Mapping probability of liquefaction using geostatistics and first order reliability method based on CPTU measurements

The probabilistic frameworks of the simplified procedure for estimating the probability of liquefaction (PL) of subsoil have gained increasing interest in the literature. A procedure based on the multivariate theory is proposed in this study to address the soil liquefaction as a point failure problem using the piezocone penetration test (CPTU) data. In the procedure, geostatistics combined with the Box-Cox transformation is utilized to evaluate the uncertainties associated with the CPTU parameters including the cone tip resistance (qt) and sleeve frictional resistance (fs) at each spatial point over a site. Then the first order reliability method (FORM) modified to incorporate the Box-Cox transformation is conducted at each point to estimate the probability of liquefaction (PL) of the soil with the consideration of the parameter uncertainties. Detailed steps of the procedure include the trend removal, data transformation, semivariogram analysis, Kriging interpolation, back transformation, validation of the predictions, and reliability-based analysis. An illustrative example for implementing the procedure is presented. It is shown that the deterministic nature of the CPTU observations can be captured in the probabilistic analysis if the proposed procedure is applied. Moreover, the spatial map of PL of the soil over a regional site can be provided. The PL map can provide useful information for the design of ground improvement at the site.