Statistical review of CPT data and implications for pile design

Abstract

Natural soil variability can result in a range of pile capacities at a specific test site. Soil variability is ideally determined using in-situ testing techniques such as the CPT. This paper presents a series of CPT profiles from a dense sand site in Wicklow, Ireland. The qc value is detrended using a quadratic profile with depth and the scatter in the remaining data is quantified using the coefficient of variation (COV). In addition, the autocorrelation of each CPT trace is quantified using the scale of fluctuation. The measured values from the site, including the trend, COV and scale of fluctuation are subsequently incorporated into a Monte Carlo simulation that generates multiple realisations of the CPT data from an underlying lognormal distribution. These realisations are applied to a simple pile design model to calculate the variability in capacity that arises from the site specific variability. Spatial averaging is shown to be highly influential in reducing the uncertainty in pile resistance, with long piles and small scale of fluctuations demonstrating the lowest COV in the calculated capacity.