Neural networks relating alloy composition, microstructure, and tensile properties of α/β-processed TIMETAL 6-4

Peter C. Collins, Santhosh Koduri, Brian Welk, Jaimie Tiley, Hamish L. Fraser

Research output: Contribution to journalArticleResearchpeer-review

41 Citations (Scopus)

Abstract

Bayesian neural networks have been developed, which relate composition, microstructure, and tensile properties of the alloy TIMETAL 6-4 (nominal composition: Ti-6Al-4V (wt pct) after thermomechanical processing (TMP) in the two-phase (α + β)-phase field. The developed networks are able to make interpolative predictions of properties within the ranges of composition and microstructural features that are in the population of the database used for training and testing of the networks. In addition, the neural networks have been used to conduct virtual experiments which permit the functional dependencies of properties on composition and microstructural features to be determined. In this way, it is shown that in the microstructural condition resulting from TMP in the two-phase (α + β) phase field, the most significant contribution to strength is from solid solution strengthening, with microstructural features apparently influencing the balance of a number of properties.

Original languageEnglish
Pages (from-to)1441-1453
Number of pages13
JournalMetallurgical and Materials Transactions A: Physical Metallurgy and Materials Science
Volume44
Issue number3
DOIs
Publication statusPublished - Mar 2013
Externally publishedYes

Cite this