A Polynomial Model for Predicting Liquefaction Potential from Cone Penetration Test Data

Abstract

Liquefaction is a serious geotechnical hazard leading to catastrophic damages which cause life and property losses. In many instances it may be preferable to predict liquefaction susceptibility indirectly, by common in-situ tests such as cone penetration test (CPT). A new approach, for prediction of liquefaction susceptibility, is proposed which presents a polynomial model to correlate cyclic resistance ratio (CRR) predicated on subsoil geotechnical properties from CPT tests, viz. normalized cone tip resistance (qC1) and friction resistance (fs). The derived model is applied to a total of 182 data sets, including field investigation records from eighteen earthquakes. The performance of the proposed approach is compared to other available methods within a quantitative validation framework (e.g., Precision, Recall, andF-Score). Results indicate the accuracy and generalization of the proposed new approach in predicting liquefaction susceptibility.