PTQ Q2 2024 Issue

Figure 5 CokeNil trend analysis

• The ‘Total Naphtha Vs Time’ graph shows the amount of naphtha (feedstock) that has been used over time. • The ‘COT Vs Time’ graph displays the COT of the furnace. • The ‘Economics’ section highlights the economic benefits of using the AI model, such as increased production and reduced costs. CokeNil alerts and alarms These key elements of the dashboard include: • TMT prediction and possible alert: Forecasts the TMT and generates alert to take necessary action. • Alert for critical parameters: If any critical parameters exceed safe limits, an alert will be triggered, indicating a potential concern. It is important to monitor and con- trol these parameters to ensure efficient operation of the furnace. All triggered alarms due to specific conditions will be emailed to the relevant authorities. This ensures that poten- tial risks are addressed efficiently, contributing to a seam - less and well-managed process. CokeNil what-if analysis The following elements are identified by the what-if analysis: • Plant admin – diagnostic analysis: Explains the overall purpose of the dashboard. • Status of the furnace (online/offline): Indicates if the plant is operational or not, which can affect the data displayed. • Actual vs predicted TMT: Compares real-time TMT with the AI’s prediction, aiming to optimise TMT for efficient and safe operation. • Specific input prompts: These allow the manual input of specific feedstock or cycle time data for analysis. • Recommendation and download: The AI suggest adjust- ments based on the TMT comparison, which can be down- loaded for implementation. • Recommendation acceptance: Shows the user acceptance

rate for the AI’s recommendations regarding TMT, indicating their effectiveness. • Similar structure for CPR and SHC ratio: These follow the same pattern as TMT, analysing and potentially optimising parameters for improved performance. Possible implications • CokeNil’s AI tool plays a crucial role in analysing plant per- formance, predicting key parameters and suggesting opti- mising actions. • By optimising TMT, CPR, and SHC ratio, CokeNil could potentially improve process efficiency, reduce waste, and enhance safety. • The high recommendation acceptance rate suggests user confidence in the CokeNil insights and its potential to deliver positive outcomes. Conclusion It becomes imperative to strategically optimise operations within steam crackers, with an emphasis on enhancing furnace efficiency to address the economic constraints and ensure sustainable performance. Central to this discussion is the continuous learning capability of AI, empowering the system to evolve and adapt based on new data and expe- riences. In the context of ethylene furnaces, this continuous learning contributes to data-driven decision making. The AI system becomes not just a static tool but also a dynamic partner in improving furnace performance and supporting extended run lengths. Surabhi Thorat is Head of Data Science at Dorf Ketal Chemical (I) Pvt Ltd. Vivek Srinivasan , is Senior Manager, Global Technical Services at Dorf Ketal Chemicals India Private Limited. Email viveks@dorfketal.com Sudarshan Vijayaraghavan is District Manager at Dorf Ketal Chemicals PTE Ltd.

PTQ Q2 2024