Back to overview

Plenary Lecture

On the design of high strength aluminium alloys

Tuesday (05.11.2019)
12:10 - 12:50
Part of:

Design of aluminium alloys is complex. Alloy properties are closely linked to composition, and in the case of high strength aluminium alloys, the careful addition of as many as 10 (ten) alloying elements is typical. Additional complications include factors that range from the safety of alloying elements, to processing and solidification constraints. The design of aluminium alloys is presently difficult to achieve from first principles calculations and thermodynamic analyses. In an attempt to emulate the knowledge embodied in the portfolio of aluminium alloys to date, a machine learning approach is demonstrated for the design of high strength aluminium alloys via deep learning. For a given (target) yield strength, a model based on deep learning predicts the chemical composition and the temper of an aluminium alloy required to achieve a target yield strength. Unique alloys produced from such predictions are presented and discussed. The outcomes of this approach (including the challenges and prospects), along with attendant discussion and factors for general consideration are elaborated.

Prof. Dr. Nick Birbilis
The Australian National University