Seismic Performance Enhancement of RC Buildings Using Shape Memory Alloy Reinforcement Integrated with Bayesian Digital Twin Structural Health Monitoring: Experimental Validation and Real-Time Residual Capacity Assessment

Abstract

Post-earthquake functionality and rapid re-occupancy of reinforced concrete (RC) buildings depend critically on two converging innovations: self-centering structural systems that suppress residual drift, and real-time structural health monitoring (SHM) frameworks that provide quantitative post-event condition assessment. This paper presents an integrated experimental and computational investigation combining superelastic NiTi Shape Memory Alloy (SMA) bar reinforcement for seismic self-centering with a Bayesian digital twin framework for real-time structural condition monitoring and residual capacity prediction. Eight RC columns (150 × 150 × 600 mm, fck = 30 MPa) four with conventional HYSD steel reinforcement and four with superelastic NiTi SMA bars (6 mm diameter, σfwd = 420 MPa, εrecov = 6%) were tested under combined axial load (0.1f'cAg) and reversed cyclic lateral loading following the FEMA 461 incremental protocol to 5% lateral drift. A digital twin prototype was simultaneously deployed, integrating FBG fiber-optic strain sensors, MEMS accelerometers, and corrosion potential probes, with a particle filter Bayesian model updating engine updating the calibrated OpenSees nonlinear model in real time at each loading step. SMA-reinforced columns demonstrate 5.5-fold reduction in residual drift (0.42% vs 2.30% at 5% demand), stable flag-shaped hysteretic loops to 4% drift, and less than 8% stiffness degradation over 20 loading cycles confirming superelastic self-centering mechanisms. The digital twin Bayesian updating reduces flexural stiffness uncertainty from ±18% (prior) to ±4.2% (posterior) after 45 loading cycles, and successfully detects the onset of SMA bar slip at 72% of peak load providing a 28% early-warning margin before structural failure. Post-earthquake residual capacity is predicted within ±8% of experimentally determined values, enabling quantitative, sensor-informed post-event occupancy decisions. A probabilistic post-earthquake safety assessment framework is proposed integrating SMA self-centering performance with digital twin remaining capacity estimation.