Estimation of Timing of Liquefaction Using Spectral Energy Ratio

Abstract

The significant reduction in the stiffness of liquefied soil is accompanied by a decrease in the shear wave velocity, which ultimately results in the softening of the liquefied site. Time–frequency response analysis can identify the sudden drop in the frequency of the liquefied site, which has been widely employed to determine the onset of liquefaction. However, using the modal frequency (corresponding to the maximum power at each time step) to identify the timing of liquefaction (tL) captures the reduction in frequency during earthquakes, but it does not encompass the entire range of frequencies that have changed. Furthermore, previous literature defines tL as the boundary separating the modal frequency into pre- and postliquefaction time segments, but this estimate does not consider the generation of pore water pressure. Two representative case histories are presented to highlight the limitations of identifying tL by solely relying on the modal frequency approach that uses a two-step function. As a result, this study introduces an innovative method to identify tL utilizing the spectral energy ratio (SER), which captures the entire frequency shift. A step-by-step procedure using SER is detailed, and the new estimates of tL are compared with those derived from previous literature using 30 case histories. To validate the approach, a sensitivity analysis was performed using centrifuge test data from the Liquefaction Experiment and Analysis Projects. Results indicated that incorporating a ramp that accounts for pore water pressure buildup in the trilinear function improved tL estimation. An optimized SER value of 0.92 was determined for the proposed method. The notable contribution of this study is an enhanced approach of identifying the timing of liquefaction triggering by only utilizing acceleration records without requiring pore water pressure responses.