A multi-objective decision making on reliability growth planning for in-service systems

Abstract

New designs prioritized for the fast time-to-market usually can not carry out sufficient in-house reliability growth testing due to the stringent delivery deadline. Reliability improvement for those systems can be achieved by implementing corrective actions (CAs) on in-service systems. In this paper, three types of effectiveness functions are proposed to measure the reduction rate of failure modes given different CA resources. Integrated with the effectiveness function, a new failure intensity model is proposed for predicting the mean-time-between-failures (MTBF) of field systems. Finally, a multi-objective optimization model is formulated to maximize the system reliability and to minimize the reliability uncertainty with the constraint of the CA resources. Genetic Algorithms combined with greedy heuristic are applied to search the optimal CA decisions that lead to the maximum reliability growth while minimizing the reliability uncertainty. Results show that the proposed reliability growth program can effectively guide decision-makers to find the most effective corrective actions for achieving the reliability goal for a large fleet of in-service systems. Throughout paper, systems and products will be used interchangeably.

A Multiphase Decision Model for System Reliability Growth With Latent Failures

“…This paper aims to extend the preliminary findings [22] to a multiphase reliability growth environment. The study is motivated to address the stochastic behavior of latent failures and further mitigate their impact on product reliability.…”

“…Ellner and Hall [21] proposed an RGP model to estimate system reliability growth taking into account latent failures. Jin and Wang [22] formulated a multicriteria programming model to maximize field system reliability through optimal CA decisions. These studies show that GRP is an effective approach to drive the product reliability if extended in-house testing is infeasible in early design and prototyping phase.…”

“…Two types of CA are generally applied to repairable systems: retrofit and engineering change order (ECO) [21], [22]. Retrofit uses spare modules (i.e., line replaceable units) to proactively replace inservice modules that will fail due to a known failure mode.…”

“…In the last decade, reliability growth planning (RGP) has emerged as a new methodology to resolve these challenging issues [20]- [22]. RGP differs from RGT in that it drives the reliability across design, manufacturing, and field operation.…”

Research on the testing duration of reliability growth planning

“…Assuming that the reliability growth rates for all components conform to the AMSAA model, the Coit model was presented which allocated the reliability growth testing time of components in series as a whole [20]. Jin and Wang provided three validity functions to express corrective actions and applied them to failure intensity models for projecting the mean time between failures (MTBF) of systems, a two-objective reliability growth model was provided for maximizing the reliability of the system and minimizing the uncertainty of reliability under limited corrective actions [21]. Calomeris and Herbert [22] pointed out that the experiment duration, primary MTBF, and goal MTBF all have a big influence on the growth pattern, but no further research has been found.…”