N9 O- 12 55 5
DESIGN AND FABRICATION REQUIREMENTS FOR LOW NOISE
SUPERSONIC/HYPERSONIC WIND TUNNELS
I. E. Beckwith
NASA Langley Research Center
Hampton, Virginia
F.-J. Chen and M. R. Malik
High Technology Corporation
Hampton, Virginia
947
PROPOSED SUPERSONIC LOW-DISTURBANCE TUNNEL
A schematic diagram of the new proposed Supersonic Low-Disturbance Tunnel at NASA
Langley is shown in figure 1. Existing high pressure air and vacuum systems will be
used. The specifications and design of two high quality air filters for the new
facility are based on experiences and data in the pilot tunnel (ref. 1). Aerodynamic
analysis and engineering details of the 8-ft diameter settling chamber and other
tunnel components are given in reference 2. The tunnel is designed to accommodate
nozzles of various lengths with Mach numbers ranging from 2 to 6. The first nozzle to
be operated in the tunnel will be the Mach 3.5 rapid expansion, two-dimensional nozzle
illustrated in the lower part of figure 1. The two-stroke model injector is required
to place the model into the quiet test core in the upstream part of the uniform flow
test rhombus. A 1/3-scale version of this nozzle has been developed and tested
extensively in the Pilot Low-Disturbance Tunnel at NASA Langley (refs. 1-4). The
techniques for obtaining laminar boundary layers on the nozzle walls, which is the key
requirement for quiet test section flow, will be presented in the next several
figures.
Filters
_-_-- High pressure and
II I IJJ [ [;[[_ high temperature air
II L j _ I0 vacuu m
i l _ -- or atm. A -.
/_ /_",',; _ _-__n um Vacuum
__!'., _'I! "_'_--_-_,_-'---'-'----_-_ ---,-- soheres
Acoust,_c Honey_--_om b I m_;_Zrin;:ctor _- Variable diffuser
odel z _ector
baffles and screens I
I
I
M = 3.5 contoured 2-D nozzle
(exit dim. = 18 in. x 30 in.)
Transition-_ _ Radiated noise
I I
Figure 1
948
QUIET TEST CORE IN M = 3.5 RAPID EXPANSION NOZZLE
The dominant source of test-section disturbances in conventional supersonic/
hypersonic tunnels is the acoustic radiation from eddies in the turbulent boundary
layers on the nozzle walls. In supersonic flow, this noise is in the form of finite-
strength wavelets which are propagated along Mach lines. Hence, as illustrated in
figure 2, the location of transition onset in the wall boundary layers is sensed with
a hot-wire probe at any point along a Mach line extended downstream from that location
which is then the "acoustic origin" for the onset of radiated noise in the nozzle flow
field. As the unit Reynolds number is increased, transition moves upstream along the
contoured walls in this nozzle. The quiet test core region then becomes smaller and
tends to approach some minimum size of streamwise length AX and height AY. When the
nozzle walls are very clean and highly polished, the minimum value of AX is about
4.5-inches (refs. 1 and 2). At high Reynolds numbers, the sidewall boundary layers
are generally turbulent. For these conditions, radiation from the side walls is
minimized by the large width of the nozzle (see lower part of fig. 2) and the small
local Mach numbers (M^ < 2.5) at the acoustic origin locatiDns The large width of
the quiet test core, AZ, allows the testing of swept wings and models at large angle
of attack.
/
6xi0 18x30
5 to 10 16 to 33
1.5 to 3.0 5 to 10
7.5 23.0
- Mach lines
Figure 2
949
TRANSITION REYNOLDS NUMBERS ON SHARP CONES
Local Reynolds numbers at transition onset, ReT, determined from recovery
temperatures measured on a sharp tip 5° half-angle cone in the pilot tunnel are
plotted against local unit Reynolds number, Re, in figure 3 (see refs. 3 and 4). For
values of Re/in < 8 x 10_, these data are in the range of atmospheric flight data
which are much higher than conventional wind-tunnel data due to the high-stream noise
levels in these latter tunnels (sources of these flight and conventional wind-tunnel
data are given in ref. 3). The cone used in the quiet tunnel tests was 15-inches long,
and for these lower values of Re/in, transition usually occurred on the cone well
downstream of the acoustic origin boundary of the quiet test core. Analysis and
correlation of these results (refs. 3 and 5) indicate that the cone boundary layer is
much more sensitive to wind-tunnel noise in the vicinity of the neutral stability
point than further downstream. For Re/in > 8 x 10v, the transition Reynolds numbers
decrease more or less rapidly towards the levels for previous conventional wind-tunnel
data depending on the nozzle wall finish. The effect of surface finish on wall
transition and quiet test core sizes in the pilot nozzle and the corresponding
quantitative requirements on the wall finish will be considered next.
2
1078
ReT 6
4
106
(I: 0 °
Quiet tunnel data, Moo= 3.5
- Nozzle Max k,
surface polish p-inch
o _B__-- 100 --__
Me = 1.4to4.6 __
Conventional wind
t unneldata_//////_lff////////' "_
_ Me = 2"5t° 4"
I I I t t Ill t I I I I t ill
2 4 68 2 4 68
105 106
Re/in
I
2
Figure 3
950
EFFECT OF TRANSITION ON QUIET TEST CORE LENGTH
Figure 4 shows how the axial distance from the throat to transition on the nozzle
wall, XT, is related to the length of the quiet test core, ,IX. Values of these
parameters and corresponding free-stream Reynolds numbers based on AX are given in
the table for large values of R/in > 9 x 10_ and for different surface finish
conditions. The original blocks were made of 17-4 PH stainless steel with a nominal
surface finish of 4 to 6 rms u-inches. After more polish work was completed the
finish was improved to about 1 rms u-inch and the values of AX then approached the
"minimum" values observed for this very good surface finish with the corresponding
much larger values of R AX . As part of an attempt to im])rove this finish a new set
of blocks was machined Trom 15-5 PH vacuum remelt stainless steel. The before-polish
data on these blocks indicates that AX was very s_all or zero. After preliminary
was Incr.ased by a factor of aboutpolish work, the value of AX for R /in _ 10 x 10_ " ,:
7 times. However, at R/in = 12.5_x 10b there was still no usable quiet test
core. To understand the reasons for these poor results, we will consider next the
effects of known roughness magnitudes and characteristics on nozzle-wall transition.
Mach 3.5 two-dimensional pilot nozzle
w__ Xl----_/-Transi!ion,-_--_Im /Radiatednoise
-'. -.\.
FIo _ _ -_- - <-_" " .--:" -'- -".... "" :--"'_" "- " -
,_ * I- .----___- Centerllne
//'_ _'__'n. _ X'_ _-"_ Quiet test core
Mach
lines
'-- Contour wall
Original blocks I
New blocks I
RoJin. xT, Quiet AX, Roo, L_X
x 10-5 inches inches x I0 -6
10.2 1.7 2.5 2. 6 Before
15.0 .9 .5 .8 Polish
10.4 2.7 4.6 4.8 After
15.5 2.6 4.5 7.0 Polish
I0.0 1.0 0.7 0.7
15.1 0.6 0 0 Before
9.3 2.9 4.9 4.6 After
12.5 0.4 0 0
Fi gure 4
951