A New Frequency Domain Framework of Inverse Ground Response Analysis for the Determination of Dynamic Soil Properties in a Two-Layered System

Abstract

Ground response analysis (GRA) requires the use of dynamic soil property curves (DSPCs). However, because regional DSPCs are not always available, globally recorded DSPCs are utilized in regional GRA the majority of the time. For those regions where regional DSPCs, based on laboratory estimations, exist, the DSPCs are affected in a variety of ways. For this reason, inverse GRA (IGRA), performed on instrumented earthquake (EQ) data, appears to be a realistic alternative for determining in situ DSPCs. Existing IGRA studies, based on downhole EQ data from around the world, show constraints, particularly in determining damping ratio (β) properties. Furthermore, previous frequency domain IGRA studies have only detected DSPCs for the surficial layer. Whereas a few of the frequency domain IGRA studies have determined the shear modulus degradation (G/Gmax) curve for multiple soil layers, they have been unable to determine the β curve for those layers. In the current study, we propose a novel frequency domain framework for determining the G/Gmax and β curves for a two-layered soil system positioned between successive accelerometer levels. The proposed framework was applied to 28 EQ records from the Lotung downhole array in Taiwan to determine the G/Gmax, β, and shear strain (γ) values for the top two layers from the ground surface. In addition, average DSPCs for the two layers are proposed. The effectiveness of the proposed DSPCs was tested by using them in a GRA and comparing them with recorded ground motions. It has been noted that, at present, there are limited downhole array records available. Therefore, there is ambiguity involved in applying the proposed framework presented in this paper to a larger number of records from downhole arrays. However, physical modeling of the wave propagation mechanism through layered media can be done on a smaller scale to determine dynamic soil properties in the future. Several attempts have been made to estimate the G/Gmax and β based on centrifuge testing using recorded acceleration time histories at different levels in the centrifuge model. In addition, the one-dimensional propagation of shear waves could be determined in those studies by taking special measures. However, these studies were based on a stress–strain imaging method in the time domain, which has its own limitations. However, recently, a frequency domain method was used to determine the G/Gmax and β from centrifuge modeling for a single soil layer. In that study, the β values obtained were scattered. Thus, the frequency domain framework proposed herein could be a viable alternative for determining the G/Gmax and β of a multilayered soil profile (here, two layers) when used in centrifuge testing.