The Effectiveness of Spatial Interpolation of Sparse PCPT Data to Optimise Offshore Design

The geotechnical data available to offshore infrastructure designers is invariably sparse, necessitating the use of engineering judgement in deriving (or estimating) soil design properties at untested (unsampled) locations. This task becomes more challenging when dealing with seabeds having a complex (layered) soil stratigraphy. Recent (and growing) interest in data-centric methods and their application to sparse datasets has seen progress in the spatial interpolation/extrapolation of geotechnical data using statistical and analytical approaches. This paper describes a case study involving one such approach where Bayesian Compressive Sensing and Markov Chain Monte Carlo methods are applied to a sparse two-dimensional PCPT dataset obtained from an offshore deep-water location comprising a layered (non-uniform) seabed. Results from the study are used to examine the ability of the considered approach in addressing soil variability and uncertainty in PCPT parameter estimation and highlights the difficulty in applying such approaches to complicated real-world settings.

Keywords: Site characterisation, offshore foundations, uncertainty, spatial variability