An integrated probabilistic approach for optimum maintenance of fatigue-critical structural components

Abstract

Inspection and maintenance are important means to validate or recover reliability of metallic structural systems, which usually degrade over time due to fatigue, corrosion and other mechanisms. These inspection and maintenance actions generally account for a large part of lifetime costs, which necessitate an efficient maintenance strategy to satisfy the requirements on reliability and costs. Most often, an optimum maintenance/repair crack size criterion is derived by probabilistic cost-benefit analysis, e.g. by minimization of expected lifetime costs, which are assessed based on cost models. This paper proposes an integrated approach to derive an optimum range for maintenance criterion using both reliability-based and cost-based optimization. It is found that an optimum repair (crack size) criterion exists which leads to the maximum lifetime fatigue reliability. A smaller repair criterion than the reliability-optimum do not lead to a higher lifetime fatigue reliability but leads to higher lifetime costs. Hence, a limit for repair criterion is defined by the reliability-optimum criterion, which can be obtained without cost models. The reliability-optimum criterion is found to be smaller than the costoptimum criterion and thus an optimum range between the reliability-optimum and cost-optimum criterion is established.

A fatigue-reliability approach using ultrasonic non-destructive inspection

“…In this paper, the Bayesian updating method has been used for updating failure probability in cases of detection or nondetection of a crack under uncertainties for a coupon test. Equations ( 15) and ( 16) define the updated failure probability P f ,up in each case: [20][21][22] • Indication case:…”

“…Moreover, Zareei and Iranmanesh [18] have used the Bayesian updating concept along with Markov Chain Monte Carlo (MCMC) method and Metropolis-Hasting algorithm to update material parameters and enhance fatigue life prediction. These outcomes will be helpful in order to determine optimal inspection schedule and maintenance operations, that's what several researchers are working on these years [19][20][21][22].…”