Simulated ground motions for seismic risk assessment of structures

• Main Author: Tsioulou, Alexandra
• Published: University College London (University of London) 2018
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.763328

The recent advances in computational efficiency and the scarcity/absence of recorded ground motions for specific seismicity scenarios have led to an increasing interest in the use of ground motion simulations for seismic hazard analysis, structural demand assessment through response-history analysis, and ultimately seismic risk assessment. Two categories of ground motion simulations, physics-based and stochastic site-based are considered in this study. Physics-based ground motion simulations are generated using algorithms that solve the fault rupture and wave propagation problems and can be used for simulating past and future scenarios. Before being used with confidence, they need to be validated against records from past earthquakes. The first part of the study focuses on the development of rating/testing methodologies based on statistical and information theory measures for the validation of ground motion simulations obtained through an online platform for past earthquake events. The testing methodology is applied in a case-study utilising spectral-shape and duration-related intensity measures (IMs) as proxies for the nonlinear peak and cyclic structural response. Stochastic site-based ground motion simulations model the time-history at a site by fitting a statistical process to ground motion records with known earthquake and site characteristics. To be used in practice, it is important that the output IMs from the developed time-histories are consistent with these prescribed at the site of interest, something that is not necessarily guaranteed by the current models. The second part of the study presents a computationally efficient framework that addresses the modification of stochastic ground motion models for given seismicity scenarios with a dual goal of matching target IMs for specific structures, while preserving desired trends in the physical characteristics of the resultant time-histories. The modification framework is extended to achieve a match to the full probability model of the target IMs. Finally, the proposed modification is validated by comparison to seismic demand of hazard-compatible recorded ground motions. This study shows that ground motion simulation is a promising tool that can be used for many engineering applications.