Department

Chemical Engineering

First Advisor

Dr. David M. Bagley

Description

Model predictive control has been used in chemical engineering processes since the 1980s. The chemical process industry still uses model predictive control; however, it is expensive, time-consuming to install and calibrate, computationally demanding, and must be retuned when process conditions change. The chemical process industry has long been looking for a simple, robust, inexpensive replacement for model predictive control. Allan Kern, a 1981 University of Wyoming alumnus, has recently patented a new process control system that may eliminate the problems of model predictive controllers. Mr. Kern developed the rate predictive controller (RPC) to address the challenges with model predictive control. Instead of using highly empirical models, RPC simply adjusts the controller output based on the rate of change of the controlled variable. The goal of this project was to evaluate the theoretical performance of RPC. Excel, MATLAB, and Simulink were used to simulate RPC. Experiments were run in these simulation environments to test the performance of RPC under different conditions. Additionally, the effects of changing key RPC parameters, such as process response time and controller band size, were examined. These experiments yielded varying results. While theoretically instability could be achieved, by following Mr. Kern’s guidelines RPC was found to be operationally stable. The simulation experiments were also able to answer questions customers of Mr. Kern have had about RPC, such as the effect of disturbances and higher order transfer functions.

Included in

Education Commons

Share

COinS
 

Rate Predictive Process Control

Model predictive control has been used in chemical engineering processes since the 1980s. The chemical process industry still uses model predictive control; however, it is expensive, time-consuming to install and calibrate, computationally demanding, and must be retuned when process conditions change. The chemical process industry has long been looking for a simple, robust, inexpensive replacement for model predictive control. Allan Kern, a 1981 University of Wyoming alumnus, has recently patented a new process control system that may eliminate the problems of model predictive controllers. Mr. Kern developed the rate predictive controller (RPC) to address the challenges with model predictive control. Instead of using highly empirical models, RPC simply adjusts the controller output based on the rate of change of the controlled variable. The goal of this project was to evaluate the theoretical performance of RPC. Excel, MATLAB, and Simulink were used to simulate RPC. Experiments were run in these simulation environments to test the performance of RPC under different conditions. Additionally, the effects of changing key RPC parameters, such as process response time and controller band size, were examined. These experiments yielded varying results. While theoretically instability could be achieved, by following Mr. Kern’s guidelines RPC was found to be operationally stable. The simulation experiments were also able to answer questions customers of Mr. Kern have had about RPC, such as the effect of disturbances and higher order transfer functions.