Energy-based model for predicting liquefaction potential of sandy soils using evolutionary polynomial regression method

Abstract

This study presents a comprehensive, energy-based model for the evaluation of the liquefaction potential of all types of sandy soils, including clean, silty, and clayey sands. Unlike previous energy-based capacity models, this study considers the effect of Plasticity Index (PI) and loading Frequency (Frq) in the proposed model. Introducing an innovative method, the model was developed in two steps. At first, a small-scale model was developed using Evolutionary Polynomial Regression method (EPR), considering PI, Frq, Fine Content (FC), mean grain size (D50), and uniformity coefficient (Cu). For this aim, 63 laminar shaking table tests were performed on different fine-grained soils mixed with sand, with a wide range of PI (from 0 to 26) to provide a comprehensive data bank for EPR modeling. Then, by introducing an appropriate scaling factor, the model was scaled up to the field condition considering effective stress (σv′) and relative density (Dr). The model was successfully verified using a number of 100 case-histories worldwide with various soil properties. The results showed that the present model is considerably superior to previous energy-based models. Parametric analysis demonstrated that the liquefaction capacity energy of sandy soils with high fine content is mainly influenced by FC and PI, respectively.