Improvement of a numerical two-phase simulation model for single-screw plasticizing extruders based on experimental investigations

Abstract The accurate prediction of throughput and the melting process in single-screw extruders are crucial for the design and optimization of plasticizing units, as the melting process is often the rate-limiting factor and markedly influences final product quality. While the design of these units has long relied on empirical values and simplified analytical models, Computational Fluid Dynamics (CFD) now offers a powerful alternative for process analysis. However, its predictive accuracy strongly depends on the precision of the underlying physical models. This work aims to improve and validate a numerical two-phase simulation model for plasticizing extruders. To this end, extensive experimental investigations were conducted using screw pull-out tests, following the method of Maddock, on an extruder with a 30 mm diameter. Utilizing High-Density Polyethylene (HDPE) and two different three-zone screws, a broad dataset was generated through a statistical design of experiments by systematically varying screw speed and barrel temperature. Based on this experimental data, targeted adjustments were made to a CFD model that builds upon the continuum approach by Karrenberg and further advanced by Dörner and Schall. The key improvements include the implementation of models for the bulk density and thermal conductivity, which account for the porosity of the bulk material in the screw channel as a function of pressure and temperature. The validation of the improved model shows good agreement between simulation and experiment. Throughput is predicted with high precision, and the simulated melting profiles reproduce the expected melting mechanism, although the melting lengths appear to be underestimated when compared with the cross-sections from the screw pull-out experiments, the accuracy of which is subject to discussion. Analyses at the process limits, where unmelted material was observed exiting the extruder, likewise revealed that the model overestimates the melting performance, although the extent of overestimation is smaller. This study demonstrates that by adapting the physical sub-models, particularly by considering porosity, the predictive accuracy of CFD simulations can be markedly increased, especially with respect to throughput, thus providing an alternative tool for extruder design.