PHYSICS BASED MODELING AND CONTROL OF REACTIVE EXTRUSION

Loading...
Thumbnail Image

Files

umi-umd-1453.pdf (3.54 MB)
No. of downloads: 3949

Publication or External Link

Date

2004-04-30

Citation

DRUM DOI

Abstract

Kinematic modeling has been shown to be important for the understanding and control of co-rotating twin screw extruders. The residence time distribution (RTD) is often used to characterize the steady-state behavior of an extrusion process. Due to the complex rheological behavior of polymer flow in the extruder, few have felt that the RTD would be independent of changes in operating conditions for the same screw configuration. To investigate, we are asserting that resident distributions could be independent of operating conditions for certain types of polymers. Four different polymers, two polyethylenes and two polypropylenes, were processed on the same 30mm Werner and Pfleiderer co-rotating twin-screw extruder (CoTSE) equipped with reflectance optical probes to compare their RTD's. Additionally, each material was tested to determine its complex viscosity, to better understand the phenomena involved.

Using physically motivated models to control reactive extrusion processes is attractive because of the flexibility and robustness it could provide. This thesis uses residence distribution analyses to characterize the material flow through a co-rotating twin-screw extruder. Furthermore, we examine the applicability of residence distributions as the basis for kinematic modeling of the extrusion process. This demonstration of using a steady-state model - the residence distribution - as a basis for kinematic behavior is unique. The signals have been deconvoluted to kinematically characterize the flow in the different regions of the extruder, such as the melting, mixing and metering zones. Studies of step changes have shown that the steady state value of extrudate viscosity is dependent on the peroxide concentration, volume mixing, and on the residence time from the specific throughput. This data has also provided plant models of the peroxide initiated degradation reaction using system identification techniques. Although a specific example of vis-breaking of polypropylene is studied, the techniques are general. A proportional and integral controller (PI) with a Smith predictor was used to track set point changes and regulate the viscosity.

Notes

Rights