Analyzing the Effect of Prognosis on Spare Part Inventory Management

Abstract

For an effective inventory management, the accurate prediction of demand is critical. While advanced prognostic algorithms theoretically exist in aviation, their adoption has been limited thus far, partially due to high development costs, extensive data requirements, computational complexities and their impact assessment. This study, addresses the question of whether a benchmarking method can be defined, to determine if the prognostic failure predictions provide measurable advantages over statistics-based approaches for spare part inventory management. A metric-driven surrogate framework is proposed that emulates prognostic models using two key performance/maturity indicators: prediction accuracy (α) and prognostic horizon (PH). Implemented within a discrete-event simulation (DES) environment, the framework evaluates how variations in these metrics affect total inventory cost and logistics service level (i.e., the share of fulfilled material requests without delays). The results provide practitioners with breakeven points between predictive and traditional reliability-based approaches. They further highlight which aspects of prediction capability have the highest influence on operational benefits, enabling targeted improvements in predictive performance