Abstract

In order to qualify CFD codes for accurate numerical predictions of transient evolution of flow regimes in a vertical gas-liquid two-phase flow, suitable closure models are needed. The current study focuses on detailed numerical investigation of the interfacial driving force models and assessment of two population balance model approaches viz. the MUltiple-Size-Group (MUSIG) and one-group Interfacial Area Transport Equation (lATE) using the two-fluid modelling approach. Numerical predictions of five primitive variables: gas volume fraction, interfacial area concentration, Sauter mean bubble diameter, gas velocity and liquid velocity; have been validated against experimental data of Monros et al., (2013). Three specific objectives have been completed in this study. Firstly, under the assumption of mono-disperse bubbles, a consistent set of interfacial force models have been investigated. The effect of drag, lift, wall lubrication and turbulent dispersion forces has been assessed. New parameters have been introduced in the wall lubrication force models of Antal et al., (1991) and Frank et al., (2004, 2008) as well as implementing additional drag coefficient models using CFX Expression Language (CEl). The Tomiyama, (1998) lift coefficient model has been modified in this study. In general, the predictions from the sets of interfacial force models yielded satisfactory agreement with the experimental data. A set of Grace drag coefficient model, Tomiyama lift coefficient model, Antal wall force model, and Favre averaged turbulent dispersion force were found to provide the best agreement with the experimental data. Secondly, a model validation study to assess the performance of existing coalescence and breakup models of the MUSIG model in simulating bubbly flow in vertical configuration has been conducted. The breakup model of Luo and Svendsen, (1996) and coalescence model of Prince and Blanch, (1990) have been implemented. Detailed analysis has been performed for the wall lubrication and turbulent dispersion force models. Overall, the comparison has shown that the MUSIG model yielded satisfactory agreement with the experimental data. The transition from wall peak to core peak gas volume fraction profiles has been successfully captured and encouraging results clearly exemplified the capability of the model in capturing the dynamical changes of bubble size due to coalescence and breakup processes. The observed agreement with the gas volume fraction profiles indicates a level of confidence in the interfacial force models (especially the lift coefficient model) used, while on the other hand, the agreement seen on the interfacial area concentration indicates that the birth and death processes modelled are reasonably adequate to describe the bubble dynamics. Nevertheless, noticeable discrepancies in simulating bimodal bubble size distributions were found revealing the plausible imperfection of existing coalescence and breakup kernels. Thirdly, the performance of the interfacial area transport equation has been assessed by implementing some typical constitutive models for bubble coalescence and breakup taken from literature. The interfacial area transport equation has been successfully implemented into the CFD code and the constitutive model formulations from Hibiki and Ishii (2000a), Wu et al. (1998) with coefficients from Ishii and Kim (2001) and Wang (2010) have been implemented and validated. The combined effects of lift and wall lubrication force have been investigated. Both models of the lATE source and sink terms were able to reasonably capture the experimental data especially the gas volume fraction profiles. More experimental and theoretical work needs to be done in this field to increase the prediction capability of the simulation tools regarding the distribution of the phases along the pipe radius. (au)