Physical Based Models in Combination with Artificial Intelligence as a Basis for Computational Alloy DesignWednesday (06.11.2019) 16:40 - 17:00 Part of:
Future design of new alloys will to an increasing extent depend on advanced computer simulations. This is particularly the case in applications where the products must offer a variety of properties where some are increasingly demanding and some are often considered incompatible like for instance strength and ductility. In addition, the products need to meet the highest possible production speed. In production of aluminium products such as extruded products, it is well known that each operation along the process chain affects the properties of the final product. Equally important is the chemical composition window of the alloys which may become larger when more post-consumed scrap is being used. It is therefore obvious that the complexity soon becomes impossible to handle by traditional methods.
The present article describes how computational alloy design is about to become an established methodology for alloy design in Hydro. The basis for the methodology is advanced physical based models with a high predictive power, which are coupled by an optimization tool to allow fully automatic through process simulations. The physical based models predict the evolution of the microstructure in terms of primary particles, dispersoids, precipitates, dislocation sub-structures, grains, as well as concentration of elements in solid solution. Dislocation mechanics is then used to convert the predicted structure parameters to mechanical properties like yield strength and work hardening, which in turn, are input to the final product simulations. The optimization tool runs the models in iterations until user-defined acceptance levels on the combination of properties, are obtained. In the present paper, examples will be shown on applications of this software tool, named PRO 3 TM, for various aluminium products.