Stochastic evaluation of static liquefaction in a predominantly dilative sand fill

Abstract

The paper investigates the potential for liquefaction in a predominantly dilative hydraulic sand fill, and, in particular, whether pockets of loose material could have been instrumental in the failure of the Nerlerk underwater berm. This has been achieved stochastically, by linking random field theory for modelling spatial variability with finite elements for computing geostructural response. Specifically, a double-hardening soil model has been calibrated against data from 74 triaxial tests on Erksak sand, for a wide range of sand states, while state parameter statistics have been derived from 71 cone penetration tests from two Beaufort Sea sand islands. Multiple realisations of structural performance have then been carried out, by generating univariate random fields of state parameter, from which the soil model parameters have been inferred. The results demonstrate that it may be possible for a predominantly dilative fill to liquefy, due to the presence of semi-continuous loose zones arising from deposition-induced anisotropy.