Engineering Structures 22 (2000) 793–808

[Link]/locate/engstruct

Practical analysis and design of mechanically-stabilized earth

walls—I. Design philosophies and procedures
*
Yohchia Chen
Civil Engineering Program, Penn State at Harrisburg, 777 W. Harrisburg Pike, Middletown, PA 17057-4898, USA

Received 1 December 1997; received in revised form 22 February 1999; accepted 22 February 1999

Abstract

Design methods with the important design/construction considerations for mechanically-stabilized earth (MSE) walls are
described. A general analysis/design procedure for MSE wall systems is discussed in detail. The effects of design methods, parti-
cularly the impact of the new load and resistance factor design (LRFD) method, on MSE walls are discussed in the companion
paper (Part II) through the use of realistic examples. To ease the comparisons among the various design approaches, a unified
notation is used to express all equations.  2000 Elsevier Science Ltd. All rights reserved.

Keywords: Analysis; ASD; Computation; Design; LFD; Limit state; Load factor; Load group; LRFD; MSE wall; Resistance factor

Conversion of units condition, and hence they are usually assumed to be

1 mm ⫽ 0.03937 in inextensible in practice [1].
1m ⫽ 3.2808 ft MSE walls are considered as proprietary systems.
1N ⫽ 0.2248 lb Their analysis or design has been relied on the limited
1 N/m ⫽ 0.0685 lb/ft past experience. Current US practice on MSE walls
1 Pa ⫽ 0.000145 psi employs either the allowable stress design (ASD)
method or the load factor design (LFD) method [2]. In
1994, American Association of State Highway and
Transportation Officials (AASHTO) introduced the load
1. Introduction and resistance factor design (LRFD) method [3], which
is generally known as the limit states design in Europe
Fig. 1 shows a typical mechanically-stabilized earth and the rest of the world. This new design philosophy
(MSE) wall system which consists of three major struc- has the following key features: (a) it is a probability-
tural components: the vertical facing element, the level based code considering the degree of variability of dif-
pad, and the soil reinforcement. Application of such ferent loads and materials; (b) it includes all possible
retaining wall systems has been increasingly wider in limit states (i.e. the conditions representing the limits of
the USA since about two decades ago. MSE walls are structural usefulness); (c) the load and resistance factors
particularly suitable where substantial settlements are were derived based on a comprehensive calibration pro-
anticipated. cedure and a statistically meaningful database; and (d)
The soil mass behind the precast concrete wall (Fig. it recognizes the importance of the effects of ductility
1) generally consists of select granular backfill material and redundancy on loads. Since the LRFD method is
reinforced by metal strips or metal grids. These soil rein- relatively new, its impact on MSE wall design, as com-
forcing elements have tensile strains generally smaller pared to the other design methods, requires detailed
than those of the surrounding soil under the same stress evaluation as it has not yet been done. Also, as a result
of the increasing interest and applications, it is important
to rationalize and standardize the MSE wall
* Tel: ⫹ 1-717-948-6146; fax: ⫹ 1-717-948-6401; e-mail: analysis/design procedure for practical purposes and
yxc2@[Link] quality control.

0141-0296/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 9 9 ) 0 0 0 2 1 - 8
794 Y. Chen / Engineering Structures 22 (2000) 793–808

Fig. 1. General MSE wall system.

Part I of this two-part paper discusses the various Exact quantification of n factors has not yet been com-
design methods, and presents a general analysis/design pleted, but they are believed to be between 0.95 and
procedure for MSE walls. The companion paper (Part 1.05. n has been inherently assumed to be 1.0 in the past
II) discusses the effects of various design methods on practice. The left-hand side of Eq. (1b) represents a parti-
MSE walls, particularly the impact of the LRFD method, cular limit state (i.e. a condition that a structure ceases
through the use of realistic examples. its intended function), or a statistics-based load combi-
nation, while the right-hand side of Eq. (1b) means the
available structural strength/capacity for design purpose.
2. Design philosophies and methodologies ␥i and ␾ factors for the LRFD method were determ-
ined through a calibration process in which the structural
2.1. Load and resistance factor design (LRFD) reliability index was preselected as 3.5 (corresponding
to the probability of failure of approximately 0.0002 i.e.
The fundamental design requirement of the LRFD two failures in every 10 000 cases) and a significantly
method can be expressed by: large database was used [3]. By introducing such ␥i and
␾ factors, the desirable uniform structural safety can
Required factored load ⱕ Design strength (1a) therefore be achieved.
The loads, load factors and load combinations associa-
or ted with the LRFD method are summarized in Table 1
[3]. The unfactored loads, Qi, consist of three groups
n
⌺(␥iQi) ⱕ ␾Rn (1b) (Table 1):

where n is a load modifier depending upon the impor- 쐌 Permanent loads: DC, dead load due to structural
tance, ductility and redundancy of the structure, ␥i the components and attachments; DD, dead load due to
load factors (normally ⱖ 1.0 depending on the degree downdrag; DW, dead load due to wearing surfaces
of load uncertainty), Qi the individual loads such as dead and utilities; EH, horizontal earth pressure; EV, verti-
load, live load, earth load… etc., ␾ the resistance factor cal earth pressure; and ES, earth surcharge.
accounting for material uncertainties ( ⱕ 1.0 normally), 쐌 Transient loads: LL, vehicular live load; IM, vehicular
and Rn is the nominal resistance/strength of material. impact load; CE, vehicular centrifugal load; BR,
 Y. Chen / Engineering Structures 22 (2000) 793–808 795

Table 1
Load factors and load combinations for LRFD

DC LL WA WS WL FR SE
DD IM
Limit state DW CE
EH BR
EV PL
ES LS EQ IC CT*

Strength-I ␥p 1.75 1.00 – – 1.00 ␥SE – – –

Strength-II ␥p 1.35 1.00 – – 1.00 ␥SE – – –
Strength-III ␥p – 1.00 1.40 – 1.00 ␥SE – – –
Strength-IV
DW, EH, EV, ES ␥p – 1.00 – – 1.00 – – – –
DC only 1.50
Strength-V ␥p 1.35 1.00 0.40 0.40 1.00 ␥SE – – –
Extreme event-I ␥p ␥EQ 1.00 – – 1.00 – 1.00 – –
Extreme event-II ␥p 0.50 1.00 – – 1.00 – – 1.00 1.00
Service-I 1.00 1.00 1.00 0.30 0.30 1.00 ␥SE – – –
Fatigue
LL, IM, CE only – 0.75 – – – – – – – –

*Use one of these three loads, EQ, IC and CT, at a time. ␥p, extreme load factors for permanent loads, Table 2.

vehicular braking force; PL, pedestrian live load; and Table 2

LS, live-load surcharge. Extreme load factors for the permanent loads in Table 1 (␥p)
쐌 Other loads: WA, water load; WS, wind load on struc- Type of load Load factor (␥p)
ture; WL, wind on live load; FR, friction; SE, settle-
ment; EQ, earthquake load; IC, ice load; CT, vehicu- Maximum Minimum
lar collision force; and CV, vessel collision force.
DC 1.25 0.90
It should be mentioned here that temperature, creep DD 1.80 0.45
and shrinkage effects are negligible for MSE walls and DW 1.50 0.65
that vehicular impact (IM) usually does not affect a sub- EH:
Active 1.50 0.90
structure design. Therefore, these effects need not be At-rest 1.35 0.90
considered in MSE wall designs. It should also be noted EV:
that in Table 1 the load factor reflecting the simultaneous Overall stability 1.35 N/A
occurrence of an earthquake and the transient load(s), Retaining structures 1.35 1.00
␥EQ, has not been finalized, but it is believed to be ES 1.50 0.75
between 0 and 1.0. As an approximation, ␥EQ may be
taken as 0.50. Moreover, the load factor for settlement,
␥SE, in Table 1 is project oriented, which must be
determined through a specific study. stability computations maximum EH and EV load fac-
The load combinations listed in Table 1 include: (a) tors generally result in the maximum stress of soil
five strength limit states which are related to the safety reinforcement and minimum EV load factor usually con-
of a structure during its intended life; (b) two extreme trols pullout computations. ␾ factors for the LRFD
limit states which are also related to the structural safety, method are shown in Table 3 [3].
but against extreme loads such as earthquake, ice impact
and vessel collision; (c) one serviceability limit state 2.2. Load factor design (LFD)
which is related to the structural performance under nor-
mal service load conditions; and (d) the fatigue limit The LFD method has the merit of the LRFD method
state which concerns the fatigue due primarily to vehicu- (Eq. (1b)). Its design dilemma can be written as:
lar live load with impact. For permanent loads,
maximum and minimum loading cases should be con-
sidered in order to examine the possible extreme struc- ␥o⌺(␤iQi) ⱕ ␾Rn (2a)
tural effects that may exist. The associated extreme load
factors (␥p) are summarized in Table 2 [3]. For example, or
for external stability computations maximum EH and EV
load factors generally govern design, while for internal ⌺[(␥o␤i)Qi] ⱕ ␾Rn (2b)
796 Y. Chen / Engineering Structures 22 (2000) 793–808

Table 3
N factors for LRFD methoda

Bearing resistance Sand

쐌 Semi-empirical procedure using SPTb data 0.45
Rock
쐌 Semi-empirical procedure 0.60
Sliding Precast concrete placed on sand
쐌 Using ␾ cs from SPT data 0.90
Concrete cast-in-place on sand
쐌 Using ␾s from SPT data 0.80
Overall stability 0.85

Tensile resistance of inextensible reinforcement Strip reinforcements

쐌 Yielding of gross section less sacrificial area 0.85
쐌 Fracture of net section less sacrificial area 0.70
Grid reinforcements
쐌 Yielding of gross section less sacrificial area 0.75
쐌 Fracture of net section less sacrificial area 0.60
Connectors
쐌 Yielding of gross section less sacrificial area 0.75
쐌 Fracture of net section less sacrificial area 0.60
Ultimate pullout resistance in soil 0.90
Concrete 쐌 Flexure 0.90
쐌 Shear 0.90
쐌 Axial tension 0.90
쐌 Axial compressiond 0.75

a
Clay being undesirable and excluded from this table. bStandard penetration test. cInternal friction angle of soil. dExcluding Extreme event I limit
state with high seismicity.

where ␤i is a load factor modifier, and ␥o is the overload notation for loads is being used [2]. The ␾ factors for
factor depending upon the load combination. the LFD method are contained in Table 5 [2].
In Eq. (2b), ␥o times ␤i is equivalent to ␥i of Eq. (1b),
and thus ⌺[(␥o␤i)Qi] represents a load combination. 2.3. Allowable stress design (ASD)
However, the load and resistance factors of the LFD
method are not statistically representative since they
The design equation for the ASD method can be
were calibrated against the ASD method by considering
rewritten from Eq. (2b) as:
only one limited case—a simply supported bridge sys-
tem with relative short span length (15.24 m). The ␥o and
␤i factors and load combinations for the LFD method are group load ⫽ ⌺Qi ⱕ
Rn
(3)
summarized in Table 4, in which the standard LRFD ␥ave/␾

Table 4
Load factors and load combinations for LFD method

Load EH
group ␥o DL (LL ⫹ (LL ⫹ CE EV BR WA WS WL FR EQ IC
IM)na IM)pb

I 1.3 ␤D 1.67 – 1 ␤E 1 1 – – – – –
II 1.3 ␤D – – – ␤E 1 1 1 – – – –
III 1.3 ␤D 1 – 1 ␤E 1 1 0.3 1 1 – –
IV 1.3 ␤D 1 – 1 ␤E 1 1 – – – – –
V 1.25 ␤D – – – ␤E 1 1 1 – – – –
VI 1.25 ␤D 1 – 1 ␤E 1 1 0.3 1 1 – –
VII 1.3 ␤D – – – ␤E 1 1 – – – 1 –
VIII 1.3 ␤D 1 – 1 ␤E 1 1 – – – – 1
IX 1.2 ␤D – – – ␤E 1 1 1 – – – 1

␤D ⫽ 0.75 when checking maximum bending moment and 1.0 when checking maximum axial load; ␤E ⫽ 1.0 for vertical earth pressure and 1.3
for horizontal earth pressure.
a
Normal vehicles. bSpecial vehicles.
 Y. Chen / Engineering Structures 22 (2000) 793–808 797

Table 5
Safety and resistance factors for ASD and LFD methodsa

Bearing resistance Sand and rock 3.0

Sliding 1.5b
Overturning 2.0/1.5bc
Overall stability 1.5

Tensile resistance of inextensible reinforcement Strip reinforcements

쐌 Yielding of gross section less sacrificial area 1.82
쐌 Fracture of net section less sacrificial area 2.0
Grid reinforcements
쐌 Yielding of gross section less sacrificial area 1.82
쐌 Fracture of net section less sacrificial area 2.0
Connectors
쐌 Yielding of gross section less sacrificial area 1.82
쐌 Fracture of net section less sacrificial area 2.0
Ultimate pullout resistance in soil 1.5b
Concrete 쐌 Flexure 0.90
쐌 Shear 0.85
쐌 Axial tension 0.90
쐌 Axial compressiond 0.70

a
Clay being undesirable and excluded from this table, and the listed resistance factors pertain to the LFD method. bReduced to 75% if seismic
loading present. c2.0 for mass on soil, and 1.5 for mass on rock. dWith ties as the transverse/confinement reinforcement.

where ␥ave is an implicit average load factor defined by: typically). The ASD method permits certain percentage
of increase for allowable stresses/loads (Table 6).

冘 冋␥ ␤ 册 As can be inferred, different safety factors would be

N

␥ave ⫽
o i
(4) required for different load combinations using the ASD
i⫽1
N
method. This results in a non-uniform safety which is
the main drawback of the ASD method.
where N is the number of loads involving in a particular
load combination. 2.4. Summary
␥ave/␾ in Eq. (3) implies a factor of safety (FS). There-
fore, the right-hand side of Eq. (3) represents an allow- The LRFD method is clearly more rational and scien-
able load, Qall (Rn/FS). Table 6 summarizes the ␥o and tific as it has the merit of both the probability theory
␤i factors and load combinations for the ASD method (load combinations) and the reliability theory (structural
with the standard LRFD notation for loads [2]. In the safety). For practical reasons, the designers are, how-
ASD, service loads are typically employed (i.e. ␥o ⫽ 1.0 ever, not expected to carry out the details of probability

Table 6
Load factors and load combinations for ASD method

Load EH Increase of
group allowable
stress or
load (%)*
␥o DL (LL ⫹ (LL ⫹ CE EV BR WA WS WL FR EQ IC
IM)n IM)p

I 1.0 1 1 – 1 1 1 1 – – – – – 0
II 1.0 1 – – – 1 1 1 1 – – – – 25
III 1.0 1 1 – 1 1 1 1 0.3 1 1 – – 25
IV 1.0 1 1 – 1 1 1 1 – – – – – 25
V 1.0 1 – – – 1 1 1 1 – – – – 40
VI 1.0 1 1 – 1 1 1 1 0.3 1 1 – – 40
VII 1.0 1 – – – 1 1 1 – – – 1 – 33
VIII 1.0 1 1 – 1 1 1 1 – – – – 1 40
IX 1.0 1 – – – 1 1 1 1 – – – 1 50

*100 ⫻ (1—maximum stress for operating rating/allowable basic stress). Note: (LL ⫹ IM)n is for AASHTO vehicular loading, and (LL ⫹ IM)p
is for specially permitted vehicular loading.
798 Y. Chen / Engineering Structures 22 (2000) 793–808

and reliability theories as they are implictly reflected in properties (soil/rock, concrete, steel reinforcement),
the load and resistance factors and load combinations wall height (H), applicable loads, load and resistance
(i.e. Eq. (1b)) decided by the codewriters through the factors, soil reinforcement type, and design life.
calibration process. The important fact is that this new 쐌 Step 2. Identify the special conditions required for
design method makes designers realize and understand evaluation (subflowchart 5a). Possible special con-
the variability in loads and materials, the probability of ditions for a MSE wall system include the potential
one load combined with certain other loads, and the loss of support at the base of wall due to the removal
inherent structral reliability/safety. For a detailed of soil, scour, drainage, erosion, and seismic loading.
description of the LRFD method, it is suggested that These should be considered during the
readers refer to the available literature, such as Refer- design/analysis process.
ences [3,4]. 쐌 Step 3. Calculate the inclination angle of earth press-
ures (I) and the slope of backfill surface (i)
(subflowchart 5a). I ⫽ tan−1[a/(2H)] for the broken
3. General analysis/design procedure backfill surface condition (Fig. 2), i ⫽ tan−1(a/b) for
the sloping backfill surface condition (Fig. 3), and i
In general, MSE walls should be proportioned by con- ⫽ 0 for the horizontal backfill condition (Fig. 4),
sidering the following limit states: (a) sliding resistance where a is the vertical projection of backfill slope, H
of the reinforced soil zone; (b) bearing resistance of soil is the height of the wall and b is the horizontal projec-
or rock upon which the reinforced soil zone resides; (c) tion of backfill slope (Fig. 1).
overturning of the reinforced soil zone; (d) overall stab- 쐌 Step 4. Design/analyse the reinforced soil mass
ility (against deep-seated failure) of foundation materials (subflowcharts 5a and 5b). Details for this step and
below the structure; (e) structural capacity of wall facing steps 5 and 6 are described in separate sections fol-
and connections; (f) rupture and pullout resistance of lowing this section.
reinforcing elements; and (g) vertical and differential 쐌 Step 5. Check the settlement of MSE wall system
settlement of foundation. Limit states (a–f) stated above (subflowchart 5b).
are strength type, while limit state (g) is a serviceability 쐌 Step 6. Check the failures of soil reinforcements
one. The condition of backfill surface could be broken (subflowcharts 5b and 5c).
(Fig. 2), sloping (Fig. 3), or horizontal (Fig. 4). The pro- 쐌 Step 7. Design/analyse the panel-reinforcement con-
posed complete analysis/design procedure for a MSE nections (subflowchart 5c). The connections between
wall system emphasizing the LRFD method is depicted the wall panel and soil reinforcements should be
by the flowcharts shown in Fig. 5 (a–d). This general designed with a horizontal force at least equal to 85%
procedure involves the following key steps: of the maximum horizontal force at a given depth of
쐌 Step 1. Establish the required basic design data 0.5 H from the top of the wall, and then staying at
(subflowchart 5a). The required basic design data 100% for the lower half (ASD and LFD) or increasing
includes general geotechnical conditions (foundation linearly to 100% at the bottom of wall (LRFD).
bearing material, sloping, groundwater level), material 쐌 Step 8. Design/analyse the wall panels (subflowchart

Fig. 2. Earth pressure and live load surcharge diagrams for the broken backfill condition (BC).
 Y. Chen / Engineering Structures 22 (2000) 793–808 799

Fig. 3. Earth pressure and live load surcharge diagrams for the sloping backfill condition (SC).

Fig. 4. Earth pressure and live load surcharge diagrams for the horizontal backfill condition (HC).

5d). The wall panels should be designed with suf- 4. Step 4. Design/analyse the reinforced soil mass
ficient redundancy to protect against catastrophic fail- (subflowcharts 5a and 5b)
ure of each discrete facing panel. Adequacy under this
condition should be evaluated using a load factor of Sizing of the reinforced soil mass is generally based
1.1 for EH and 1.0 for EV for both pullout and rupture on stability and geotechnical strength requirements,
of soil reinforcements using the LRFD method [3]. structural resistance requirements, and the development
The design of a wall generally involves: (1) determin- of soil reinforcement. This step involves the tasks:
ing the unfactored loads and moments at the bottom 쐌 Determine the length of soil reinforcement (L, Fig. 1),
center of the wall; (2) calculating the factored loads the active earth pressure coefficient (ka), the height
and moments at the bottom center of the wall; (3) from the bottom of the wall to the top of earth press-
determining the required reinforcing steel for the wall; ure diagram (h, Figs. 2–4), and the earth pressure
and (4) checking the shear and axial resistances for (EH).
the wall. 쐌 Compute the live-load surcharge load (LS), and the
쐌 Step 9. Design/analyse the footing (subflowchart 5d). maximum acceleration coefficient at the centroid of
The design of a footing supporting the MSE wall gen- the investigated mass (Am).
erally involves: (1) calculating the factored loads and 쐌 Determine the forces and moments at the toe of the
moments at the critical sections of the footing; (2) reinforced soil mass.
determining the required reinforcing steel for the foot- 쐌 Check the external stability for the reinforced soil
ing; and (3) checking the shear resistance for the foot- mass.
ing. 쐌 Check the safety against the possible failure modes
800 Y. Chen / Engineering Structures 22 (2000) 793–808

Fig. 5. (a) Subflowchart; (b) Subflowchart; (c) Subflowchart; (d) Flowchart describing the general analysis/design procedure for a MSE wall system.

including overturning, sliding, bearing, overall stab- H

H1 ⫽ for the LRFD, (5a)
ility and deep-seated shear. 1 ⫺ 0.3tani

4.1. Determine L, ka, h and EH or

The length of reinforced soil mass, L, is generally 2H ⫹ a

H1 ⫽ for the LFD and the ASD. (5b)
determined based on the pullout resistance of soil 2
reinforcements. For inextensible reinforcing elements,
the pullout resistance should be analysed by dividing the The practical minimum value for L is 2.4 m or 0.7 H
in-situ reinforced zone as two zones—the active and whichever is greater. L should be increased when sur-
resistant zones as shown in Fig. 6. The practical mini- charges or other possible external loads are present, and
mum length of the resistant zone (Fig. 6) is 900 mm. should remain constant throughout the entire height of a
Therefore, the minimum required L is 0.3 H1 plus 900 MSE wall for practical reasons. Furthermore, for design
mm where H1 is defined by (Fig. 6): purposes the tension in the soil reinforcement can be
 Y. Chen / Engineering Structures 22 (2000) 793–808 801

Fig. 5. Continued.

assumed to be resisted by a uniformly distributed earth and

pressure on the back of the panel.
ka, which is for evaluating the safety against soil fail- EH ⫽ 0.5ka␥bh2 (8)
ure, may be computed according to Rankine’s theory:
where ␥b is the unit weight of random backfill.
cosI ⫺ √cos I ⫺ cos ␾b
2 2 The horizontal component of EH (EHh) and the verti-
ka ⫽ cosI (6) cal component (EHv) are:
cosI ⫹ √cos2I ⫺ cos2␾b
EHh ⫽ EHcosI (9a)
where ␾b is the internal friction angle of random back-
fill; h and EH are determined by: and

h ⫽ H ⫹ Ltani, (7) EHv ⫽ EHsinI (9b)

802 Y. Chen / Engineering Structures 22 (2000) 793–808

and

Am ⫽ (1.45 ⫺ A)A (11)

where heq the equivalent soil height for the design

vehicular loading, and A is the maximum earthquake-
induced acceleration obtained from the isoseismal map
[3].
In the ASD and LFD methods, heq remains as a con-
stant value of 600 mm [2], while in the LRFD method
as shown in Table 7 heq depends on the height between
the footing bottom and wall top when designing the wall
or depends on the height between the footing bottom and
the top of the earth pressure diagram when analysing the
reinforced soil mass [3]. Linear
interpolation/extrapolation may be used to estimate the
heq value for a design height not listed in Table 7.

4.3. Determine the forces and moments at the toe of

the reinforced soil mass

Unfactored forces and bending moments are first

determined at the toe (O, Fig. 1) of the reinforced soil

Fig. 5. Continued. Table 7

Suggested heq values for LRFD method

4.2. Compute the live-load surcharge (LS) and the Wall height (mm) heq (mm)
maximum acceleration coefficient at the centroid of the
mass (Am) ⱕ 1500 1700
3000 1200
6000 760
ⱖ 9000 610
LS ⫽ ka␥bheqh, (10)

Fig. 6. Active earth zone and lateral earth pressure diagram for the pullout analysis.
 Y. Chen / Engineering Structures 22 (2000) 793–808 803

mass, followed by the factored ones if the LFD or LFRD

method is chosen. The following limit state or load
group is associated with an earthquake: extreme event-
I limit state for the LRFD (Table 1), and load group VII
for the LFD (Table 4) and the ASD (Table 6).

4.4. Check the external stability for the reinforced soil

mass

The examination of the external stability for the

reinforced soil mass must be done for all applicable limit
states by assuming the mass as a ‘rigid body’. In the
stability computation, live load surcharges should be
applied from a vertical plane beginning at the back of
the reinforced zone for pullout calculations, while it
should be deemed to extend over the reinforcement zone
for maximum stress calculations [3,5].
Furthermore, in addition to the static forces stability
Fig. 7. Location of EQEA and EQIR.
determinations should be made by including the possible
seismic forces: EQIR (horizontal inertial force), and 60%
(ASD and LFD) or 50% (LRFD) of EQAE (dynamic hori-
B
zontal thrust). EQAE can be evaluated using the pseudo- e⫽ ⫺ x0 (15)
static Mononabe–Okabe method provided that the soil is 2
not liquefied.
For a MSE wall with the broken or sloping backfill where B is the width of the soil mass (L, as defined in
surface condition (Figs. 2 and 3), the seismic forces the above), and x0 is the location of the vertical load at
should be determined based on an effective mass having the base of the mass measured from the toe (O, Fig. 1),
a height H2 and a base width equal to 0.5 H2 where H2 which can be found from:
is defined by:
兩Mv ⫺ Mh兩
0.5Htani H x0 ⫽ for the ASD and the LFD, (16a)
V
H2 ⫽ H ⫹ ⫽ . (12)
1 ⫺ 0.5tani 1 ⫺ 0.5tani
or
EQIR and EQAE are computed by [3]:
兩Muv ⫺ Muh兩
EQIR ⫽ 0.5Am␥rH22, (13) x0 ⫽ for the LRFD (16b)
Vu
and
where V and Vu are the unfactored and factored vertical
EQAE ⫽ 0.375Am␥rH22 (14) loads, Mv and Muv the unfactored and factored moments
due to vertical loads, and Mh and Muh are the unfactored
where ␥r is the unit weight of reinforced backfill. and factored moments due to horizontal loads, respect-
EQAE should be applied to the back surface of the ively.
reinforced fill at the height of 0.6 H2 from the base, and In the LRFD method, the overturning is checked by
EQIR at the height of 0.5 H2 (i.e. mid-height) of the [3]:
effective soil mass, Fig. 7.
For the horizontal backfill surface condition (Fig. 4), e ⱕ B/4 for a mass resting on soil, (17a)
H2 ⫽ H.
or
5. Check the safety against the possible failure
e ⱕ 3B/8 for a mass resting on rock. (17b)
modes (subflowchart 5b)

5.1. Overturning In the ASD and LFD methods, instead the overturning
is checked by [2]:
To check overturning, the eccentricity of the vertical
load off the centerline of the base of the reinforced soil Mv
ⱖ FSover, (18a)
mass, e, must be determined first: Mh
804 Y. Chen / Engineering Structures 22 (2000) 793–808

and 5.3. Bearing

e ⱕ eo (18b) The bearing equations given in the AASHTO LRFD

[3] are not so convenient for use. Instead, the equations
where FSover is the factor of safety against overturning, presented in this section are suggested.
and eo is a certain limit for e. The design bearing strength, qr, is determined by:
For a mass resting on soil, FSover is equal to 2.0 and
eo is equal to B/6. While for a mass resting on rock, they qr ⫽ ␾brgqult ⱖ qu (21)
are equal to 1.5 and B/4, respectively [2]. However,
when seismic loading is present FSover may be reduced where qu is the factored bearing pressure, ␾brg the resist-
to 75% (in this case, FSover ⫽ 1.5 for soil condition or ance factor for bearing (0.45) [3], and qult is the ultimate
1.125 for rock condition) [2]. It is noted that in the ASD bearing resistance that can be determined from the fol-
and LFD methods unfactored loads should be used for lowing general equation for foundations on soil:
stability computation.
qult ⫽ 0.5␥B⬘N␥s␥i␥ ⫹ cNcscic (22)
5.2. Sliding
⫹ (q0 ⫹ ␥Df)Nqsqiq
The following equations are used to investigate the
sliding of the reinforced soil mass: where N␥, Nc and Nq are the bearing capacity factors, s␥,
sc and sq the shape factors, and i␥, ic and iq the inclination
Qr ⫽ ␾␶Q␶ ⫹ ␾epQep ⱖ Hu for the LRFD, (19a) factors for the applied load, ␥ an effective unit weight
for the soil, B⬘ the effective width (B ⫺ 2eB, eB the
and eccentricity of the resultant load off the center of dimen-
sion B), c the soil cohesion, Df the soil cover depth in
Qall ⫽ fV ⱖ 1.5Hunfac for the ASD and the LFD (19b) front of the wall (Fig. 1), and q0 is the applied sur-
charge load.
where Qr is the design sliding strength, Qall the allowable For foundations on rock, qult can be estimated by the
sliding resistance, Hu the factored horizontal load, Hunfac following semi-empirical formulae [6]:
the unfactored horizontal load, f the friction coefficient,
Qep the nominal passive resistance from the earth qult ⫽ (√s ⫹ m√s ⫹ s)1/2Co (23)
material in the vicinity of the toe (O, Fig. 1), ␾ep the ⫽ NmsCo
resistance factor related to Qep, ␾␶ the resistance factor
related to Q␶ (␾␶ ⫽ 0.90) [3], and Q␶ is the nominal
where s is a rock mass quality parameter, m is also a
shear resistance which can be computed by:
rock mass quality parameter, Nms is a coefficient for esti-
mating qult for rock, and Co is the uniaxial compressive
Q␶ ⫽ Vutan␦, (20a)
strength of rock.
If ␾s ⬍ 37°, then ␥ in Eq. (22) can be determined by
where
one of the following three equations depending upon zw
(the depth from the base of the investigated mass to the
tan␦ ⫽ 0.8tan␾s (20b)
highest anticipated groundwater level, positive
downward):
where ␾s is the smaller angle of friction of random back-
fill and the soil underneath the foundation.
␥ ⫽ ␥m for zw ⱖ B, (24a)
It should be noted that Qep in Eq. (19a) is usually
neglected in practice for conservatism—the front soil
zw
cover is only 300–600 mm resulting in a small Qep value, ␥ ⫽ ␥⬘ ⫹ (␥ ⫺ ␥⬘) for zw ⬍ B, (24b)
and is likely to be removed for maintenance/repairing B m
reasons. In Eq. (19b), the constant of 1.5 is the factor
of safety against sliding (FSslid) [2], and f should be and
evaluated based on ␾s. In general, f is taken as tan␾s for
concrete cast against soil or 0.8 times tan␾s for precast ␥ ⫽ ␥⬘ for zw ⱕ 0 (24c)
concrete [3]. In the absence of specific strength data for
the foundation material, a maximum value of 30° may where ␥⬘ is the buoyant unit weight of soil (saturated
be assumed for ␾s. Similar to FSover, FSslid may also be unit weight of random backfill minus unit weight of
reduced to 75% when there is seismic loading water), and ␥m is the moist unit weight of reinforced
involved [2]. backfill.
 Y. Chen / Engineering Structures 22 (2000) 793–808 805

On the other hand, if ␾s ⱖ 37°, then ␥ can be com-

puted by: 冉
i␥ ⫽ 1 ⫺
Hunfac
V 冊 2
(28b)

zw␥m ␥⬘ and
␥ ⫽ (2D ⫺ zw) 2
⫹ 2 (D ⫺ zw)2, (25a)
D D
1 ⫺ iq
ic ⫽ iq ⫺ for ␾s > 0 (28c)
where D is defined by: Nctan␾s

D ⫽ 0.5Btan2(45° ⫹ ␾s/2). (25b) or

Several investigators have developed the bearing Hunfac

capacity factors involved in Eq. (22). However, the fol- ic ⫽ 1 ⫺ for ␾s ⫽ 0. (28d)
BL⬘cNc
lowing equations are recommended.
For foundations on soil with general shear behavior, Earthquake loads should be excluded from Hunfac
appearing in Eq. (28a), (28b) and (28d). In determining
Nq ⫽ e␲tan␾tan2(45° ⫹ ␾s/2), (26a) the bearing resistance of the geological material underly-
ing the reinforced soil mass, an equivalent width equal
Nc ⫽ (Nq ⫺ 1)cot␾s for ␾s > 0, ⫽ 2 (26b) to the soil reinforcement length (L) and a length equal
to the wall length are usually assumed in design. Bearing
⫹ ␲ for ␾s ⫽ 0, stresses should be computed using a uniform base press-
ure distribution over the equivalent width of the soil
and mass [11]:
For the ASD and LFD methods, the bearing is
N␥ ⫽ 2(Nq ⫹ 1)tan␾s. (26c) checked by:

If the soil experiences the undesirable local or punch- qall ⫽ qult/FSbrg ⱖ q (29)
ing shear, then Eq. (26a) [7–9], Eq. (26b) [7–9], and Eq.
(26c) [9,10] should be modified by replacing c with 0.67 where qall is the allowable bearing pressure, qult calcu-
c and ␾s with tan−1 (0.67␾s). lated by Eq. (22) or Eq. (23), FSbrg the factor of safety
Based on load tests, the following semi-empirical for- against bearing (3.0 typically) [2], and q is the actual
mulae are recommended for determining the shape fac- average pressure determined by:
tors in Eq. (22):
VTOT VTOT
q⫽ ⫽ (30)
B⬘ Nq B ⫺ 2e 2x0
sc ⫽ 1 ⫹ , (27a)
L⬘ Nc
where VTOT is the total unfactored vertical load.
B⬘
sq ⫽ 1 ⫹ tan␾s, (27b) 5.4. Overall stability and deep-seated shear
L⬘

and Overall stability and deep-seated shear failure should

be evaluated at all applicable limit states or load groups
(Tables 1, 4 and 6) by using a limiting equilibrium
B⬘
s␥ ⫽ 1 ⫺ 0.4 (27c) method such as the Modified Bishop, Janbu, Spencer, or
L⬘ Swedish Circle methods when the wall footing is built
on a slope or when the wall system is loaded with slop-
where L⬘ is the effective length (L ⫺ 2eL,eL ⫽ the eccen- ing surcharge.
tricity of the resultant load off the center of dimension
L).
Note that for continuous walls/footings, B/L ⬇ 0 6. Step 5. Check the settlement of MSE wall
which leads to: s␥ ⫽ sq ⫽ sc ⫽ 1.0. The inclination fac- system (subflowchart 5b)
tors in Eq. (22) can be determined by:
The settlement of the reinforced soil mass, ⌬v, should
Hunfac be evaluated at a limit state or load group similar to
iq ⫽ 1 ⫺ , (28a)
V Service-I limit state in Table 1 regardless of the design
806 Y. Chen / Engineering Structures 22 (2000) 793–808

method chosen. The settlement can be estimated by limit states (LRFD) or loading groups (LFD and ASD).
using the following empirical formula [11,12]: The possible failure modes for soil reinforcements are
pullout and rupture.
qB␮1␮2
⌬v ⫽ (31)
Ms 7.1. Determine Pis, Shi and Hm

where Ms is the soil modulus, ␮1 an influence factor The internal inertia force per unit length of the struc-
depending on the depth-to-width (Df/B) and mass length- ture, Pis, should be considered equal to the mass of the
to-width ratios of the reinforced soil mass, and ␮2 is also active zone times the maximum wall acceleration coef-
an influence factor related to soil depth-to-mass width ficient (Am). Pis should be distributed to the soil
(Hs/B) and mass length-to-width ratios. reinforcements proportionally to their resistant areas,
The tolerable settlement of MSE walls is generally namely:
limited by the longitudinal deformability of the facing
and the ultimate purpose of the structure. Limiting toler- AReff
able differential settlement for MSE wall systems with Hm ⫽ Pis , (32a)
⌺AReff
wall panels less than 2.8 m2 (recommended) in area and
a maximum joint width of 19 mm are presented in Table
and
8. These limits are intended to prevent open spaces in
the joints. Structures subjected to greater differential
settlements could be damaged by concrete spalling from biLei
AReff ⫽ (32b)
panel corners. Where foundation conditions indicate Shi
large differential settlements are possible over a short
horizontal distance, a system of open vertical slip joints where Hm is the incremental dynamic inertia force at
should be provided at suitable intervals or the foundation level i (per length of wall), AReff the effective area of
improved by various ground improvement techniques, reinforcement per vertical dimension, ⌺AReff total effec-
such as over-excavation and replacement with com- tive area of reinforcement, bi the reinforcement width for
pacted backfill using select material. layer i, Lei the effective reinforcement length for layer i,
Special exploration, testing and analyses may be and Shi is the horizontal reinforcement spacing for
required for MSE walls constructed over soft deposits layer i.
since they could be subjected to consolidation and/or lat-
eral flow that would result in unacceptable long-term 7.2. Determine the total factored horizontal forces at
settlements or horizontal movements. soil reinforcement levels (Pi), vertical spacing of soil
The MSE wall is actually a system consisting of flex- reinforcements (S), and tensile forces of soil
ible composites of granular soil and tensile reinforce- reinforcements (Tstrip)
ment, and hence behaves similarly to earth embankments
with vertical or near vertical faces. They exhibit a con- The factored loads at the reinforcement levels due to
siderable degree of freedom to deform without structural soil weight can be precalculated (subflowchart 5c). Then,
distress, and, therefore, bearing resistance, rather than Pi, which is used to evaluate the internal failure by
settlement, generally governs the design. slip/rupture of soil reinforcements, is computed by:

Pi ⫽ ␴hhi, (33)
7. Step 6. Check the failures of soil reinforcements
(subflowcharts 5b and 5c)
and
The examination of the internal stability of the soil
reinforcing elements should be made for all applicable ␴h ⫽ k␥p␴v (34)

where hi is the height of reinforced soil zone contributing

Table 8
Relationship between joint width and limiting distortion of MSE horizontal load to the reinforcement at level i, determ-
wall facing ined as the vertical distance from the mid-point between
layer i and the next overlying layer to the mid-point
Joint width (mm) Limiting vertical distortion between layer i and the next underlying layer, ␴h the
horizontal stress at layer i, ␥p the load factor for vertical
19 1/100
13 1/200 earth pressure EV (Table 2), k a lateral earth pressure
6 1/300 coefficient, and ␴v is the pressure due to the resultant
vertical forces at the reinforcement level determined
 Y. Chen / Engineering Structures 22 (2000) 793–808 807

using a uniform pressure distribution over the effective zone. For seismic loading, the ␾fs factor in Eq. (36) can
width B⬘( ⫽ B ⫺ 2eB) [11]. be reduced to 80% [3].
As shown in Fig. 6, k equals to ko (at-rest earth press-
ure coefficient) at the height above the top of the level- 7.4. Calculate the correction losses, and check the
ling pad (H1) and decreases linearly to ka value at 6 m breakage/yielding of soil reinforcements
level from the top, in which ka is determined by tan2
(45° ⫺ ␾b) [2,3]. Below a depth of 6 m, k remains as Soil reinforcing elements in MSE walls should be
ka, Fig. 6, ko can be taken as 1-sin␾b. It is noted here designed to have a corrosion resistance durability of a
that ka and ko are independent of the external loading minimum life of 100 years for permanent structures. The
conditions. structural design of galvanized metal soil reinforcements
In determining ␴v in Eq. (34), the evaluation of equi- and connections should be made on the basis of a thick-
librium should consider only the forces acting at that ness, te, defined as:
level. The maximum friction angle used to determine the
horizontal force within the reinforced soil mass should te ⫽ tn ⫺ ts (37)
be 34°, unless the select backfill is tested for frictional
strength by triaxial or direct shear testing. Live loads
where te is the thickness of steel reinforcement at the
should be positioned for extreme force effect within the
end of service life, tn the nominal thickness of steel
zone subjected to live loads.
reinforcement at construction, and ts is the sacrificial
Tstrip is computed by:
metal thickness expected to be lost by uniform corrosion
during the service life of the structure.
Pi For structural design, ts is computed for each exposed
Tstrip ⫽ , (35a)
Nstrip surface by considering: (1) the loss of galvanizing is
0.015 mm/year for the first 2 years, and 0.004 mm/year
and for subsequent years; and (2) the loss of carbon steel is
0.015 mm/year (ASD and LFD) or 0.012 mm/year
Nstrip ⫽ 1/Shi (35b) (LRFD) after zinc depletion [3]. Other corrosion-resist-
ant coatings, if specified, should be electrostatically
where Nstrip is the number of metal strips per unit length applied, resin-bonded epoxy type, with a minimum
of wall, and Shi is a horizontal reinforcement spacing application thickness of 0.40 mm.
specified by the design engineer. In determining the requirements for corrosion protec-
tion, the following should also be considered: (a) galvan-
7.3. Determine the design pullout capacity/resistance izing is preferred to the use of epoxy-bonded coatings
of soil reinforcements, PfR because the long-term performance of epoxy-bonded
coatings buried in soil is unknown; (b) if used, epoxy
The pullout resistance of soil reinforcements should coatings should have a minimum thickness of 0.38–0.46
be evaluated at each level using only the effective pul- mm when placed in granular fills with sharp angular
lout length (Fig. 6), which extends beyond the theoretical fragments; (c) the sacrificial thickness should be pro-
failure surface. For inextensible reinforcing elements, vided in addition to the epoxy coating; and (d) alloys,
PfR can be determined by: such as aluminum and stainless steel, should not be used
as the soil reinforcements.
PfR ⫽ ␾fsPfs ⫽ ␾fsf*␥rz⬘As (36) The breakage/yielding of soil reinforcements is
checked by:
where Pfs is the nominal pullout resistance (0.67 for the
ASD and the LFD, or 0.90 for the LRFD) [2,3], ␾fs is ␴t ⱕ ␴all or ␴r ⫽ minimum of ␾yFy and ␾frFu, (38)
the resistance factor, f* is the apparent coefficient of fric-
tion at each reinforcement level, z⬘ is the vertical coordi- where ␴t is the actual tensile stress, ␴all is the allowable
nate of a soil reinforcement measured from the top of tensile stress, ␴r is the design tensile strength, Fy is the
wall, and As is the total top and bottom surface area of yielding strength, Fu is the minimum tensile strength, ␾y
a soil reinforcement along the effective pullout length is the the resistance factor for yielding (0.55 for the ASD
less the sacrificial thickness. and the LFD, or 0.85 for the LRFD) [2,3], and ␾fr is the
In the absence of pullout data, a maximum value of resistance factor for fracture (0.50 for the ASD and the
1.5 (ASD and LFD) or 2.0 (LRFD) for f* should be LFD, or 0.70 for the LRFD) [2,3].␴t is determined by:
taken at ground level, and may be assumed to decrease
linearly to a value of tan␾r at a depth of 6 m, where ␾r ␴t ⫽
Tstrip
, (39a)
is the friction angle of the backfill within the reinforced Acorr
808 Y. Chen / Engineering Structures 22 (2000) 793–808

and Washington, DC: American Association of State Highway and

Transport Officers, 1996.
[3] AASHTO. LRFD bridge design specifications, 1st edn. Wash-
Acorr ⫽ tewe (39b) ington, DC: American Association of State Highway and Trans-
port Officers, 1994.
where Acorr is the corrected cross-sectional area of strip, [4] Nowak AS. Calibration of LRFD bridge design code. Final
and we is the width of metal strip at the end of service Report. Ann Arbor, MI: Department of Civil and Environmental
Engineering, The University of Michigan, 1993.
life. [5] Mitchell JK, Villet WCB. Reinforcement of earth slopes and
embankments. National Cooperative Highway Research Program
Report 290. Washington, DC: Transport Research Board, 1987.
8. Conclusion [6] Carter JP, Kulhawy FH. Analysis and design of drilled shaft foun-
dations socketed into rock. Report EL-5918. Prepared by Cornell
University for Empire State Electrical Energy Research Corpor-
Three different design methods and the important ation and Electrical Power Research Corporation, 1988.
design/construction considerations for MSE wall sys- [7] Prantl L. Spannungsverteilung in plastichen Koerpern. Proceed-
tems are described. The general analysis/design pro- ings of the First International Congress on Applied Mechanics.
cedure for MSE walls is discussed in detail. The effects Delft: Technische Boekhandel en Druckerij, 1925:43–54.
[8] Mendelson A. Plasticity: theory and application. New York: The
of these design methods, particularly the impact of the
Macmillan Co., 1970.
new LRFD method on MSE wall design, are discussed [9] Hunt RE. Geotechnical engineering techniques and practices.
in the companion paper (Part II) through the use of New York: McGraw–Hill Inc., 1986.
realistic examples. [10] Caquot A, Kerisel J. Tables for the calculation of passive press-
ure, active pressure and bearing capacity of foundations, trans-
lated by Bec MA. Paris, France: Gauthier–Villars, 1948.
[11] Barker RM, Duncan JM, Rojiani KB, Ooi PSK, Tan CK, Kim
References SG. Manuals for the design of bridge foundations. National
Cooperative Highway Research Program Report 343. Wash-
[1] Christopher BR, Holtz RD. Geotexile engineering manual. ington, DC: Transport Research Board, 1991.
FHWA Report. Washington, DC: Federal Highway Adminis- [12] D’Appolonia DJ, D’Appolonia E, Brisette RF. Settlement of
tration, 1985. spread footings on sand (closure). J. Soil Mech. Found., ASCE
[2] AASHTO. Standard specifications for highway bridges, 16th edn. 1970;96:754–61.