| Abstract | | Taiwan is located at an active plate boundary and is frequently affected by earthquakes. Many urban areas contain aging reinforced concrete (RC) buildings that may experience significant nonlinear response and damage during strong ground shaking. Therefore, assessing their seismic risk before future earthquakes is important for retrofit prioritization, disaster mitigation planning, and post-earthquake resource allocation. Conventional regional risk assessments often rely on empirical or simplified fragility curves, which may not fully capture the structural characteristics and nonlinear seismic behavior of individual buildings. Although probabilistic seismic risk assessment can provide more physics-based building-level risk estimates through nonlinear dynamic analyses and fragility functions, its application to city-scale building inventories requires a large number of repeated analyses and thus faces substantial computational challenges. To address this issue, this study develops a high-performance computing (HPC)-based framework for city-scale probabilistic seismic risk assessment. The framework integrates building inventory processing, automated structural model generation, incremental dynamic analysis, response post-processing, and fragility curve development. It is implemented on the National Center for High-performance Computing (NCHC) Forerunner 1 supercomputer, where dynamic task allocation is used to distribute building-level analyses across multiple central processing unit (CPU) cores. A case study is conducted for 62,454 RC buildings constructed before 1999 in Hualien County. Representative low-, mid-, and high-rise frame models are developed, and incremental dynamic analyses are performed using multiple ground-motion records and peak ground acceleration levels. The resulting fragility curves are further combined with ground-motion data from the 2024 Hualien 0403 earthquake to estimate building damage and compare the predictions with observed damage data. The results demonstrate the feasibility of using HPC for high-resolution city-scale seismic risk assessment. |
|
|---|