Beruflich Dokumente
Kultur Dokumente
MANUAL
Civil & Structural
TRM 234
Rev 1
Date 12/07
Page 1 of 5
INTRODUCTION
Any structural design model is based on a number of assumptions. It is particularly important to
recognise the existence of these assumptions when modelling supports and boundary conditions.
When designing a structural element or frame, we generally assume that the structure is elastic and
that it is supported on rigid supports. Conventional design of pile caps and pad foundations is based
on the assumption that the foundation is rigid and the supports are elastic. These assumptions give
good results providing there is sufficient difference in stiffness between the two elements being
considered.
For larger foundations, such as large piled bases, rafts and piled rafts, the flexibility of the foundation
becomes significant. Take for example a large piled base. If we use the conventional assumption
that the foundation is rigid and the piles elastic, we can significantly overestimate the contribution of
the piles at the edge of the raft. This leads to overdesign of the reinforced concrete and
underdesign of the piles directly under the superstructure. Conversely, if we model the piles as rigid,
we can significantly underestimate the contribution of the piles at the edge of the raft, leading to
underdesign of the reinforced concrete and edge piles and overdesign of piles directly under the
superstructure.
In finite element analysis, the second erroneous assumption is more commonly adopted than the
first. It is also potentially more dangerous.
In order to model a raft foundation using finite element analysis, supports can be modelled as
springs, but it must be recognised that soil behaviour is non-linear; soil does not actually behave as
an elastic spring.
It is important to use realistic values of spring stiffness. With excessively stiff springs behaviour
tends towards the rigid support assumption. If the springs are too flexible, the errors associated with
considering the foundation as a rigid body are introduced.
The duration of the load also needs to be considered. Long-term stiffness values should be used for
behaviour under gravity loads, whereas immediate values are generally appropriate for wind loading
conditions. Depending on the soil conditions, the immediate stiffness could be 3-4 times higher than
the long-term stiffness. The effect of cracking and creep in the reinforced concrete foundation must
also be taken into account.
UNITS FOR SPRING STIFFNESS
At first appearance, the units for spring stiffness can be confusing.
For a point spring, the spring stiffness (K) is defined as the force required to produce a unit
displacement. It is therefore expressed in kN/m or more often kN/mm or MN/mm.
Some computer programs such as RAM Concept allow area springs to be defined. In this case, the
2
spring stiffness is defined as the pressure (kN/m ) to displace the slab by a unit distance (m) and is
3
therefore expressed in kN/m . The user specifies the area supported and the spring constant in
3
kN/m and the program automatically assigns the appropriate spring value to each point in the finite
element mesh. (In other programs the user must assess the area of slab supported at each point,
and then calculate the appropriate point spring value (kN/m); in this case a regular finite element
mesh should be used.)
2
Following the same logic, the units for line spring supports are kN/m .
WSP Group
TECHNICAL REFERENCE
MANUAL
Civil & Structural
TRM 234
Rev 1
Date 12/07
Page 2 of 5
SOIL-STRUCTURE INTERACTION
It is important to recognise that soil stress/strain behaviour is non-linear. Settlement depends on
pore water migration which is time-dependent and not directly reversible.
Soil providing continuous support to a slab does not act as a matrix of independent elastic springs,
as load applied at one point on the surface of the soil results in settlement at surrounding points.
Modelling of soil support as a matrix of independent elastic springs is therefore a crude
approximation and must be supported by parallel geotechnical modelling of the behaviour of the soil.
The apparent stiffness of the soil changes with the size of the foundation. For the same applied
bearing pressure, the settlement beneath a large raft is greater than the settlement below a small
pad. This effect can be modelled using VDISP or a similar analysis package. (VDISP is a relatively
simple geotechnical program that assesses the settlement of loaded areas of varying shapes
treating the soil as an elastic half space using the solutions proposed by Boussinesq and Mindlin.)
PLAXIS, a geotechnical FE program updated by WSP Geotechnics, allows full soil-structure
interaction to be considered.
PROCEDURE FOR ITERATIVE ANALYSIS OF FOUNDATION BEHAVIOUR
The following procedure has been used successfully for modelling foundations taking account of
soil-structure interaction. It involves a finite element analysis of the foundation, typically using RAM
Concept, and a geotechnical analysis of the soil behaviour using a programme such as VDISP for
rafts or PLAXIS for piled foundations and piled rafts.
1. A structural model is created using a set of initial spring values. The model is analysed and
a set of spring reactions is obtained. These spring reactions provide initial soil loading data.
(Note that for ground-bearing rafts, any support which is shown resisting uplift is invalid.
Some computer programs allow lift-off springs, which addresses the problem. In other
cases, springs that are in uplift need to be eliminated and the model re-run.)
2. The initial soil loading data is input into the geotechnical model which calculates resulting
settlements.
3. A new set of spring stiffness values are calculated from the spring reactions and the
settlements calculated from the geotechnical model.
4. The new spring stiffness values are compared with the initial values and if there are
significant differences the whole process is repeated.
Several iterations between the structural and geotechnical programmes may be required before
convergence of the spring stiffness values is achieved.
There are a number of techniques that can be adopted for speeding this process up. These involve
the selection of an appropriate matching structural and geotechnical grid and ensuring that this grid
is of a sensible size.
It should be noted that spring stiffness values are at best an approximation. It is advisable to carry
out sensitivity checks. What is the effect of doubling (or halving) the stiffness? If there is a hard (or
soft) spot under part of the slab how is the structure affected? Does long term deflection due to
creep change the results?
WSP Group
TRM 234
TECHNICAL REFERENCE
MANUAL
Rev 1
Date 12/07
Page 3 of 5
Soil type
Sandy soils
Loose sand
Medium dense sand
Dense sand
Clayey medium dense sand
Silty medium dense sand
Clay soils
cu = 40 75
75 < cu < 150
cu > 150
4,800 16,000
9,600 80,000
64,000 128,000
32,000 80,000
24,000 48,000
(q = 75 150)
(q = 150 300)
(q = 300 600)
12,000 24,000
24,000 48,000
> 48,000
2
TRM 234
TECHNICAL REFERENCE
MANUAL
Rev 1
Date 12/07
Page 4 of 5
A more accurate assessment can be obtained from the results of a pile test or from a simulation
using PLAXIS. Using the same example of a 600 mm diameter pile with 2000 kN safe working load,
figure 1 is a plot of load/settlement response under specific soil conditions.
4000
3500
Load kN or K Value
3000
2500
Settlement
Elastic Shortening
K Value x 1000
2000
1500
1000
500
0
0
10
15
20
25
30
35
40
Settlement mm
Figure 1 Pile load/settlement response for a 600 mm diameter pile, 2000 kN safe working load
The total capacity of a pile is the sum of the frictional resistance of the shaft and the bearing capacity
of the base. In practice, the bearing capacity of the base is not mobilised until the frictional
resistance of the shaft has been overcome. The pile behaviour is close to linear until this point is
reached. In this particular case, the ultimate shaft load is 3400 kN and the ultimate base load is
600 kN. As the ultimate shaft load is greater than the pile working load, the load settlement
response of the pile up to the working load is essentially linear.
The elastic shortening of the pile, which is both linear and reversible, is shown on the plot, and is
assessed from three components:
1) any free or low friction length below the pile head
2) a pile length over which friction is transferred
3) the whole pile shortening as a column after the ultimate shaft friction has been reached.
It is clear that the total settlement (soil settlement + pile shortening) and hence the spring stiffness
will vary considerably from project to project, depending on the soil conditions, the pile length and
the applied loading.
The response plot in figure 1 shows that in this particular case the spring stiffness of the 2000 kN
6
2
capacity 600 mm diameter pile at working load is 1.1 x 10 kN/m which is five times the initial
estimate. It should be noted that this is an immediate value and, depending on the soil conditions,
the long-term value may be significantly lower.
WSP Group
TECHNICAL REFERENCE
MANUAL
Civil & Structural
TRM 234
Rev 1
Date 12/07
Page 5 of 5
KEYWORDS
Foundations; rafts; piles; settlement; soil-structure interaction; spring stiffness.
Authors:
Andrew Smith and Matthew Sharratt, WSP Cantor Seinuk and WSP Geotechnics
Sponsor:
Group Technical Centre
Revision record
12/07
First issue (rev 1)
WSP Group