Risk-informed sensitivity analysis and optimization of seismic mitigation strategy using Gaussian process surrogate model

Abstract

This work aims to develop a method to derive the optimal parameters of seismic mitigation devices for achieving the minimum system seismic risk, which is derived by accounting for the correlation between each individual components risk. Gaussian process surrogate model (GPSM) is used to illustrate the functional relationship between seismic risk and the variables of the mitigation devices, and then both sensitivity analysis and parameter optimization are carried out. Specifically, global sensitivity analysis (GSA), which is powerful for assessing the relative importance of parameters of the isolation devices, is achieved through the use of GPSM; the parameter optimization can be efficiently conducted to find the minimum seismic risk. A cable-stayed bridge model is established to demonstrate the application of the proposed GPSM-based method of sensitivity analysis and optimization of seismic isolation devices. The fluid viscous damper (FVD) which provides additional damper and stiffness is selected to illustrate the procedure of the multi-object optimization of seismic isolation devices on cable-stayed bridges. The seismic risk is calculated by integrating probabilistic seismic hazard analysis (PSHA), probabilistic seismic demand analysis (PADA), and probabilistic damage states, incorporating important sources of nonlinearity and uncertainties. Component-level objectives and one system-level object are considered, and the optimal parameters (damping coefficient Cd and exponent α) are determined. The results demonstrate that the GPSM-based is effective and efficient for sensitivity analysis and optimization of seismic isolation devices.