A multi-stage stochastic programming for joint condition-based replacement and spare parts ordering towards complex manufacturing system with uncertain lead times

Rapid development of manufacturing systems towards growing structural complexity has posed challenges to the corresponding maintenance planning for economic improvement. Rational planning of component replacement and spare parts ordering for such a multi-component multi-machine system is intractable due to the necessity of considering multiple dependencies and various spare parts attributes. Moreover, the supply unpredictability results in the prevalence that the lead times of spare parts orders become more uncertain. Existing methods cannot adequately tackle this issue since they basically ignored the lead time or assumed it as constant. This work thus develops a joint optimization framework of condition-based component replacement and spare parts ordering for a multi-component multi-machine system considering multiple dependencies and stochastic lead times (JOCRSP-M&S) for the total cost reduction. We formulate this problem as a multi-stage stochastic programming model using sample average approximation method. In this model, the structural and economic dependencies are reflected by the impact on replacement decision of disassembly sequence among components and multi-unit system configuration respectively. Various attributes of critical and consumable spare parts are incorporated into the spare parts management, and the uncertainty of lead times is introduced by a scenario tree construction procedure. Facing the model solving challenge, we reformulate the built model with linearization and simplification. A rolling horizon solution strategy is applied to facilitate the efficient decision-making output. Numerical experiments validate the stability and effectiveness of the proposed optimization model. Compared to deterministic programming and traditional policy, JOCRSP-M&S shows the advantages in terms of total cost reduction.