Abstract
Insights into the yield strength of metallic alloys have recently been revealed in terms of microstructure entropy (ME) per grain S* [Acta Materialia 195 (2020) 531–540]. Here, we report the decay of ME, which is taken as a phenomenological parameter, in steel alloy systems characterized by a bimodal grain size distribution and a combination of high-strength and high-ductility as the dynamics of the grain growth approaches a self-similar regime. The self-similar bimodal grain distribution was realized by an innovative phase reversion strategy of the 60 groups steel alloys. An increase in tann and/or Tann resulted in an increase in S* and a decay in the total entropy, Sm, due to the more significant effect of the grain number decreases than that of the S* growth. The Sm, the average grain perimeter (p¯), and the average grain volume (v¯) exhibited close relationships with the mean grain surface area (a¯) during the grain growths and the ME degenerations. The decay of ME, caused by the system approaching a self-similar regime during the grain growth, is well explained following the first and second laws of thermodynamics, which has further been studied via machine learning. Three linear regression models were established with high prediction accuracy. Using these well trained machine learning models, with any continuous or random Ln (a¯, μm2) as the input, we can predict the Ln (Sm, μm−3), Ln (p¯, μm), and Ln (v¯, μm3), which are related to the microstructure and properties of different materials, within millisecond.
Accepted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call