Structural Reliability Methods Applied in Analysis of Steel Elements Subjected to Fire
- 1 December 2021
- journal article
- research article
- Published by American Society of Civil Engineers (ASCE) in Journal of Engineering Mechanics
- Vol. 147 (12) , 04021108
Abstract
The accuracy and efficiency of an approach using surrogate models are investigated, in comparison with traditional structural reliability methods, in the analysis of two steel structures subjected to natural fire. For this purpose, an algorithm is used, coupled to a finite-element computational package, which includes the first-order reliability method (FORM), the approximate method most used in this type of problem, and Monte Carlo simulation (MCS), which is usually taken as a reference. On the other hand, the literature indicates surrogate models as a promising alternative to maintain, at acceptable levels, the computational cost in similar analyses. Therefore, two approaches using surrogate models are also applied to the algorithm, more specifically adaptive and nonadaptive artificial neural networks. The analyses led to differences of up to 21.83% between the failure probabilities obtained via FORM and via MCS. Furthermore, while the nonadaptive approach fails to achieve sufficient accuracy, the adaptive approach is confirmed to be a viable alternative, with differences, on average, close to 1.00% and computational times less than 1.60% of the time required by MCS.Keywords
This publication has 44 references indexed in Scilit:
- Global structural optimization considering expected consequences of failure and using ANN surrogatesComputers & Structures, 2013
- Metamodel-based importance sampling for structural reliability analysisProbabilistic Engineering Mechanics, 2013
- Stochastic Analysis of Structures in Fire by Monte Carlo SimulationJournal of Structural Fire Engineering, 2013
- Probabilistic Evaluation of Structural Fire ResistanceFire Technology, 2012
- Probabilistische Bewertung von Stahlbetonbauteilen unter NaturbrandbeanspruchungBeton- und Stahlbetonbau, 2008
- Load Combination Requirements for Fire-resistant Structural DesignJournal of Fire Protection Engineering, 2005
- Structural reliability analysis for implicit performance functions using artificial neural networkStructural Safety, 2005
- Training feedforward networks with the Marquardt algorithmIEEE Transactions on Neural Networks, 1994
- Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networksNeural Networks, 1990
- Adaptive sampling — an iterative fast Monte Carlo procedureStructural Safety, 1988