The impact of soil autocorrelation on pile load displacement behaviour

Foundation design is often controlled by the serviceability limit state and the mobilised settlement under operational conditions is often the governing design condition. Accurate predictions of pile displacements are often hampered by the inherent soil variability. This paper describes an analysis which incorporates the uncertainty in soil properties directly into the pile settlement calculations through a monte-carlo simulation. A t-z analysis is performed which assumes the axial load in a pile is resisted by non-linear uncoupled spring elements, which are dependent on the properties of the surrounding soil. The input soil parameters are modelled by log normally distributed variables. The ultimate friction mobilised by the soil springs is calculated using the Cone Penetration Test based Imperial College pile design approach. CPT data from an Irish dense sand test site is used in the analysis. The springs are assumed to be auto-correlated with depth in a similar manner to the CPT profile, with the degree of correlation defined by the scale of fluctuation. In the final section, the results are discussed in light of previous research which assumed uncorrelated soil properties.