*2.1.3. Nozzle box and test section*

The Stagnation pressure control system was deliberately chosen to be manual in orer to sim‐ plify control although a hybrid system could be incorporated at a later stage if desired. This system would consist of manual start and initial stabilisation with switch over to automatic operation once the stagnation pressure has stabilised. Some of the problems of supersonic tunnel automatic stagnation pressure control have been discussed by Pugh and Ward [1]

The stagnation chamber has an inside diameter of 13.5 inches and is connected to the 6 inch inlet pipe work by a 300 conical rapid expansion. Flow stabilisation and smoothing devices consist of a conical perforated plate and flow smoothing and turbulence reduction screens (Fig. 6). A parallel settling length is provided downstream of the screens and is fitted with

The mean velocity at the screens is approximately 20 ft/sec for Mach 3 operation. The cham‐ ber is also sized to permit operation down to Mach 1.5 using the same test section area when the velocity at the screens could increase to about 70 ft/sec. This remains within the range of

The perforated cone has an apex angle of 900 and is manufactured from a ¼ inch plate with 3/8 inch diameter holes on 9/6 inch centres. The perforations have an open area ratio of 40%. The mean Mach number through the perforations under the worst conditions, which occur at the lowest test section Mach number, is less than 0.1. The cone is welded into the wide angle expansion. In operation, it appears to have eliminated any pressure fluctuations gen‐ erated by the stagnation pressure control valve as well as assisted in 'filling' the wide angle

The four stainless steel flow smoothing screens are of 24 mesh by 34 gauge wire and have an open area ratio of 49.5%. The screens are fixed in individual aluminium retaining ring frames. These frames are clamped together by long bolts passing through large frames at‐ tached to the rapid expansion and settling length sections of the stagnation chamber. The individual frames are spigoted together to ensure internal surface alignment and are sealed

The parallel settling length downstream of the screen is 1500 screen wire diameters long, or

A two dimensional contraction and section change transition region is provided at the down‐ stream end of the settling chamber. This region has a rectangular outlet area of 12 inch x 4 inch and a circular 13.25 inch diameter inlet section. The area contraction ratio is 2.9:1. A further two-dimensional contraction, of ratio 10:1, is built into the nozzle blocks to contract the air‐ stream to a sonic throat 1.185 inch high by 4 inches wide. The method of contraction design presented by Gibbings [4] is recommended. This method is also applicable to contractions in

which there is an appreciable axial trigger between the plan and elevation profiles.

and Conolan [2].

diffuser.

by O rings at each joint.

a length of approximately 18 inches.

*2.1.2. Stagnation chamber*

stagnation temperature and pressure tappings.

82 Wind Tunnel Designs and Their Diverse Engineering Applications

10 to 80 ft/sec as recommended by Pope [3].

The nozzle box is of conventional construction and is manufactured from steel plate with internal surfaces ground after welding and stress relieving. Heavy stiffening ribs prevent deformation under pressure forces, particularly in the throat region. Dowels are fitted to en‐ sure accurate and repeatable alignment of adjacent parts. Circumferential 'O' ring seals are provided at each end of the nozzle box. One side wall opens downwards on hinges to facili‐ tate nozzle block changes. The complete assembly of nozzle box and settling length section of the stagnation chamber can be moved on rollers to permit easy screen removal. The roll‐ ers are brought into operation by four jack screws. An axial movement of 3 inches in the downstream direction is possible.

There are circular Schlieren window positions in the nozzle box walls, one pair at the throat and one pair at the test section. The windows have a clear diameter of 7.5 inches and thick‐ ness of 1 inch and are held in a sub-frame which is fixed to the tunnel by a clamping ring. This arrangement permits easy removal and rotation of each window. Rotation of the win‐ dow assembly permits selection of the optimum orientation for optical characteristics of the glass fitted. This allows the use of cheap and comparatively low grade plate glass. The glass is sealed to the sub-frame with Dow Corning Silastic 732 RTV silicon rubber compound. A special jig has been developed for window assembly which ensures that the glass is com‐ pletely 'floating' in Silastic and is also flush to within 0.001 inches with the frame edges. A set of high quality glass windows obtained from Optical Works, UK are also available for specially sensitive Schlieren applications.

The supersonic nozzle blocks are manufactured from extruded AA28S aluminium alloy. The contours were generated by a programmed 'Hydroptic jig borer and were finally hand fin‐ ished to remove machining marks. Each block is fitted with a continuous, circular cross sec‐ tion rubber seal which runs as close to the contoured surface as is possible using straight line approximations. The nozzle blocks are held in the nozzle box with the bolts passing through the top and bottom walls of the box into barrel nuts inside each block. Location is by integral machined pads on the basis of each block. It is now realised that the provision of separate ground pads on the base of each nozzle block configuration would have permitted the fitting of permanent shims clamped between pad and block, thereby simplifying the fit‐ ting and accurate setting up of each nozzle block configuration within the nozzle box.

The nozzle block co-ordinates are those derived by McCabe [5] for operation at Mach 3 in a nominal 5.5 inch x 5.5 inch test section. The design method used by McCabe divides the noz‐ zle inviscid core flow into five main regions, as follows:

**•** A subsonic contraction


McCabe also divides each of the 3rd and 4th regions into further regions so as to take advant‐ age of more precise computational methods and eliminate discontinuities in nozzle curva‐ ture. The nozzle contour boundary layer corrections are based on the data of Sibulkin [6] for the throat region, with corrections on the contoured and flat side walls obtained from the data of Rogers and Davis [6]. Because the nozzle contours so derived did not have zero slope at the test section location, a cubic curve was fitted to permit smooth transition to the parallel block region downstream of the test rhombus. The boundary conditions were con‐ tinuity of ordinate, slope and second derivative at the upstream and zero slope at the down‐ stream end. Before feeding to the jig borer, the complete set of nozzle block co-ordinates was smoothed by computer using a 6th order polynomial. Data on nozzle profile design may be obtained from Refs 8-15.

### *2.1.4. Diffusers and model support system*

Downstream of the nozzle box are a string chamber, supersonic diffuser, first stage subsonic diffuser, parallel make up duct and a set of cascade turning vanes. The complete assembly downstream of the nozzle box to the corner cascade is mounted on a flanged wheel and rail system and may be moved 18 inches in the downstream direction away from the test sec‐ tion. This allows easy access for model changes. The face of the sting chamber is aligned with, and closed against, the nozzle box pressure seal by four dowel pins and tapered lock‐ ing wedges. Four cam clamps are used to close the pressure seal between the cascade corner and the vertical second stage subsonic diffuser.

The sting chamber can be removed entirely, if required. It is fitted with a pair of circular openings of the same diameter as those in the nozzle box side walls and, can, therefore, ac‐ cept the interchangeable Schlieren windows or metal window blanks. A side mounted sting may be fitted to either of the side window blanks. A vertical, full span, wedge nose strut is also in use as a model sting support.

The optimum design of a supersonic diffuser for a small wind tunnel presents a difficult problem as it must operate over a wide range of inlet flow conditions and Reynolds num‐ bers. Moreover, at Mach numbers in excess of 2, optimum starting and running diffuser geo‐ metries diverge significantly. It is assumed in the following discussion that the supersonic diffuser is followed by a subsonic diffuser of small divergence angle.

In general, the airstream entering the diffuser is highly turbulent because it contains the wake flow from the test section model and its support system. Moreover, diffuser performance is in‐ fluenced by boundary layer effects which are a function of tunnel Reynolds number.

Supersonic diffusers for small wind tunnels may be one, or a combination of the following types:


**•** The high subsonic, low supersonic throat region

84 Wind Tunnel Designs and Their Diverse Engineering Applications

**•** The straightening or 'Busemann Region' **•** The parallel flow or test section region

obtained from Refs 8-15.

*2.1.4. Diffusers and model support system*

and the vertical second stage subsonic diffuser.

also in use as a model sting support.

types:

**•** An initial expansion region wherein the contour slope increases to its maximum value

McCabe also divides each of the 3rd and 4th regions into further regions so as to take advant‐ age of more precise computational methods and eliminate discontinuities in nozzle curva‐ ture. The nozzle contour boundary layer corrections are based on the data of Sibulkin [6] for the throat region, with corrections on the contoured and flat side walls obtained from the data of Rogers and Davis [6]. Because the nozzle contours so derived did not have zero slope at the test section location, a cubic curve was fitted to permit smooth transition to the parallel block region downstream of the test rhombus. The boundary conditions were con‐ tinuity of ordinate, slope and second derivative at the upstream and zero slope at the down‐ stream end. Before feeding to the jig borer, the complete set of nozzle block co-ordinates was smoothed by computer using a 6th order polynomial. Data on nozzle profile design may be

Downstream of the nozzle box are a string chamber, supersonic diffuser, first stage subsonic diffuser, parallel make up duct and a set of cascade turning vanes. The complete assembly downstream of the nozzle box to the corner cascade is mounted on a flanged wheel and rail system and may be moved 18 inches in the downstream direction away from the test sec‐ tion. This allows easy access for model changes. The face of the sting chamber is aligned with, and closed against, the nozzle box pressure seal by four dowel pins and tapered lock‐ ing wedges. Four cam clamps are used to close the pressure seal between the cascade corner

The sting chamber can be removed entirely, if required. It is fitted with a pair of circular openings of the same diameter as those in the nozzle box side walls and, can, therefore, ac‐ cept the interchangeable Schlieren windows or metal window blanks. A side mounted sting may be fitted to either of the side window blanks. A vertical, full span, wedge nose strut is

The optimum design of a supersonic diffuser for a small wind tunnel presents a difficult problem as it must operate over a wide range of inlet flow conditions and Reynolds num‐ bers. Moreover, at Mach numbers in excess of 2, optimum starting and running diffuser geo‐ metries diverge significantly. It is assumed in the following discussion that the supersonic

In general, the airstream entering the diffuser is highly turbulent because it contains the wake flow from the test section model and its support system. Moreover, diffuser performance is in‐

Supersonic diffusers for small wind tunnels may be one, or a combination of the following

fluenced by boundary layer effects which are a function of tunnel Reynolds number.

diffuser is followed by a subsonic diffuser of small divergence angle.

**•** Oblique shock diffuser with centre body

The diffuser throat must be sized must be to 'swallow' a normal shock when starting at the design Mach number in a second throat type of diffuser. Faro [11] obtained the theoretical minimum starting contraction ration required, together with experimental points from sev‐ eral sources. For Mach numbers in excess of 2, the experimental values are lower than the theoretical values. It is suspected that this is due the starting shock passing through the sec‐ ond throat before the design Mach number has been reached. The experimental data of Lu‐ kasiewicz [16] indicates that at the area ratio of A2/A1 =0.7, which is required for stating at Mach 3, the starting pressure ratio requirement of a fixed geometry second throat diffuser, might be 0 to 30% less than that required by an optimum length constant area diffuser. If a variable geometry second throat diffuser is used, significant reductions in optimum running contraction ratio may be obtained particularly at Mach numbers in excess of 2.

Variable geometry diffusers, set at optimum running contraction ratios after starting, enable reductions in running pressure ratio of up to about 45% when compared with optimum length constant area ducts [17]. This reduction is achieved, however, at a considerable in‐ crease in complexity of construction and operation when compared with the simple constant area diffuser.

The advantages of the fixed and variable area second throat diffusers over the constant area diffuser are less certain when allowance is made for the presence of the model and its sup‐ port system within the test section. Faro [11] states that in nearly all cases the presence of a model reduces supersonic diffuser efficiency. This is confirmed by De Leo and Huerta [17] who found that a 5% blockage model increased starting and running pressure ratios by ap‐ proximately 7 and 13 % respectively for optimum empty tunnel diffuser geometry and Mach numbers of 2.5 and 3.4. However, experiments have also shown that the presence of a model tends to reduce tunnel instability through stabilisation of the diffuser shock system. This appears to be due to interaction between the wall reflections of the model bow wave and the diffuser shock system. In summary, therefore, it seems that the second throat diffus‐ er types do not offer significant advantages over the constant area duct for small supersonic tunnels. This conclusion does not apply to very large intermittent type tunnels, however, as then a reduction in running pressure ratio could lead to significant increases in tunnel run time for a given pressure storage capacity or reduction in capital cost of the storage system for a given run time.

The supersonic diffuser of the University of New South Wales is a constant area duct system although the sting strut, when used probably acts as a centre body with oblique shock diffu‐ sion. Although such a constant diffuser is a dissipative system, it does have the considerable advantage of being stable over a wide range of inlet flow conditions and of being simple to construct and maintain.

The only design decision required for a constant area supersonic diffuser is the optimum length of duct. In such a diffuser the normal shock appears as a shock system which is strongly affected by the state of the boundary layer. Experimental dat quoted by Lukasiewcz [16] confirms that for best efficiency the shock compression process should be completed in the constant area duct and not in the divergent subsonic diffuser downstream of the super‐ sonic diffuser. Faro [11] illustrates the gain in isentropic efficiency with increasing length to height ratio for a constant area duct at Mach 2. The significant reduction in operating pres‐ sure ratio with increasing length of parallel duct may be seen in curves plotted for length to height ratios of 0, 2 and 7. Further design data for constant area diffusers is demonstrated by Faro [11]. Two points are noted in connection with his work:


Faro [26] indicates that a single wedge such as the leading edge of a sting support strut may be used to provide an oblique shock system which will improve diffuser efficiency over the simple normal shock case. The benefits for this type of device are limited, however, is that high efficiency can only be obtained with a large number of oblique shocks which in turn implies design for a specific Mach number and thus a narrow range of off-design conditions. The simple strut type oblique shock generator gives moderate efficiency gains over a wider range of Mach numbers.

To summarise, little data for the design of constant area supersonic diffusers or for the effect of a model and strut system on diffuser efficiency can be found. The available information suggests that:


The supersonic diffuser of the University of New South Wales tunnel is a parallel wall rec‐ tangular duct fabricated from 4 inch x 1 inch extruded aluminium bar top and bottom walls and 0.5 inch aluminium plate side walls. The top and bottom walls may be easily replaced with a set of contoured blocks so as to provide a fixed area second throat, if so desired. The parallel diffuser length is 8.4 diffuser heights from the rear of the model support strut and 11.4 heights from the end of the supersonic nozzle with model support system removed. A removable parallel subsonic make-up duct permits the fitting of an additional 4.3 diffuser heights. The window blanks alongside the model support strut also allow for experiments involving 'de-blocking' of the area around the strut.

The design of the subsonic diffuser was straightforward. The data of Lukasiewicz [16] indi‐ cates that, in the Mach number range 0.4 to 0.9, total pressure recovery is virtually constant at about 0.88 for open ducts without models and that the diffuser divergence angle should be less than 70 . Data on subsonic diffuser design is available from Refs 14 and 16.

There are two stages of subsonic diffusion separated by the corner cascade (Fig. 6). The first stage of subsonic diffusion is separated by the corner cascade (Fig. 6). The first stage has an area ratio of 5.6 and divergence angle of 60. Maximum Mach number at the subsonic diffus‐ er exit is approximately 0.13. The diffuser is constructed from 3/16 inch steel plate reinforced at 6 inch x 3 inch centres with 1 x 0.25 inch flat bars on edge.

The second stage subsonic diffuser has a 60 divergence angle and 3:1 area ratio. Maximum Mach number at exist to this diffuser is approximately 0.04. The diffuser is manufactured from plywood and incorporates part of the tunnel silencing system. The corner cascade uti‐ lises sheet metal circular arc turning vanes.
