Multi-objective optimization of maintenance programs in nuclear power plants using Genetic Algorithm and Sensitivity Index decision making

Maintenance planning is a critical issue for all heavy industrial sectors such as aeronautics, automobile factories and power plants. The study herein is aimed towards improvement of safety systems reliability in a nuclear power plant. Optimization of maintenance and surveillance test activities is one of the best strategies to improve the reliability of the related safety systems in this kind of plants. Maintenance programs can be formulated in terms of a multi-objective optimization where unavailability, cost and Exposure Time (ET) act as decision criteria and surveillance tests, Allowed Outage Time (AOT) and Preventive Maintenance (PM) intervals act as decision variables. In this paper, Genetic Algorithm (GA) is used to find the best series of answers in the form of Pareto front curve. Sensitivity Index (SI) is applied as a decision making tool to extract the most optimized and promising solution. The unavailability, cost and ET functions, for the most optimal solutions, were reduced by 86%, 58% and 30%, respectively.