CFD Simulations of Multiphase Flows with Particles
MetadataVis full innførsel
This Master s thesis describes two CFD studies particle plug propagation in bent pipes and critical velocity determination in dilute slurry flows. The objective of this study is to extend an understanding of these phenomena and to test the capabilities of the CFD software ANSYS Fluent in modelling of particle flows. In addition, contribution to 1D modelling of particle flows is considered as another important goal of this project. The particle plug propagation is simulated by Eulerian Granular model while the critical velocity determination is simulated using Discrete Phase Model (DPM). The obtained results regarding the propagating particle plug show that Eulerian Granular model in ANSYS Fluent is capable to reproduce the particle plug movement in pipes, specifically the dispersion and transition zones are well captured and show a good qualitative fit with the experimental measurements. The detailed model selection study is conducted which could be potentially used in similar research works. In addition, a slip relation for 1D particle modelling is proposed which is capable to reproduce the particle dispersion zone at the plug front in an accurate way. DPM model in ANSYS Fluent demonstrated a reasonably good performance in predicting critical velocity in slurry flows, but the value of the velocity is underestimated. Also, it is obtained that the critical velocity is independent on the particle volume fraction which is not the same as in the experimental observations. The obtained results need more investigations with regards to mesh dependency and missing physics in the default DPM model. Further work might address to predicting the value of the particle volume fraction in the dispersion and transition zones of the particle plug using different specularity coefficients. It could useful to obtain a grid independent solution in terms of the value of the first layer thickness. The proposed slip relation for 1D modelling might be further improved and generalized if all the parameters are described by a single variable, for example, average particle fraction. As for the critical velocity modelling, the better predictions might be obtained if important factors such as particle-particle interaction and Shields parameter are included in the simulations.