Energy efficiency modeling of fire tube boiler package 900-08B-2 at PT kilang Pertamina Internasional (KPI) Refinery Unit II Production S.Pakning using adaptive neuro fuzzy inference system

Abstract Energy efficiency is crucial in oil and gas operation where boilers play a central role in steam generation. This study evaluates the performance of Boiler Package 900-08B-2 at PT Kilang Pertamina Internasional (KPI) Refinery Unit II Sungai Pakning using a quantitative approach based on both the indirect method and computational intelligence modelling. The indirect method was applied to identify and quantify heat losses, revealing that major inefficiencies originate from flue gas, fuel moisture, air moisture, and radiation losses. To enhance predictive capability, the Adaptive Neuro-Fuzzy Inference System (ANFIS) was integrated, enabling efficiency estimation across varying operational parameters which is the objective of this research. The result of the energy efficiency calculation using the indirect method is 87%. The ANFIS based model achieved a high prediction accuracy with a Mean Absolute Percentage Error (MAPE) of 1.75% and a Root Mean Square Error (RMSE) of 0.028 and energy efficiency is 88%, showing strong consistency with indirect calculations. These results confirm the suitability of ANFIS for energy efficiency modelling. The research highlights that increasing input variables and expanding experimental datasets can further improve predictive robustness. This study provides a quantitative framework that supports modelling of boiler energy efficiency.