Effect of Design Models and Test Numbers on the Design of Pad Foundations

It is widely known that there are significant uncertainties in the design of foundation systems. Specifically these uncertainties are categorised into model errors, uncertainties due to the spatial variation of soils and test and procedural effects. Geotechnical engineering professionals have aimed to minimise the magnitude of these uncertainties by adopting refined testing techniques, numerical design procedures and increased site investigation schemes. However the magnitude of the uncertainty still remains unknown. This paper introduces a method to quantify the effectiveness of undertaking site investigations of increasing scope as well as the use of varying testing techniques and foundation design models. The process involves simulating soil profiles using random field theory with nominated statistical parameters. This allows the soil properties to be known in detail at all locations – a situation that is not possible with real sites. Several design methods are employed using either complete knowledge of the soil profile or data derived from simulated site investigations. Designs are compared on a failure and over-design basis, where conclusions regarding the sensitivity of the design model and scope of site investigation are drawn. A Monte Carlo analysis is undertaken where numerous soil profiles are generated to develop probabilities of failure and over-design.