Multi-Objective Reliability-Redundancy Allocation for Non-Exponential Multi-State Repairable Components

Abstract

In recent years, little work has been done over the joint reliability-redundancy allocation and in all of that, exponential failure and repairs are assumed. In this paper we focus on investigating the reliability-redundancy allocation where the repair failure distributes are non- exponential. When failure/repair times follow a distribution other than exponential, analytical methods become too complicated or even useless. In these cases simulation methods like Mote Carlo and discrete event simulations are typically preferred, but they usually take relatively long time to estimate the availability of the system. This weakness gets more important when we want to use them in optimization algorithms. Here we offer a new discrete event computer simulation package which can estimate availability values in a reasonable time. Then we combine the mentioned simulation method with genetic algorithm to obtain optimal designs. These methods are illustrated with their application using some numerical examples.