Full probabilistic geotechnical design under various design scenarios using direct Monte Carlo simulation and sample reweighting

common to encounter a diversity of design scenarios with different values and statistics of loads and/or geotechnical parameters in design practice of engineering geology and geotechnical engineering. Maintaining a uniform level of reliability close to the target reliability under various design scenarios is a key goal of geotechnical reliability-based design (RBD), which is a fundamental challenge for semi-probabilistic RBD methods because a wide range of design scenarios can be involved in geotechnical design practice. This paper develops an efficient RBD updating approach under a full probabilistic design framework, which combines direct Monte Carlo simulation (MCS) with a sample reweighting technique to efficiently obtain final designs under different design scenarios by a single simulation run, avoiding repeatedly performing direct MCS for different design scenarios. This leads to considerable computational saving, particularly as the number of design scenarios concerned is large. The proposed approach is illustrated through a pad foundation design example and a rock slope design example. Results show that the proposed approach properly calculates failure probabilities of different possible designs under different design scenarios by a single run of direct MCS, provided that sufficient failure samples are generated in the direct MCS run to cover failure regions of different design scenarios, which are used to determine the updated final designs. Reliability of the final designs obtained from the proposed approach for various design scenarios is generally uniform and is close to the target reliability level prescribed for the design purpose. Moreover, the proposed approach is able to link site investigation efforts to design saving in an efficient and straightforward manner. It also provides insights into the sensitivity of the final design to uncertain parameters involved in design.