Neural network analysis of the influence of chemical composition on surface cracking during hot rolling of AISI D2 tool steel

The reasons for the formation of surface cracks during the hot rolling of tool steels are not well understood. However, we know that apart from the parameters of the thermo-mechanical processing, the chemical composition of the tool steel has a big influence on the formation of these surface cracks. The majority of examinations of the hot deformability (appearance of surface cracks) of various steel grades made so far were limited to studying the influence of a minor number of chemical elements on the formation of the surface cracks, where the databases were based on laboratory tests. This paper proposes a new approach to the study of hot workability by analysing crack formation during the hot rolling of AISI D2 tool steel. The database was formed on the results from the surface cracking of rolling stock in an industrial rolling process and the rolling stock’s chemical composition. The analysis of the spatial influence was performed with CAE neural networks, and included an analysis of the influence of carbon and carbide-forming elements, of manganese and sulphur, copper, tin, aluminium, etc. The results of the analyses revealed a new understanding of the influences, and thus also the possibility to reduce the amount of surface cracking if the chemical concentrations of the elements were to be closer to the exactly determined values, or closer to the more exactly determined ratios.