CFD modeling of void distribution in two-phase flows

Abstract

Cross-flow boiling on tube/tube-bundles commonly found in horizontal steam generators, kettle reboilers, etc. presents substantial complexity due to the local two-phase flow structures in the presence of phase-change heat transfer, but a modelling capability would be of considerable practical use. To use component-scale ensemble-averaged Eulerian-Eulerian CFD multiple additional closure models are required to define the interactions between bubble-liquid and bubble–bubble interfaces, and the choice of these models inevitably affects the predictions. This thesis reports investigations into these closure models, focusing mainly on the interfacial forces for modeling two-phase adiabatic and subcooled boiling flows, and the effects that they have upon the predictions of the flow, with the objective of attempting to identify so far as possible a "best combination". Detailed CFD investigations on adiabatic and boiling upward two-phase flows in a simple vertical pipe geometry, using the ensemble average Eulerian-Eulerian two-fluid framework of the commercial CFD code STAR-CCM+, with various interfacial force models, are reported. This is done for both adiabatic and subcooled boiling flows. One of the main findings of these studies was the model inadequacies associated with the non-drag interfacial forces that are primarily responsible for the lateral void distributions. Among the interfacial force models investigated for adiabatic cases, the use of the Ishii and Zuber (1979) drag coefficient, the Behzadi et al. (2004) lift coefficient and the Antal et al. (1991) wall lubrication force model combination gives the best predictions for all the primitive variables considered. However, the need for a more appropriate model for the lift force coefficient, along with a matched/paired model for the wall lift/lubrication force, is identified. For the subcooled boiling investigation, a negative lift coefficient (-0.1) in conjunction with the default turbulent dispersion and wall lubrication force models predicted a reasonably accurate axial void distribution, as well as an acceptable velocity profile for both phases, and this is identified as the “best combination”. In addition, many other shortcomings regarding the closures related to the two-fluid framework have been identified and are discussed. Finally, the ‘best combination’ of models has been used successfully to simulate the boiling heat transfer on a single tube under cross-flow, which is a novel application of the Eulerian-Eulerian two-fluid approach in conjunction with the WHFP approach on such flow geometry.