Evaluating model uncertainty for reliability analysis of liquefaction potential of soils

The goal of this dissertation research is to improve existing methods for evaluating model uncertainty for reliability analysis of liquefaction potential of soils. The research deals with several closely related topics. First, probabilistic liquefaction triggering analysis using a Cone Penetration Test (CPT)-based model is conducted by means of the First Order Reliability Method (FORM). The basis for the reliability-based framework developed in this study is the ability to estimate the uncertainty of a limit state model (i.e., the deterministic boundary curve for liquefaction assessment). This model uncertainty along with the parameter uncertainty is required for a rigorous reliability analysis.; Secondly, the study of the CPT-based model as a limit state for reliability analysis of soil liquefaction is then followed by an extension to using a Standard Penetration Test (SPT)-based model as the limit state. In the meantime, the study of model uncertainty is continued with an emphasis on the treatment of prior probability. Effect of the assumed prior probability on the calculated final probability of liquefaction obtained through the FORM analysis is examined in detail, and an iterative procedure is developed to obtain a unique solution of the final probability of liquefaction through continuous calibration of both model uncertainty factor and the prior probability. The SPT-based analysis procedure, along with the method for estimating model uncertainty, is then applied to two other SPT-based limit state models, and the results from the three SPT-based models are compared. It should be noted that reliability analysis is conducted for a given pair of seismic parameters, the peak ground surface acceleration and the moment magnitude, and thus, the result of the FORM analysis is a condition probability of liquefaction.; Thirdly, a framework that integrates the conditional probability of liquefaction with the variation of earthquake sources and the underlying model uncertainty, as specified in the U.S. Geological Survey (USGS) Seismic Hazard Maps, for a fully probabilistic analysis of soil liquefaction in a given time exposure is evaluated and refined in this dissertation research. Field cases are employed in the demonstration analysis using this framework.; Finally, the conclusions are drawn based on the results obtained in this research and recommendations for future research are formulated and presented.