Study On Logistic Regression Model-based Procedure For Evaluating Liquefaction Potential

Sand liquefaction is one of typical phenomenon of earthquake disasters,the probability of liquefaction discrimination method is one of the hot issues of geotechnical earthquake engineering research.Therefore,exploration of sand liquefaction fundamentals and development to consider the uncertainty of liquefaction expression method,has important theoretical application value.Based on the existing liquefaction survey data in mainland China,this paper uses the method of logistic regression analysis to establish the probability of sand liquefaction potential evaluating formula,and makes the critical standard penetration formula under the given number of liquefaction probability.The proposed probability formula and critical standard penetration value equation are verified by histories data from Taiwan Chi-Chi earthquake,and the calculation results of the proposed formula and the current Chinese Code for Seismic Design of Buildings method are compared.The main work and research results are as follows:1.The research status of sand liquefaction at home and abroad are reviewed.The seismic damage phenomenon of sand liquefaction,the liquefaction fundamentals and liquefaction influencing factors are summarized in this paper.The research status of sand liquefaction discrimination method at home and abroad are reviewed and commented.Based on analysing of the present study results,the existing problems of evaluating liquefaction potential are pointed out,and the proposals for further research are discussed.2.The existing liquefaction survey data in mainland China are collected and statistical analyzed.The background information of liquefaction date form 8 major earthquakes in mainland China are introduced,which have been brought about great seismic damage.The existing liquefaction survey data in mainland China are collected,and the distribution characteristics of existing liquefaction data such as intensity,standard penetration test data,soil depth and ground water depth are analyzed.This work provides the basic data for putting forward the probability of sand liquefaction potential evaluating method.3.The probability of sand liquefaction discriminant formula is established by using logistic regression model.The basic principles and its application fields of the logistic regression model are briefly introduced.The liquefaction probability discrimination formula using 4evaluation index such as intensity,standard penetration test data,soil depth and ground water depth,which is developed based on logistic regression analysis which consists of215 case of liquefaction/non-liquefaction occurrence form mainland Chinese major earthquakes.According to the results of regression model checking,and comparing thecalculation results of the current Chinese Code for Seismic Design of Buildings method,the reliability of the proposed formula is tested in this paper.4.The proposed formula is checked analysis by sand liquefaction data of Taiwan Chi-Chi earthquake.In this paper,the background information of liquefaction date form Taiwan Chi-Chi earthquake is briefly introduced.And the distribution characteristics of liquefaction data are analyzed,which are tested in Taiwan Chi-Chi earthquake.Based on those liquefaction data,the applicability of the proposed liquefaction probability discrimination formula and the reliability of the discriminant results are tested The results show that the computing results of the proposed formula are slightly conservative,but the overall corrective rate by the liquefaction probability equation is higher,which can better reflect the actual situation of site liquefaction.