Improved Prediction of Spudcan Penetration Resistance by an Observation-Optimized Model

Abstract

The prediction models of penetration resistances of spudcan footings on sand overlying clay have been established by extensive centrifuge tests and large-deformation numerical simulations. However, most existing models have been based on deterministic analyses, where the measurement deviation of soil strength profiles was not considered. A parameter estimation technique to improve the predictions of the deterministic models is developed by using the penetration resistance data measured at depths shallower than the current. The divergence between the prediction of the deterministic model and the measurement is attributed to the uncertainties of soil properties. A deterministic model, termed HU15, is used to provide the original prediction, and two properties of the clay layer are adjusted with measured resistances prior to the peak resistance in the upper sand layer and after punch-through in the underlying clay layer. The parameter estimation technique combined with the HU15 model shows satisfactory performance when it is verified against the existing data from centrifuge and field tests.