Stochastic Forecasting of Lumpy-distributed Aircraft Spare Parts Demand

An approach is proposed to forecast lumpy spare parts demand associated with non-routine aircraft maintenance based on a stochastic model. The model assumes demand arrival in the form of a Homogeneous Poisson Process (HPP). The developed model is applied to a dataset from an aircraft maintenance operator consisting of lumpy spare parts demand for nine Line Replacable Units (LRUs) from a shared customer pool of parts, which are associated with significant repair turnaround time delays. To meet the operator's service level requirements and associated stocking strategy, the model is evaluated for its capacity to forecast peak demand, rather than its capacity to minimize forecasting error. Under these terms, the proposed model outperforms the existing company approach.