Sensitivity Analysis of Phase Change Modelling for Non-Ideal Fluids in Turbomachinery Applications

Investigating phase change phenomena with numerical methods reveals a strong dependence on the selected model equations. The phase change from gas to liquid, or condensation, is primarily governed by nucleation and droplet growth. Numerous model equations for both processes exist in the literature, typically based on the law of ideal gases and empirical factors historically fitted for water. Despite this fluid-specific calibration, results show significant model sensitivity even for water. This effect increases for other fluids, such as CO2, which exhibit strong non-ideal thermophysical properties. To enable reliable modelling for such fluids, it is essential to first identify which model parameters exert dominant influence. This study contributes by systematically varying model approaches and parameters. For both water and CO2, a test case of supersonic flow through a Laval nozzle is considered. The sensitivity study examines variations in the calculation of the critical energy barrier for nucleation, droplet growth rate, and type of modelling the droplet size distribution. A comparison of two fundamentally different numerical schemes further distinguishes the influence of physical modelling from that of numerical variability. Results for CO2 show a sensitivity to droplet size distribution modelling comparable to water, but with a broader spread for phase change modelling due to fluid non-idealities. These findings support a systematic uncertainty estimate for phase change modelling in turbomachinery operated with CO2 and point to the need for experimental validation of condensation models in non-ideal fluids.