(387a) On Identification and Control of Reactive Extrusion Processes

Reactive extrusion is an important commercial process used for the production and modification of a wide variety of polymers and blends (e.g. Ethylene co-polymers, Polyamides, etc.). Its primary distinguishing characteristic is that chemical reactions are deliberately carried out during continuous melt extrusion to achieve desired product properties. The dynamics of the process are determined primarily by the strong interactions between the fluid mechanics, reaction kinetics, heat transfer and the complex extruder geometry, making effective control of product quality and end use properties very difficult [1]. It is important to note that the usual challenges associated with the control of conventional polymer reactors (e.g. availability and low frequency of physical property measurements, and frequent grade transitions) are also encountered in the control of reactive compounding.

Many recent attempts at controlling reactive extrusion have focused on the control of a single variable, such as viscosity [2], which is inadequate to guarantee desired product quality and end use characteristics. To meet the customer's demands efficiently, it has become important to develop a comprehensive methodology for effective control of not only process variables ?y' (Melt pressure, melt temperature, motor power etc.), and product properties ?q' (Melt index, viscosity, density, etc), but more importantly, end-use physical characteristics ?w' (toughness, UV/chemical resistance, etc.). Only by such a comprehensive approach can one guarantee acceptable end-use product performance. Our ultimate objective is to develop such a framework for controlling key product properties and assuring acceptable end-use performance.

Our approach to this challenging problem is to begin with an appropriate mathematical representation of the relationships between variables across the entire processing chain, and then to use these models for two crucial tasks: (i) to translate the customer requirements on end-use performance to set points for process variables, and (ii) to make appropriate modifications (i.e. take control action) wherever appropriate along the ?manufacturing chain? on the basis of all available feedback information. In the proposed framework, the mathematical representation translates into a hierarchical sequence of models, which are based on first-principles, empirical data or both.

In this presentation we will first discuss the framework proposed to achieve the stated objectives, and then present our current work on modeling the relationships between process manipulated inputs ?u' and process output variables ?y', specifically for an experimental system involving the reaction of a functionalized ethylene co-polymer, ?Elvaloy? (Ethylene/n-Butyl Acrylate/Glycidal Methacrylate Terpolymer) with an acid co-polymer ?Nucrel? (Ethylene/Methacrylic Acid Copolymer) supplied by DuPont, in a Coperion W&P ZSK-30mm co-rotating, intermeshing twin screw extruder. Due to the complexity of the interacting process mechanisms, first-principles modeling is impractical for control at this level; system identification from carefully obtained input/output data is a more practical alternative.

We will present and illustrate a strategy for effective model identification of this class of processes. The strategy consists of carefully designed identification experiments involving preliminary process tests and a final test, followed by judicious model structure and model order selection. The details of what each test entails, how the results of the preliminary tests are used to design the final test and the procedures for model structure and order selection will be discussed in the presentation. For example, an analysis of the preliminary tests carried out on our experimental system revealed that the nature of the process response is significantly different at low and high viscosity operating regimes (indicating significant nonlinearity). The final test was therefore designed to generate a different model (of appropriate structure and order) for each operating regime.

References:

1] Janssen L. P. B. M., ?On the stability of reactive extrusion?, Polym. Eng. Sci., 38 (12), 2010 ? 2019.

2] Pabedinskas, A.; Cluett, W. R., ?Controller design and performance analysis for a reactive extrusion process?, Polym. Eng. Sci., 34 (7), 585-597 (1994).