Comparing Conventional and Buffered Failure Probability from the Perspective of Risk-based Decision-Making

Abstract

Failure probability has been widely used as a reliability metric to provide a building block of reliability-based decision-making. Such use of failure probability has implications for both risk management and computation. In this study, we revisit the familiar concept of a reliability metric by comparing the conventional failure probability with a recently proposed alternative metric, the buffered failure probability. Comparisons are made from the perspectives of risk management, failure probability evaluation, and reliability-based optimization. In addition to analytical examinations, we present numerical investigations revealing several intriguing distinctions between the failure probabilities. We also contemplate the prospects of using the two metrics in parallel.