Proc. Instn Civ. Enyrs, Part 1, 1991, 90, Dec.

, 1255-1277
ENGINEERING
9585 DISCUSSION
GROUND GROUP

Use of sheet pile retaining walls for deep

excavations in stiff clay
D. M. Potts and R.A. Day

J. D. Findlay, Stent Foundations

With regard to the parametric studies, when the Authors looked at the difference
between K , = 0.5 and 2, did they equally increase the parameters of the other
parts of the soil? Was thestiffness increased in the sameway so that it matched a
realistic soil?
51. It looks as if constant soil stiffness was used in each of the detailed studies.
Was any account takenof the fact that with increasingly flexible walls there might
havebeenrather less stiffness in the soil, whichin turn mightlead tomore
movement? In other words,is there any possibilitythat the movement of the very
flexible walls wasunderestimated?
52. It is stated in 4 28 that a reduction was taken in the pile modulus from the
high modulus 4N,whichwas presumablytaken astheequivalent tothedia-
phragm wall, resulting in a maximum of 60% increase in movement. In many cases
it would not matter if 10 mm were a starting point, but I think many would find
difficulty if 30 mm were the starting point and there was a 60% increase in the
movement.
53. Therefore, each particular problem dictates whether or not a very flexible
wall is chosen. Walls are often built within 2 m of existing structures and this
makes the client extremely sensitive about the movement he thinks might take
place. As a result, one is obliged to guarantee or to underwrite in some way that
very large movements(e.g. of not more than5 mm) will not occur; becauseof these
contractual obligations, thereis a great dealof incentive notto be wrong.
54. In my experience, if one suggests to main contractors, in particular, that
they could choose more flexible walls with the addition of props, they will press
extremely hard to take out proppinglevels rather than put morein because of the
practicalities of excavating quickly; theydo not want prop levels there. One thing
to be taken into account, especially in a multi-prop situation,is how much extra
the whole projectwill cost if more proppinglevels are putin.
55. On secantwalling, I agreed with theAuthorsthatthere is nopointin
putting actual money figures against it because the requirement changes with time
and also particular caseswill alter it, but if a secant wall with 100 kg/m3 of steel is
put in, then a secantwall might be 7 4 0 % more expensivethan sheet piles. In many
conditions if the cost difference were of this order it might be cheaper in the
particular circumstances.
56. No account was taken of using contiguous pile walls. If contiguous walls,
or evenhard-softsecant walls such as Stentwall are used,thecostbasis will
decrease dramatically, and a contiguous bored pile wall is less stiff than a dia-
phragm at 1.2 m. So one has the advantage of more flexibility without much
Paper published: Proc. Instn Civ. Enyrs, Part 1, 1990,88, Dec., 899-927.
1255
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 pISCUSSION
movement. A contiguous bored pile wall of 900 mm diameter with perhaps 10
T40s is quite heavily reinforced. Certainly for a commercial building it is heavily
reinforced although it is not very heavily reinforced so far as the Dtp is concerned.
But that would give a modulus between a lBXN and a 4N (i.e. still quite a stiff
wall) and extremely little movement comparedwith the Authors’ figures. It would
certainly be advantageous to prop only at one level.
57. So Iwouldsuggestagain that theparticularcircumstancesarecritical.
There are many cases if there is no water where the contiguous wall needs to be
matched against the sheet pile wall. If there is water, the sheet pile needs to take
into account whether or not dewatering is necessary; the hard-soft secant might
do that at little costso that the cost canbe kept as low as for a sheet
pile wall.
58. The Paper shows that flexibility is important and that many walls are
probably too stiff for their own good. However,all solutions need to be examined
and themost efficacious one selected for the client.

D. Thompson, Dew Group Ltd

Noisewhensheet piles are beingdrivencan be controlled wellby using the
Hushrig, which is operated by the Piggot Bachy Group. It is a sound-proof box
into which the sheet piles are pitched and inside which a drop hammer is lifted by
the crane hoist rope to drive pilesthe down to groundlevel. The box is then moved
along guide rails approximately one metre and the next pair of piles or single pile
is pitched and driven. The method is very effective from the noise point of view, but
it does not cure the problem of vibration.
60. The machine that caneliminate both noise and vibration is theTaylor
Woodrow Pilemaster operated by Serf Piling Ltd. This machine pushespiles into
the groundby means of hydraulic rams attached to the top of the sheetpiles which
each have a 200 ton capacity to overcome soil resistance.There is no impact and
therefore no noise or vibration is created. The problem with this machine is that it
will not operate in gravels and coarse sands-it operates best in clays or silts. As
such it is not a panacea for the problemof driving although it can drive quite deep
piles through stiff clay, particularly London clay. A solution to driving piles in
granular soils such as sandsand gravels (e.g. in Thames Ballast)is to predrive sheet
piles usingavibrator and then to completedrivingin clay by means of the
Pilemaster. However, the vibrator creates a certain amount of vibration that can
give rise to surface crackingof pavements, roads andeven buildings.
61. The latest development for driving piles into clay is a machine which has
been used extensively in Japan and hasrecently been introduced into the UK. The
machine (called a Silent Piler) workseffectively the same as the Pilemaster in that
it presses piles into the ground with a hydraulic ram, but it has a smaller ram
capacity (130 tons) than the Pilemaster. It is nevertheless a more sophisticated
machine and is very wellengineered.

I. F. Symons, Transport and Road Research Laboratory

The Paper draws attention to the need to take account of wall flexibility in the
assessment of the structural loading on embedded retaining structures. I been have
involved in a field study of the behaviour of an anchored sheet pile wall used at
temporary works for the Al(M) improvement project, carried out in 1985-86 at
Hatfield.31 A major part of this scheme consistedof an 1150 m long cut-and-cover
tunnel with the approaches to the tunnel in retaining cutting.
1256
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585
63. Figure 18 shows the form of construction in which over 2 km of sheet pile
walling, with prestressed ground anchors at 3 m centres, was used to retain the
ground duringconstruction of the permanent structures.On completion, the space
between the temporary andpermanent retaining walls was backfilled, the ground
anchors were cut and the sheet piles extracted. The groundconditions consistedof
glacial sands andgravel overlying a layer of stiff Boulder clay. A 20 m long section
of the sheet pile wall on the northern approach to the tunnel was extensively
instrumented to study the behaviour of the wall and anchorage system during its
construction and over a ten-month period in service. As part of this instrumen-
tation, inclinometer tubes and strain gauges were attached to the sheet piles, so
that the deformations and bending moments couldbe determined. Load cells were
also used to measure the head loads in the ground anchors.
64. The 13 m longLarssen 4/20 sheet pileswere driven percussively, in
clutched pairs to full depth using a Hushrig. Excavation was then carried out in
stages to a final depth of 9.3 m for construction of the base slabfor the reinforced
concrete wall. The measurements showed that, as construction progressed, the
maximum pile deflexions and bending moments occurred at about the mid point
between excavation and anchorlevels.
65. A maximum bending moment of 75 kNm/m was obtained from the mea-
surements of pile strain on completion of excavation, and this increased to 110
kNm/m prior to backfilling. These values can be compared with a design value of
422 kNm/m using the approach given in the Piling handbook3’ Rowe’s method of
design,33 which takes into account the stiffness of the wall and propping system,
gave a maximum bending moment of about 200 kNm/m which is closer to, but
still inexcess of, the values obtained from the strain measurements.
66. One of the schemes described in the Paper is the Bell Common tunnel on
theM25in Essex. Monitoring of thebehaviour of thistunnelduringits

Larssen sheet piles

. .
l Made ground
I
I

I
l
1
I
I I Upper glacial
gravel and sand

I
l
I
-
-GWL
y
I
r---L
Ground anchors

O 2
U, Boulder clay

Scale of m

Fig. 18. Cross-section of the sheet pile wall

1257
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
c o n s t r ~ c t i o nshowed
~~ that significant movement and reduction in stress in the
ground were caused by the installation of the bored secant pile walls. This prob-
ably contributed to the smaller bending moments measured on completion or
tunnel construction compared with the class A predictions from finite element
analyse^,'^ as shown in Fig. 19. In these analyses, like those in the Paper, thewall
was 'wished in place'and thereforeanymovements or stresschanges In the
ground caused by the method used to form the wali were neglected. Better agree-
ment was obtained between the measured maximum bending moment and the
results of further finite element analysesz6 (also shown in Fig. 19) in which an
attempt was made to model the installation of the wall.
67. If the walls of the Bell Common tunnelhad been formedfrom driven.
rather than boredpiles, one might have anticipateda local increase in lateral stress
in the groundwith a consequent rise in theinitial earth pressure coefficient.
68. How might installation effects influence the bending moments calculated
by the Authors'?

C . Raison, Keller Foundations

Figure 20 illustrates the results of computer analyses carried out by Ove Arup &z
Partners for the diaphragmwall originally proposed for the new British Library.3'
This compares two computer runswith wall stiffness varied by a factor of 3. There
is almost no change in soil pressures or wall displacements. The significant change
in bending momentis a function primarilyof the wall stiffness.
70. This effect applies to propped retaining walls, particularly multi-propped
walls. Retaining wali movements and soil pressures are far more sensitive to the

Depth below ground level m

c \
Measurements
c' \
Revlsed
prediction

\
l//

/
20 ( I
1 I l l I 1
60G 400 200 0 -200
Bending moment In wall- kNmim

Fig. 19. Measured and predicted bending moment projilesfor stage

1258
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585

1259
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
assumptions made about initial in-situ stresses and prop stiffnesses than to the
wall stiffness.
71. Where retaining walls have to act as unpropped cantilevers or to remain
unsupported in basement core or ramp areas, there are often good reasons for
using stiffer walls. For the moregeneral stiuation the Paper makes a good case for
the use of more flexible walls-an approach which I hope can be adopted more
widely.
72. Steel sheet piling is not the only alternative for more flexible walls. Bored
cast in place piling can also providefull therange of stiffnesses.
73. Rowe’s flexibility number H4/EI has traditionally been used for compari-
sons of wall stiffness. Fig. 21 is a copy of Fig. l(a) annotated to show the more
common bored pile stiffnesses for which a value of H = 20 m has been taken.
Because of the wall length term this comparison is not idealand theuse of In(E1) is
preferable, as in Table1.
74. Figure 22 showsadirectcomparisonwithTable 1. Stiffnessesfor steel
sheet piling are shown compared with those for bored piling. E1 for the concrete
piling has been computed using an uncracked moment of inertia and a Young’s
modulus of 25 OOO MN/mZ, andis given in terms of stiffness per metre runof wall.
As can be seen, bored piling canoffer stiffnesses covering a similar rangeto those
given by steel sections. Use of mini piles can result in a wall stiffness less than the
Frodingham I N section, although mini piles are usually used with a permanent
steel liner to give increased stiffness.
75. The results shown in Fig. 22 do not tell the whole story. Thewall stiffness is
based on a momentof inertia calculated for an uncracked section. Unfortunately,

240[ 120 mm
Bored piles

900 mm
I 750 mm 450 mm
nm

I
- 0.5 1.5
Log (Q),p = H4/E/:m3/kN

Fig. 21. Common bored piled stiffnesses

1260
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585
for reinforced concrete thereis a distinct change instiffness of the section with the
onset of cracking which causes a reduction in the moment of inertia.
76. Methods existfor determining the moment/curvature relationships for a
particular reinformed section. The example in Fig. 23, carried out by Ove Arup
and Partners, is based on the CEB-FIP method applied to a 1 m thick diaphragm

Sheet piling Bored piling

220 mm dla.
Frodingham 1 N

Frodingham 4N 450 mm dla.

600 mm dia.

Hlgh modulus 1BXN 750 mm dia.

900 mm dia.

1200 mm dla.
High modulus 4N
E, = 25 X 106 kN/m2
Fig. 22. Retaining wall stzjiness per metreof wall

2000 -

E
2 ,
3
Earth face
S+
c
m
g 1000 -
I
A,' = Y32 at 200

100 cover to
main bars

€,,,,= 20 000 MN/m2

= 200 000 MN/m2
I I I
0.001 0.002 0,003
Curvature: m-'
Fig. 23. Moment-curvature relationshipsfor diaphragm wall, CEB-FIP method35
1261
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
Three cases are shown with varying reinforcement and axial forces. The
bi-linear relationship should be noted.
77. This form of analysis is important, particularly for serviceability checks
where it is necessary to predict the likely maximum crack width. It is also impor-
tant in order to model correctly the non-linear behaviour of the structure as well
as that of the soil in any soil-structure interactionanalysis.
78. The example in Fig. 24 shows results for a 600 mm bored pile with a 6Y32
reinforcement cage.36 The behaviour is again non-linear and shows two distinct
phases. Fig. 24 alsoshows the effective pile stiffness EIjmoment relationship for
the same boredpile, and illustrates the difficulty in choosing a representative value
for use in soil-structure analyses. Similar relationships can also be computed using
long-term concrete properties.
79. I hope that the approach proposed inthePapercan be adoptedmore
widely, not just when using steel sheet piling but also with reinforced concrete
bored piles. It is to be hoped that comments about the difficulties of choosing

L
OO 80 160
Moment: kNm
240 320

(b)
Fig. 24. Short-term ultimate bending moment, load case 1 : ( a ) moment-curvature
plot: (h)moment-stiflness plot
1262
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585
representative stiffnesses for the wallwill actasatimelyreminder.Although
sophisticated computer analysis can be carried out to investigate soil-structure
interaction, there arestill areas where further investigation
is required.

T. Paul, Ove Arup and Partners

With regard to the new British Library, sheet piles were used as retaining struc-
tures during the construction of both single and doublelevel basements.
81. Two separate packagesof sheet pilingwere carried out in roughly the same
area of the site at different times. In both cases,the sheet piles were driven from a
ground level of + 16.0 m OD. Theground profile consisted of London clay
overlying Woolwich and Reading beds with the interface at approximately - 1.0
m OD. Undrained strength values for the London clay were 50-250 kPa and SPT
N values were 20-60, both parameters increasingwith depth.
82. The first package of sheetpiling was carried out usingthePilemaster
system.Approximately 250 Frodingham 5N sections were installed to forma
retaining wall of 110 m plan length. The sheet pile sections were driven 11-15 m
below ground level, retainedheightsbeingup to 7 m. Aspecialmeasure was
adopted to improve the driveability of the sheet piles. This consisted of a metal
block welded to each sheet pile approximately 3 m from thetoe of the pile. As the
pile was driven the block forced the soil away from the pile face, reducing the
adhesion forces and thereby ensuring easier penetration. Less than 3% of the total
number of piles failedto reach the design toe levels.
83. In 1989, a second sheet pile retaining wall was installed in a similar region
of the site. Frodingham 4Nsections were used for this contract due to the unavail-
ability of 5N sections. Analysisof the driving records showsthat for those sections
with design toe levels 13-14 m below ground level, all were driven to the design
levels. However, approximately 15 sheet piles with deeper design toelevels (14-16
m OD) reached refusal morethan 2 m short of the design levels. To accommodate
this problem with minimum delay to the overall programme the basement con-
struction sequence was modified to reduce the retained heightin the critical areas.
84. These case histories illustrate the importance of assessing sheet pile drive-
ability and choice of sheet pile section. InLondon,the secondcasehistory
described shows that driving problems may occur in London clay, particularly at
depth where the &/fine sand content increases towards the interface with the
Woolwich and Reading beds.

Dr H. D. St John, Geotechnical Consulting Group

I want to describe a site where steel sheet piles have beenused as a remedial
measure because the bored piles gave problems. Thesite originally contained two
adjacentbuildings, both founded on shallowfootings at basement level. One
building was demolished and an additional basement level constructed beneath
the new structure. In orderto be able to excavate the basement itwas necessary to
install a peripheral cut-offwall which extended through the upper alluvial deposits
into the Londonclay.
86. The London clay was about 8 m below the original basement level. The
upper alluvium comprised loose to medium dense sands, soft clays and peats. A
thin gravel layerwas present just abovethe London clay. A significant feature was
the proximity of the (tidal) water-table to the original basementlevel, which made
it difficult to install a bored pile wall without losing ground into the bore and
1263
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
causingtheadjacentfoundations to settle. Thesituation was madeworse by
having to pile around the corner of the building facing into the site being devel-
oped, the rear face of the new wall being within a metreof the external face of the
existing wall.
87. Settlement of the adjacent foundations was monitored, and movements
during piling were significant. (The maximum settlement wasabout 40 mm.) Thus
piling was stopped whenitwasless than 50% completein critical areas. An
alternativescheme was sought which wouldallowthe wall to be completed
without causing significant additional damage. The solution adopted was to use
the Silent Pilerto jackin fairly stiffsheet piles through the soft and loose materials
into the more competent strata, and thencomplete to the boredpile wall to its full
depth on the excavation side of this temporarywall.
88. The piles went in like butter to a depth of about 10 m, and it took only a
short time to install a length of about 30 m over thecritical area. Subsequently, the
bored pile was put in with little additional movement. The installation of the sheet
piles had caused no measurablemovement.Commonsenseindicates that this
remedial measure could not entirely prevent loss of ground, but the installationof
the stiff membrane allows local loss of ground to be spread both laterally and
vertically.
89. Stiff walls are stiff because they are large. They are large because they are
designed to resist predicted bending moments rather than to reduce movements.
Usually the design of a wall involves calculating a maximum bending moment,
then taking an appropriatefactor of safety (from a structural code) and designing
the reinforcementfor the size of pile or wall chosen.
90. In the case of a multi-prop wall, themaximumbendingmomentsmay
occur only when the wall is already propped at several levels; in fact it usually
occurs below the level of the bottom prop. This is particularly the case in t o p
down construction. The wall itself is rarely watertight; all walls leak because they
have vertical joints. Therefore, if cracking due to excessive bending moment were
to occur, the function of the wall would not be significantly impaired. Failure of
the wallin bending would make the wall more flexible in the latter stages of
excavation, but by this time the groundbehind the wall would be supported by the
upper props, and movement beneath the excavation would be controlled mainly
by the properties of the ground rather thanthose of the wall unless the ground is
weak. It is not really the wall that is doing the work. Therefore,failure in bending
is unlikely to be a problem.
91. A slightly different problem is the magnitude of the bending moment that
can be induced at the lowestprop after completion of the structurewhen clay soils
drain. At this stage the basementis a rigid box in the ground, probably supported
on piles. The magnitude of the moment in the wall below this box is irrelevant
unless the wall is having to work to prevent loss of ground behind the wall, which
is rarely the case. Even if the wall were required to take a vertical load, this might
be possible on acracked section.
92. I believe that there may be significant savings in reinforcement costs by
rationalizing the approach to the design of concrete walls. It does mean, however,
that designs would haveto incorporate the development yield within thewall.

Dr W. H. Ward, Consulting Engineer

With regard to the question of whether or not a steel sheet pile wall driven into a
stiff clay causes displacement, here
is an example.
1264
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585

94. About 33 yearsagotworows of steel sheet piles, formingaperimeter

cofferdam for the construction of the deep basement retaining wall, were driven
around the futureShell Centreat SouthBank alongside theRiver Thames.37
95. The piling was driven through made ground, alluvium and gravel, and into
the London clay to form a cut-off. However, where it crossed the Bakerloo Line
tunnels, lyingone above the other, the depth of cut-off had to be reduced to avoid
them. It was understood that the straight rows of piles were to cross the tunnel
approximately normal toits axis with the lower endsof the piles stepped up in the
form of an archedprofile about 3 m clearand above thetunnel. Special care was to
be taken in driving thepiles in this region. Theywere to be driven progressivelyin
pairs at night when thetrains were notoperatingand whenthestrains and
displacements in the tunnel lining could be monitored at frequent interval^.^'
96. At the first crossing operation, the crown of the uppermost tunnel (4 m
OD) lay 3.8 m below the London clay surface and about 13.6 m below ground
level. Monitoring of the tunnel lining ensued as the piling approached daily from
one side. By Christmas 1957, it had been driven down almost to axis level and
stopped about 3.3 m away from the lining with only a minor elongation of its
vertical axis being recorded. Drivingof the crossingat night was anticipated in the
new year. It was therefore with much surprise and concern that during remote
observation of the vibrating-wire strain gauges on the lining when the trains were
running early in 1958, a strong distortion was found to be taking place at the
crossing.
97. This distortion was caused by a plan length of 3 m of piling (Larsen 4B
section) driven parallel, about 3.3 m to one side, and almost to axis level to form
an offset in the piling to one side of the tunnel. It caused a local inward displace-
ment of the nearest side of the lining which, at the ring monitored in detail,
amounted to a reduction in horizontal diameterof about 3.7 mm and asimilar rise
in the crown. The major part of this distortion took place as the toeof the piling
descended the2 m belowthe tunnel crown.
98. The monitored ring would not have recorded the maximum effect of the
piling being driven parallelon the tunnel because it was located opposite one end
of the offset. The fulleffectislikely to be about double that observed, i.e. a
displacement of, say, 7-8 mm, which is about half the thickness of the Larsen 4B
section.
99. Therefore, sheetpiles driven into London clay do displace the ground and
tunnels away fromtheir sides.
100. Theamount of noise andvibration withinthetunnels when the pile
crossings were driven at night was alarming, but no incidents occurred. When the
piling was kept clear of the top periphery of the tunnel by 3 m and it did not
descend below crown levelin the vicinity, a small permanent rise (less than a
millimetre) of the tunnel crown arose progressively as the piling crossed over. It
was associated with a corresponding contractionin the horizontal diameterof the
tunnel. This distortionis perhaps contrary to what one might expect, but probably
represents some dynamic dilation of the clay beneath the piles.
101. These observations provide a useful guide to the driving of further sheet
pile walls near tunnels instiff clay. However, for the case of a contiguous boredpile
wall or a diaphragm wall built to become a basement retaining wall, there have
beennumerousobservations to show thattheadjoiningLondon clay moves
towards thewall and settles as thewalls or single pile shafts are excavated.
102. Indeed a bodily settlement and a squatting of several millimetres of the
1265
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
Bakerloo Line tunnels at the Shell Centre was recorded when foundation piles
were excavated either side of them. The lateral extension of the escalator shaft at
Hyde Park Corner Station towards both diaphragm and contiguous bored pile
walls was recorded when they were used, probably for the first time in London, to
build the cut-and-cover vehicular underpasses either side of the shaft. M ~ o r e ~ ~
records a settlementof up to 24 mm of the London clay surfacebetween groups of
large augeredpiles at the Houseof Commons car park.

A. S. O'Brien, formerly W . S . Atkins Consultants Ltd. now Mott MacDonald U K

Ltd
The finite element analyses used by the Authors assume that thesoil behaved as a
linear elastic-perfectly plastic material, i.e. before yield the elastic stiffness is con-
stant (at a particular depth) irrespective of the changein stress which occurs at the
element under consideration. It is now well understood that soil stiffness is highly
non-linear, even at small strains prior to yield,40 and is significantly affected by
changesinrecent stress h i ~ t o r y . ~The
' undrainedsecant Young's modulus of
overconsolidated clays, when expressed as a ratio of undrained shear strength
e,/c,, can vary typically fromabout 1500 at shear strainsof 0.01% to 150 at shear
strains of about 1.0%. The elastic stiffness usedby the Authors for the finite
element analyses can be regarded as an intermediate value between small and
large strain stiffness, with eJc, ratios varying typically from about 550 to 850 for
the sites considered. Hence, it is likely that in regions of small strain the type of
constitutive model used by the Authors will tend to overpredict ground deforma-
tion, whereas in regions of large strain the modelwill tend to underpredict ground
deformation.
Dlstance from edge of excavatlon: m

Normalized dlstance from edge of excavation XIH Reralnlng wall geometry

Linear elastic-

_____.-----
z
(b)
_ _ _ removal
Prior
to of -Prlor to removal of
/\ non-linear
-x-x-
temporary
elastlc-
temporary
linear
prop
After removal of
temporary
prop
1\
1
perfectly
plastic analysis - After
prop
removal
temporary
prop
of \, elastic-
perfectly
analysls
plastlc
Fig. 25. Settlement behind retaining walls, efect of type of constitutive model on
settlement predictions: ( a )absolute settlement behind retaining walls; ( b )normalized
settlement behind retainingwalls,afterremoval of temporaryprop; 6v,,, =
maximum settlement
1266
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 958.5

104. Recently, ICFEP has been used to analyse a retained cutting which will
form part of a proposed major urban motorway scheme. Two different types of
constitutive model have been used to analyse the same type of retaining wall; thz
predicted settlement troughs behind the wall are compared in Fig. 25. The wall is
permanently propped at road slablevel, and is supported temporarily by a prop at
the top of the wall. Fig. 25(a) shows the absolute displacements; in Fig. 25(b) the
settlementshave been normalized by themaximumpredictedsettlement. As
shown in Fig. 25(b), the linear elastic-perfectly plastic analysis underpredicts dii-
ferentlal settlements behind thewall. whereas the differential settlements predicted
by the non-linear elastic-plastic analysis are similar in form to those which are
normally observed adjacent to deep excavations in heavily overconsolidated clay.
Simpson et d 4 ’ observed similar effects: satisfactory ground deformation predic-
tionsbehindretaining walls (the examples usedwereNew PalaceYardand
Neasden) were made only when small strain high stiffness behaviour (so-called
threshold effects) was modelled. In the back-analysis of Bell Common,26 theuse oT
a non-linear elastic-perfectly plastic model gave better estimatesof the maximum
measured ground surface movement than those given by a linear elastic-perfectly
plastic model.
105. The Authors assume that, during construction. the clayis undrained. This
is probably reasonable for construction periods of short duration (say, two to
three weeks). but it is unlikely to be correct for construction periods of several
months. In the London clay, large negative excess pore water pressures are like11
to be generated following excavation. Water seepage into the excavation, from
whatever source, will lead to significant increases in ground deformation close to
the excavation; Fig. 26 illustrates this effect. It shows ICFEP predictions of wall
deformation for varying degrees of excess pore water pressure dissipation. The
important point to note 1s not the absolute values of displacement, but the changes
in displacement as negative excess pore water pressures are partially dissipated.
Hence, to minimize ground deformation, the designer must be certain that the
retaining wall is as watertight as possible, particularly for the common situation
when water-bearinggravelsoverlieoverconsolidated clay. Drivingsheet piles
through dense gravel is often difficult and declutching of sheet piles can occur. If
the sheet piles declutch or are damaged during penetrationthen the wall will lose
its watertightness, and greater ground deformations, particularly close to the exca-
vation. are likely to occur. In practice. diaphragm and secant pile walls are more
likely to be watertight than sheet pile walls. It is also worth noting that sheet
a pile
cofferdam failure in London clay occurred partly as a result of water seepage
through the sheetpile interlock^.^^
106. Somepractisingengineersmayalsoquestion the wished-in place’
assumption for the insertion of retaining walls, because driven sheetpiles are likely
to lead to increases of in-situ horizontal stress, whereas the installation of rein-
forced concrete walls islikely to lead to decreases of in-situ horizontal stress.
However, unless the retaining wallisstiff and propped, the bending moments
induced in the wall are unlikely to be particularly sensitive to the initial in-situ
horizontal stresses. Could the Authors comment on this?
107. Could the Authors explain how noise and vibration during sheet piling
driving can be kept so low during diaphragm walling operations? Ground defor-
mations during sheetpile installation can alsobe significant.
108. I would suggest that the Authors’ predictions of the change in the struc-
tural forces which will be imposed on retaining walls of varying rigidity are likely
1267
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
dhiH: %
0.5 1 1.5 2
O0 \\
\, Tempbrary
I I

0.5 -
c'-
\ X

\ X
prop

Undramed

II After partial disslpation of excess

pore water pressures, see inset

1 2
l
1
I
l
\
X
Floor of excavatlon

I
I I

-
/ 7 dh = horizontal displacement

5 1.5

/ /
Temporary prop

T------T
!
(10%) 0.37ZIH

3 L

Figures In brackets give degree of

-
excess pore pressure dlsslpatlon to be
allowed durmg construction

Fig. 26. Normalizeddisplacement of retainingwallbeforeremoval of temporary

prop; injuenceof partial drainage on wall displacement during construction

to be reasonable. However, the increase in differential ground movement, particu-

larly close to the excavation, from that associated with a stiff reinforced concrete
wall to that associated with a flexible sheet pile wall is likely to be underpredicted
by the Authors, the degree of underprediction being influencedby the number of
props being usedto support the excavation.

D. P. Nicholson and K. Yah, Ove Arup and Partners

TheAuthorsprovideaninterestingnumericalstudy of theinfluence of wall
stiffness on the behaviour of propped retaining walls. It is of interest to compare
the calculated wall deflexions with
the field measurements at the end of excavation
for the New Palace Yard sitez8(see Fig. 27). The maximum calculatedwall deflex-
ions are slightly less than those measured, whereas at the toe of the wall the
1268
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585
Lateral movement towards excavation: mm
0 10 20 30

Fig.27. Comparison of movement at theend of excavation in New Palace Yard

calculated deflexion is more than those measured in situ. This results in tighter
measured curvatures and hence larger bending moments for the in-situwall.
110. On this basis it can be seen that underestimating the small strain stiffness
of the Thanet sand and chalk below the toeof the wall can result in underestimates
of the bending moments.
111. The Authors consider the bending stiffness of a 1.0 m thick New Palace
Yard diaphragm wall to be equivalent to the high modulus 4N sheet pile (4 24).
The behaviour of a reinforced concretewall in bendingis complex. At low bending
moments, the tensile strainsand hence tensile stresses are relatively small, and the
concrete remains uncracked. As the bending moment increases, tension cracks
develop in the concreteand the load is gradually transferred to the reinforcement.
The stiffness of the cracked concrete section therefore reduces, unlike the steel
section which remains constant in the elastic working range.
112. Various design procedures have been proposed for assessing the stiffness
of cracked concrete sections (Fig. 28). In thisdiscussionthebehaviour of the
1269
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
concrete in compression is based on BS 81 10, Part 2.44 A short-term concrete
modulus of 28 X 106 kN/m' has been adopted. Based on BS 8110, Part 2, Table
7.2, this corresponds to a cube strengthof 35 N/mmz.
113. For comparison with the New Palace Yard case history, a 1.0 m thick
diaphragm wall has been considered. Details of the reinforcement were unknown
and therefore T40 bars at 167 mm centres on both faces have been assumed (1.5%
steel). Based on BS 8 110, Part 2, the calculated ultimate bending moment is 2650
kN m per metre run. Therefore, the working bending moment is 1900 kN m per
metre run, assuming a partial factor of 1.4.
114. The OASYS program ADSEC has been used to study the variation of wall
stiffness with increasing bending momentfor the three design procedures shown in
Fig. 28. The results are shown in Fig. 29. These results show an uncrackedstiffness
of 3.1 X 106 kN m' per metre run, reducing to 1.0 X 106 kN m' per metre run at
working condition (1900kN m per metre). It can be seen that the stiffness reduces
rapidly using the BS 8110, Part 1, no tension design procedure.44 Fig. 29 also
shows the uncracked wall stiffness of 2.33 X 106 kN m per metre run adopted by
the Authors(4 24).
115. The bending moments for the New Palace Yard diaphragm walls based
on a uniform uncracked wall stiffness are shown in Fig. 9. They vary from 0 to
loo0 kN m per metre. From Fig. 29, it can be seen that the above design pro-
cedureswouldpredictsignificantchanges inwallstiffness over this range of
bending moments.
116. The OASYS program ADSEC also calculates the crack widths on the tension
face for the different design procedures(see Fig. 29). Limited guidance is available
on acceptable crack widths for diaphragm wallsin normal ground conditions.

f, = compresswe stress f, = tensile stress

f , = lhmltlng stress fcl = characterlstlc tenslle strength
Eo = concrete modulus rp = creep factor
fc

Neutral Axis

Steel
/I
t--i
1 N/rnm2 f, varies as
X- shown below

(No tensile strength)

Strain Straln at tenslon steel Stratn at extreme fibre

(a) (b) (C)

F i g . 28. Short-term tensile stresslstrain for concrete in bending: ( a )BS 8110, Part I ,
clause 3.4.4.1;44 ( b )BS 8110, Part 2,Fig. 3.1 ;44 (c)S ~ o t t ~ ~
1270
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585
l m

...m.

Uncracked wall stiffness

adopted by Authors

\
’\ ScottJ6f,, = 2.37 Nimm’

BS 8110. Part 144

no tension € E

g b
-3
l I I I I

Moment: kNm per metre run

Cornpresslon stress block modelled usmg BS 81 10, PartP‘‘

E, = 28 X 106 kN/rn2. f,, = 35 Nimm’

Fig. 29. Variation of wall stiffness with bendingmoment

Recent research thinking suggests that even large cracks are unlikely to lead to
corrosion of the reinforcement bars. However, BS 8110, Part 2, clause 3.2.4.2,
recommendsacrackwidth of 0.3 mmfor corrosion effects onconcretein
aggressive environments. The bending moment at which this crack width occurs
varies from 1060 kN m per metre using the BS 8110, Part l , no tension procedure
to 1500 kN m per metre using Scott’s procedure.36 These are significantlyless than
the working bending moment of 1900 kN m per metre discussed previously, and
can therefore influence the acceptability of the design in serviceability conditions.
117. The Authors’ analysis showsthat the lateralwall movements in stiff clays
are insensitive to wall stiffness. Similar results have been reported by Clough et
for undrained conditions inclay (Fig. 30). These show that provided the
factor of safety againstbasalheave(definedinreference 46) exceeds 2.0, the
maximum deflexions are insensitive to wallstiffness. The Authors’ results cor-
respond to a factor of safety against basal heave of about 3 according to Clough
et al.
1271
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
118. The numerical results reported by the Authors for New Palace Yard and
Bell Common arealso basedon undrained shear strengths in the London clay and
have been plotted in Fig.30. The factors of safety against basal heave for the New
Palace Yard and Bell Common sites have been calculated as 4.7 and 6.5 respec-
tively. This calculation assumed rigid
a base at the bottomof the Londonclay.
119. TheAuthorshavecalculatedbendingmoment profiles on which they
base sheetpile cost comparisons. However, these bending moment profiles are less
thanthoseenvisagedwhenthe wallswere originallydesignedbecause of the
different design parametersand method^.'^ By adopting theAuthors’ calculations
it is possible to put forward alternative reinforced concrete retainingwall designs
which wouldalsoshowsignificantsavingsovertheoriginaldesignsfor New
Palace Yard and Bell Common. This would have given a uniform basis for cost
comparison.
120. For example, the wall stiffness E1 of 0.8 m and 0.6 m thick diaphragm
wall panels has been assessed(Fig. 31) using the same quantity of reinforcement as
adopted previously. Thestiffnesses of the sheet pile sections are also shown. Fig. 31
shows that the high modulus lBXN section is equivalent to a 0.6 m diaphragm
wall. The predicted working bending moments for the high modulus lBXNsection
are shown in Fig.9. These d o not exceed 600 kN m per metre runand aretherefore
within the capacity of a 0.6 m thick diaphragm wall. In practice, it would be
possible to reduce and curtail the reinforcement.
121. The capacity of wall calculated using the program ADSEC, allowing for a
maximum crack widthof 0.3 mm, would have been 750 kN m per metre run. For

15 45 75 105 150 750 450 1500 3000

Wall stiffness N X 103

New Palace Yard Bell Common Tunnel

A High
modulus 4N 0 Highmodulus
4N
v Highmodulus l B X N + Highmodulus l B X N
X Frodlngham 4N * Frodingham 4N
0 Frodingham
1N A Frodingham1N
Fig. 30. Design curves to obtain maximum lateral wall movement
for
1272
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585

0.3

400 800 1600 1200 2000

Moment: kNm

diaphragm
wall: m

Twelve T40 2030 1450

2.5 1375 980

Fig. 31. Variation of wall stiffness with bending moment for three diaphragm wall
sections

cost comparison purposes, a 0.6 m thick diaphragm wall could be considered at

New Palace Yard insteadof the 1.0 m thickwall.
122. From a consideration of section properties, a 0.6 m diaphragm wallis
similar to a contiguous bored pile wall comprising 0.75 m dia. piles at 0.9 m
centres. On the basis of the Authors’ calculations it would have been possible to
adopt this type of contiguous wall at Bell Common instead of the 1.2 m dia. secant
wall.

N. J. Brooks, Stent Foundations

The Authors point out the importance of the ratio of horizontal to vertical effec-
tive stress K O ,and yet the three examplesgiven show three different distributions
1273
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION
of K , in depth. Is this a function of the sites or of the approachof different authors
to K,,?

D r Potts and Mr Day

The paper presents the results of the analyses in terms of wall stiffness. The wali
stiffness values chosen for the analyses relate to those of available sheet pile wall
sections, but the results are not restricted to such walls. They can be used in a more
general sense to indicate the effects of wallstiffness and, as suggested by M r
Raison, M r Nicholson and M r Yah, can be applied to concrete walls. However, as
noted by these discussers, the flexural stiffness of concrete walls is not constant and
care must beexercised when assigning a single wall stiffness value. If the variatlon
of wall stiffness with increased loading is defined, it can be modelled in a finite
element analvsis.
125. Inpractice,there is anabundance of stiff concrete walls. ThePaper
suggests that a more flexible sheet pile wall could be viable from an engineering
point of view and also more economical. It may also be possible to provide a
flexible concrete wall that is cost-efTective. As academics, we are not in a position
to judge the relative economiesof such alternatives.
126. With regard to the effect of a variable soil stiffness on the results present-
ed in the Paper, it is now accepted that soil stiffness is both effective stress and
strain level d e ~ e n d e n t . ~
The
' effects that such a non-linearitv has on the behaviour
of a variety of boundary value problems have been investigated4' and its impor-
tance when predicting the behaviour of engineering structures has been estab-
l i ~ h e dSuch
. ~ ~effects are not included in the Paper,however, because the previous
analysis was performed several years ago when understanding of the effects of
straln level on stiffness was limited. The behaviour of soils under low stress levels,
similar to thatwhich might exlst near the excavatedsurface on thepasslve side of a
retaining wall, were not, and arestill not, well understood. Soils data. from which
the parameters necessary to define the variable stiffness behaviour could be deter-
mined, were not available for all the case studies examined.The mainobject of the
present work was to identify trends of behaviour associated with changing wall
stiffness. Inclusion of a variable soil stiffness would have complicated both the
analyses and their interpretatlon.
127. We have performed analyses, not reported in the Paper. in which a vari-
able soil stiffness was used. As would be expected, these analyses show that the
magnitudes of displacements andstructural forces are affected. However,the
results indicate that the trend of behaviour associated with changing the wall
stiffness is similar to that presented in the Paper. The results presented by M r
O'Brien (Fig. 25) were obtainedfromananalysis we carried out at Imperial
College. They show that a variable soil stiffness can have a significant effect on the
shape of the profiles of movement of the retained soil surface. The maximum
values of the settlement of the retained soil surface and thewall displacements and
structural forces are affected to a lesser degree. We would strongly support the use
of a variable soil stiffness for any design analysis. However, it is our opinion that
the inclusion of such behaviour in the analyses presented in the Paper would not
have greatly affected the basic trends of behaviour associated with changing wall
stiffness.
128. The influence of wall installation on subsequent wall behaviour is clearly
dependent on the methods and procedures used to construct the wall. For con-
1274
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585

Cretewalls this will depend on whetherthey are of the diaphragm, secant or

contiguous bored pile type, and in the later two cases by the degree of support
provided by temporary casings and drilling fluid.
129. For such concrete walls, movements in the surrounding soil are likely to
occurtowardsthe wall. Suchmovements are likely to be associatedwitha
reduction of horizontal soil stresses. If the wallisstiff and the soils are over-
consolidated, this will result in lower structural forces on bulk excavation (Fig. 1).
There is therefore a trade-off between larger soil movements and smaller structural
forces in the wall. If the wall has been designed to be stiff to prevent adjacent soil
movements then extreme care must be exercised to prevent significant movement
occurring duringwall installation. If this is achieved therewill be little reduction in
soil stresses, and if the soil is overconsolidated the structural loads are likely to be
high. It is interesting that M r Findlay quotes being obliged to guarantee that
movements will not exceed 5 mm and uses this to justify the use of a stiff wall.
However,both Dr Ward and Dr Sr John reportsituationswheremovements
substantially inexcess of this amount have been observed as a result of wall
installation alone.
130. When sheet piles are installed in overconsolidated soil they induce move-
ments away from the wail. The magnitude of these movements islikely to be
related to the volume of the sheet piles themselves. Intuitively, the stresses in the
ground and acting on thepiles once installed will increase. However, it is difficult
to calculate suchstress changes.49p51
131. If the wall had been flexible, the structural loads in the wall after bulk
excavation is completed and in the long term would not be greatly affected by
movements and stress changes occurring as result a of wall installation (Fig. 1).
132. With regard to M r Findlay's question concerning the results in Fig. l, we
confirm that only the wall stiffness and K , value were varied in the analyses. All
other soil properties remained the same. This was done so that the effects of
changing the wall stiffness and K , value could be isolated. We appreciate that for
real soils, other parameters, such asstiffness, will accompany changes in K , , but if
all these changes had been included it would have been difficult, if not impossible,
to isolate the effect of each variable. This simplification applies onlyto the results
presented in Fig. 1 ; it does not apply to the main textof the Paper. For each case
study, only one K,, profile was considered and it is consistent with the other soil
properties.
133. M r Brooks raises the question of the origin of these K , profiles. They were
interpreted from site investigation data by the original designer^.^^,^'.^^ K is a ,
difficult parameter to determine from eitherin-situ or laboratory data.The differ-
ences in the profiles at the three sites reflect the difficulty to some degree but are
also likely to be affected by the present and past geological history at each loca-
tion. Inparticular,thevaluesaresensitivetoanyredeposition which has
occurred.52
134. Mr O'Brien raises a question concerning noise and vibration during sheet
piling operations. This has essentiallybeen answered by Mr Thompson. Both the
Pilemaster and Silent Piler machines jack the sheet piles into the ground, thus
minimizing noiseand vibration.
135. M r Symons briefly describes the anchored sheet pile wall forming part of
the temporary worksfor the Al(M) improvement project at Hatfield. We also have
carried out design calculations and finite element analysesfor this wall. In particu-
lar, we have compared the maximum bending moment, anchor force and factor of
1275
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 DISCUSSION

safety from eight different conventional design procedures with the results from
finite element analysis and with the observed b e h a ~ i o u r The
. ~ ~ design methods
included both theoretical and empirical approaches, the latter providing the best
agreement with observed behaviour. The distributionof bending moment and the
deflexion of the wall from the finite element analyseswere in good a.greement with
the observation^.^^

References
31. SYMONS I. F. et al. Behaviour of a temporary anchored sheet pile wall on the Al(M) at
Hatfield. Transportand Road Research Laboratory,Crowthorne, 1987, Research
report 99.
32. BRITISHSTEEL CORPORATION. Piling handbook. British Steel Corporation, 1984.
33. BARDENL. Sheet pile wall design based on Rowe’s method. Construction Industry
Research and Information Association, London, 1974, Technical Note 54.
34. OVEARUP & PARTNERS. Diaphragm wall analysis and design: British Library Euston. Ove
Arup, London, 1979,1980, Reports S14 and S22.
35. COMITE EURO-INTERNATIONAL DU BET6N and FEDBRATION INTERNATIONALE DE LA
PRECONTRAINTE. Model code for concrete structures. CEB and FIP,1978.
36. SCOTTR. H. Theshort-term moment-xrvature relationship for reinforced concrete
beams. Proc. lnstn Civ. Engrs, Part 2,1983,75, Dec., 725-734.
37. MEASOR E. 0. and WILLIAMS G. M. J. Features in the design and construction of the
Shell Centre, London. Proc. Instn Ciu. Engrs,1962,21,475-502.
38. WARDW. H. Displacements and strains in tunnels beneath a large excavation in
London. Proc. 5th. Int.Con$ Soil Mech., Paris,1961,2,749-753.
39. MOOREJ. A. F. Discussion on Settlements associated with piles. Settlement ofstructures.
Pentech, 1975,718-719.
40. JARDINER. J. et al. The measurement of soil stiffness in the triaxial apparatus. Gbotech-
nique, 1984,34, Sept., 323-340.
41. ATKINSON J. H. et al. Effect of recent stress history on the stiffness of overconsolidated
soil. GCotechnique, 1990,40, Dec., 53 1-540.
42. SIMPSON B. et al. A computer model for the analysis of ground movements in London
clay. Gbotechnique, 1979,29, June, 149-175.
43. ROWEP. W. The relevance of soil fabric to site investigation practice. Gbotechnique, 1972,
22, June, 195-300.
44. BRITISH STANDARDS INSTITUTION. Structural use of concrete. BSI, London, 1985, BS 8110,
Parts 1 and 2.
45. CLOUCHG. W. et al. Movement control of excavation support systems by iterative
design. Foundation engineering: current principles and Practices, 2. American Society of
Civil Engineers, New York, 1989.
46. TERZAGHI K. Theoretical soil mechanics. Wiley, New York, 1943.
47. JARDINER. J. et al. Studies of the influence of non-linear stress-strain characteristics in
soil-structure interaction. Gtotechnique, 1986,36, Sept., 377-396.
48. JARDINE R. J. et al. Some practical applications of a non-linear ground model. Deforma-
tion of soils and displacements of structures. Proc. 10th Eur. Conf: Soil Mech., Florence,
1991,1,223-228.
49. KAVVADAS M. and BALICHM. M. Non-linear consolidation analyses around pile shafts.
Proc. 3rd Int. Conf: Behaviour Oflshore Structs, Boston, 1982,1,325-337.
50. WHITTLE A. J. et al. A model for predicting the performance of TLP piles in clay. Proc.
Conf: Behaviour Offshore Structs, Trondheim, 1988,97-112.
51. JARDINE R. J. and BOND A. J. Behaviour of displacement piles in a heavily over-
consolidated clay. Proc. 12th Int. Conf:Soil Mech., Riode Janeiro, 1989,2,1147-1152.
52. BURLANDB.J.et al. Movements around excavations in London clay. Design parameters
in geotechnical engineering,1. British Geotechnical Society, London, 1979,13-29.
1276
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.
 9585

53. DAYR. A. and P o n s D. M. A comparison of design methods for propped sheet pile walls.
Steel Construction Institute,Ascot, 1989, Publication 077.
54. DAYR. A. and POTTSD. M. (BEERet al. (eds)). Finite element analysis of the Hatfield
wall. Computer methods and advances in geomechanics. Balkema, Rotterdam, 1991.

1277
Downloaded by [ Jegatheesan Mathanabalan] on [11/01/19]. Copyright © ICE Publishing, all rights reserved.