Probabilistic Framework for Strength Limit and Service Limit Checks of Drilled Shafts considering Soil Spatial Variability

This paper presents a performance-based, probabilistic framework for design of a drilled shaft under axial and lateral loading that can consider spatial variability of soil properties at a project site. The performance criteria of a drilled shaft are stated in relation to the limiting tolerable deformations for strength limit and service limit, respectively. The computational algorithm for calculating the deformation of a drilled shaft is based on the commonly adopted load transfer method and the p-y method. “Geotechnical failure” is defined as an event in which the specified performance criteria are not met. Three failure modes are considered: axial movement, lateral deflection, and angular distortion. The spatial variability of soil properties is considered by using random field modeling techniques in which correlation length is introduced to account for site variability in addition to mean and variance. The method of fitting a sample autocorrelation function to a prescribed correlation function by using the method of ordinary least squares is introduced for determining site-specific correlation length for soil parameters. Geostatistical principles known as kriging are employed to estimate unknown parameters at unsampled locations from neighboring sampled locations. A numerical example is given to illustrate the application of the proposed methodologies. The example demonstrates that correlation length is an important statistical descriptor for characterizing site variability. Performance-based design provides unified consideration for both strength limit and service limit. Finally, the overall probability of failure for a drilled shaft when all three failure modes are considered is greater than the failure probability for any individual failure mode.