Abstract

Infrastructure failure due to soil liquefaction has been repeatedly observed in past megathrust earthquakes, causing significant material and structural functionality losses. In most seismic regions, soil liquefaction potential is assessed using updated versions of the cyclic-stress-based simplified procedure initially proposed by Seed and Idriss in 1971. However, the application of these procedures to large-magnitude (Mw > 7.5) subduction earthquakes has shown discrepancies between forward predictions and field observations, particularly regarding liquefaction triggering and manifestation. This paper proposes an alternative model to assess soil liquefaction due to large-magnitude subduction earthquakes based on excess pore water pressure ratios and shear deformations. The triggering criteria are based on the peak values of excess pore pressure ratio and shear strain anticipated within the critical, potentially liquefiable soil layer. The model considers liquefiable layer thickness and relative density, along with input motion's Cumulative Absolute Velocity (CAV), as the main predictors of soil liquefaction. To this end, a numerical model was first developed and validated against results from a dynamic centrifuge test simulating free-field conditions. The calibrated numerical model was then used to perform a numerical parametric study to identify the trends and key predictors of liquefaction in layered soil deposits subjected to large-magnitude subduction earthquakes. Finally, a simplified probabilistic procedure, validated against available case histories, was developed to estimate the probabilities of full, marginal, and no liquefaction occurrence within each critical layer. © 2024 Elsevier Ltd