Application of the random finite element method to assess traditional techniques used to analyse free-field ground response and liquefaction triggering

Traditional one-dimensional (1D) techniques for analysing free-field ground response and liquefaction triggering rely on the assumption of ideal, homogeneous soil deposits, which are hardly ever encountered. This paper highlights the inaccuracies and limitations of 1D schemes and the motivation for two-dimensional (2D) strategies using the random finite element method (RFEM). Through Monte Carlo simulations, the 2D dynamic response of various soil domains, considering the impact of spatial variability of void ratio on liquefaction potential, is analysed. Each 2D realisation has been re-examined by splitting the domain into 1D soil columns while preserving similar variability attributes. The results reveal significant differences. While 2D schemes show a reduced variability in the ground surface responses and more realistic liquefaction spreading compared to 1D simulations, 2D schemes indicate more severe impacts on ground surface accelerations, response spectra peak values, and energy released. For site response analysis using a homogeneous soil profile, a characteristic void ratio value based on the mean minus 2 standard deviations is suitable for high PGA potential scenarios. However, the differences in responses between 1D and 2D schemes diminish if the input earthquake acceleration is not strong enough to cause liquefaction.