Model uncertainty of SPT-based method for evaluation of seismic soil liquefaction potential using multi-gene genetic programming

In this paper, the model uncertainty of the developed standard penetration test (SPT)-based model for evaluation of liquefaction potential of soil is estimated within the framework of the first-order reliability method (FORM). First, an empirical model to determine the cyclic resistance ratio (CRR) of the soil is developed, based on the post-liquefaction SPT data using an evolutionary artificial intelligence technique, multi-gene genetic programming (MGGP). This developed resistance model along with an existing cyclic stress ratio (CSR) model forms a limit state function for reliability-based approach for liquefaction triggering analysis. The uncertainty of the developed limit state model is represented by a lognormal random variable, in terms of its mean and the coefficient of variation, estimated through an extensive reliability analysis following a trial and error approach using Bayesian mapping functions calibrated with a high quality post-liquefaction case history database. A deterministic model with a mapping function relating the probability of liquefaction (P-L) and the factor of safety against liquefaction (F-s) is also developed for use in absence of parameter uncertainties. Two examples are presented to compare the present MGGP-based reliability method with the available regression-based reliability method. (C) 2015 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.