Abstract
This paper shows how hybrid modeling combining a physics-based Precipitation Model and a data-driven Gaussian Process Regression model can be used to predict flow stress curves of different variants of AlMgSi1 alloys. The approach can be a step towards a methodology to manage higher variability in input material, such as remelted contaminated post-consumer aluminum scrap. Data from laboratory compression tests of six different compositional variations of AlMgSi1 with different contents of Si, Cu, and Mg was used. The proposed hybrid model aligns well with experimental results both within the training data range and inputs beyond the training range.