Kautschuk Gummi Kunststoffe, Vol.55, No.5, 232-235, 2002
Process control PUR - Statistical process models and neural networks control the reaction moulding of polyurethanes
Modern production machines offer a wide range of supervisions of all set values; the reaction in the mold itself during the production cycle of reactive polymethane can not be monitored with these systems because of missing technology. With the use of the dielectric spectroscopy an online control of the reaction of the most important process value is now possible. Monitoring the electrical conductance of the reacting polyurethane system leads to an improvement of quality calculations based upon statistical process models and neural networks. By using information of the dielectric behavior of the reaction the accuracy for the calculation of the E-modulus has been increased by 2-4 %. Process deviations can be detected and removed early during the production cycle. Consequences are lower reject rates and improved productivity.