CPT-based probabilistic characterization of effective friction angle of sand

Recent development and implementation of reliability-based design codes around the world underscore the needs of probabilistic characterization of soil properties. A Bayesian approach is presented in this paper to characterize probabilistically the effective friction angle φ′ of sand using CPT data. The Bayesian framework is developed in conjunction with random field theory to model the inherent spatial variability and regression model to relate CPT measurements with φ′. The likelihood function and posterior distributions of uncertain parameters are derived. A sensitivity study is carried out to explore the effects of prior mean, variance, and correlation length on posterior mean.