Data-Driven Models for Predicting Drift Ratio Limits of Segmental Post-tensioned Precast Concrete Bridge Piers

Segmental post-tensioned precast concrete (SPPC), an important technique of Accelerated Bridge Construction (ABC), has been proven its advantages over monolithic cast-in-place concrete in rapid bridge rehabilitation and construction. To deploy its implementation in bridge substructures for seismic applications, proposing performance-based design specifications is much needed. However, one main challenge of achieving this goal stems from defining quantitative criteria for a series of damage states that are associated with different performance levels, in relation to the functionality of the bridge. In recent years, data-driven models have been recognized as a powerful tool for making rational predictions in several structural engineering applications. Multiple linear regression (MLR) offers great potential to develop equations that are capable of identifying the maximum drift ratio limits for the four performance levels (i.e., immediate service, limited service, service disruption, and life safety) for SPPC piers. In this respect, based on a database generated from validated finite element models, MLR with stepwise backward elimination is performed in the current study using key design parameters, including concrete strength, aspect ratio, gravity load ratio, post-tension force, post-tension strand ratio, and energy dissipation bar ratio. For each damage state, a predictive equation for the threshold drift ratio is developed by seeking a balance between accuracy and simplicity. Sensitivity analysis is also performed to evaluate the effect of design parameters on the variability of the predicted drift ratio limits.