[SY-F2] Symposium F-2

[SY-F2] The use of thermodynamic tensor models for phase-field simulations of spinodal decomposition in quaternary alloys coupled with CALPHAD data.

^○Yuri Amorim Coutinho¹, Nico Vervliet², Lieven De Lathauwer^2,3, Nele Moelans¹ (1.Dept. of Materials Engineering, KU Leuven, Belgium, 2.Dept. of Electrical Engineering, KU Leuven, Belgium, 3.Group Science, Engineering and Technology, KU Leuven - Kulak, Belgium)

The successful coupling of the phase-field and CALPHAD methods for a thermodynamic consistent description of the system free energy in a phase-field model is challenging for multicomponent alloys. The many coupling schemes presented in the literature all tend to suffer from inefficiencies and limitations when applied to higher-order systems. When collecting calculated thermodynamic data in a multiway array or a tensor, we can observe that the number of entries grows exponentially as a function of the number of components and thus the direct use of tensors to provide thermodynamic information to a phase-field simulation is not feasible for quaternary or higher order systems. However, when a canonical polyadic decomposition is applied, we can represent the data contained in the tensor with a small number of coefficients, which grows only linearly if new components are included. This approach allows the construction of thermodynamic tensor models, which can be efficiently used to approximate individual entries of the original tensor with good accuracy. Furthermore, the gains in data reduction obtained with the tensor decomposition technique increases when more elements are considered in the simulations. The efficiency of this novel coupling scheme is verified with spinodal decomposition simulations of quaternary alloys.