Probabilistic evaluation of liquefaction analysis in performance based design framework

Abstract

Soil liquefaction during earthquakes poses a persistent challenge in geotechnical engineering, particularly in translating advanced numerical simulations into reliable, performance-based damage predictions. This study presents a novel framework that incorporates the maximum excess pore pressure ratio (PPR_max)—a simulation–derived yet underutilized Engineering Demand Parameter (EDP)—to directly predict liquefaction–induced damage under site–specific seismic loading conditions. Dynamic effective–stress finite element simulations were performed for soft alluvial soils in the seismically active İzmir–Karşıyaka region. Using logistic regression and receiver operating characteristic (ROC) analysis, PPR_max thresholds were statistically calibrated against observed damage levels to define transition points between minor and moderate damage. This calibration enabled the derivation of fragility curves linking peak ground acceleration (PGA) to probabilistic damage states within a regional hazard–consistent framework. The study further demonstrates the critical role of liquefiable layer thickness in controlling seismic pore pressure response. Even under identical ground motion intensities, variations in stratigraphy produced significantly different damage outcomes—highlighting a major gap in current seismic codes, which often neglect subsurface variability. The proposed framework enhances the predictive capacity of liquefaction risk assessments by bridging physics–based numerical modeling and empirical damage observations. It provides a scalable foundation for integrating simulation–compatible EDPs into performance–based seismic design and risk mitigation strategies.