Page 337 - Numerical Analysis and Modelling in Geomechanics

318 ANALYSIS AND DESIGN OF PILE GROUPS
which is evident in current revisions of national and regional design codes, may
partially be attributed to the common belief that predicting deformations is more
difficult and less reliable than predicting capacity. In reality, however, the
reverse is often true for pile foundations (Randolph, 1994; Mandolini and
Viggiani, 1997).
Thus, provided there is a minimum factor of safety, which may be as low as 1.
5, pile group design should be approached in terms of satisfying the settlement
criterion, rather than being based on a notional factoring of the ultimate state of
each pile (Fleming et al., 1992). If this design philosophy is adopted, and hence
low safety factors are employed, consideration of non-linear soil behaviour
becomes essential. This would result in an improved understanding of pile group
behaviour and hence in more effective design techniques.
In this chapter, the effects of soil non-linearity on pile group response, as
measured experimentally and as predicted by current numerical analyses, have
been discussed. A computer program, called PGROUPN, for pile group analysis
and design has been presented. It has been shown that the proposed method, by
taking into account the continuous nature of pile-soil interaction, removes the
uncertainty of empirical t-z and p-y approaches and provides a simple design tool
based on conventional soil parameters.
Use of the program may lead to a number of significant advantages in
practice. For example, even for a purely linear analysis, the PGROUPN solution
is capable of modelling important features of group behaviour which are
normally disregarded by the other numerical codes. Consideration of such
features is essential in order to obtain a more realistic prediction of the load
distribution between the individual piles of the group.
Another significant aspect of group behaviour which is not treated adequately
by the other numerical procedures is the effect of soil non-linearity. The main
advantage of a non-linear group analysis system over a linear one is that it has
the desirable effect of demonstrating a relative reduction of the corner loads in
large groups in both the horizontal and vertical senses. It has been shown that,
even at typical working load levels, this reduction is significant. These
observations are of basic importance in practice and may lead to tangible
improvements in design procedures and worthwhile savings in construction costs.

Acknowledgements
The author is grateful to Dr Ken Fleming, Cementation Foundations Skanska,
and Dr Nick Hurt, Halcrow Group, for reading the manuscript and providing
valuable comments.