An improved notion for the computation of strain ratio in equivalent linear site response

Abstract

The geotechnical engineering community widely uses the one-dimensional (1D) Equivalent Linear Site Response Analysis (EQLSRA). However, the literature review reveals that surface ground motion characteristics predicted employing EQLSRA are often overdamped, especially at high frequencies. Recent studies have pointed out that the algorithm used within EQLSRA for obtaining the Equivalent Shear Strain (ESS) is one of the primary reasons for its unsatisfactory performance. Usually, for a given strain time history, the corresponding ESS is taken as Strain Ratio (STR) times the absolute peak strain. Conventionally, a constant value (usually ranging from 0.40 to 0.75 and typically 0.65) is manually set as STR. Though ESS plays a pivotal role in EQLSRA, not much attention has been given to the estimation of STR. This work aims to resolve such issues. Here, a new notion for the computation of STR is proposed, which can be easily integrated within EQLSRA. The coined idea takes into account the frequency content and intensity in the induced seismic strain waveform for the determination of STR. The suitability of the modified EQLSRA (mod-EQLSRA) incorporating the proposed technique of STR calculation is tested with the help of 15 Kiban Kyoshin Network (KiK-net) stations and 110 ground motions. It is found that the average absolute error in the surface Peak Horizontal Acceleration (PHA) predicted employing the mod-EQLSRA reduces by approximately 54% when compared to that estimated via the traditional EQLSRA.