DEGRADATION‑AWARE SPARES OPTIMIZATION WITH WEIBULL/PHM UNCERTAINTY AND LOGISTICS DELAYS

Abstract

Spare parts inventory optimisation for critical industrial equipment requires accurate characterisation of both equipment failure demand and the lead-time distribution over which that demand must be covered. Conventional inventory models treat failure rates as stationary Poisson processes calibrated from historical mean demand, systematically ignoring two sources of uncertainty that cause costly stockout events in practice: the time-varying, condition-dependent hazard rate structure captured by Proportional Hazard Models (PHM) and the stochastic variability of procurement lead times arising from supply chain disruptions, customs clearance delays, and supplier capacity constraints. This paper presents a degradation-aware spares optimisation framework that jointly models equipment failure demand through a Weibull-baseline PHM with continuously updated degradation covariates and procurement lead-time uncertainty through a log-normal mixture distribution calibrated from CMMS purchase-order records. The integrated demand-over-lead-time distribution is computed via numerical convolution of the PHM-driven compound Poisson demand process with the stochastic lead-time distribution, enabling derivation of optimal reorder points and safety-stock levels that correctly account for both failure-rate non-stationarity and logistics delay variability. Validation is performed on seven critical spare-part classes at a refinery facility, demonstrating a 24.7% reduction in the total sparing cost index (holding plus stockout plus ordering cost), a 59.2% reduction in critical-spare stockout probability at the target 95% service level, and a 17.8% reduction in on-hand stock levels relative to conventional static min-max policies. The proposed framework is directly implementable within SAP MM inventory modules and ISO 14224-aligned CMMS environments.