Studies on Mixing Characteristics in the Critical Flow Regime of Supersonic Ejectors

Abstract

Ejectors are passive devices having ubiquitous applications in several engineering domains, including modern-day applications in supercritical CO2-based power and refrigeration cycles. The ejector compresses a low-enthalpy fluid (secondary flow) through gasdynamic interactions with a co-flowing high-enthalpy fluid (primary flow). The ejector operates in the critical mode when the secondary flow is choked aerodynamically, and the entrainment ratio (ER, ratio of secondary to primary mass flow rate) becomes independent of the compression ratio (CR, ratio of the ejector exit pressure to the secondary flow stagnation pressure). In the mixed mode of operation, the ER decreases with CR. Applications of ejectors in energy conversion systems prefer the critical mode of operation. Ejectors have been studied using experimental, analytical, and computational tools with an emphasis on evaluating performance parameters. The gasdynamic mixing which drives the performance of the ejector is not well understood. An optical diagnostic evaluation of mixing in the ejector has been performed for the mixed mode of operation. The mixing characteristics and flow structures inside an ejector are not reported for the critical mode operation, which is the prime motive of this work. With an aim to achieve the critical flow regime, eight mixing duct geometries of different lengths and heights and three supersonic nozzles are designed and fabricated. Experiments are conducted in the supersonic ejector facility at Laboratory for Hypersonic and Shockwave Research, Indian Institute of Science, Bengaluru. Mass flow rate measurements yield the ER, and the static pressure profile indicates the mixing progress and compression. High-speed schlieren images are captured to observe unsteady flow features. Mixing characteristics in terms of non-mixed length (Lnm) is quantified using the planar laser Mie-scattering technique. Modal analyses (proper orthogonal decomposition and dynamic mode decomposition) are implemented on schlieren images to determine the spatial modes and associated frequencies of the re-compression shock structures observed in the diffuser. Additionally, Fourier spectrum analysis and continuous wavelet transformation are performed on the pressure signals. The designed ejectors operate in the critical flow regime, which is confirmed by experimentally measured ER response with CR. Lnm in the critical flow regime is 55% higher than in the mixed flow regime. The ER remains unchanged for mixing duct length less than Lnm; after that, it increases until L/H (mixing duct length to height ratio) of 15 and then decreases. There is a maximum area ratio corresponding to maximum ER for a given primary nozzle, beyond which ER decreases. The oscillation of re-compression shock structures is multimodal, with frequencies ranging between 100 – 250 Hz. A new non-dimensional frequency scaling of the dominant re-compression shock frequency is proposed, which is constant at 4.12+/-18. An artificial neural network (ANN) model is developed with a topology of seven input parameters representing the geometry, operating conditions, and working fluid characteristics for two output parameters of entrainment ratio (ER) and the operational regime (OR) of the ejector. The trained ANN model predicts the ER with +/-10% error and classifies the operating regime with 91% accuracy.