Replacing primary aluminum with recycled aluminum holds great potential for reducing CO2 emissions and by that further promotes lightweight constructions. The recycling process consumes only a small fraction of the energy required for primary aluminum production. However, the recycling process introduces increased impurity levels, which can significantly degrade the material's performance.[1, 2] Solely relying on experimental evaluation for each aluminum batch would impede economic efficiency. Therefore, integrating recycled aluminum into automotive parts necessitates a digital twin as a decision support system to predict mechanical and corrosive properties. As part of this agenda, understanding the precipitation behaviour in secondary aluminum alloys is crucial, as low levels of impurities already leading to precipitations which could significantly degrade the quality of the alloy. Thermodynamic simulation tools enable a glance into which phases will form to which extend, although these simulations are typically bound to high computational effort. Nevertheless, high throughput thermodynamic simulations of phase fractions for AlSi7Mg0.3 with Fe, Mn and Cu as impurities can be performed. The resulting data is then complemented by tree-based supervised machine learning to surrogate the thermodynamic calculations at low computational costs. This enables a high-volume screening of precipitation behaviour for investigating the complexity of phase forming in secondary AlSi7Mg0.3 alloys with common impurities.