For many years, engineers have designed foundations, walls and culverts for highway and other geotechnical applications using allowable stress design (ASD) methods. In ASD, all uncertainties in the load and resistance are combined into a global factor of safety which, unfortunately, leads to uncertain safety margins in the design. The determination of system failure probability requires a coherent method of design for the geotechnical system. The Load and Resistance Factor Design (LRFD) approach allows designs to be targeted to
acceptable failure probability levels, which depend on the limit state being avoided. This research proposes Load and Resistance Factor Design provisions for the ultimate limit state punching shear failure of deep foundations. The load factors currently used are as specified by the National Building Code of Canada. The geotechnical resistance factors required to achieve a certain acceptable failure probability are estimated as a function of the spatial variability of the soil and of the degree of site understanding. A mathematical theory is developed to analytically estimate the failure probability of deep foundations in which the spatially random soil field is modeled using random field theory. The analytical results are validated by simulation and then used to estimate failure probabilities and geotechnical resistance factors required for design.