Artificial Intelligence in Autonomous Operation of Oil and Gas Facilities

摘要

Objectives/Scope: Remote, inaccessible locations, hazardous environments and cost are some of the drivers for unmanned or remotely-operated facilities. With insight from facilities which are already operating remotely, it is time to take the next step towards running autonomous plants with no people. With reference to external sources and client case studies, this paper will discuss the preconditions and technological developments which now make it possible to do this safely and economically and how Artificial Intelligence is a key component. Methods, Procedures, Process: The paper describes key criteria and technologies needed for successful autonomous operations. In particular, companies must ensure operations are safe through process safety management. They must also be able to run operations with no physical presence via automatic controls for startup, change and shutdown. Maintenance and inspection must also be done with no-one there, using plant-wide condition monitoring and predictive analytics, as opposed to only tracking critical equipment as the latter is insufficient in processes with so many interdependencies. Robots and drones must also be leveraged where manned intervention is required e.g. for pipeline scraper handling. Results, Observations, Conclusions: The importance of zero operators is significant. Even just one employee, actually means 20 are needed for a site running all day, every day by the time rest, vacation and sick leave cover requirements are factored in, not to mention support functions like catering and cleaning. This is not only expensive but also means a significant number of people remain exposed to a hazardous working environment. Underpinning the above are Internet of things technologies to sense, measure and control operations, connectivity to share the information in a timely manner, the analytics to process it ? with a significant reliance on artificial intelligence? and appropriate dashboards to bring the information to the remote operator, domain expert or vendor partner. The importance of digital engineering and digital twins in facilitating such activities will also be discussed. Novel/Additive Information: We show that new safety, operational and maintenance procedures are needed to implement autonomous unmanned operation. Artificial intelligence with machine learning is needed to reach this goal. The paper will provide examples of howunmanned operations are becoming a reality. The safety and cost-saving benefits achieved from pilots and wider field applications of technology will be presented. Many of the technologies and operational changes will in themselves deliver reduced costs by eliminating operator errors, reducing maintenance expenditure and improving availability.