A multi-algorithm approach for optimizing collapse margin ratio in seismic design of reinforced concrete structures

This study presents a comprehensive multi-algorithm framework for optimizing the Collapse Margin Ratio (CMR) of reinforced concrete (RC) structures subjected to seismic loading, in accordance with the FEMA P695 methodology. A hybrid approach combining Artificial Neural Networks (ANNs) with Genetic Algorithms (GAs) is employed, alongside standalone optimization techniques including Particle Swarm Optimization (PSO), Simulated Annealing (SA), and Bayesian Optimization (BO), to improve key seismic performance parameters. A dataset of 114 RC archetype structures is analyzed through more than 5,000 Incremental Dynamic Analyses (IDA) using far-field ground motion records. Surrogate models and metaheuristic algorithms are used to efficiently identify optimal values for input parameters such as fundamental period, yield and ultimate displacements, overstrength factor, and spectral acceleration. The results demonstrate that ANNs and PSO deliver the most robust performance, achieving a maximum CMR of 5.99. Sensitivity analysis further underscores the dominant influence of the fundamental period and overstrength factor. The study also incorporates uncertainty quantification and outlier detection to enhance the reliability of the optimization process. This data-driven methodology not only improves seismic resilience and cost-efficiency in structural design but also advances the integration of computational intelligence into performance-based earthquake engineering.