Abstract

In the literature, several simplified methods can be found to assess nonlinear liquefaction potential of soil. Derived from several field and laboratory tests, various procedures, also named as conventional methods, have been developed by utilizing case studies and undisturbed soil samples. In order to examine the collective knowledge built up in the conventional liquefaction methods available in the literature, a General Regression Neural Network (GRNN) model is proposed herein, which incorporates the parameters ignored in the past and accordingly will eliminate the shortcomings of the existing design formulae. Two, separate sets of field data, based on the standard penetration test, SPT, and the cone penetration test, CPT were used to develop the GRNN model. The proposed GRNN model predicted the occurrence/nonoccurrence of soil liquefaction well in these sites. Furthermore, liquefaction decision supported by SPT test results is incorporated into CPT based soil and seismic data. Therefore, the model supports the data conversion of an SPT-to-CPT throughout the liquefaction potential analysis, which believed to be the primary limitation of the simplified techniques. Thus the proposed model provides a viable tool to geotechnical engineers in assessing seismic condition in sites susceptible to liquefaction