Development Of Numerical Modeling For Gas-Liquid Two-Phase Flows Based On Unstructured Grids And Parallel Computing

Gas-liquid two phase flow phenomena are present extensively in industries of energy, nuclear, oil, and chemistry. Research on two phase flows is meaningful to make guideline for industry design. There are a lot of experimental and theoretical researches in the literature. However, numerical studies are less published and reports in this area are mainly focused on applying the commercial software or within the framework of structured grids which has limited applications for two phase flows. The present thesis focuses on the development of numerical methods for gas-liquid two phase flows with unstructured grids.Based on the experiences in Computational Fluid Dynamics (CFD), the linear equation solver takes more than 60% CPU time of the whole process, therefore acceleration of the linear equation solver is a key element in enhancing the whole performance of CFD. An agglomeration algebraic multi-grid method is developed and implemented in the solver which accelerates the linear equation set solution. Another advantage of this linear equation solver is that it is convenient for parallel computing, which is also the purpose of this research.Numerical algorithm for general transport equation based on unstructured grid is presented, which includes discretization, boundary condition and velocity-pressure coupling. Some factors influencing the face flux calculation are also discussed in detail. Couple solver of multi-material is constructed and its implement is also given.An N-S equation solver based on unstructured grids is developed. The solver is validated in a number of classical flow problems and the results agree well with other experimental data in literature. Comparison of present solver with commercial software Fluent shows that the developed solver works well.Two-fluid method of gas-liquid two phase flows is discussed in detail, including turbulent enclosure model, interfacial force model and phase change model. Meanwhile the axi-symmetric model has been implemented into the solver and used to save resources and enhance the performance for axi-symmetric flows. In order to simulate detailed large scale industrial problems, the parallel computing method based on MPI has been developed and ideal speedup is achieved.