Velocity measurements in an S-shaped diffusing duct

Abstract

The intake ducts of gas-turbine engines can involve curved ducts and diffusion and the need for information of the flow characteristics, and particularly of the velocity distribution at the compressor face, has led to a number of exprimental investigations and to the desire to develop calculation methods to predict performance. The experiments have made use of small scale ducts and the calculations of numerical solution of differential equations; examples of the former include the measurements of Agrawal et al. (1978), Humphrey et al. (1981), Taylor et al. (1982), Rojas et al. (1983), Vakili et al. (1986) and of the latter the calculations of Humphrey et al. and Vakili et al., Briley and McDonald (1984) and Anderson (1991). The present investigations is concerned with a particular S-shaped diffusing duct designated RAE 2129 and described, for example, by Willimer et al. (1981). Two ducts were devised with the same diffusing characteristics and with intake and outlet displaced by 0.3 and 0.45 times the length. They were intended to served as standard ducts against which others might be compared and wall static and total pressure measurements were recorded by Willimer et al. The total pressure measurements were confined to the engine compressor face (i.e. downstream of the exit of the duct) and it was expected that both ducts would give rise to flow separation and this was confirmed by early calculations of Horton (1989) though with uncertainties associated with the use of a parabolised form of the Navier-Stokes equations and an algebraic eddy-viscosity assumption. Thus, the results of the third section of this paper encompass the entire flow and pay particular attention to the possibility of flow separation. They were obtained for the duct with the smaller offset since this is closer to practice particularly in that any region of separated flow should be comparatively small. Related measurements in S-shaped diffusers have been reported by