Large Deformation Finite Element Analysis of Spudcan Penetration in Sand Overlying Clay Profile

Abstract

Abstract Sand overlying clay soil profile is typically associated with punch-through risk during preloading. Computation of load-penetration response is usually performed in accordance with methods detailed in SNAME (2008) and ISO (2012). These methods are based on "wished-in-place" simplification that neglects the changes in soil stratigraphy during spudcan penetration. It has been demonstrated in both physical modelling experiments and large deformation finite element analysis (LDFE) that a sand plug is extruded as the spudcan penetrates through the upper sand layer. The presence of this sand plug beneath an advancing spudcan significantly alters the load-penetration response. This paper presents a case study on the application of LDFE analysis as recourse to compute the load-penetration response of a spudcan in sand overlying clay profile. The LDFE analysis is performed using the Coupled Eulerian-Lagrangian functionality in ABAQUS. In an Eulerian analysis, the spatial position of the nodes is fixed and materials are allowed to flow from one element to another element. Owing to the large deformation characteristic of the spudcan penetration problem, the Eulerian approach is an attractive solution strategy. In order to simulate the phenomena of progressive strength degradation with increasing accumulation of plastic deformation in the clay layers, a modified Tresca material model with strain-softening is implemented via a user-defined subroutine. The installation of the spudcan is modelled as a continuous spudcan penetration problem ("pushed-in-place") with the spudcan initially positioned at the mudline and progressively pushed downward to the final penetration. When compared against the spudcan penetration curves based on conventional methods, the numerically simulated load-penetration response is different in three important aspects. Firstly, the LDFE model indicates that rapid leg penetration will occur over a distance of approximately 8m as compared to 20m computed based on conventional methods. Secondly, the LDFE model shows a modest 10% reduction in bearing capacity over the depths at which rapid leg penetration is expected to occur while calculations based on conventional methods indicates a 45% reduction. Lastly, the LDFE model predicted a final penetration depth that is significantly shallower than that computed using conventional methods. With a LDFE model, it is not necessary to prescribe the failure mechanisms a-priori. The governing failure mechanism, which is a manifestation of the natural tendency for failure planes to develop along the weakest admissible path, is an output from the simulation. In contrast, a conventional spudcan penetration analysis is performed based on a "wished-in-place" approach without taking into account of the occurrence and effect of the composite footing formed by the spudcan and the sand plug. In the conventional approach, the sand layer is assumed to diminish as the spudcan penetrates beyond the clay-sand interface.