RCU optimization in a DCS gives fast payback. [Residue Cracking Unit; Distributed Control System]

The BP refinery at Kwinana in Western Australia recently installed a multivariable predictive controller on the reactor/regenerator section of their residue cracking unit (RCU). This application is unique in that the controller resides entirely within the distributed control system; there is no host computer. The controller consists of seven manipulated variables and 22 controlled variables. It is required to effectively control a process which is subject to severe feedstock variation. Sophisticated product value optimization, integrated with the regulatory and constraint handling capabilities of the multivariable controller resulted in a payback period for the project of less than three months. The paper discusses the process, functional requirements, control design, project implementation, and results.