Technical Paper by M.C. Alfaro, S.

Hayashi,
N. Miura and K. Watanabe

PULLOUT INTERACTION MECHANISM OF

GEOGRID STRIP REINFORCEMENT
ABSTRACT: Soil-reinforcement interaction parameters used in the design of reinforced soil systems are typically evaluated by measuring pullout resistance. A combined two-dimensional and three-dimensional interaction mechanism is proposed to
evaluate the pullout resistance of geogrid strip reinforcement embedded in dense granular soil. Pullout tests were conducted using uniaxial geogrid reinforcement of various
specimen widths. Results from the pullout tests confirmed that a three-dimensional interaction mechanism had developed at both edges of the strip reinforcement. This interaction mechanism is caused by a confinement of the dilating zone of soil around the
reinforcement as described by Schlosser and Elias (1978). For the type of geogrid reinforcement and backfill material used in this investigation, the combined interaction
mechanism is applicable for a strip reinforcement width, B, greater than 20 cm. For strip
reinforcement with B less than 20 cm, the influence of restrained soil dilatancy results
in the development of a purely three-dimensional interaction mechanism across the
strip reinforcement width.
KEYWORDS: Geosynthetic, Geogrid, Pullout test, Soil-reinforcement interaction,
Restrained dilatancy, Granular soil.
AUTHORS: M.C. Alfaro, Graduate Student, S. Hayashi, Professor, N. Miura,
Professor, and K. Watanabe, Graduate Student, Department of Civil Engineering, Saga
University, 1 Honjo, Saga 840, Japan, Telephone: 81/952-24-5191, Ext. 2583, Telefax:
81/952-23-6759.
PUBLICATION: Geosynthetics International is published by the Industrial Fabrics
Association International, 345 Cedar St., Suite 800, St. Paul, MN 55101, USA,
Telephone: 1/612-222-2508, Telefax: 1/612-222-8215. Geosynthetics International is
registered under ISSN 1072-6349.
DATES: Original manuscript received 16 February 1995, revised manuscript received
and accepted 16 May 1995. Discussion open until 1 March 1996.
REFERENCE: Alfaro, M.C., Hayashi, S., Miura, N. and Watanabe, K., 1995,
Pullout Interaction Mechanism of Geogrid Strip Reinforcement, Geosynthetics
International, Vol. 2, No. 4, pp. 679-698.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

679

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

INTRODUCTION

The use of geosynthetics as reinforcement for earth structures has grown rapidly in
the past two decades. When used for retaining walls or steep slopes, geosynthetic reinforcement can be laid continuously along the width of the reinforced soil system (sheet
type) or laid in intervals (strip type). Sheet reinforcement is usually employed when the
geosynthetic reinforcement itself forms a part of the wrap-around facing, while strip
reinforcement is commonly associated with panel type facings.
Soil-reinforcement interaction parameters for the design of reinforced soil systems
are usually evaluated by measuring pullout resistance. The primary pullout interaction
mechanism of sheet reinforcement is classical soil-reinforcement interface friction
which will be called the two-dimensional (2-D) interaction mechanism in this paper.
When strip reinforcement is placed in dense granular soils another interaction mechanism occurs which is called the three-dimensional (3-D) interaction mechanism in this
paper. This mechanism develops as a consequence of the restrained soil dilatancy effect
described by Schlosser and Elias (1978). As the strip reinforcement is pulled out and
shear displacement occurs along the interface, the zone of soil surrounding the reinforcement tends to dilate. However, the volume change is restrained by the surrounding
nondilating soil, inducing an increase in normal stress on the soil-reinforcement interface.
Early reinforced soil systems utilized a very narrow metallic strip reinforcement in
which an idealized 3-D interaction mechanism is applicable. However, present reinforced soil practice includes the use of geosynthetic reinforcement (e.g. geogrids) strips
which range in width from 0.2 to 1 m. In such cases, the soil-reinforcement interaction
can be a combination of 2-D interaction mechanisms developing over the middle section, and 3-D interaction mechanisms developing at both edges of the strip reinforcement as shown in Figure 1.
In this paper, an investigation of the combined 2-D and 3-D pullout interaction mechanisms of geogrid strip reinforcement embedded in dense granular soils is reported. The
paper also describes the laboratory testing program and the procedures that were
employed to evaluate the parameters involved in these interaction mechanisms. The results obtained are presented and discussed.
2

CONCEPTUALIZED PULLOUT INTERACTION MECHANISM

As shown in Figures 1a and 1b, the nondilating zone in the backfill soil surrounding
the strip reinforcement functions as a restraint against soil dilatancy in the dilating zone.
This generates shear stresses at the border between the dilating and the nondilating
zones and results in an increase in normal stresses at both edges of the strip reinforcement. A 3-D interaction mechanism develops at both edges of the strip reinforcement
while its middle section experiences 2-D interaction behavior (Figure 1c). As the width
of reinforcement becomes narrower, the influence of restrained dilatancy results in the
development of what is considered a purely 3-D interaction mechanism (Figure 1d).
The increase in normal stress at both edges of the strip reinforcement further suppresses
soil dilatancy during interface shear displacement. Based on this combined interaction
mechanism, the distribution of normal stresses imposed on the strip reinforcement is

680

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

(a)

(b)
Dilating zone
Shear stress occurred
at the border between
the dilating zone and
the nondilating zone

Nondilating
zone

Nondilating
zone
Dilating zone

(c)

(d)

3-D interaction mechanism

(restrained
dilatancy effect)

Applied
normal
stress

Increase in
normal stress
n

n
n

Be

B --- 2Be

Be

B 2Be

2-D interaction mechanism

(interaction effect)

Figure 1. Conceptualized pullout interaction mechanism of strip reinforcement: shear

stress and strain mobilized around (a) wide strip, and (b) narrow strip; distribution of normal
stress on (c) wide strip, and (d) narrow strip.

idealized as shown in Figure 2. Another aspect of this interaction mechanism that requires due consideration is that only part of the reinforcement length will be mobilized
when an extensible reinforcement, such as geogrid, is pulled out from the backfill soil.

B
Be

Be
n
n

Strip reinforcement
Figure 2. Idealized distribution of normal stress on soil-reinforcement interface.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

681

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

This effect is particularly prevalent under high normal stresses whereby nodal displacements tend to localize near the pullout load application which results in a shorter mobilized reinforcement length. Thus, the evaluation of maximum pullout resistance should
take into account only the effective/mobilized reinforcement length.
For the above interaction mechanism, the following relationship is proposed for the
maximum effective pullout resistance, PTE :
P TE = P 2-D + P 3-D

(1)

where: P2-D = pullout force from 2-D interaction resistance; and P3-D = pullout force from
3-D interaction resistance. Equation 1 can also be expressed as:
P TE = 2BLe n tan p + 4B e Le n tan p

(2)

where: B = width of reinforcement; Le = effective/mobilized reinforcement length; n

= applied normal stress; Be = width along the edge of reinforcement influenced by restrained dilatancy effect; n = increase in normal stress at the soil-reinforcement interface due to restrained dilatancy on width Be ; and p = friction angle at soil-reinforcement interface. Equation 2 neglects interface adhesion and assumes that the friction
angle at the soil-reinforcement interface is the same for 2-D and 3-D interaction mechanisms; thus, only the increase in normal stress contributes to the 3-D interaction resistance as quantified by the product Be n , while Be remains constant. The values of Be
and n were obtained from pullout tests using reinforcement specimens of various
widths under different applied normal stresses.
3

LABORATORY TESTING PROGRAM

3.1

Equipment and Materials Used

The test apparatus used in this investigation is made of rolled steel plates, angles,
channels, and H-sections, welded or bolted together to give the dimensions shown in
Figure 3. Its dimensions were selected as a compromise between the need to minimize
the effects of the apparatus boundaries and to minimize the difficulty of preparing an
extremely large volume of backfill soil. The thickness of the soil layers above and below the reinforcement was 15 cm and 25 cm, respectively. A thicker soil layer below
the reinforcement was chosen to minimize the restraining effect of the rigid bottom
boundary. The thinner soil layer above the reinforcement together with a flexible top
boundary was preferred for the purpose of measuring soil dilatancy closer to the soil-reinforcement interface. The reinforcement clamping system was positioned inside the
compacted soil. This arrangement ensured that the geogrid specimen, which was considered to be the extensible reinforcement, was confined during the pullout test. It also
ensured that coupling of the unconfined and confined portions of geogrid reinforcement
would not occur during the pullout test. The pullout resistance of the clamping system
was determined by conducting pullout tests on the clamping system alone and subtract-

682

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Reinforcement specimen for

pullout test
3230
1630
Clamping plates for pullout test
Universal joint

300

850
700
600
Varies

Air pressure inlet

Air pressure gauge

700

1600
1500

Vertical
displacement
rods

Load cell
Sleeve
Wooden spacer
Rollers for direct shear test

PLAN
Speed control device
Load cell
Universal joint
Sleeve

Air pressure inlet/outlet

20

Reaction column
910

LVDT

Clamping plates
for pullout test

Reinforcement
Compacted soil

50

LVDT

200

200

200 100 150 150 100

Vertical displacement
rods
Air bag

Adjusting handle for

vertical movement
of pulling jack

1300

Clamping plates for

direct shear test

Pulling jack

Inextensible wires

50

Clamping plates for direct shear test

1500
1600

50
1630
3230

SECTION

Pulling jack

Figure 3. Schematic diagram of pullout/direct shear test apparatus.

(Note: Dimensions in mm.)

ing its resistance from the measured pullout loads on geogrid specimens. This apparatus
is capable of performing both pullout and direct shear tests. A typical test setup is shown
in Figure 4. A rubber air bag was used to produce a uniformly distributed vertical pressure on top of the backfill soil. A rubber membrane lubricated with silicone grease was
employed to minimize friction between the soil and the side walls of the testing box.
Linear variable differential transformers (LVDTs) were used to measure the displacements along the reinforcement length and the soil dilatancy during pullout. Load cells
were used to measure the applied pullout load. Measuring the displacements along the
reinforcement length allows for proper interpretation of the interface load transfer
mechanism and also provides an appropriate evaluation of the pullout resistance. Nodal
displacements along the reinforcement length were measured using cables connected

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

683

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Figure 4. Typical geogrid pullout test setup.

to the LVDTs mounted on the rear wall of the testing box. These cables, which run inside
stiff tubing to provide protection from direct contact with the soil, were always kept tensioned by spring-loaded LVDTs. The nodal displacements were used to calculate the
average tensile strains between two adjacent nodes. Tensile strains inferred from nodal
displacement measurements are considered to be reliable, and this technique has been
used by a number of researchers. All instrumentation was monitored by a personal computer connected to an electronic datalogger which was programmed to scan the instrumentation at desired time intervals. Additional details of the test equipment and instrumentation can be found in Alfaro et al. (1995).
A well-graded sandy gravel was used as the backfill soil with the following particle
size properties: average particle size, d50 = 4.74 mm; uniformity coefficient, Cu = 15;
and coefficient of curvature of the gradation curve, Cc = 1.67. The maximum and minimum dry unit weights were 19.10 kN/m3 and 14.32 kN/m3, respectively. The internal
friction angle of the compacted soil determined from triaxial compression testing was
45_ at a relative density of 95%. Tensar SR-80, a uniaxial geogrid normally used for
reinforced soil walls and steep slopes, was employed as the reinforcement specimen.
Figure 5 shows the results of in-air load-extension tests (wide-width strip tensile tests)
performed at a temperature of 19_C and at a displacement rate of 1 mm/min - the same
rate of displacement used in pullout tests. The results are similar to the test data found
in the manufacturers technical brochures, and indicate no significant difference between the load-strain curves for the 10 and 30 cm width specimens.
3.2

Testing Conditions and Procedures

The backfill soil was compacted in the box by manually tamping layers not exceeding
10 cm in thickness so as to provide adequate compaction. A relative compacted density
of 95% was chosen on the basis of the weight of the soil and the volume of the testing
box. Friction between the soil and the side walls of the box was minimized by the use
of a lubricated rubber membrane. This was verified by installing pressure cells near the

684

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

Load (kN/m)

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

10 cm width
} This study
30 cm width
Manufacturers
30 cm width }
literature
Strain (%)
Figure 5. Results of geogrid in-air load-extension tests.

reinforcement levels along the half-width of the testing box. The pressure cells were
used to detect any relative differences in the measured normal stress at the centerline
and side walls of the testing box. Figure 6 shows the pressure cell results, indicating
similar measured values at the center and side wall locations, and thus demonstrating
minimal wall friction. These measured normal stresses are greater than the applied normal stresses, which may be attributed mainly to the stress concentration, or arching effect, caused by higher cell stiffness relative to the soil.
A total of 24 pullout tests were conducted using six specimen widths (B = 10, 15, 20,
30, 45, and 58 cm) under four applied normal stresses (n = 20, 30, 40, and 50 kN/m2).
The series of tests with specimen widths of B = 58 cm, which is slightly smaller than
the width of the testing box, corresponds to a 2-D interaction mechanism. This interaction mechanism was envisaged to be appropriate for this series of tests because the lubricated side walls would not induce a restraining effect which could be caused by the
presence of dilating and nondilating zones within the backfill soil. Thus, the assumed
2-D interaction behavior across the width of the reinforcement is appropriate. On the
other hand, the narrower specimens are assumed to experience a combination of the 2-D
and 3-D interaction mechanisms as discussed earlier. Pullout tests were terminated
when a pullout displacement of 75 cm was attained, or when reinforcement breakage
occurred. A pullout displacement of 75 cm was chosen based on the experience that a
pullout displacement of 75 cm is generally found to be sufficient to reach, or exceed,
the peak pullout capacity of a soil-geogrid system.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

685

Pressure cell reading (kN/m )

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Applied normal pressure

n = 20 kN/m2
n = 50 kN/m2
Rubber membrane
C
L
and silicone grease

Side
wall

Distance from centerline (cm)

Figure 6. Verification of minimal side wall friction.

The pullout resistance was evaluated using the Japanese standard method (Hayashi
et al. 1994) as described below. The evaluation of the maximum pullout shear resistance
at an appropriate applied normal stress expressed in terms of interface shear stress is
given as:
Pmax =

F TE
2Le

(3)

where: FTE = maximum effective pullout force per unit width of specimen; and Le = effective/mobilized reinforcement length. The interaction parameters, cp and p , are determined using the Mohr-Coulomb strength equation:
Pmax = c p + n tan p

(4)

where: cp = apparent adhesion at the soil-reinforcement interface; and p = friction angle

at the soil-reinforcement interface as defined in Equation 2. The pullout test data analysis procedure as proposed by Hayashi et al. (1985), and illustrated in Figure 7, is used
in evaluating the above relationships. The pullout resistance in terms of interface shear
stress is assumed to be concentrated at the nodes (junctions) of the geogrid as depicted
in Figures 7a and 7b. The tensile force per unit width acting along the reinforcement
length (Figure 7f) is determined based on Figures 7c, 7d, and 7e. The concentrated shear
force at the geogrid nodes is back-calculated from the difference in adjacent tensile
forces as shown in Figure 7f. Using the measured nodal displacements (Figure 7c), the
average strain between the adjacent nodes (Figure 7d) is calculated by:

686

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

(a)
Ft

01 T 1 12 T 2 23 T 3 34 j-1 j T j j j+1

Ft

(b)

T2

T3

Tj

T1

(c)
Nodal displacement

Xi

X1
1

01

j-1 j

12
1

(e)

X3

2
3
Node number

(d)
ij

Xj

X2

23

j j+1

34

Node number
Load-strain
relationship

FTij
FT

FT

ij

(f)
T1
Ti , Fij

T2

FT F01

F12
1

F23

Tensile force/pullout force

along the reinforcement length
T3
F34

2
3
Node number

Figure 7. Procedure for geogrid pullout test data analysis (after Hayashi et al. 1985): (a)
mobilized resistance forces at geogrid junctions, ribs and longitudinal elements; (b) geogrid
junction forces (Tj ); (c) measured nodal displacements; (d) corresponding strain () between
geogrid junctions; (e) load-strain curve of geogrid; (f) tensile force per unit width acting along
geogrid length.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

687

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

ij =

xi xj

(5)

where: a = distance between two adjacent nodes; x = nodal displacement; and i, j = subscripts indicating the position of a node. The pullout force or tensile force between
nodes is then determined from the load-strain curve of the reinforcing material (Figure
7e). There is still no easy and accurate method of determining the load-strain behavior
of geogrid reinforcement in a confined environment, although there have been proposed methods for geotextile reinforcement. Therefore, the load-strain curve inferred
from an in-air extension test was employed.
An example determining the maximum effective pullout force per unit width, FTE ,
and the effective/mobilized reinforcement length, Le , is given in Figure 8 following the
method of constructing Figure 7f. When the reinforcement is pulled out without tension
breakage, Le is taken to be the total reinforcement length. Yet, when tension breakage
occurs, Le is reduced to a limiting reinforcement length. The limiting reinforcement
length, Ll , is a function of the strengths of the backfill soil and reinforcement as expressed in the following equation:
Ll =

Fu
2 (c + n tan )

(6)

where: Fu = ultimate strength of reinforcement per unit width; c = apparent soil cohesion; n = applied normal stress; and = internal friction angle of the soil. The value
of FTE is equal to the maximum applied pullout force per unit width minus the pullout

Pullout force (kN/m)

Effective/mobilized
length (Le )
FTmax

n = 30 kN/m2
B = 30 cm

Effective pullout
force
(FTE = FTmax --- FR )

FR

Distance from point of load application (cm)

Figure 8. Determination of maximum effective geogrid pullout force.

688

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

force, or tensile force per unit width generated at the end of the mobilized reinforcement
length (FTmax - FR ). The value of FR corresponds to the value of the horizontal projection
of the intersection between the vertical line drawn at the end of effective/mobilized reinforcement length and the fitted curve of the tensile force/pullout force along the reinforcement as shown in Figure 8. It can be seen that the determination of FTE implicitly
considers the value of Le .
4

TEST RESULTS AND DISCUSSIONS

4.1

Pullout Test Results and Analyses

Typical pullout test results are given in Figure 9 and depict the progressive movement
of nodal displacement along the reinforcement from the point of pullout load application. This is mainly due to reinforcement extensibility which consequently creates a
nonuniform mobilization of pullout resistance. Results from all test series are summarized in Table 1 together with the derived contributions of 2-D and 3-D interaction resistances. As mentioned earlier, the series of tests with specimen widths of B = 58 cm correspond to the 2-D interaction mechanism. Interaction parameters are given in Figure
10 and show that the measured interface strength envelope is slightly curved, but for
convenience, is represented by a straight line over the applicable stress range. To simplify the analysis, the apparent adhesion at the soil-reinforcement interface which is rel-

Applied pullout load (kN/m)

6
Node 5
Node 6

Front

6 @ 170 mm
Node 4 Node 3
Node 2

Node 1

Front

Legend:

n = 30 kN/m2
B = 30 cm

Front
Node 1
Node 2
Node 3
Node 4
Node 5
Node 6

Displacement (mm)
Figure 9. Typical geogrid pullout test results.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

689

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

atively small can be neglected (the results from all series of tests are plotted in Figure
11). The contribution of 2-D interaction resistance for narrower specimen widths (B =
45, 30, 20, 15, and 10 cm) was derived from test results on specimen width, B = 58 cm
based on specimen width proportions. The difference between the maximum effective
pullout force measured for narrower widths and their corresponding 2-D interaction resistance is considered to be the contribution of 3-D interaction resistance. This difference is represented by the second term in Equation 1 and is quantified by the product
Be n in Equation 2.
Table 1. Summary of geogrid pullout test results.

Reinforcement
width
(cm)
B

Applied
normal
stress
(kN/m2)
n

Max.
effective
pullout force/
unit width
(kN/m)
FTE

Mobilized
length
(m)
Le

Max.
effective
pullout
force
(kN)
PTE

2-D
effect
(kN)
P2-D

3-D
effect
(kN)
P3-D

Be n b

58.8

20

(58)a

34.5

1.17

20.28

20.28

30

49.90

1.17

29.34

29.34

40

57.50

1.17

33.81

33.81

Remarks

50

60.50

1.17

35.57

35.57

45.7

20

41.60

1.17

19.01

15.76

3.25

1.11

(45)

30

52.00

1.17

23.76

22.80

0.96

0.32

40

58.7

1.17

26.82

26.28

0.54

0.18

50

61.00

0.65

27.88

27.64

0.24

0.15

32.2

20

50.20

1.17

16.16

11.11

5.05

1.73

(30)

30

59.10

1.17

19.03

16.07

2.96

1.01

40

59.50

0.81

19.16

18.51

0.65

0.32

T.B.

50

61.13

0.65

19.68

19.48

0.20

0.12

T.B.

20.8

20

58.29

1.17

12.12

7.17

4.95

1.70

(20)

30

62.50

1.08

13.00

10.38

2.62

0.97

T.B.

40

59.00

0.81

12.27

11.96

0.31

0.15

T.B.

50

61.00

0.65

12.70

12.58

0.12

0.07

T.B.

15.5

20

61.20

1.17

9.49

5.35

4.14

1.41

(15)

30

62.40

1.08

9.67

7.73

1.94

0.71

T.B.

40

58.20

0.81

9.02

8.91

0.11

0.05

T.B.

T.B.

50

60.50

0.65

9.38

9.38

0.00

0.00

T.B.

11.5

20

65.70

1.17

7.55

3.97

3.58

1.22

T.B.

(10)

30

65.50

1.08

7.53

5.74

1.79

0.66

T.B.

40

58.50

0.81

6.73

6.61

0.12

0.05

T.B.

50

60.10

0.65

6.91

6.96

0.00

0.00

T.B.

Notes: a number in parentheses refers to the reinforcement width indicated in the graphs; b Be n = (P3-D ) /
(4Le tanp ), where p = 32_; and T.B. = tension breakage occurred (reinforcement rupture).

690

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

Shear stress, Pmax (kN/m2 )

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Geogrid reinforcement width, B = 58 cm

p = 32_

Coulomb envelope

Actual envelope
cp = 3.0 kN/m2
Normal stress, n (kN/m2)

Figure 10. Interaction parameters under 2-D conditions.

From Figures 11, 12 and 13 the following trends can be identified:

S The contribution of 3-D interaction mechanisms diminishes with increasing applied

normal stress. This further confirms that the 3-D interaction mechanism is a consequence of the restrained dilatancy effect. Dilatancy decreases with increasing applied
normal stress as can be seen in the normal displacement measurements given in Figure 12 for tests on specimens with width B = 58 cm under the applied normal stresses
of 20, 30 and 50 kN/m2. Test results under the applied normal stress of 40 kN/m2 were
omitted to minimize congestion of data points and to facilitate clarity of presentation.
Shear displacement was calculated using the average nodal displacement of adjacent
pairs of nodes and were calculated at the same locations where normal displacement
measurements were made (Figure 12). The locations of normal displacement measurements are given in the same figure. It should be noted that normal displacement
measurements were taken at the centerline of the specimen width and on specimens
having widths slightly smaller than the box width. Thus, the restraining effect does
not influence the measured normal displacements caused by the nondilating zone of
soil at both edges of the specimen.

S Specimen width has an influence on the pullout resistance particularly in the development of the 3-D interaction mechanism. A test specimen which has a width approaching the width of the pullout testing box (e.g. B = 45 cm) develops minimal 3-D interaction resistance. This is attributed to the minimal restraint provided by the dilating soil
which is partly due to the lubricated membrane on the side walls, and by a smaller
nondilating zone of soil at both edges of the specimen (Figure 13a). Minimal 3-D interaction resistance is also observed in specimens with widths smaller than 20 cm.
For this investigation, this is considered to occur because the extent of Be at both edges
of the specimen overlapped each other (Figure 13b) which could have reduced the
magnitude of the 3-D interaction resistance.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

691

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Effective pullout force (kN)

Effective pullout force (kN)

(b)

(a)
n = 20 kN/m2
P2-D
PTE

P3-D
P2-D

PTE

Reinforcement width (cm)

Reinforcement width (cm)

n = 40 kN/m2
P2-D
PTE

Effective pullout force (kN)

(d)
Effective pullout force (kN)

(c)

n = 30 kN/m2
P2-D
PTE

Reinforcement width (cm)

n = 50 kN/m2
P2-D
PTE

Reinforcement width (cm)

Figure 11. Pullout resistances for various geogrid specimen widths at different applied
normal stresses.
(Notes: PTE = maximum effective pullout force; P3-D = pullout force from 3-D interaction resistance; and
P2-D = pullout force from 2-D interaction resistance.)

The contribution of the 3-D interaction resistance, as quantified by the product

Be n , is plotted in Figure 14. This figure provides a sense of the reduction in rate of
increase of normal stress with increasing applied normal stress. The increase in normal
stress is observed to be very small under the applied normal stress of 50 kN/m2. In the
field this condition would mean that the effect of 3-D interaction resistance caused by
restrained soil dilatancy can be neglected for a strip reinforcement laid with an overbur-

692

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Location of normal displacement measurements

Reinforcement node
a
b

Normal displacement (mm)

15 cm
25 cm

6 5 4

3 2 1 Front

Test 1a
Test 1b
Test 2a
Test 2b

n = 20 kN/m2

n = 30 kN/m2

n = 50 kN/m2

Shear displacement (mm)

Figure 12. Measured dilatancy during geogrid pullout.

den embankment height of 2.5 m or higher. It can also be seen that the threshold value
of 2Be below which the 3-D interaction mechanism develops across the specimen width
can be taken as 20 cm. Having fixed the value of Be to equal 10 cm, the value of the
increase in normal stress, n , can be determined based on Figure 14. The increase in
normal stress was found to equal 17 kN/m2 and 10 kN/m2, or an increase of 85% and
33% for the applied normal stress of 20 kN/m2 and 30 kN/m2, respectively. It should
be noted that these values represent the average increase in normal stress at both edges
of the specimen along the effective/mobilized reinforcement length. The above procedure of determining the increase in normal stress is an alternative method to directly
measuring the increase in normal stress using earth pressure cells. Experience suggests
that earth pressure cells are difficult to install correctly and often provide unreliable
quantitative values. The calculated increase in normal stress, however, is qualitatively
reasonable in comparison with those determined in previous similar investigations.
Pullout tests on tie rods embedded in a sand embankment (Plumelle 1984) showed an
increase in normal stress of approximately 67% for an initial normal stress of 30 kN/m2
based on pressure cell readings located 20 cm from the tie rod. Matsui and San (1989)
conducted restrained direct shear tests on a soil-structure interface and complimented
these tests with a finite element analysis. These tests results indicated an increase in normal stress of 90% from an initial normal stress of 10 kN/m2.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

693

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

(a)

Lubricated side walls of pullout test box

Smaller nondilating
zone to restrain
dilatancy

Dilating zone
Shear stress is
eliminated by
lubricated
side walls

Reinforcement

2 mm thick rubber membrane

(b)

Silicone grease

B < 2Be
Be
Be
n
n

Strip reinforcement
Figure 13. Soil-reinforcement interaction: (a) geogrid specimen width slightly smaller
than the pullout testing box width; (b) narrow geogrid specimen width.

4.2

Practical Implications of the Test Results

The results of this investigation demonstrate that there is an enhanced pullout resistance due to the development of a 3-D interaction mechanism at both edges of a strip
reinforcement. This provides a technical, as well as, economical advantage for geosynthetic reinforcement which is not laid continuously along the width of a reinforced soil
structure (i.e. strip) as compared with those which are laid continuously (i.e. sheet). The

694

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Normal stress:
n = 20 kN/m2
n = 30 kN/m2

Influence shown
in Figure 13a

Be n

Influence shown
in Figure 13b

n = 40 kN/m2
n = 50 kN/m2

Reinforcement width, B (cm)

Figure 14. Variation of restrained dilatancy effect on different geogrid specimen widths.

enhanced pullout resistance of a strip reinforcement will develop, but only under low
normal stresses, and when dense granular soils are used as backfill materials.
When considering design stability, a limit equilibrium analysis is commonly
employed and requires that the maximum pullout resistance per unit width of the reinforcement be calculated. Most laboratory pullout test results on geogrid reinforcement
generally indicate an increasing pullout resistance per unit width with decreasing specimen width. This was also observed in this investigation (see Column 3 of Table 1). A
probable reason for this observed behavior is the combined 2-D and 3-D interaction
mechanisms on the strip reinforcement, whereby the contribution of 3-D interaction resistance is presumably imposed across the specimen width. This implies that the designed pullout resistance per unit width of reinforcement must be derived from pullout
tests in which the specimen width is equal to the width that will be used in the actual
construction. This condition proved to be difficult to satisfy because of the limited size
of the pullout testing box. Using the parameters contributing to the combined 2-D and
3-D interaction resistance, which can be quantified from an appropriate pullout test program and analysis, a more logical value of pullout resistance of strip reinforcement can
be evaluated. The parameters can be used in conjunction with Equation 2 to estimate
the pullout resistance of geogrid strip reinforcement. A separate evaluation of 2-D and
3-D interface resistances would also facilitate interpretation of pullout test results using
different specimen widths.

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

695

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Laboratory test arrangements should correspond to the field reinforcement layout because the specimen width relative to the box width influences the measured pullout resistance. If the field layout is a sheet type of reinforcement, the specimen width for laboratory tests should be slightly smaller than the width of the pullout testing box. The side
walls of the testing box should be free from wall friction by providing a lubricated rubber membrane. On the other hand, if the field layout is strip reinforcement, the development of the 3-D interaction mechanism should be accounted for in the laboratory test
arrangement.
5

SUMMARY AND CONCLUSIONS

A combined 2-D and 3-D pullout interaction mechanism has been developed for geogrid strip reinforcement embedded in dense granular soils. The methodology for evaluating the parameters involved in these mechanisms has been presented and discussed.
Results from pullout tests conducted in this investigation verified the existence of the
combined interaction mechanism. A three-dimensional (3-D) interaction mechanism
has been confirmed that is the result of confinement of the dilatating zone of soil around
the reinforcement.
For the type of geogrid reinforcement and backfill material used, the combined 2-D
and 3-D interaction mechanisms are applicable for a strip width, B, greater than 20 cm.
As the strip becomes narrow, such that B 20 cm, the 3-D interaction mechanism develops across the width of the strip. The calculated increase in normal stress at the edges
of the specimen due to restrained dilatancy was 85% and 33% for the applied normal
stresses of 20 kN/m2 and 30 kN/m2, respectively. The increase in normal stress was observed to be very small under the applied normal stress of 50 kN/m2. In a field condition
this would mean that the effect of a 3-D interaction mechanism caused by restrained
dilatancy can be neglected at an overburden embankment height of about 2.5 m or
greater.
The width of test specimen relative to the width of the pullout testing box has some
influence on the development of 3-D interaction resistance, and on the overall pullout
resistance. Awareness on the part of the designer to the test arrangement is therefore
important for proper application of interaction parameters.
REFERENCES
Alfaro, M.C., Miura, N. and Bergado, D.T. , 1995, Soil-Geogrid Reinforcement Interaction by Pullout and Direct Shear Tests, ASTM Geotechnical Testing Journal, Vol.
18, No. 2, pp. 157-167.
Hayashi, S., Makiuchi, K. Ochiai, H., Fukuoka, M. and Hirai, T., 1994, Testing Methods for Soil-Geosynthetic Frictional Behavior - Japanese Standard, Proceedings of
the 5th International Conference on Geotextiles, Geomembranes and Related Products, Singapore, Vol. 1, September 1994, pp. 411-414.

696

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

Hayashi, S., Yamanouchi, T., Ochiai, H. and Song, G., 1985, Mechanism of Pullout
Resistance of Polymer Grid in Soil, Journal of the Japanese Society of Soil Mechanics and Foundation Engineering (Tsuchi-to-kiso), Vol. 33, No. 5, pp. 21-26. (in Japanese)
Matsui, T. and San, K.C., 1989, An Elasto-Plastic Joint Element with Its Application
to Reinforced Slope Cutting, Soils and Foundations, Vol. 29, No. 3, pp. 95-104.
Plumelle, C., 1984, Improvement of the Bearing Capacity of Soil by Inserts of Group
and Reticulated Micropiles, Proceedings of the International Conference on In-Situ
and Rock Reinforcement, Paris, France, pp. 113-119.
Schlosser, F. and Elias, V., 1978, Friction in Reinforced Earth, Proceedings of the
ASCE Symposium on Earth Reinforcement, ASCE, Pittsburgh, PA, USA, April 1978,
pp. 735-763.
NOTATIONS
Basic SI units are given in parentheses.
B
Be

=
=

c
Cc

=
=

Cu
cp
d50
Fij
FR

=
=
=
=
=

FT
FTE
FTmax
FU
Le
Ll
P2-D
P3-D
PTE
Ti
xi

=
=
=
=
=
=
=
=
=
=
=

width of reinforcement specimen (m)

width along the edge of reinforcement influenced by restrained soil
dilatancy (m)
apparent soil cohesion (N/m2)
coefficient of curvature of soil particle size distribution curve
(dimensionless)
coefficient of uniformity of soil particle size distribution (dimensionless)
apparent adhesion at soil-reinforcement interface (N/m2)
average particle size (m)
pullout force or tensile force per unit width between nodes i and j (N/m)
pullout force per unit width generated at end of mobilized reinforcement
length (N/m)
applied pullout force per unit width (N/m)
maximum effective pullout force per unit width (N/m)
maximum applied pullout force per unit width (N/m)
ultimate strength of reinforcement per unit width (N/m)
effective/mobilized reinforcement length (m)
limiting reinforcement length (m)
pullout force from 2-D interaction resistance (N)
pullout force from 3-D interaction resistance (N)
maximum effective pullout force (N)
concentrated shear force per unit width at node i (N/m)
nodal displacement at node i (m)

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4

697

ALFARO, HAYASHI, MIURA AND WATANABE D Geogrid Strip Reinforcement Pullout

xj
n

=
=

p
ij

n
Pmax

=
=
=
=
=

698

nodal displacement at node j (m)

increase in normal stress at the soil-reinforcement interface on the width
of Be (N/m2)
friction angle at soil-reinforcement interface (_)
average strain between nodes i and j (dimensionless)
soil internal friction angle (_)
applied normal stress (N/m2)
maximum interface shear resistance (N/m2)

GEOSYNTHETICS INTERNATIONAL

S 1995, VOL. 2, NO. 4