Poisson Based Sparing Analysis with Forecastability and Performance Evaluation

SUMMARY & CONCLUSIONSInventory management of spare parts is a critical issue in the aerospace and defense industry. Better forecasting ensures improved maintenance, helps meet operational availability requirements, minimizes downtime costs, and reduces the risks associated with acquiring and holding costly inventory.A traditional approach to spare parts inventory, Poisson-based spares analysis, uses average demand rates and desired part availability over a target lead time. This method, however, does not capture the effects of failure distribution models, various support system parameters (like repair and lead times), the influence of the part reorder interval on operational availability, or part stockouts, or overstocks.This study implemented Monte Carlo simulations in Python, integrating diverse distributions for time to failure, repair, and resupply. The research evaluated the forecastability of a Poisson-based spare part analysis method, categorizing demands into four types, and assessed its performance via operational availability and overstock ratio for a single-indenture, single-echelon system over a target operational period.Changes in failure distributions from infant mortality to wearout mode resulted in more conservative Poisson-based spare prediction results (e.g., higher overstock). Similarly, increasing the probability of spare sufficiency also led to more conservative Poisson predictions, while simultaneously increasing availability and decreasing its variation. Conversely, backorder counts exhibited trends opposite to availability when considering failure distributions, average failure rate, and probability of spare sufficiency. An increase in the replenishment interval shifted the demand category from intermittent to smooth for a case of wearout mode. While overstock rates were lower for smooth demands and higher for intermittent demands, no clear trends were observed in the availability results based on demand types.This study provides valuable support for decisions like setting safety stock levels and developing optimal inventory control solutions. These solutions aim to ensure timely fulfillment of parts demand in a cost-effective manner.Future work will expand on the foundation by incorporating repair models with dynamic modeling capabilities, creating redundant system models, and seeking to leverage AI models to optimize inventory management.