No abstract

No abstract

A computational study of an air-air ejector system, utilizing a primary S-bend transition duct, was compared with experimental results. Two primary ducts, differing in offset, consisted of an annular-to-circular-to-oblong transition which incorporated a total area increase of 62.4% with the duct performing 16% diffusion. The ducts were analyzed both alone and in ejector configuration under varying degrees of inlet swirl. The ejector geometry consisted of the duct with a downstream mixing tube. Several mixing tubes, of oblong cross sectional shape, differing in both length and area, were tested in various parametric configurations. Ejector performance was established on the basis of pumping capability, duct back pressure, and outlet effective area. Experimental work commenced on a cold flow test rig, with duct inlet conditions being measured with four 3-hole pitot probes. The duct outlet profiles were measured using a 7-hole probe which traversed the entire exit area. Three conditions of inlet swirl were analyzed: 0°, 20° and 40°. Experimental results showed an increase in pumping performance with increased inlet swirl, mixing tube length, area ratio and standoff. An optimum standoff value of 0.25Dh2 was observed. CFD simulations were based on experimental mass flow inlet conditions utilizing the realizable k-ε turbulence model. CFD results showed that the realizable k-ε turbulence model was quite capable of modeling the complex flow for the associated geometry, and correctly predicted flow features as well as performance trends for all geometrical configurations tested. However, the CFD was unable to properly predict the duct inlet static pressure leading to erroneous back pressure results.

In some gas turbine applications, it is desirable to redirect the exhaust flow through 90 degrees and mix this flow with the ambient air for the purposes of structural integrity and heat signature suppression. A method to achieve this is to transform the flow from a circular profile to a rectangular slot of high aspect ratio. The increase in wetted perimeter allows for greater mixing with the ambient air; however the shape of such a duct causes significant amounts of flow distortion and poor pressure recovery. This paper presents preliminary experimental results of the performance of such a duct and discusses the ability of a commercial CFD software package to numerically predict this performance. Significant crossflows and reversed flows were observed at the duct outlet leading to inefficient use of the outlet area, high back pressure and consequently a high loss coefficient. These trends are exacerbated with an increasing inlet swirl angle. The preliminary numerical predictions captured the general trends of the flow but could not capture the extent of the reversed flow, leading to over-prediction of the effective area ratio, E, and under-prediction of the loss coefficient, k.

An experimental study has been undertaken on a 3-stage entraining diffuser with a distorted inlet flow. Two different circular-to-slot transition ducts were used as driving nozzles. Both transition ducts included a 90 degree bend. Varying degrees of inlet swirl were also considered. A 7-hole pressure probe was used to traverse the diffuser outlets. With a longitudinal duct the measured flows showed severe impingement of the primary flow on the walls of the entraining diffuser. Similar outlet flow distributions were observed for all cases of inlet swirl considered. In contrast, only minimal primary flow impingement was observed with a transverse duct. The use of the transverse duct also resulted in significant secondary flows including multiple large-scale vortices. These secondary flows intensified when swirl was added. Four performance parameters were calculated: the entrainment ratio, back pressure coefficient, pressure recovery coefficient and the entraining diffuser efficiency. The results showed that peak performance was attained at 20° of inlet swirl for both configurations. The addition of an entraining diffuser was found to result in minimal improvements of performance for the longitudinal duct configuration and significant improvements for the transverse orientation.