Robust Maintenance Scheduling for Reliability Maximization

Abstract

The present paper aims to single out the maintenance actions to perform on a system constrained to be maintained only during some planned stops. Since system failure implies costs and risks for people and/or the environment, then it is necessary to minimize the probability of its occurrence. Thus, maintenance actions need to maximize the system reliability up to the next planned stop, in respect to some constraint. The originality of the problem discussed in the present paper lies in considering reliability values affected with uncertainty within a range of vagueness. Consequently, a further problem is added to the constrained reliability maximization, that is the search for a robust solution: the selected solution also needs to guarantee a low sensitivity to the real position of the component reliability within the variability range. This implies the formulation of some parameter able to express the solution robustness at best. To solve the problem, the authors have developed an exact dynamic programming algorithm that is also suitable for complex series-parallel systems, usually intractable by the mathematical programming. The algorithm requires short computational times and thus quickly allows to individuate the optimal solutions for different sets of components reliability values, so constituting a valid decision support tool. Finally a numerical example applied to a complex system composed by a large number of components is reported.