Analytical Framework for Spare Parts Inventory Control Using K-Means and Adaptive Forecasting

Abstract

The increasing complexity of power plant operations has made spare parts inventory management increasingly critical. At PLTU XYZ, inventory value increased from IDR 11.85 billion in 2023 to IDR 20.79 billion in 2025, while inventory turnover decreased from 4.23 to 1.99 times per year and deadstock remained around IDR 3–4 billion. The existing system still relies on material classification and engineering evaluation, while demand forecasting has not been systematically applied. This study develops an analytical framework based on K-Means clustering and adaptive forecasting. Clustering of 1,215 items using criticality, availability, usage value, ADI, and CV² produced four clusters with a mean silhouette coefficient of 0.45. In the forecasting stage, nine sample items were evaluated using SES, Croston, SBA, TSB, mSBA, and mTSB through grid search. The results show that SBA performed best for four items, mSBA for three items, and Croston and TSB for one item each. Rolling-origin evaluation indicated that five items achieved method dominance of 85.7%–100%. This framework supports adaptive inventory control and Inventory Policy development.