2025 Spring Meeting and 21st Global Congress on Process Safety

(7a) Enhancing Furnace Performance through AI and Convection TMT Predictions

Authors

Arun Pitchaiahrajan, Ingenero Technologies
Deepesh Sawhney, Ingenero Technologies
Amrita Sachan, Ingenero Technologies
Ethylene furnace operations is vital to the overall performance of entire petrochemical complex, with enhancements in capacity, efficiency, and reliability directly contributing to increased production while lowering maintenance and operating costs. A smart data analytics solution transforms furnace data into actionable intelligence, addressing critical challenges such as differential coking, decoke effectiveness, instrumentation issues, scheduling, and process control variations—key factors for optimizing run length and yield.

This paper focuses on the challenges within the ethylene furnace especially in convection bank which typically has more unknowns and less data for decision making. And how applied AI provides much-needed visibility to address typical challenges including any capacity constraints and higher maintenance cost. Effective monitoring of convection section operations is essential for achieving optimal thermal performance, maintaining efficiency, managing fouling, and tracking coil Tube Metal Temperatures (TMT), all of which are crucial for safe and efficient operations.

The paper will showcase applications of AI-driven convection models that enhance operational performance and reliability by evaluating individual bank performance and identifying potential fouling and flue gas channeling. The approach includes a modeling solution for furnace convection section for estimating coil TMTs on LIVE basis, which facilitates proactive risk management through continuous monitoring against design limits and timely alerts for potential overheating. The presentation will highlight the benefits of an AI-powered dashboard that provides real-time insights, performance indicators, actionable recommendations, and targeted alerts to support effective decision-making without overwhelming operators. Additionally, the advantages of integrating the convection bank prediction model with radiant section models will be explored to further optimize overall furnace operations, promoting sustainable production while minimizing operational risks.