International Conference on Civil Engineering

Architecture & Urban Sustainable Development

18 &19 December 2013, Tabriz , Iran

Comparison of current method for internal stability

design of Mechanical Stabilized Earth walls
Mahboobeh aboutalebi *, Hosein ghiasian

1. M sc, graduated in science & technology university of Tehran,mitra_abutalebi@yahoo.com

2. Associate Professor, school of civil engineering of Iran University of Science and Technology,
H_Ghiassian@iust.ac

Abstract
This paper describes a working stress design methodology for steel reinforced soil walls
(K-Stiffness Method) it includes the effects of global wall stiffness, soil strength,
Reinforcement layer spacing, and wall height in design. It also discusses a comparison of
the method to other methods found in US design codes and guidelines for internal
stability design (e.g. AASHTO Simplified Method, Coherent Gravity Method and FHWA
structure Stiffness Method). All these three methods use limit equilibrium concepts to
develop the design model. It also introduces a large database of full-scale steel reinforced
soil and some general information about them. These walls was used to investigate the
accuracy and shortcomings of the various reinforcement load prediction methods,
reinforcement load and other measurements can be compared to predictions. The primary
focus of this paper is on steel reinforced MSE walls with granular backfills.

Key words: K-stiffness method, limit equilibrium, reinforcement load, MSE walls

1. Introduction
Accurate prediction of reinforcement loads and their distribution within the backfill is
necessary to produce cost effective, internally stable reinforced soil wall designs. The
predicted reinforcement loads are used in design to evaluate the reinforcement strength and
spacing requirements as well as the length required to resist pullout.
The three primary methods existing in design codes and guidelines included the Coherent
Gravity Method [1], the Federal Highway Administration (FHWA) Structure Stiffness
Method [2], and the Simplified Method. The differences in the predictions from these
methods are the result of both differences in the case studies used to develop each method and
differences in the assumptions for each method. All three methods also use limit equilibrium
concepts to develop the design model and working stress observations to adjust the models to
fit what has been observed in full-scale structures. Small-scale gravity and centrifuge models
taken to failure have been used to evaluate design models at true limit equilibrium conditions.

1
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

This approach has worked reasonably well for typical steel reinforced soil walls. The FHWA
Structure Stiffness and Simplified Method provide different empirically derived envelopes for
different reinforcement types in an attempt to broaden their applicability to a wider range of
reinforcement types. This paper describes a working stress design methodology for steel
reinforced soil walls (K-Stiffness Method) [3] that it includes the effects of global wall
stiffness, soil strength, reinforcement layer spacing, and wall height in design. It also
introduces a large database of full-scale steel reinforced soil and some general information
about them. These walls were used to investigate the accuracy and shortcomings of the k-
stiffness method and other various reinforcement load prediction methods. Reinforcement
load and other measurements can be compared to predictions.

2. Calculation of Tmax for steel reinforced wall systems by K-stiffness

method
This new methodology considers, directly or indirectly, the stiffness of all wall components
relative to the soil stiffness to estimate the distribution and magnitude of Tmax. As such, it uses
working stress principles to estimate the load and strain in the reinforcement. However, the
method is empirical in nature, since it was calibrated to accurately predict the reinforcement
loads in nine full-scale field geosynthetic wall cases (13 different wall sections and surcharge
conditions with 58 individual data points) and 19 full-scale field steel reinforced MSE wall
cases (24 different wall sections and surcharge conditions with 102 individual data points).
The K-Stiffness Method considers the following variables:
1. For wall geometry, H (the total height of the wall), S (the average surcharge height
above the wall), Sv (tributary area, equivalent to the average vertical spacing of the
reinforcement near each layer location when analyses are carried out per unit length of
wall), and Φfb (a factor to account for the effect of wall face batter).
2. For reinforcement properties, Slocal (the local reinforcement stiffness), and Sglobal
(the global wall stiffness)
3. For facing stiffness, Φfs (facing stiffness factor)
4. For soil properties, γ (the backfill soil unit weight), and K0 (the at-rest earth pressure
coefficient based on the peak plane strain soil friction angle).
These factors are introduced analytically in the following general expression for the
maximum load per running unit length of walls in reinforcement layer i:

2
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Tmax  S i v . h .Dt max  (1)

Where S vi  tributary area (equivalent to the vertical spacing of the reinforcement near each
layer location when analyses are carried out per unit length of wall);  h = lateral earth
pressure acting over the tributary area; Dtmax = load distribution factor that modifies the
reinforcement load on the basis of layer location; and  = influence factor that account for the
influence of local and global reinforcement stiffness, facing stiffness, and face batter.
The lateral earth pressure is calculated as the average value acting over the height of the wall
according to conventional earth pressure theory, hence:

 h  K H  S 
1
(2)
2
Where K = lateral earth pressure coefficient; γ = unit weight of the soil; H = height of the
wall; and S=equivalent height of uniform surcharge pressure q (i.e., S = q/γ). This is
equivalent to taking the total lateral earth force acting on the wall and dividing it by the height
of the wall. For reinforced soil structures, conventional practice is to assume that the lateral
earth pressure acting at any level in the wall is directly proportional to the overburden stress.
The coefficient of lateral earth pressure, K, is calculated by using the Jaky equation for “at
rest” earth pressure [4]:
K 0 1  Sin ps and K 0  0.3 (3)

Where,  ps is the peak plane strain friction angle .

Substitution of equations 2 and 3 into equation 1 leads to:

Tmax  0.5Svi K0 ( H  S ) Dt max 
(4)
Equation 4 contains the conventional expression for calculating reinforcement loads in current
limit equilibrium methods of analysis. The modifiers Dtmax and Φ are empirically determined
parameters, or functions, that reflect actual mechanisms. They are used to improve the
correspondence between measured and predicted reinforcement loads at working stress
conditions on the basis of a large number of case studies. Parameter Dtmax is a load
distribution factor that modifies the reinforcement load as a function of normalized depth
below the top of the wall (z+S)/(H+S), and it varies over the range 0≤Dtmax≤1. For brevity in
equation 4 the influence factor, Φ, is used to represent the product of four factors as follows:
   local   fb   fs   g
(5)
Where Parameter Φg is a global stiffness factor that accounts for the influence of the stiffness
and spacing of the reinforcement layers over the entire wall height. It has the following
general form

3
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran


 S global 
 g    
 p a  (6)
Where Sglobal = global reinforcement stiffness and  and  are constant coefficients (for steel
strip walls,     0.25 ). The non-dimensionality of the expression is preserved by dividing
the global reinforcement stiffness by pa=101 KPa (atmospheric pressure). The global
reinforcement stiffness value for a wall is calculated as follows:
n

J ave  Ji
S global   i 1

H / n  H (7)
Where Jave = average tensile stiffness of all “n” reinforcement layers over the wall height and
Ji = tensile stiffness of an individual reinforcement layer expressed in units of force per unit
width of wall.
Parameter Φlocal is a local stiffness factor that accounts for relative stiffness of the
reinforcement layer with respect to the average stiffness of all reinforcement layers. It is
expressed as follows:
a
S 
 local   local 
S 
 global  (8)
The coefficient term “a” is taken as a = 0 for steel reinforcement and a = 1 for geosynthetic
reinforced soil walls. Slocal is the local reinforcement stiffness for reinforcement layer i, equal
to J/Sv where J is the reinforcement stiffness layer i and Sv is used to quantify the local
combined influence of the individual layer stiffness and spacing on reinforcement load.
The high global stiffness of steel reinforced walls minimizes the influence of the stiffness of
the facing and wall toe on reinforcement loads. So the facing stiffness factor for all steel
reinforced soil walls is recommended to be  fs   1 .

The influence of wall facing batter on maximum reinforcement loads is adjusted in the
proposed working stress method by using an empirical facing batter factor  fb  expressed as:
d
K 
 fb   abh 
 K avh  (9)
Where Kabh is the horizontal component of the active earth pressure coefficient accounting for
wall face batter, and Kavh is the horizontal component of active earth pressure coefficient,
assuming the wall is vertical and d is a constant coefficient (d=0.25 is reasonable for steel
reinforced soil walls [5]). The form of the equation shows that as the wall face batter angle ω
→ 0 (i.e., the wall facing batter approaches the vertical), the facing batter factor  fb  1 .

4
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Parameter Dt max is load distribution factor. To generate a normalized reinforcement load

distribution envelope for the steel reinforced soil walls Allen et al [5] performed empirical
approach. The results are plotted in Figure 1.

Figure 1: Normalized distribution of measured values of Tmax: (a) geogrid and geotextile
reinforced soil walls; (b) steel strip reinforced soil walls

3. Summary of case histories evaluated

The key properties and parameters for each of the case histories referenced in this paper are
summarized in table1. Additional details for each of these case histories, including wall type,
reinforcement geometry, reinforcement type, soil properties, and construction history, are
provided by Allen[6].

Table 1: Summary of steel strip reinforced soil wall case histories (from Allen et al. 2001)
Case Case Wall Backfill Backfill Backfill Surcharge Ka Typical Sh Reinforcement Global Wall Reference
No. description height φtx φps γ condition Sv stiffness J Stiffness
and date (m) (degree) (degree) (kN/m³) (m) (m) (MN/m) Sglobal
built (MN/m2)
Lille,
France steel Bastick
6 44 48 18.1 None 0.18 0.75 0.5 48 64
strip wall, 1984
SS1 1972
UCLA steel
Richardson
strip test 6.1 38 40 19.8 None 0.24 0.76 0.76 63 104
et al. 1977
SS2 wall, 1974
WES steel Al-Hussaini
Uniform
strip test 3.66 36 40 18.5 0.26 0.61 0.77 18 30 and Perry
(90 kPa)
SS3 wall, 1976 1978
Fremersdrof
Thamm
steel strip 7.3 37 40 19.6 None 0.25 0.76 0.76 79 103
1981
SS4 wall, 1980
Guildford
bypass steel Hollinghurst
strip wall, 6 48 53 22.3 None 0.15 0.3 0.9 83 264 and Murray
sections 1986
SS6 A&B, 1981

5
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

0.75 for
Asahigaoka, top 10
Sloping
Japan steel layers, Bastick
12 36 40 17.7 ≤1 m 0.26 0.75 80-121 128
strip MSE and 0.5 for 1984
high
wall, 1982 bottom 6
SS7 layers
0.76 for
top 12
Ngauranga 5 degrees
layers,
steel strip 12.6 50 50 a 21.5 negative 0.13 0.76 79-118 122 Boyd 1993
and 0.51
wall,1985 slope
for bottom
SS10 5 layers
Algonquin
Christopher
steel strip 6.1 40 43 20.4 None 0.22 0.76 0.73 55 72
1993
SS11 wall,1988
Gjovik
1.5:1
(Norway) Vaslestad
12 38 40 19 slope ≤3 0.24 0.76 0.76 53 70
steel strip 1993
m high
SS12 wall, 1990
0.76 for
Bourron top 10
Marlotte layers,0.61
Steel Strip for 11th Bastick et
10.5 37 40 16.8 None 0.25 0.76 79-118 137
Rectangular layer, al. 1993
Test wall , and 0.51
1993 for bottom
SS13 3 layers
0.76 for
top 5
Bourron
layers,0.61
Marlotte
for 6th
Steel Strip Bastick et
10.5 37 40 16.8 None 0.25 0.76 layer, 79-118 118
trapezoidal al. 1993
and 0.51
Test wall ,
for
1994
bottom8
SS14 layers
Note: SS5 steel strip reinforced soil wall. The stiffness J was determined assuming a steel modulus of 200 GPa, with the exception
of SS3 which used the value reported in the reference. The stiffness was computed as force per unit width of wall, based on
geometry and horizontal spacing of the reinforcement.
a -Estimated from in-soil pullout tests.

The database of steel reinforced soil wall case histories includes a total of 11 instrumented
wall sections that were built to full scale in the field (the same wall with and without a
surcharge was considered to be one case). These wall cases included a variety of wall
geometries and materials, surcharge conditions, and granular backfill. Wall reinforcement
products included only steel strip reinforcement. Reinforcement vertical spacing varied from
0.3 to 0.76 m and all of the walls were vertical. Wall heights varied from 3.66 m to 12.6 m,
with surcharge heights of up to 3.9 m of soil. Facing types included pre-cast concrete panels,
and modular concrete blocks. Estimated plane strain peak soil friction angles varied from 37o
to 54o.

6
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Note that the plane strain friction angle uses to characterize soil shear strength in this method,
while other methods use triaxial or direct shear friction angles. Plane strain conditions
typically exist in MSE walls, and recent work indicates that the plane strain soil friction angle
correlates best to reinforcement loads. Measured plane strain friction angles,  ps  , were not
available for these case histories, but were estimated from measured triaxial or direct shear
friction angles. Peak friction angles reported in the source references from triaxial
compression tests, tx  , were adjusted to peak plane strain friction angles using the equation
by Lade and Lee [7]
 ps  1.5tx  17  if tx  34 
 ps  tx  if tx  34 

4. Comparison of the prediction methods to measured behavior

1-4- General observations
Figures 2(a) to 2(h) show the predicted reinforcement loads as a function of depth below the
wall top for some of case histories. These were determined with the various prediction
methods, allowing direct comparison to the measured reinforcement loads.

7
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Figure 2: Predicted and measured reinforcement peak loads for steel strip reinforced wall (a)
ss3-no surcharge; (b) ss3-24 kPa surcharge; (c) ss3-48 kPa surcharge; (d) ss3-72 kPa surcharge;
(e) ss6; (f) ss7; (g) ss10 and (h) ss11

The measured triaxial or direct shear soil friction angle was used for three limit equilibrium
methods predictions rather than an estimated plane strain soil friction angle or a constant
volume friction angle, as current design specifications (AASHTO, 1999) refer to direct shear
or triaxial shear strength for use with these methods, while the plane strain friction angle uses
to characterize soil shear strength in K-stiffness method. Though there is a considerable
amount of scatter in the measured results relative to the predicted reinforcement loads, the
following general trends can be observed:
 All of the methods provide predictions that are close. In general, all of methods
assume that reinforcement stresses increase as a function of depth below the wall top,
but K-stiffness method assumes that after a specific height this increase became
constant, especially near the base of the wall. This assumption lead to prevent from
over predict in the lower half of the wall.
 In the cases with any surcharge loads, three limit equilibrium methods tend to
slightly underestimate the reinforcement loads, but K-stiffness method provides better
predictions than other methods and its profile intersect measure data in some points.
 When a significant soil surcharge is present, three limit equilibrium methods tend to
overestimate the reinforcement load especially in the lower half of the wall, but the
profile of K-stiffness method is approximately fitted to measured data.
If the measured reinforcement loads are significantly different than the predicted loads, all
methods tend to error on the same side relative to the measured loads.

2-4- Comparison of the limit equilibrium methods

8
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Table 2 and figures 3a to 3c provide an overall view of how well each limit equilibrium
method predicts reinforcement stresses for steel strip reinforcement, for all granular backfills.
A log-log scale has been used to better display the wide range of load levels in the walls. The
reinforcement load Tmax in the figures is the maximum load in the reinforcement layer. Table
2 summarizes a statistical analysis of the ratio of the predicted to measured loads for each
method for this wall reinforcement type. A normal distribution was assumed. This
information suggests that the Simplified Method provides the best prediction, on average, of
the reinforcement loads for steel strip reinforced walls, while the Coherent Gravity and
FHWA Structure Stiffness methods tend to underestimate the reinforcement loads, on
average. Although the FHWA Structure Stiffness Method appears to under-predict the
reinforcement loads for steel strip reinforced walls, it also has a lower coefficient of variation,
indicating a slightly tighter distribution of the data.

Figure 3: Predicted Tmax versus estimated Tmax from measured strains using (a) the AASHTO
Simplified Method; (b) Coherent Gravity method and (c) The FHWA method

9
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Table 2: Summary of the average and coefficient of variation for the ratio of the predicted to the
measured reinforcement loads, assuming a normal distribution, for each prediction method for
all granular backfill soils.
MSE Wall Ratio: Predicted/Measured Reinforcement Load
Reinforcement
Coherent Gravity FHWA Structure Simplified Method
Type
Method Stiffness Method
Average COV Average COV Average COV
Steel Strip 0.89 0.57 0.86 0.51 0.91 0.58
COV = coefficient of variation = (standard deviation of ratio of modulus values/mean of ratio of modulus values)

The Simplified Method does slightly under-predict the reinforcement loads for steel
reinforced MSE walls. A possible explanation is that during the original development of the
Simplified Method, peak friction angles were interpreted as peak triaxial or peak direct shear
soil friction angles and these values were not corrected for plane strain field conditions.
With these investigating, it concluded that the AASHTO Simplified Method produces results
similar to those of the other methods, yet it has the advantage of being simpler to use and
more broadly applicable. Therefore, the Simplified Method will be used herein as the baseline
of comparison for reinforcement loads predicted with this working stress method.

3-4- Comparison of the simplified method and K-stiffness method

Current AASHTO design specifications require a maximum friction angle of 40o even if the
measured friction angle is higher, so that specifications are consistent with the empiricism in
the Simplified Method for steel reinforced walls [6]. Therefore, to maintain consistency
between the Simplified Method and the K-Stiffness Method (which effectively caps the plane
strain soil friction angle at 44o), the triaxial or direct shear friction angle was capped at a
maximum value of 40o for steel reinforced walls.
Figure 4a and 4b indicate how well the Simplified Method and K-stiffness method predict
reinforcement loads in steel reinforced soil walls built in the field, respectively. A log-log
scale has been used to better display the wide range of load levels in the walls. The
reinforcement load Tmax in the figure is the maximum load in the reinforcement layer.

11
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

Figure 4: Predicted Tmax versus estimated Tmax from measured strains using (a) the AASHTO
Simplified Method and (b) K-Stiffness method

Table 3: Summary of the average and coefficient of variation for the ratio of the
predicted to the measured reinforcement loads, assuming a normal distribution, for K-
stiffness and simplified methods for all granular backfill soils.
Ratio: Predicted/Measured Reinforcement Load
MSE Wall
Reinforcement Simplified Method K-stiffness method
Type
Average COV Average COV
Steel Strip 1.16 0.51 1.11 0.36

The improvement in predicted loads versus measured values when the K-stiffness method is
used instead of the AASHTO Simplified Method is more efficient and reasonable. This is
demonstrated in figures 4a and 4b. The same conclusion is reached by examination of values
for the mean and spread (COV) of the ratio of predicted to the measured reinforcement loads
for the two methods in Table 3, the following remarks can be observed:.
 Even with the friction angle capped at 40o (triaxial or direct shear), or 44o (plane
strain), the Simplified Method does not appear to predict the loads estimated from
measured strains as well as the K-Stiffness Method.
 K-Stiffness Method matches the measured data more consistently than does the
AASHTO Simplified Method, as shown by the COV values in Table 3, although both

11
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

methods produce an average ratio of measured to predicted Tmax that is near 1.0,
provided that triaxial or direct shear soil friction angles are used with the Simplified
Method.

5. Acknowledgments
Below are the concluding remarks on this paper:

 All of prediction methods (limit equilibrium and working stress methods) capture the
general trend in reinforcement loads, which can be seen to increase with depth below
the top of the wall.
 The AASHTO Simplified Method produces results similar to those of the other limit
equilibrium methods, yet it has the advantage of being simpler to use and more
broadly applicable. Therefore, the Simplified Method will be used as the baseline of
comparison for reinforcement loads predicted with K-stiffness method.
 The K-Stiffness Method has been shown to improve the accuracy of predicted
reinforcement loads in comparison to the Simplified Method and provides a way to
estimate reinforcement loads on the basis of measured parameters.
 This methodology considers, directly or indirectly, the stiffness of all wall
components relative to the soil stiffness to estimate the distribution and magnitude of
Tmax. As such, it uses working stress principles to estimate the load and strain in the
reinforcement.
 K-Stiffness Method matches the measured data more consistently than does the
AASHTO Simplified Method, as shown by the COV values in Table 3, although both
methods produce an average ratio of measured to predicted Tmax that is near 1.0,
provided that triaxial or direct shear soil friction angles are used with the Simplified
Method.
 The K-Stiffness Method provides a more accurate estimate of reinforcement loads,
and its use can result in substantial cost saving.

References
[1] AASHTO. Standard Specifications for Highway Bridges, with 1999 Interims. Washington, D.C
:American Association of State Highway and Transportation Officials, Sixteenth Edition, 686p,1996
[2] Christopher, B. R., Gill, S. A., Giroud, J.-P., Juran, I., Mitchell, J. K., Schlosser, F., and Dunnicliff,
J. Reinforced soil structures, Vol. 1 Design and construction guidelines.” FHWA Rep. FHWA-RD-
89-043, 1990.

12
 International Conference on Civil Engineering
Architecture & Urban Sustainable Development
18 &19 December 2013, Tabriz , Iran

[3] Allen .T.M, Bathurst .R.J. Prediction of Reinforcement Loads in Reinforced Soil Walls.
Washington State Department of Transportation and in cooperation with Federal Highway
Administration, Report Number WA-RD 522.2, 2003.
[4] Holtz, R. D., and Kovacs, W. D. An introduction to geotechnical engineering, Prentice-Hall.
Englewood Cliffs, N.J, 1981.
[5] Allen .T.M, Bathurst .R.J, Holts .R.D, Lee .W.F and Walters .D. New Method for Prediction of
Loads in Steel Reinforced Soil Walls. Journal of Geotechnical and Geoenvironmental engineering,
ASCE, VOL. 130, issue 11, pp. 1109-1120,2004.
[6] Allen .T.M, Christopher .B, Elias .v and DiMaggio .J. Development of The Simplified Method for
internal stability design of Mechanically Stabilized Earth Walls. Washington State Department of
Transportation and in cooperation with Federal Highway Administration, Report Number WA-RD
513.1,2001.
[7] Lade, P. V., and Lee, K. L.Engineering properties of soils.” Rep. UCLA-ENG-7652 [as cited by
Holtz and Kovacs (1981)],1976.

13