Towards Improved Reliability of the Analysis of Factors Influencing the Properties on Steel in Industrial Practice

Abstract

Due to the large number of parameters that influence the properties of steel and lower accuracy of some measured data, it can take several years for industry to collect a database large enough to carry out reliable analysis. In this paper a new approach is presented to overcome these problems. The Latin hypercube sampling technique (LHS) was used for modelling of the uncertainty of measured industrial data and CAE NN was used for analysis of the mutual dependence of influencing parameters. Using the example of AISI D2 tool steel, the proposed method was applied for determination of relationships between chemical composition and yield δ by considering the corresponding coefficients of variation. New insights of relationships were directly applicable in the industrial practice. Despite using small database, it was possible to considerably increase the yield δ for industrial hot rolling.