ANALYSIS OF INPUT MULTIPLICITIES IN CONTROLLED CHEMICAL PROCESSES

KURT ALAN RICKARD, Purdue University

Abstract

Controlled chemical processes may exhibit more than one steady operating state for some sets of controller set points. This behavior has been termed input multiplicity and can cause poor static and dynamic performance of typical control systems. In some cases, process models with input multiplicities may exhibit multiple stable steady states, oscillatory states, or saturation of manipulations. At each of the multiple stable steady states, the controlled variables are maintained at their set point values. The manipulations and other process variables are at significantly different values, resulting in greatly differing economics of process operation. The research is directed at demonstrating numerical examples of the occurrence and consequences of input multiplicities, and at the development of a method of detection of these problems. The existence of a catastrophe set is shown to be a necessary condition for the occurrence of multiple solutions for the general multi-control loop process. The catastrophe set contains steady state solution points for which the linearized Jacobian matrix of the nonlinear process is singular. A general sufficient condition is derived. The steady state process model is used to determine which steady states may be asymptotically stable under a multiloop linear control system containing resetting action. A method is developed to determine a limit on the number of these steady states which may be stable under any one multiloop control system.

Degree

Ph.D.

Subject Area

Chemical engineering

COinS