Contribution to efficient structural safety assessments: A comparative analysis of computational schemes

Abstract

The design of structural elements requires the use of advanced analysis techniques that are capable to closely model reality throughout their expected lifetime. However, the use of stochastic principles might be necessary to address the impact of uncertainties in structural engineering. Probability-based safety methods, as Crude Monte-Carlo simulations, can evaluate structural safety with high precision in exchange of numerous simulations. However, for complex problems, Crude Monte-Carlo simulations can demand significant computational power and be very time-consuming. Sampling (or variance reducing) techniques – as Importance Sampling and Subset Sampling – can minimise such drawbacks and improve computational efficiency. This paper investigates the potential offered by a newly developed R-package built in the programming language R, to evaluate the computational performance of these sampling techniques. To this, these techniques are compared in terms of accuracy and computational efficiency. This investigation shows that both techniques are capable to generate a wide range of solutions with high precision and satisfactory computational efficiency. Finally, this study explores the potential of using the newly developed R-package to select probability-based safety methods based on the features of reliability problems. The R-package offers a robust platform to conduct and compare different safety assessments of complex structural problems.