Parametric identification of Timoshenko-beam model for shear-wall structures using monitoring data

Abstract

The feasible modeling of the shear-flexure coupling behavior of a shear wall is desired and needed for long-term structural health monitoring of shear-wall structures. To address this issue, a Timoshenko-beam finite-element model is proposed for referencing a shear-wall structure. A parametric identification framework along with a particle swarm optimization is developed for updating the target model using the vibration data. Numerical examples are analyzed with particular focuses on the selection of objective functions, and the potential effects of ground motions, sensor allocation, model selection, and measurement noise. Results indicate the preference of adopting the acceleration-based errors as the objective function, the robustness of identified optimal parameters regarding ground motion and sensor allocation, and the importance of choosing an appropriate model for referencing the structure. The accuracy of the proposed model is further validated using the data obtained from a Walnut Creek building under eight earthquakes from years 2015 to 2021. The averages of normalized root-mean-square acceleration prediction errors of the proposed model, along the longer and the shorter sides of the building, are 0.480 and 0.767, respectively. Comparisons with the shear-beam model, the Euler-Bernoulli-beam model, and the parallel connection of these two models further show that the Timoshenko-beam model outperforms other candidate models in predicting the response of the examined building.