LIMIT EQUILIBRIUM ANALYSES OF TWO-TIERED GEOSYNTHETICREINFORCED SOIL WALL MODELS IN A GEOTECHNICAL

CENTRIFUGE
Suliman B. A. Mohamed1, Kuo-Hsin Yang2, Wen-Yi Hung3

ABSTRACT
The FHWA guidelines limit the use of limit equilibrium (LE) method for
designing reinforced slopes (face inclinations less than 70o); however, this
limitation is somewhat arbitrary and there is no reason why LE method could not be
theoretically applicable in the design of reinforced walls and those with complex
geometry such as multi-tiered walls. This study therefore evaluated the use of LE
for predicting failure in centrifuge two-tiered geosynthetic-reinforced soil (GRS)
wall models with varying offset distances, D. The LE analysis results revealed good
agreement between LE and centrifuge models in locating failure surfaces. This
study also examined the effect of offset distance on the confined (or in-soil)
ultimate tensile strength of reinforcement, Tult. The confined Tult was backcalculated from the LE analyses at the failure of centrifuge wall models (i.e.,
FS=1.0). The analytical results showed that offset distance correlated negatively
with the effective overburden pressure on reinforcement and the resulting confined
Tult. In addition, this study evaluated the effect of offset distance on the normalized
reinforcement tension summation coefficient, KT, an equivalent earth pressure
coefficient derived from the confined Tult. The analysis results indicated that single
and independent wall models yielded a single consistent KT value, while in the
compound wall models the KT value decrease as the offset distance D increases. The
presented experimental results and discussions improve the understanding of limit
equilibrium analysis of multi-tiered reinforced walls.
KEYWORDS: Geosynthetic-reinforced soil wall, Limit equilibrium analysis,
Centrifuge model, Offset distance.
1

Ph.D. Candidate, Department of Construction Engineering, National Taiwan University of

Science and Technology, 43, Sec. 4, Keelung Rd., Taipei 106, Taiwan, Tel: +886981424577, Fax:+886-2-2737-6606, E-mail: kobrest@yahoo.com
2
Assistant Professor, Department of Construction Engineering, National Taiwan University
of Science and Technology, 43, Sec. 4, Keelung Rd., Taipei 106, Taiwan, Tel: +886-22730-1227, Fax:+886-2-2737-6606, E-mail: khy@mail.ntust.edu.tw
3
Associate Technologist, National Center for Research on Earthquake Engineering, 200,
Sec. 3, Xinhai Rd., Taipei 10668, Taiwan, Tel: +886-955329820, Fax:+886-3-425-2960,
email: wyhung@ncree.narl.org.tw

1. INTRODUCTION
Geosynthetic-reinforced soil (GRS) structures are applied in tiered
configurations for various reasons, including aesthetics, stability and construction
constraints. A lateral earth pressure method is widely used in the design of these
retaining structures. The earth pressure method for designing multi-tiered reinforced
walls is an extension of the design method for single-tiered reinforced walls. Some
researchers, however, have questioned the use of this empirical approach
(Leshchinsky and Han 2004). Very limited studies have confirmed the effectiveness
of earth pressure method for designing multi-tiered reinforced walls, and few have
investigated the behaviors and performance of GRS walls in tiered configuration.
Stuedlein et al. (2010) reported design and monitoring data for a four-tier 46m tall
wall reinforced with ribbed steel strips. Yoo and Jung (2004) and Yoo and Kim
(2008) investigated the performance and behavior of full-scale two-tiered GRS
walls. Reduced-scale wall models have been used to study the internal stability of
two-tiered GRS walls with varying offset distances (Yoo et al. 2011). Numerical
analyses have also been used to investigate the performance and stability of multitiered GRS structures under static (Leshchinsky and Han 2004; Yoo and Kim 2008;
Stuedlein et al. 2010; Yoo et al. 2011) and seismic loading conditions (Liu 2011).
Notably, current FHWA design guidelines (Berg et al. 2009) recommend lateral
earth pressure method for designing reinforced walls (e.g., single or multi-tiered
walls) and limit equilibrium (LE) method for designing reinforced slopes. Walls and
slopes are differentiated by a facing inclination of 70o. However, this limitation
based on face inclination is somewhat arbitrary, and there is no reason why LE
method could not be theoretically applicable in the design of reinforced walls. A
comparison of finite element and LE analyses by Leshchinsky and Han (2004)
concluded that LE analyses may be applicable for analyzing multi-tiered walls.
However, the LE predictions of performance at failure in multi-tiered GRS walls
have not been fully validated in physical models.
The above problems prompted the current LE analyses of centrifuge two-tiered
GRS wall models with various offset distance, D. The objective was to test the
assumptions and predictions of LE analysis to evaluate its validity for designing
multi-tiered GRS walls. To achieve this aim, experimental results and LE
predictions were compared specifically for failure surface locations. This study also
examined the influence of offset distance on the confined (or in-soil) ultimate
tensile strength of reinforcement, Tult. The confined Tult was back-calculated from
the LE analyses at the failure in centrifuge wall models (i.e., FS=1.0). This effect
can be interpreted as the effective overburden pressure on reinforcement, which
depends on the offset distance. Finally, normalized centrifuge test results and their
implications for the design of multi-tiered GRS walls are discussed.
2. CENTRIFUGE TESTS AND LIMIT EQUILIBRIUM ANALYSES
2.1 Centrifuge Tests
Hung (2008) performed centrifuge tests to investigate performance and failure
mechanisms in single and two-tiered GRS walls. To perform the LE analyses in the
current study, eight centrifuge model tests were selected from the Hungs centrifuge

testing program. Centrifuge models were constructed in a rigid aluminum container

with internal dimensions of 820 mm 450 mm in plan 580 mm in height. For all
models, the wall heights of the upper and lower tiers were H1 = H2 = 160 mm, and
additional layer of 20 mm of soil was deposited on the top of the upper tier to cover
the topmost reinforcement layer. Therefore, the wall models have an equivalent
height of 340 and were built on a foundation layer 150 mm thick. Figure 1 shows a
profile view of model walls. Each model was built using the same number of
reinforcement layers: 9 for the upper tier and 8 for the lower with 20 mm vertical
spacing. The reinforcement layers, which were folded back at the face of the wall
models, formed a wrap-around facing and a secondary (overlapping) layer (Lo=
40% of reinforcement length for each tier).
In all centrifuge models, offset distance D ranged from 0 mm to 270 mm. Table
I summarizes the geometrical configurations and test results for the two-tier GRS
wall models. The wall models were grouped into three test series (S, C, and I) in
accordance with FHWA design guidelines (2009):
(1) S-series: two wall models were single wall designs with D(H1+H2)/20 and
L1=L2 = 0.7(H1+H2).
(2) C-series: four wall models were compound wall designs with
(H1+H2)/20<DH2tan(90o- ), L2=0.6(H1+H2), and L1=0.6(H1+H2). Notably,
to prevent pullout failure during testing, the selected L1 value slightly
differed from the L1=0.7H1 recommended in the FHWA design guidelines
(2009).
(3) I-series: two wall models were independent wall designs with D>H2 tan(90o), L2 = 0.7H2, and L1 = 0.7H1.
In the centrifuge tests, all models were loaded by gradually increasing the glevel until failure. Table I gives the failure g-level, Nf, recorded for each model.
Figure 2 shows the initial and failure conditions observed in Tests S2, C5, and I8.
The centrifuge testing program is discussed in further detail in Hung (2008).

Figure 1. Schematic profile view of a centrifuge two-tiered GRS wall model (D is the offset
distance; L1 and L2 are the reinforcement length of the upper and lower tier; Lo is
the overlap length of reinforcement)

TABLE I. GEOMETRICAL CONFIGURATIONS AND TEST RESULTS OF TWO-TIERED GRS

WALL MODELS

Wall parameters
Test No.

S-series

C-series

I-series

1
2
3
4
5
6
7
8

Reinforcement
length

Results

D
(mm)

(deg.)

(deg.)

L1
(mm)

L2
(mm)

Nf
(g)

0
10
30
40
50
70
250
270

59.0
57.8
55.1
53.6
52.1
48.9
59.0
59.0

90.0
88.2
84.7
82.9
81.1
77.7
90.0a
90.0a

224
224
192
192
192
192
112
112

224
224
192
192
192
192
112
112

16
18
18
16
18
19
19/18b
18/18b

Back-calculated

Tult

(kN/m)
0.115
0.123
0.112
0.092
0.098
0.092
0.079/0.066b
0.076/0.066b

Note: D =Offset distance; =Failure plane angle in FHWA design guidelines; =Equivalent wall
inclination in FHWA design guidelines; L1 =Reinforcement length of upper tier; L2 =Reinforcement
length of lower tier; Nf =Failure g-level of centrifuge model; Tult =Back-calculated ultimate tensile
strength of reinforcement; aI-series is analyzed as two vertical single walls separately; bResults for
upper / lower tier

(a)
Initial condition

Failure

Initial condition

Failure

Initial condition

Failure

(b)

(c)

Figure 2. Photos of centrifuge tests: (a) Test S2; (b) Test C5; (c) Test I8

2.2 Material Properties

The soil used in the centrifuge test was clean and uniform Fulung beach sand,
which is classified as poorly graded sand (SP) in the Unified Soil Classification

System. The effective size D10, uniformity coefficient Cu, and coefficient of
curvature Cc for the sand are 0.17mm, 1.78, and 1.05, respectively. The sand was
pluviated from a hopper to achieve a uniform and dense state. The backfill unit
weight of sand and the friction angle obtained in a series of triaxial compression
tests at the target relative density of Dr = 70% were = 15 kN/m3 and tx = 39.5o,
respectively. As reported by Lade and Lee (1976), the plane strain peak friction
angle (ps = 42.3o) was estimated by finding the correlation between the triaxial
compression friction angle and the plane strain friction angle:

ps 1.5tx 17

(1)

A nonwoven geotextile was used in the centrifuge study. The average

unconfined tensile strengths measured in the standard wide-width tests (ASTM
D4595) and zero-span tests (with clamps 6 mm apart), were 0.05 and 0.12 kN/m,
respectively. As the nonwoven geotextile tensile strengths were found to be affected
by soil confinement and impregnation of geotextile by soil particles (Boyle et al.
1996), both wide-with and zero-span tensile tests may not accurately represent insoil tensile strength values. However, experimentally quantifying the in-soil
mechanical properties of low-strength nonwoven fabrics is difficult. This study
therefore performed a back analysis to calculate the confined ultimate tensile
strength of reinforcement, Tult. Table I summarizes the confined Tult values. The
highest value of confined Tult was obtained in a single wall because of the small
offset distance, which induced a high overburden pressure on the reinforcement.
Independent walls had the lowest values because the large offset distance decreased
the overburden pressure on the reinforcement. The unconfined tensile strength
values obtained in wide-width and zero-span tests represent the potential range of
confined Tult values. The back analyses used to obtain the values for confined Tult
and to evaluate the influence of offset distance on the confined Tult are discussed
further in sections 2.3 and 3.2 below.
2.3 Limit Equilibrium Analyses
Limit equilibrium analyses are typically used to analyze the stability of natural
and reinforced slopes. A series of centrifuge tests of GRS slopes by Zornberg et al.
(1998) showed that LE is effective for predicting failure in GRS slopes. In the
current study, LE analyses were performed to predict the locations of failure
surfaces in two-tier GRS wall models and to assess the confined ultimate tensile
strength of reinforcement. The LE calculations were performed using Spencers
method with circular and noncircular critical failure surfaces as coded in the Slide
v.6.0 program. This method, which is sufficiently rigorous to satisfy all equilibrium
conditions, assumes that all inter slice forces are parallel. The shear strength of the
test sand in the centrifuge model was characterized by the plane strain friction angle.
Centrifugal force was simulated by increasing the unit weight of backfill Nf times
until it corresponded to the target g-level at failure. One critical failure surface was
identified for each of the single and compound wall models while two critical
failure surfaces were identified separately for each tier wall of the independent wall
models.
The limit equilibrium analyses in this study assumed a uniform distribution of

reinforcement tensile forces with depth, horizontal orientation of reinforcement

forces, and overlapping geotextile layers were modeled as additional reinforcement
that increased stability in the wall. The tensile strength of reinforcement used as
input in LE analysis was adjusted until a factor of safety was reached FS =1.0 in
each centrifuge test model. The estimate accounted for the confined ultimate tensile
strength of the reinforcement and was expected to equal the average in-soil
reinforcement tension at the moment of failure. Finally, reduction factors such as
creep, installation damage and degradation were excluded because the centrifuge
model tests were meticulously constructed to ensure that no installation damage
occurred. The test duration was also kept sufficiently short to avoid long-term
behavior such as creep or degradation
3. RESULTS AND DISCUSSIONS
3.1 Comparison between Predicted and Experimental Location of Failure
Surfaces
Figure 3 compares the locations of failure surfaces obtained experimentally
from centrifuge tests and the locations of critical failure surface predicted by LE
analyses. The failure surface of the centrifuge wall model was identified by the
tears (ruptures) observed in each reinforcement layer. The tears are indicated by the
black triangles in Fig. 3. The failure surface in LE analysis was identified by
searching for both circular and noncircular critical surfaces. Figure 3 shows the
comparison results, which indicate the very good agreement between the critical
noncircular failure surfaces predicted by the LE analyses and those actually
observed in the experiments. Notably, however, the predicted critical circular
failure surfaces slightly differ from the actual failure surfaces. Figure 3 also
compares between experimental failure surfaces and the maximum tension lines
developed by FHWA design guidelines for a given triaxial compression friction
angle. In most cases, the comparative results show that the maximum tension lines
in the FHWA design guidelines depict the failure surfaces with a longer distance
from the wall face, in particular for the upper part of the upper tier. Therefore, using
the maximum tension lines recommended in the FHWA design guidelines would
result in overestimating the required reinforcement embedment lengths against
pullout.
Overall, the LE analysis with a noncircular failure surface is superior to the two
other methods. It can be concluded that LE approach with a noncircular failure
surface is competent of accurately predicting the failure surfaces of multi-tiered
walls. These comparison results also support the modeling assumptions (e.g., use of
noncircular failure surface, a uniform distribution of reinforcement forces with
depth, and horizontal orientation of reinforcement forces). Finally, in the
geosynthetic reinforced soil structures, the reinforcement length that extends
beyond the failure surface should provide sufficient pullout resistance. Therefore,
accurate identification of the failure surface location by the LE analysis confirms
that LE is effective for evaluating whether the internal stability of these structures is
sufficient to withstand reinforcement pullout.

Figure 3. Predicted and measured locations of failure surfaces from centrifuge tests:
(a) Test S1; (b) Test S2; (c) Test C4; (d) Test C5; (e) Test I8

3.2 Effect of Offset Distance on Confined Tensile Strength

Ultimate tensile strength of reinforcement is the most important parameter when
designing reinforced structures to resist reinforcement rupture. As noted above, the
ultimate tensile strength measured in the centrifuge test may differ from that

measured in the standard unconfined tensile test due to soil confinement and
impregnation of the geotextile by soil particles. One alternative is to evaluate in-soil
geotextile strength by back-calculation from the centrifuge model results at failure.
Table I summarizes the back-calculated values of confined Tult. Since overburden
pressure can affect the confined Tult, the Grey elastic solution modified by Wright
(2005) (Eq. 2) was used to calculate additional vertical stress v from upper tier
wall. Afterward, effective overburden pressure v on a given reinforcement layer
can be obtained (i.e., overburden pressure z from lower tier wall plus additional
vertical stress v from upper tier wall).

q
(x D )z
a

R a2

(x D )z
i
R i2

(2)

where v is the additional vertical stress from the upper tier, q is the equivalent
uniform surcharge from the upper tier wall, x is the horizontal distance from the
wall face, D is the offset distance, z is the depth below the top of lower wall, and a,
i, Ra and Ri are indicated in Figure 4.
D

q
x

z
Ra

Ri

i
v

Flexible Wall

H2

Reinforcement

Figure 4. Effective overburden pressure behind a flexible wall

Figure 5 illustrates the effect of offset distance on the confined Tult and the
corresponding values of average v along the reinforcement at the bottom layer of
the lower tier wall for all tests. Figure 5 clearly shows that confined tensile strength
decreases as offset distance increases. These analytical results imply that, as the
offset distance increases, the effective overburden pressure on reinforcement
decreased, which then decreases the confined Tult. In Fig. 5, the intersection of two
trend lines indicates the critical offset distance, Dcr, which is the offset distance
beyond which the two tiers act independently. The critical offset distance of Dcr =
0.7H2 observed in this study agrees with the Dcr = 0.8H2 reported previously
(Leshchinsky and Han 2004, Yoo et al. 2011). However, these Dcr values are much
smaller than the critical offset distance criterion (Dcr = H2 tan(90o- o) = 1.2H2)
recommended in the FHWA design guidelines. Therefore, use of the Dcr value
given in the FHWA design guidelines would likely result in a conservative design.
It can be concluded that the limit equilibrium results are consistent with the
literature, the data in this study support the use of critical offset distance of 0.8H2 in

two-tiered walls, which is much less than the value given in the FHWA design
guidelines.

Figure 5. Effect of offset distance on confined ultimate tensile strength for all the tests

4. NORMALIZATION OF REINFORCEMENT TENSION SUMMATIONS

In the case of multi-tiered walls, equilibrium between reinforcement forces and

horizontal soil stresses along the potential failure surface is essential. This
equilibrium relationship can be further characterized by an equivalent earth pressure
coefficient, KT, derived from the confined Tult, defined as
2 nTult
K T ,
H 2

1
.
Nf

(3)

where KT is the equivalent earth pressure coefficient, so known as the normalized

reinforcement tension summation coefficient at the moment of failure (Zornberg et
al. 1998); = unit weight of soil; H = total height of tiered walls, n = number of
broken reinforcement layers (both primary and overlaps); Tult = confined ultimate
tensile strength of reinforcement back-calculated from the LE analysis; Nf = failure
g-level. This coefficient KT depends on the backfill friction angle and the
configuration of multi-tiered wall system (i.e., number of tiers and offset distance
D) which can be represented as an equivalent wall inclination
Figure 6 shows the normalized centrifuge test results. Figure 6 shows the linear
relationship established for centrifuge models of a single and independent wall
(with the same and). A single value of KT = 0.19 is obtained. This KT value is
consistent with the value obtained from the active earth pressure coefficient Ka for
vertical walls (i.e., Ka = 0.195). Figure 6 also gives the normalization results for
centrifuge models of a compound wall (with the same but different ). The KT

value (the slope of the dashed line) decreases as wall offset D increases (or
equivalent wall inclination decreases). These KT values are further compared with
those calculated using the Coulomb active earth pressure equation by inputting the
correspondingvalues. The comparison results in Table II indicate that the KT
values from compound wall models are only slightly lower than those calculated by
the Coulomb equation for different D or values. The overall good agreement
between the KT values and the Coulomb active earth pressure coefficients confirms
that the predicted confined Tult results back-calculated from the limit equilibrium
analyses are reasonable and can be normalized as discussed in this section.

Normalized reinfcement tension summation,

2nTult / H2

I-series
S-series
4

C-series

Ka = 0.195

Ka

KT

KT = 0.19
= 90o

D = 30 mm, = 84.7o
D = 40 mm, = 82.9o
D = 50 mm, = 81.1o

D = 70 mm, = 77.7o

0
0

10

15

20

25

G-level at failure, Nf

Figure 6. Normalized reinforcement tension summation values from centrifuge test results
TABLE II. COMPARISON OF KTVALUES FOR COMPOUND WALL MODELS

Test No.
C-series

3
4
5
6

D
(mm)
30
40
50
70

(deg.)
84.7
82.9
81.1
77.7

Normalized centrifuge test results

KT
0.15
0.14
0.13
0.12

Coulomb
Ka
0.16
0.15
0.14
0.13

5. CONCLUSIONS

A series of LE analyses was presented for centrifuge modeling of two-tiered

GRS walls. This study evaluated the use of limit equilibrium analyses for predicting
failure in two-tiered geosynthetic reinforced soil wall models. The conclusions of
this study are summarized below:

The location of the critical noncircular failure surface predicted by the limit
equilibrium analysis agrees well with the actual location of the critical
failure surface obtained experimentally.
The centrifuge test results provide insight into the evaluation of the in-soil
tensile strength of geotextiles. The observed influence of offset distance, D,
on the ultimate confined tensile strength, Tult, suggests that an increased
offset distance reduces the effective overburden pressure and thus decreases
the confined Tult.
The critical offset distance identified in this study (Dcr = 0.7H2) agrees with
the Dcr = 0.8H2 reported in the literature. Additionally, these Dcr values are
much lower than the critical offset distance specified in the FHWA design
guidelines.
The normalization of the centrifuge test results obtains a consistent
reinforcement tension summation coefficient KT for single and independent
walls (with the same friction angle and equivalent wall inclination). In
compound wall models (with the same but different), however, the KT
values decrease as the offset distance D increases. The KT values also agreed
well with the Coulomb active earth pressure coefficients with varying .

REFERENCES
1. ASTM D4595. Standard test method for tensile properties of geotextiles by the wide-width strip
method. American Society for Testing and Materials. West Conshohocken, PA, USA.
2. Berg, R., Christopher, B. R., and Samtani, N. (2009). Design of mechanically stabilized earth
walls and reinforced soil slopes. Vol. I and II. Report No. FHWA-NHI-10-024, Federal
Highway Administration.
3. Boyle, S. R., Gallagher, M. and Holtz, R. D. (1996). Influence of strain sate, specimen length
and confinement in measured geotextile properties, Geosynthetics International, 3(2), 205-225.
4. Hung, W.Y. (2008). Breaking failure behavior and internal stability analysis of geosynthetic
reinforced earth walls, Ph. D. Dissertation, National Central University, Jhongli, Taiwan.
5. Lade, P.V., and Lee, K.L. (1976). Engineering properties of soils. Report UCLA-ENG-7652,
University of California, Los Angeles, Calif.
6. Leshchinsky, D. and Han, J. (2004). Geosynthetic reinforced multitiered walls. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 130(12), 1225-1235.
7. Liu, H. (2011). Comparing the seismic responses of single- and multi-tiered geosynthetic
reinforced soil walls. Geo-Frontiers 2011, ASCE, 211, 3478-3486.
8. Porbaha, A. and Goodings, D. J. (1996). Centrifuge modeling of geotextile reinforced cohesive
soil retaining walls. Journal of Geotechnical Engineering, ASCE, 122(10), 840848.
9. Stuedlein, A.W., M. Bailey, D. Lindquist, J. Sankey, and Neely, W. J. (2010). Design and
performance of a 46-m-high MSE wall. Journal of Geotechnical and Geoenvironmental
Engineering, 136(6), 786796.
10. Wright, S. (2005). Design procedures for multi-tiered mechanically stabilized earth walls. GeoFrontiers 2005, ASCE, Austin, Texas.
11. Yoo, C., Jang, Y. S. and Park, I. J. (2011). Internal stability of geosynthetic-reinforced soil walls
in tiered configuration. Geosynthetics International, 18(2), 7483.
12. Yoo, C. and Jung, H. S. (2004). Measured behavior of a geosynthetic reinforced segmental
retaining wall in a tiered configuration. Geotextiles and Geomembranes, 22(5), 359-376.
13. Yoo, C. and Kim, S. B. (2008). Performance of a two-tier geosynthetic reinforced segmental
retaining wall under a surcharge load: full-scale load test and 3D finite element analysis.
Geotextiles and Geomembranes, 26(6), 447518.
14. Zornberg, J. G., Sitar, N. and Mitchell, J. K. (1998). Limit equilibrium as basis for design of
geosynthetic reinforced slopes. Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 124(8), 684698.