Stochastic seismic analysis and reliability assessment of singly curved RC shell structures

Singly curved reinforced concrete (RC) shell structures are increasingly used for large column-free spans and post-earthquake emergency shelters, yet their seismic assessment has largely relied on linear dynamic analysis and deterministic approaches that neglect material nonlinearity and uncertainty. This study presents an automated design-to-reliability framework, implemented through a Python-MATLAB-ABAQUS workflow, in which the Developed Advanced Sandwich Model (DASM) generates the element-wise reinforcement separately for each realization, the nonlinear response is computed using a multi-axial plasticity-damage model implemented through a VUMAT subroutine, and the probability density evolution method (PDEM) is coupled with the equivalent extreme-value event (EEVE) method to quantify reliability. Uncertainty is thereby propagated through both the reinforcement-design and the seismic-response stages. The framework is applied to a singly curved RC shell subjected to the Northridge-01 record, with uncertainty in the concrete and steel properties, the gravity loads and the excitation amplitude. For the adopted displacement-based limit state, the framework yields a reliability of 65.36%. The deterministic analysis predicts a maximum displacement below the adopted limit, whereas the 95th percentile of the stochastic response exceeds it, showing that deterministic assessment can underestimate the range of possible seismic responses. The framework thus provides a basis for performance-based stochastic assessment of singly curved RC shell structures.