An Optimal Maintenance Spare Parts Prediction Model and Its Complex Applications

This paper introduces an optimal stochastic maintenance spare parts prediction model under general conditions including intermittent uses. This model overcomes the drawback in the existing spares prediction models by using queueing theory. Specifically, this model, based on the M/M/c queue, establishes that if spares demand cannot be fulfilled it will be queued (waiting), unlike other existing models which assume that it is lost. Obviously, this M/M/c queueing model is realistic since any spares demand needs to be fulfilled eventually in practice. This M/M/c queueing model assumes that the spare parts demand follows a Poisson process and that repair turnaround times are modeled by an exponential distribution. Model parameters are estimated by using big data analytic techniques. Spare parts prediction with complex hierarchical spares allocation systems and scrap rates are also discussed in this paper. Complex spares availability is defined and derived. When the spare parts necessary to perform the maintenance action are not available, the incurred shortage or unavailability costs could be substantial. This new model and tool can be used to provide an optimal number of spares to ensure high system operational availability and to enhance customer satisfaction at the lowest costs. The relationship between spare parts availability and system operational availability are also established to support optimal spare parts predictions. An application example is presented. This M/M/c queue spares prediction model has been applied to Raytheon Technologies products and built into Raytheon logistics support system EAGLE. Software tools have been developed to facilitate computing and implement digital engineering. Numerical validation for various test cases and comparisons to existing spare prediction models have been performed. Future research is discussed.