High Fidelity Cfd-Trained Machine Learning To Inform Rans-Modelled Interfacial Turbulence

In aero-engine bearing chambers, two-phase shearing flows are difficult to predict as Computational Fluid Dynamics (CFD) RANS models tend overestimate interfacial turbulence levels, leading inaccuracies in the modelling of flow. Turbulence damping methods have been developed address this problem, such Egorov’s correction, however, method is mesh dependent and results differ considerably according choice coefficient. addition, approach assumes a smooth interface between air oil phases when reality they wavy. paper, Machine Learning used inform an unsteady modelling. It trained using high fidelity quasi-DNS simulation data provide appropriate correction popular Wilcox’s standard