Adaptive Experimental Design of Real-Time Hybrid Simulation for Parameter Calibration of Rate-Dependent Devices toward Optimal Response Prediction

Abstract

Real-time hybrid simulation (RTHS) provides a cyber-physical testing method for system evaluation of global or local system evaluation of civil subjects to dynamic loading. Physical testing of experimental substructures is integrated with numerical simulation of analytical substructures, thus facilitating large-scale testing in size-limited laboratories. Current practice of RTHS often focuses on structural response evaluation under selected ground motions. This study presents an innovative application of RTHS for parameter calibration of rate-dependent devices for optimal structural response prediction. An adaptive experimental design approach is integrated with RTHS to account for realistic seismic demands on the rate-dependent device due to structural uncertainties. A kriging metamodel is utilized to surrogate the response discrepancy between RTHS and computer simulation with experimental substructure numerically modeled. Efficient global optimization (EGO) is then applied to sequentially determine new sample points for parameter calibration through laboratory testing to minimize the difference between RTHS and model-predicted results. The proposed approach is then applied for model parameter calibration of a self-centering viscous damper (SC-VD) device through RTHS. The parameters of the SC-VD device are calibrated from sequentially designed RTHS to optimize the error in maximum structural response prediction. The proposed adaptive approach is further compared with traditional experimental design methods and demonstrates better performance in maximum response prediction especially when resources are only available for a limited number of experiments.