Data-driven multi-criteria framework for the seismic and post-earthquake performance assessment of corroded RC bridge piers

This study proposes an artificial intelligence-enhanced framework to assess the seismic and post-earthquake performance of corroded reinforced concrete (RC) bridge piers, addressing the limitations of conventional design methods that prioritize maximum drift while neglecting residual drift and corrosion effects. By integrating machine learning (ML) with numerical simulations, the framework introduces a performance matrix, a novel classification system that replaces traditional binary performance levels. This framework evaluates structures by dual criteria: maximum drift (safety during shaking) and residual drift (post-earthquake recoverability), categorizing structural response into seven performance classes. The ML-driven model predicts maximum and residual drifts of corroded RC piers, trained with two separate databases, each with 2000 input-output pairs, which are created by randomized nonlinear dynamic analyses on numerical models of corroded RC piers. The numerical modeling technique is capable of capturing corrosion-induced shear-associated behavior. The results showed the critical importance of the performance matrix in modern bridge engineering, as they unify seismic safety and resilience, enabling resilient infrastructure management in corrosion-prone seismically active regions.