Probability-based derivation of resistance factors for bearing capacity prediction of shallow foundations under combined loading

Abstract

ABSTRACTThe practical use of probabilistic methods in geotechnical code development is illustrated by the methodology adopted in a research project aimed at establishing resistance factors for the AASHTO Load and Resistance Factor Design Specifications for the Ultimate Limit State (ULS) of shallow highway bridge foundations. The backbone of the work were databases of shallow foundations tested to failure with more than 500 load test cases with different types of load combinations in different ground conditions assembled within this study. A main focus of the research was the analysis of the model uncertainty involved in the bearing capacity prediction. The model uncertainty was evaluated in a lump sum procedure by a model factor defined as the ratio of measured bearing capacity from load tests over calculated bearing capacity from a pre-defined design method. Using statistical procedures major sources contributing to the uncertainties in the bearing capacity prediction were identified. With the derived model factor statistics, resistance factors for different boundary conditions were established from probabilistic analyses. In this paper, the adopted methodology is critically reflected in the light of improvements especially in model uncertainty assessment available today.