Identification of ground motion intensity measure and its application for predicting soil liquefaction potential based on the Bayesian network method

Abstract

In an attempt to reduce the parameter and model uncertainties of the Bayesian network model for predicting earthquake-induced soil liquefaction, 31 candidate intensity measures were investigated by the analyses of correlation, efficiency, proficiency, and sufficiency based on a large database of historical ground motion records. Two new Bayesian network models were developed using the identified intensity measures by combining the measured liquefaction-related data and the prior knowledge of soil liquefaction based on a large dataset of standard penetration tests. The results reveal that the root-mean-square acceleration is the optimal intensity measure for assessing soil liquefaction, whereas the peak ground acceleration is second best for liquefaction potential evaluation. The two new Bayesian network models with interpretability both in physical mechanisms and mathematics perform better than both the existing Bayesian network model and the Idriss and Boulanger model. The possible bias in this study is also discussed and pinpoints the importance of quantifying excess pore pressure development in the evaluation of soil liquefaction.