| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|---|---|---|
|
Closed-circuit tunnels need corners with guide vanes of some kind, to deflect the flow without the boundary-layer separations that normally occur in all but very gentle bends. The "first corner" is the first one downstream of the test section, and so on.
The most popular type of corner is a (usually) 90 deg. bend, equipped with curved blades ("vanes": e.g. Fig. 4) or other devices to turn the flow smoothly. The shapes of the inner and outer walls are chosen to match the outer and inner profiles, respectively, of the blades. The assembly of blades is often called a "cascade", following turbomachinery terminology of uncertain origin. Some tunnels have 180 deg. "racecourse" bends, with vanes running full length or with separate sets of vanes at intervals, but these are difficult to build and the only apparent advantage is greater strength in the case of pressure tunnels: see Fig. 5. Internal pressure tends to straighten out sharp bends, as in the case of an inflated bicycle inner tube. There does not seem to be any information on performance, but the pressure drop through several sets of vanes is presumably larger than that through the two sets of a pair of 90 deg. corners, and in principle secondary flows may lead to the accumulation of slow-moving fluid near the outer side of the bend.
The design rules for 90 deg. corner vanes are in principle the same as for turbomachine cascades. The gap/chord ratio is usually chosen as about 0.25: the current practice is to use circular-arc camber lines of 85-86 deg. subtended angle, with a leading edge angle of attack of 4 or 5 deg. and trailing edges set along the centerline of the tunnel. For all but the biggest tunnels, vanes can be pressed or rolled from sheet metal with rounded edges: vanes rolled in a conventional 3-roller mill must have straight leading and trailing edge extensions: the latter feature, at least, is aerodynamically desirable. It is a useful precaution to ensure that the trailing-edge angle can be changed if the flow does not come out straight (with resulting asymmetric flow, and danger of separation, in the diffuser). |
The total-pressure drop through a 90 deg. corner with thin vanes is only 0.12-0.15 of the dynamic pressure, compared with a drop equal to or greater than the dynamic pressure in the case of a square elbow with no vanes: thin vanes have proved just as successful as the thick aerofoil-section vanes used formerly (e.g. Fig. 1). However use of modern airfoil design methods can lead to more efficient design (see A. Sahlin & A.V. Johansson, Design of guide vanes for minimizing the pressure loss in sharp bends,
Phys. Fluids A3, 1934, 1991). This allows vane gap/chord ratio to be increased, but the increase can lead to strong crossflow and large total-pressure losses in the end-wall boundary layers (not considered in the abovementioned work).
Attempts have been made to combine corner vanes with an expansion in area round the corner: in general the efficiency of diffusion is low and there is a danger of flow separation, but London et al. (ASME J. Engg for Power 90, 271, 1968) describe a "one-dimensional" design, with a tapering inlet duct and an outlet honeycomb at 90 deg. to the inlet, which produces nearly-uniform exit flow.
Large high-speed tunnels need a heat exchanger somewhere in the circuit. It is a very bad idea to use the fourth corner vanes as a "radiator", because the screens and/or honeycomb downstream of the corner will reduce the velocity fluctuations which would otherwise mix out the spatial and temporal variations in temperature. (It is of course even worse to put the heat exchanger in the settling chamber.) Temperature fluctuations in a test section lead to spurious indications of velocity fluctuations in hot-wire readings, and can also cause unsteady refraction of the light beams of laser Doppler or Schlieren systems (the same mechanism makes stars twinkle).
 |
|
