(231e) A Systematic Approach to Automation & Control of Batch Processes

McFarland, T. P., Carnegie Mellon University
Suresh, P., Dow Chemical Company
Ydstie, B. E., Carnegie Mellon University

Batch and semi-batch operations are prevalent in many branches of the chemical manufacturing industry: food, pharmaceuticals, and specialty chemicals.  Batch plants are suitable for small volume-high value chemical products, flexible manufacturing, making products to order and reducing time to market.  Because process chemistry development takes place in laboratory batches, new process scale-up is less time consuming and less expensive than for continuous processes.  While typically less labor and capital intensive to scale-up from bench scale experiments, batch processes present unique issues for process control.  The dynamic characteristics of batch processes are inherently different from continuous processes in four distinct ways: absence of a steady state, repetitiveness of batch runs, finite processing time, and the nonlinear time dependency of process variables. 

However, the control literature predominately deals with continuous processes, and batch control research is often focused on the application of advanced control and optimization algorithms to specific case studies.   The goal of the current work is to examine the characteristics of batch processes and devise a systematic approach for control design.  The main objective is to develop control system architectures that maintain ease of design and flexibility so that specific process equipment can be applied to very different chemistries without significant change to the control strategy.

Recent research efforts define theoretical control properties such as stability and controllability for batch processes in terms of sensitivity functions[1].  In the current paper we extend the application of these ideas through the application of the theory of incremental passivity.  The incremental passivity based control addresses the control of continuous time process variables and can be used to analyze stability along trajectories rather than stability of fixed reference points as is common in Lyapunov and classical applications of passivity theory.  In this approach the batch process is seen to progress through a series of discrete states, defined by logic decisions defined by the batch recipe. In between these discrete states the batch system may “coast” along a trajectory given by initial conditions and maybe a few additional controls if we consider a semi-batch process. With computer automation of the entire batch process, the control design approach we develop take into account the stability along the trajectories as the process transitions from one discrete state another. 

The procedural approach (work process) for batch automation we propose derives the continuous and discrete control actions directly from the batch recipe. We follow the modeling framework defined in the ISA-88 Standard for batch control. In this standard a distinction is made between the process model, which contains the recipe and descriptions of chemical transformations, and the equipment model, which describes vessels and instruments used to carry out reactions and separations needed to make the final product. The separation of these models in the control design makes it possible to see which part of the controller is connected with the recipe and which part is connected to the specific equipment. The separation of process and equipment models facilitates the development of control strategies that can be reconfigured to take into account modifications of the control design as a function of the recipe as the plant changes from producing one product to another.

[1] Srinivasan, B., & Bonvin, D. (2007). Controllability and stability of repetitive batch processes.  Journal of Process Control , 285-295.