Process automation, from basic to advanced is deployed widely within the refining industry. As a component of organizations’ competitive business strategies, modern DCS systems and associated regulatory process controls, inferential control, multivariable control and real-time optimization are all automation technologies employed in the industry to varying degrees. It is well documented that these technologies deliver economic benefits. Unfortunately however, as management now understand, these benefits are not generated without the long term costs and operations management issues associated with sustaining what can be a fragile infrastructure.
All of these assets, from the valves and sensors to the optimization applications, work in a coordinated fashion to operate the plant optimally. Today, control engineers and maintenance technicians are challenged with sustaining the performance of multiple layers of automation. Sticky valves, controller tuning, and disturbance problems for example are all commonly reported at the regulatory level while utilization, constraint, tuning, and model problems are normal maintenance issues with widely adopted technologies like Model Predictive Control (MPC).
The large number of these assets in a typical refinery has dictated that a complaint driven or failure based maintenance model be employed, with only the most important of these devices receiving regular attention. The result has been sub-optimal control performance and ultimately a negative impact to business performance. As a significant opportunity exists to improve the business, management has turned an eye toward the reliability and operations management of these automation assets.
This paper looks at the current problem of control performance with the assets in the plant control hierarchy up to and including the model predictive controllers in detail. Limitations of the existing maintenance approaches are discussed and an alternative to the most common methods is presented. A condition based approach to the problem will be discussed with case studies and industrial results presented.