Application of Design of Experiments for Alloy Development of an Aluminum Copper Casting Alloy

Abstract

Design of experiments (DoE) based on a linear regression model was used to develop an Aluminum Copper-based casting alloy. The main objectives of the development were the achievement of (1) a high strength at elevated temperatures with (2) a low hot tearing tendency. Within the DoE, 17 different chemical compositions of the newly developed alloy AlCuMnCo(Ni) were cast, tested regarding hot tearing tendency and characterized in tensile tests up to 300 °C. Test results showed that the AlCuMnCo(Ni)-alloys from the DoE have high mechanical properties from ambient temperature up to 300 °C and thus feature a high thermal stability. It was found that the alloying elements Cu and Co increase the yield strength whereas Mn and Ni tend to increase the attainable elongation. Furthermore, some of the alloys showed no or a very low tendency to hot tearing—a remarkable feature for Al-Cu alloys which are otherwise highly susceptible to hot tearing. The regression model that was developed from the test results fulfils a set of quality criteria and is therefore expected to provide reliable predictions. The predictive ability of the model was validated by casting and testing a sweet spot alloy. Results show that the model is sufficient for predicting the mechanical properties from ambient temperature to 250 °C. Furthermore, the sweet spot alloy surpasses the reference alloy AlCuNiCoSbZr (RR30) in its mechanical properties up to 250 °C. It was shown that by applying design of experiments, time and effort for an alloy development can effectively be reduced and simultaneously a high degree of information density about the alloying system considered is generated.