Subsonic intake duct flows
Here both S-shaped and singly curved (here classified as S-shaped) duct diffusers for intakes in aeronautical propulsion systems are studied. The results are applicable in other situations where similar ducts occur; for example on V/STOL aircraft employing re-direction of thrust, intercomponent ducting in high bypass ratio engines, etc. An open circuit static test rig, capable of mass flow rates of 5 kg/s, and three-dimensional instrumentation were established. Flow measurements were made in S-shaped intake duct diffusers for rear mounted gas turbine engines in both aircraft and air-breathing missiles. These designs are intended for ventral type inlet installation. These ducts possess cross-sectional shape transitions, from oblate to circular, with area increase and annular ducts at the engine face. The work was aimed at both fundamental understanding of the flows and at establishing test data for the prediction methods. Tests were performed at throat Mach numbers of nominally 0.15 and 0.6 and in the unit Reynolds number range of 3x10_6/m - 2x10_7/m for three different ducts each having different upstream bends but common downstream bends. Detailed boundary layer surveys were made to establish plane of symmetry growth of the viscous region and the extent of three-dimensionality away from the plane of symmetry. Data are presented in the form of velocity profiles, streamwise and cross-flow, integral thicknesses and surface pressure fields. Engine face distortion is assessed from full outlet flow surveys. Flow visualization was recorded using surface oil flow techniques. Evidence is presented of a trend towards three-dimensional separation as the upstream bend increases in severity. For the most extreme case large regions of complex three-dimensional separated flow occur and topological analysis of the recorded surface oil flow pattern allows reconstruction of the separating flow. Clear correlations are established between flow visualization results and flow measurements yielding better understanding. Finally, results were compared with a three-dimensional compressible prediction method.