Reliability Analysis of One-Dimension Seismic Site Response Analysis Programs for Deep Soil Sites

Seismic site response analysis for deep soil is a big challenge for ground-motion prediction in seismic design and seismic hazard evaluation. One-dimension (1D) seismic site response analysis methods, which are principally adopting equivalent linear theory, are still prevailing tools to analyse seismic soil response. The feasibility of such tools to evaluate deep soil response has not been commonly recognized and usually ignored. To straightforwardly address such issue, 30 strong-motion stations installed on soil layers deeper than 50 m from KiK-net strong-motion seismograph network are selected, and then 1218 seismic data recorded by the stations are collected. 5 widely-used seismic site response analysis programs, i.e., Shake2000 (SHAKE), SoilQuake (SQ), SoilResponse (SR), LSSRLI-1 (LS), and DeepSoil (DP), are used to calculate ground acceleration and corresponding 5%-damped response spectra. The characteristic parameters of a standardized response spectrum, which are peak acceleration (amax), characteristic period (Tg) and plateau value (βmax), are designated as the target indices. The analytical results demonstrate that the calculation accuracy of amax values exhibits noticeable discrepancy within programs, while those of Tg and βmax show little dependence on numerical programs. The consistency in calculating target indices implies the selection of 1D site response analysis programs for deep soil is not the prior consideration. Through variability and residual analysis, the scattering of calculation, which is generated by the 1D soil response analysis programs, comparing to the records stays remarkable. However, the predictions of the target indices basically remain in unbiased trends with the records. The analytical findings are expected to recognize the performance of 1D soil response analysis programs on deep soil, and to give advice on adopting 1D analysis programs for deep soils response analysis in engineering practice.