Large Eddy Simulation of a supersonic air ejector

The supersonic ejector has attracted considerable research attention as a potential candidate for reducing compressor load in refrigeration systems, among other applications. This can be attributed to its relatively simple design that does not involve moving parts. The primary flow in supersonic ejectors is accelerated through a converging-diverging nozzle, and the resulting supersonic jet get discharged into a mixing chamber, where it entrains and compresses a secondary flow. The performance of these ejectors is often evaluated based on their entrainment ratio and compression ratio. These parameters have a significant influence on the overall efficiency of the refrigeration cycle, and their optimization is vital in improving the system performance.

Applying data-driven models like artificial neural networks can improve the design and optimization of both single-phase and condensing steam supersonic ejectors. However, extrapolating these tools outside their training range is difficult and challenging. Additionally, they fail to provide vital information about the underlying physics of the ejector systems. On the other hand, different wall pressure measurements and flow visualization techniques have been used to study the influence and relationship between different parameters, including entrainment ratio, compression ratio, ejector geometry and specific flow characteristics.

Recently, Large Eddy Simulation (LES) has been proposed as an effective modeling approach for overcoming the limitations of the existing methods. LES has been successfully used in different applications involving supersonic jects, such as supersonic mixing layers and acoustic wave propagation of supersonic jets. Unfortunately, there are very few studies that employ LES to analyze supersonic ejectors. This requires a thorough understanding of the flow topology of the supersonic ejectors.

On this account, Canadian scientists: Dr. Sergio Croquer, Professor Sébastien Poncet and Professor Stéphane Moreau from Université de Sherbrooke in collaboration with Dr. Olivier Lamberts and Professor Yann Bartosiewicz from Université Catholique de Louvain in Belgium studied the flow topology and mixing layer characteristics of a supersonic ejector using LES. Subsequently, a specialized finite-element code was adopted to model a supersonic air ejector of squared crossed-section. The device and the related flow characteristics were further described and discussed. The results were presented in terms of instantaneous structures and time-averaged profiles in the mixing layer, while the flow features were determined via instantaneous temperature fields and pressure profiles of the device. Their work is currently published in the research journal, Applied Thermal Engineering.

The research team showed that under the provided conditions, the mixing layer was initially laminar before transitioning towards turbulence characterized by ? vortices in the first quarter of the mixing chamber. These vortices evolved into hairpin vortices before breaking down at about half the mixing chamber. The time-averaged primary jet velocity profiles exhibited self-similarity before the first half of the mixing chamber. In Comparison with the unconfined mixing layer, the supersonic ejector mixing layer grew slower initially before developing a similar rate after the transition region. A shock train was observed towards the end of the mixing chamber and its interaction with the recirculation zone in the diffuser contributed to narrowing the flow passage, breaking vertical flow symmetry, and enhancing mixing.

In a nutshell, this is a pioneer application of LES to study the flow topology in the mixing chamber of a supersonic air ejector. A good agreement, in terms of wall pressure measurements and primary jet shock cell structures, was obtained when the results were compared with the experimental data. The results showed the potential of LES in detailed analysis of supersonic ejectors in terms of visualization, quantification and flow description. For instance, it can fully resolve the energy transport in large turbulent structures. In a statement to Advances in Engineering, Dr. Sergio Croquer explained their new findings would contribute to the generalization of LES applications in industrial ejectors.