Advanced Computational Algorithms for Designing and Dynamically Maintaining Industrial Systems: An NSGA-II Algorithm and Discrete Event Simulation Approach

Profit maximisation is an important objective for industries in a competitive world, and this can be achieved by improving the reliability, availability and life cycle cost (LCC) performance of repairable industrial systems. Engineers have used many techniques to improve the availability of these systems, such as adding redundant devices, using components that perform better in terms of reliability and maintainability, or programming dynamic maintenance strategies. However, the idea of using these techniques simultaneously has not received sufficient attention. The authors of this paper have recently studied the simultaneous optimisation of system design and maintenance strategy in order to achieve both maximum reliability and minimum life cycle cost: the Nondominated Genetic Sorting Algorithm II (NSGA-II) has been coupled with event simulation to obtain a set of nondominated solutions. In this work, this study is extended to the optimisation problem with three objectives, namely reliability, unavailability and life cycle cost. This combinatorial approach is successfully demonstrated through an industrial case study providing non-dominated solutions that balance reliability, system unavailability and life cycle costs, thus contributing to informed decision making in the choice of system configuration and specifications and its maintenance policy in competitive environments.

Keywords: Multi-objective evolutionary algorithms, Discrete Event Simulation, Reliability, Unavailability, LCC, Design, Maintenance.