• What is the roadmap to achieve autonomous refining in a complex oil refinery using digital technology?



  • Anant Chourikar, HCL Technologies Limited, anandashok.chourikar@hcl.com

    To achieve autonomous refinery, a company has to go through many steps, some of it in sequence and some in parallel. The 7 step process helps in creating a detailed roadmap for companies to achieve autonomous refining. Step 1: IT-OT integration, Step 2: IIOT & Edge deployment, Step 3: Implementation of Complex Event processing Engine Step 4: SCADA, HMO, DCS integration with Complex Event Processing, Step 5: Enablement Layer consolidation & collaboration Step 6: Define integrated functional Engagement Layer (KPI) & Troubleshooting, Step 7: Self-healing Autonomous Operations

    Each step will have multiple stages and may need orchestration layer for control operations in the event of malfunctions.

    Core decision will be taken up at self-healing and KPI layer while operations risks are managed at complex event processing stage.

    Most of the models will be defined at the edge level within refinery.








  • Doug Morgan, Searles Valley Minerals, morgan@svminerals.com

    Start with including computer process control experience as a qualification for employment. My experience is that some, not all, refiners proactively limit the number of field employees with process control skills and responsibilities. Those with it, are usually forced to be on call to those in operations who don't know what is happening nor what to do, as if not having skills gives one the right to take command.



  • Alon Davidy, TOMER, alon.davidy@gmail.com

    In my opinion it may be possible to implement the idea of autonomous refining in complex oil refinery by using the multi- agent reinforcement learning (one of branches of the machine learning algorithms). Each agent may represent one of the refinery compnent (such as: FCC unit, Hydro-Cracker unit, Delayed coker unit). A typical multi agent system contains several interacting agents, where each agent is a computer system or an algorithm capable of performing autonomous activities to achieve certain tasks in an environment. The ability of agents to interact with each other through collaboration, while adapting to changes and disruptions, makes them suited for distributed control. Due to its conformity with the complex and distributed nature of the Oil and Gas Industry. This approach has already gained attention of researchers as a means to solve a number of tasks in the Oil and Gas Industries. One of the significant applications of this system is to detect anomalies. This system may predict failures based on relationships between valid readings.



  • Nagarathinam S Murthy, McKinsey & Company, Chennai, nssvdvr@gmail.com

    I look at this issue in a macro level as "read and or write" access to any common digital platform by all employees & managers who have stake in day to day refining. To illustrate, gasoline octane ex FCC can be attributed as a result of 'right feed quality, FCC E Cat Mat, Operating severity and Cut point. The scheduler of the refinery makes a decision based on the LP runs and offers a programme digitally on daily or weekly basis for Operations to adhere to. If at the end of the day, deviations have been noticed wrt Octane barrel of gasoline, the management will be certainly keen to know where the change (ether error or deliberate move) got introduced on the digital display which could be either live or on cumulative 24 hr basis. An autonomous system should be able to capture such change automatically and facilitate management to take corrective measures thereof.