350913 Plantwide Dynamics and the Integration of Production Scheduling and Supervisory Process Control

Tuesday, April 1, 2014: 10:05 AM
Quarterdeck (Hilton New Orleans Riverside)
Michael Baldea, McKetta Department of Chemical Engineering, The University of Texas at Austin, Austin, TX

Modern chemical and petrochemical plants make extensive use of material and energy recycling to improve operating economics and minimize their environmental footprint. Such process integration measures are introduced at the steady-state design stage; yet, they have a profound impact on process dynamics. Integration is at the origin of strong feedback interactions, which lead to the emergence of a complex, process-wide dynamic behavior that must be accounted for in controller design. In the first part of this presentation, the main dynamics and control challenges and solutions related to process integration will be reviewed. A control framework for integrated processes, consisting of a hierarchy of fast, distributed controllers at the unit level, and a supervisory controller addressing plant-wide objectives, as proposed in our previous work (Baldea and Daoutidis, Int. J. Robust Nonlin. Contr., in press; Baldea and Touretzky, Syst. Contr. Lett., 62, 723–731, 2013) will be presented.

In the second part of our presentation, a connection will be established between supervisory control and the formulation and solution of the production scheduling problem. The aforementioned control  results serve as the basis for deriving an explicit low-order model of  the closed-loop process-wide dynamics, which is then utilized for formulating the scheduling problem as a dynamic mixed-integer optimization that is tractable in real-time. We show that this approach leads to economic improvements and significant computational cost savings over existing scheduling formulations, while preserving the closed-loop stability properties of the plant-wide control system. Implementations using existing supervisory control technology (e.g., model predictive control) and integration into enterprise resource planning (ERP) systems will be discussed.

Illustrative examples will be presented in support of the theoretical developments.

Extended Abstract: File Uploaded