A Metric-Driven Framework for Evaluating Prognostic Based Failure Estimation on Spare Part Inventory Management

Abstract

In recent years, the aviation sector has been at the forefront of adopting Industry 4.0 technologies, including artificial intelligence, additive manufacturing, cyber-physical systems, big data analytics, and the Internet of Things (IOT). These technologies have accelerated the development of advanced main- tenance strategies, such as Predictive and Prescriptive Maintenance, especially in the field of spare part inventory management. By leveraging insights from Prognostic Health Management (PHM) technologies, logistics and maintenance serviceproviders can optimize inventory levels to reduce costs while maintaining service levels, thereby minimizing aircraft downtimes. Despite these potential advantages, the widespread adoption of PHM strategies in the Maintenance, Repair and Overhaul (MRO) industry remains a challenge, primarily due to their modelling complexity, high cost of adoption, regulatory challenges, data availability, and impact assessment. However, a more targeted allocation of development resources, to address these barriers, can be achieved if economic benefits can clearly be demonstrated for individual stakeholders (such as logistics) and different PHM technology maturity levels. Therefore, the aim of this study is to quantify the benefits of prognostic-based inventory policies in comparison to traditional reliability-based approaches across different demand patterns. Specifically, this study investigates the influence of different prognostic accuracy and prognostic horizon levels on key performance indicators, such as total cost and service level. It also evaluates the robustness of the proposed methodology against noise factors like prediction biases and false alarms. Based on these comparisons, minimum performance requirements for prognostics based policies can be established to ensure tangible benefits. Consequently, this study not only provides the readers with a methodology to quantify the impact of prognostics-based failure prediction on spare part inventory management, but it also proposes a lightweight framework which could act as surrogate for prognostic models to assist in the development of future prescriptive maintenance strategies