DETERMINING KEY FACTORS OF NOX EMISSION IN H2 RICH GAS FIRED - NATURAL DRAFT HEATER OPERATION USING DATA MINING TECHNIQUE

• Main Author: Martua Raja Hasibuan, Togar
• Format: Theses
• Language: Indonesia
• Subjects: Manajemen umum
• Online Access: https://digilib.itb.ac.id/gdl/view/76074
• Institution: Institut Teknologi Bandung

Crude oil refining is vital in the oil industry, involving the breaking and combining of hydrocarbon chains. Oil refineries heavily rely on heaters, impacting fuel consumption, NOx emissions, and operating costs. However, controlling heaters can be challenging, leading to compliance issues with strict emission regulations. Hydrogen-rich gas is an attractive fuel source in heater operation due to its ability to reduce carbon and NOx emissions. Yet, combustion of nitrogen in the air used for hydrogen-rich gas remains a significant NOx contributor. Elevated oxygen levels worsen NOx emissions. This research investigates factors causing NOx emissions in hydrogen-rich gas-fired natural draft heaters used in refineries using data mining technique to understand why some low-NOx burners fail to meet emission limits despite hydrogen-rich gas benefits. The approach involves a literature review on NOx emissions in such heaters, and data mining to analyze operational data, fuel characteristics, and burner parameters. The aim is to develop NOx emission models and identify influencing factors. By using data mining, practical solutions for mitigating NOx emissions and meeting regulations can be developed based on random forest approach. The research offer short-term and long-term recommendations to address high NOx levels. Based on the result using random forest modeling, NOx emission can practically be adjusted through excess air levels in the heater as well as the stack pressure. The model can provide a promising mean absolute error (MAE) of NOx emission by 4.28%.