deformations occurring during the first 25 load cycles, TBR values of 1.8,
2.0, and 0.9 were obtained for these same test sections.
The test results were also evaluated using the method of analysis described in Section
4.1 that accounts for variations in properties and the applied load magnitude from section
to section (i.e. comparison of allowable ESAL to applied ESAL). Traffic benefit ratios
ranging from 2.1 to 3.5 were calculated for sections containing Geotextile A, while a val-
ue of 2.5 was calculated for the only section containing Geotextile B. For the three geo-
grid sections, TBR values of 0.34, 0.56, and 1.1 were obtained. Most of these TBR values
are considerably less than one (which is equivalent to no improvement in performance)
and outside the range of values obtained for the control sections (0.78 to 1.2), indicating
that the geogrid sections performed worse than an equivalent unreinforced section.

570 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

These results suggest that the separation function of the geotextile was the most criti-
cal feature in terms of improving test section performance for the type of subgrade used.
This does not help explain, however, why the geogrid-reinforced sections performed
worse than the control sections. If the improved performance of the geotextile sections
was due to separation and filtration functions, then it would be expected that the perfor-
mance of the geogrid sections would be no worse than the control sections and perhaps
better to a much lesser extent than the geotextile sections, due to the same functions.
These results tend to suggest that the differences in layer properties between the sec-
tions was greater than that which could be accounted for by the parameters used to de-
fine these properties and/or the analysis technique used to compare sections was not
able to account for the property differences observed. The base course soil used con-
tained particles as large as 50 mm which may have prevented positive interaction with
the 25 mm by 33 mm aperture size of the geogrid.

4.2.2 Barksdale et al. (1989)

Barksdale et al. (1989) compared the performance of one type of geogrid and one type
of geotextile rather than the influence of the geosynthetic modulus. The modulus of the
geotextile was greater than 2.5 times that of the geogrid. For test sections using multi-
ple-track loading and geosynthetics placed in the middle of the 200 mm thick base layer,
the geogrid (Section 4, Figure 3a) performed better in terms of rut depth development
than the geotextile (Section 3, Figure 3a). A control section was not included in this test
series. Using the control section from the preceding test series (Section 1, Figure 3a),
TBR values of 17 and 2.5 for the geogrid and geotextile sections were calculated, re-
spectively, for a 12.5 mm rut depth.
An examination of the constructed layer thickness and subgrade properties of Section
1, 3, and 4 provides insight as to whether or not these sections are in fact comparable.
The thickness of the pavement section layers and the subgrade strengths are listed in
Table 7. The average subgrade strength was reported for a given test series and not the
subgrade strength of each test section. Despite the apparent differences in section thick-
ness and subgrade properties, a geotextile-reinforced section in the same test series as
the control section with the geotextile in the middle of the 200 mm thick base (Section
2, Figure 3a) resulted in TBR = 2.1. This TBR value is comparable to the TBR value
obtained for the geotextile-reinforced section from the next test series that had a slightly
stronger subgrade (TBR = 2.5).
Traffic benefit ratios of 2.2 and 2.8 were calculated for two single-track test sec-
tions reinforced with geotextile and geogrid layers placed in the middle of the 200
mm thick base layer, respectively, for a 12.5 mm rut depth. These TBR values are
closer than the TBR values obtained for the multiple-track tests on sections that also
contained a geosynthetic in the middle of the base layer. These two sections were
compared to a control section from the previous test series, as was done for the multi-
ple-track tests. A TBR value of 4.7 was calculated for a geotextile-reinforced test sec-
tion in this previous test series.
The single-track test results are similar in trend to the multiple-track tests, but TBR
values of different magnitude were obtained. It is possible that the close proximity of
the single-track tests to the multiple-track tests (which were performed first) had an im-
pact on the consistency of test results.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 571


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

4.2.3 Brown et al. (1982)

Brown et al. (1983) demonstrated that there was no improvement in performance for
the two geotextiles used. For five of the seven test sections, the development of rut depth
with load cycle was worse for the sections reinforced with a geotextile. The transient
stress and strain response of the test sections also worsened due to the geotextile rein-
forcement.

4.2.4 Cancelli et al. (1996)

For a weak subgrade with CBR = 1, Cancelli et al. (1996) demonstrated that the geo-
synthetic modulus and manufacturing process affects the reinforcement benefit (Figure
4a). For four of the geogrid products used and for a 25 mm rut depth, Geogrid J, I, A,
and H required 4280, 2480, 1020, and 1020 cycles of load, respectively. The modulus
of these geogrids decreased from Geogrid J, I, A, to H, with Geogrid I and A having
identical modulus values. An identical unreinforced section reached a 25 mm rut depth
in approximately 80 load cycles, resulting in TBR values well in excess of those ob-
tained by other studies. Geogrid A was more torsionally rigid than Geogrid H, indicat-
ing that torsional rigidity was less important than that noted by Webster (1993).
For test sections built on a subgrade with CBR = 3 and for a 25 mm rut depth, TBR
= 1.5 for Geogrid D and Geotextile F, while TBR = 5.1 and 6.7 for Geogrid H and J,
respectively (Figure 4b). In the unreinforced sections, very little mixing of the base
course and the subgrade occurred for subgrades other than the subgrade with CBR =
1, indicating that the benefit from the geotextile was most likely due to reinforcement.
The improvement in performance due to increased geogrid stiffness decreased as the
subgrade strength increased.

4.2.5 Collin et al. (1996)

Collin et al. (1996) showed that increased geogrid modulus was insignificant in im-
proving performance when the geogrid was placed at the bottom of the base and when
the base layer was either thin (< 175 mm) (Figure 5a) or thick (> 275 mm) (Figure 5c).
For an intermediate base layer thickness, the more stiff geogrid produced TBR values
approximately twice those of the less stiff geogrid (Figure 5b).
The difference in TBR values for different base layer thicknesses did not appear sig-
nificant until a rut depth of approximately 15 mm was reached. It is important to note
that the thick base layer sections were placed on a subgrade that was only 350 mm thick
in some areas; therefore, potentially resulting in a section which would be stronger than
one where the thickness of the subgrade layer was more deep.

4.2.6 Miura et al. (1990)

Miura et al. (1990) used two laboratory test sections with the same properties, but dif-
ferent types of geogrid to compare and evaluate the performance improvement due to
geogrid modulus. The subgrade for both sections was consolidated under a 10 kPa pres-
sure. In Section 2 (Figure 7a), Geogrid B was placed between the subbase and the sub-
grade, while in Section 3 (Figure 7a), Geogrid C was placed in the same location.

572 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Geogrid B has the same modulus in the machine direction as Geogrid C and a modulus
18% greater than Geogrid C in the cross-machine direction. For rut depths greater than
approximately 2 mm, the section with Geogrid B shows less rut depth for the same load
cycle number. Using Section 1 (Figure 7a) as the control section, at a 5 mm rut depth,
TBR = 4.2 and 3.0 for the test sections containing Geogrid B and Geogrid C, respectively.
The maximum geogrid tensile strain measured was 0.3% directly beneath the load
center. At a radius of 2.5 times the radius of the load plate, the geogrid strain was equal
to zero, and at greater radial distances the strains were compressive. These test results
are similar to those presented by Haas et al. (1988). There was a correlation between
the performance of the test sections and the amount of strain developed in the geogrid
during cyclic loading, indicating that the tensile strength and modulus of the geogrid
was responsible for the observed improvement.
In the field tests performed by Miura et al. (1990), the criteria used to evaluate section
performance revealed different levels of improvement between Geogrid B and C rein-
forced sections. These results are complicated by the apparent differences in pavement
layer stiffness from section to section as described in Section 2. In terms of rut depth,
the sections reinforced with Geogrid B and C gave similar performances and slightly
better responses than the unreinforced section which contained 50 mm less base course
and 50 mm less subbase. The deflection data from the Benkelman beam tests indicate
that the sections containing the more stiff geogrid, Geogrid B, performed better than
those sections containing Geogrid C. Sections reinforced with Geogrid B performed
slightly worse than the unreinforced section with the thinner base and subbase layers.
An evaluation of the pavement crack percentage revealed the same results as the Ben-
kelman beam tests, although there is a greater scatter in the data.

4.2.7 Webster (1993)

Webster (1993) constructed two lanes using channelized traffic to compare differ-
ences in performance between Geogrid A, B, D, E, F, and G (Figure 8a). Each section
contained 50 mm of AC, 350 mm of base on top of a subgrade with a CBR = 3, and
geogrids placed at the subgrade-base interface. Test sections containing Geogrid A, B,
and G were constructed in one lane, while the remaining geogrid-reinforced sections
were constructed in the other lane. The study has been criticized because the first lane
did not contain a control section (Better Roads 1995), and that differences in perfor-
mance were due to differences in subgrade strength and base course density and not
geosynthetic type.
The base course and subgrade CBR values that were measured after traffic loading
on in-place material were reasonably consistent from section to section. Furthermore,
the section with the weakest subgrade performed the best, while the section with the
strongest subgrade gave the poorest performance. Also, the control section reached a
25 mm rut depth in 106 passes, which is comparable to a control section in another lane
with the same subgrade and 50 mm less gravel where 90 passes were necessary to reach
this same rut depth.
For a 25 mm rut depth, TBR values of 2.7, 4.7, 1.1, 0.9, 0.9, and 1.6 were reported
for Geogrid A, B, D, E, F, G, and H, respectively. With the exception of Geogrid F, these
results suggest that the woven geogrids perform poorly in this application. These test
results also suggest that the modulus of the geogrid was not the overriding factor gov-

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 573


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

erning performance. A secant aperture stability modulus was developed and correlated
to TBR values. The secant aperture stability modulus describes the torsional rigidity
of the geosynthetic. With the exception of Geogrid F, greater TBR values were obtained
for geogrids with a greater stability modulus.

4.3 Geosynthetic Location and Layering

Several studies have examined the importance of proper geosynthetic location within
the base course layer, and whether additional benefits can be achieved by placing more
than one layer of geosynthetic in the base.

4.3.1 Barker (1987)

Barker (1987) constructed a one-layer, reinforced test section containing Geogrid B


placed in the middle of a 150 mm thick base resting on a strong subbase-subgrade sys-
tem. The traffic load applied to the section was heavy in comparison to the majority of
the other studies. The reinforced section performed worse than the control section up
to a 18 mm rut depth; for greater rut depths improvement was observed. These results
are compared to the results obtained by Moghaddas-Nejad and Small (1996) and Web-
ster (1993) at the end of this section to illustrate the importance of geosynthetic place-
ment position.

4.3.2 Barksdale et al. (1989)

Barksdale et al. (1989) placed geosynthetics at the bottom and middle of the 200 mm
thick base course layer. For the sections utilizing multiple-track loading, superior re-
sults were obtained when the geosynthetic was placed in the middle of the base layer
as compared to the bottom of the base. For a 12.5 mm rut depth, TBR = 1 and 1.6 for
the test sections with the geogrid (Section 6, Figure 3b, Section 5 = control section) and
geotextile (Section 8, Figure 3b, Section 7 = control section) placed at the bottom of
the base, respectively. Traffic benefit ratios of 17 and 2.5 were obtained when these ma-
terials were placed in the middle of the base (Section 1, 3, and 4, Figure 3a).
Similar to the comparisons made in Section 4.2.2, the single-track loading TBR val-
ues showed similar trends to the multiple-track loading TBR values, but varied in mag-
nitude. For the two geotextile sections,TBR = 1.0 when the geotextile was placed at the
bottom of the base and TBR = 2.2 and 4.7 when placed in the middle of the base. For
the geogrid sections, TBR = 0.7 when the geogrid was placed at the bottom of the base
and TBR = 17 when the geogrid was placed in the middle of the base. The test section
which resulted in TBR = 0.7 was the only geogrid section that performed worse than
the control section. It is not clear why the multiple-track loading sections and the single-
track sections performed differently. As mentioned in Section 4.2, the close proximity
of the wheel paths between the two sections may have had an influence on the single-
track results.
Vertical strains in the subgrade were less when a geosynthetic was placed at the bot-
tom of the base layer; however, the overall performance in terms of rut depth was better
when the geosynthetic was placed in the middle of the base layer. Generally, there were
greater deformations in the base than in the subgrade and AC layers because a greater

574 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

reinforcement benefit was achieved with base reinforcement as opposed to subgrade


reinforcement which limited strain in the subgrade. This also demonstrates the prob-
lems associated with making an evaluation of pavement performance based on one
component of stress or strain at a particular location in the pavement structure rather
than a variable or set of variables indicative of the state of stress or strain in the structure
as a whole. Chan (1990) measured 46% less permanent deformation in a geogrid-rein-
forced base layer section when compared to the same geotextile-reinforced section,
where the reinforcement was placed in the middle of the base layer.

4.3.3 Brown et al. (1982)

Brown et al. (1982) constructed one geotextile-reinforced test section which con-
tained two layers of Geotextile I, one at the bottom of the subgrade and the other in the
middle of the base layer. The reinforced section performed slightly worse than the con-
trol section.

4.3.4 Cancelli et al. (1996)

Cancelli et al. (1996) examined the effect of geosynthetic layering. Geogrid H was
placed at the bottom and middle of the base layer over a subgrade with CBR = 1 (Figure
4a). The section required 18,500 cycles of load to reach a 25 mm rut depth, whereas an
identical section containing one layer of Geogrid H at the bottom of the base required
1020 cycles of load. The unreinforced control section reached a 25 mm rut depth in
approximately 80 cycles. Even when the most inflexible geogrid (Geogrid J) was used
for the single-layer system, the two-layer system still performed substantially better.

4.3.5 Collin et al. (1996)

Collin et al. (1996) did not vary the position of the geosynthetic as it was always
placed at the bottom of the base layer, but did examine the effect of base thickness,
which provides indirect information on the required geosynthetic position for optimum
reinforcement. The reinforcement effect was more significant when the base layer
thickness varied between 175 and 275 mm. For a 25 mm rut depth, the optimal geosyn-
thetic position appeared to be 255 mm below the top of the base. These results suggest
that an optimal reinforcement benefit will be achieved when the geosynthetic is placed
at an appropriate distance away from the applied load.

4.3.6 Haas et al. (1988)

Haas et al. (1988) showed that for a moderately strong subgrade (CBR = 8) and for
moderately thick test sections, the difference between placing a geogrid at the bottom
in comparison to the middle of the base layer was slight (Section 2 and 3, Figure 6a),
with the bottom position providing the better results. For a 25 mm rut depth, TBR = 2.4
and 2.1 for the bottom and middle positions, respectively. When the geogrid was placed
at the interface of the AC and base (Section 4, Figure 6a), the section performed slightly
worse than the control section (Section 1).

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 575


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

This behavior was substantiated by measurements of strain on the reinforcement


which showed that when the geogrid was high in the base layer, the geogrid experienced
small compressive strains until a rut depth of over 15 mm was developed, after which
it was able to provide reinforcement by developing positive tensile strains. Proper loca-
tion of the geogrid in the middle to bottom of the base layer allowed for positive tensile
strains to develop immediately upon load cycle application.
For a weak subgrade (CBR = 1) and a 300 mm thick base, TBR = 11 was observed
for a 25 mm rut depth when the geogrid was located in the middle of the base (Section
5 and 6, Figure 6b). This value was much higher than those obtained for other configura-
tions, indicating that, for thicker base layers, the middle of the base is superior to the
bottom. A test using these same variables but with the geogrid placed at the bottom of
the base was not performed to make this comparison directly.
For a very weak subgrade (CBR = 0.5) and a thick base layer, placing the geogrid
slightly into the subgrade provided better results in comparison to placing the geogrid
at the subgrade-base interface (Section 9 and 8, Figure 6c). For a 25 mm rut depth and
for a subgrade of this strength, placing a single layer of geogrid at the bottom of the base
resulted in slightly poorer performance as compared to the control section (TBR = 0.8)
at a 20 mm rut depth (Section 8, Figure 6c). Placing two layers of geogrid (middle and
bottom of the base, Section 10, Figure 6c) resulted in slightly better performance (TBR
= 2.7) as compared to the single layer placed in the subgrade (TBR = 2.3) for a 25 mm
rut depth. These values were seen to increase moderately for greater rut depths.

4.3.7 Miura et al. (1990)

The laboratory test results reported by Miura et al. (1990) show a strong dependency
of improvement on the geosynthetic position within the base and subbase layers. Sec-
tion 2 and 4 (Figure 7a) correspond to one layer of Geogrid B placed at the bottom of
the subbase and the bottom of the base layer, respectively. For a 5 mm rut depth, TBR
= 4.2 and 8.0 for Section 2 and 4, respectively. This difference in TBR values is consid-
erably greater than that for Section 2 and 3, which reflect a change in geogrid stiffness.
There was less geogrid strain directly beneath the centerline of the load in Section 4.
While this may seem contradictory because a greater TBR value was calculated for Sec-
tion 4, the distribution of geogrid strain was different for the two sections.
In Section 4, the geogrid strain was uniformly distributed away from the centerline
of the load, whereas in Section 2 the majority of the geogrid strain was concentrated
near the vertical projection of the loaded area. These test results indicate that the way
in which the geogrid carries the strain, or load, was dependent on the geogrid position
and in turn influences the geogrid reinforcement potential.
Section 6 and 7 (Figure 7b) also demonstrate the importance of geogrid placement
position. In Section 6, two layers of Geogrid A were used - one at the bottom of the base
and the other at the bottom of the subbase. Two layers of Geogrid A were also used in
Section 7 - one in the middle and the other at the bottom of the base. For a 5 mm rut
depth, TBR = 5.4 and 1.9 for Section 6 and 7, respectively. These results indicate that
the geogrid reinforcement potential is not fully realized if it is placed too high within
the base layer.
For Section 6 and 7, it appeared that placing the geogrid 125 mm below the pavement
surface (75 mm into the base) was not sufficiently deep to realize the reinforcement po-

576 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

tential of the geogrid. The Section 6 geogrid strain measurements appear to contradict
the Section 2 and 4 strain measurements. The strain distribution of the two geogrid lay-
ers in Section 6 show that the lower geogrid carried more load than the upper geogrid,
while the strain distribution of the two layers was similar. This indicates that the lower
geogrid position was optimal, while the results from Section 2 and 4 indicate the oppo-
site. Consideration should be given to the fact that the geogrid modulus and subgrade
strength of these two sets of tests (Section 2 and 4, and Section 6 and7) were different,
which may indicate that the optimal geogrid position was not unique, but was dependent
on the geosynthetic type and pavement layer mechanical properties.
The field test results reported by Miura et al. (1990) are ambiguous with regard to the
effect of geogrid position in the test section. Test sections containing one layer of Geo-
grid B or C at either the bottom of the base or the bottom of the subbase were evaluated.
In terms of rut depth, placing the geogrid at the bottom of the subbase was optimal, al-
though the difference in rut depth between these sections was no more than 2 mm. In
terms of the Benkelman beam tests, one test series revealed that the bottom of the base
was the optimal geogrid position, while the second test series revealed the opposite.
Differences in pavement crack percentage were also ambiguous. These results corres-
pond to a monitoring period of only six months.

4.3.8 Moghaddas-Nejad and Small (1996)

Moghaddas-Nejad and Small (1996) constructed reinforced and unreinforced test


sections with a significantly thin base layer (40 mm) and a thin AC layer (20 mm) and
used a relatively light applied load of 0.42 kN. The two reinforced sections contained
Geogrid B in the middle and bottom of the base layer, respectively. Single-track load
tests were performed on the unreinforced and the two reinforced sections and termi-
nated after 5000 load cycles. Rut depths of 22, 12.5, and 6 mm developed in the unrein-
forced, bottom reinforced, and middle reinforced sections, respectively. For the
multiple-track load tests, a 15 mm rut depth developed in the unreinforced section and
the two reinforced sections behaved similarly to the single-track load tests, developing
7 mm rut depths after 4000 load cycles.
Considerably less deformation occurred in the base layer when the geogrid was
placed in the middle of the base, indicating that the predominant reinforcement mecha-
nism was one of lateral restraint. More deformation occurred in the subgrade, however,
when the geogrid was placed in the middle of the base. The difference in deformations
in the base was greater than those in the subgrade, giving rise to the overall better behav-
ior of the section with the geogrid placed in the middle of the base. These results indi-
cate that geogrid reinforcement can aid in the reinforcement of both the base and the
subgrade and that proper placement position should be dictated by an examination of
which layer requires the most reinforcement.

4.3.9 Webster (1993)

For a base thickness of 350 mm, Webster (1993) showed greater benefits by placing
Geogrid B at the bottom of the base (TBR = 4.7) rather than at the middle of the layer
(TBR = 2.2, Figure 8b) for a 25 mm rut depth. All of the other reinforcement products
used in this study were placed at the bottom of the base. These results are contrasted

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 577


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

against those of Barksdale et al. (1989), Collin et al. (1996), and Haas et al. (1988)
where, for sections of this thickness, optimum reinforcement was seen by placing the
geosynthetic higher in the base layer and where a much smaller cyclic load was applied.

4.3.10 General Comments

The majority of the results discussed in Section 4.3 indicate that the proper location
of the geosynthetic within the base course was dependent on the magnitude of the ap-
plied traffic load and the subgrade strength. In general, it appears that as the load magni-
tude increases the optimal geosynthetic location occurs at greater depths.
Moghaddas-Nejad and Small (1996) used the lightest load (0.42 kN) and the thinnest
base layer (40 mm) and observed better performance when the geogrid was placed in the
middle of the base. The data presented by Barker (1987) and Webster (1993) represent
the other extreme, where loads of 120 and 130 kN were used, respectively. Barker (1987)
found that placing the geogrid in the middle of a 150 mm thick base resulted in poor per-
formance. Webster (1993) found that placing the geogrid at the bottom of a 150 mm
thick base resulted in a modest improvement (Figure 8c), and that for a 350 mm thick
base a better performance was observed when the geogrid was placed at the bottom of
the base. Haas et al. (1988) demonstrated that for a moderate load (40 kN) and geogrid
placement high in the base layer, small compressive strains develop in the geogrid until
an appreciable rut depth develops. This indicates that, for such a configuration, the geo-
grid provides reinforcement only through the tensioned membrane function. Therefore,
the geogrid used by Barker (1987) could not prevent lateral spread of the base course soil
because it was placed too high in the base layer, and the optimal distance between the
geogrid and the applied load increases as the load magnitude increases.

4.4 Base Course Layer Equivalency

Several studies have suggested that geosynthetic reinforcement can be used to reduce
the base course layer thickness of pavement, thereby contributing a structural compo-
nent of support to the pavement system. In this section, direct experimental findings
which demonstrate this concept are described.

4.4.1 Anderson and Killeavy (1989)

Anderson and Killeavy (1989) constructed three sections with different base course
thicknesses that were designed to perform identically. The results of the study indicate
that the sections essentially performed the same. Therefore, the geotextile-reinforced
base layer thickness could be reduced from 450 to 350 mm, while the geogrid- and geo-
textile-reinforced section thickness could be reduced from 450 to 200 mm.

4.4.2 Cancelli et al. (1996)

Cancelli et al. (1996) demonstrated that by using the most flexible geogrid of the
study (Geogrid H), the base course thickness could be reduced from approximately 425
mm to 300 mm for test sections with a subgrade CBR = 3. This results in a 30% savings

578 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

in aggregate base material (Figure 4c). Other test results suggest that an even greater
savings could have been achieved if the more stiff geogrids were used.

4.4.3 Collin et al. (1996)

Only limited conclusions can be made regarding base course layer equivalency using
the data presented by Collin et al. (1996). For test sections reinforced at the bottom of
the thin base layer (180 mm) with Geogrid A and B and a 25 mm rut depth, an equiva-
lent performance was obtained for an unreinforced section with no more than 55 mm
of additional base. A similar unreinforced section equivalent performance was ob-
tained for the test section reinforced with Geogrid B at the bottom of a 235 mm thick
base. For test sections reinforced with Geogrid A and B and a 15 mm rut depth, the base
layer thickness can be reduced from 235 to 180 mm, with failure occurring in approxi-
mately 250 load cycles.

4.4.4 Haas et al. (1988)

For a subgrade with CBR = 3.5, Haas et al. (1988) demonstrated that a 200 mm thick
base layer (Section 11, Figure 6d) was equivalent to a 100 mm thick base layer rein-
forced at the bottom with Geogrid A (Section 14, Figure 6d). The test results showed
that the two sections were nearly equivalent at all rut depths. The results of another test
series with a weak subgrade, CBR = 1, revealed that a section with a 150 mm thick geo-
grid-reinforced base layer (geogrid at the bottom of the base, Section 17, Figure 6e) did
not perform as well as an unreinforced section with a 200 mm thick base (Section 15,
Figure 6e). For this same test series, a reinforced section with a 200 mm thick base layer
(Section 16, Figure 6e) performed better than the unreinforced section (TBR = 2.2).
Therefore, the reinforced base layer thickness required for an equivalent performance
to the 200 mm thick section varied between 150 and 200 mm, suggesting that the base
course equivalency was not as thick for test sections with weaker subgrades.

4.4.5 Miura et al. (1990)

Miura et al. (1990) constructed reinforced field test sections that contained 50 mm
less base course and 50 mm less subbase than the unreinforced section. In terms of rut
depth, all of the reinforced sections performed better than the unreinforced control sec-
tion, implying that the geogrid reinforcement was equivalent to in excess of 100 mm
of combined base and subbase material. Benkelman beam tests showed that the section
containing Geogrid B located at the bottom of the subbase performed better than the
unreinforced section, implying the same equivalency as stated above. The other rein-
forced sections performed either the same or slightly worse than the unreinforced con-
trol section.

4.4.6 Webster (1993)

For a subgrade with CBR = 8, Webster (1993) showed that a section containing Geo-
grid B and a 150 mm thick base was equivalent to an unreinforced section with a 250
mm thick base, which resulted in a 40% reduction (Figure 8c). The sections were seen

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 579


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

to be equivalent for all rut depths. For a subgrade with CBR = 3, a reinforced section
with a 300 mm thick base performed better than an unreinforced section with a 350 mm
thick base, but not as well as a section with a 450 mm thick base (Figure 8d). Thus the
percentage reduction of base course material was greater for the stronger subgrades.

4.5 Section Layer Thickness and Subgrade Strength and Stiffness

4.5.1 General

In this section, the test results are analyzed to determine if the improved performance
of flexible pavements was dependent on the overall thickness of the AC and base sec-
tions, and the mechanical properties of the subgrade. The results of this analysis may
have design implications, such as whether or not this type of application was only ap-
propriate for low-volume roads incorporating relatively thin sections, whether higher
volume roads can also benefit from reinforcement, and whether the application was
only appropriate for soft subgrades.
This type of comparative analysis is difficult to perform for the studies discussed in
the current paper because of the other variables that influence improved performance,
such as geosynthetic placement and magnitude of the applied load. In an attempt to state
broad conclusions concerning section thickness, thin and thick sections are classified
as sections with a structural number less than 2 and greater than 3, respectively. The
structural numbers are listed in Table 2. Sections with structural numbers between 2 and
3 are classified as moderately thick. It may be expected that a thin section subject to
a light load would give a similar reinforcement benefit as a thicker section subject to
a greater load. In many of the studies discussed in the current paper, thick versus thin
sections were not compared, making it possible to only comment on the levels of im-
provement that were measured for the different test sections.

4.5.2 Collin et al. (1996), Haas et al. (1988), and Webster (1993)

Collin et al. (1996), Haas et al. (1988), and Webster (1993) constructed thick and thin
test sections. The test results reported by Collin et al. (1996) suggest that the influence
of section layer thickness on performance was influenced by the rut depth at which the
evaluation was made.
For a 12 mm rut depth, the thicker base layer sections tend to show higher TBR values.
As rut depth increases, TBR increases for the thicker base layer sections and reaches a
peak value at a rut depth of approximately 20 mm, where after it begins to decrease. The
same trend is seen for sections with smaller base layer thickness, only with the rut depth
at which the TBR reaches a peak being greater with decreasing base thickness. On the
average, greater TBR values are seen for sections with base course layer thickness rang-
ing from 200 to 270 mm, where TBR is as great as 10. At a 25 mm rut depth, the greatest
TBR value is seen for a base layer thickness of 255 mm. These results may have been
different had the geogrid been placed up into the base layer for the thicker sections.
Haas et al. (1988) showed no particular trend of TBR value with structural number
but did show that sections which required a larger number of cycles to failure exhibited
greater TBR values. Webster (1993) obtained greater TBR values for sections with
lower structural numbers. Haas et al. (1988) and Webster (1993) showed that greater

580 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

TBR values were obtained for sections with a greater subgrade CBR value. Both stud-
ies also demonstrated that greater base course layer reductions could be achieved from
sections with greater subgrade CBR values. These results suggest that optimal rein-
forcement is obtained for sections that are properly designed to avoid premature fail-
ure and that under designed sections fail at essentially the same rate with or without
geosynthetic reinforcement.

4.5.3 Cancelli et al. (1996)

Cancelli et al. (1996) constructed thick test sections. Traffic benefit ratios ranging
from 4 to 15 were calculated for sections with the same structural number and rein-
forced with Geogrid H. The TBR values decreased with increasing CBR values (depen-
dent on the rut depth criterion used). A nearly constant TBR value of 4.5 was obtained
for CBR values ranging from 1 to 18 and a 12.5 mm rut depth. The TBR values de-
creased from 15 to 3.5 for a 25 mm rut depth criterion and increasing CBR values. Can-
celli et al. (1996) also demonstrated that, as the subgrade strength increases (i.e. greater
CBR values), the improvement in test section performance decreases with increasing
geogrid stiffness.

4.5.4 Al-Qadi et al. (1997)

Al-Qadi et al. (1997) used the interpretation and analysis scheme described in Section
4.1 to analyze test section data. Traffic benefit ratios ranging from 2.1 to 3.5 were calcu-
lated for the subgrade with a nominal CBR = 4, while TBR = 2.9 for a section with a
nominal CBR = 2. For the geogrid-reinforced sections, TBR = 0.34 and 1.1 for the sub-
grade with CBR = 4, while TBR = 0.56 for the subgrade with CBR = 2. For the geotextile
sections, these results indicate no particular trend of improvement with subgrade CBR,
which is unexpected given that separation and filtration were the controlling functions
of the improvement observed. An improvement in performance is generally thought to
be greater for decreasing subgrade strength when the controlling functions are separa-
tion and filtration (Holtz et al. 1995).

4.6 Separation and Filtration

The geosynthetic functions of separation and filtration are distinct from geosynthet-
ic reinforcement. Whenever possible, it is important to distinguish between these
functions when evaluating the reinforcement potential of geosynthetics. It is possible
that a geogrid can improve section performance when placed in a subgrade where sep-
aration and filtration functions are not required, yet, not improve the performance of
an identical section where a separator and filter are required to prevent base course
layer contamination.

4.6.1 Cancelli et al. (1996), Collin et al. (1996), and Webster (1993)

Of the studies examined, Cancelli et al. (1996), Collin et al. (1996), and Webster
(1993) noted no problems with regard to base layer contamination in the control and

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 581


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

geogrid-reinforced sections. Each of these studies demonstrated significant improve-


ment in pavement performance.
A fine sand subgrade was used by Cancelli et al. (1996). The third author of this study
stated that the base layer became contaminated with subgrade material only for the
weakest subgrade (CBR = 1), and the contamination was limited to bottom 10 to 15 mm
of the base (Rimoldi 1997). The second author of the study performed by Collin et al.
(1996) stated that no contamination of the base layer with the silty sand subgrade was
observed in any of the sections (Kinney 1996). Webster (1993) noted that the separation
function was not a factor with regard to failure for any test sections with a highly plastic
clay subgrade. Control sections failed due to lateral displacements of the aggregate at
the subgrade interface and not because of deterioration of the structural properties of
the base layer caused by separation or filtration problems.

4.6.2 Al-Qadi et al. (1994), Barksdale et al. (1989), and Halliday and Potter (1984)

Al-Qadi et al. (1994), Barksdale et al. (1989), and Halliday and Potter (1984) reported
contamination of the base layer in the control and geogrid-reinforced test sections.
These studies report conflicting data with regard to the amount of measured improve-
ment in the geogrid- and geotextile-reinforced sections.
Al-Qadi et al. (1994) stated that base material penetrated into the subgrade and sub-
grade fines migrated into the base layer in the unreinforced and geogrid-reinforced test
sections with a silty sand subgrade. There was no mixture of the subgrade and base ma-
terials in the geotextile-reinforced sections (i.e. the geotextile acted as a separator and
filter). There was little to no improvement in the geogrid-reinforced sections and, de-
pending on which method was chosen to interpret the data, it could be shown that the
geogrid-reinforced section performed worse than the control section.
Halliday and Potter (1984) excavated the two sections when testing was complete.
In the control section, approximately 70 mm of base material had penetrated into the
subgrade. The subgrade penetration was uniform across the roadway and, as a result,
the authors concluded that mixing occurred during construction and was not caused by
traffic loading. This was also confirmed by the lack of subgrade rutting. The height of
the top of the base was pre-established, thus an additional 70 mm of base material had
to be used in the control section. Base and subgrade layer mixing was not observed in
the geotextile-reinforced section. The behavior of the two sections was identical which
implies that the geotextile acted as a separator during construction and less base soil
had to be used.
Barksdale et al. (1989) noted contamination of the base layer in all sections except
the section containing the geotextile at the bottom of the base layer. Contamination ex-
tended into the base by 25 to 38 mm. When the geotextile was placed at the bottom of
the base TBR = 1.6 (Section 8, Figure 3b), while TBR = 2.1 when the geotextile was
placed in the middle of the base (Section 2, Figure 3a). This observation contradicts the
separation and filtration functions of geotextiles. If the primary function of the geotex-
tile was separation and filtration, it would be expected that Section 8 would have a high-
er TBR value than Section 2. In this case, separation and filtration were required to
prevent base layer contamination, but the consequence of contamination was not severe
and, thus, the reinforcement effect of the geotextile could still be observed under such
conditions.

582 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

4.6.3 Brown et al. (1982)

Brown et al. (1982) reported that no fines migrated though the geotextiles. No com-
ments were made regarding the amount of fines contamination observed in the control
sections. It was observed that the base layer soil above the geotextile was not as com-
pacted as the base layer soil in the control section. This indicates that compaction was
more difficult to achieve when the section contained a geotextile. Also, slip occurred
between the base layer soil and the geotextile in several sections.

4.6.4 Anderson and Killeavy (1989)

Anderson and Killeavy (1989), using the measured FWD data, showed that the geo-
grid-reinforced section had a base layer modulus equal to 2.3 times that of the unrein-
forced section. The section with only the geotextile separator had a base layer modulus
of 1.4 times that of the unreinforced section. In addition, the deflection basins indicate
that the section with only the geotextile experienced the highest tensile strains in the bot-
tom of the AC layer and was the weakest section in light of this criteria. These findings
suggest that the geotextile separator provided stabilization during and possibly after
construction, which prevented mixing of the base and subgrade. This is more than likely
the reason for the higher modulus seen in the section containing only the geotextile.
The additional improvement in modulus for the thinner geogrid section with the same
geotextile separator can be attributed to the reinforcement effect of the geogrid. The
authors also note that observations made during construction clearly showed the benefit
of the geotextile and geogrid in preventing mixing and mobility problems during
construction. These results indicate that for soft site conditions, both the reinforcement
and separation functions are important considerations.

4.7 Other Observations

Several studies examined the effect of prestressing and/or prerutting the geosynthetic
prior to the application of traffic loads. Barksdale et al. (1989) examined the effects of
prerutting the control and reinforced sections, and geogrid pretensioning. Barksdale et
al. (1989) performed multiple- and single-track loading tests on a prerutted section re-
inforced with a geotextile at the bottom of the thin weak base layer. There was no im-
provement in performance when this section was compared to the same section without
prerutting. For the multiple-track tests, prerutting worsened section performance. Pre-
rutting the section containing a geogrid at the bottom of the thicker, stronger base re-
sulted in TBR = 4.6 for a 12.5 mm rut depth in comparison to TBR = 1 obtained for the
same section without prerutting. For the single-track tests, a similar trend was ob-
served, however, the TBR values were lower and increased from TBR = 0.7 for no pre-
rutting to TBR = 1.8 with prerutting. Also, prerutting an unreinforced section resulted
in TBR = 12.
The test results reported by Barksdale et al. (1989) suggest that geogrids facilitated
greater compaction of the base course soil if the test section was prerutted. The inter-
locking effect of the geogrid appeared to prevent soil spreading, such that improved ver-
tical compaction could be achieved. This phenomenon did not occur in
geotextile-reinforced sections because there was no interlocking effect between geo-

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 583


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

textiles and soils. It is not understood why prerutting an unreinforced section resulted
in such an improvement. It would be expected that prerutting a relatively soft subgrade
would cause separation problems and that the section performance would be compro-
mised. The base course was typically compacted to 95% of its maximum dry density,
thus, the amount of improvement in density that could be achieved by prerutting was
not significant. Geogrid prestressing resulted in large TBR values ranging from 5 to 50
for a 12.5 mm rut depth. Geogrid prerutting offered more improvement, however, in
comparison to prestressing.
Geogrid pretensioning did not offer any significant improvement in performance
(Haas et al. 1988). Anderson and Killeavy (1989) stated that geogrid pretensioning
through traffic loading of the unsurfaced section was beneficial; however, no experi-
mental data was obtained to support this argument.
Barksdale et al. (1989) showed that, in general, the resilient vertical strain was the
greatest at the top of the base and subgrade layers and was slightly less for the reinforced
sections as compared to the control sections. Prerutting and geogrid prestressing signifi-
cantly reduced the resilient strain at the top of the subgrade. Transient vertical stress
at the top of the subgrade did not appear to be influenced by the presence of a geosyn-
thetic layer. The presence of a geosynthetic decreased the development of transient lon-
gitudinal stress and lateral resilient strain as the tests progressed.
For the subgrade with CBR = 3.5, Haas et al. (1988) showed that the vertical perma-
nent stress at the top of the subgrade was reduced by approximately 22% with the inclu-
sion of reinforcement. This reduction was observed during the first load cycle and
continued until a 20 mm rut depth was reached. For rut depths greater than 20 mm, the
vertical permanent stress was only reduced by 12%. For very weak subgrade materials
(CBR = 1 and 0.5), greater stresses were measured in the reinforced section in compari-
son to the control section. As the number of load cycles increased, the stresses in the
reinforced and control sections became equal.
The laboratory tests reported by Miura et al. (1990) showed that the modulus of sub-
grade reaction of each reinforced section, measured after cyclic loading, was approxi-
mately 30% greater than the unreinforced section. A test section containing a geogrid
placed in a slightly convex configuration was also constructed, which caused a slight
to moderate improvement in rut depth with load cycle. The improvement was signifi-
cantly less than that obtained when the geogrid was placed in a slightly concave confi-
guration. Miura et al. (1990) attributed the improvement of the convex configuration
to geogrid-soil interlock, and stated that additional improvement of the concave config-
uration was due to membrane action. Thus proper geosynthetic placement in the field
was important.
In several of the studies discussed, the geogrids were instrumented with bonded re-
sistance strain gages to measure the tensile strains induced during loading. Haas et al.
(1988) measured significant strains in the geogrid below the load plate. The geogrid
strain was mobilized at small cycle numbers, thus, the geogrid reinforcement response
was essentially immediate. The maximum strain values of 1 to 1.8% were recorded at
a 20 mm rut depth. The geogrid strain was equal to zero at a radius of 1.5 times the load
plate radius and became compressive for a small distance beyond this radius; therefore,
geogrid anchorage was not responsible for the improved section performance. When
compressive strains were measured outside of the area experiencing tensile strains, lat-
eral spread was the main function causing improved performance.

584 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Haas et al. (1988) also reported an improved section performance when the geogrid
strain gradient measured away from the load center was high. This signifies that the area
beneath the load was reinforced, and beyond this region there was little need for addi-
tional reinforcement.
Miura et al. (1990) also instrumented geogrid specimens with strain gages and, simi-
lar to Haas et al. (1988), measured appreciable geogrid strains beneath the load center
that reduced to zero at a radius of 2.5 times the radius of the load plate. Miura et al.
(1990) measured smaller geogrid strains than Haas et al. (1988) because Miura et al.
(1990) used a more inflexible geogrid. The test section performances were correlated
to the development of strain in the geogrid specimens.

4.8 Evaluation of TBR as a Function of Rut Depth

The results of rut depth versus the number of load cycles from three studies were ana-
lyzed by computing the TBR value for different rut depth values. This analysis was per-
formed using the test results reported by Haas et al. (1988), Miura et al. (1990), and
Webster (1993). The purpose of this analysis is to examine the manner in which im-
provement, defined in terms of TBR values, changes during the development of perma-
nent deformation in the paved roadway.
Figure 9 is a plot of TBR values versus rut depth for four test sections (Haas et al.
1988). The reinforced and control sections used to calculate the plotted data are given
in Figure 9. The TBR values increase with increasing subgrade strength and base course
thickness. The curves representing Section 15 and 16, and Section 11 and 12 correspond
to sections where the subgrade CBR = 1 and 3.5, respectively. These curves show a gen-
eral trend of moderately increasing TBR values with increasing rut depth, where great-
er TBR values were obtained for the stronger subgrade. The higher TBR values
obtained for Section 1 and 3 (Figure 9) correspond to sections with a subgrade CBR =
8. For Section 1 and 3, the TBR reaches a peak value of approximately 5 at a 8 mm rut
depth and decreases to less than 3 at a 25 mm rut depth. The greatest TBR values were
obtained for Section 5 and 6 and correspond to a subgrade CBR = 1; however, Section
5 and 6 contained an additional 100 mm of base material and the geogrid was placed
in the middle of the base layer. For Section 5 and 6, the TBR values continue to increase
with increasing rut depth.
Figure 10 is a plot of TBR values versus rut depth for two reinforced sections from
the laboratory study by Miura et al. (1990). The TBR values in Figure 10 are greater
at lower rut depths. Four reinforced sections from the study by Webster (1993) were
used to create the four curves shown in Figure 11. In Figure 11, the TBR values for the
thicker sections moderately increase with increasing rut depth and increase substantial-
ly for a relatively thin section. The TBR values measured by Webster (1993) (Figure
11) show an opposite trend to the TBR values measured by Haas et al. (1988) (Figure
9) in that the TBR values increased with decreasing base course thickness. The load
magnitude used by Webster (1993) was greater than three times the load magnitude
used by Haas et al. (1988).
Figures 9 to 11 show that improved test section performance (defined in terms of a
TBR value), and the way in which improvement changes with permanent deformation
of the pavement surface, is a function of subgrade strength, base course thickness, and
load magnitude. Similar data were reported by Collin et al. (1996) as discussed in Sec-

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 585


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

20
Section 5 and 6
Section 1 and 3
15 Section 11 and 12
Section 15 and 16
TBR
10

0
0 5 10 15 20 25
Rut depth (mm)
Figure 9. TBR values versus rut depth from Haas et al. (1988).

20
Section 4
15 Section 6
TBR

10

0
0 1 2 3 4 5 6
Rut depth (mm)
Figure 10. TBR values versus rut depth from Miura et al. (1990).

20
350 mm base
300 mm base
15 250 mm base
150 mm base
TBR

10

0
0 5 10 15 20 25
Rut depth (mm)
Figure 11. TBR values versus rut depth from Webster (1993).

586 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

tion 4. Results from these studies show that improvement is possible for all levels of
rut depth when the section has been properly designed. Large rut depths are not a neces-
sary condition for improvement.

5 CONCLUSIONS

Results from the 12 studies reviewed show that reinforcement of the base layer of flex-
ible pavements through the function of preventing lateral spread of this material is feasi-
ble when the section has been properly designed. A “proper” design might include the
use of a separator/filter and/or the incorporation of structural layers of sufficient thick-
ness to prevent the development of excessive rut depth for a small number of load cycles.
Improvement can be defined in terms of a traffic benefit ratio, indicative of an in-
crease in the service life of the pavement structure, where common TBR values have
ranged between 3 and 10, or in terms of a base layer equivalency, where reductions of
base layer thickness have ranged between 22 and 50%. Improvement due to geosynthet-
ic reinforcement was observed in laboratory-scale experiments involving stationary
circular plates to which a cyclic load has been applied (Cancelli et al. 1996; Haas et al.
1988; Miura et al. 1990), test tracks incorporating moving wheel loads (Barksdale et
al. 1989; Collin et al. 1996; Moghaddas-Nejad and Small 1996; Webster 1993), and
full-scale roads constructed with normal construction equipment (Anderson and Killea-
vy 1989; Miura et al. 1990).
Perhaps of greatest importance is the observation that improvement, whether it be in
terms of an extension of service life or equivalent performance of a reduced-base course
thickness, is seen for all levels of rut depth and not only for conditions of excessively
large rut depth. Only for several test sections from different studies were TBR values
seen to continue to increase with increasing rut depth.
Haas et al. (1988), Collin et al. (1996), and Miura et al. (1990) reported conditions
under which TBR values reach a maximum well before a 25 mm rut depth. These same
studies also contain sections where TBR values were seen to remain essentially
constant for all levels of rut depth. Anderson and Killeavy (1989), Cancelli et al. (1996),
Collin et al. (1996), Haas et al. (1988), Miura et al. (1990), and Webster (1993) have
shown that when a reinforced section has been designed to yield equivalent perfor-
mance to an unreinforced section with an increased base thickness, equivalent perfor-
mance is seen for all levels of rut depth.
For those studies where strains were measured on geogrid reinforcement layers (Haas
et al. 1988; Miura et al. 1990), strains were observed in the geogrid immediately upon
load application and well before any appreciable rut depth developed, provided the re-
inforcement was properly placed in the base layer. Strain measurements on geogrid
specimens have indicated that these materials are engaged in a tensile capacity and that
the level to which strain develops is closely related to the amount of improvement ob-
served. The above results indicate that base course geosynthetic reinforcement is well
suited for flexible pavements that cannot tolerate significant surface deformations and
remain operational.
The range of different studies examined has provided insight into the importance of
the roles of separation, filtration, and reinforcement for flexible pavements. Al-Qadi
et al. (1994) stated that appreciable mixing of the base course and the subgrade soils

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 587


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

occurred in the control sections. Al-Qadi et al. (1994) also showed using the analysis
approach described in Section 4.1, which was designed to account for differences be-
tween the sections, that the geogrid-reinforced sections offered no improvement and in
most cases performed worse than control sections. Consistent with these findings, Al-
Qadi et al. (1994) showed that for the subgrade-base system used, geotextiles offered
superior performance as compared to geogrid sections where mixing was predominant.
Other studies which reported a more moderate amount of mixing, and indicative of con-
ditions for which separation and filtration functions were not as critical (Anderson and
Killeavy 1989; Barksdale et al. 1989) indicate that reinforcement is still possible, but
perhaps not to the extent which might have been observed had separation and filtration
functions been incorporated into the section designs. On the other extreme, studies ex-
hibiting no problems with mixing (Cancelli et al. 1996; Collin et al. 1996; Haas et al.
1988; Miura et al. 1990; Webster 1993) demonstrated significant improvement with
geogrid reinforcement layers for properly designed sections. Many of these later studies
involved subgrades with CBR values less than 3 and soil types ranging from a poorly-
graded sand (SP) to a clay containing plastic fines (CH), indicating that an evaluation
for the need for separation and filtration cannot be made solely in terms of the subgrade
CBR value. While not the focus of the current study, these results suggest that other fac-
tors, such as subgrade plasticity and grain size, load magnitude, base layer thickness
and grain size distribution, asphalt layer thickness, and construction technique are all
important factors when considering the need for a separator/filter. These results demon-
strate that separation and filtration functions cannot be overlooked when designing for
reinforcement and that improvements due to reinforcement will be fully realized only
when problems of mixing are addressed.
Several studies have used both geogrids and geotextiles, while others have focused
only on one geosynthetic type. For those studies involving both materials, Anderson
and Killeavy (1989), Barksdale et al. (1989), and Cancelli et al. (1996) demonstrated
that geogrids were superior to geotextiles when used as reinforcement, while Al-Qadi
et al. (1994) showed that superior performance was seen when a geotextile was used.
As mentioned above, this result is attributed to the severe need in these sections for a
geosynthetic acting as a separator/filter.
Significant improvement was seen in studies involving only geogrids (Collin et al.
1996; Haas et al. 1988; Miura et al. 1990; Webster 1993). Studies involving only geotex-
tiles (Brown et al. 1982; Halliday and Potter 1984) tended to show that little improve-
ment could be attributed to geotextile reinforcement, with a separation function being
the only discernible effect seen in the paper by Halliday and Potter (1984). Brown et
al. (1982) pointed to several problems that may be encountered when using nonwoven
geotextiles and a relatively thin base course layer. These problems concerned the diffi-
culty associated with achieving adequate compaction of the base layers adjacent to the
geotextiles and with possible slip along the interface during traffic loading. Barksdale
et al. (1989) alluded to similar problems during prerutting operations of geotextile sec-
tions. Problems have also been encountered with geogrids where the base course soil
was sized excessively large, thereby preventing proper interlock. The lack of an appre-
ciable reinforcement function in the studies of Al-Qadi et al. (1994) and Barker (1987)
may be due in part to this characteristic of the base layer soil used. Taken as a whole,
these results indicate that the ability of the base layer soil to interlock with the geosyn-

588 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

thetic is critical for the function of preventing lateral spread and that geogrids appear
to be a more suitable product for meeting this criterion.
Secondary to the importance of whether geogrids versus geotextiles are superior for
this application is the importance of geogrid type, stiffness, placement position, and
layering. Each of these variables is difficult to evaluate by itself since it appears to be
influenced by the other variables as well as factors such as section thickness, subgrade
strength, and load magnitude. Cancelli et al. (1996) and Webster (1993) indicate that
torsionally rigid geogrids were superior to torsionally flexible ones, such as woven geo-
grids. The results of Cancelli et al. (1996) are somewhat inconclusive on this point since
two geogrids with different torsional rigidities were seen to perform identically. For tor-
sionally rigid geogrids, an increase in axial stiffness resulted in increased improvement
(Cancelli et al. 1996; Collin et al. 1996; Miura et al. 1990; Webster 1993). The magni-
tude of improvement was seen to be dependent on the thickness of the base layer and
the subgrade strength.
Optimal placement position in the base course layer was dependent on base layer
thickness and load magnitude. For a very light load and thin sections, Moghaddas-Ne-
jad and Small (1996) showed that the middle of a 40 mm thick base offered optimal per-
formance. For a more moderate load, Barksdale et al. (1989) showed that the middle
position of a 200 mm thick base was better than the bottom. For a heavier load, Haas
et al. (1988) showed that optimal position was at the bottom of bases up to 250 mm in
thickness and in the middle of bases thicker than 250 mm. For the geogrid placed at the
bottom of the base, Collin et al. (1996) showed that optimal performance was achieved
when the base layer thickness was 255 mm. For an even heavier load, Webster (1993)
showed that the bottom of a 350 mm thick base was better than the middle, while Barker
(1987) showed that little reinforcement occurred when a geogrid was placed in the
middle of a 150 mm thick base. Cancelli et al. (1996) and Miura et al. (1990) showed
that two layers of geogrid could be as effective as one layer of a more stiff geogrid.
Similar to the considerations of geosynthetic type, stiffness, placement position, and
layering, the influence of section layer thickness and subgrade strength on improve-
ment potential is difficult to evaluate due to the interrelationship with other variables.
In general, it appears that reinforcement was not as effective for sections that have been
grossly under designed. An under designed section may be a result of an excessively
thin section on a competent base or a section of more moderate thickness on a very weak
subgrade. For sections designed to carry normal traffic, reinforcement appears to be
feasible for sections involving various section thicknesses and subgrade strengths, pro-
vided that the geosynthetic has been properly placed within the base course layer.
In summary, experimental work performed to-date indicates the promise and feasibil-
ity of using geosynthetics, mainly geogrids, to reinforce the base layer of flexible pave-
ments, where improvement appears to be through the prevention of lateral spreading
of the base layer. These experimental results taken by themselves appear to be insuffi-
cient for the development of an accepted design procedure due to the many dependent
variables impacting the problem. The authors of the current study believe that addition-
al experimental work is necessary to clearly define the mechanisms associated with this
type of geosynthetic reinforcement and that the development of analytical models is a
necessary step in reaching the goal of providing a methodology for the design of such
roadways. Literature pertaining to existing design solutions and finite element models
is discussed in a companion paper (Perkins and Ismeik 1997).

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 589


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the financial support of the Montana Department
of Transportation through Grant Number 8126. The authors are indebted to many of the
authors of the papers cited and other members of the research community who reviewed
this paper for accuracy and/or provided additional details on their work. Included in this
list are I. Al-Qadi, A. Anderson, R. Bathurst, S. Brown, J. Collin, K. Henry, M. Killeavy,
T. Kinney, G. Richardson, P. Rimoldi, S. Webster, and A. Zhao.

REFERENCES

American Association of State Highway and Transportation Officials (AASHTO),


1993, AASHTO Guide for Design of Pavement Structures, Washington, DC, USA.
Al-Qadi, I.L., Brandon, T.L., Valentine, R.J., Lacina, B.A. and Smith, T.E., 1994, “Lab-
oratory Evaluation of Geosynthetic Reinforced Pavement Sections”, Transportation
Research Record 1439, pp. 25-31.
Al-Qadi, I.L., Brandon, T.L. and Bhutta, A., 1997, “Geosynthetic Stabilized Flexible
Pavements”, Proceedings of Geosynthetics ’97, IFAI, Vol. 2, Long Beach, California,
USA, March 1997, pp. 647-662.
Anderson, P. and Killeavy, M., 1989, “Geotextiles and Geogrids: Cost Effective Alter-
nate Materials for Pavement Design and Construction”, Proceedings of Geosynthet-
ics ’89 , IFAI, Vol. 2, San Diego, California, USA, February 1989, pp. 353-360.
Barker, W.R., 1987, “Open-Graded Bases for Airfield Pavements”, Technical Report
GL-87-16, U.S. Army Corps of Engineers, Waterways Experiment Station, Vick-
sburg, Mississippi, USA, 76 p.
Barksdale, R.D., Brown, S.F. and Chan, F., 1989, “Potential Benefits of Geosynthetics
in Flexible Pavement Systems”, National Cooperative Highway Research Program
Report No. 315 , Transportation Research Board, National Research Council, Wash-
ington, DC, USA, 56 p.
Bender, D.A. and Barenberg, E.J., 1978, “Design and Behavior of Soil-Fabric-Aggre-
gate Systems”, Transportation Research Record 671, pp. 64-75.
Better Roads, 1995, “Corps Report Validity Questioned”, Vol. 54, No. 2, p. 26.
Brown, S.F., Jones, C.P.D. and Brodrick, B.V., 1982, “Use of Non-Woven Fabrics in
Permanent Road Pavements”, Proceedings of the Institution of Civil Engineers,Part
2, Vol. 73, pp. 541-563.
Brown, S.F., Jones, C.P.D. and Brodrick, B.V., 1983, “Discussion of Paper: Use of Non-
Woven Fabrics in Permanent Road Pavements”, Proceedings of the Institution of Civ-
il Engineers, Part 2, Vol. 75, pp. 343-358.
Cancelli, A., Montanelli, F., Rimoldi, P. and Zhao, A., 1996, “Full Scale Laboratory
Testing on Geosynthetics Reinforced Paved Roads”, Earth Reinforcement, Ochiai,
H., Yasufuku, N., and Omine, K., Editors, Balkema, Proceedings of the International
Symposium on Earth Reinforcement, Fukuoka, Kyushu, Japan, November 1996, pp.
573-578.

590 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Carroll, R.G., Jr., Walls, J.C. and Haas, R., 1987, “Granular Base Reinforcement of
Flexible Pavements Using Geogrids”, Proceedings of Geosynthetics ’87, IFAI, Vol.
1, New Orleans, Louisiana, USA, February 1987, pp.46-57.
Chan, F.W.K., 1990, “Permanent Deformation Resistance of Granular Layers in Pave-
ments”, Ph.D. Thesis, University of Nottingham, United Kingdom, 146 p.
Christopher, B.R. and Holtz, R.D., 1985, “Geotextile Engineering Manual”, Report No.
FHWA-TS-86/203, U.S. Department of Transportation, Federal Highway Adminis-
tration, Washington DC. USA, March 1985, 1044 p.
Collin, J.G., Kinney, T.C. and Fu, X., 1996, “Full Scale Highway Load Test of Flexible
Pavement Systems With Geogrid Reinforced Base Courses”, Geosynthetics Inten-
tional, Vol. 3, No. 4, pp. 537-549.
Douglas, R.A., 1993, “Stiffnesses of Geosynthetic Built Unpaved Road Structures: Ex-
perimental Programme, Analysis and Results,” Proceedings of Geosynthetics ’93,
IFAI, Vol. 1, Vancouver, British Columbia, Canada, pp. 21-34.
Giroud, J.P. and Noiray, L. 1981, “Geotextile Reinforced Unpaved Road Design” Jour-
nal of the Geotechnical Engineering Division, Vol. 107, No. GT9, pp. 1233-1254.
Haas R., Walls, J. and Carroll, R.G., 1988, “Geogrid Reinforcement of Granular Bases
in Flexible Pavements,” Transportation Research Record 1188, pp. 19 - 27.
Halliday, A.R. and Potter, J.F., 1984, “The Performance of a Flexible Pavement
Constructed on a Strong Fabric”, Transport and Road Research Laboratory, Report
1123, Crowthorne, Berkshire, United Kingdom, 15 p.
Holtz, R.D., Christopher, B.R. and Berg, R.R., 1995, “Geosynthetic Design and
Construction Guidelines”, Report No. FHWA-A-HI-95, U.S. Department of Trans-
portation, Federal Highway Administration, National Highway Institute Course No.
13213, Washington DC, USA, 396 p.
Houlsby, G.T. and Jewell, R.A., 1990, “Design of Reinforced Unpaved Roads for Small
Rut Depths” Proceedings of the Fourth International Conference on Geotextiles,
Geomembranes and Related Products, Balkema, Vol. 1, The Hague, Netherlands,
May 1990, pp. 171-176.
Huang, Y.H., 1993, “Pavement Analysis and Design”, Prentice Hall, Englewood Cliffs,
New Jersey, USA, 805 p.
Kennepohl, G., Kamel, N., Walls, J. and Haas, R., 1985, “Geogrid Reinforcement of
Flexible Pavements: Design Basis and Field Trials”, Proceedings of the Association
of Asphalt Paving Technologists, San Antonio, Texas, USA, February 1985, Vol. 54,
pp. 45-70.
Killeavy, M., 1997, Personal communication, February 1996.
Kinney, T.C., 1996, Personal communication, October 1996.
Miura, N., Sakai, A., Taesiri, Y., Yamanouchi, T. and Yasuhara, K., 1990, “Polymer
Grid Reinforced Pavement on Soft Clay Grounds”, Geotextiles and Geomembranes,
Vol. 9, No. 1, pp. 99-123.
Moghaddas-Nejad, F. and Small, J.C., 1996, “Effect of Geogrid Reinforcement in Mod-
el Track Tests on Pavements”, Journal of Transportation Engineering, Vol. 122, No.
6, pp. 468-474.

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 591


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Montanelli, F., Zhao, A. and Rimoldi, P., 1997, “Geosynthetic-Reinforced Pavement


System: Testing and Design”, Proceedings of Geosynthetics ’97, IFAI, Vol. 2, Long
Beach, California, USA, March 1997, pp. 619-632.
Penner, R., 1985, “Geogrid Reinforcement of the Granular Base Layer in Conventional
Three-Layer Pavement Sections”, M.S. Thesis, University of Waterloo, Kitchener,
Waterloo, Ontario, Canada, 130 p.
Penner, R., Haas, R., Walls, J. and Kennepohl, G., 1985, “Geogrid Reinforcement of
Granular Bases”, Paper Presented at the Roads and Transportation Association of
Canada Annual Conference, Vancouver, British Columbia, Canada, September 1995.
Perkins, S.W. and Ismeik, M., 1997, “A Synthesis and Evaluation of Geosynthetic Rein-
forced Base Layers in Flexible Pavements: Part II”, Geosynthetics International, Vol.
4, No. 6, pp. 605-621.
Rimoldi, P., 1997, Personal communication, March 1997.
Ruddock, E.C., Potter, J.F. and McAvoy, A.R., 1982, “A Full-Scale Experiment on
Granular and Bituminous Road Pavements Laid on Fabrics”, Proceedings of the Sec-
ond International Conference on Geotextiles, IFAI, Vol. 2, Las Vegas, Nevada, USA,
August 1982, pp. 365-370.
Smith, T.E., Brandon, T.L., Al-Qadi, I.L., Lacina, B.A., Bhutta, S.A. and Hoffman,
S.E., 1995, “Laboratory Behavior of Geogrid and Geotextile Reinforced Flexible
Pavements”, Final Report Submitted to Atlantic Construction Fabrics, Inc., Amoco
Fibers and Fabrics Company, and the Virginia Center for Innovative Technology,
February 1995, Virginia Polytechnic Institute and State University, Department of
Civil Engineering, Blacksburg, Virginia, USA, 94 p.
Steward, J.E., Williamson, R. and Mohney, J., 1977, “Guidelines for use of Fabrics in
Construction and Maintenance of Low-Volume Roads”, USDA, Forest Service report
PB-276 972, Portland, Oregon, 172 p.
Webster, S.L., 1993, “Geogrid Reinforced Base Courses For Flexible Pavements For
Light Aircraft, Test Section Construction, Behavior Under Traffic, Laboratory Tests,
and Design Criteria”, Technical Report GL-93-6, U.S. Army Corps of Engineers,
Waterways Experiment Station, Vicksburg, Mississippi, USA, 86 p.
White, D.W., Jr., 1991, “Literature Review on Geotextiles to Improve Pavements for
General Aviation Airports”, Technical Report GL-91-3, U.S. Army Corps of Engi-
neers, Waterways Experiment Station, Vicksburg, Mississippi, USA, 58 p.

592 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


Table 1. Test section location and load type.

Type of Facility Test section Applied cyclic Applied cyclic Load frequency
Author Load type
facility dimensions* (m) length (m) pressure (kPa) load (kN) or wheel speed
Stationary circular plate,
Al--Qadi et al. (1994) Laboratory tank 3.1 × 1.8 × 2.1 NA 550 39 0.5 Hz
300 mm diameter
Anderson and Killeavy Field truck Outdoor staging 65 - 70
NA Loaded truck traffic Random 65
(1989) staging area area vehicles/week
Barker (1987) Outdoor test track 21 × 4.6 × 1.1 4.6 Moving single wheel 1826 120 NR
Barksdale et al. (1989) Indoor test track 4.9 × 2.4 × 1.5 1.6 Moving single wheel 460 ~ 500 6.6 1.3 m/s
Brown et al. (1982) Indoor test track 4.9 × 2.4 × 1.5 NR Moving single wheel 530 5 -- 11 1.3 m/s
Stationary circular plate,
Cancelli et al. (1996) Laboratory tank 0.9 × 0.9 × 0.9 NA 570 40 5 or 10 Hz
300 mm diameter
Collin et al. (1996) Indoor test track 14.6 × 2.4 × 1.2 3.4 Moving single wheel 550 20 1.2 m/s

GEOSYNTHETICS INTERNATIONAL
Halliday and Potter Two-axle,
Outdoor test track 20 × 4.25 × 1.5 10.0 760 49 -- 68 1.4 - 2.2 m/s
(1984) dual wheel truck
Stationary circular plate,
Haas et al. (1988) Laboratory tank 4.5 × 1.8 × 0.9 NA 550 40 8 Hz
300 mm diameter
Miura et al. (1990): Stationary circular plate,
Laboratory tank 1.5 × 1.5 × 1.0 NA 200 6.3 0.18 Hz
laboratory 200 mm diameter

S 1997, VOL. 4, NO. 6


Miura et al. (1990):
Public roadway 300 m road 50.0 Random traffic Random Random Random
field
Moghaddas-Nejad and
Indoor test track 1.4 × 0.5 × 0.8 1.4 Moving single wheel 210 0.42 0.74 m/s
Small (1996)
Webster (1993) Outdoor test track 44 × 3.8 × 1.0 11.0 Moving single wheel 470 130 NR

Notes:

593
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

* Length × Width × Depth; NA = not applicable; NR = not reported.


Table 2. Test section layers and properties.

594
Layer thickness (mm) Layer material types Structural
Author
AC Base Subbase Base Subbase Subgrade type [CBR] number
Al-Qadi et al. (1994) 70 150, 200 None Crushed granite (GW-GM, A-1-a) None Silty sand (SM, A-4) [4, 2] 2, 2.3
Anderson and Soft silt and clay
105 200 None Crushed limestone None 2.8
Killeavy (1989) with organic pockets
Cement treated
Barker (1987) 75 150 150 Crushed limestone (GP, A-1-a) Sandy silt [27] NA
sandy gravel
Rounded gravel and sand
Barksdale et al. Silty clay (CL, A6)
25, 38 150, 200 None (GP, A-1-b) and crushed dolomitic None 1.3, 1.8
(1989) [2.6 - 3.2]
limestone (GP-GM, A-1-a)
37 - 53 107 - 175 Silty clay (CL, A6)
Brown et al. (1982) None Crushed limestone (GW, A-1-a) None 1.7
(50 nominal) (150 nominal) [2 - 8]
Rounded sand (SP, A-3)
Cancelli et al. (1996) 75 300 None Crushed rock (GW, A-1-a) None 3.0
[1, 3, 8, 18]
Collin et al. (1996) 50 150 - 460 tapered None Crushed rock (GW, A-1-a) None Silty sand (SM, A-2) [1.9] 1.7 - 3.5

GEOSYNTHETICS INTERNATIONAL
Halliday and Potter London clay (CH, A-7-6)
160 300 None Crushed granite None 4.3
(1984) [0.7 - 4.3]
Fine sand (SP, A-3) 1.7, 1.8,
100, 150,
Haas et al. (1988) 75, 100 None Crushed stone (GW, A-1-b) None [8, 3.5], and fine sand 2.1, 2.4, 2.7,
200, 250, 300
mixed with peat [1, 0.5] 2.8, 3.0
Miura et al. (1990): Sensitive clay slurried and
50 150 200 NR NR 2.5
laboratory consolidated by 5 or 10 kPa

S 1997, VOL. 4, NO. 6


600 - 800 mm compacted
Miura et al. (1990): 200,
50 150, 200 NR NR mine tailings [4 - 6] over 2.5, 3.0
field 250
soft Ariake clay
Moghaddas-Nejad
20 40 None Crushed basalt (SP, A-1-a) None Silica sand (SP, A-3) 0.55
and Small (1996)
150, 250, Crushed limestone 1.7, 2.3, 2.6,
Webster (1993) 50 None None Clay (CH, A-7-6) [3, 8]
300, 350, 450 (SM-SC, A-1-a) 2.9, 3.4
Notes: NA = not applicable; NR = not reported.
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Table 3. Physical properties of the geosynthetic reinforcement products tested.

Mass/
Geosynthetic Polymer Aperture size 5% secant modulus
Structure Unit area
type composition MD/XMD (mm) MD/XMD (kN/m) 1
(g/m2)
Geogrid A Punched sheet drawn biaxial Polypropylene 203 25/36 180/260
Geogrid B Punched sheet drawn biaxial Polypropylene 306 25/33 220/400
Geogrid C Punched sheet drawn biaxial Polypropylene 247 46/64 220/340
Polyester/PVC
Geogrid D Woven 235 20/20 248/167
coating
Geogrid E Woven Polyester 270 30/33 227/124
Geogrid F Biaxial Polypropylene 230 32/40 160/216
Polyester/
Geogrid G Woven 193 18/19 218/161
PVC coating
Geogrid H Multilayer, biaxial, continuous extrusion, and orientation Polypropylene 220 20/25 160/240
Geogrid I Biaxial, continuous extrusion, and orientation Polypropylene 250 30/40 180/260

GEOSYNTHETICS INTERNATIONAL
Geogrid J Biaxial, continuous extrusion, and orientation Polypropylene 350 30/40 200/370
Geotextile A Woven Polypropylene 120 NA NR
Geotextile B Woven Polypropylene 190 NA 228/420
Geotextile D Woven Polypropylene 970 NA NR/750
Geotextile E Woven, multifilament Polyester NR NA NR

S 1997, VOL. 4, NO. 6


Geotextile F Woven Polypropylene NR NA NR
Geotextile G Woven Polypropylene NR NA NR
Geotextile H Nonwoven, melt-bonded substrate reinforced with polyester yarn Polyester NR NA NR
Polypropylene
Geotextile I Nonwoven, melt-bonded NR NA NR
and polyethylene

Notes: 1 Values not reported by the manufacturer; NA = not applicable; NR = not reported; MD = machine direction; XMD = cross-machine direction.

595
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Table 4. The type of geosynthetics used in each study.

596
Author Geosynthetic Location with respect to the base course layer
Al-Qadi et al. (1994) Geogrid B Bottom
Geotextile A Bottom
Geotextile B Bottom
Anderson and Killeavy (1989) Geogrid A Bottom
Barker (1987) Geogrid B Middle
Barksdale et al. (1989) Geogrid A Middle, bottom
Geotextile D Middle, bottom
Brown et al. (1982) Geotextile H Bottom
Geotextile I Bottom, bottom and middle
Cancelli et al. (1996) Geogrid A, D, I, J Bottom
Geogrid H Bottom, bottom and middle
Geotextile F Bottom
Collin et al. (1996) Geogrid A Bottom
Geogrid B Bottom

GEOSYNTHETICS INTERNATIONAL
Halliday and Potter (1984) Geotextile E Bottom
Haas et al. (1988) Geogrid A Top, middle, bottom
Miura et al. (1990): Geogrid A Bottom and bottom of subbase, bottom and middle of base
laboratory Geogrid B Bottom, bottom of subbase
Geogrid C Bottom of subbase
Miura et al. (1990): Geogrid B Bottom, bottom of subbase

S 1997, VOL. 4, NO. 6


field Geogrid C Bottom, bottom of subbase
Moghaddas-Nejad and Small (1996) Geogrid B Bottom, middle
Webster (1993) Geogrid A Bottom
Geogrid B Middle, bottom
Geogrid D Bottom
Geogrid E Bottom
Geogrid F Bottom
Geogrid G Bottom
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Table 5. Instrumentation and/or measurement parameter for test sections.

Author Surface of test section Asphalt concrete Base and subbase layers Geosynthetic Subgrade Other

Rut depth and surface


Al-Qadi et al. (1994) None None None Vertical stress None
deflections

Anderson and
Rut depth Crack percentage None None None FWD
Killeavy (1989)

Barker (1987) Rut depth, profilometer None None None None FWD

Barksdale et al. Longitudinal Vertical and longitudinal


Rut depth, profilometer Transverse strain Vertical stress and strain FWD
(1989) strain strain, longitudinal stress

Brown et al. Longitudinal Vertical and longitudinal Longitudinal and Vertical and longitudinal
Rut depth, profilometer None
(1982) strain stress and strain transverse strain stress and strain

Rut depth and surface


Cancelli et al. (1996) None None None None None
deflections

GEOSYNTHETICS INTERNATIONAL
Collin et al.(1996) Rut depth None None None None None

Halliday and Potter Permanent and Longitudinal and transverse Longitudinal and Longitudinal and Vertical and transverse
None
(1984) transient rut depth strain, temperature transverse strain transverse strain strain, vertical stress

Rut depth and surface


Haas et al. (1988) None None Strain Vertical stress None
deflections

Miura et al. (1990): Rut depth and surface Vertical displacement at Vertical displacement at
None Strain None

S 1997, VOL. 4, NO. 6


laboratory displacements bottom of layer top of layer

Miura et al. (1990): Rut depth, Benkelman Longitudinal and


Crack percentage None Vertical stress None
field beam, profilometer transverse strain

Moghaddas-Nejad Vertical displacement Vertical displacement at Vertical displacement at


Rut depth None None
and Small (1996) at bottom of layer top and bottom of layer top of layer

Webster (1993) Rut depth None Vertical displacement None Vertical displacement FWD

597
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Table 6. Test section data for Al-Qadi et al. (1994).

Section Geosynthetic AC thickness (mm) Base thickness (mm) Subgrade CBR Average load (kPa)

6 Unreinforced 74 147 4.4 40.7

10 Unreinforced 71 146 4.3 37.5

3 Geotextile A 71 135 4.5 38.0

11 Geotextile A 74 150 4.4 39.8

4 Geogrid B 79 135 5.7 35.4

12 Geogrid B 79 142 4.6 39.0

15 Unreinforced 74 132 2.0 30.9

16 Geotextile A 94 122 2.2 32.0

17 Geotextile A 74 132 2.0 31.8

18 Geogrid B 74 132 2.0 27.9

Table 7. Test section data for Barksdale et al. (1989).

Section Geosynthetic Geosynthetic position AC thickness (mm) Base thickness (mm) Subgrade CBR

1 Unreinforced Unreinforced 30 211 2.7

2 Geotextile D Middle of base 33 196 2.7

3 Geotextile D Middle of base 38 211 3.2

4 Geogrid A Middle of base 34 216 3.2

5 Unreinforced Unreinforced 30 211 2.5

6 Geogrid A Bottom of base 28 206 2.5

7 Unreinforced Unreinforced 30 211 2.7

8 Geotextile D Bottom of base 30 206 2.7

598 GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6


PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

Table 8. Test section data for Haas et al. (1988).

Section Geosynthetic position AC thickness (mm) Base thickness (mm) Subgrade CBR

1 Unreinforced 100 200 8

2 Bottom of base 100 200 8

3 Middle of base 100 200 8

4 Top of base 100 200 8

5 Unreinforced 75 300 1

6 Middle of base 75 300 1

7 Unreinforced 75 300 0.5

8 Bottom of base 75 300 0.5

9 In subgrade 75 300 0.5

10 Middle and bottom of base 75 300 0.5

11 Unreinforced 75 200 3.5

12 Bottom of base 75 200 3.5

13 Bottom of base 75 150 3.5

14 Bottom of base 75 100 3.5

15 Unreinforced 75 200 1

16 Bottom of base 75 200 1

17 Bottom of base 75 150 1

Table 9. Test section data for Miura et al. (1990).

Section Geosynthetic Geosynthetic position Subgrade consolidation pressure (kPa)

1 Unreinforced Unreinforced 10

2 Geogrid B Bottom of subbase 10

3 Geogrid C Bottom of subbase 10

4 Geogrid B Bottom of base 10

5 Unreinforced Unreinforced 5

6 Geogrid A Bottom of base and subbase 5

7 Geogrid A Bottom and middle of base 5

GEOSYNTHETICS INTERNATIONAL S 1997, VOL. 4, NO. 6 599


Table 10. Summary of the major findings of each study.

600
type
layer
Other

criteria
filtration

Author
layering
thickness
suggested

Base course

location and
equivalency
and stiffness
observations

Performance
Section layer

Geosynthetic
Geosynthetic
Separation and

characteristics
Design method

Subgrade strength
Geotextile provides
Range of
more improvement
improvement of Mixing of
over geogrid; Geogrid
geotextile sections subgrade and
provided little Large base
approximately same base material
improvement; particles may
for two subgrades observed in
Rut depth Increased geotextile NA NA NA prevent proper No
used; Lack of unreinforced and
stiffness does not interlocking
improvement in geogrid-
improve performance; with geogrid.
geogrid sections also reinforced
Separation most

Al-Qadi et al. (1994)


no different between sections
critical function for
two subgrades used
subgrade used
Use of only
geotextile

GEOSYNTHETICS INTERNATIONAL
showed
moderate
Geotextile allowed Geotextile allowed Authors state
NA, however improvement,
100 mm reduction of 22% reduction of geogrid
reinforcement NA, however while use of
base material, while base material, NA, however prestressing
Rut depth, effective when reinforcement geogrid and
combination of while combination reinforcement though traffic
cracking, placed at bottom of possible for geotextile No
geotextile and geogrid of geotextile and effective for soft loading of
FWD 200 mm base and relatively thick showed greater
allowed 250 mm geogrid allowed subgrade unsurfaced
when placed directly sections improvement,
reduction of base 56% reduction of section as
on geotextile thus, both
material base material beneficial
separation and

S 1997, VOL. 4, NO. 6


reinforcement

Anderson and Killeavy (1989)


functions
important

For load magnitude


No improvement
used, locating
Rut depth, observed for single
geogrid in middle of NA NA NA NA None No
FWD geogrid used for rut
150 mm base was
depth < 18 mm
ineffective

Barker (1987)
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Prerutting
improved
Fines
performance of
contamination
For relatively low unreinforced
Rut depth, extended 25 - 38
load magnitude and geogrid
longitudinal Slight evidence to mm into base
Geogrid performed applied, middle of sections but not
strain at indicate that more layer; TBR
better than geotextile 200 mm base layer geotextile
bottom of improvement is values higher,
with stiffness 2.5 times provided better sections;
asphalt NA possible with NA however, for No
that of geogrid; TBR performance; Prerutting more
concrete, thinner sections geotextile
values as high as 17 for Differing results beneficial than
vertical stress and poor quality sections with
geogrid seen between two prestressing,
at top of base material geotextile in

Barksdale et al. (1989)


types of traffic loads thus soil
subgrade middle as
applied densification
opposed to
was major
bottom of base
effect taking
place

Base layer soil


more loose
above
Rut depth, Two geotextile geotextile; Base
No improvement No fines
transient layers performed layer soil
observed for two NA NA NA migrated through No
stress and worse than control thought to have
geotextiles used geotextile
strain section slipped relative

GEOSYNTHETICS INTERNATIONAL
to geotextile in

Brown et al. (1982)


several test
sections

For moderate rut


depth, TBR values
Performance increased Two layers of less nearly constant for
with increasing stiff geogrid For less stiff different subgrade
geogrid stiffness; A performed geogrid, base layer strengths; For larger Very little base

S 1997, VOL. 4, NO. 6


flexible woven substantially better could be reduced rut depths, TBR course
Rut depth geogrid performed than one layer of 30%; Results NA increases with contamination None Yes
worse than stiff same material and suggested greater decreasing subgrade occurred in
geotextile, which better than single savings for stiffer strength; Less sections
performed worse than layer of stiffer geogrids improvement due to

Cancelli et al. (1996)


other geogrids used material increase in geogrid
stiffness as subgrade
strength increased

601
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Table 10. Continued

602
Influence
dependent on rut
depth at which
Optimum improvement
Significant TRB Limited results
reinforcement when evaluated; Base course
increase for stiffer indicate possiible
geogrid within 200 - Greatest contamination
Rut depth geogrid and only for reduction of NA None No
270 mm below top improvement for not observed in
base thickness approximately
of base, with optimal 200 - 270 mm any section
between 175 - 275 mm 25%
depth of 255 mm base layer

Collin et al. (1996)


thickness, with
optimal thickness
of 255 mm
Transient and Geotextile
permanent rut prevented
depth, mixing of base
No
transient and Single geotextile Geotextile layer with soft
improvement in
permanent ineffective as NA NA ineffective for all NA subgrade during No
performance
stress and reinforcement member AC thicknesses construction; No
noted
strain in the structural
soil, AC, and improvement
geotextile noted

Halliday and Potter (1984)

GEOSYNTHETICS INTERNATIONAL
Reinforcement For CBR = 3.5
should be at bottom subgrade, 50%
Measured
of thin base layers base layer
geogrid strains
and at midpoint of thickness reduction Reinforcement
and
bases 250 mm possible; benefits in terms of
demonstrated
thick; No benefits Reinforced TBR values, base
Single geogrid ‘lateral spread’
expected when reduced base course equivalency,
produced TBR values function;
Rut depth single layer of section performed NA and vertical stress at NA Yes
as great as 3.3 for 20 Reinforcement

S 1997, VOL. 4, NO. 6


reinforcement same as control for top of subgrade
mm rut depth reduced vertical
placed within all levels of rut appeared to decrease
stress on

Haas et al. (1988)


compression zone depth; For weaker as subgrade strength
subgrade;
(i.e. near top of base subgrades, decreased
Pretensioning
layer or high in percentage
not effective
section over very reduction did not
soft subgrades) appear as great
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Significant TRB
increase with proper
placement location,
while location not
unique but
dependent on Modulus of
Rut depth, Moderate TRB
geosynthetic type subgrade
modulus of increase with moderate
and pavement layer NA NA NA NA reaction 30% No
subgrade increase in geogrid
mechanical greater for each
reaction stiffness
properties; Layering pavement layer
can be beneficial but
proper geosynthetic
type selection and

Miura et al. (1990): laboratory


location more
important
In terms of rut
With stiffer geogrid, depth and
no increase in rut depth Benkelman beam
Rut depth, performance, and displacements,
Results reveal either
Benkelman moderate increase in geogrid No information No information No information
little difference or None No
beam, crack performance as reinforcement available available available
ambiguous
percentage defined by Benkelman equivalent to > 50
Beam and Crack mm base course

GEOSYNTHETICS INTERNATIONAL
Percentage and > 50 mm

Miura et al. (1990): field


subbase

Observation of
deflection
bowls
For very light load
suggested that
Rut depth, and thin pavement
No information No information No information No information greater load
deflection One geogrid used layers, Geogrid B No
available available available available

S 1997, VOL. 4, NO. 6


distribution on
bowl placement in middle
the subgrade
of base was optimal
was achieved
with
reinforcement.

Moghaddas-Nejad and Small (1996)

603
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I
Table 10. Continued

604
Stiff geogrids perform
better than flexible
geogrids; Performance
related to torsional 40% base layer
rigidity of reduction possible,
FWD
geosynthetic, with For base thickness where performance Greater TBR No problems
Greater TBR values measurements
more torsionally stiff  350 mm, optimal equivalent at all rut values obtained with separation
Rut depth obtained for stronger do not detect Yes
geogrids offering more location was bottom depths; Percentage for thinner base or filtration
subgrades influence of
improvement; For of the base reduction appeared course layers observed
geogrid
given stiff geogrid, greater for stronger

Webster (1993)
increase in axial subgrades
stiffness resulted in
additional
improvement

Table 11. Traffic bearing ratios (TBR) values for a test section with a 150 mm thick base and a subgrade CBR = 4 for a 25 mm rut depth

GEOSYNTHETICS INTERNATIONAL
(Smith et al. 1995).

Geosynthetic TBR values


and section Control Section 6 Control Section 10
number Raw Corrected Raw Corrected
Geogrid B
3.2 2.1 0.7 0.9

S 1997, VOL. 4, NO. 6


Section 4
Geogrid B
4.8 3.9 1.1 1.7
Section 12
Geotextile A
5.9 4.1 1.3 1.8
Section 3
Geotextile A
11.6 11.6 2.6 5.0
Section 11
PERKINS AND ISMEIK D Geosynthetic-Reinforced Bases in Flexible Pavements: Part I

You might also like