Reliability Analysis of Soil Liquefaction Using Improved Hypercube Sampling (IHS) Method

Soil properties are uncertain due to various sources of variability of soil deposit formation. Currently, there is less reliable procedure to account for all pertinent uncertainties in predicting the occurrence of soil liquefaction phenomena when subjected to strong ground motion. In the present research, a series of reliability analyses using improved hypercube sampling (IHS) method have been performed to the latest modifications on liquefaction assessment proposed by Boulanger and Idriss for SPT-based simplified procedure to examine the soil liquefaction potential considering the soil and earthquake loading uncertainties. The analyses have been performed on 252 well-documented historical cases. Comparison of Monte Carlo (MC) and IHS results indicates that the IHS method is more accurate, reliable, and simple over MC for analyzing the reliability of liquefaction phenomenon. The sensitivity analysis illustrates that different parameters may have the most effect on the probability of liquefaction depending on soil conditions. A probabilistic relation has been obtained and a classification is suggested to predict the likelihood of liquefaction occurrence by the probability of liquefaction. It is found that the proposed model yields more accurate results in comparison to other similar previous models. In the absence of the knowledge of parameter uncertainty, the recommended correlation can be used to calculate the probability of liquefaction pertinent to a safety factor. The results show that the proposed approach is a favorable and useful tool for the reliability analysis of liquefaction potential estimations.