On Global Equilibrium in Design of Geosynthetic Reinforced Walls

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

On Global Equilibrium in Design of Geosynthetic

Reinforced Walls
Dov Leshchinsky1
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Common design of MSE walls is based on a lateral earth pressure approach. A key aspect in design is the determination of the
reactive force in each reinforcement layer so as to maintain the system in equilibrium. This force leads to the selection of reinforcement
with adequate long term strength. It is also used to calculate the pullout resistive length needed to ensure the capacity of each layer to
develop strength. Lateral earth pressures used in design may or may not satisfy basic global equilibrium of the reinforced soil mass.
Hence, the present work establishes a benchmark test using a simple statically determinate approach, in order to check if different design
procedures satisfy equilibrium. Basic statics indicate that such a test is necessary, but not sufficient, to ascertain the validity of the
calculated reactive force. Three existing design methods are examined: AASHTO, National Concrete Masonry Association, and
Ko-stiffness. AASHTO, which is the simplest to apply and generally considered conservative, satisfies the benchmark test. However, it
may yield very conservative results if one considers the facing to play a major role. NCMA is likely satisfactory if one explicitly accounts
for the facing shear resistance in assessing the reaction in the reinforcement. The emerging Ko-stiffness approach, which is empirical, may
violate statics potentially leading to underestimation of the reinforcement force.
DOI: 10.1061/共ASCE兲1090-0241共2009兲135:3共309兲
CE Database subject headings: Equilibrium; Walls; Geosynthetics; Earth pressure; Design; Soil stabilization.

Introduction 共or limited eccentricity instead overturning兲, and has sufficient


capacity against bearing failure. While internal stability deter-
Geotechnical design reflects a process in which different analyses mines the required embedment length to resist pullout, both inter-
are used resulting in a synergetic outcome that is safe against nal and external stabilities are used to determine the
collapse while displacing within tolerable limits. In some earth reinforcement layout.
structures, such as slopes and walls, the predominant objective in Formally, Tmax is a function of the elongation of the geosyn-
the design process is to adequately satisfy stability. The same thetic as it interacts with the confining soil along its interfaces.
objective is usually applied to geosynthetic reinforced soil struc- Determining Tmax based on displacements, however, is compli-
tures. Experience shows that safe reinforced structures are pro- cated and generally impractical. Consequently, simplified calcula-
duced using simple limit state approach, and if safety factors tion of Tmax in common designs of walls is done based on lateral
satisfy typical design criteria, the displacement will likely be earth pressures multiplied by a layer’s tributary area; i.e., the
within tolerable limits. overburden pressure and the lateral earth pressure coefficient at
Common design of geosynthetic reinforced walls in the United the elevation of each layer are assumed to be known. Subse-
States is the result of a synergistic approach. That is, the rein- quently, the resulting Tmax in each design method will depend on
forced soil mass is analyzed for internal and external stability the utilized assumptions.
using different analyses to render the required strength and length
of the reinforcement. In internal stability, the analysis produces
the maximum reactive force, Tmax, in each reinforcement layer, Objective
the load at the connection to the facing, T0, which is typically
related to Tmax, and the pullout resistive length enabling the de- The objective of this paper is to establish a simple benchmark test
velopment of Tmax. Consequently, Tmax is the basis for selecting a
and assess each of the three design procedures used in the United
geosynthetic with proper long term strength, potentially the rein-
States: AASHTO 共1999, 2007兲, FHwA 共Elias et al. 2001兲, de-
forcement length, and the required connection strength. Clearly,
noted as the AASHTO method, NCMA 共1997兲, denoted as the
Tmax is a central element in design. External stability analysis
NCMA method, and the Allen and Bathurst 共2001a,b兲 method,
considers the reinforced soil as a coherent mass that, like a grav-
denoted as the Ko-stiffness method. This benchmark test uses a
ity wall, has adequate resistance against sliding and overturning
global static limit equilibrium analysis in which the sum ⌺Tmax
1
can be evaluated. While the benchmark test is not suitable for
Professor, Dept. of Civil and Environmental Engineering, Univ. of assessing the individual value of Tmax, it is useful in evaluating
Delaware, Newark, DE 19716. E-mail: dov@UDel.edu the level of global conservatism of established 共AASHTO,
Note. Discussion open until August 1, 2009. Separate discussions
NCMA兲 and emerging 共Ko-stiffness兲 designs. Such a test can also
must be submitted for individual papers. The manuscript for this paper
was submitted for review and possible publication on February 22, 2008; be used to examine, in a global sense only, the common criticism
approved on May 19, 2008. This paper is part of the Journal of Geotech- about the AASHTO and NCMA methods of being overly conser-
nical and Geoenvironmental Engineering, Vol. 135, No. 3, March 1, vative. The simple benchmark can be useful in assessing the
2009. ©ASCE, ISSN 1090-0241/2009/3-309–315/$25.00. emerging Ko-stiffness method as implemented in a design manual

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 309

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


In other words, this basic benchmark test is a necessary, but not
sufficient, condition to verify whether a calculated Tmax distribu-
tion is feasible.
In design that is based on lateral earth pressures or on slope
stability analysis 共e.g., AASHTO, NCMA, or Leshchinsky et al.
1995兲, it is commonly assumed that at its intersection with each
reinforcing layer, a potential failure plane AB defines the location
of Tmax. If one considers the locus of Tmax not to be along a single
failure surface such as plane AB 共Baker and Klein 2003兲, statics
requires that the design ⌺Tmax be equal to or larger than ⌺T cal-
culated along any plane AB as ⌺T is the minimum summed force
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

needed for global equilibrium of ABC. That is, if for any surface
AB the condition ⌺T ⬎ ⌺Tmax exists, the design ⌺Tmax is smaller
than required to maintain global equilibrium, implying that Tmax is
underestimated for some layers. Later a log spiral geometry is
used and its results are only slightly more conservative 共i.e., criti-
cal兲 than the planar surface, thus implying that plane AB is a
reasonable approximation for walls 共i.e., very steep slopes兲. For
situations such as complex geometries, very short reinforcement
or complex layout of reinforcement, neither planar nor log spiral
surface would be adequate; however, such cases are beyond a
simple benchmark test.
Conducting force equilibrium for the free body diagram in Fig.
1共b兲—see force polygon—and solving for ⌺T 共essentially, it is a
Culmann method兲, one gets

⌺T = 共 2 ␥H2 + qH兲共cot ␪ − tan ␻兲tan共␪ − ␾兲


1
共1a兲
Fig. 1. Basic statics for benchmark test: 共a兲 test body ABC represent-
ing summation all reinforcement forces along AB as ⌺T; 共b兲 free where ␥ = unit weight of reinforced soil; q = uniform surcharge
body diagram and force polygon load on the crest; H = height of wall; ␪ = assumed inclination of
planar surface AB; ␻ = face batter of wall; and ␾ = design internal
angle of friction of the reinforced soil. For an assumed value of ␪,
共WSDOT 2006兲. This paper indicates that, indeed, such a bench-
the value of ⌺T can be calculated using Eq. 共1a兲. This process
mark is needed.
needs to be repeated for various ␪ until maximum ⌺T, 共⌺T兲max, is
produced; this maximization represents the critical results: ␪cr and
共⌺T兲max. The benchmark test requires that
Benchmark Test
共⌺Tmax兲 艌 共⌺T兲max 共1b兲
While each design method yields different force Tmax for each
reinforcing layer, one can select a test body through the rein- Eq. 共1b兲 states that the sum of the design values of Tmax must be
forced soil layers to ascertain that the summation of Tmax over all equal to or larger than the maximum sum of T needed to maintain
layers meets or exceeds the requirement for static global equilib- the wedge ABC in a limit state equilibrium.
rium in the limit state 共refer to Fig. 1兲. AB is a line along which Allen and Bathurst 共2001b, a兲 state that the Ko-stiffness
each intersecting layer mobilizes force T and T for each layer method is for working stress conditions implying that no slip
could be different. ⌺T represents the summation of all forces in surfaces develop in the soil. Looking at the wedge ABC in Fig. 1
reinforcement layers intersecting AB. Consider the test body ABC and assuming that the soil strength is not fully mobilized along
in Fig. 1共b兲. A limit equilibrium analysis can then be conducted. AB implies that the average friction angle along AB is less than ␾.
Selection of a general shaped surface rather than AB results in a The force polygon in Fig. 1共b兲 implies therefore that ␨ ⬎ 共␪-␾兲,
statically indeterminate problem which cannot be solved by static thus resulting in 共⌺T兲max which is larger than the one required for
considerations alone. However, if one selects a planar or log spi- a limit equilibrium state. As the friction mobilized for such a case
ral surface that forms a simple single wedge, no statical assump- is unknown, the value of 共⌺T兲max for the working stress condi-
tions are needed in writing the global limit equilibrium equations tions cannot be found using the benchmark test approach; how-
for the test body, thus making it possible to evaluate the ⌺T value. ever, the 共⌺T兲max must be larger at working stresses than at
To avoid consideration of connection loads and resultant of lateral failure, thus requiring in design even larger 共⌺Tmax兲 than in Eq.
earth pressure at the face, ABC includes the facing, thus making 共1b兲. That is, some Tmax at working stress would be larger than at
these unknown force components internal, irrelevant in global a limit state. It is analogous to the well-known state of active
equilibrium. However, facing may affect ⌺T and its global impact versus at-rest lateral earth pressure.
can be considered as shown in the section “Benchmark Case.”
Consequently, the ABC test body becomes a simple benchmark
for checking whether the design Tmax distribution yields a glo- Overview
bally unstable structure. Such a benchmark cannot serve as con-
firmation of the validity of an individually calculated Tmax; it can For the sake of completeness, the expression for calculating Tmax
only serve to exclude the calculated distribution of Tmax if its in each design method is reproduced. It was used for generating
summed resistance is insufficient to render limit state equilibrium. the numbers presented in the “Benchmark Case” section.

310 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


sile modulus of all reinforcing layers; the exponent a is taken as 1
for geosynthetics and 0 for steel; ⌽fb = facing batter factor that is
constant for a given wall= 共Kabh / Kavh兲d where Kabh = horizontal
component of the active earth pressure coefficient accounting for
wall batter; and Kavh = horizontal component of active earth pres-
sure coefficient assuming the wall is vertical; and d = constant
coefficient with a recommended value of 0.5 based on curve fit-
ting; ⌽fs = facing stiffness factor with recommended values for
walls up to 6 m height of ⌽fs = 0.5 for modular block and propped
concrete panel faced walls; and ⌽fs = 1.0 for all other type of wall
facings such as wrapped face, welded wire, or gabion faced, and
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

incremental precast concrete facing; ␣ and ␤ = constant coeffi-


cients and, through curve fitting, their values are set to 0.27 and
0.24, respectively; and pa = atmospheric pressure to preserve non-
dimensionality.
Note that in Eq. 共3兲, the influence factor is ⌽ = ⌽local⌽fb
⫻⌽fs⌽g where ⌽local = 共Slocal / Sglobal兲a and ⌽g = ␣共Sglobal / pa兲␤. The
Ko-stiffness approach is based on regression analysis on data ob-
tained from field tests on walls conducted by various parties. It
assumes that the various factors are uncorrelated.
Fig. 2. Dt max versus normalized depth below top of wall following
Allen and Bathurst 共2001a兲: 共a兲 geosynthetic reinforced wall; 共b兲 me-
tallic reinforced wall
Benchmark Case

Without affecting the generality of the conclusions, consider a All designs implement factors in the analysis 共factors of safety,
simple wall with horizontal crest subjected to uniform surcharge reduction factors, and load and resistance factors兲. For an unob-
load—see Fig. 1. For AASHTO and NCMA, the calculated Tmax scured benchmark case, all factors are set to one. Such an ap-
for each layer is proach tests the key element in internal design of walls, Tmax.
It is convenient to utilize an example problem published by
Tmax = ␴hSv = K共␥Z + q兲Sv 共2兲 Allen and Bathurst 共2001a兲 as the benchmark case for two rea-
where for a particular layer Sv = tributary area; ␴h⫽representative sons. First, the Tmax values in the Ko-stiffness method are calcu-
horizontal stress; K = lateral earth pressure coefficient; ␥ = unit lated by Allen and Bathurst 共2001a兲, step by step, and thus do not
weight of reinforced soil; Z = distance from the crest to the middle require an intimate familiarity with all the details of this emerging
of the tributary area; and q = uniform surcharge load on the crest. method. Many of the details, however, deal with the implementa-
For geosynthetics, AASHTO uses K = Ka = Coulomb’s active lat- tion of Tmax in design 共e.g., calculating strains in the reinforce-
eral earth pressure coefficient with interface friction ␦ = 0. NCMA ment based on the empirically derived Tmax兲 and, though
uses K = Kah = Ka cos共␦-␻兲 = horizontal component of Coulomb’s important, are not relevant in the context of this paper; the inter-
active lateral earth pressure coefficient, Ka, with interface friction ested reader is referred to Allen and Bathurst 共2001a兲. Second, the
of ␦ = 共2 / 3兲␾ where ␾ = internal friction angle of the reinforced analyzed case is one of several tested walls that have been used to
soil and ␻ = wall face batter. calibrate the Ko-stiffness method 共Allen and Bathurst 2001b, a兲.
The Ko-stiffness approach 共Allen and Bathurst 2001b兲 utilizes Hence, one may assume that such field data are the “Rosetta
the following expression to determine Tmax for each layer of re- Stone” to assess the Tmax, unless it conflicts with fundamental
inforcement: statics; i.e., the benchmark test.
Fig. 3 shows the benchmark problem. Relevant data 共Allen
Tmax = ¯␴hSvDt max⌽ and Bathurst 2001a兲 are as follows:
1
= Ko␥共H + S兲SvDt max
2
冉 冊
Slocal a
Sglobal
⌽ fb⌽ fs␣
pa
冉 冊
Sglobal ␤
共3兲
1. Reinforced soil: unit weight, ␥ = 20.4 kN/ m3; internal fric-
tion angle corresponding to plane strain conditions, ␾ps = ␾
= 43°;
where ¯␴h = lateral earth pressure calculated as the average value 2. Block: unit weight, ␥u = 18.9 kN/ m3; depth Wu = 0.60 m; face
acting over the height of the wall 关i.e., ¯␴h = 0.5Ko␥共H + S兲兴; batter, ␻ = 2.9°;
Dt max = load distribution factor that modifies the reinforcement 3. Reinforcement layout: length/spacing of reinforcement is as
load on the basis of layer location 关it represents a trapezoidal shown in Fig. 3. However, summing the spacings yields a
envelope with depth varying over the range of 0 艋 Dt max 艋 1 共see wall height of 6.0 m and not 6.1 m as marked on the figure;
Fig. 2兲兴; ⌽ = influence factor which is the product of factors that and
account for the influence of local and global reinforcement stiff- 4. Backslope: geometry is shown; however, for calculation of
ness, facing stiffness, and facing batter; Ko = Jaky’s at-rest lateral Tmax, Allen and Bathurst 共2001a兲 follow AASHTO’s proce-
earth pressure coefficient 共Ko = 1 − sin ␾ps where ␾ps = ␾ corre- dure using an equivalent height of S = 1.3 m producing uni-
sponding to plane strain conditions兲; H = height of the wall; S form surcharge pressure of q = ␥S = 26.52 kPa.
= equivalent height of soil representing uniform surcharge pres- The actual analyzed benchmark is as shown in Fig. 4. Note
sure q 共i.e., S = q / ␥兲; Slocal = local reinforcement stiffness= J / Sv that the height was corrected to 6.0 m to keep consistency with
where J = tensile modulus of an individual layer; Sglobal = global the specified geosynthetic spacing. To ensure that the reported
reinforcement stiffness= ⌺J / H; where ⌺J = summation of the ten- Tmax by Allen and Bathurst 共2001a兲 remains approximately the

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 311

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


ABC remains stable. While the individual reactive horizontal
force, T, along AB requires assumptions, ⌺T mobilized by the
reinforcement layers to maintain the wedge ABC in a limit equi-
librium state can be determined using simple statics 关see Eq.
共1a兲兴.
Results for the Ko-stiffness method in Table 1 were obtained
using values employed or recommended by Allen and Bathurst
共2001a兲 in conjunction with Eq. 共3兲. Dt max values are as shown in
Fig. 2共a兲. Ko = 1 − sin 43° = 0.318. Reinforcement stiffness is J
= 420 kN/ m enabling one to find Slocal = J / Sv needed for each lay-
er’s Sv. Sglobal = ⌺J / H ⬇ 551 kPa 关for H = 6.0 m, Sglobal = 560 kPa;
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

the value of 551 kPa corresponds to H = 6.1 m and is used so as to


be consistent with the value used by Allen and Bathurst 共2001a兲;
however, the difference in results is insignificant兴. The exponent
value for geosynthetics is a = 1.0. The factors ⌽fb and ⌽fs are
0.952 and 0.5, respectively; note that for the wrapped-face wall,
⌽fs = 1.0. Finally, the values of ␣ and ␤ are 0.27 and 0.24, respec-
tively. All the values in Table 1 can now be reproduced using
Eq. 共3兲.
Maximizing ⌺T versus ␪ in Eq. 共1a兲, one note that for wedge
ABC the critical results are ␪cr ⬇ 65° and 共⌺T兲max = 90.4 kN/ m.
Hence, for a wrapped-face wall that is a variation of the wall in
Fig. 4, and the benchmark test requires that the summation of
Fig. 3. Cross section for Algonquin geogrid segmental concrete Tmax, 共⌺Tmax兲, would be equal to or greater than 90.4 kN/ m. A
block-faced wall following details by Allen and Bathurst 共2001a兲 log spiral analysis, briefly presented later, indicates for the same
case a sum of 92.3 kN/ m, marginally 共⬃2 % 兲 a more critical
value.
same, S was adjusted from 1.3 to 1.4 m 共i.e., q = 20.4⫻ 1.4 Table 1 shows Tmax for a wrapped-face wall using the
= 28.56 kPa兲. Ko-stiffness and the AASHTO designs. Note that NCMA method
Ignoring the facing 共considering a wrapped-face wall兲, the test is limited to block facing. Allen and Bathurst 共2001b兲 categorize
body to be evaluated is illustrated in Fig. 4 as wedge ABC. AB wrapped face as a flexible wall with ⌽fs = 1.0. In such systems
represents a planar slip surface along which the reinforcement there are no facing effects on 共⌺T兲max thus making a straightfor-
must develop sufficient reactive force to ensure that the wedge ward comparison with the static analysis of wedge ABC.
AASHTO calculates Tmax using Coulomb’s with soil-block in-
terface friction angle ␦ equal to zero. It considers walls with
batters smaller than 8° as vertical; i.e., ignore the batter of such
walls. Modifying AASHTO to consider the actual wall batter
共2.9°兲, one can see a drop of about 10% in Tmax. However, com-
paring respective Tmax calculated using the AASHTO and
Ko-stiffness methods, one realizes substantial differences.
Experience indicates that the distribution but not the summa-
tion of Tmax by AASHTO leads to excessively large maximum
共Tmax兲. As is often the case in design, maximum 共Tmax兲 is used to
select reinforcement with adequate long-term strength, thus re-
sulting in a selection of stronger than needed reinforcement.
However, the benchmark test looks only at the global equilibrium
of the system. Hence, 共⌺T兲max, considered as a horizontal force
keeping the wedge ABC stable, is of interest. The modified
AASHTO yields 共⌺Tmax兲 = 92.3 kN/ m, marginally larger than
needed for limit equilibrium 共90.4 kN/ m兲 using a planner surface
or the same if a log spiral is considered. Only for zero batter will
AASHTO yield identical summation to Culmann’s 关Eq. 共1a兲 and
共1b兲兴. That is, with ␻ ⬎ 0 and ␦ = 0, Coulomb’s resultant is in-
clined upward at ␻ to the horizontal, and this resultant is then
taken as ⌺Tmax. The AASHTO recommended procedure in which
zero batter is used for any ␻ ⬍ 8° produces a larger than needed
value 共⌺Tmax = 101.8⬎ 90.4 kN/ m兲. Realistically, one may con-
Fig. 4. Equivalent problem with horizontal crest subjected to uni- sider this excess as a reasonable tradeoff for more complex cal-
form pressure analyzed by Allen and Bathurst 共2001a兲. 关Note: 共1兲 culations.
wall height, H, is corrected to 6.0 m from reported value of 6.1 m The Ko-stiffness presents a dilemma: its summed value is
and height of uniform surcharge, S, equivalent to broken backslope, ⌺Tmax = 47.7⬍ ⬍ 90.4 kN/ m. There is a large deficit or imbalance
is 1.4 m, whereas reported value is 1.3 m; and 共2兲 line AB illustrating of stabilizing force. It is noted that Tmax values for the wrapped-
boundary of test body is unrelated to Allen and Bathurst 共report兲.兴 face case were obtained using the Ko-stiffness procedure; it does

312 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


Table 1. Tmax Calculated Using Three Design Methods
Ko-stiffness: Tmax AASHTO: Tmax NCMA: Tmax
共kN/m兲b 共kN/m兲 共kN/m兲
Reinforcement ␻ = 2.9°; ␦ = N / A ␦ = 0° ␻ = 2.9°; ␦ = 28.7°

Elevation Tributary ␻ = 0° ␻ = 2.9°


Layer 共toe is 0兲 area, Sv Block Wrapped block or block or
number 共m兲 共m2 / m兲 facing facing wrapped facingc wrapped facingc Block facing
a a
1 0.20 0.50 共0.40兲 1.76 共1.40兲 3.58 13.79 12.50 10.36
2 0.80 0.60 2.79 5.58 15.27 13.85 11.48
3 1.40 0.60 3.54 7.08 13.88 12.59 10.44
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

4 2.00 0.60 3.54 7.08 12.50 11.33 9.39


5 2.60 0.70 3.54 7.08 12.82 11.63 9.64
6 3.40 0.90 3.54 7.08 13.71 12.43 10.31
7 4.40 0.90 3.18 6.36 10.59 9.60 7.96
8 5.20 1.20 1.94 3.88 9.26 8.39 6.96
Total= 共⌺T兲max = 90.4 kN/ m for 23.8 47.7 101.8 92.3 76.5
benchmark test—Eq. 共1a兲
a
Numbers in parenthesis are considered as typographical error—check spacing in Fig. 3.
b
Values of Tmax for the block facing were calculated by Allen and Bathurst 共2001a, p. 60兲. Values of Tmax for wrapped face are inferred by the “facing
stiffness factor” provided by Allen and Bathurst 共2001b, p. 245兲.
c
AASHTO simplifies the calculation of Tmax: if the batter ␻ 艋 8° use ␻ = 0°.
d
Wrapped facing is not an available option in NCMA.

not reflect the results for the calibrated case as the tested wall had imbalanced horizontal force is significant 共90.4-23.8= 66.6
facing 共Fig. 3兲. However, Tmax values are the result of a recom- kN/ m兲 and, as in the NCMA case, it must be complemented by a
mended extrapolation of an empirical method. The benchmark force due to the block column over BD in Fig. 4. Using the block
test indicates that the necessary, but not sufficient, condition of data for the problem in Fig. 3 and assuming that the weight of the
⌺Tmax 艌 90.4 kN/ m is substantially not satisfied for the entire column of blocks is transmitted through BD, this force
Ko-stiffness method. would be W = 18.9⫻ 0.60⫻ 6.0= 68 kN/ m. Further assuming in-
AASHTO is silent about a procedure that considers the impact terface a friction angle along BD 共block to soil or block to block兲
of facing on Tmax. In calculating Tmax, NCMA accounts for facing of 30° renders sliding resistance of about 40 kN/ m. Hence, the
by only using the horizontal component of Coulomb lateral earth horizontal force deficit is reduced to 共66.6− 40兲 = 26.6 kN/ m. If
coefficient based on an interface friction of ␦ = 共2 / 3兲␾ between only the vertical portion of the column of blocks exerts vertical
the reinforced soil and blocks. The downward component of Cou- load to BD, it would be 51 kN/ m, rendering sliding resistance of
lomb force is arbitrarily ignored. Table 1 shows a drop of about about 30 kN/ m and horizontal force deficit of 36.6 kN/ m. The
20% in Tmax relative to AASHTO with a recommended large in- required interface friction to eliminate the imbalance is about 45°
terface friction of ␦ = 28.7°. The summation 共⌺Tmax兲 = 76.5 for the full weight of column of blocks, and 53° if only the ver-
⬍ 90.4 kN/ m. NCMA does not consider a benchmark test, tical portion of the column transmits the load. While one may
共⌺T兲max, for global stability. However, facing units can develop question whether such interface resistance is feasible, it is more
limited horizontal force along the interface between the leveling important to note that for the problem shown in Fig. 3, the
pad and foundation soil or between stacked blocks; see segment Ko-stiffness method would produce the same Tmax for other, more
BD in Fig. 4. This resistance can be assessed and represented in common, modular block facing 共i.e., ⌽fs = 0.5 is taken the same
wedge ABC, 关Fig. 1共b兲兴, to act in the same direction as 共⌺T兲max, for all modular block units兲. For example, for typical blocks that
thus reducing the need for reinforcement. Inclusion of this resis- are only Wu = 0.30 m deep with unit weight of ␥u = 22 kN/ m3, and
tance to sliding in the wedge analysis will likely reduce or elimi- an assumed interface friction of 30° along BD, the resulting slid-
nate the NCMA’s deficit. As an example, consider typical modular ing resistance would be 23 kN/ m, thus having a force deficit of
blocks having a unit weight of ␥u = 22 kN/ m3 and depth Wu about 43 kN/ m. An interface friction of about 60° along BD is
= 0.30 m. Assume that the entire weight of the 6 m high column needed to resolve the imbalance for typical modular blocks.
of blocks 共W = 22⫻ 0.3⫻ 6.0= 39.6 kN/ m兲 is transferred to the It is noted that Leshchinsky et al. 共1995兲 employed a log spiral
base 共BD in Fig. 4兲. Following NCMA recommendations, ignore analysis to explain the potential impact of facing in the test wall
the downdrag force on the blocks. Assume that the friction angle shown in Fig. 3. The analysis considered the broken backslope as
along BD is 30°; the resistance to sliding along BD would be is, without translation to an equivalent surcharge. Their analysis
Ts = 39.6⫻ tan共30° 兲 ⬇ 23 kN/ m. The deficit in the imbalanced showed that for Wu = 0.60 m, a large interblock friction and soil-
horizontal force is 共90.4-76.5兲 = 13.9 kN/ m ⬍ Ts = 23 kN/ m. block friction are needed 共40 and 20°, respectively兲 so as to pro-
Hence, the potential contribution of practically all segmental duce the apparent measured force in the reinforcement. However,
blocks is sufficient for this example. If the weight of the vertical they stated that these friction values are speculative. Though they
column of blocks, only above BD, is considered, W and hence Ts could have used a “rational” explanation by attributing the re-
would be half, thus resulting in marginal but practically accept- duced force in the reinforcement to the blocks, they stated that
able deficit. other factors beside friction can have the same effects, and thus
Table 1 indicates that Tmax for block facing in the Ko-stiffness they suggested further verifications before adaptation of interfa-
method is a small fraction of the NCMA value. The deficit in the cial frictions in the design of block walls. The “Commentary”

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 313

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


next shows that “other factors” such as a trace of cohesion 共due, Table 2. Effect of Cohesion in Benchmark Test 共Log-Spiral Analysis兲
for example, to capillary suction or soil matrix potential兲 can Assumed apparent
create the same effects. Conversely, it seems that the K-stiffness cohesion 关kPa兴 0 5 10 15 17.6
approach adopted a substantially reduced force in the reinforce- a a a a
共⌺T兲max 共kN/m兲 92.3 66.4 40.5 14.7 0
ment based solely on empirical correlations, ignoring statical con- a
siderations. Location of 共⌺T兲max in log-spiral analysis is at elevation 2.70 m; lower
locations have less than 2% influence.

Commentary versely, values of Tmax cannot be measured directly. That is, one
can measure strain and convert it to force. Allen and Bathurst
The Ko-stiffness approach was implemented in a public sector 共2001b兲 collected results of published field tests and made a com-
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

design manual 共WSDOT 2006兲. It includes a few refinements mendable effort to translate the data into Tmax values. Though
compared with Allen and Bathurst 共2001a,b兲 which are detailed in such values may have large statistical range, it is assumed that it
Allen and Bathurst 共2003兲. In the term ⌽g = ␣共Sglobal / pa兲␤ the val- was accounted for when Eqs. 共3兲 and 共4兲 were established. One
ues of ␣ and ␤ were changed from 0.27 and 0.24 to 0.25 and 0.25, can suggest at least two hypotheses why a conflict between reality
respectively. For the benchmark case it means that ⌽g drops from and theory does not necessarily exist:
0.406 to 0.382. Eq. 共3兲 implies that the calculated Tmax, and hence 1. There is a possibility of a trace of cohesion in the benchmark
共⌺Tmax兲, in Table 1, would be about 6% smaller for the refined test. Such a trace could be due to elements such as capillary
case. Furthermore, the Dt max versus depth for the geosynthetics suction or soil matrix potential. A trace of cohesion may
used in the benchmark case is slightly different than in Fig. 2共a兲. dramatically reduce the measured Tmax. As an example, con-
Rather than using Dt max = 0.2 and 1.0 at Z / H = 0.0 and 0.3, respec- sider the benchmark test for the wrapped-face problem.
tively, the manual recommends Dt max = 0.0 and 1.0 at Z / H = 0.0 Using log spiral analysis, Table 2 shows the required 共⌺T兲max
and 0.4, respectively. Following Eq. 共3兲, this represents a reduc- as a function of apparent cohesion 共the unit weight and fric-
tion in Tmax of the upper few layers leading to a further overall tion angle were maintained as in the benchmark test prob-
reduction of 共⌺Tmax兲. Finally, to account for actual facing, the lem兲. The instructive aspect of Table 2 is that an apparent
manual introduces an expression for facing stiffness ⌽fs 共note that cohesion of only 17.6 kPa 共about 370 psf兲 could provide suf-
it was taken before as 0.5 for block facing and 1.0 for wrapped ficient stability so as to render nil the measured Tmax. Smaller
facing; it remains as 1.0 for all flexible facing such as wrapped values of apparent cohesion will imply low measured Tmax
facing兲 needed to maintain static equilibrium. Ignoring apparent co-
hesion in design is prudent; however, not considering it in
⌽fs = ␩ 冉 1.5H5
ELW3uheff
pa 冊 ␬
共4兲
interpretation of measured field data may lead to unconser-
vative conclusions, attributing to the facing more than they
actually contribute. The large force deficit in the wrapped-
where E = modulus of the facing material; L = unit length of the face benchmark case 共90.4− 47.7= 42.7 kN/ m兲 potentially
wall; heff = equivalent height of an unjointed facing column that is supports the hypothesis of apparent cohesion which makes
100% efficient in transmitting moment throughout the facing col- up for the deficit.
umn, recommended as the average vertical reinforcement spacing 2. The Ko-stiffness method suggests its applicability to the
for modular block faced wall system; and the coefficients ␩ and benchmark test also in conjunction to a wrapped-faced wall;
␬, determined from regression analysis, are 0.5 and 0.14, respec- however, it was actually calibrated for the facing shown Fig.
tively. All other parameters were defined in conjunction with Eq. 3. The benchmark test for the block facing indicates that
共3兲. Utilizing Eq. 共4兲 with the data for the benchmark case 共Wu there still exists a force imbalance, though smaller than that
= 0.6 m兲, one can assess ⌽fs to be about 0.43 共less than the 0.5 for the wrapped-face wall. This imbalance may potentially be
value used in generating Table 1兲. For the hypothetical case of eliminated if a downdrag force is exerted by the reinforced
Wu = 0.3 m discussed in the previous section, ⌽fs would be ap- soil onto the facing blocks. Such a force will reduce the
proximately 0.58. However, with the further reductions due to ⌽g weight of wedge ABC 关i.e., 共W + Q兲 in Fig. 1共b兲兴, thus de-
and Dt max, the calculated values in Table 1 will be close to those crease 共⌺T兲max. It will also increase the sliding resistance
obtained before for a typical block that is 0.30 wide. Conse- along BD 共Fig. 4兲 through an increased vertical force exerted
quently, the discussion related to the benchmark case holds, by the blocks. As per the NCMA, however, it seems prudent
whereas the deficit in 共⌺Tmax兲 could even be slightly larger. to ignore in design the downdrag force on the back of the
Therefore, the following commentary uses the observations as facing blocks, at least in terms of its effects on the sliding
related to the benchmark case. resistance along BD. Hence, even if a downdrag force is used
The Ko-stiffness method implies that for the benchmark case, to explain the force deficit when block facing is considered,
3
about 4 共66.6 kN/ m兲 of the horizontal load was carried by the it seems that in the context of design its impact on Tmax
1
blocks and only about 4 共23.8 kN/ m兲 by the geosynthetic. Since should be ignored. In any event, extrapolation of the cali-
the Ko-stiffness method yields results close to measured field val- brated case to a wrapped-face wall cannot rely on downdrag
ues 共Allen and Bathurst 2001b, a兲, there is seemingly a potential force and, as such, the results of the Ko-stiffness method
conflict between reality 共measured values兲 and basic physics remain unexplained from the perspective of basic statics.
共theory兲. This commentary suggests that one should not be dismissive of
The validity of the benchmark test could be questioned if one “theory” by blindly embracing “reality.” The same experimental
rejects the Mohr–Coulomb failure criterion in practice. However, observation can be attributed to different causes and it is quite
it does appear that all three design methods use ␾ that corre- easy to end up with a physically impossible theory. The purpose
sponds to peak strength conditions. The mechanics used in the of the simple benchmark test is to provide one with a tool guard-
benchmark test deals with a statically determinate problem. Con- ing against possible misinterpretation of observed results. Per-

314 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

J. Geotech. Geoenviron. Eng. 2009.135:309-315.


ceived reality has some potential uncertainties which cannot be the Ko-stiffness method results in a substantial deficit of 共⌺T兲max.
resolved by statistical correlations alone. The NCMA method yields just slightly less than the required
Eq. 共3兲 reflects an apparent lateral earth pressure envelope that value. Since the NCMA is used only for segmental facing blocks,
when combined with tributary area yields Tmax. It would be in- the unaccounted resistance to sliding by these blocks will likely
structive to point out Peck’s approach to an apparent lateral earth compensate for the imbalance in horizontal force. Large block
pressure envelope for braced cuts in sand. Peck’s equation, ␴ facing or low walls with typical modular facing blocks may ren-
= 0.65␥HKa, is empirically based. It exhibits a uniform apparent der AASHTO very conservative in a global sense; however, if the
lateral stress envelope along the height of the supported excava- facing effect on 共⌺T兲max is ignored, AASHTO meets the bench-
tion. Hence, this envelope renders a resultant force P mark test. The Ko-Stiffness method would require large facing
= 0.65␥H2Ka. Since Rankine’s active state is at a limit state, the units to make up for an imbalance in global horizontal equilib-
benchmark criterion for this problem would be a resultant force of rium. With smaller blocks it will have to count on long-term large
Downloaded from ascelibrary.org by UNIV OF CONNECTICUT LIBRARIES on 05/12/13. Copyright ASCE. For personal use only; all rights reserved.

Pa = ␥H2Ka / 2. Consequently, the global inequality P ⬎ Pa exists downdrag force as well as high interblock friction. The NCMA
in Peck’s case, thus satisfying a relevant benchmark criterion; ignores downdrag force in design as its viability in block walls
however, if it were P ⬍ Pa, static equilibrium would be violated. over time is questionable.
Note that while Rankine’s predicts lateral pressure that is linearly It seems that a potential reconciliation between statics and the
increasing with depth, the empirical value indicates it is uniform. empirical statistical approach is to integrate the statical bench-
This distribution of pressure can be affected by local factors 共flex- mark with the empirical formulation. The benchmark constraint
ibility of wall, stiffness of braces, compaction兲; however, the re- would not guarantee prediction of an individual Tmax that is nec-
sultant global force must exceed a minimal value to ensure static essarily correct; however, it can be used to readjust Tmax distribu-
equilibrium. In fact, the resultant in Peck’s is 30% larger than for tion to eliminate deficit in global equilibrium. While it is likely
Rankine’s active state. This larger value is likely due to the high that factors typically used in design 共e.g., load factors, resistance
stiffness of the bracings and facing which limits the movement of factors, reduction factors, factors of safety兲 combined with the
the retained soil, thus preventing Rankine’s state from being fully readjusted Tmax distribution would produce a safe structure in the
developed. That is, the tradeoff for restricting soil movements is sense of equilibrium, the use of unfactored values of individual
lesser soil contribution to shear resistance resulting in a need for Tmax to calculate strains in each reinforcement layer 共e.g., as is
larger than limit state auxiliary forces to maintain global equilib- required in WSDOT 2006兲 would render questionable strain val-
rium. The analogous case with reinforced wall is that “at working ues from a design perspective, especially when related to under-
stress conditions” the reaction in the reinforcement, 共⌺T兲max, estimated Tmax values.
would be larger than predicted at a limit equilibrium state where
the soil shear strength is fully mobilized. The Ko-stiffness method
implies just the opposite, rendering a deficit in force to maintain
equilibrium even in a limit state. References

AASHTO. 共1999兲. Standard specifications for highway bridges, 16th Ed.,


Washington, D.C.
Concluding Remarks
AASHTO. 共2007兲. LRFD bridge design specifications, 4th Ed., Washing-
ton, D.C.
The general problem of mechanically stabilized walls is statically
Allen, T., and Bathurst, R. 共2001a兲. “Application of the Ko-stiffness
indeterminate. In particular, determination of the maximum reac- method to reinforced soil wall limit state design.” WA-RD 528.1,
tive force in each layer, Tmax, requires assumptions. Yet, the value Washington State Department of Transportation, Olympia, Wash.
of Tmax in each layer is the key for economical design of a wall Allen, T., and Bathurst, R. 共2001b兲. “Prediction of soil reinforcement
that is internally stable. The need to assess the value of Tmax loads in mechanically stabilized earth 共MSE兲 walls.” WA-RD 522.1,
resulted in the development of the AASHTO, NCMA, and Washington State Department of Transportation, Olympia, Wash.
Ko-stiffness methods. The Ko-stiffness method yields vastly Allen, T., and Bathurst, R. 共2003兲. “Prediction of reinforcement loads in
smaller results than the other two. It is purely empirical based on reinforced soil walls.” WA-RD-522.2, Revised, Washington State De-
correlations of data collected from various case histories. partment of Transportation, Olympia, Wash.
This paper presents a statically determinate benchmark test Baker, R., and Klein, Y. 共2003兲. “An integrated limiting equilibrium ap-
that is necessary, but not sufficient, for a predictive method to be proach for design of reinforced soil retaining structures: Part
I—Formulation.” Geotext. Geomembr., 22共30兲, 119–150.
acceptable. That is, it suggests checking global equilibrium where
Elias, V., Christopher, B. R., and Berg, R. R. 共2001兲. Mechanically sta-
共⌺T兲max for all reinforcement layers is considered using a limit
bilized earth walls and reinforced soil slopes—Design and construc-
equilibrium analysis. If this summation is larger than the design
tion guidelines, No. FHWA-NHI-00–043, Federal Highway
value of 共⌺Tmax兲, the design approach underestimates the reactive
Administration, Washington, D.C.
force Tmax. However, if 共⌺Tmax兲 艌 共⌺T兲max, the approach is glo- Leshchinsky, D., Ling, H. I., and Hanks, G. 共1995兲. “Unified design
bally acceptable. approach to geosynthetic reinforced slopes and segmental walls.”
The use of a benchmark test for a case that was also used to Geosynthet. Int., 2共5兲, 845–881.
calibrate the empirically-statistically based Ko-stiffness method National Concrete Masonry Association 共NCMA兲. 共1997兲. Design
indicates that as is, it results in a large deficit of global horizontal manual for segmental retaining walls, 2nd Ed., J. G. Collin, ed., Hern-
force. Ironically, the same benchmark case was one of several don, Va.
tests sponsored by the FHwA to establish the mechanistically Washington State Department of Transportation 共WSDOT兲. 共2006兲. Geo-
based AASHTO method. For a wrapped-face wall, the AASHTO technical design manual, M 46–03, Chap. 15, Abutments, retaining
method produces marginally larger than required 共⌺T兲max whereas walls, and reinforced slopes, Olympia, Wash.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 315

J. Geotech. Geoenviron. Eng. 2009.135:309-315.

You might also like