Assessment of Liquefaction Potential Based on a Probabilistic Model and Performing Reliability Analysis with Evaluation the Relative Importance of Model Parameters Uncertainty

Investigating the potential of soil liquefaction plays an important role in reducing earthquake damages. Prediction of this phenomenon is difficult due to the complexity of the nature of soil and earthquakes. Previous studies had many errors that include inaccurate modeling, inadequate databases and disregarding the uncertainties that are caused by soil and earthquake complexity. In this research Bayesian inference method is used as a probabilistic modeling method. This method used a comprehensive database of standard penetration test (SPT). For the first time, first-order reliability method (FORM) and importance sampling method were used to estimate the probability of failure and the reliability index of the limit state function of liquefaction. Then with the help histogram sampling, probability density function (PDF) and cumulative probability function (CDF) were obtained to investigate the probability of transgression. A sensitivity analysis of the model was also performed to estimate the most effective parameters. As a result of this study, a robust and efficient probabilistic model was developed to evaluate the liquefaction potential of soils. Comparing the results of this probabilistic model with other deterministic and probabilistic models showed a significant reduction in model uncertainty and standard deviation, increased accuracy and a better understanding of the relationship between failure probability and safety factor of liquefaction. Monte Carlo sampling and importance sampling methods were closed to each other. In the sensitivity analysis of the proposed model, the uncertainty of the magnitude of the earthquake parameter was identified as the most important uncertainty of the model.