Bi-objective Optimization of Site Investigation Program for Liquefaction Hazard Mapping

Abstract

The development of a liquefaction hazard map generally requires field data from site investigations. In this study, a bi-objective optimization framework is proposed for selecting an optimal site investigation program considering both the accuracy of the liquefaction hazard map and the site investigation efforts. To validate the proposed framework, a three-dimensional synthetic soil field with extremely detailed soil properties is generated and the corresponding liquefaction hazard map is used as the benchmark. Both regular and random sampling-based site investigation programs with varying site investigation efforts are considered and are used to infer input parameters of the subsequent random field-based liquefaction hazard mapping. It is found that the random field-based liquefaction hazard maps generally overestimate the hazard when validated against the benchmark liquefaction hazard map. When site investigation efforts (quantified by the number of sounding sites) are the same, regularly spaced site investigation programs yield more accurate hazard maps than those by random sampling-based investigation programs. An optimal site investigation program is recommended for the study site and the proposed framework can be applied to optimize site investigation of other sites.