Evaluation of Finite Element, Finite Difference and Elasticity Methods for Hypothetical Raft Foundations Installed on Layered Strata

Abstract

Different numerical methods including 2-D finite element (FE) analysis were implemented for hypothetical raft foundations of high length/breadth ratios to evaluate the reliability of (1) Boussinesq’s method, (2) “rigid-punch” (RP) elastic methods and (3) finite difference (FD) incorporating subgrade reaction theory. A dedicated computer program was developed and used in addition to professional software packages to analyse uniformly loaded rafts of various length/width ratios and stiffnesses, interacting with multi-layered soils of different properties. It was found that the greater the flexibility of the raft the closer was the agreement between the deformations computed from FE and FD methods and the settlements predicted from FE and RP methods. Additionally, the FE method yielded the most realistic predictions principally because of its capability to model and account for the interactions between soil layers as well as the continuity of the soil support. The results revealed that soil stresses computed from elastic methods were inaccurate in cases where a relatively incompressible bedrock existed within depths equivalent to 4 times the raft width or less. This led to a recommendation that in order to predict raft deformations accurately, only those methods that account for both soil-structure and soil–soil interaction should be used in designing raft foundation in practice.