Uncertainty quantification of large-scale Magneto-Rheological damper models in seismic hazard mitigation

Failure probabilities estimations are often important and challenging in civil application problems. High-Dimensional and multivariable problems often suffer from accurate sampling. Thus, a robust method Metropolis-Hasting (MH) algorithm which is a sampling from high- dimensional and multivariable distribution and solving many modern-day statistical and computational problems is significantly important to complex model. The MH algorithm could also be used for uncertainty estimation for the phenomenological model application. With estimation on the uncertainty of system, sensitivity analysis and probabilistic numerical simulation could be used to evaluate the performance of the numerical model. Many different deterministic numerical models were simulated for Magneto-Rheological (MR) damper, a particularly promising type of semi-active control device. However, the studies of probabilistic numerical models were insufficient due to scarce data and experiments for MR damper. Therefore, this study focuses on uncertainty estimation for various numerical models. Sensitivity analysis, variable correlation estimation and numerical simulation for 3DOF were performed based from the estimated uncertainties; an innovative procedure and recommendation for probabilistic analysis system were proposed, which increase efficiency and accuracy for under limited resources.