Executive Summary : | Non-Newtonian two-phase flows are crucial in various industrial applications, leading to increased interest in their numerical simulation. However, modeling rheology and interfacial dynamics is challenging due to instabilities, timestep adaptivity, and capturing the sharp phase interface. Current numerical modeling approaches are limited to structured grids, making their extension to irregular geometries difficult. This project proposes a computational framework based on the moment of fluid (MOF) method for non-Newtonian two-phase flows, which can accurately capture interfacial dynamics, alleviate rheological challenges, and efficiently couple complex physical phenomena even in irregular flow domains. A multidimensional geometric advection procedure for phase fraction and centroid, along with an efficient minimization algorithm for interface reconstruction, will be implemented for accurate capturing of the sharp interface on non-orthogonal/unstructured grids. A machine learning (ML)-based technique for curvature estimation will be integrated with the MOF method for accurate surface tension calculations. A non-Newtonian two-phase flow solver will be developed for complex geometries, incorporating different rheology and strategies like the log-conformation technique to mitigate stability issues. The computational development will be done within a parallel framework using OpenMP/MPI to handle the intensive nature of computations. The ML augmented MOF framework is likely the first attempt to study non-Newtonian two-phase flows involving complex geometries, making it a significant step towards understanding these flows. |