Liquid oxygen (LO₂) delivery losses can reduce its distribution efficiency, increase operational costs, and affect delivery reliability. This study analyzes delivery-loss patterns, evaluates forecasting methods, and identifies operational factors contributing to loss variability at PT XYZ. A quantitative case-study approach was applied using delivery-loss data from February to July 2025, with monthly averages of 10.00%, 39.20%, 9.00%, 8.20%, 9.20%, and 9.30%, respectively. Linear regression, three-period moving average, single exponential smoothing, and Holt’s linear trend were evaluated using Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE). Linear regression indicated a downward tendency (Ŷ = 23.58 − 2.694X; R² = 16.83%), but with limited explanatory power. The three-period moving average achieved the lowest MAD (5.45) and MAPE (63.37%), whereas single exponential smoothing yielded the lowest MSE (45.25). Holt’s method produced the largest forecasting errors, indicating its unsuitability for the short and highly fluctuating dataset. The findings highlight the importance of selecting forecasting methods according to data characteristics and strengthening operational controls. Improvements in data recording, SOP compliance, equipment maintenance, calibration, and routine monitoring are recommended to enhance the reliability and efficiency of liquid oxygen distribution.
Copyrights © 2026