Multi-objective optimization assisting three-dimensional quantitative Cahn-Hilliard simulations of microstructure evolution in Fe-Cr alloys during spinodal decomposition

Phase separation occurring via spinodal decomposition is considered responsible for the "475 degrees C embrittlement" in Fe-Cr alloys. It is thus critical to gain quantitative descriptions of the microstructure evolution in Fe-Cr alloys during spinodal decomposition. However, quantitative in-situ or ex-situ experimental observations of spinodal decomposition processes in Fe-Cr alloys are generally scarce, and most numerical simulations are still not completely quantitative. In this paper, the Cahn-Hilliard simulations regarding spinodal decomposition in Fe-Cr alloys were systematically summarized. We employed the Pareto optimal driven automation framework to perform quantitative three-dimensional Cahn-Hilliard simulations of microstructure evolution in Fe-Cr alloys during spinodal decomposition process. The sampling efficiency of newly developed exploration strategies and different searching algorithms were extensively examined and discussed. The uncertain material/model parameters of the Cahn-Hilliard model were derived by considering multiple characteristic microstructure data. The remarkable consistency between the simulated multiple microstructure characteristics and the experimental observations further validated the generalization ability of the parameters set. It shows a massive potential that the parameters set can quantitatively simulate microstructure evolution in Fe-Cr alloys under various conditions. Furthermore, the capability of the Pareto optimal driven automation framework was reconfirmed by its successful application to Fe-Cr alloys.