Designing control strategies from a plant-wide perspective

Chemical manufacturing plants nearly always consist of many manufacturing stages, made up of unit operations such as mixers, reactors, and separators. Each unit operation requires at least a minimum control system to insure proper operation. Typically, these unit control systems are designed separately, if not independently. As higher demands are placed on manufacturing for increased quality, productivity, and efficiency, it is increasingly apparent that more attention needs to be given to the plant-wide context in the design of the control system for each unit operation. This study presents a straightforward approach to designing control strategies for multi-unit plants. The work follows a plant-wide approach, which emphasizes process knowledge, plant objectives, and management of inventories and variation. Current steady-state tools for control strategy design do not adequately address the first-level material balance control decisions (selection of throughput manipulator, design of level controls) that must be made early in the process of designing control strategies. This work uses a simple plant module to develop heuristics for the design of these material balance controls. The current literature also does not describe a systematic procedure for selecting sets of manipulated and controlled variables from large variable sets. This work describes a mathematical procedure for identifying variables that do not contribute to the control of the plant, and for grouping variables that represent essentially the same information. The material balance heuristics, the variable selection procedure, and existing literature tools were evaluated by using them to design a multi-loop control strategy for the Eastman Challenge Plant. The control strategies suggested by this procedure compare favorably to other published solutions to this test problem.