Response spectrum-based intensity measure for underground structures

Abstract

The purpose of this study is to present a novel approach to constructing intensity measures (IMs) for underground structures using seismic response spectrums. The proposed IM takes the form of an integral of the product of the response spectrum and an arbitrary function that reflects the relative importance of each range of period to the seismic response of underground structures. To obtain the arbitrary function, an artificial neural network (ANN) technique is employed. To demonstrate the effectiveness of this method, we develop two different soil-subway station dynamic systems, referred to as Case I and Case II. Based on both cases, the probabilistic seismic demand model (PSDM) for the proposed IM and eight existing IMs are developed. The performance of the proposed IM is compared with that of the existing IMs in terms of their proficiency, practicality, and efficiency, as well as their ability to describe the probabilistic distribution of damage measure (DM). The results indicate that the proposed IM outperforms the existing IMs since it has the highest proficiency for both cases. Moreover, the fragility curves calculated using the proposed IM are more effective than those calculated using traditional IMs, with narrower uncertainty ranges, which can help reduce the uncertainty in evaluating seismic risk. Comparing the function solved by the ANN with the seismic response of subway stations under different periods of harmonic excitation, it can be found that the solved function and the seismic response reach their peak points at the same period. This finding suggests that the proposed IM can account for the interaction between the surrounding soil and underground structures, as well as the softening of soil layers during earthquakes, making it a suitable IM for underground structures.