Validation and generalisation of hybrid models for flow stress and recrystallisation behaviour of aluminium–magnesium alloys

Numerical models for materials properties prediction require validation to assure the developer and the user that the mechanics and numerical algorithms implemented in the model are correct and consistent with the experimental information available in the literature. Validation of computer models for hot deformation of aluminium–magnesium alloys is described in this paper. The models utilise data-driven neuro-fuzzy models, which describe the relationships between changing process histories and the evolution of the internal state variables comprising dislocation density, subgrain size and subgrain boundary misorientation. These integrated models are further combined with physically based models to give the effects of the internal state variables on the flow stress and recrystallisation behaviour. Genetic Algorithms are used to optimise the overall hybrid-model constants, which are then generalised for a range of aluminium–magnesium alloys of both high and commercial purity. It is shown that this hybrid-modelling methodology supported by a knowledge base offers a flexible way to iteratively develop the microstructrural modelling as more data and better understanding of the evolution of the internal state variables become available.