PAPER

Embedded Retaining Wall Design

– Engineering or Paradox?
A personal viewpoint by A.Y. Chmoulian, associate at Royal Haskoning

Introduction stresses are lower than those in the General design conditions but to allow
Retaining wall design theory is a working conditions. The analyses presented in this paper modelling of both the elastic and
complicated subject with a long his- The amended edition of BS8002 represent real design situations. plastic soil behaviour, most of the
tory. The problem was, for a long states that the earth pressures at But to make the effect of different calculations for this paper were
time, solved exclusively in terms ultimate limit state are actually design parameters clearer, only performed using a standard industry
of Coulomb’s approach, that is, lower than those under working simple geometries are presented, software suite that allows modelling
through equilibrium calculations conditions. with the varying factors applied non-linear soil structure interaction
based on post-failure soil pressures. It is noted that the British Standards independently. The basic initial with coupled subgrade reactions
The geometry of the wall reflect- for structural design, for example conditions always implied that the (Software Suite A). References to
ing its equilibrium could then be BS5950 [1] and BS8110 [2], have soil conditions are uniform and all the results attained through other
enhanced to create a desired mar- also introduced additional partial structural elements are absolutely types of numerical modelling are
gin of safety on wall stability. In a factors of safety on loads calculated stiff, although the effect of finite given where necessary.
similar way, direct factoring of the using the BS8002 design approach. wall stiffness was also studied. All of the analyses have first
calculated stresses would ensure the This, in the author’s experience, The groundwater pressures were been carried out using the factored
necessary margin of wall structural sometimes causes substantial ignored here to ensure that the sets of soil parameters to establish
strength. increases in the overall structural primary effects caused by the soil required wall embedment. The
The main disadvantage of this strength demand compared with the pressures are clearer. factored analyses would also
method was its oversimplified traditional (unfactored) retaining It is a normal design practice that give the structural design forces
approach to the soil-structure wall design approach. the majority of retaining wall analyses consistent with BS8002 or with EC7
interaction. However, most of While it is logical to accept within a single project are carried out [4]. Analyses were then repeated
the sophisticated methods of soil- that soil pressure reduces with the using the same design software. It using the same wall geometry
structure interaction analysis were increasing wall deflection, it is would be unusual in a routine design but with the representative soil
not capable of dealing with the more difficult to comprehend that to find the same engineer using parameters. This would indicate the
conventionally used definitions of a structure can deflect less under different analysis methods when structural strength demand using
factors of safety on wall stability. greater earth pressures. carrying out the ultimate limit state the traditional design approach. It
Hence, the advances in numerical This paper will show examples (ULS) calculations using factored would also give the designer’s best
analysis required modification of from the author’s experience, soil parameters, compared with estimate of the real situation, which
the safety factor approach to ensure representing just a small sample the serviceability limit state (SLS) should attract a load factor when
compatibility of the results. of possible effects that may be calculations using representative using structural design codes.
An easy way forward was available caused by the use of factored soil parameters. When factoring the soil
through factoring soil strength and parameters for designing embedded As the analyses presented here parameters in accordance with
was often adopted. The idea is very retaining walls. are supposed to represent normal BS8002 it was assumed that
simple and clear for wall stability: the critical state effective stress
if the soil strength is reduced by a The following notations are used here: strength is greater than the factored
factor and the wall is still standing Subscripts “f” and “r” refer to any results of representative strength value.
– the problem is solved. “f” and “r” the analysis using respectively a factored or a Otherwise, the findings presented
The problem becomes less clear representative set of design parameters may be exaggerated even further, as
when dealing with structural forces the ratio between the factored and
induced in the retaining wall by soil ø Internal friction angle of soil the representative values increases.
pressures. Common sense suggests Where references are made to
Adhesion angle on the interface
that reducing the soil strength would δ BS8002 δ-values it was assumed
between soil and retaining wall
cause consequential increases in soil that δr = 2/3ør in the analysis using
pressures on the interface with the d Retaining wall embedment representative soil parameters and
wall and therefore in the structural tan(δf) = 0.75tan(øf) for the factored
forces, providing the ultimate design h Retaining wall height, measured from top to toe soil parameters.
loading conditions. This idea was D Excavation depth The effect of factoring soil
suggested in the original (1994) parameters is addressed here in the
edition of BS8002 [2]. Respectively, active, passive and at- form of ratios Am and Ar, which
Ka, Kp, Ko
But it was noticed that when rest earth pressure coefficients gives an indicative comparison of
factored soil parameters are used, the structural strength demand when
the stability requirement is for the M Maximum calculated bending moment in the retaining wall using different design approaches.
external forces to be in equilibrium R Maximum calculated prop force Issues related to the analysis of
only, that is by calculation the wall retaining wall stability are avoided
may actually be as near to the point Ratio between maximum bending moments calculated in if possible, although it should be
Mf
of failure as possible. Am = retaining structures of the same geometry using factored acknowledged that different design
Mr (Mf) and representative (Mr) sets of soil parameters
For a cantilever wall or a wall approaches, for example, J.B.
with a single prop near the top Burland and D.M. Potts (1981), may
Ratio between maximum prop forces calculated for
this may mean that the soil on the Rf result in different wall embedment
Ar = retaining structures of the same geometry using factored
interface with the wall has reached Rr requirements, which can ultimately
(Rf) and representative (Rr) sets of soil parameters
the point of failure and so interface affect the structural design of the

GROUND ENGINEERING JULY 2007 31

[Link] 31 22/6/07 12:58:34

 PAPER
δ-values
Partial factor 1.46

Am
0˚ 2/3 ø ø BS8002 δ=0 δ = 2/3ϕ δ=ϕ
1.2 1.406 ± 0.02 1.396 ± 0.033 1.376 ± 0.03 1.306 ± 0.044 1.45
1.25 - 1.498 ± 0.064 1.491 ± 0.057 -
1.44
Table 1: Am ratios for different analysis parameters
1.43

wall. of 1.25, as per Table A.2 of EC7. It 1.42

The size of this paper does not can be seen that bending moments
permit inclusion of all available for both representative and factored 1.41
data. In particular, only uniform sets of parameters are dependent on 1.40
granular soil deposits are discussed. the soil strength and on the adhesion
It should be noted that the factor. However, the curves are very 1.39
20 25 30 35 40
design is very sensitive to the wall similar in shape and as a result the
ϕ, degrees
embedment; the calculated value Am ratios are virtually independent
was always rounded up to the of the varied factors as shown in
Figure 1: Errors arising from scaling Ka and Kp values from charts
nearest 100mm rather than the usual Table 1.
0.5m to 1m. Therefore, the designer can expect
7500

Bending moment, kNm/m run

The results do not always align good predictability of the results Mf (δ = 0)
into smooth curves as they, like all whether the representative (or best 6500
Mr (δ = 0)
numerical analyses, contain different guess) or the factored soil strength Mf (δ = 2/3ϕ)
Mr (δ = 2/3ϕ)
types of small random errors due to design is carried out in uniform 5500 Mf (δ = ϕ)
various numerical inaccuracies. ground conditions. The Table 1 Mr (δ = ϕ)
4500 Mf (BS8002)
An example in Figure 1 shows values do not apply to layered soils. Mr (BS8002)
a plot of Am versus ø-value for a
3500
cantilever retaining wall in a uniform Effect of overexcavation on
granular deposit. The values are cantilever walls 2500
hand calculated using a simple Figures 3 and 4 show the effect
approximation of perfectly plastic of overexcavation on the analysis 1500
soil behaviour, with the maximum results for absolutely stiff retaining
500
bending moment corresponding to walls. The depths of overexcavation 20 25 30 35 40
the zero shear point. were selected as the lesser of 0.1D or ϕ, degrees
The results of these calculations 0.5m and the overexcavations were
are not dependent on the depth of the only included in the analysis using Figure 2: Maximum bending moments for a cantilever retaining wall
excavation and the calculation errors the factored soil parameters, which with 9m upstand
are only caused by the inaccuracy of is consistent with BS8002. For all
reading the Ka and Kp values from equations here, δ = ø was used. 5000
Bending moment, kNm/m run

Mf (without overexcavation)
the BS8002 charts. Naturally, Am- As can be seen, the effect of 4500
values for δ = 0 were calculated overexcavation depends on the Mf (with overexcavation)
4000
directly and contain no error so they depth of the main excavation, with Mr (with or without overexcavation)
3500
are aligned horizontally. It can be the Mf-values between 55% and 80%
seen that a few per cent calculation greater than the Mr-values for the 3000
error is possible just through the same wall geometry of absolutely 2500
rounding of chart data. stiff walls. This compares with 2000
about 38% when modelling the wall
1500
To “design what is meant” or without overexcavation.
to “mean what is designed”? 1000
Effect of embedment depth on 500
20 25 30 35 40
General analysis of cantilever walls cantilever walls
ϕ, degrees
A large sample of analyses was Figure 5 shows the effect of extra
carried out for a simple cantilever depth of embedment, in excess
retaining wall. The analysis covered of the stability requirement, on Figure 3: Effect of overexcavation for a cantilever retaining wall
various excavation depths from Am-values for stiff walls. As the with 9m upstand
3m to 12m, ø-values from 20º to embedment depth increases, the
1.9
Am

40º, wall stiffnesses ranging from wall gradually approaches the fixed
2 x 104 kN.m2/[Link] (conversion earth condition. As a result, earth 1.8
is difficult to achieve for smaller pressure on the active side becomes 1.7
stiffnesses) to the virtual infinity closer to the at rest value for both 1.6
and soil stiffnesses from 10MPa to representative and factored analyses 1.5
100MPa. and Am-ratio reduces. But, this 1.4
The values of Ko = 0.6 were used effect develops very slowly and is 1.3
in all analyses. unlikely to affect the results unless No overexcavation (all depths)
1.2
A sample of analysis results for the embedment is substantially Overexcavation (D = 3m)
1.1 Overexcavation (D = 6m)
9m deep excavations in soils of greater than is necessary.
1.0 Overexcavation (D = 9m)
different strength is presented in Naturally, the effect caused Overexcavation (D = 12m)
Figure 2. These curves are based by extra wall embedment will be 0.9
20 25 30 35 40
on the analyses assuming a partial reduced for walls of finite stiffness.
ϕ, degrees
factor of 1.2 on the ø values. The analysis results presented
A very similar plot can be above for cantilever walls do not Figure 4: Effect of overexcavation on Am ratios for a cantilever
produced for the partial factor value represent anything unexpected. retaining wall

32 GROUND ENGINEERING JULY 2007

[Link] 32 22/6/07 12:59:42

 The picture changes, however, the analysis. This does not allow
9000

Bending moment, kNm/m run

when dealing with tied or propped a direct comparison between the Mf (δ = 0)
retaining walls. results for Software Suites A and Mr (δ = 0)
8000 Mf (δ = 2/3ϕ)
B. Nevertheless, the comparison Mr (δ = 2/3ϕ)
General analysis of propped walls of bending moments calculated for 7000 Mf (δ = ϕ)
Figures 6 and 7 show, respectively, representative and factored designs 6000
Mr (δ = ϕ)
Mf (BS8002)
bending moments and strut forces was confirmed by Software Suite B Mr (BS8002)
for an absolutely stiff wall with analysis, as shown in Figure 10. 5000
an absolutely stiff prop at the top. It can be seen that factored
4000
These analyses were based on a soil analysis yields higher values of
strength factor of 1.2. The use of bending moments for a relatively 3000
absolutely stiff structural elements weak soil (ø = 25º), whereas for
2000
allows avoiding distortions of stronger soil (ø = 40º) representative
diagrams that may potentially be bending moments were marginally 1000
caused by the secondary effects. higher than those from the factored 20 25 30 35 40
It can be seen that the plots are analysis. ϕ, degrees
similar for different δ-values, but, A sensitivity analysis for the wall
Figure 6: Bending moments diagram for a tied retaining wall with
significantly, for soils of greater embedment can help to explain the
15m upstand
strength the forces calculated in apparent paradox of greater soil
the factored analyses are smaller strength causing greater structural
than those from the representative loads. This was carried out using Strut force, kN/m run 1000
Rf (δ = 0)
analyses. It looks as if an increase in Software Suite A, assuming an 900 Rr (δ = 0)
Rf (δ = 2/3ϕ)
the soil strength causes an increase absolutely stiff wall and using soil 800 Rr (δ = 2/3ϕ)
in the forces in the retaining wall, parameters as above and δ = ø. The Rf (δ = ϕ)
700 Rr (δ = ϕ)
which sounds as paradoxal as results are presented below. Rf (BS8002)
“Hambly’s stool”. 600 Rr (BS8002)
As a result, the respective Am and Effect of embedment depth on 500
Ar ratios vary from about 1.2 for propped walls 400
the weaker soils to 0.8 to 0.9 for the A plot of maximum bending 300
stronger soils, as shown on the plot moments calculated for different
200
in Figure 8. wall embedments is presented
A similar set of results is shown in in Figure 11. It can be seen that 100
20 25 30 35 40
Figure 9 for the partial soil strength relatively small increases in the
ϕ, degrees
factor of 1.25, as per EC7. The wall embedment cause significant
amplitude of variation of Am and Ar increases in the bending moments. Figure 7: Strut forces diagram for a tied retaining wall with 15m
ratios is just slightly greater than for This effect, which is particularly upstand
the BS8002 factor, but the overall transparent for the soils of greater
pattern remains the same. strength, is caused by relative 1.3
Am and Ar

To demonstrate that the above stiffening of the embedded part of Am (δ = 0)

Am (δ = 2/3ϕ)
effects were not caused by software the wall. This effect occurs similarly Am (δ = ϕ)
1.2
error, a numerical analysis of a to the cantilever walls but is much Am (BS8002)
Ar (δ = 0)
similar retaining wall supporting more pronounced for the propped Ar (δ = 2/3ϕ)
a 15m excavation was carried out walls. A stiffer toe response would, 1.1 Ar (δ = ϕ)
Ar (BS8002)
using standard industry numerical naturally, reduce wall deflections
analysis software using finite and increase the bending moments. 1.0
differences approach (Software A plot of the respective strut forces
Suite B). is very similar and is not presented 0.9
The wall was modelled with an here.
absolutely stiff support at the top. Therefore, the phenomenon of 0.8
As numerical analysis software increased soil strength resulting in 20 25 30 35 40
cannot properly converge to a greater structural forces is caused by a ϕ, degrees
solution for an absolutely stiff wall, combination of effects of the
a reduced stiffness was used in increased soil strength and the Figure 8: Am and Ar ratios for a tied retaining wall (partial factor1.2)

1.4
Am and Ar

1.5 Am (δ = 2/3ϕ)
Am

Am (δ = ϕ)
1.3
1.4 Ar (δ = 2/3ϕ)
Ar (δ = ϕ)
1.2
1.3

1.2 1.1

1.1 1.0
Embedment = d
Embedment = 2d 0.9
1.0
Embedment = 3d
Embedment = 4d 0.8
0.9
20 25 30 35 40 20 25 30 35 40
ϕ, degrees ϕ, degrees

Figure 5: Effect embedment depth on Am ratios for a cantilever Figure 9: Am and Ar ratios for a tied retaining wall
reataining wall (partial factor1.25)

GROUND ENGINEERING JULY 2007 33

[Link] 33 22/6/07 13:02:23

 PAPER
relatively increased wall embedment. including an overexcavation in the Bending moment, kNm/m run
Had two different walls been analysis. Although it was carried out -2000 0 2000 4000 -1000 0 1000 2000
analysed separately, using either for an absolutely stiff wall, similar 0 0

Depth, m
factored or representative strength effects will be encountered for walls Mr Mr
-2
for both stability and structural of finite stiffness. It can be seen that Mf -2 Mf
loads calculations, the one using the Mf values do indeed increase as a -4
representative soil strength would result of overexcavation. But due to -4
have a smaller embedment and the excess embedment, Mr values -6
smaller design bending moments. increase to an even greater extent. -6
-8
The above effect seems only to be Figure 17 shows the respective
encountered when using software ratios Am and Ar for different -10 -8
capable of numerical modelling overexcavation scenarios. There is
-12
of soil-structure interaction. The very little variation of the ratio values -10
software suites that use a simplified for different depths of excavation -14
analysis approach, even those based checked here, that is, within the range -12
on non-coupled subgrade reaction of 9m to 24m upstands. Comparing -16
analysis, just show a conventional the Am and Ar ratios for analyses -14
-18
increase in structural forces with with and without overexcavation,
reduced soil strength. Figure 12 it can be seen that introduction -16
-20
shows the effect of extra wall of overexcavation increases the (a) (b)
-22 -22
embedment on Am and Ar. It can be relative factored structural forces for
seen that for wall embedment just weaker soils. For stronger soils the Figure 10: Bending moment diagrams from Software Suite B
twice the design value, the ratios are effect is the opposite – introduction analysis: (a) for uniform soil strata with ø = 25°, (b) for uniform soil
very close to 1. of overexcavation causes Am strata with ø = 40°
Naturally, as wall stiffness and Ar ratios to reduce further
reduces, Am and Ar ratios become compared with the results without
Bending moment, kNm/m run 8000
Mf (Embedment = d)
less dependent on the relative overexcavation. 7000 Mr (Embedment = d)
increase in the wall embedment. In the same way, structural Mf (Embedment = 1.1d)
Mr (Embedment = 1.1d)
It should be noted that the relative forces developing during gradual 6000 Mf (Embedment = 1.25d)
wall stiffness depends on the depth excavation of soil on the passive Mr (Embedment = 1.25d)
5000 Mf (Embedment = 2d)
of excavation. Figure 13 shows side of the retaining wall may be Mr (Embedment = 2d)
the effect of wall stiffness on the greater than when the excavation 4000
calculated M-values for a propped is completed. This is only relevant
retaining wall supporting a 15m for relatively stiff walls in stronger 3000
deep excavation. soils, but ignoring this factor in 2000
It can be seen from Figure 14 the design cases when it is actually
that for a sufficiently soft wall Am is applicable will certainly not increase 1000
20 25 30 35 40
about 1.4 to 1.5, that is, similar to the conservatism of the design.
ϕ, degrees
the values calculated for cantilever Similarly to the cantilever wall
walls. The Am values for soft walls analyses, sensitivity to soil stiffness Figure 11: Effect of embedment depth on bending moments for a
do not seem to be affected by the was checked for E-moduli range tied retaining wall with 15m upstand
soil strength or by the depth of from 10MPa to 100MPa and no
excavation. effect on the analysis results was 1.3
Am and Ar

Rf/Rr (Embedment = d)
To put the above sensitivity encountered. Rf/Rr (Embedment = 1.1d)
analysis into perspective of real No sensitivity analysis for strut Rf/Rr (Embedment = 1.25d)
1.2 Rf/Rr (Embedment = 2d)
walls, Figure 15 shows a sensitivity stiffness is presented here. But this Mf/Mr (Embedment = d)
analysis for wall stiffness. Stiffnesses factor rarely causes a significant 1.1
Mf/Mr (Embedment = 1.1d)
Mf/Mr (Embedment = 1.25d)
of some real walls are shown on the effect on the bending moments. The Mf/Mr (Embedment = 2d)
same plot for comparison. It should greatest effect is normally caused by
1.0
be noted that AZ12 sheet pile introduction of pre-stressed anchors,
section is the lightest available from which increases the stiffness of the
its manufacturer. system, thus reducing the deflections 0.9
It can be seen that Am-ratios and increasing the anchor forces.
may be close to or even less than 1 This is similar to the way it works in 0.8
20 25 30 35 40
for quite real structures. Strut force pre-stressed concrete.
ϕ, degrees
plots are similar and not presented
here. Discussion & concluding remarks Figure 12: Effect of embedment depth on Am and Ar ratios for a tied
This phenomenon can be found in When designing a retaining wall retaining wall
real design situations. For example, the designer would be concerned
Bending moment, kNm/m run

6000
the author has been involved to provide a safe as well as efficient Md (all stiffnesses)
Mr (stiffness 2 x 10^8)
in reviewing designs in ground design. 5000 Mr (stiffness 2 x 10^7)
conditions comprising sands over It is also usually desirable to Mr (stiffness 2 x 10^6)
Mr (stiffness 2 x 10^5)
completely weathered sandstone, ensure that differences between 4000 Mr (stiffness 2 x 10^4)
where an increase in design values various designs of the same structure
of sandstone strength caused an done by different designers are not 3000
increase in the design bending just governed by their willingness to
2000
moments. take risks.
Based just on the limited analysis 1000
Effect of overexcavation on propped presented in this paper, it is possible
walls to evaluate the range of structural 0
20 25 30 35 40
The above effect will be exaggerated design forces that are implied by
ϕ, degrees
by other factors increasing the current design codes. This is only
relative embedment of the wall. a sample of the possible design Figure 13: Effect of wall stiffness on bending moments for a tied
Figure 16 shows the effect of outcomes. retaining wall with 15m upstand

34 GROUND ENGINEERING JULY 2007

[Link] 34 22/6/07 13:00:19

 1.6
Am

6000

Bending moment, kNm/m run

Mf (No overexcavation)
1.4 5500
Mr (No overexcavation)
5000 Mf (Overexcavation)
1.2
4500 Mr (Overexcavation)
1.0
4000
D = 6 (stiffness 2 x 10^8)
0.8 D = 15 (stiffness 2 x 10^8) 3500
D = 24 (stiffness 2 x 10^8)
0.6 D = 6 (stiffness 2 x 10^7)
3000
D = 15 (stiffness 2 x 10^7) D = 6 (stiffness 2 x 10^5) 2500
0.4 D = 24 (stiffness 2 x 10^7) D = 15 (stiffness 2 x 10^5)
D = 6 (stiffness 2 x 10^6) D = 24 (stiffness 2 x 10^5) 2000
0.2 D = 15 (stiffness 2 x 10^6) D = 6 (stiffness 2 x 10^4) 1500
D = 24 (stiffness 2 x 10^6) D = 15 (stiffness 2 x 10^4)
0 1000
20 25 30 35 40 20 25 30 35 40
ϕ, degrees ϕ, degrees

Figure 14: Effect of wall stiffness on Am ratios for a tied retaining Figure 16: Effect of overexcavation on bending moments for a tied
wall with 15m upstand retaing wall with 15m upstand

Sheet pile wall AZ12 1.2m thick RC diaphragm wall

Am and Ar
Contiguous RC pile wall d600mm at 900mm c/c RC T-diaphragm wall: flange 4x1m, web 3.6x1m 1.4
1.6 Am (No overexcavation - all depths)
Am

D = 6m (ϕ = 20˚) D = 15m (ϕ = 40˚) 1.3

1.5 D = 6m (ϕ = 40˚) D = 24m (ϕ = 20˚) Am (Overexcavation - all depths)
D = 15m (ϕ = 20˚) D = 24m (ϕ = 40˚) Ar (No overexcavation - all depths)
1.4 1.2
Ar (Overexcavation - all depths)
1.3
1.1
1.2
1.1 1.0
1.0
0.9
0.9
0.8 0.8
0.7 0.7
1.00E 1.00E 1.00E 1.00E 1.00E 1.00E 20 25 30 35 40
+04 +05 +06 +07 +08 +09
ϕ, degrees
Wall stiffness, kN m^2

Figure 15: Am ratios versus wall stiffness for a tied retaining wall Figure 17: Effect of overexcavation on Am and Ar ratios for a tied
retaining wall
BS5950 (1) (Clause BS8110 (3) (Table 2.1 without BS8110 (Table 2.1 with
2.2.4) Yf = 1.2 overexcavation) and BS5950 (Table overexcavation) Yf = 1.0 on earth Table 2: Ratio of the structural design moments to the
s

on nominal loads
determined in
2) Yf = 1.2 on earth pressures
obtained from BS8002 including
pressures obtained from BS8002,
when unplanned excavation is
serviceability moments (soil strengh factor M=1.2)
accordance with CP2 appropriate mobilisation factors included in the calculation
For a cantilever 1.55-1.70, depending 1.55-1.80, depending environment when retaining wall ● Any conservative changes made
retaining wall, for on the δ-value on excavation depth
a wide range of design can be done by engineers to the geometry of the structure
input parameters
and managers with a very basic after the design is completed are not
For very stiff 1.00-1.25, depending on the 0.75-1.30, depending understanding of geotechnics, who causing an increase in the structural
1.4
retaining walls ø and δ-values, but converges on the excavation depth
propped near to about 1.2 for walls with (variation of δ-values not just follow the published design forces (this includes, for example,
the top excessively large embedment addressed in this paper) recommendations. This creates a increasing embedment of the wall,
For propped wall 1.00-1.80, depending Not addressed in this paper large variety of designs, some of its cross-section).
of finite stiffness on the ø-values
which may be more pragmatic than ● The conservative assumptions
The Table 2 values are quoted Using this value, the design desired. made with regard to, for example,
directly from the results presented engineer will find the required It is hoped that the above analysis soil strength or construction staging
in this paper, for walls in a uniform embedment through analysis and gives some understanding of the are indeed conservative.
granular stratum. then round it up to make the design difficulties arising from direct The issues discussed in this paper
There are other recommendations more conservative. The detailer will application of standard requirements will not disappear when using EC7
on the selection partial load factors then increase the wall strength (and for the use of factored soil parameter unless engineers understand the
that are currently available, however, stiffness too) for the same reason. designs. importance of using a combination
some of them are not sufficiently The result may be that the wall The author suggests that when of several design approaches in their
explicit and some others imply Yf- embedment will be two to three times using a factored soil strength design practice.
values lower than those applied to what was actually required for the approach (whether BS8002, or EC7,
the self weight of the structures. real soil conditions and the bending or other) the designer may wish to References
The author believes that a moment may be up to twice greater ensure that: 1. BS5950-1:2000. Structural use of
comparison based on the above than that calculated by design. ● They understand that the analysis steelwork in building – Part 1: Code of
three options would give a sufficient Introduction of overexcavation based on factored soil parameters practice for design – rolled and welded
review here. into the design may increase the may result in a set of structural sections.
Just to make the designer’s task underdesign in structural strength design loads little related to the real 2. BS8002:1994. Code of practice for
more challenging, it is interesting by a further 10% to 15%. working conditions. earth retaining structures.
to see how the understanding of a It should be remembered ● They understand the assumptions 3. BS8110-1:1997. Structural use of
conservative design translates into that the above refers just to the implied by the method of analysis or concrete – Part 1: Code of practice for
a paradox: a geotechnical engineer designs capable of modelling type of software employed, together design and construction.
interpreting the ground conditions variation in soil structure interface with the consequences caused to the 4. BS EN 1997-1:2004, Eurocode 7:
may believe that by ascribing ø = stresses, depending on the relative structural design of the wall. Geotechnical design – Part 1: General
40º to a soil whose real ø-value is deflections. The use of simpler types ● The comparative increase rules.
45º, makes the design conservative. of analysis may introduce much in structural strength demand 5. J.B. Burland, D.M. Potts. The overall
That would translate into a factored different results. against the analysis based on “best stability of free and propped embedded
ø-value of about 35º using a The progress in computer estimate” reflects the factual design cantilever retaining walls. Ground
mobilisation factor of 1.2. development has created the uncertainties faced. Engineering, 1981, July, pp. 28-38.

GROUND ENGINEERING JULY 2007 35

[Link] 35 22/6/07 13:00:53