A first attempt at genuine quality control based on experience in predicting the quality of continuously cast products is being undertaken in an ongoing research project. Prerequisites in this respect include the comprehensive compilation of process, machine and quality parameters and the development of suitable models for predicting the main quality characteristics. A CSP plant has been selected for this project because of the way in which the individual process stages are closely coupled. The surface quality of the produced hot strip, which is monitored by means of an automatic surface inspection sys-tem, is serving as the central quality characteristic for the time being. The feedback loop for the result-ing data extends as far as the casting machine.

This approach, which necessarily encompasses all the process stages, is an attempt to exert a posi-tive influence on the quality of the end product as early as possible. With the aid of a correlation analy-sis, it is possible to compare the recorded process variables offline with the quality data measured in downstream process stages, so as to ultimately develop a suitable, predictive model.
These predictive quality models can then serve as a basis for controlling the process in such a way that the quality of the end product in question complies with the given specifications. Control of the quality can be realised either as a so-called open or closed loop, thereby enabling the information and correlations contained in the accumulating mass data to be used continuously for selective process optimisation.