(81a) RTO has not Realized its Potential Due to Conflict with MPC

RTO has not realized its Potential due to Conflict with MPC

                                             By:  Charles Cutler – Cutler Technology

                                                     Don Golden – Cutler Technology Consultant

                                                     Charles Johnston – Mustang Engineering

Real Time Optimization (RTO) has not realized the value to the process industries that it should.  There have been political and management problems that account for some of this failure to realize the value of the technology; however, there are still some technical issues that need to be resolved.  The primary technical problem is the interfacing of the RTOs to the Multivariable Predictive Controllers (MPC).  One issue that should be clear to everyone is that having two different optimizations, with two different models trying to solve the same problem is not correct.  A linear program (LP) optimizer is used in the MPC and a non-linear optimizer is used in the RTO.  The conflict between these optimizers has routinely created infeasibilities in the controller’s LP.  The general solution to the infeasibility problem is to rank the controlled variables and drop out the lower ranked variables from the LP until it becomes feasible. The flip flopping of the LP solution when variables are dropped out leads to the unstable behavior of the MPC. This paper will present a solution to the conflict between the MPC and RTO.

About the Authors

Charles Cutler obtained his PhD from the University of Houston in Chemical Engineering.  His dissertation topic was “Dynamic Matrix Control an Optimal Control Algorithm”.  He was the founder of Dynamic Matrix Control Corporation, which he sold to Aspen Technology in 1996.  He has been active in the application and development of Real Time Optimization and Multivariable Predictive Control systems since 1963.  He is currently president of Cutler Technology. 

Email Address   Charles.Cutler@Cutler-Tech.com

Charles Johnston did his undergraduate work in Chemical Engineering.  He holds a doctorate in dentistry from the University of Texas and a MS degree in mathematics from Rice University.  While going to dental school he worked as a contractor part time in Shell Oil’s Control Research Group.  After practicing dentistry for two years, he joined Cutler in the development of the first configurable software for MPC in 1985.  He has been active in developing control software and applying it for the last twenty eight years. 

Email Address    Charles.Johnston@Mustangeng.com

Don Golden received his BS and Master degrees in Electrical Engineering from Rice University.    He has worked in the Multivariable Predictive control area for Shell Oil, Dynamic Matrix Corporation and Aspen Technology.  At Aspen Technology he was responsible for the Advanced Control Software Development organization and then later was the Principal Instructor in the Advanced Control area.  Most recently he has served as a consultant to Cutler Technology.

Email Address   Don Golden <don.golden@cutler-tech.com> (don.golden@cutler-tech.com)