Abstract |
: |
One of the most important efficiency losses in a vaporcompression refrigeration system, especially in a transcritical CO2 cycle, is caused by the throttling process. The most promising solution is an application of a two-phase ejector as a compressor booster. The schematic diagram of this solution is presented in Figure 1. Efficient ejector operation is mostly dependent on ejector geometry, where the converging-diverging motive nozzle plays an important role. In this nozzle, the pressure potential energy of the fluid is partly converted into kinetic energy. The lower pressure at the motive nozzle outlet, the higher motive stream velocity and the better potential of momentum exchange with the secondary stream farther downstream. For given operating conditions (inlet, outlet pressure) the maximum achievable velocity for a simple converging nozzle is fixed by the propagation velocity (the speed of sound). For a converging-diverging nozzle, the velocity at the nozzle outlet could be larger and depends on the divergent design. For two-phase flows this issue is additionally complicated by an occurrence of non-equilibrium effects: although the fluid pressure decreases below the saturation pressure, a phase change process does not occur and the flow remains single-phase (metastable liquid flow). After the pressure reaches an enough low value the phase change (flashing) starts but the formed two-phase flow is still far from equilibrium. In case of the CO2 transcritical cycle, the described flow is even more complicated because of the fluid transition from supercritical to subcritical state. All those phenomena significantly influence the motive mass flow rate, which is a key parameter for an ejector performance. Therefore, an accurate prediction of a local speed of sound in non-equlibrium two-phase flows is a crucial issue for an ejector motive nozzle design. Aforementioned circumstances were a motivation for research of a relatively simple and accurate way to describe the sound speed in two-phase flows. For this purpose, three approaches were analysed and compared, namely: homogenous equilibrium model, relaxation equilibrium model and delayed equilibrium model. Authors found that the absence of equilibrium can be relatively simply included in speed of sound predictions by usage of the relaxation models. Moreover, calculations performed on the base of an own implementation of mentioned models were compared to available experimental data. |