NUMERICAL INVESTIGATION OF FLASHING FLOW IN A CONVERGENT-DIVERGENT NOZZLE

In order to reduce CO2 emissions, it is important to utilize renewable energy sources to their fullest. The output of renewable energy sources fluctuates which can be addressed through reliable storage. Liquid air energy storage can be used as an alternative energy storage method. Replacing the Joule-Thompson valve in the liquefaction process with a turbo-expander can increase the round-trip efficiency of the liquid air energy storage system. When the expansion occurs in two-phase conditions, the flow undergoes a flashing expansion with simultaneous acceleration and nucleation of vapor. To perform an efficient design of the turbo-expander, it is imperative to understand the flashing phenomenon and to model it accurately. The main objective of this paper is to identify the correct modeling approach available in commercial computational fluid dynamics software to predict the flashing phenomenon of water. Super Moby Dick and Abuaf nozzles were investigated using 2D Reynolds-averaged Navier-Stokes equations coupled with a two-phase mixture model. The Zwart-Gerber-Belamri model, Schnerr and Sauer model, and Lee model were used to model the interphase mass transfer. The results of the numerical simulations were compared with experimental data in order to evaluate which approach is the most suitable to model the flashing phenomenon. The results indicate that all models are strongly sensitive to the values of the model constants, as they govern both the onset of vapor nucleation and the vaporization rate. Therefore, in order to accurately predict the flashing phenomenon, a tuning of these parameters is needed. Overall, the error in the pressure profile increases as the inlet temperature increases, due to an increasing influence of non-equilibrium effects. Despite the fact that flashing and cavitation appear to have similar macroscopical characteristics, their underlying phase-change mechanisms are different. Additionally, owing to the absence of a wall nucleation model, it is not possible to accurately capture localized flow phenomena.