Loads have been measured on wedges placed symmetrically in supersonic jets of air. The jets were created by a nozzle with a radially divergent exit section having a lip Mach number of 2.2. The underexpansion ratio was varied from 1 to 2.2 and the distance between the nozzle exit section and the wedge apex was varied from 0 to 2 nozzle exit diameters. All wedges had a base width equal to the nozzle exit diameter: their total included apex angles covered the range 30° to 180°. Pressures were measured on the front faces and the bases for three of the wedges, hence enabling individual contributions to the overall force to be evaluated. Overall loads were measured by means of strain-gauged supports for all six wedges. It was found that the overall load coefficient is only weakly dependent on underexpansion ratio and wedge location but depends strongly on wedge angle. The maximum load coefficient recorded corresponded to 73% of the jet momentum. The base pressures contribute up to 59% of the overall load on a 45° wedge but rather less for larger wedge angles.

The paper presents the results of measurement of the static pressure, total pressure and swirl in the flow through an S-shaped duct of typical air intake proportions mounted in a wind tunnel and tested at different incidences and different through-flow ratios. In order to reduce the magnitude of swirl at high incidence, two methods have been studied - one, to change the distribution of pressures by means of a spoiler and two, to re-energize the separated flow with an inflow of free stream air through auxiliary inlets. Measurements were also made, at 0° incidence, with a perforated spoiler intended to simulate roughly the taking in of low energy flow from the fuselage boundary layer. The results show a small degree of swirl at low incidence which takes the classical pattern of two contra-rotating flows, and a large degree of swirl at high incidence, in the form of a single rotating flow. Of the anti-swirl devices, the spoiler is the more powerful and can be sized either to reverse the swirl direction or to eliminate the swirl completely. A parameter of swirl coefficient, SC60, has been suggested. Values of SC60 at 30° incidence are 0.188 for the duct as designed, 0.068 for the arrangement with auxiliary inflow and −0.039 with a solid spoiler of width 0.15 times throat width. The various arrangements tested, together with results of an earlier study, shed useful light on the general nature of swirl in an S-duct, its method of generation and its final form and magnitude. A further experiment will be made on a duct having both a horizontal and a vertical offset.

As part of a series of investigations of the swirl in S-ducts, the present paper gives the results of tests on a duct with both horizontal and vertical offsets, for comparison with earlier work on ducts with horizontal offset only. Results obtained bear a similarity to those of Reference 2, for which the duct had the same cross-sectional shapes and areas, but the addition of a vertical offset magnifies the resultant swirl considerably. As before, the swirl is found to be sensitive to entry conditions, in that a solid spoiler, blocking off 15% of entry width from the inside wall, reverses the sense of the final swirl. Taken overall, the results throw additional light on the nature of the swirl problem. A particularly interesting and potentially important correlation is obtained between the final swirl (in magnitude and sense of rotation) and a pressure difference between the top and bottom walls of the duct, taken in the end plane of the second bend.

Consideration is given to the relationship H1 = f(H) linking the common shape factor H and the mass-flow shape parameter H1 which is used in entrainment models of boundary-layer development. A formula suggested by Green et al is found to be most nearly consistent with the measurements presented. However, a more exact prediction of H1 is obtained by introducing a factor involving the Reynolds number based on the local momentum thickness θ; thus H1 = f(H, Reθ). Predictions obtained by incorporating the appropriately modified entrainment equation into the well-known method of Green et al prove not to give an improved representation of the development of boundary layers studied experimentally by the authors and others. It is concluded that the modified formula for H1 is primarily useful in giving an improved specification of the overall boundary layer thickness δ = θ(H1 + H), and hence of other features of the developing profile.