Session: SE-04-01 Machine Learning for Seismic Analysis of Industrial Facilities

Paper Number: 123237

123237 - State-Dependent Seismic Fragility Functions for Bolted-Flange Joints on Special-Risk Industrial Substructures 

Seismic risk assessment of industrial plants and process equipment has gained attention lately. Past earthquakes demonstrated the high vulnerability of industrial facilities to experience serious damage to process components. Consequently, multiple and simultaneous releases of hazardous substances represent a major threat to the environment and society. Nevertheless, the design of industrial plants is inadequately described in recent codes and guidelines, as they do not consider the dynamic interaction between the structure and the equipment. Thus, the effects of the seismic response of the process components on the response of the structure and vice versa are neglected. This is strictly related to the complexity of the problem and its modelling, combined with a general scarcity of available data. On these premises, a new methodology to assess the vulnerability of such complex systems is proposed. Specifically, it combines data from physics-informed finite element FE models with the latest advances in surrogate model techniques to numerically derive state-dependent fragility curves. Those are functions capable of modelling the probability of exceedance of damage thresholds conditioned not only by seismic hazard but also by the initial damage state conditions. Hence, a real case-study application is presented. As a result, empirical state-dependent fragilities are evaluated for one of the most critical components of the experimental campaign on the SPIF industrial MRF frame, i.e., the monitored bolted flange joints BFJs.

Presenting Author: Chiara Nardin University of Trento

Presenting Author Biography: Structural engineer and postdoctoral fellow at the Hazard Mitigation, Structural Dynamics and Control (HMSDC)
group of the University of Trento. She received her Ph.D. in Modelling and Simulation in 2022 at the University of
Trento with a thesis on “Seismic experimental analyses and surrogate models of multi-component systems in special-risk industrial facilities”. Her research interests rely on risk analysis and fragility assessment of civil and process plant infrastructures, particularly concerning industrial facilities under severe seismic events. Passionate about design and experimental testing of complex systems, she was actively involved in the EU SERA Project SPIF on shaking table tests of a real multi-component industrial facility. During her PhD studies, she became fascinated with the world of UQ, metamodeling and advanced machine learning algorithms throughout her research stay at the Chair of Reliability, Safety and Uncertainties Quantification (RSUQ) of ETH, Zurich. As a result, she is currently collaborating with the RSUQ and HMSDC groups on developing analytical and numerical methods to evaluate seismic state-dependent fragility functions for special-risk industrial systems.

Authors:

Chiara Nardin University of Trento
Oreste Salvatore Bursi University of Trento
Marco Broccardo University of Trento
Stefano Marelli ETH Zürich