at the Royal Aircraft Establishment, Farnborough, and a number distributed amongst the private aeronautical firms of Britain. 1 America has a number of channels of generally similar type, 2 but with a unique example in one instance where the speed of the air current is very high. 3 The oldest of the wind tunnels of importance in the development of aviation is that of Eiffel, 4 and from it in 1909-10 came a number of experiments on wing forms at a time when flying-machines were becoming realities. The Eiffel type of wind tunnel has been used elsewhere and in France a new installation has been erected at St. Cyr. s The other European wind tunnels of note are in Italy (Rome), at Gottingen University (Germany) and Koutchino (Russia). Owing to the general upheaval in Russia the last-named labora- tory is closed, but it earned distinction in the years of its activity particularly in dealing with interesting experiments on funda- mental points in the theory and practice of the day.
In general conception all wind tunnels agree in attempting to obtain a uniformly distributed, non-fluctuating air stream; and the tendency has been to increase the dimensions and the velocity attained in passing from one installation to a succeeding type. Economy of power for a given extension of experimental range is, by the principles of dynamical similarity, more readily obtained with large dimensions than with high speed. The best criterion, other things being unchanged, is the product of diameter and velocity, and judged on this standard the largest installations of the various countries do not differ materially.
At the Royal Aircraft Establishment (formerly called the Royal Aircraft Factory), Farnborough, a speed of 100 m.p.h. (nearly 1 50 ft. per sec.) is reached in an air stream 7 ft. square. At the National Physical Laboratory a speed of 1 10 ft. can be pro- duced in a stream 7 ft. deep by 14 ft. in width and forces on a model of the order of 200 Ib. are there contemplated.
The larger Eiffel tunnel gives an air speed of 40 metres per second (130 ft. approximately) on a circular section about two metres in diameter. The tunnel at McCook field (America) gives the very high speed of 500 ft. to a circular stream of air about 3 ft. in diameter.
The experimental section of an Eiffel type wind tunnel con- sists of an air stream as it crosses an open room from wall to wall, through a specially devised nozzle and collector. The National Physical Laboratory type and others use a working section of the stream in the centre of a chute with solid walls. There are no striking advantages of either type so far as can be seen at the present time. The great desiderata are uniformity of distribution of velocity across the stream and freedom from large pulsations. Uniformity of distribution is almost auto- matically secured by using a straight air stream. Once curvature has been introduced by the turning of corners the difficulties of producing uniformity are formidable. On the other hand the delivery of large volumes of air nearly half a million cub. ft. per minute in the large tunnels requires special consideration if large eddies in the room with consequent pulsations in the flow are to be avoided. There is an opinion, supported as yet only by crude experiments, that the N.P.L. type of channel is somewhat less fluctuating than the Eiffel type. For the delicate adjustments required in the measurement of stability coefficients high value attaches to the steadiness of the air stream.
In dealing with efficient wing forms, where the lift may be more than 20 times the resistance, it is important that the direction of the air stream be accurately known and remain fixed; one-tenth of a degree is considered to be the maximum permissible error. It is found by experience that in a parallel walled channel the wind sets itself parallel to the walls with the accuracy desired. Freedom from large variations of velocity across the section depends not only on the straightness of the chute but also on the distance over which the air has been in contact with solid walls. From some experiments by Stanton it appears that the final distribution of velocity in tubes is not reached for some 20 to 50 diameters behind the open end. On the score of space required and power needed such proportions are unrealizable in wind channels and in other respects would be dis-
1 Report, A. C. A., 1912-3, R and M, 68.
2 Mass. Inst. of Technology. McCook Field.
4 Eiffel, La Resistance de I'air et I' Aviation (Dimod & Pinet, 1910). 6 La Nature, Oct. 2 1921.
advantageous. Some variation of velocity distribution from point to point along a wind channel is then to be expected, there being -a retardation of flow at the walls and an acceleration in the centre. This change of flow is accompanied by a fall of static pressure along the working section of the channel. For experiments on wings, struts, etc., these departures from uniformity are unimportant but in the case of long models of airship forms there is introduced a spurious resistance large in comparison with that proper to the air- ship model. It has been suggested, and experiments are being car- ried out to give effect to it, that the objectionable effects of the wind channel might be minimized by the substitution of a slightly diverg- ing chute in the working section for the usual parallel part. It appears to be possible by such device to increase substantially the ease and accuracy of tests on airship forms.
The motion of the air in the wind tunnel is eddying and on this account a difference from motion through still air exists. So far, however, no suspicions have been aroused as to the inapplicability of model tests on this ground. Some eddies produced in the working of a tunnel are worthy of mention. If light sawdust be sprinkled over the floor of the building housing a wind tunnel, below the intake, it will be noticed that isolated miniature whirlwinds are produced. Some of these are vigorous and the base will clear a track amongst the sawdust whilst the core extends upwards to the tunnel intake. The spin in such eddies is great and the effect of the forces experi- enced by a body in the air flow is considerable. Being spasmodic, the effect is easily differentiated from that of the mean flow and an observer at an aerodynamic balance is conscious of a sharp blow on his apparatus. To eliminate these whirlwinds sufficiently a honey- comb is placed across the intake, the cells being small compared with the dimensions of the whirlwind. Some 10 % to 20 % of the en- ergy of the power plant may be dissipated by the frictional resist- ance of the honeycomb and some appreciable length of tunnel is required to permit of the levelling-up of the flow before reaching the working section.
The design of a wind tunnel will be seen to involve much study if more than a very moderate degree of refinement of experiment be contemplated. The following brief description of a tunnel intro- ducing modern knowledge may be of interest (see fig. 16).
The wind tunnel is housed in an unobstructed chamber a little longer than itself, a space of one and a half diameters between the intake and wall being sufficient for the satisfactory admission of air from the chamber to the tunnel. The cross section of the room should be 25 to 30 times that of the channel, otherwise the return flow of air from delivery to intake will produce fluctuations of undesirably large magnitude. The tunnel proper is straight and is placed symmetri- cally in the building, this being effective in securing symmetry of air flow in the working section. Taking the diameter of the section whether square or circular as a standard, the tunnel would have an overall length of 10 to 15 diameters made up of a parallel working section and intake four or five diameters long, having a rounded entrance and honeycomb, a cone connecting this working section to a circular race enclosing the airscrew, which may be of similar length, and a discharge section to the end of the room.
The airscrew giving steadiest flow is one of small pitch-diameter ratio but otherwise similar in characteristics to those used in aerial locomotion. The pitch-diameter ratio may be 0-4 upwards, the higher values giving rather greater economy of power and less steadiness. With careful design of airscrew and cone the divergence from channel to airscrew can be made large with resulting economy of power and no loss of steadiness.
The most modern method of dealing with the delivery stream is to divide the building into two parts by an openwork brick wall. Eddies in the return flow are thereby broken up to dimensions which do not greatly affect the steadiness of the air when it again enters the intake. In one instance, in addition to the partition wall, there is a structure closely surrounding the delivery from the airscrew; this delivery is in the form of a jet which impinges on the .end wall of the building, and splashing over it, reaches the corners and forms rollers along the four walls. The structure over the jet is designed to break up the stream more completely than the porous wall alone. Instead of the free jet spreading at the wall it is distributed through holes in the covering structure, the spacing being such that equal volumes of air are delivered through each unit of area of the distributor. The number of openings per unit area is small near the wall of the building and increases to cover the whole area just before the airscrew section. It is possible to reduce the velocity at which the air returns to the room to 5% of that in the jet without the introduction of appreciable back pressure at the airscrew.
Methods of Measurement of Velocity of Air. Having secured uniformity of distribution and a degree of steadiness sufficient for the type of experiment to be performed, it is necessary to be able to measure the air speed. No simple means is known of obtaining a standard of reference using a wind channel alone, and only one measure possibly two of absolute air speed appears to have been made under precision conditions. The particular measurements made on a whirling arm and in the William Froude National Tank at the National Physical Laboratory
