ABSTRACT
The mechanical properties of steel depend strongly on chemical composition and the parameters used during thermomechanical processing. Understanding how each variable affects such properties is indispensable for obtaining high-quality steel products at a lower cost. However, the large number of variables involved in the manufacturing process makes this a difficult task. It is possible to use statistical tools combined with predictive modelling to identify the most relevant parameters and to create a mathematical function that can adequately describe the mechanical properties of the rebar from the selected input–output pairs. In the present work, information about the chemical composition and the thermomechanical processing variables were collected at steel mills and used to predict the mechanical properties of rebar using linear regression analysis and an artificial neural networks approach. The coefficient of determination between the measured and estimated values was calculated for yield strength (YS), ultimate tensile strength (UTS), UTS/YS ratio, and percent elongation. The estimation performed using the artificial neural network was better than the one calculated by linear regression analysis for all four properties studied. The results showed that an artificial neural network can be useful in evaluating and choosing the most adequate parameters to achieve the desired steel properties.
Acknowledgments
We gratefully acknowledge the financial support of the Brazilian agencies CAPES, FUNCAP and CNPq. Authors also gratefully acknowledge fruitful discussions with Prof. Hamilton Ferreira Gomes de Abreu.
Disclosure statement
No potential conflict of interest was reported by the author(s).