Maintenance optimization of series systems subject to reliability constraints

The extension of maintenance optimization methodologies used for single component to multiple component systems must take into account the interdependencies that may exist between the components. Such dependencies could arise when the maintenance optimization of the system over the time is subject to constraints. In this paper, a methodology using Lagrangian relaxation techniques embedded in dynamic programming is proposed for minimizing the maintenance costs of reliability constrained series systems. The methodology could be applied to deterministic and probabilistic dynamic programming problems, as well as to partially observable Markov Decision process. The computational complexity of the proposed approach is polynomial in the number Q of the system components. Theoretical and practical issues related to the existence, and the computation of the Lagrange multipliers are considered. The proposed methodology is illustrated by a numerical application considering maintenance planning of a pipeline.