305

Practical solutions to soil-structure

interaction problems
John C Small
University of Sydney, Australia

Summary in the raft and for moments induced into the

lnteraction of a structure and its foundation with the piles. lt is also shown that the use of thin plate theory
soil is discussed in this paper, and some of the for the raft can give good results for thick rafts and
numerical and analytical methods that have been can be used for raft and piled raft analysis. The effect
developed for the analysis of raft and piled raft of including the stiffness of the suPerstructure in the
foundations are presented. lt is shown that the use of analysis of the foundation is examined, and it is
simple spring models for the soil behaviour can lead shown that the type of structure and its stiffness can
to erroneous result and it is recommended that their have an effect on the deformation of the foundation.
use should be discontinued. Simple finite layer Finally, an analysis of an instrumented structure is
techniques are also examined, and results are carried out, and it is shown that reasonably good
compared with those of three-dimensional finite predictions of the behaviour of the foundation can be
element techniques. lt is shown that the finite layer made through the use of soil-structure interaction
techniques can yield good results for displacements theory and finite layer techniques.

Key words: soil-structure interaction; rafts; piled rafts; piles; foundations

Prog. Struct. Engng Mater.200 l; 3: 305-3 l4 (DOl: I0. 1002/pse.87)

l. lntroduction used in the early stages of desig^ to obtain some

understanding of foundation behaviour before more
The term 'Soil-structure interaction' covers a wide complex analysis is undertaken.
field, and can include the interaction of all types of
structures with the soil that they are constructed in or
upon. Foundations for structures such as buildings, 2. Surface raft foundations
bridges, silos and storage tanks, form the most
common type of problem considered, although arch or For raft or mat foundations constructed on or slightly
rectangular culverts and retaining walls of all types below the soil surface, many different analytic
form a large part of interaction analysis. approaches have been developed, and these are listed
Because the field is so broad, the topic is restricted in in the followi^g subsections.
this paper to analyses of foundations for structures
that involve raft or piled raft foundations under static 2.1. SpRtNG oR \ /INKLER FouNDATtoNs
load. Such foundations can be as simple as on-ground One of the earliest and simplest approaches to
slabs that are loaded by traffic, or wearhouse racking soil-structure interaction was to represent the
systems, or may extend to complex problems resistance of the soil by u spring. This aPProach was
involvitg pited raft foundations that are constructed applied to laterally loaded piles as well as raft
in non-uniform soils. foundations. However, for raft foundations, this
The development of methods for the analysis of approach can lead to serious error because:
soil-structure interaction involvi^g foundations has
taken place over many years and convenient accounts 1. The springs are independent and do not interact.
can be found in the books by Selvaduritrt and Therefore the compression of one spring does not
Hemsl eytz,st. Solution methods range from simple influence other parts of the foundation. To illustrate
hand calculation techniques to semi-analytical and this, consider the case of a uniformly loaded raft.
numerical methods that require complex computer Such a raft will undergo a uniform displacement
analysis. Each type of analysis has its field of and therefore there will be no bending moment
application, and in general, simple techniques can be predicted in the raft. This is obviously wrong, ds

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200 | ; 3:305-3 l4
305 SrnucruRAl ANALYSIs AND cAD

it is observed that such a loading would make

a rectangular raft (for example) deform into
winkler
a dished shape, and the raft would then carry Continuum
--
bending moments.
2. It is difficult to establish the stiffness values for the
springs that are used in analysis because the spring
constants are dependent on the scale of the
9?
tr"
G)

boL
il: -

foundation. For example, if a modulus of subgrade €

reaction is determined from a plate loading test, the c)
ol
load-deflection behaviour is specific to the size of
plate used in that test. It should not be applied to
loaded areas that are different in size from that of
the plate.
3. A Winkler or spring model cannot directly take
account of soil layering.
4. A vertical loading on a foundation may cause
lateral displacements. A spring model cannot be
used for such predictions.
Because of the limitations listed above, it is desirable
to use continuum models for the soil (i.e. treat it as
(.)
being an elastic or elastoplastic material). An example tr
o
of the differences in solutions obtained by using ....-E:{.01-.t
a spring model and a continuum model has been
ho
--6.1
=
presented by Brownt+l to illustrate the difference in the o
ca

choice of soil model. The problem involves unit point

loads applied to a strip raft (L/B:10). In order to
compare the two models, the modulus of elasticity of winkler
Continuum
the soil (continuum model) and of the subgrade
reaction (spring model) were chosen so that the -
settlements of a rigid strip foundation with a central
point load are equal. Fig. 1 shows the computed
moments in the raft where the raft stiffness is
lllll
defined as

-
16E1(1 \4)
I\ : ---------=--:--;= (1) Ir-tzl
ltE"L' q)
t|+.
x

and where EI :
bending stiffness of the raft; zr
:
v" Poisson's ratio of the soil; L : length of the raft; uol
tr
B: width of the raft; and E" : Young's modulus of the A)
co
soil.
From Fig. 1, it may be seen that the calculated moments
in the raft show reasonable agreement for the central Wnkler
Continuum
point load only. For the multiple point load cases there
is a large difference in the calculated moments. -
It may therefore be concluded that the use of spring (c) xlL
models may lead to large errors and that they should
not be used for raft foundation design. Fig. I Moments in strip raft: Winkler and continuum solutions

2.2. Sor-urroNs FoR uNrFoRM coNTtNuA For vertical loading the deflection of the soil at the
Solutions that treat the soil as an elastic continuum are interface can then be assumed equal to the deflection
superior to spring models in that they allow for of the raft at selected locations, and enough equations
interaction between loaded portions of a foundation or can be established to solve for the magnitudes of the
adjacent foundations. Early solutions to the problem of blocks of pressure or the unknown coefficients of the
a foundation on an elastic continuum involved terms in the arithmetic series. Solutions obtained using
assuming that the contact stress between the raft and these approaches include those of Cheung & NagtTl,
the foundation could be approximated, either as and Cheung & Zienkiewicztat for structures on
a series of blocks of uniform pressuretsl or as an infinitely deep soils, Browntel for strip footings
arithmetic seriestet. carrying point loads and for circular raft

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200 | ; 3:305-3 l4
 Sou-srRucru RE I NTERAcT oNs 307

2.0 2.4. MerxoDs FoR RAFTs oN LAvERED sorLs

I vs =o'5 For soils that are horizontally layered so that the
V5 -0 properties of the layer do not vary in the horizontal
\ --l
direction, either Fourier series or Fourier transforms
\j
cd
d
\- \.-- \- oo
may be applied to the field variables (i.e.
ts \\. displacements and stresses) to obtain solutions.
\\ \
I tJ

e{ c.,
\\_
I
5
Transforms can be applied to the contact stress,
\\\ \_ 5
represented either as uniform blocks of pressure or as
r!
0
\\
x \_____ an arithmetic function. Blocks of pressure may be used
g 2
I more generally, as they can be used with any shape of
()
1.0 '*--\ \ \ raft foundation and any loading pattern. Arithmetic
o \\\ hla:2
g.
a
functions can only be used in certain cases, for
I
A example a circular raft with a uniform load, where the
(!
Lr form of the functions can be chosen to suit the
,q 0.5 +----=.- tl problem.
0.5 An early solution to this type of problem was
obtained by Fraser & Wardletret who used integral
_0.2
hla: 0 0 5 transform techniques (Fourier transforms) to obtain
0.2
1 the response of the soil to the contact stress applied by
l0-2 l0-l I lol lo2 the raft. The raft was analysed by finite element
K
techniques. They presented solutions for the
settlement and bending moment in uniformly loaded
Fig. 2 Central displacement of uniformly loaded circular raft on rectangular rafts on layered soils of finite thickness.
uniform soil layer of depth h Their method of computing the behaviour of the
layered soil was approximate, involvitg a weighting
foundationstro, 111 on elastic soils of infinite or finite of the elastic parameters of each layer to obtain an
depth. Selvaduraitrzt and Rajapaksetrgl have also 'average' set of parameters.
presented solutions for circular rafts on infinitely deep Tham et al.Uzt first used finite layer methods to
soils. obtain a more rigorous solution to the problem of a raft
An example of such a solution is shown in on a layered soil. Zhang & Smalltrat also demonstrated
Fig. 2lot, where the differential deflection in a circular the use of finite layer methods to analyse a raft on
raft foundation is presented as a function of the a layered soil. Fourier transforms were used to obtain
stiffness of the foundation K, where: the response of the soil to blocks of uniform pressure,
and finite element analysis was used for the raft. This
K_E,(l
t\ -v3)f{) e) approach allowed a rigorous analysis of rafts on
E, \ot ) layered anisotropic soils of finite depth, and could
and where: E, is the modulus of the raft; E, is the easily incorporate lift-off of the foundation, or could be
modulus of the soil; v, is the Poisson's ratio of the used to limit the contact stress to a maximum value in
soil; f is the thickness of the raft; a is the radius of order to approximately model soil yield.
the raft; h ts the layer depth; and q is the uniform These semi-analytical approaches have appeal in
loading. that they may be used to analyse what is essentially
In most of these analyses, the raft was analysed by a three-dimensional problem, with fairly simple data
treating it as a plate or thin shell, so the theory of input. Their limitation is that they deal only
Timoshenko & Woinowski-Kriegertl4t could be used. approximately with soil yield, and soil layers must be
Whether the use of thin-shell theory is justified when horizontal. However, in most practical cases these
real foundations may be very thick (i.e. several metres) limitations are not of great significance as, in general,
is examined in Section 3.7.2. Treatment of the soil as loading is well below the failure load of the soil, and
a uniform material also has its limitations, and so sedimentary soils are often layered horizontally.
semi-analytic solutions were also developed for An example of a solution of the problem of a raft on
layered materials (see Section 2.4). a layered soil is shown in Fig. 3. The program FEAR
(finite element analysis of rafts), based on finite layer
2.3. BouNDARY ELEMENT TEcHNreuEs theory was used to compute the resultstlet. With this
Bound ary element techniques have also been used to technigue, the contact stress beneath the raft is treated
compute the soil deflection and also to look at rafts as a series of uniform blocks of pressure that
buried at depthtrst. Their approach is to model the raft correspond to each element in the raft. The deflection
by finite element methods and find the deflection of of a layered soil can then be calculated for each of the
the soil by boundary element techniques. However, if rectangular blocks of uniform pressure by the finite
the soil is layered, approximate techniques need to be layer techniquet2ot. The method is very simple to use,
used. as the raft can be of any shape, and can carry point,

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200 | ; 3:305-3 l4
308 SrnucruRAl ANALYsts AND cAD

SFT CN) CPT q" (MPa)

CONTOURS OF VERTCAL DISPLACEMENI 0 102030 24f6E
0
Cgnlour L.0.nd
YZ*'
tf$fl 5.oooE-0l
n'fiI 1.000ErOO
" -......, t.5008+00 E
2.000E+00
SNN ()
2 500E+00 tr
iffffti .J
3.000E+00
r*".1 q)
3.5008'00 v)
I
1'oooE:ao
r//ln
4.500E+00
ffi f00
o Mcasurcd
5.000 E
x hcrlictcd with interaction
. hcdicted without intcraction

Lay nrnd s<tll

Fig. 4 Deformation of foundation calculated with and withoutthe
suPerstructure

Fig. 3 Example of graphical output from finite layer analysis

(deflections in mm)

uniform or point moment loadings and (for the soil)

only the thickness of each soil layer and its elastic
properties are required. +0.02 ie (no wall)

2.5. FrNrrE ELEMENT METHoDs +0.01

The most powerful method for analysing rafts is the
finite element method. A full three-dimensional mesh 0
can be developed, and the raft and complete structure
on top of the raft can be incorporated. Different
constitutive laws can be used for the soil, such as
advanced elastoplastic models.
An early demonstration of the technique was made
by Smithtzrt, but for an axisymmetric problem only,
while Cheung & Zienkiewicztst looked at rafts
of general shape. With the development of computers, Fig. 5 Moments generated by the fluid in a liquid storage tank
with a wall
it is now possible to analyse quite complex
three-dimensional problems by using desk-top
displacement in the raft. Zhang and Smallt3ol analysed
computerslzzl.
three-dimensional framed buildings on raft
foundations, and demonstrated that the larger the
2.6. EFFEcT oF suPERsrRUcruRE relative stiffness of the building frame, the smaller the
In most of the analyses reported in the previous differential deflections in the raft.
sections, only the raft foundation has been considered Brown & Yutzzt also showed that as a building is
in the analysis. The actual superstructure is not constructed, the stiffness of the overall structure
considered, and column loads or distributed loads and increases and this affects the differential displacement
moments are applied directly to the foundation. If in the raft. Gusmau Filho & Guimaraesl3l] have also
there are extremely stiff structural elements such as looked at construction sequence and have noted that
shear walls or solid cores (i.e. as used for lift shafts), the loads in columns reach a maximum (or minimum)
these can be approximated by using very stiff elements value as more storeys are added to the building,
for the raft. These elements will then behave in a very leading to the idea of the building reaching a
rigid fashion, attracting moment and allowing little 'limit stiffness'.
differential deflection in the region of the stiff element. An example showitg how incorporation of the
Several researchers have examined the effects stiffness of the structure into the analysis can improve
of superstructures that do not behave very the predicted behaviour of a foundation has been
rigidlytzs-271. presented by Lopes & Gusmaot32l. For a lS-storey
Fraser & Wardletzst presented results for a two buy structure in Brazil, supported by u system of strip
portal frame, where they showed that the differential footings, the settlement distribution was shown to be
deflections in the frame depended on the stiffness of predicted more closely if the stiffness of the structure
the frame. Brownt2gl has also shown , for a strip raft is included in the settlement analysis (FiB . 4).
beneath a two-dimensional frame, that the relative Most of the papers cited in this section have
stiffness of the structure has an effect on differential contained the conclusion that the stiffness of the

Copyright O 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200l; 3:305-3 l4
 Sor l-srRucruRE NTERAcIoNS
I 309

structure does have an effect on the differential Many different means of analysi^g pited raft
deflection in the raft foundation, although for flexible foundations have been developed over the years (an
framed structures the effect is smalltsst. excellent review has been provided by Randolphtaal.
One type of structure where analysi^g just the Some of the methods that can be used for piled rafts
foundation is not adequate is a storage tank. The walls are similar to those used for surface rafts, and so, once
at the edges of the foundation resist the rotation of the again, it is convenient to group them into the followitg
base, and so create a large moment there. If the walls classes:
are not included in the analysis, then the moment at 1. Simple plate-on-springs aPProaches: these methods
the edge has to be assum ed zero, to zero rotation has to
treat the piles as springs, with the raft is treated
be assumed at the edge. This problem has been as a plate, and include the methods of Clancy
examined by Booker & Smallts+l, and Fig. 5 shows the & Randolphtsel, Poulostaot and Viggianitarl.
moments per unit length in the base and wall of 2. Bound ary element methods: these employ the
a liquid storage tank. From the figure it may be seen technique described above and include solutions
that there is a restraining moment generated at the obtained by Butterfield & Banerjeet+zl, Brown
edge of the base by the wall. In the figure, d - depth of
& Weisner[4s], Hain & Leel44l, Kuwabarat4sl and
tank; a - radius; tp : floor thickness; t* : wall Chowt+at.
thickness; T :
unit weight of fluid; M, - radial 3. Finite layer techniques: Ta and Smallt4Tl used finite
moment; z - distance from base of tank; and layer techniques to compute the behaviour of piled
K - stiffness of tank as defined in eq. 2 with t : tp.
rafts,where the piles were driven into layered soils.
Cheun g et al.t+sl had previously used series to
analyse the behaviour of pile grouPs in layered
3. Piled raft foundations soils, and the method can be extended to piled rafts.
Zhang & Smallt+st have extended these techniques
If a surface foundation is not adequate to carry to horizontal loading of a piled raft.
structural loads without excessive differential 4. Simplified finite element or finite difference
deflections, piles may be needed. Both the raft and the analyses: analyses can be carried out by
piles then transfer load to the soil, and the interaction approximating the piles as a two-dimensional or
problem involves both the raft and the piles. In some axisymmetric body and assigni^g 'smeared'
cases, the piles are placed beneath the raft solely to
material properties to the piles in order to
provide differential settlement control, and are approximate the actual three-dimensional
allowed to fail under loadtgsl. behaviour. That is, the solid continuous 'pile' in an
It is important to realize that piles do not need to be axisymmetric or two-dimensional analysis is given
uniformly placed over a foundation, but can be a lower modulus to make it comPress the same
judiciously placed so as to carry the larger loads or to amount as the actual individual piles. Analyses of
limit the differential deflections. In this regard, this sort include those of Desai et al.lsol and
it is useful to have a quick and simple computer Hoopertslt. Lin et al.tszt have used a finite difference
program or simple design method that can be used in technique to compute the behaviour of the soil
the design stage to determine the best layout of the beneath a piled raft, and applied the theory of piled
piles beneath the foundation. For examPle, Horikoshi rafts in Bangkok clay, using a two-dimensional
& Randolphtgot have shown that the optimum desig. finite difference grid.
of a piled raft carrying a uniform load would involve 5. Three-dimensional finite element analyses: as
piles placed under the central 76-25% of the raft area. computer storage has increased, full three-
Initially, piles were treated as grouPs that were dimensional analyses of piled rafts have been
rigidly joined at the head or carried equal loads, and carried out, and examples of this are given by
the flexibility of the raft that joined the pile heads was Zhuang et al.tsst, Katzenbach & Reults+I, Katzenbach
ignored. The book by Poulos & Davistgzl includes et al.1s5), Ottavianitsat and Reultszt.
many of the methods for comPuting the settlement of
piles or pile groups when the raft is assumed to be
totally rigid or totally flexible. (i.e. raft flexibility is one 3. l . NuMERTcAL MoDELLING
of two extremes). These solutions are based on treating In the previous section, many different methods of
the shear forces acting down the pile shaft as a series of pited raft analysis were listed, and the model chosen
uniform shear stresses acti^g over sections of the pile for a particular application would depend on the
shaft. Mindilin's equation for a sub-surface point load degree of sophistication required in the analysis. It is
is integrated over the section of pile to obtain the desirable to know the effects of assumptions made in
solution for the effect of the uniform shear stress on the different types of analyses and so, in the followi^g
deflections of the soil at other sections of pile for the subsections, a limited examination is made of some
pile itself or for other piles. Interaction between piles aspects of the analyses listed. The accuracy of finite
can therefore be found by this technigue, often called layer solutions, and of using thin-plate theory for the
a 'boundary element' technique. raft are the aspects examined.

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater.200l; 3:305-3 l4
3t0 SrnucruRAL ANALYSIs AND cAD

P*i +Y
-@Fp*o-
DS
ri
Fig. 7 Finite element mesh used for lateral loading of piled raft
Ovcrhang problem

Quantity Value

Pile diameter 0.5 m

Pile length l0m
Depth of soil 15 m
Raft width l,. SID - 3;9m
Raft breadth 8,. SID - 3;9m
Overhang of raft 0.5 m
Raft thickness 0.25 m
l-l o Soil modulus l0 MPa
Soil Poisson's ratio 0.3
Raft modulus 30 000 MPa
Fig. 6 Layout of laterally loaded piled raft Raft Poisson's ratio 0.3

3. l.l. Finite layer techniques

Simple techniques, such as the finite layer methods of shown in Fig. 8. It can be seen that the raft rotates
Zhang & Smallt+qt that are used to compute soil under the horizontal loading and, at its centre (x - 0),
movements from analytic or semi-analytic techniques, does not undergo vertical movement. The comPuted
may produce errors because of the aPProximations results from the finite layer and finite element
made in the analysis. In order to test the accuracy of methods can be seen to be in reasonably close
such methods, solutions were obtained from a agreement.
three-dimensional finite element program, and from The moments in the piles may also be computed,
a finite layer program (APRAF) for a piled raft with and are shown in Fig 9. Moments down the pile shaft
a horizontally applied loadi^g. are shown for pile 1 (the corner pile) and pile 5 (the
The raftis shown in Fig. 5 and consists of a 3 x 3 pile centre pile), and it may be seen that there is very close
group with a raft in contact with the ground surface. agreement between the finite element and finite layer
The raft overhangs the piles by one pile diameter values.
(around the perimeter). The finite element mesh used
to model this raft is shown in Fig. 7 where only
one-quarter of the raft is modelled because of 3. 1.2. Effect of raft modelling
symmetry. The mesh extends further in the r-direction It is of interest to know the effects that the assumptions
because loadi^g is to be applied to the raft in that made in the numerical modelling of raft and piled raft
direction, and the bound ary should not affect the foundations have uPon the solutions. The method of
results by being too close. All the ProPerties of the modelling the raft can vary a great deal, and can affect
piled raft are given in Table 1. the accuracy of the calculated behaviour.
Two horizontal point loads were applied to the The raft can be analysed by the use of finite
heads of each pile (18 loads in all) making a total difference or finite element techniques. Often it is
horizontal loading of 18 MN. For PurPoses of treated as being a thin plate, so that thin-plate theory
comparison, no slip was allowed between the raft and can be used in the analysis. For rafts 2-3 m thick, this
the soil, or the piles and the soil. The deflection of the may lead to inaccuracy, although it is the width-to-
raft can be calculated from the finite layer method, and thickness ratio of the raft that determines its flexibility.
a section (A-B in Fig. 6) through the deformed raft is To examine this problem, a raft was analysed by

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200 | ; 3:305-3 l4
 Sor-srRUcruRE NTERAcIoNS I 3tl
-60

.!r-CF APRAII
\ \a\ .l
40 - 6- Ei-i{ool
l.lulv wlw, tA/Ir]
f

\
E
tr
i>
.30
()
G)
-20

_0 50 -0 40 -0 l0
a,

-0 h \
X) 0. 0 0. !0 0. i0 0. 10 0.

G
q) \ ormalise l distaru x/8,
o
20
{ 'a

\
\"
40
[nt"arat*d..fi@ \
lloadinesS/D:3 |
\
60

Fig. 8 Vertical displacement of laterally loaded piled raft foundation along section A-B of Fig. 6

Bending monrent in pile (MN.m) -t5

-2 -l 0 t -16

I
a

-t7
f,

+) -18
\ + I So[d elenrn thbk raft

q)
It -19
f,
\ -+ Pbte ebrrcrls
o)
g,

-- -20
a
0
-21
-c -22
At
\ I
CJ

L
(l) -23
.1 S-,
-24
€"
c)
0.4 -25
! 0 3.2s
-3.25
o
.2 (a) Distance (m)

E
zo 0.6 -t5.1

-r5.3
E
E -r5.s
--
3 -rs.7
X
ti
-t5.e
E
€L
q)
- 16.1 + 2 So[d ehrnent thick raft t-
I -+ Plate elements
I_l
_l t 2m thir
raft
l+lSolll"@ I I
- 16.5
Fig. 9 Moment variation with depth for piles beneath laterally -3.25 0 3.25
loaded piled raft (b) D istance (m)

a finite element program, where the soil was modelled Fig. I 0 Vertical displacement of a thin and a thick raft, computed
with three-dimensional finite elements (20 nodes) and by shell or solid elements
the raft was modelled either by the same type of 20
node elements, or by thin-plate elements (havi^g
8 nodes). No slip along the pile shafts was allowed and 5 diameters spacing (see Fig. 5). Tests were carried out
the soil was assumed to be elastic, in order to model to make sure that the boundaries of the mesh were
piles in the same way as some of the simpler theories. sufficiently far away from the pile group so as not to
The finite element mesh used in the analysis was affect the result. Firstly the problem of a 9-pile raft
similar to that shown in Fig. 7, although not of such carrying point loads was analysed by the thin-shell
great lateral extent in the r-direction. The pile grouP elements for the raft, and then the problem was
was a 9-pile group (3 x 3) where the piles are at reanalysed with solid three-dimensional elements for

Copyright @ 2001 John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 2001; 3:305-3 l4
3t2 SrnucruRAl ANALYSIs AND cAD

700
J tl

!E
qtE
g''-
600
500
400
300
I
I
I
-:F Phte elenenrs
-+- I Sohd
"h*l !!! -t tu
-lo.zsm ttrick f

!.tr
tra
q) 200
trc)
oa r00
EE
b1 Z
7r
.&
0 -\ r A-* Settlement measuring Position

2.23 0.57
- 100

-200
\ I 2.5r 1.08

(a)
-300
-3.25 0

Distance (m)
3.25
iiIr.+zl ,oa:
700
i/ il
{--t
600
2m tHck I
raft
!El
cE
500
A R I

3.02
t'i 400
\ I
tra E
'rr
q,
300
200
f (pile load in MN)
E!
oa
EE
x3
q)
r00
0 .J
r \
T
ilr.06

- r00
f-6 2 SotiO elenrenrs thick raft
I + Pbte ebnrns
-200 I + | Solid elerrpnt thick raft
-300
-3.25 0 Fig. l2 Layout of pile positions and monitoring positions on
(b) Distance (m) the raft

Fig. | | Moments per unit length in thin and thick rafts, computed
by shell elements and solid elements

the raft. The properties of the raft were the same as

shown in Table 1, except that the piles were sPaced at
5 diameters, making the raft 5.5 m square in plan. The
point loads were applied in pairs midway between the
centre and outer piles along the x-axis. The loads
making up each pair were applied 0.25 m either side
of the x-axis. Each load had a magnitude of 1 MN,
making 4 MN, in total for all four point loads.
Fig. 10 shows the results for the case where the raft Fig. | 3 ldealization of H-pile cross-section
thickness is altered from 0.25 to 2 m. The calculated
deflections that are presented for section A-B of the
raft show that the deflections that were comPuted in loading, and this makes verifying theoretical
the raft are very close, regardless of whether the raft is computations for horizontal loadi^g more difficult.
modelled by solid elements or by thin-plate elements. As an example of the application of theory to a
The moments in the raft may also be computed for full-scale structure, the piled raft shown in plan in
both the thin-shell and solid elements. For the solid Fig. 72 was analysed. This structure has been
elements, the moments are computed from the stresses described by Yamashita & Kakuraitsst and is
in the elements representing the raft. The moments are supported on a piled raft. Sixteen H-piles of different
presented in Fig. 11 and again, even though very sizes were used to support the structure, as shown in
different means of analysis are used to compute the the figure. The soil, although layered, was treated as
moments, the thin-shell elements and the solid being uniform by Yamashita & Kakurai, who took the
elements give much the same result. modulus of the soil as E" :70.6 MPa and the Poisson's
ratio as vs : 0.4. The H-piles used beneath the piled
raft were 15 m long and were placed beneath the
3.3. AppLtcATroN columns of the building that applied a concentrated
There is not a great deal of data involving the load to the raft. The program PIRAFvzt was used for
performance of foundations for structures. This is the analysis, but as it is based on the piles being
partly due to the fact that it is difficult to fully circular, the H-piles were treated as a circular pile with
instrument foundations, and so often deflections at the radius ro and modulus Eo, where:
foundation surface are all that is measured. Often the
only data is for vertical loadin1, and not for horizontal (3)

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater. 200 | ; 3:305-3 l4
 Sor l-srRucru RE INTERAcIoNs 313

from the data presented the followi.g may be

o Measured (Yamashita and Kakurai l99l) concluded:
Calculated (Yamashita and Kakurai t99l)
2.0 PIRAF
1. The use of spring or Winkler models can lead to
z erroneous results and should not be used.
Eq)
t.5
Continuum models (for the soil) are a more rational
c)
o
lY
way to model the foundation soil, and a linear or
I-. O
E
x
l.o nonlinear continuum model is desirable.
2. When analysi^g rafts, or piled rafts, inclusion of
0.5
t--9. the stiffness of the superstructure will reduce the
differential deflections in the raft. The relative
0
stiffness of the superstructure will determine the
810 t2 L4 l6
effect, but for very flexible structures the raft alone
Depth (m)
without great error. Neglecting the
can be analysed
Fig. |4 Axial force distributed along pile B superstructure can be conservative, but there are
exceptions, such as the case of liquid storage tanks
where the moments at the edge of the tank are
large, owing to the restraining effect of the walls.
3. The use of thin-plate or thin-shell theory to model
the foundation raft can yield acceptable results, &s
Settlement (mm) Pile load (MN) was demonstrated by the examPle involving rafts
up to 2m thick.
Location sl s2 5354AB c 4. Simple models, such as those based on the finite
Measured 8.0 8.0 r0.5 3.0 0.53 2.08 t.35 layer method, can yield results of acceptable
Calcu lated 1sz1 9. I ll.5 t3.2 lt.4 l.ls 1.77 1.73 accuracy without the need to use full three-
PIRAFI+21 4.8 8. I t0.5 7.4 0.86 | .95 .63 |
dimensional numerical methods.
5. Reasonable predictions of raft behaviour can be
made for full-scale structures by simple techniques,
Ep : (EA)H/ n4 (4) provided the loads placed on the piles is not such
that pile yield or yield of the soil occurs.
where ro is the equivalent pile radius, and
Eo equivalent Young's modulus of the circular pile;
bs and a)s are the dimensions of the H-pile as shown References and recommended reading
in Fig. 13.
Fig. 14 shows the axial force calculated down the I I I Selvaduri APS. Elastic Analysis of Soil-Foundation lnteraction. New York:
Elsevier 1979.
shaft of pile B by both the PIRAF program and by
t2] Hemsley JA. Developments in raft analysis and design. In HemsleyJA (ed)
Yamashita & Kakurai, along with the measured Design Applications of Raft Foundations. London: Thomas Telford. 1998: 486-605.
values. It may be seen that for this pile, the calculated t3I Hemsley lA. Elastic Analysis of Raft Foundations. London: Thomas

values are less than the measured ones. However they Telford. 2000.

do have the same general shape, with the axial force t4l Brown PT. Strip footings. Lecture 7, GeotechnicalAnalysis and Computer
Applications, Department of Civil Engineering, University of Sydney. 1977: l.
reducing with depth. Differences may be due to the t5] Zhemochkin BN & Sinitsyn AP. Practical Methods of Designing
fact that the soil is not uniform as assumed by Foundation Beams and Slabs Resting on an Elastic Foundation. 2nd edn. Moscow:

Yamashita & Kakuraits8t. State Publishing House for Literature on Structures, Architecture and Structural
Materials, 1962.
Values of the computed settlements at various Gorbunov-Possadov Ml & Serebjanyi RV. Design of structures on
t6]
points on the raft are shown in Table 2, along with the efastic foundations, Proceedings of the Sth lnternational Conference on Soil
predicted and measured pile head loads. Although the Mechanics and Foundations of Engineerings, Paris, July 196l: | : 643-648.

pile head loads calculated with the PIRAF Program are t7lCheung YK & Nag DK. Plates and beams on elastic foundations - linear
and non-linear behaviour. Geotechnique 1968: l8: 250-260.
in reasonable agreement with the measured values, Cheung YK & Zienkiewicz OC. Plates and tanks on elastic foundations-
t8l
the deflections tend to show a different pattern. Both an application of finite element methods. lnternationalJournal of Solids and Structures
the PIRAF result and Yamashita & Kakurai's results | 965: l: 45 | -461.
predict larger settlements near the larger column t9I Brown PT. Strip footing with concentrated loads on deep elastic
foundations. 6eorechnical Engineering 1975: 6: l-13.
loads, but the measured values show a larger
I l0] Brown PT. Numerical analysis of uniformly loaded circular rafts on elastic
settlement at 51 than 52. However the predicted layers of finite depth. G6otechnique 1969: | 9(2): 30l-306.
settlements are of the correct order of magnitude. fl l] Bown PT. Numerical analysis of uniformly loaded circular rafts on deep
elastic foundations. Gdotechnique 1969: l9(3): 399-404.
fl 21 Selvaduri APS. Elastic contact between a flexible circular plate and
a transversely isotropic elastic halfspace. lnternationalJournal of Solids and Structures
4. Conclusions 1980: l6(2): 167-176.
fl 3] Rajapakse RKND. Interaction between a circular elastic plate and
In the previous sections, several aspects of a transverseley isotropic elastic half-space. lnternational Journal for Numerical and
soil-structure interaction have been examined, and Analytical Methods in Geomechanics 1988: l2(): 419-436.

Copyright @ 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater.200l; 3:305-3 l4
3t4 SrnucruRAl ANALYsts AND cAD

[f4] TimoshenkoS&Woinowsky-KrlegerS. Theoryof PlatesandShells. [37] PoulosHG&DavlsEH.Pilefoundationamlysisanddesign. NewYork:

McGraw-Hill. 1959. Wiley 1980.

[ | 5] Mandaf JJ & Gosh DP. Prediction of elastic settlement of rectangular raft [38] RildolPh MF. Design methods for pile groups and piled nfts. Proceedings
foundation - a coupled FE-BE approach. lntemational lournal for Numerical and of the Xlll lnternatioml Conference on Soil Mechanics and Foundation Engineering'
Analytial Methods in Geomechanics 1999: 2l: 263-273. New Delhi, 5- | 0 January 1994: 6l-82.
[ | 6] Fraser RA & Wardle Ll. Numerical analysis of rectangular mfts on [39] Clancy P & Randolph MF. An approximate analysis procedure for piled
layered foundation s. G*otechnique | 976; 26(4): 6 | 3-630. raft foundations. lnternational lournal of for Numerical and Analytical Methods in

[f7] ThamLG,ManKF&CheungYK.Analysisof footingrestingon Geomechanics 1993: 17(12):849-869.

non-homogeneous soil by double spline elemeng Computers and Geotechnics 1988: [40] Poulos HG. An approximate numerical analysis of pile-mft interaction.
5:249-268. InternationalJournal for Numerial and Analyical Methods in Geomechanics 1994:

fl 8l Zhang BQ & Small lC. The analysis of rectangular mfts of finite flexibility | 8(2): 73-92.
subiected to concentmted loads. Proceedings of the 6th Austnlia-New Zealand [4 | ] Yiggiani C. Pile groups and piled raft behaviour' ln: van lmpe WF and
Conference on Geomechanics, Christchurch 1992: 205-210. Haegman W (eds). DeeP Foundations on Bored and Augered Piles' BAP lll.

If9] SmaffJC.FEARlJser'sManual.Sydney:CentreforGeotechnicalResearch, Rotterdam:Balkema.1998:77-90.

University of Sydney, 2000. [42] Butterfield R & Baneriee PK. The elastic analysis of compressible piles
[20] Small lC & BookerJR, Finite layer analysis of layered elastic materials and pile groups. G&>technique l97l: 2l(l):43-60.
using a flexibility approach, Pan 2 - circular and rectangular loadings. Internatioml [{3] Brown PT & Weisner TJ. The behaviour of uniformly loaded Piled striP
Joumal for Numerial Methods in Engineering 986: 23: 959-978.
| fatings- Soils and Foundations | 975: | 5(4): | 3-2 l.
[2f ] Smith lM. Afinite elementapproach to elastic soil-structure intemction. [44] Hain S & Lee lK. Theanalysis of flexible raft-pile systems. G6otechnique
Canadian Geotechnicalloumal 1970: 7(2):95-105. 1978:28(l):65-83.
[22] Milovic SD. A comparison between obseryed and calculated large [45] Kuwabara F. An elastic analysis for Piled raft foundadons in
settlements of raft foundations. Canadian GeotuhniatJournal 1998:35(2): 25 -253. |a homogeneous soil. Soils and Foundations 1989: 29(l): 82-92.

[23] Lee K & Brown PT. Structure-foundation intemction analysis. lournal

f [46] Chow YK. Analysis of venically loaded pile groups. lntematiomlloumal
of Structunl Division (ASCE) 972: 9S(ST | | ): 2413-2430.
| for Numerial and Analytical Methods in Geomechanics | 985: | 0: 59-72.
[24] HainSJ&LeelK.Rational analysisof nftfoundation.loumal of [47I TaLD&Small JC.Analysisof piledraftsystemsinlayeredsoils.
Geotechnical Engineerrhg (ASCE) | 974: | 00(GT7): 843-860. lntemational lournal for Numerical and Analltial Methods in Geomechanics 1996:

[25] Lee lK. Structure-foundation-suPPorting soil intemction analysis. In 20:57-72.

Vaffiappan S, Hain S & Lee lK (eds) Soil Mechania, Recent Developments. Sydney: [,18] Cheung YK, Tham LG & Guo DJ. Analysis of pile grouP by infinite
University of New South Wales. 255-294. element layer method. G&ttechnique | 988: l8(3): 4 5-43 l. |

[26] Poulos HG. Settlement analysis of structuml foundation systems. [49] Zhang HH & Small JC. Analysis of capped pile groups subiected to
Proceedings of the 4th South East Asian Conference on Soil EngineeringKuala horizontal and vertiol loads, Computerc and Geotechnics2000: 26(l): l-21.
Lumpur, 1975: 4.544.61. [50] Desal CS, lohnson LD & Hargett CM. Analysis of pile supported
[27] Brown PT & Yu SKR. Load sequence and structure-foundation graviq lock. lournal of the Geotechnical Division ASCE | 00(GT9): | 009-1029.
intenctjon. Journal of Structural Engineering ASCE 1985: I l2( l); 48l-488. [5 | ] Hooper JA. Obsemtions on the behaviour of a piled-mft foundation on
[28] Fruer RA & Wardle Ll. A mtional analysis of shallow footings London clay. Proceedings of the lnstitution of Civil Engineerc 1973: 55(2): 855-877.
consideringsoil-structureinteraction. AustnlianGeomechanicsloumal1975: (Discussion 1974:51(2):547-552).
G51):20-25. [52] Lin DG, Bergado DT & Balaubramanium AS. Soil-structure

[29] Brown pT. The significance of structure-foundation interaction. intemction of piled raft foundation in Bangkok subsoil' Proceedings ofthe I lth Asian
proceedings ofthe Austnlia-New Zealand Conference on Geomechania,Brisbane, Regional Conference on Soil Mechanics and Foundation Engineering 1999: 183-187.
fEAust., f 975: l.l 79-gZ. [53] Zhuang GM, Lee lK & Zhao XH. Intenctive analysis of behaviour of
[30] Zhang BQ & Small JC. Finite layer soil-raft-structure
analysis of mft-pile foundation s. Proceedings of the lnternational Conference on Geotechnical
interaction. Proceedings of the Xlll lntemational Conference on Soil Mechanics and Engineering for Coastal DeveloPment - Theory and Practice on 1oft Ground - GEO-
Foundation Eryineering, New Delhi. 5-l0January 1994: 2: 587-590. COAST'91 Yokohama, (3-5 SePtember) l99l: l:759-764.
[3 | ] Gusmau Filho lA & Guimaraes LlN. Limit stiffness in soil-structure [54] Katzenbach R & Reul O. Design and performance of piled rafts.
intemction of buildings. Proceedings of the t4th lnternational Conference on Prcceedings of the l4th lnternatioml Conference on Soil Mechanics and Foundations

Soil Mxhanics and Foundations EngineeringHamburg 1997: 2: 807-808. EngineringHamburg 1997:1:2253-2256,

[32] Lopes FR & Gusmao AD. On the influence of soil-structure interaction [55I Katz€nbach R, Arslan U, Gutwald J & Holzhiuser J. Soil-structure-
in the distribution of foundation loads and seftlements. Proceedings of the t0th intenction of the 300m high Commezbank tower in Fnnkfun am Main.
European SMFE Conference Florence, l99l:2: 475-478. Measurements and numerical studies. Proceedings of the l4dt lntematioml

[33] YaoZE&ZhangJR.Anassessmentof theeffectsof structure/raft/soil ConferenceonSoil MechanicsandFoundationsEngineeringHamburgh, l99T:2:

intemction. Prcceedings ofthe Sth lnternational Conference on Numerical Methods l08l-1084.
in Geomechanics, Nagoya, Japan. I -5 April | 985: 2: I | 3-8 | 9. [56I Ottaviani M. Three-dimensional finite element analysis of venically loaded

[34] BookerJR & Small JC. The analysis of liquid storage tanks on deep elastic Pile trouPs. G&ttechnique 1973: 75(21: 159-174.
foundations. lnternational lournal for Numerial and Analytial Methods in [57] Reul O. Soil-structure-intenction of a piled nft foundation. Proceedings
Geomechanics l9g3: l: lg7-207. of the l2th European Young Geotxhnical Enginerc Conferencg Tallinn, Estonia,
[35] Hansbo S & Killstriim R. A case study of two alternative foundation 1988: l-12.
f -g: 23-27.
principfes. vi!-och vattenbyggaren lgg3: [58] Yamchita K & Kakurai M, setdement behaviour of the raft foundation

[36] HorikoshiK&RandolphMF.Optimumdesignofpiledmfrfoundations. withfrictionpiles.Proceedingsofthe4thlnternationalConferenceonPilingandDeep

Proceedings ofthe l4th lnternational Conference on Soil Mechania and Foundations Foundations, Stresa, ltaly, l99l: l: 461-466.
Engineering. Hamburg 1997: 2: 1073-1075.

John C Small
Associate Professor,
Department of Civil Engineering,
The University of Sydney,
NSW 2006, Australia.
E-mail: J.smal l@civil.usyd.edu.au

Copyright O 200 | John Wiley & Sons, Ltd. Prog. Struct. Engng Mater.200l; 3:305-3 l4