Numerical design methods have the potential to accelerate the development process of alloys significantly. In this contribution, we present a versatile and efficient way to design new alloys by combining models of alloy properties and multicriteria optimization. To this end, we view the alloy design task as an optimization problem where we aim to find the best-performing alloys with regard to specific design criteria and under given constraints. These criteria can be inherent material properties such as phase fractions and transition temperatures or performance indicators of, e.g., strength or processability. The result of such an optimization is a Pareto front, from which compositions can be selected for further experimental investigations.

The CALPHAD method is especially important in alloy design since it allows for the prediction of properties in thermodynamic equilibrium as well as non-equilibrium solidification with high speed and reasonable accuracy. However, in the case of Ni-base superalloys, which are employed as turbine blades in jet engines, predicting the creep performance accurately is crucial for designing alloys that can sustain high temperatures for a long time, thus improving the fuel efficiency and sustainability of the engines. A machine learning approach may more accurately model this complex phenomenon.

We present several case studies on the design of Ni-, CoNi-, and NiAl-based alloys, showcasing the potential and limits of the above design methodology. Ni-base superalloys were designed for maximum creep strength and minimal density using simple performance indicators based on equilibrium phase fractions and compositions. Creep experiments show that such alloys can compete in terms of creep life with state-of-the-art commercial alloys but may not perform as well concerning, e.g., oxidation resistance. Furthermore, we compare the efficacy of a machine learning-based creep model to the simpler performance indicators.

CoNi-base alloys were optimized for high homogeneity in the as-case state to prevent the formation of undesired intermetallic phases in the late stage of solidification during casting. Finally, a eutectic NiAl-(Cr,Mo) alloy with a minimum solidification interval was designed and shown to achieve a higher microstructural homogeneity than its non-optimized counterpart.