*2.1.5. Silencer*

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

**•** The Mach number Mav is the average Mach number at the supersonic diffuser inlet and would be less than the test section Mach number because of the presence of the model and its support system and boundary layer growth between test section and diffuser.

**•** The effect of the free stream Reynolds number is not accounted for. Some indication of the reduction in the length of the shock compression system at high Reynolds numbers may be obtained from [16]. This data is applicable to a Mach number of approximately 2.

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

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

**•** The shock system compression process should be completed within the parallel diffuser

**•** The optimum length of parallel duct required to complete the compression process is a strong function of Mach number and Reynolds number. This length is probably within the 5 to 12 diffuser heights of Mach numbers of 1.5 to 3.5 and Reynolds numbers of 2 x 105

. Design data for supersonic diffusers may be obtained from Refs 16-22].

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

Faro [11]. Two points are noted in connection with his work:

86 Wind Tunnel Designs and Their Diverse Engineering Applications

range of Mach numbers.

duct for best efficiency

suggests that:

to 6 x 106

Preliminary investigations on an existing M3.5, 4 inch diameter conical nozzle indicated that the noise level for an unsilenced tunnel would be unacceptably high at about 120 db in the frequency band of 100 to 2000 Hz. Accordingly, a silencer was designed for the supersonic tunnel to the following requirements:


After investigation, an attenuating duct design was chosen as best fulfilling these require‐ ments. This type of silencer requires an absorbent material as dense as possible with a thick‐ ness of 2 inch to 12 inches to absorb the low frequency noise below 500 Hz. Attenuation at the lower frequencies is increased considerably by the use of a perforated duct facing mate‐ rial having about 3 to 10% open area perforations. Low frequency attenuation is further as‐ sisted by providing airspace behind the absorbent material and increasing the amount of absorbent around the duct periphery. When compared with splitter type duct attenuators, low frequency attenuation can be improved by arranging a given amount of attenuation ma‐ terial such that it forms thick layers. This latter arrangement gives a lower peak but better average attenuation over the 100 to 1000 Hz frequency band. Design information can be found in the literature [23-29].

The silencer for the University of New South Wales tunnel is constructed in two sections: the first of which is built around the second stage subsonic diffuser. The first section com‐ prises 6 to 12 inch thickness of polythene wrapped rockwool batts and loose rockwool fill around around all four sides of the diffuser. The rockwool density varies from 4 to 6 lb/cubic feet for the loose fill. The duct interline is surfaced with 3/16 inch thick perforated plywood and the outside of the silencer is sealed with 1 inch thick, exterior quality waterproof ply‐ wood. Both internal and external surfacing materials are heavily glued, screwed and nailed to substantial connecting framing. The second section of the silencer, which is 16 ft long is a rectangular duct lined on two sides with 6 inch thickness of rockwool batts backed by a 3 inch airspace. The remaining two sides of this duct are 1 inch thick exterior plywood. Other constructional details are similar to those of the first section silencer. The second diffuser section is run in the laboratory ceiling space and is supported from the roof structure on 'Si‐ lentbloc' vibration isolators.

Initial tests on completion of the tunnel indicated a large direct sound transmission through the walls of the first stage subsonic diffuser. This was found to be caused by high frequency resonance of the 3/16 inch thick flat steel plate walls. The vibration was almost completely eliminated and the noise level reduced by decreasing the spacing of the existing 1 inch x 0.25 inch stiffening bars from approximately 12 inch x 6 inch to 6 inch x 3 inch centres as descri‐ bed in section 4.4.

In the final form, the silencer has reduced the noise level in the vicinity of the tunnel to about 75 to 90 dB, for the 100 to 2000 Hz band, depending to some extent upon the operating stagnation pressure. It is estimated that the duct silencer provides an attenuation of about 2 to 3 dB per foot of length in the frequency range 100 to 1000 Hz.
