CPT-based liquefaction analysis, Part 1: Determination of limit state function

Abstract

This paper is the first of a set of two papers that address the issue of earthquake-induced liquefaction potential. In this paper, a CPT-based limit state function for assessing the cyclic liquefaction resistance of sandy soils is presented. The liquefaction resistance of a soil is generally expressed as cyclic resistance ratio (CRR), while the seismic load that causes liquefaction is expressed as cyclic stress ratio (CSR). By definition, CRR is equal to the maximum CSR that a soil can sustain without the occurrence of cyclic liquefaction. In the present study, a new method to establish a limit state function for evaluating cyclic liquefaction resistance is developed. This new method is based on an artificial neural network (ANN) modelling and analysis of 225 field liquefaction performance records. First, an ANN model is developed to predict the occurrence of liquefaction based on historic field performance records. Second, a search procedure is developed to locate data points on the limit state surface. Third, another ANN model is created to approximate the multi-variable limit state function. The established approximate function, an ANN model, can be used to determine the CRR of a soil using CPT data. The developed CPT-based limit state function forms the basis for the development of a risk-based method for assessing cyclic liquefaction potential.