This article was downloaded by: [University of Tennessee, Knoxville]

On: 24 December 2014, At: 15:05

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Pavement Engineering

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gpav20

Evaluation of geogrids for stabilising weak pavement

subgrade
a a a
Xiaochao Tang , Ghassan R. Chehab & Angelica Palomino
a
Department of Civil and Environmental Engineering , The Pennsylvania State University ,
University Park, PA, USA
Published online: 29 Oct 2008.

To cite this article: Xiaochao Tang , Ghassan R. Chehab & Angelica Palomino (2008) Evaluation of geogrids for stabilising
weak pavement subgrade, International Journal of Pavement Engineering, 9:6, 413-429, DOI: 10.1080/10298430802279827

To link to this article: http://dx.doi.org/10.1080/10298430802279827

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 International Journal of Pavement Engineering
Vol. 9, No. 6, December 2008, 413–429

Evaluation of geogrids for stabilising weak pavement subgrade

Xiaochao Tang1, Ghassan R. Chehab* and Angelica Palomino2
Department of Civil and Environmental Engineering, The Pennsylvania State University, University Park, PA, USA
( Received 24 February 2008; final version received 7 June 2008 )

This study attempts to identify mechanical and physical properties of geogrids that are critical to their effectiveness in the
stabilisation of pavement subgrade. Geogrid properties, including aperture size, wide-width tensile strength and junction
strength, for four geogrid products are correlated with bench-scale interface test results, including direct shear and pull-out,
and accelerated pavement testing (APT) results. APT is conducted through the use of a one-third scale APT device, the
model mobile load simulator (MMLS3) on geogrid-reinforced pavement sections. The pavement sections are constructed on
a subgrade soil with a low California bearing ratio. The performance of each pavement section is evaluated by measuring
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

surface rutting at various trafficking stages. The analysis reveals a strong relationship between performance and junction and
tensile strength of geogrids at small strains, whereas aperture size has a positive correlation with pull-out test results.
Keywords: geogrid; subgrade stabilisation; pavement; bench-scale testing; accelerated testing, MMLS3

1. Introduction Results from direct shear and pull-out tests are the most
Geogrids have been widely used as reinforcement in commonly used parameters depicting the soil–geogrid
structures with unbound materials, such as pavements, interaction characteristics. Characteristics identified through
slopes, retaining walls and embankments. It has been shown those tests are a function of the geogrid material mechanical
that, in a pavement system, the inclusion of geogrids at the properties, shape and texture, as well as soil and aggregate
interface between a pavement base course and subgrade can properties, such as gradation, plasticity, density and
significantly improve the performance of the pavement on a moisture content, applied normal stress and loading rate
weak subgrade according to both laboratory tests and full- (Ingold 1983, Jewell et al. 1984, Farrag et al. 1993).
scale field experiments (Barksdale et al. 1989, Al-Qadi et al. Identification of geogrid properties critical for
1994, 2007, Perkins 1999, Hufenus et al. 2006). Given a optimising the effectiveness of a specific reinforcement
proper ratio of geogrid aperture size to aggregate grain size, mechanism can be achieved by correlating the known
geogrids provide lateral confinement to the pavement geogrid properties, e.g. junction strength, with the
aggregate base course (ABC) through shear resistance and response of geogrids in bench-scale interface tests and
friction between the geogrid and surrounding aggregate. model-scale performance tests. Accelerated pavement
testing (APT) has been used successfully for evaluating
The confinement due to the geogrid increases the modulus of
geogrid performance for pavement applications in the
the aggregate, which leads to an improved vertical stress
USA since 1909 (Metcalf 1996). The advantages of APT
distribution over the subgrade and consequent reduction in
are the ability to conduct performance tests at relatively
vertical subgrade deformation (Love 1984, Hass et al. 1988).
low costs over a short time period, and the ability to
Reduction in deformation also results from reduction in shear
control the loading and environmental conditions.
stress transfer at the geogrid–subgrade interface (Perkins This study focuses on correlating the index properties
1999). Additional types and mechanisms of reinforcement – of geogrids with the results of bench-scale testing and
surface friction along the geogrid, passive thrust against accelerated testing. Testing includes standard index
the geogrid’s bearing ribs, aggregate interlocking between property tests, interface tests, including direct shear and
the apertures, and/or soil–soil friction – are mobilised by the pull-out tests and APT. An in-depth analysis is then
presence of geogrids depending on the application and type conducted to correlate the values of index properties with
of environmental and loading conditions (Shukla 2002). the results of bench-scale testing and performance-based
The effectiveness of geogrid reinforcement is highly APT. Through the correlation analysis, the geogrid index
dependent on physical and mechanical properties, i.e. index properties that most influence the performance of geogrids
properties, of the geogrid and on the properties of the as a means for flexible pavement subgrade stabilisation are
interface between the geogrid and the surrounding materials. identified.

*Corresponding author. Email: [email protected]

ISSN 1029-8436 print/ISSN 1477-268X online
q 2008 Taylor & Francis
DOI: 10.1080/10298430802279827
http://www.informaworld.com
 414 X. Tang et al.

2. Experimental design
2.1 Materials
2.1.1 Geogrids
Four commonly used biaxial geogrid products are selected
for this study and are herein designated as Grids A – D.
Grids A and D are composed of high tenacity polyester
multifilament yarns and coated with a proprietary polymer
and polyvinyl chloride coating, respectively. Grid B is
made of woven polypropylene (PP) yarns, while Grid C is
made of extruded PP sheets. Based on the measured
flexural rigidity as per ASTM D1388, Grids A, B and D are
classified as flexible geogrids, while Grid C is classified as
a stiff geogrid (Koerner 1998).
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

2.1.2 Subgrade soil

Figure 2. CBR variation with water content for the subgrade soil.
The subgrade material is a silty sand. Table 1 provides a
summary of the subgrade soil properties. The soil grain
size distribution is shown in Figure 1. The soil is classified
with ASTM D2487 and A-2-4(0) in accordance with
as SW-SM, which is a well-graded silty sand in accordance
AASHTO M145.
A set of laboratory unsoaked California bearing ratio
Table 1. Subgrade soil properties. (CBR) tests (ASTM D1188) are performed on the soil at
different water contents, as shown in Figure 2. CBR
Property Measurement methods Result decreases significantly with an increase in water content
Classification AASHTO M145/ A-2-4(0)/SW-SM beyond the optimum water content (10%), indicating the
ASTM D2487 soil is water sensitive. Hence, the soil is compacted at a
Per cent passing ASTM D422 6.2
water content greater than optimum to induce weak soil
#200 sieve (%)
Liquid limit (%) ASTM D4318 17.6 conditions.
Plastic limit (%) ASTM D4318 14.7
Optimum moisture ASTM D698 10.0
content (%) 2.1.3 Base course aggregate
Maximum dry ASTM D698 2066 Dense-graded crushed stone is used for the pavement ABC
density (kg/m3) layer. A standard Proctor test for the aggregates yields an
Friction angle, ASTM D3080 31.8
f (deg) optimum moisture content of 3.9% and maximum dry
density of 2329 kg/m3.

2.1.4 Hot mix asphalt

The asphalt mixture is provided by a local hot mix asphalt
plant. The asphalt mixture has a theoretical maximum
specific gravity of 2.510. No laboratory tests are conducted
for asphalt mixture properties since the same asphalt
concrete (AC) layer is used for all sections and thus is not
considered a variable in this study. Air voids (AVs) of the
constructed asphalt layer are measured as they are
pertinent to overall pavement deformation as discussed
in subsequent sections.

3. Experimental tests
3.1 Index tests
Index tests are performed on the four biaxial geogrid
Figure 1. Grain size distribution of subgrade soil and base products to determine their physical and mechanical
course aggregates. properties. Testing was conducted following ASTM
 International Journal of Pavement Engineering 415

Table 2. Tested index properties of the geogrids (Chehab et al. 2007).

Geogrid
Grid A Grid B Grid C Grid D
Index property Test method MD* TD** MD TD MD TD MD TD
Aperture size (mm) Calipers 27.18 28.96 35.05 41.15 25.65 36.58 25.65 26.42
Rib thickness (mm) Calipers 0.76 1.12 1.98 1.09 0.76 1.07 1.42 2.03
Junction thickness (mm) ASTM D5199 1.17 2.29 3.94 1.55
Mass per unit area (g/m2) ASTM D5261 298.37 252.26 319.06 350.93
Tensile strength at 2% strain (kN/m) 7.5 10.1 14.8 15.0 9.8 15.6 10.3 11.2
Tensile strength at 5% strain (kN/m) ASTM D6637 13.1 14.1 30.1 30.0 16.8 29.2 18.1 17.4
Ultimate tensile strength (kN/m) 33.3 57.8 36.5 35.7 23.9 32.9 39.5 52.8
Elongation at break (%) 10.5 14.0 7.1 6.7 20.6 10.9 10.5 12.0
Junction strength (kN/m) GRI GG2 6.1 7.6 10.2 4.3 17.7 28.1 7.4 7.1
Flexural rigidity (mg cm) ASTM D1388, modified 146,119 271,509 1,429,355 452,671
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Torsional stiffness (cm kg/deg) COE/GRI GG9 3.47 3.97 7.50 3.43

*MD, machine direction; **TD, transverse/cross-machine direction.

standards as well as standards set forth by the Geosynthetic of the pressure imparted on the pavement subgrade during
Research Institute (GRI). Properties measured from index the APT based on the applied traffic loading. Shear forces
tests are the most commonly used criteria in specifications are applied at a constant displacement rate of 0.1 cm/min,
for use of geogrid products by state highway agencies. slow enough to dissipate soil pore pressure.
Table 2 lists the index tests conducted on the four geogrid Figure 3(a) shows the direct shear test results for the
products, standard test protocols followed, along with the control interface, i.e. subgrade soil – base aggregates
measured values of physical and mechanical properties. without geogrids. The shear stresses at zero displacement
indicate the resistance to sliding. As expected, the applied
shear stress increases with increasing normal pressure.
3.2 Bench-scale tests The shear strength parameters of the interface, adhesion
The objective of the bench-scale tests is to evaluate the and friction angle, can be obtained from the Mohr –
performance of the geogrid under the conditions and in the Coulomb failure envelope derived from the peak values of
medium in which it will be installed, as opposed to index direct shear test results (Figure 3(b)).
tests where the geogrid is tested in isolation. Direct shear Given the shear strength parameters of the control
and pull-out tests are conducted in this study to interface, the interface efficiency factor can be calculated
characterise the interaction properties of the various as (Koerner 1998)
types of geogrids installed between the subgrade and the
aggregates used in the aggregate base layer. tan d
Ef ¼ ; ð1Þ
tan f

3.2.1 Direct shear test where d is the friction angle of the geogrid reinforcement
The direct shear test is conducted in conformance with interface and f is the friction angle of the control interface.
ASTM D3080 to measure the friction angle and adhesion at The efficiency factor for geotextiles varies from 0.6 to 1,
the interface between the subgrade and the aggregate base but can be greater than 1 for geogrids (Juran et al. 1988).
layer, with and without a geogrid in place. The soil, Table 3 provides a summary of the tested properties from
aggregate and geogrid products of the test specimens are the direct shear tests for the geogrid-reinforced interfaces.
same as those used in the APT. The geogrids are placed The four tested geogrids have interface efficiency factors
between the upper aggregate box and the lower soil box. greater than 0.5.
The dimensions of both the boxes are 5.1 cm £ 30.5 Figure 4 illustrates the shear stress –displacement
cm £ 10.2 cm. The base aggregate is remoulded and behaviours for the geogrid-reinforced soil – aggregate
compacted to 100% of maximum dry density at optimum interfaces under the same normal pressure of 27 kN/m2.
moisture content. The subgrade soil for these tests is Although, the interface characteristics during direct shear
compacted to 92.5% of maximum dry density at optimum tests can be influenced by many factors, such as applied
moisture content (10%). Direct shear tests are performed normal pressure, geogrid material characteristics and
under three different normal pressures: 12, 27 and drainage conditions, for this study, the geogrid material
36 kN/m2. The selected pressure of 27 kN/m2 is an estimate properties are expected to be the only factor affecting the
 416 X. Tang et al.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 3. Direct shear tests: (a) shear stress – displacement under three normal pressures and (b) friction angle of the control interface.

interface since all other factors are held constant among in Figure 5. The geogrid samples are cut into
the four tests. 1.2 m £ 0.6 m sections and inserted into a 0.4 m thick
compacted aggregate layer with the machine direction ribs
oriented parallel to the pull-out direction. All pull-out tests
3.2.2 Pull-out test
are carried out under a normal pressure of 6.9 kN/m2 and at
Pull-out tests are conducted on four geogrids in a medium a displacement rate of 0.1 cm/min. The geogrid displace-
consisting of the base course aggregates used in the ments are measured at the front and at 31, 61, 89 and
pavement section as per ASTM D6706 in the machine 116 cm away from the front through a telltale system
direction of the geogrid. The pull-out test set-up is shown having steel wires connecting geogrids to linear variable
differential transformers.
Table 3. Results of the direct shear and pull-out tests. The interaction coefficient, Ci, represents the ratio of
the average interface strength to the internal shear strength
Property Grid A Grid B Grid C Grid D Control
of the base course aggregates and is used herein to quantify
Friction angle, 28.6 44.0 48.0 32.7 44.2 the reinforcement effectiveness for pull-out tests. Ci is
dpeak (deg) calculated according to Bergado and Chai (1994) and
Efficiency 0.56 0.99 1.14 0.66 N/A
factor, Ef Tatlisoz et al. (1998) as
Adhesion, 1.72 0.00 0.00 3.69 0
c (kN/m2) P
Ci ¼ ; ð2Þ
Interaction 0.86 1.00 0.82 0.62 N/A 2 WLðc þ sn tan fÞ
coefficient, Ci
where Ci is the coefficient of interaction, P is the maximum
pull-out load, sn is the applied normal pressure, c is the
adhesion of soil medium tested, f is the friction angle of
soil medium tested, W is the width of the geogrid specimen
and L is the embedded length of the geogrid in the soil.
Ci is a function of frictional characteristics between the
geogrids and the surrounding unbound materials, the
strength of the geogrid junctions, the flexural stiffness of
the transverse ribs and the geogrid per cent open area.
The calculated interaction coefficients for the four geogrid
cases are presented in Table 3. A strong bond between the
soil and the geogrid corresponds to an interaction
coefficient value greater than 1. An interaction coefficient
less than 0.5 implies a weak bond between the geogrid and
the surrounding materials and/or possible breakage of
geogrid cells. Most values reported for geosynthetics range
Figure 4. Shear stress – displacement at the subgrade – between 0.5 and 1 (Mohiuddin 2003).
aggregate base interface under a pressure of 27 kN/m2 for the Figure 6 shows the pull-out force – displacement
control case and with each of the four tested geogrids. relationships for Grids A – D at the nearest location
 International Journal of Pavement Engineering 417
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 5. Pull-out test set-up: (a) schematic plan view of the pull-out box; (b) top view of the pull-out box showing the geogrids on the
soil and tubes housing steel wires and (c) connection of steel wire to a geogrid rib (courtesy by TRI Inc.).

(31 cm) from the front face of the pull-out box. Grid B reinforcement since traffic-induced deformation of geo-
exhibits the highest peak pull-out force. Although, Grid grids in pavements is minimal. From that standpoint, the
C’s interaction coefficient, derived from the maximum coefficient of interaction results should be used cautiously.
pull-out load, is the second lowest among the four The magnitude of the necessary pull-out force to induce
geogrids, Grid C has the best pull-out resistance at small small displacements is more indicative of performance in
displacements (up to 11 mm in this case). Similar trends pavements.
are observed at the other locations: 61, 89 and 116 cm from Figure 7 demonstrates the relationship between the
the front face. Note that the attributes of geogrids at small pull-out force and the displacement at different distances
strains are important when geogrids are used as pavement from the front of the pull-out box for Grids C and D, which
represent stiff and flexible geogrid behaviours, respect-
ively. Along the pull-out direction, the portion of Grid C
furthest from the pulled end (back end of the pull-out box)
does not show significant movement until the occurrence
of pull-out failure. By contrast, significant displacement at
all the telltale locations indicates possible slippage of Grid
D at the interface. This again indicates that Grid C has
better pull-out resistance in spite of its low interaction
coefficient.

3.3 Accelerated pavement testing

Accelerated testing is conducted on laboratory-fabricated
slabs for the four types of geogrid products using the one-
third scale model mobile load simulator (MMLS3). Two
sets of accelerated pavement tests are performed, each for
a different subgrade CBR value. The first set of testing,
Figure 6. Pull-out load– displacement for Geogrids A – D at denoted as APT I, corresponds to the testing of all four
locations 31 cm from the front of the pull-out box. geogrids on a subgrade with a CBR value of 3, while the
 418 X. Tang et al.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 7. Relationship between the pull-out force and the displacement: (a) Geogrid C and (b) Geogrid D.

subgrade for the second set of testing (APT II) has a CBR 3.3.2 Pavement slabs
value of 1.5. Only three of the four geogrids are tested in 3.3.2.1 Configuration. The pavement slabs are con-
APT II: Grids B – D, in addition to a control section with no structed in a test pit with reinforced concrete walls and
geogrid. foundations. The pit is backfilled with aggregate base and
compacted to serve as pavement bedrock. Figure 9 shows
the layout of the constructed pavement sections for APT
3.3.1 Testing equipment I. There are four constructed sections labelled R1, R2, R3
The MMLS3, as shown in Figure 8, is an APT device that and R4 that are reinforced with Geogrids A – D,
applies unidirectional trafficking to the pavement in a respectively. Similarly, four pavement sections, P1 – P4,
controlled laboratory environment or on full-scale are constructed for APT II. Section P1 is a control section,
pavements in the field. i.e. no geogrid, while sections P2, P3 and P4 are reinforced
The MMLS3 has four tyres, each with a diameter of with Geogrids B, C and D, respectively. The four
30 cm and a width of 8 cm. The actual wheel path pavement slabs are constructed using the same materials
generated by the MMLS3 is approximately 137 cm long. for the individual layers. The subgrade is placed on top of
The load exerted by each wheel of the MMLS3 is 2.7 kN, a waterproof membrane to avoid moisture loss to the
with a corresponding tyre pressure of 621 kPa. bedrock layer below. Subgrade soil and ABC are placed
The MMLS3 suspension system is designed so that the and compacted uniformly and at the same moisture
wheel load is independent of the wheel vertical content for all sections. The only difference among the
displacement; thus, the applied load remains constant sections is the geogrid type at the subgrade – ABC
even if rutting occurs. The traffic speed is set to 7200 axles interface.
(wheels) per hour or two axles (wheels) per second.
Testing is conducted at room temperature under dry
conditions with no wandering, i.e. channelised trafficking. 3.3.2.2 Structure. The MMLS3 applies a wheel load of
The P900 contact-type electronic profilometer (MLS 2.7 kN with a contact pressure of 690 kPa (100 psi) roughly
Test Systems Pty Ltd, Stellenbosch, South Africa) is used to representing one-ninth of the loading conditions applied
measure pavement rutting profiles. Measurements are made by a standard full-scale single tyre (one-quarter dual-tyre
through a 25 mm diameter, 10 mm wide steel wheel that ESAL). In an effort to attain similitude between the scaled
rolls on the pavement surface. The vertical measurement slabs and the actual field slabs, the thickness of each layer
resolution is approximately 1 mm and the repeatability is in should be scaled approximately to one-third of that in the
the range of 10 mm. The maximum vertical displacement field (Martin et al. 2003).
that can be measured is 40 mm. The increment of the In this study, the Kenlayer program (Huang 2004) is
measuring wheel is 2 mm; that is, profile measurements are used to determine the required scaling ratio that is
recorded every 2 mm across the wheel path. expected to be close to 1/3. The thicknesses of AC and
In accordance with ASTM D7113, a non-nuclear base course layers of 3.8 cm (1.5 in.) and 7.6 cm (3 in.),
density gage, Pavement Quality Indicatore (PQI) Model respectively, are found to lead to vertical stresses on top of
301 (Engius, LLC), is used to provide quick readings of the subgrade and horizontal strains in the geogrid similar
density, temperature and moisture content of the AC layer. to those for typical full-scale flexible pavements compris-
The PQI works through a constant voltage, low frequency ing 10 cm of AC and 20 cm ABC. This corresponds to a
and electrical impedance mechanisms. scaling ratio of 1/2.6.
 International Journal of Pavement Engineering 419
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 8. One-third scale MMLS3: (a) schematic, units in mm and (b) MMLS3 mounted on top of a laboratory pavement slab.

3.3.2.3 Construction. The pavement subgrade soil is dominates the overall pavement rutting. Table 4 presents
compacted in three lifts with a vibratory plate compactor. the as-constructed layer properties for both APTs I and II.
The compactor travel direction ensures consistency of the The AC layer density is measured along the wheel path
soil density throughout the pit. The geogrids are carefully for each section using the PQI non-nuclear density gage
unrolled to avoid folds and wrinkles and placed directly on prior to MMLS3 trafficking. Figure 10 presents a contour
the subgrade layer. The geogrids from adjacent sections plot of the measured AV distribution for each section
have a 7.6 cm overlap to ensure adequate development measured after construction and before trafficking.
length, and the ends of the geogrid sections are folded The asphalt layer in sections R3 and P3 reinforced by Grid
against the pit walls to obtain necessary anchorage and C has significantly higher AVs than the other sections in both
slight pre-tensioning, as well as to prevent shifting of the test sets. It is highly likely that the high AC AVs in these two
geogrids out of position (Figure 9). The geogrid is overlaid sections result from the attributes of the geogrid itself, which
with an ABC layer compacted using a vibratory plate has a significantly higher stiffness than the other geogrids
compactor. A 6% AV content is targeted for the AC layer. tested, given the uniform compaction technique and
The target AV is similar to that targeted in field consistent pavement structure and materials used. Further
construction and is expected not to significantly contribute studies that incorporate different types of geogrids and
to the overall rutting of the pavement, since the low CBR subgrade soils are needed to confirm and further understand
of the subgrade would lead to extensive deformation that the mechanisms causing this phenomenon.
 420 X. Tang et al.

for section R3 at various traffic cycles is also plotted in

Figure 11(c).

3.3.3.2 APT II. Rutting failure in the APT II test sections

occurs well before 140,000 MMLS3 traffic cycles can be
applied. Hence, MMLS3 trafficking is applied to each
section of the APT II slab until rutting failure occurs.
The control section P1 exhibits extensive deformation
after 40,000 wheel applications, while similar deformation
is observed at 70,000 applications for section P2.
Trafficking is stopped at 100,000 wheel applications for
sections P3 and P4. Figure 11(b) shows the rutted wheel
paths for the fours sections. The extensive rutting and
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

resulting cracking at the edge of the wheel path for the

control section P1 is clearly visible. The measured rutting
profile for section P3 is also shown in Figure 11(d).

4. Observations and analysis of results

Except for the difference in geogrid products, the
experimental variables are minimised as much as possible
in the bench-scale testing and APT. The subgrade soil,
ABC and asphalt materials are consistent for all tests. It is
Figure 9. Pavement section layout: (a) dimensions, units in cm therefore reasonable to presume that most of the difference
and (b) plan view of the section and geogrid installation. in the performance is due to the intrinsic properties of
geogrids and their interaction characteristics with the
medium in which they are placed. This section first covers
3.3.3 Accelerated trafficking of slabs the observation and analysis of the performance of the
3.3.3.1 APT I. A total of 140,000 MMLS3 traffic geogrids from accelerated trafficking. An attempt is then
cycles are applied to each section. Rutting accumulation at made to find possible correlations among physical and
specific locations throughout each section is measured and mechanical properties of geogrids (index properties),
averaged. Figure 11(a) shows the rutted wheel path for each mechanistic parameters from bench-scale testing and
section. To shed light on the accumulation of deformation performance from accelerated testing of the scaled slabs.
as trafficking cycles are applied, the measured rutting profile The comparisons conducted are limited to particular index

Table 4. As-constructed pavement layer properties.

Subgrade Base course

APT thickness (cm) thickness (cm) AC thickness (cm) Subgrade moisture content (%) Subgrade CBR value (%)
I 23.6 6.6 3.8 14 3
II 15.2 6.6 3.8 14.8 1.5

R4

Figure 10. AV content distribution in the asphalt layer: arrows represent compaction path (same for both test sets) and shading
represents the range of AV content: (a) APT I and (b) APT II.
 International Journal of Pavement Engineering 421
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 11. Rutting of the wheel paths: (a) APT I and (b) APT II; measured rutting profile at various stages of trafficking for sections:
(c) R3 in APT I and (d) P3 in APT II based on the average from all measured locations along the wheel path.

properties that are deemed important for pavement between 70,000 and 100,000 cycles). This illustrates the
applications, i.e. low strain conditions. ability of geogrids to stabilise weak subgrade and potential
to minimise pavement deformation under the traffic load.
Similar to APT I, section P3 reinforced with Geogrid C
4.1 Performance of geogrids – accelerated testing exhibits relatively higher rutting among the reinforced
4.1.1 APT I sections, probably due to the higher AVs and consequent
In APT I, the most significant rutting is observed for section rutting of the AC layer (Figure 12(b)).
R3 (Grid C), as shown in Figure 12(a). However, this could In order to minimise the effect of variation in the AV
partially be attributed to the high AV of the AC layer and the content of the asphalt layer, the rutting accumulation at
possible densification of that layer (Figure 10(a)). locations with similar compacted asphalt AVs is presented
By comparing the other three sections R1, R2 and R4 in Figure 12(c). Section P2 (Grid B) has the second highest
which had relatively similar AV, it is evident that Grids A cumulative rutting among the four test sections. The rate of
and B exhibit similar rutting for this type of subgrade soil, rutting accumulation for sections P1 and P2 is slightly
with higher rutting in section R4 reinforced with Grid D. higher than that for sections P3 and P4. Significantly
After trafficking, the slab is trenched across the travel higher rutting is observed for section P4 (Grid D) when
direction to expose the pavement cross-section. Figure 13 compared to that for P3 (Grid C). Overall, the relatively
shows extensive densification occurred in the AC layer of stiffer geogrid, Grid C, shows the best performance with
section R3 with minimal densification occurring in R4. respect to the rutting resistance for the tested pavement on
A significant amount of soil slurry is observed underneath this specific weak subgrade.
the wheel path at the geogrid – aggregate base interface,
probably due to soil pumping from the subgrade into the
aggregate base layer during traffic loading. 4.1.3 Comparison between APTs I and II
The significant variable in designing the sections of APTs I
and II is the subgrade CBR, with a subgrade CBR of 3 and
4.1.2 APT II 1.5 for APTs I and II, respectively. The ranking of the
The control section P1 exhibits significantly higher rutting geogrid products in terms of the ability to minimise
that accumulated relatively quickly (failure occurs at subgrade rutting is somewhat different between APTs I
40,000 cycles) compared to the reinforced sections (failure and II. One noticeable difference is the switch in ranking
 422 X. Tang et al.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 12. Rutting accumulation based on the average from all locations for each section: (a) APT I average rutting accumulation;
(b) APT II average rutting accumulation and (c) APT II rutting accumulation for locations with similar asphalt mixture AVs.

of rutting performance between Grids B and D, although support for a particular subgrade, it might not perform as
both are considered flexible grids. well when used for a different type of soil and/or aggregate
While more replicates are necessary, particularly for base. In the case of this study, the section with Grid B
different subgrade soil types, the measured rutting and experiences significantly higher rate of rutting accumulation
observation from trenching provide evidence that the in APT II than in APT I, indicating that Grid B might not be
effectiveness of geogrid reinforcement and strengthening of suitable for reinforcing a subgrade as weak as that in APT II.
weak subgrade is dependent on the interaction between the Due to the construction variability, sections in APT I
reinforcement and the surrounding materials. Thus, proper exhibit overall higher asphalt AVs than that in APT II
selection of the geogrid type for a given subgrade is (Figure 10). Hence, the rutting measured at locations with
essential. While a specific geogrid provides adequate similar AC AVs should be considered as points of direct

Figure 13. Trench cross-section of the pavement: (a) extensive deformation in APT I-R3 and (b) slight densification in the asphalt layer
of APT I-R4.
 International Journal of Pavement Engineering 423
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 14. Rutting for geogrid reinforcement in APT I and APT II: (a) Grid B; (b) Grid C and (c) Grid D.

comparison. Figure 14 shows the comparison between the interface efficiency factor for the types of materials used in
rutting of Geogrids B –D from the APTs I and II tests. The this study. There is no correlation between the rib thickness
rutting resulting from APT II is greater than that of APT I and the interface efficiency factor. Both Grids B and C
for all geogrid types, given the weaker subgrade in APT II. exhibit factors greater than 1.
Note, however, that the sections reinforced with Geogrid C The passive resistance exerted on the bearing members
exhibit very similar rutting in both the APTs I and II tests of the geogrid to some extent depends on the junction
despite the difference in subgrade CBR. This indicates that strength when considering the sliding movements against
Grid C is more effective in reinforcing weak subgrade the ribs between junctions, and possibly tensile strength at
compared to other geogrids used in this study. small strains. The combination of the junction strength and
the tensile strength at 2% strain in the machine direction
has a strong correlation with the interface efficiency factor
4.2 Correlation between results of index (Figure 15(b)). Junction thickness as the indictor of the
and bench-scale testing junction strength also shows a good correlation with the
4.2.1 Index and direct shear test results interface efficiency factor (Figure 15(c)).
The shear resistance of the soil–geogrid–aggregate inter-
face against direct sliding movement consists of geogrid’s
skin friction, soil–aggregate friction and passive resistance 4.2.2 Index and pull-out test results
against the geogrid’s transverse ribs. However, the Determining the correlation between geogrid index proper-
contribution from the geogrid skin friction is likely to be ties and pull-out test results is evaluated through the
minimal due to its relatively small surface area with respect coefficient of interaction. The coefficient of interaction,
to the total area of the interface. On the other hand, the area of however, is calculated at a maximum pull-out force at the
the geogrid’s aperture determines the contact surface onset of failure. For pavement applications, it is more
between the subgrade soil and the aggregates from the appropriate to investigate the effect of index properties on the
base, thus affecting the overall soil–aggregate interface behaviour of geogrids in the pull-out test at small
friction. Figure 15(a) illustrates the relationship between the displacements. The displacement experienced by the
interface efficiency factor (Equation (1)) and the aperture geogrids in pavement applications at the subgrade is much
area. Geogrids with a larger aperture area result in higher smaller than that resulting in other applications such as slope
 424 X. Tang et al.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 15. Correlation between Ef (interface efficiency factor) and geogrid index properties: (a) aperture area; (b) sum of the junction
strength and the tensile strength at 2% strain in the machine direction and (c) junction thickness.

stabilisation and retaining walls; moreover, a complete Although the index properties of Grid D and its
pull-out of the geogrid does not occur. Thus, for pavement behaviour from direct shear test are comparable to other
applications, evaluating the load–displacement relationship geogrids, particularly Grid A, it is highly likely that the
at small displacements is more meaningful to the pull-out significantly lower pull-out forces measured are due to
behaviour compared to the coefficient of interaction derived experimental errors, and thus the pull-out results of Grid D
from a single value of the maximum pull-out load. can be considered as an anomaly. This proposition
With regard to the fact that small strains in geosynthetics
are typically developed for reinforced soil structure
(McGown et al. 2005), it is more relevant to compare the
geogrid’s behaviour and interaction properties, such as
in a pull-out test, at small displacements when they will
be subjected to in the medium and under normal operational
conditions. The displacements of the geogrids for subgrade
stabilisation are expected to be minimal. Figure 16 shows the
load–displacement relationship for the four geogrids up to
5 mm displacement. Due to limitations in instrumentation,
unit displacements less than 0.3 mm cannot be measured.
Figure 16 shows that within the first 5 mm of displacement,
the slope of the pull-out force–displacement curves is nearly
the same for all tested grids. Grid C exhibits the highest pull-
out force at small displacements in spite of its low
interaction coefficient (Table 3). Furthermore, the rankings
in Figure 16 reveal a direct correlation with the friction angle
and the efficiency factor from direct shear results, with the Figure 16. Pull-out force versus displacement for the various
exception of Grid D. geogrid types at small displacements.
 International Journal of Pavement Engineering 425

is confirmed by the fact that the average rutting for the correlations are observed for tensile strength at 2% strain
section with Grid D is not as low as the pull-out force results and efficiency factor from direct shear.
suggest, and is actually less than that for all other sections in Figure 18 shows that there is no evident correlation
APT II (Figure 12(b)). Thus, Grid D will not be incorporated between the index properties and the coefficient of
in the investigation of correlations linked to the pull-out interaction. However, it is worth to further discuss the
force. effect of the aperture area on the interaction coefficient. In
Figure 17 illustrates the correlation between the index order for effective interlocking of the aggregate– geogrid
properties and the pull-out force at 5 mm. It can be skeleton to take place, the ratio of aperture size to
concluded that an insignificant correlation exists between aggregate particle size should allow aggregates to strike
the increase in aperture area and flexural rigidity and an through the aperture. Maximum interaction efforts can be
increase in the pull-out force measured at 5 mm. Stronger achieved when the grain size is similar to that of the
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 17. Effect of geogrid properties on the pull-out force measured at 5 mm displacement.
 426 X. Tang et al.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

Figure 18. Correlation between index properties and coefficient of interaction.

geogrid aperture size (Shukla 2002). As can be seen in 4.3 Correlation between index, bench-scale testing
Table 2, the aperture sizes of Geogrids A – D are greater and accelerated testing results
than 25.4 mm (1 in.). The course aggregate gradation Figure 19 shows the relationships between the geogrid
suggests that , 87% of the material passes the 19 mm bench-scale test results, namely the coefficient of
(0.75 in.) sieve. The surrounding material grain size interaction and the interface efficiency factor, and the
distribution, relatively dense graded with nominal measured subgrade rutting from the accelerated testing
maximum aggregate size of 17.5 mm (1/2 in.), may APTs I and II, while Figure 20 relates the geogrid index
positively impact that interlock. Grid B has the largest properties to the measured rutting. It is observed that the
apertures among the other geogrid products used in this effects of ultimate strength, junction strength and bench-
study (Table 2), which could partially explain its highest scale testing properties on measured rutting are not
interaction coefficient, Ci (Equation (2)). strongly evidenced, possibly due to the dominance of low
 International Journal of Pavement Engineering 427

Figure 19. Correlation of mechanistic properties from bench-scale testing with average rutting from APT.

cyclic strains occurring in the geogrids at the subgrade Flexural rigidity and aperture area seem to affect the
level when compared to high strain levels applied under a rutting according to trends that are dependent on the CBR.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

monotonic loading condition until failure in index testing. For the subgrade soil CBR of 3.0 in APT I, the general
In none of the accelerated testing runs are any of the trend shows increased rutting with increases in aperture
geogrid samples stretched to a strain limit close to or at area and flexural rigidity. The trend is reversed for APT II,
failure based on the observations of geogrids after in which the subgrade soil CBR is 1.5.
trenching. This is in line with observations made from The observations above do not offer conclusive
full-scale sections in the field (Perkins 1999). correlations between some index properties, bench-scale

Figure 20. Correlation of geogrid properties from index testing with average rutting from APT.
 428 X. Tang et al.

Table 5. Relationship between selected index and bench-scale properties with subgrade rutting from accelerated testing.

Correlation and observed trend

Efficiency Pull-out Coefficient of Rutting Rutting
Property factor @ 5 mm interaction @ 1.5 CBR @ 3.0 CBR
Aperture area I I I D I
Junction strength I I N D I
Tensile strength at 2% strain I I N N N
Ultimate strength N I N N N
Flexural rigidity N I N D I
Efficiency factor I N N I
Pull-out force @ 5 mm N N
Coefficient of interaction N N

I, increasing trend; D, decreasing trend; N, no observed trend.

Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

testing properties and measured rutting under trafficking. geogrid tensile strength at 2% strain, junction strength and
More trials and replicates of APT need to be conducted to results of direct shear tests, while aperture opening does not
evaluate the statistical significance of any observed show a strong correlation with the direct shear test results.
correlations and to assess the possibility of statistical As another indicator of the effectiveness of using
interaction between the properties and their effects on geogrids as reinforcing elements for subgrade stabilis-
rutting. Table 5 summarises the observed correlations ation, the coefficient of interaction between the geogrid
among the index properties, bench-scale testing properties and the surrounding material is determined through
and measured rutting from the APT. pull-out tests. A strong correlation between aperture
It should be noted that during the construction of the opening and interaction coefficient indicates that the
pavement sections for both the sets of APT, the sections aperture opening of the geogrid plays an important role in
that are reinforced with a stiff geogrid exhibited difficulty its interaction with the ABC materials. Although ultimate
in achieving proper compaction of the asphalt layer. It is tensile strength and junction strength are not strongly
speculated that the high stiffness of the geogrid itself led to correlated with the coefficient of interaction, they are
the insufficient compaction since the stiffness is the major expected to be important intrinsic properties of geogrids
difference between Grid C and other geogrids used in this for gaining high pull-out resistance. Grid C with the
study; nevertheless, care should be taken to ensure the highest junction strength shows the best pull-out resistance
proper construction of thin geogrid-reinforced pavements. according to the relationship between the pull-out force
Interpretations and any consequent conclusions presented and the telltale displacements.
here are based solely on the types of soil, geogrids, The APTs I and II tests are conducted using the
structural design and loading used in the two sets of APT. one-third scale MMLS3 on pavement sections with
cross-sections scaled to one-third of actual pavement
dimensions in the field. Pavement sections are constructed
5. Conclusions and recommendations in a test pit and tested on two types of subgrade soils. For
The objective of this study addresses the evaluation of the APT I test, although, the lowest rutting is recorded for
geogrids for weak subgrade stabilisation. Evaluation is the section with the best interaction performance, it does
done through a spectrum of experiments ranging from not show strong consistency with bench-scale testing
index testing to determine physical properties of the results, partially due to the variation in the AV content of
geogrid, passing through bench-scale tests to understand the AC layer of that section. In the APT II test, rutting
the mechanical behaviour of the geogrids in an unbound accumulation with the exclusion of asphalt AV variation
medium, culminating in performance evaluation under demonstrates the substantial benefits of geogrid stabilis-
accelerated traffic loading. Results from the various levels ation for weak subgrade soil. Furthermore, a comparison
of testing are investigated to determine correlations of rutting performance differentiates the effectiveness of
between index properties, bench-scale testing and geogrid reinforcement for subgrade stabilisation among
ultimately, the measured rutting under APT. the three geogrid products in APT II. This difference is
The interface efficiency factor parameter is used to correlated with certain geogrid index properties and
evaluate the performance of geogrids in the direct shear tests. interaction characteristics through a correlation analysis.
Tensile strength at 2% strain and junction strength contribute Aperture size, tensile strength at small strains, junction
to the characterisation of geogrids in a complex manner, strength and flexural rigidity are recognised as the most
affecting the interaction properties of the reinforcement important attributes of geogrids in pavement subgrade
interface. A good correlation is found between combined stabilisation.
 International Journal of Pavement Engineering 429

By comparing the rutting behaviour among sections in highway engineering applications. Report No. FHWA-PA-
the two sets of APT, certain geogrid products are found to 2007-009-050110.
Farrag, K., Acar, Y.B., and Juran, I., 1993. Pull-out resistance
be more suitable for use in conjunction with a stronger
of geogrid reinforcements. Geotextiles and Geomembranes,
subgrade. It is therefore possible that geogrids that meet 12 (2), 133– 159.
the criteria in physical and mechanical properties Hass, R., Wall, J., and Carroll, R.G., 1988. Geogrid
identified above will provide sufficient reinforcement for reinforcement of granular bases in flexible pavements.
weak subgrade soils. Transportation Research Record 1188. Washington, DC,
19 – 27.
Observations and correlations made are limited to the
Huang, Y.H., 2004. Pavement analysis and design. Upper Saddle
two types of subgrade soils and one aggregate type used for River, NJ: Pearson Prentice Hall.
subbase. A high number of variables exist in conducting the Hufenus, R., et al., 2006. Full-scale field tests on geosynthetic
accelerated testing, including soil types, geogrid types, base reinforced unpaved roads on soft subgrade. Geotextiles and
layer aggregate gradation, moisture conditions and AVs of Geomembranes, 24 (1), 21 – 37.
Ingold, T.S., 1983. Laboratory pullout testing of grid
asphalt layers, among others. Important variables to
reinforcements in sand. Geotechnical Testing Journal, 6 (3),
consider include the extent of mixing between the subgrade 101– 111.
Downloaded by [University of Tennessee, Knoxville] at 15:06 24 December 2014

soil and the base layer aggregates in addition to the exact Jewell, R.A., et al., 1984. Interaction between soil and geogrids.
location of the interface with respect to the geogrid after Proceedings of the conference on polymer grid reinforce-
compaction. Thus, a wider scale study should be conducted, ment. London: Thomas Telford, 18 –30.
Juran, I., et al., 1988. Pull-out response of geotextiles and
minimising these variables as much as possible and
geogrids (synthesis of available experimental data). Proceed-
including instrumentation for displacement measurement ings of symposium on geotextiles for soil improvement.
within each layer, strain measurement on the geogrid and ASCE, GSP No.18. Reston, VA: ASCE, 92 – 111.
stress at the subgrade. Koerner, R.M., 1998. Designing with geosynthetics. 4th ed.
New Jersey: Prentice Hall.
Love, J.P., 1984. Model testing of geogrids in unpaved roads.
Dissertation (Doctoral). University of Oxford, Oxford, UK.
Notes Martin, A.E., et al., 2003. Pavement response and rutting for
1. Email: [email protected] full-scale and scaled APT. Journal of Transportation
2. Email: [email protected] Engineering, 129 (4), 451– 461.
McGown, A., et al., 2005. Testing biaxial geogrids for
specification and design purposes. GRI-18 Geosynthetics
Research and Development in Progress. Conference at
References GeoFrontiers, Reston, VA, ASCE.
Al-Qadi, I.L., et al., 1994. Laboratory evaluation of geosynthetic- Metcalf, J.B., 1996. Application of full-scale accelerated
reinforced pavement sections. Transportation Research pavement testing. NCHRP Synthesis of Highway Practice
Record 1439, Washington, DC, 25 – 31. No. 235. Washington, DC: Transportation Research Board,
Al-Qadi, I.L., et al., 2007. Accelerated full-scale testing of National Research Council.
geogrid-reinforced flexible pavements. TRB 2007 Annual Mohiuddin, A., 2003. Analysis of laboratory and field pull-out
Meeting (CD-ROM). Washington, DC: Transportation tests of geosynthetics in clayer soils. Thesis. Louisiana State
Research Board, National Research Council. University, Baton Rouge, Louisiana.
Barksdale, R.D., Brown, S.F., and Chan, F., 1989. Potential Perkins, S.S., 1999. Geosynthetic reinforcement of flexible
benefits of geosynthetics in flexible pavement systems. pavements: laboratory based pavement test sections. Report
National Cooperative Highway Research Program (NCHRP) No. FHWA/MT-99-001/8138. US Department of Transpor-
Report No. 315. Washington, DC: Transportation Research tation, Federal Highway Administration, Washington, DC.
Board, National Research Council. Shukla, S.K., 2002. Geosynthetics and their applications.
Bergado, D.T. and Chai, J.-C., 1994. Pullout force/displacement London: Thomas Telford.
relationship of extensible grid reinforcements. Geotextile Tatlisoz, N., Edil, T.B., and Benson, C.H., 1998. Interaction
and Geomembrane, 13, 295– 316. between reinforcing geosynthetics and soil-tire chip
Chehab, G.R., Palomino, A.M., and Tang, X., 2007. Laboratory mixtures. Journal of Geotechnical and Geoenvironmental
evaluation and specification development for geogrids for Engineering, 124 (11), 1109– 1119.