Predictive Age-Based Model for Spare Parts Inventory Management with Intermittent Demand Patterns Considered Repair or Replace Decision

Abstract

The demand for spare parts in diesel power plants in isolated areas, such as those owned by PT PLN (Persero) UP3 Ambon, exhibits an intermittent pattern that rarely occurs but carries significant consequences if unavailable. One critical component that frequently experiences failures is the Automatic Voltage Regulator (AVR). A more responsive and predictive inventory management approach is required to address this issue. This research proposes a forecasting model based on the age of spare parts and equipment, utilizing maintenance records and installed base data. A probabilistic forecasting model is developed by considering the survival probability of the equipment and part failures using the Weibull distribution. Then, age-based machine policy calculations will determine whether the damaged part should be repaired or replaced, affecting the forecasted demand. The forecast results are then integrated into a dynamic order-up-to level inventory control policy, allowing stock levels to be periodically adjusted according to actual demand risk. This approach aims to improve the service level above the minimum target while minimizing holding costs. This single-item model is expected to be applied to other spare parts in real-world cases to manage different types of intermittent demand.