Structural Fragility of Piping Systems using Equivalent Elastic Time-history Simulations under Bayesian Framework

摘要

The main objective of this study is to effectively evaluate seismic fragility of nonlinear inelastic connections in a piping system using linear elastic time-history analyses and Bayesian inference. Typical piping systems are characterized to have (a) essentially linear elastic stress-strain behavior; (b) damage due to excessive inelastic deformation at specialized nonlinear inelastic locations such as T-joint connections, elbows, valves, etc.; (c) low ductility (<3.0) corresponding to failure. Since the system is largely linear elastic, we utilize the equivalent linearization method (ELM) to describe the localized inelastic nonlinearities. The reason to use this is that one of the critical limitations in the seismic fragility analyses of structures under probabilistic simulation framework is the enormous computational costs associated with real structures described by intensive nonlinear inelastic Finite Element (FE) models. The ELM in seismic fragility analyses must seek to minimize the error between the maximum responses which indicate the damage. For this, we introduce a concept called equivalent elastic limit state (ELS) into seismic fragility analyses. By studying a large number of representative structural systems which are essentially linear elastic but characterized by localized fragile inelastic nonlinearities, we propose a model to compute seismic fragility curves using only linear elastic time-history simulations. It is revealed that the seismic fragility using the model and linear elastic time-history simulation is quite close to the actual fragility, and the fragility using this approach can be successfully incorporated with Bayesian inference to obtain more accurate result. The efficacy of the approach is finally confirmed in an example of a full-scale piping system with fragile T-joint connections.