Investigation of radial flow ejector performance through experiments and computational simulations

Investigation of radial flow ejector performance through
experiments and computational simulations

Adjustability in ejector geometry is desirable to extend the operating range of ejectors at different working conditions. Current approaches to geometric adjustability in the throat sizes of conventional axial ejectors have the disadvantage of blocking the flow path by positioning additional components in the high-speed flow. Furthermore, a viable mechanical system is not available for altering the diameters of the nozzle and ejector throat while providing gas-tight seals and a smooth surface throughout. An alternative concept, the radial ejector, has been created which makes possible a practical ejector that can readily achieve variable geometry without introducing additional flow blockage in the center of the high-speed flow. Such a radial configuration allows geometry adjustment during operation to achieve optimum performance at a range of different operating conditions without additional pressure losses from high-speed flow blockage. The radial ejector conceived, designed, commissioned and evaluated in this thesis was formed from two disk-like surfaces for both the primary nozzle and the ejector body enabling the radial ejector to operate with different flow areas by simply changing the separation of the ejector duct walls or the nozzle plates. Conventional axial flow ejector design procedures were adapted in the design process, and benchmarking against an experimental axial ejector was also performed. Air was employed as the working fluid to improve fabrication options and allow experiments to be performed with an open system. The radial ejector has a nozzle throat area of 8.8 mm2, nozzle exit area of 180 mm2, giving a nozzle area ratio of 20.4, and an ejector physical throat area of 520 mm2, giving the ejector area ratio of 59. Experimental results show that the radial ejector produced entrainment ratios between 0.95 and 0.24 and critical pressure lift ratios around 1.5 for expansion ratios between 50 and 139. The relationships between the entrainment ratio and critical exit pressure and primary, secondary and exit pressures were similar to conventional axial ejectors. Similarly, trends observed in the measurements of wall pressure for the radial ejector configuration were generally consistent with those for axial flow ejectors. Based on the experimental data from the critical mode ejector operation and based on an isentropic flow calculation, a secondary stream Mach number of around 0.7 was determined at the physical throat of the ejector. When ejector operation transitioned from the critical to the subcritical mode, wall pressures in the throat and at locations upstream of the throat increased, leading to a peak in pressure prior to the final pressure rise in the diffuser. Comparing the experimental results with the simulations show that the entrainment ratio achieved from the radial ejector prototype agreed well. The entrainment ratio performance also closely matched that of a quasi-one-dimensional gas dynamic model with an error level less than 10%. Computational Fluid Dynamic (CFD) analysis based on the k-epsilon standard turbulence model showed that simulations of entrainment ratio and critical back pressure were in reasonable agreement with the experimental results with an average discrepancy of less than 16%. Using the k-omega SST turbulence model, it was demonstrated that adjustability in the radial ejector is viable and by increasing the separation of the ejector duct walls from 2.2 mm to 3 mm, an increase of 34% in entrainment ratio can be achieved. A critical back pressure increase of 40% was achieved by reducing the separation of the ejector duct walls from 3 mm to 2.2 mm. However, as there are systematic differences between the measurements and the computational simulations using both the k-omega SST or the k-epsilon standard model, the overall reliability of the CFD simulations is questionable. The main issue for the current prototype radial ejector is the low critical exit pressure relative to expected performance from an equivalent axial flow ejector. More experiments and simulation are required to improve this aspect of the radial ejector performance. The radial ejector geometry is required to be optimized and many different flow features need extensive investigation to identify ways to achieve better performance and expand the adjustability options in the radial ejector.