Hardenability modelling and the performance of computational intelligence tools in research on structural alloy steels

The paper presents and discusses various computational tools for modelling hardenability, including multiple regression analysis, neural networks, fuzzy systems, and genetic programming. The research aims to demonstrate the application of these tools for understanding and predicting the hardenability of steel based on factors such as chemical composition and cooling properties. Design/methodology/approach The objectives of the research were achieved through the application of multiple regression analysis, neural networks, fuzzy systems, and genetic programming. The paper describes the preliminary results of genetic programming to calculate the maximum hardness based on carbon concentration. In addition, the study uses the cooling time from 800 to 500 °C (t8/5) as a relevant parameter for quenching and includes it in the mathematical models for steel hardening. The developed models, including neural networks and fuzzy logic, are used for simulation studies on the influence of alloying elements on the hardenability of steel. Findings The research results demonstrate the successful application of computational tools, including genetic programming, neural networks, and fuzzy systems, in modelling and predicting the hardenability of steel. The paper presents equations and models that enable the calculation of the Jominy curve for steel within certain chemical composition ranges. The results show promising results for predicting hardness distribution and structural transformations in quenched steel samples. Research limitations/implications While the study demonstrates the potential of computational tools for modelling hardenability, it also acknowledges limitations. The models presented are limited to specific mass concentrations of alloying elements, and further research is required to extend their applicability to a wider range of compositions. The limitations highlight opportunities for future research to refine and improve the proposed models. Practical implications The research findings bear practical significance in materials science as they provide tools for predicting hardness, structural transformations, and generating stresses and strains in steel alloys. The models developed particularly the easy-to-implement multiple regression model, can be widely used in industry. Practical applications include simulations of the effects of alloying elements on hardenability, which help optimise steel manufacturing processes. Originality/value The original value of the work lies in the comprehensive research and application of various computational tools for modelling the hardenability of steel alloys. Genetic programming, neural networks, and fuzzy systems provide novel approaches to understanding and predicting steel properties. The practical value extends to the industries involved in steel production and provides valuable tools for optimising processes and predicting material behaviour.