Experimental Study on the Scalability of Planetary Roller Extruders

This contribution aims at developing scaling algorithms for planetary roller extruders (PREs). Laboratory- and production-scale experiments were carried out, using thermoplastic polymers according to a statistical design of experiments (DOE). By comparing plant size, spindle configuration, operating parameters, and material properties, their influence on pressure build-up capacity, process temperatures, and residence time distribution is analyzed. All data generated are used to train MATLAB-based machine learning models. First indications hint at Gaussian processes and artificial neural networks, predicting operating parameters with high accuracy.