Real-Time Hybrid Simulation with Online Model Updating: Methodology and Implementation

Abstract

Hybrid simulations have shown great potential for economic and reliable assessment of structural seismic performance by combining physical experimentation on part of the structural system and numerical simulation of the remaining structural components. Current hybrid simulation practices often use a fixed numerical model without considering the possible availability of a more-accurate model obtained during hybrid simulation through an online model updating technique. To address this limitation and improve the reliability of numerical models in hybrid simulations, this paper presents a method and an implementation procedure of conducting real-time hybrid simulation (RTHS) with online model updating. The Unscented Kalman Filter (UKF) was adopted as the parameter identification algorithm applied to the Bouc-Wen model that defines the hysteresis of the experimental substructure. The identified parameters are then used to update the models of the numerical substructures during RTHS. A parametric study of the UKF system model parameters is carried out first to determine the optimum values to be used in the verification experiments. Then RTHS of a three-story steel shear building model is conducted and the effectiveness of online model updating in RTHS and the proposed implementation procedure is demonstrated. Guidelines for implementing the UKF for online model updating in RTHS and research needs for further development are discussed.