On efficient seismic fragility assessment using sequential Bayesian inference and truncation scheme: A case study of shear wall structure

Abstract

The seismic performance of shear walls in nuclear power plants is critical because a failure of such structures can cause severe accidents. Thus, research on the seismic fragility of shear walls in nuclear power plants has been attempted. However, conventional seismic fragility analysis requires high computational costs. In this context, several researchers proposed an efficient seismic fragility method by reducing required numerical analyses. But, this method still demands quite an amount of numerical cost. To overcome this issue, this paper presents an enhanced method for more efficiently assessing the seismic fragility of shear walls through sequential Bayesian inference and truncation strategy. The core idea of the proposed method introduces a binary analysis result chain and combines such a chain with sequential Bayesian inference. The sub-strategy of the method employs the truncation strategy for reducing the number of analyses. Also, an approach to evaluate a convergence index of a seismic fragility curve for each step is embedded into the sequential Bayesian inference for minimizing additional analysis cost. As a result, the proposed method of this paper reduces the computational cost even though it provides similar accuracy results compared to the existing seismic fragility methods.