HyperCODA for Two-Phase Ejectors

Kurzfassung

Two-phase water ejectors are promising components for High-Temperature Heat Pumps (HTHP) because they can simultaneously pre-pressurize and cool the working fluid, reducing the necessary compressor stages. However, analyzing these devices is challenging due to the distinct interfaces between the liquid, mixture, and vapor phases. A thorough understanding of surface tension, evaporation, and condensation is essential to predict two-phase flow behavior. Furthermore, the reduced speed of sound in two-phase mixture region leads to compressibility effects at relatively low velocities. Gaining insight into these phenomena is crucial to effectively implement these ejectors in industrial processes.

Previously, such simulations were constrained to expensive commercial solvers with limited flexibility. This work builds upon HyperCODA (the Hypersonic extension of the joint ONERA, DLR, and Airbus CODA solver), utilizing its established tabular non-linear thermodynamics capabilities to enable the simulation of two-phase flows under the assumption of thermo-mechanical equilibrium.

In a preprocessing step, a two-phase multi-fluid lookup table is generated using detailed Helmholtz Equation of State (HEOS) using the open source software the CoolProp. To minimize interpolation errors near the phase boundary without inflating the table size, this work updates HyperCODA's lookup table infrastructure to support non-uniform spacing, enabling targeted refinement near the saturation line.

The methodology is then validated against three 1D literature benchmarks - a liquid-liquid n-dodecane shock tube, a liquid-vapor shock tube, and a double-rarefaction expansion tube - demonstrating the correct construction of the tables and the solver's capacity to handle sharp property gradients and phase change.

Lastly, the framework is applied to a two-phase water ejector across two distinct back-pressures. The results are compared against ANSYS Fluent, which employs a five-equation non-thermal equilibrium model within a RANS formulation, coupled with the Lee mass transfer model.