S-BORM: Reliability-based optimization of general systems using buffered optimization and reliability method

TL;DR

This paper introduces S-BORM, an efficient algorithm leveraging buffered optimization and reliability methods to solve complex reliability-based optimization problems for general systems with multiple cut-sets, significantly reducing computation time.

Contribution

It develops a novel S-BORM algorithm that efficiently handles general system RBO problems using buffered failure probability and a proximal bundle method, improving computational speed and robustness.

Findings

Solves systems with up to 108 cut-sets within a minute.

Demonstrates high accuracy and robustness through numerical examples.

Reduces computational complexity of general system RBO problems.

Abstract

Reliability-based optimization (RBO) is crucial for identifying optimal risk-informed decisions for designing and operating engineering systems. However, its computation remains challenging as it requires a concurrent task of optimization and reliability analysis. Moreover, computation becomes even more complicated when considering performance of a general system, whose failure event is represented as a link-set of cut-sets. This is because even when component events have smooth and convex limit-state functions, the system limit-state function has neither property, except in trivial cases. To address the challenge, this study develops an efficient algorithm to solve RBO problems of general system events. We employ the buffered optimization and reliability method (BORM), which utilizes, instead of the conventional failure probability definition, the buffered failure probability. The proposed algorithm solves a sequence of difference-of-convex RBO models iteratively by employing a proximal bundle method. For demonstration, we design three numerical examples with increasing complexity that includes up to 108 cut-sets, which are solved by the proposed algorithm within a minute with high accuracy. We also demonstrate its robustness by performing extensive parametric studies.