Evaluation and Optimization of Liquefaction Potential Index Using Both SPT and CPT

The liquefaction potential index (LPI) is a measure of the likelihood of liquefaction. An assessment is made of the LPI using standard penetration test data from 64 liquefied sites and 23 non-liquefied sites involving various earthquakes in Japan. Another assessment is made based on cone penetration test (CPT) soundings from 41 liquefied sites and 30 non-liquefied sites of Christchurch, New-Zealand. Based on the LPI, as the threshold for liquefaction becomes higher the likelihood of predicting liquefaction increases. However, as the LPI threshold increases, sites where liquefaction is not predicted, the likelihood of liquefaction increases to nearly 50%. To improve the prediction percentage of likelihood of liquefaction on SPT sites the constants of the resistance equation of LPI were optimized and Bayes theorem was applied. This showed marginal improvement in the prediction percentage. Further study was done to understand the factors that affect the likelihood of liquefaction. Relationships between confining pressure and relative density were plotted, which illustrated the importance of relative density and confining pressure with respect to weighing the soil layers. Thus, an attempt was made to improve the effectiveness of existing LPI equation by changing the weighting function by considering the depth factor to weigh the soil layers as a function of resistance. Later, the terms of existing LPI equation were modified as a function of resistance and were integrated over the resistance term. This new LPI equation was applied to 87 SPT sites and 71 CPT sites and Bayes Theorem was applied to check the probabilistic effectiveness of the equation. From the probabilistic analyses, it was found that the new LPI equation was marginally effective compared to the conventional LPI equation.