(371j) Visualizing Industrial Multivariable Control Constraints In Operations

A common practice in the refining and petrochemical industry is to use simple dedicated graphics to monitor Multivariable Controller (MVC) applications and understand what constraints are limiting plant operation (Richmond, 2010). This information may be used by operations to maximize feed rates and to further optimize their process. However, multiple constraints may limit the process, and, due to the large numbers of manipulated and controlled variables, it can be very difficult to understand the interaction between constraints. Thus, a simple dedicated graphic that shows the currently active constraints is not always sufficient to understand the behavior of industrial MVC applications.

In this paper we explore the use of parallel coordinate systems to visualize and interpret the affect of constraints on MVC controllers. This system shows great promise in allowing us understand the MVC trade-offs between constraints as we can used it to display the set points, present values, and limits of both controlled variables and manipulated variables on the same two dimensional plot. This method is demonstrated for a two input, two output system and compared to a two dimensional phase plane approach and temporal plot for a system described in the literature. The proposed method is also extended to a two input, three output MVC controller.

Richmond, P.C. (Survey Conducted by: Ethylene Producers Committee, Process Control Subcommittee), "Support Practices for Advanced Process Control," 2010 AIChE Spring National Meeting, San Antonio, TX, March 2010, Paper no. 50e.