Model Predictive Control (MPC) is a very attractive advanced process control technique for difficult multivariable control problems. One of the major advantages of MPC is its capability of predicting future system behavior based on the model, if accurate, and current measurements, and providing early warnings of potential problems [1]. This technique, offering a general control scheme, should be valid for product quality control as well, if the quality indicator(s) can be properly defined and measured (or estimated), and the product manufacturing dynamics can be adequately modeled. Note that in most industrial cases, product quality control must be coordinated with process efficiency control. This renders a challenging problem of how to achieve simultaneous control of both the process efficiency (in terms of energy and material use) and the product quality by means of MPC technique.

Integrated Product and Process Control (IPPC) was formally introduced by Xiao et al. [2]. The essence of IPPC design methodology is the coordination of the control of process efficiency and product quality using a classical cascade control scheme. Nevertheless, the control effectiveness by that methodology is restricted by the use of conventional PID control laws and the inability of dealing with model inaccuracy. Furthermore, the methodology does not take process capability issues into account during control design.

In this paper, we will introduce an enhanced IPPC design methodology where MPC is employed. In the resulting cascade control scheme, the inner-loop is designed to realize MPC-based process efficiency control under certain process capability constraints. The control objective of the inner loop is to ensure that the controlled process variable(s) provide the needed actions for product manufacturing. The outer loop, where MPC is also utilized, is designed to control the manufacturing of the product in a dynamic domain, which is driven by the “means” provided by the process system. By using this methodology, the product quality during the manufacturing process can be predicted more realistically, not just based on the measurement (either directly or indirectly obtained) of the product behavior, but also on the information about the process under capability constraints. In this work, a complete mathematical system will be presented with necessary derivations. The efficacy of the design methodology will be demonstrated through a case study which is about the control of polymeric coating quality and process energy efficiency under various process operational constraints.

Reference

1. Seboge, D. E., Edgar, T. F., and Mellichamp, D. A.. Process Dynamics and Control. John Wiley & Sons, 1989.

2. Xiao J, Huang YL. Integrated product and process control of single-input-single-output systems. AIChE J., 2007; 53:891-901.