Deep-penetrating spudcan foundations on layered clays: centrifuge tests

Abstract

Installation of independent-legged jack-up rigs in sea-bed sediments where a strong layer overlays weaker soil can lead to catastrophic ‘punch-through', with potential leg buckling or toppling of the unit. Although approximate methods of analysis exist for estimating the penetration resistance, these do not account for the distortion of the upper layer as it punches through into the lower layer and there has been only limited visual depiction of the failure mechanism that occurs for the case of strong clay overlying weaker soil. This paper reports results of centrifuge model tests undertaken to provide insight into spudcan foundation behaviour during undrained vertical penetration through a stronger clay layer into weaker material, varying the strength ratio between lower and upper soil layers, the thickness of the upper layer relative to the spudcan diameter and the strength gradient of the lower layer. The model tests included half-spudcan tests against a transparent window, allowing the soil flow to be captured continuously by a digital camera and subsequently quantified through particle image velocimetry (PIV) analysis, and separate full-spudcan tests to measure the penetration resistance. Four interesting aspects of the soil flow mechanisms were identified: (a) vertically downwards motion of the soil and consequent deformation of the layer interface; (b) trapping of the stronger material beneath the spudcan, with this material being carried down into the underlying soft layer; (c) delayed back-flow of soil around the spudcan into the cavity formed above the spudcan; (d) eventual localised flow around the embedded spudcan. The effect of normalised layer soil properties and geometry on the soil flow mechanisms and the form of the penetration resistance profile is discussed in the context of the likelihood and severity of punch-through failure. Typical critical failure modes involved punching shear, with clear shear planes in the shape of a truncated cone forming in the upper layer below the spudcan. Potential punch-through, with a peak in penetration resistance followed by some reduction, occurred for almost all cases investigated. The reduction in resistance became more severe as the strength ratio of the underlying layer to the upper layer reduced, and also as the upper layer thickness increased. The failure modes assumed by the currently available recommended practices are not consistent with those observed from the study, suggesting a more rational approach is needed.