Modeling of Total Decarburization of Spring Steel with Genetic Programming

Abstract

Store Steel Ltd. is one of the biggest spring steel producers in Europe. Spring steel should have proper chemical composition and microstructure and should be without surface defects. Decarburization—that is, reduction of carbon content—also influences spring steel surface quality. During regular production, the data regarding rolled spring steel bars (width, reduction rate, chemical composition, and total decarburization) and the heating furnace for heating billets before rolling (heating temperature, time, and oxygen content) were monitored. On the basis of the monitored data, a mathematical model for the total decarburization depth was developed by genetic programming and linear regression. The average relative deviations from experimental data for the genetic programming developed and linear regression models are 18.186% and 22.999%, respectively. According to the developed models, the heating temperature was lowered, and, consequently, 24.29% lower total decarburization (t-test, p < 0.05) and also 20.65% lower heating furnace natural gas consumption was achieved (t-test, p < 0.05).

Keywords:

• Decarburization
• Genetic programming
• Linear regression
• Modeling
• Spring steel
• Surface quality