The Effect the Choice of Turbulent Models Has on the Numerical Simulation Results for Condensation in Channels
摘要
The aim of the study was to develop practical recommendations on selecting the turbulence models to ensure correct description of condensation processes in two-phase flows in different flow regimes. The existing approaches to modeling condensation in channels are analyzed using the volume of fluid (VOF) method. It is shown that the numerical simulation results depend essentially on the choice of turbulence model and the way in which turbulent viscosity is damped near the phase interface boundary. It is noted that application of widely known models, such as the Launder–Sharma k–ε model and the Menter basic SST k–ω model may yield results significantly deviating from experimental data as a consequence of incorrectly described turbulent transfer in a liquid film. The effect the choice of turbulence model has on the accuracy with which the heat transfer during R-134a refrigerant condensation in a microchannel with a diameter of 1 mm is numerically simulated is studied. The calculations were carried out using the ANES CFD code, which implements the VOF method, and two models of mass transfer at the phase interface boundary—a modified Lee model and a model similar to the Tanasawa model. The low Reynolds Menter SST k–ω model (with and without damping) and the Launder–Sharma model are considered. The studies were carried out for mass flux values ranging from 100 to 1000 kg/(m2 s). A comparison with experimental data has shown that the best agreement is reached in the case of using the low Reynolds SST k–ω model with turbulence damping near the phase interface boundary in combination with the Tanasawa model. The average deviation does not exceed 15% and even 10% at high mass flux levels. It has been determined that with taking damping into account, it becomes possible to describe the condensate film laminar flow in a correct manner without the need to artificially set the turbulent viscosity coefficient equal to zero. It is shown that at low mass flux levels, the best coincidence with experimental data is ensured in stating the problem with a specified heat flux, whereas at high mass flux levels the best correspondence to experimental data is reached when specifying a constant wall temperature. The obtained results can be used during practical simulation of condensation processes in microchannels.