An Adaptive Simulation Framework for the Exploration of Extreme and Unexpected Events in Dynamic Engineered Systems - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Risk Analysis Année : 2017

An Adaptive Simulation Framework for the Exploration of Extreme and Unexpected Events in Dynamic Engineered Systems

Résumé

The end states reached by an engineered system during an accident scenario depend not only on the sequences of the events composing the scenario, but also on their timing and magnitudes. Including these additional features within an overarching framework can render the analysis infeasible in practical cases, due to the high dimension of the system state-space and the computational effort correspondingly needed to explore the possible system evolutions in search of the interesting (and very rare) ones of failure. To tackle this hurdle, in this article we introduce a framework for efficiently probing the space of event sequences of a dynamic system by means of a guided Monte Carlo simulation. Such framework is semi-automatic and allows embedding the analyst's prior knowledge about the system and his/her objectives of analysis. Specifically, the framework allows adaptively and intelligently allocating the simulation efforts preferably on those sequences leading to outcomes of interest for the objectives of the analysis, e.g., typically those that are more safety-critical (and/or rare). The emerging diversification in the filling of the state-space by the preference-guided exploration allows also the retrieval of critical system features, which can be useful to analysts and designers for taking appropriate means of prevention and mitigation of dangerous and/or unexpected consequences. A dynamic system for gas transmission is considered as a case study to demonstrate the application of the method.

Domaines

Risques
Fichier principal
Vignette du fichier
RISK_AdaptiveExploration.pdf (867.65 Ko) Télécharger le fichier
Origine : Accord explicite pour ce dépôt
Loading...

Dates et versions

hal-01652286 , version 1 (30-11-2017)

Identifiants

Citer

Pietro Turati, Nicola Pedroni, Enrico Zio. An Adaptive Simulation Framework for the Exploration of Extreme and Unexpected Events in Dynamic Engineered Systems. Risk Analysis, 2017, 37 (1), pp.147 - 159. ⟨10.1111/risa.12593⟩. ⟨hal-01652286⟩
324 Consultations
125 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More