Вы находитесь на странице: 1из 10

Diffusers for Supersonic W i n d Tunnels

J. L U K A S I E W I C Z *
National Aeronautical Establishment, Canada

SUMMARY present attempt at a systematic treatment of super-


A systematic review of published experimental data relating to sonic diffuser design and performance. While in no
performance and design of wind-tunnel diffusers for entry Mach way complete, it indicates the extent of available data
Numbers up to 10 is presented. I t is found t h a t : and gaps to be filled in.
(1) Compressibility effects are negligible in diffusers for M <
0.9. The subject has been divided up into a number of dis-
(2) Shock compression is more efficient in constant-area than tinct problems, as follows:
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

in divergent ducts and usually requires appreciable duct length (1) Performance of subsonic diffusers at high sub-
for completion. sonic Mach Numbers (i.e., effects of compressibility).
(3) Pressure recovery of diffusers without contraction is im-
proved by use of a long throat at entry but is always smaller (2) Shock compression in a simple divergent and
than the theoretical normal shock recovery. constant-area duct as a phenomenon common to all
(4) At Mach Numbers below 4, maximum or starting con- supersonic diffusers.
traction of constant geometry diffusers is closely predicted by (3) Performance of diffusers without contraction and
simple theory, and better than normal shock pressure recoveries
are achieved.
influence of a long, constant-area section at entry,
(5) With variable contraction diffusers, pressure recoveries of as (4) Maximum contraction of constant geometry dif-
much as twice the normal shock recovery are obtainable at fusers and their performance.
hypersonic Mach Numbers. In general, optimum diffuser con- (5) Contraction and performance of variable geom-
traction angles increase with entry Mach Number increasing.
etry diffusers.
SYMBOLS Only diffusers used with fully closed wind-tunnel
D = diameter or side of square working sections are considered, with no models (un-
D = hydraulic diameter, 4(area/perimeter) = less otherwise stated) mounted in the working section.
D (for circular or square sections) Insofar as possible, only reliable experimental data are
L = length of shock compression region
quoted; in fact some well-known results, obtained
M = Mach Number
MF = Mach Number based on static pressure and Fanno with wet air or with an ill-defined diffuser setup, were
line of state omitted from this review.
P absolute pressure, static In all cases the Reynolds Number of tests is given in
PQ absolute pressure, stagnation (at nozzle or diffuser terms of free-stream conditions at diffuser entry Mach
inlet)
Number and hydraulic diameter D of diffuser entry.
iY = stagnation pressure at diffuser outlet
ReD = Reynolds Number based on free-stream velocity,
density, viscosity, and length D COMPRESSIBILITY EFFECTS IN SUBSONIC DIFFUSERS
6 = total angle of divergence or contraction (all angles
quoted in the text are total angles) Although outside the scope of supersonic diffuser
xp diffuser contraction ratio = problem, the operation of subsonic diffusers at high
(diffuser entry area)/(diffuser throat area) subsonic velocities is briefly considered as a particular
\pmax. = maximum theoretical diffuser contraction ratio (for aspect of compressibility effects in diffusers in general.
flow starting)
It is found that the performance of diffusers of small
divergence angle ( < 7 total cone angle) is practically
INTRODUCTION
independent of entry Mach Number (provided M <

T HE SUBJECT OF THIS PAPER was first reviewed by


Crocco1 in 1935. At that time little experimental
data existed on supersonic diffusers, and most results
0.9) when expressed as a fraction of theoretical pressure
rise. This is shown by experimental results plotted
in Fig. 1; about 85 to 90 per cent of theoretical pressure
quoted by Crocco related to shock compression in di- rise is obtained in practice. The former corresponds
vergent ducts, such as Laval nozzles or subsonic dif- to a stagnation pressure loss of 3 per cent at M = 0.6
fusers. A second review2 appeared in 1946; it included and 6 per cent at M = 0.9. Since these results were
newer experimental data and dealt with the theory of obtained with no model in the tunnel working section
supersonic diffusers. Since then scores of supersonic and with a particularly good velocity distribution at
tunnels of various designs were built, and enough ex- the diffuser entry, they must be regarded as upper
perimental data were accumulated to warrant the limits of efficiencies attainable in actual subsonic wind-
tunnel diffusers.
Received December 11, 1952. Revised and received M a y 18,
1953. In some high-speed subsonic tunnels, an adjustable
* Head, High Speed Aerodynamics Laboratory. sonic choke or throat is fitted between the working sec-
617
618 J O U R N A L OF THE A E R O N A U T I C A L S C I E N C E S S E P T E M B E R, 19 53

Po-P -- LOW SPEED, CONICAL. SQUIRE R AE (1947) lent to supersonic tunnels operating a t low M a c h N u m -
Po-P bers.
CONE
0-95 J
ANGLE
693 R ^5 I 5x 10 AT M = 0 - 5
D SHOCK COMPRESSION I N D U C T S
617 NAUMANN F.B. 1 7 0 5 ( 1 9 4 2 )
501 Experiment shows that, in general, a continuous de-
CONE
Re
ANGLE D celeration of supersonic flow to subsonic velocity is not
7 CONE. Re a* 0 - 4 x 1 0 * AT M = 0 - 5
-6 I06 possible, the transition always occurring through shock
GOETHALS (1950)
compression. T h e corresponding simple theoretical
model of flow postulates a normal shock b u t neglects
0 90 viscous effects. I t is not surprising, therefore, t h a t
shock compression as observed in ducts, in presence of
U-4-IOIO r |
.* J wall boundary layer, usually differs from the theoretical
LHO 10*1
4 V of o
bo / model. Observations of shock compression in diver-
gent ducts (such as Laval nozzles, downstream of
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

throat) show a gradual rather than sudden increase of


pressure a t the wall, the static pressure never attaining
the values prescribed b y simple theory and stagnation
0 85
0 0-2 0-4 0-6 0-8 10 pressure losses appreciably exceeding those appropriate
FIG. 1. Pressure rise in subsonic diffusers. to a normal shock. Typical of the latter are results of
Castagna's 6 measurements of shock compression in a 6
3L
conical pipe, shown in Fig. 9 in terms of M a c h N u m b e r
based on the lowest observed wall pressure (i.e., a t the
start of compression) and assumption of isentropic flow.
DIVERGENCE Comparison with t h e normal shock curve indicates the
BASED ON EGGINK FB 1756 (1943) ANGLE 9
inadequacy of the normal shock theory. However,
R e Q ~ 0-5 xlO 6 the discrepancy is reduced, particularly a t higher M a c h
tin mint, m i milium ii,iiim
Numbers, b y making the angle of duct divergence
smaller, as shown in Fig. 2.
T h e above overall characteristics of shock compres-
20 sion in divergent ducts are supplemented b y schlieren
observations, which indicate t h a t , as a result of shock
and boundary-layer interaction, the region of shock
LOCATION 3 67
OF SHOCK compression m a y involve several curved or oblique
AT M shocks, forming regular or Mach-type reflections, and,
a t the walls, extremely thick boundary layers and flow
separations. T h e latter increase b o t h with shock
pressure ratio (i.e., flow M a c h Number) and with duct
divergence, approaching in extreme cases an essentially
free-jet configuration, the flow being completely de-

10
10 1-5 20 M
FIG. 2. Effect of duct divergence on shock compression,

tion and the difluser entry. Its purpose is to provide


control over working section M a c h N u m b e r a n d / o r
to prevent disturbances originating in the difluser or
compressor from reaching t h e model. Following the
choke, an expansion to slightly supersonic velocity and
shock compression take place in the subsonic difluser
entry. T h e difluser d a t a just reviewed are therefore
not applicable to subsonic tunnels equipped with a
choke which, insofar as difluser is concerned, are equiva- FIG. 3. Detached flow in supersonic nozzle for M = 4.4.
D I F F U S E R S FOR S U P E R S O N I C WIND T U N N E L S 619

tached from the walls following shock compression. An INTERACTION OF


REFLECTED SHOCK
example of this is shown in Fig. 3, which is a schlieren WITH TUBE
B0 UNOARY
photograph of flow in a 4.4 M a c h N u m b e r tunnel nozzle R A R E F A C T I O N WAVE 7

operating a t a pressure ratio insufficient to induce a


fully supersonic flow. Both schlieren and pressure ob-
servations suggest t h a t higher efficiencies of shock com-
k \A ! / / / ] i Z LAYER

x
pression are obtained in small divergence or constant- k \ * / / / / Ay ^
area ducts. In such ducts there exists in the subsonic
flow, downstream of the shock compression region, a \ V A // / / / / ! /-" 1 V

positive velocity gradient t h a t tends to stabilize the


boundary layer and reduce the separation and resulting
losses.
L
~*X
A number of observations of supersonic compression DIAPHRAGM REGION OF PHOTOGRAPHS

in constant-area ducts, in fact, indicate t h a t this is


Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

| 1
-^---=-~ 3 X
the case, although the compression does not necessarily
I G H
resemble a normal shock. Typical schlieren photo- I T I A L
graphs taken in a 1-in. square duct are shown in Fig. 4.
T h e compression consists of several shocks closely fol- FIG. 5. x-t diagram of shock-tube flow.
lowing one another, their shape presumably depending
on boundary-layer thickness relative to duct diameter
(at given M a c h and Reynolds Numbers).

A similar flow configuration m a y be expected to oc-


cur in shock-tube flow and is believed to cause the
hitherto unexplained appearance f of wave-speed cam-
era photographs, examples of which are shown in Fig.
6. J T h e wave-speed photographs are taken with a
schlieren optical system and a drum camera, the flow
inside the tube being viewed through narrow and long
windows parallel to the tube axis. T h e photographs
can be interpreted with the aid of Fig. 5, in which shock-
tube flow is represented diagrammatically in x, t co-
ordinates. T h e region of t u b e flow corresponding to
the photographs is marked in Fig. 5, and the tube-
f First evident in original photographs obtained by Payman
and Shepherd, reference 8, Figs. 3, 4, and 8.
I The author is indebted to Dr. I . I . Glass and the Institute of
Aerophysics, University of Toronto, for permission to make use
of wave-speed photographs originally published in reference 9.

FIG. 6 Wave-speed camera schlieren photographs of shock


compression in a shock tube. 9

length (x) and time (/) coordinates are oriented in the


same manner in Figs. 5 and 6. Following the bursting
of the diaphragm, a shock wave is propagated in the
low-pressure chamber of the shock tube and eventually
is reflected from the closed tube end, as shown in Fig.
5. On its return travel, the shock enters tube sections
in which b o u n d a r y layer has already developed and
presumably causes interaction similar to the stationary
FIG. 4. Schlieren photographs of shock compression in 1-in. shock-boundary-layer system of Fig. 4. T h e multiple
square duct (Rev ~ 0.2 X 106) in presence of thick boundary shocks, sketched in Fig. 5, appear as series of oblique
layer, a, b, c M ~ 1.5, shock compression at increasing
distances from nozzle, d M ~ 2.5. ripples in the wave-speed photographs, Fig. 6. I t will
620 JOURNAL OF THE A E R O N A U T I C A L S C I E N C E S S E P T E M B E R, 19 53

TABLE 1
Shock Compression in Tube in Presence of Thick Boundary Layer (Computed from Experimental Data of Reference 12)
1. Station Nozzle exit or 1 2 3 4 5 6 7 8
tube entry
2. L/D 0 2 3.75 8 14 21.5 26.4 30.5 40
3. (Pi/Po) X 103 observed 2.2 2.5 2.9 3.5 4.4 5.5 6.6 7.5 10.4
(shockless flow)
4. MY&nno (based on nominal 5.0 4.28 3.94 3.53 3.08 2.69 2.39 2.19 1.95
conditions at tube entry,
M = 5, Pt/Po = 0.00189)
5. M (based on measured Pi/Po 4.87 4.76 4.64 4.49 4.31 4.14 4.0 3.9 3.67
and isentropic flow)
6. Pi/P\ measured 19.6 16.6 13.4 10.1 7.6 6.1 5.1
7. {P*/PX)M Fanno 21.2 17.9 14.4 10.9 8.3 6.5 5.4
8. (P*/PI)M 26.3 24.9 23.3 21.5 19.8 38.8 38.4
9. L/D (length of shock com- 12 12.5 12 11.4 10.8 10 9
pression)

be noticed that the ripples are approximately parallel, the low Reynolds Number of tests (about 0.2 million
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

indicating that the whole compression region moves based on tube diameter), appreciable compression oc-
with a substantially constant velocity. The phe- curred in the tube even for shockless flow, the static
nomenon appears to gain strength with the initial tube- pressure Pi (third line, Table 1) increasing about four
pressure ratio (given in Fig. 6) or shock intensity and, and one-half times between the nozzle exit (or tube en-
in fact, is hardly noticeable at extremely low tube- trance) and station 8, 40 diameters downstream. Since
pressure ratios. 9 the flow was, to a good approximation, adiabatic, its
effective mean parameters could be obtained from pres-
So far, only few quantitative results are available on
sure distribution and Fanno line of state (derived from
shock compression in constant-area ducts. Most ex-
equations of conservation of mass, energy, and area).
tensive tests were carried out at M.I.T. 12 with relatively
In order to assess the boundary-layer thickness, the
thick boundary layers and therefore under conditions "Fanno line" Mach Number MF is compared in Fig. 7
not representative of conventional wind-tunnel prac- and Table 1, lines 4 and 5, with Mach Number M de-
tice. rived from static pressure ratio Pi/Po on the assump-
The M.I.T. flow setup consisted of a circular super- tion of isentropic flow. The corresponding boundary-
sonic nozzle, followed by a 50-in. long 1-in. diameter layer thickness can be roughly evaluated from the fol-
tube in which wall-pressure distribution was measured lowing consideration: Assuming that boundary layer
for different back pressures. Typical results obtained does not fill completely the tube cross section, the isen-
in M.I.T. tests are analyzed in Table 1. Because of tropic Mach Numbers M and Mach Number-area ratio
relationship can be used to estimate the boundary-layer
displacement thickness at any station. Taking a one-
seventh power-law velocity distribution in the boundary
:
i

'"1
layer and 8*/5 ~ 0.4 at M = 4, we find that at station 7
{ 1 (Table 1) the boundary layer is about 0.44 in. thick,
COMPUTED FROM I
NEUMANN 8 LUSTWERK almost filling completely 0.5-in. radius tube.f
30
M IT. (194 9)
. / / With increased tube back pressure, a shock compres-
20 sion develops; pressure ratios P2/P1 recorded for com-
l*S 'f\fI pressions that started at stations 1 to 7 are given in
Table 1, line 6. Here Pi is the pressure immediately
preceding the compression, and P 2 is the highest pres-
10

8 A sure attained downstream. As shown in the last line


of Table 1, the shock compression region (defined as
6 A
A ..
AJ\ tube length from Px to P2) is some 10 diameters long,
5 and appears to increase in length with Mach Number.
4
NORMAL SHOCKy
X Mfl
/ "is NTROPIC
U^ 0 R Y ) / F LOW
Observed pressure ratios are compared in Table 1,
3 lines 7 and 8, with the ones corresponding to a normal
2 f Incidentally, from the observed pressure rise in the tube and
MF it is possible to determine 10 the average friction coefficient
X. For entry MF = 5, the pressure rises (at 30 tube diameters)
to three and four-tenths times the value at entry. This corre-
1
-^- sponds to \L/D 0.07, whence X ~ 0.0023. This value agrees
0 I 2 3 4 5 M with X determined at Reynolds Numbers from 0.2 X 106 to
F I G . 7. Shock compression pressure ratio correlated to Fanno 5 X 105 at L/D 30 and at inlet Mach Numbers between 2
line and isentropic Mach Numbers, and4.11
D I F F U S E R S FOR S U P E R S O N I C W I N D T U N N E L S 621

On t h e other hand, more data, if less conclusive,


10' 1
are available on length of shock compression region.
8
ft M.I.T.- NEUMANN 8 LUSTWERK (1951) I t was obtained directly from the above t u b e experi-
M=2
A N.A.C.A.- COHEN S VALERINO (1950) ments (Table 1, b o t t o m line), as well as indirectly from
4
tunnel diffuser tests, in which the length of constant-
area diffuser e n t r y was determined to give m a x i m u m
1-9 9"x 8"
pressure recovery.
2
On t h e basis of t u b e experiments, a specific variation
of shock compression length with M a c h N u m b e r (MF)
was suggested, 12 whereas compilation of results ob-
I0 6
tained at M a c h N u m b e r s close to 2 was to determine
8
t h e variation of compression length with Reynolds
Number. 1 3 This is seen in Fig. 8, which indicates, as
expected, smaller compression length L/D a t higher
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

2-2 2 < l" DIA. Reynolds Numbers. However, small L/D results
need further corroboration since they were obtained
2 with M . I . T . tunnels, which show higher pressure re-
2 - 0 5 ^ i-*y<i bUUAWt
2 > 2 # 2-2 2 0 - 6 2 3" SQUARE coveries t h a n other tunnels of similar geometry (cf.
iV'DIA. Fig. 10).
>5 j. _____
5 6 7 8 9 10 II L/D Whereas pressure ratio of shock compression m a y be
adequately correlated to M a c h N u m b e r MF, the length
FIG. 8. Effect of Reynolds Number on length of shock compres-
sion in constant-area ducts, at M ~ 2. and structure of t h e compression region m a y be ex-
pected to depend on M, on Reynolds N u m b e r based on
shock at M a c h N u m b e r s MF and M. T h e agreement duct diameter, and on Reynolds N u m b e r based on
is remarkably good with respect to normal shock a t MF boundary-layer thickness, f T h e last p a r a m e t e r was
(at all tube stations), the discrepancy with ( P 2 / P I ) M not included in t h e a t t e m p t e d correlations we just
increasing downstream. This is also shown in Fig. 7. mentioned.!
Although, as seen in schlieren observations of flow, T h e information available on shock compression in
the shock compression in presence of thick b o u n d a r y ducts, although inadequate in detail, clearly indicates
layers differs vastly from normal shock configuration, t h a t (1) the efficiency of shock compression is higher
the above agreement in the overall pressure ratio is per- in constant-area t h a n in divergent ducts and (2) several
haps not surprising. T h e theoretical assumptions of duct diameters are required for compression to run its
normal shock transformation are a priori m e t with full course. T h e advantages resulting from inclusion
respect to conservation of energy (adiabatic flow) and of a constant-area section a t entry or a t t h r o a t of super-
area (constant tube diameter) so t h a t , provided the sonic diffusers are examined below.
frictional forces are small in the compression region (as
is the case with separated flow), the normal shock solu- D I F F U S E R S W I T H O U T CONTRACTION
tion is strictly applicable a t tube sections between
which compression takes place, irrespective of the de- Under this heading come subsonic geometry diffusers
tails of shock transformation mechanism. Further- operating with supersonic entry velocities. Their
more, the " F a n n o line" M a c h N u m b e r MF, represent- performance is given in Figs. 9 and 10 in terms of dif-
ing a mean value t h a t corresponds to the actual energy fuser entry M a c h N u m b e r and stagnation pressure
flux and accounts for the upstream kinetic energy ratio P o / P o ' across t h e diffuser.
losses in the b o u n d a r y layer, is the correct one to use in Least efficient are diffusers without a long constant-
the calculations. area entry section and with large divergence. Cas-
F r o m the above consideration one would expect, tagna's results (6 conical) and Stewart's tests (about
in general, a good agreement between measured and 12, two-dimensional expansion), Fig. 9, are in good
predicted (on basis of MF) shock compression in con- agreement, b u t t h e M . I . T . result, Fig. 10, obtained a t
stant-area ducts, both with thick (MF <C M) and M = 1.75 with a square, 6 divergent (in b o t h planes)
thin (MF ~ M) b o u n d a r y layers. However, in t h e diffuser, falls a long way below and lies in line with
latter case experimental d a t a are scarce. M o s t p u b -
lished data, here reviewed in t h e next section, refer to f Provided a fully turbulent pipe flow has not developed.
diffusers t h a t include a divergent subsonic portion, so t The phenomenon of shock compression in ducts is analogous
t h a t pressure ratios measured across t h e m are subject to flow at the base of bodies moving at supersonic speeds in t h a t
it involves interaction of shock waves with boundary layers and
to additional subsonic diffuser losses, in t u r n influenced separated regions. The base flow phenomena were correlated
b y t h e velocity distribution following shock compres- in terms of ratio of boundary-layer thickness to base diameter by
sion in constant-area entry. Chapman. 1 8
622 JOURNAL OF THE A E R O N A U T I C A L S C I E N C E S S E P T E M B E R , 1953

P 0 /Po
PoJi /
400 / 6 CONICAL - CASTAGNA (1935)
A - 6 CONICAL-CASTAGNA (1935) / 6 (ALL WALLS) SQUARE
0 - 1 2 IN ONE PLANE, 5"x5" TUNNEL / 2-66 CONICAL NEUMANN &
R e n ~ 0 - 5 x l 0 6 STEWART /
200 LUSTWERK
U. of T. (IS)52) 10-5D ENTRY, 6 CONE
M.I.T.
/ 11-20 ENTRY, 6 SQUARE (1949 a
/ 60 ENTRY, I8"x 2 4 " TUNNEL 1951)
100 R / 6-5D ENTRY, 8"x 9 " TUNNEL

80 I / 6-7D ENTRY, I2'x 12" TUNNEL PUCKETT


J.RL.
k /
/ X
NORMAL SHOCK (THEORY) 5D ENTRY, 9"x 12" TUNNEL (1948)
60 8-50" E N T R Y - COHEN a VALERIN0 (1950)
6 DONE-
\ /
t 8-2D ENTRY-SIMONS FB 1738 (1943)
/
40 / /

20
/
/ *
a
1
1
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

10
8

6 / /
/A /
A
Vl
,f NORMAL SHOCK (THEORY)

/ /
A
/ /
4

./ /
)
2
i

1
0 2 4 6 8 10 12 M
FIG. 9. Diffusers without contraction: pressure ratio.

I 2 3 M
other M . I . T . tests made with a small divergence (2.66) FIG. 10. Diffusers without contraction:pressure ratio.
conical diffuser. f
T h e expected favorable effect of a constant-area
M=2 I0"x 3-84 n TUNNEL
section a t diffuser e n t r y was demonstrated b y NACA 1 7
Re7j~2xl06
tests of a 10- b y 3.84-in. wind tunnel, fitted with a M =
COHEN 8 VALERIN0 N.A.CA. (1950)
2 nozzle, which was followed b y 46 in. of constant-area
1-9
section joined to a 5.5 (average, in both planes) rec-
tangular diffuser. In Fig. 11, the pressure ratios re-
corded for various distances of the start of the pressure
rise (or shock compression) in t h e constant-area duct
from divergent diffuser entry are given. T h e required
pressure ratio is reduced from 1.86 to 1.65, with length
of compression in constant-area duct increased from
17 to 46 in.
In Fig. 10, pressure ratios for diffusers with a long
constant-area entry are indicated by filled-in symbols.
M . I . T . tests show smallest pressure ratios, which
closely approach the ideal normal shock values and 0 10 20 30 40 50 IN.
indicate the favorable effect of a constant-area entry. LENGTH TO SUBSONIC DIFFUSER ENTRY FROM START OF
Comparable J P L , 1 5 NACA, 1 7 and German 1 6 results lie ' SHOCK COMPRESSION IN CONSTANT AREA DUCT
FIG. 11. Effect of length of constant-area entry on diffuser
much higher, even exceeding already mentioned M . I . T . pressure ratio.
d a t a for diffusers without a long constant-area entry.
Diffusers of the geometry here considered are not
reached from larger or smaller values. J This was de-
subject to large hysteresis effects. The pressure ratio
required to obtain and maintain a given M a c h N u m b e r | In intermittent tunnels run by means of large volume reser-
in the working section is practically the same whether voirs, the pressure ratio across a tunnel diffuser decreases during
the run. With diffusers of the above type, the minimum start-
f In general (cf. below), the M.I.T. tests show a smaller pres- ing and the actual flow breakdown pressure ratios are practically
sure ratio than obtained in other comparable experiments. equal.
D I F F U S E R S FOR S U P E R S O N I C W I N D T U N N E L S 623

TABLE 2
Maximum Theoretical Diffuser Starting Contraction Ratio and
Pressure Recovery ( 7 = 1.4)
M tmax (Po/Po'ty.maz.
1.0 1.000 1.00
1.2 1.023 1.00
1.4 1.068 1.01
1.6 1.120 1.04
1.8 1.170 1.11
2.0 1.216 1.20
2.5 1.316 1.59
3.0 1.391 2.27
3.5 1.445 3.33
4.0 1.488 5.00
4.5 1.520 7.14
5.0 1.543 10.75
6 1.577 21.79
7 1.598 41.15
8 1.614 73.53
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

9 1.624 124.69
10 1.632 202.43
CO 1.667 CO

termined for M . I . T . circular diffusers 12 a n d also in


Stewart's 1 4 tests; of t h e two values of pressure ratios
shown in Fig. 9 a t M = 5.5 a n d 6.5, t h e higher (by
less t h a n 10 per cent) ones were required to start t h e
flow.

ENTRY ANGLE
D I F F U S E R S W I T H CONSTANT CONTRACTION
DEGREES

Diffusers grouped in this category are characterized F I G . 12. Typical effects of contraction and model on diffuser
pressure ratio.
by a convergent entry section in which appreciable
compression is achieved a t supersonic speeds. T h e y
are therefore usually known as supersonic diffusers, retical \pmax.- T h e o p t i m u m contraction, a t which
in distinction t o purely divergent, subsonic ones. T h e smallest pressure ratio ( ~ 3) is required t o initiate t h e
final compression to subsonic velocity occurs in super- flow, is only slightly smaller t h a n t h e limiting contrac-
sonic diffusers b y means of shock compression of t h e tion.
type already considered near t h e minimum or t h r o a t Also in Fig. 12, a curve is included for a variable dif-
diffuser cross section, on its downstream side. fuser, whose contraction can be increased after super-
One of t h e fundamental design parameters is t h e dif- sonic flow a t entry h a s started. Comparison of this
fuser contraction ratio \p = Ai/A*', where A1 = dif- curve with t h e one for starting or, in effect, constant
fuser entry or nozzle exit area a n d A*' = diffuser contraction diffuser, shows t h a t a t t h e limiting contrac-
minimum or t h r o a t area. tion of 1.4, a pressure ratio of only 2.4 is required t o
T h e maximum possible value of \f/ a t a n y entry dif- maintain t h e flow compared with t h e starting pressure
fuser M a c h N u m b e r was derived theoreticallyf from ratio of 3.
considerations of starting t h e flow in a supersonic I n Figs. 13 a n d 14, experimental contraction a n d
nozzle-supersonic diffuser system and corresponds to a pressure ratios of constant geometry diffusers are
normal shock a t t h e diffuser entry M a c h N u m b e r a n d given in terms of entry M a c h N u m b e r . A t M < 3.5,
sonic flow in t h e diffuser t h r o a t . T h e theoretical re- maximum contraction ratios for starting, J Fig. 13,
sults are given in Table 2, where \f/max. denotes t h e agree closely with t h e theoretical \pmax. valuesa result
m a x i m u m theoretically permissible contraction a n d somewhat unexpected in view of t h e already considered
(Po/Pfo)^max. is t h e corresponding stagnation pressure large discrepancies between theoretical a n d actual
ratio across t h e diffuser on t h e assumption of isentropic "normal shock" compression in ducts. T h e agreement
flow except for shock located a t throat. is presumably due t o t h e favorable influence on t h e
Typical variation of pressure ratio with contraction boundary layer of t h e accelerated subsonic flow (during
ratio for constant geometry diffusers is shown in Fig. starting) downstream of t h e shock compression, in t h e
12. T h e t o p curve marked ''starting'' shows t h a t diffuser contraction. I n general, a smaller contraction
supersonic flow a t diffuser entry cannot be obtained is possible with diffusers having a long constant-area
for $ > 1.4, irrespective of t h e pressure ratio applied. throat, because of additional losses in t h e t h r o a t sec-
This limiting contraction closely agrees with t h e theo- tion.
t See reference 2. This criterion was first established and J Practically equal to optimum contraction (for minimum pres-
experimentally checked in Germany. 7 , 16 sure ratio), as pointed out above.
624 JOURNAL OF THE A E R O N A U T I C A L S C I E N C E S - S E P T E M B E R , 19 53

k
In square or rectangular tunnels the supersonic dif-
/ fuser contraction is usually two-dimensional and the
1
1 I ( I I I question of optimum contraction angle arises. Since,
0 ~ 3 ENTRY, 7-l"x7-l" TUNNEL, R e _ ~ l - 6 x l 0 6
DI6GINS N.O.L. (1951) for a given contraction ratio, shock losses increase and
D ~ 9 ENTRY, 47"x47" TUNNEL, R e - ~ 3 x l 0 6
WEGENER a LOBB N.O.L. 11952) D frictional wall losses decrease with increasing contrac-
A ~ 2 0 ENTRY, 2"x2-5" TUNNEL, Re ~ l - 6 x l 0 6
HEPPE GALCIT (1947) tion angle, an optimum contraction angle might be ex-
O

15 ENTRY, f DIA. TUNNEL
15 ENTRY, 10-7 D THROAT l" DIA
NEUMANN a pected to exist. The limited data availablef indicate
3 . LUSTWERK MAX. CONTRACTS
TUNNEL M.I.T. /fa)R STARTING. that at M < 4 small contraction angles, of the order of
10 ENTRY, 6-6 D THROAT, l-25"x|-25" ! (1949 a 1951)
a -4-a4-o-
D ~X\V~
3 to 4, are favorable, higher pressure ratios being

V
Re

^ required with large contraction angles, t This is in


2 OPTIMUM
CONTRACTION
agreement with further evidence obtained from tests of
variable diffusers.
A,
A
V 'max VARIABLE CONTRACTION DIFFUSERS
*? M
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

Following the establishment of supersonic flow, a


1 b
further contraction is possible with variable diffusers,
F I G . 13. Maximum contraction of constant geometry diffusers the pressure ratio at which the flow breaks down being
or for starting. much lower than the starting pressure ratio. In order
to take advantage of variable diffusers, a tunnel drive
capable of providing momentarily (for starting) a
higher than operating pressure ratio is usually required.
Intermittent tunnels run by means of pressurized
and/or evacuated reservoirs lend themselves very well
to this type of operation.
Typical characteristics of variable contraction dif-
fusers are shown in Fig. 12 at M = 2.83. Comparison
of "starting" and "variable contraction" curves indi-
cates that in the latter case much larger contractions
and smaller pressure ratios are possible. The optimum
contraction, at which a minimum pressure ratio is ob-
tained, is significantly smaller than the maximum pos-
sible contraction.
Experimental results for variable diffusers are col-
lected in Figs. 15 and 16. From Fig. 15 the maximum
and optimum contraction and the effect of model
mounted in the working section on optimum contrac-
tion can be estimated. Diffuser pressure ratios are
given in Fig. 16. The minimum values indicated by
the broken line are somewhat smaller than the
theoretical ones for diffusers having i/max. contrac-
t 1-5 2 3 4 5 6 7 8 9 10 M
tion.
FIG. 14. Diffusers with constant contraction:pressure ratio.
With regard to the optimum contraction angle of
variable diffusers, it appears that it increases with the
At Mach Numbers between 7 and 10 results from Mach Number. In the NOL tests 19 (diffuser without
only one set of experiments20 are available and give long constant-area throat), an optimum throat loca-
contraction ratios much in excess of ypmax.- No expla- tion was determined, while the total (including con-
nation of this discrepancy with results at lower Mach vergence and divergence) diffuser length was kept con-
Numbers can be offered at present. stant. The optimum throat position was found to be
The diffuser pressure ratios are given in Fig. 14 and, at 5.35> from the diffuser entry for M = 2.48 and 2.83
in general, closely follow the theoretical normal shock and at S.9D for M = 4.92, where D is the height of
values. Hysteresis effects, already considered, are square tunnel working section. The corresponding
indicated in Fig. 14 by arrows, which connect pressure
t Insufficient to separate Reynolds and Mach Number effects.
ratios required for starting and for maintaining the
t This statement is based on comparison of NOL d a t a with
flow. Although long-throat diffusers require a smaller GALCIT results, Fig. 14. T h e M . I . T . results show lower pres-
operating pressure ratio, their starting performance is sure ratios with large contraction angles but also indicate the de-
similar to diffusers without a long throat. sirability of a small angle. 12
D I F F U S E R S FOR S U P E R S O N I C WIND T U N N E L S 625

entry angles were 4.5 and 5.6 at optimum contrac- than optimum, parallel-throat diffuser gave slightly
tions for the two lower Mach Numbers. (about 10 per cent) smaller pressure ratios.
The above results were confirmed by R.A.E. tests In two investigations the effect of Reynolds Number
at M = 2.48 with incomplete diffuser,23 Fig. 17, which on diffuser performance was found to be small at hyper-
show that contraction ratio (and, hence, pressure re- sonic speeds. Bertram 22 found no change in the mini-
covery) increases with contraction angle decreasing. mum diffuser pressure ratio with the variation of
Galcit tests, 21 Figs. 15 and 16, indicate relatively
small optimum contraction ratios and pressure recov-
eries, which can be attributed to an excessively large p.

diffuser entry angle (~28). k


1 1 1 1 1 I
On the other hand, apparently extremely low pres- R
DIGGINS N.O L (1950
sure ratios were obtained in M.I.T. tests at similar <s D ~ I 3 ENTRY, 4 7 " x 4 7 " TUNNEL, Re Q ~3 x I0 6

Mach Number, Fig. 16, in spite of a 20 contraction 4


WEGENER 8 L0BB, N.O L (1952)
' ' 5
angle. It is not certain whether this was due to a long HEPPE GALCIT (1947) i
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

3
1 ~ I 8 ENTRY, 3 D THROAT, l l " x l l " TUNNEL, Re 0 M>-3 to 2-6 x 10
constant-area throat used in these tests. BERTRAM N AC A (1950)
/
2 2 0 ENTRY 10 D THROAT, l-3"xl-3" TUNNEL Re D ~0-2xl0 6 -1 /
Both NACA and NOL tests indicate that at hyper-

sonic speeds (say M > 5) much larger diffuser entry


NEUMAN a LUSTWERK M.I T (1951)
/ L
angles (between 13 and 18) should be used. The 100 6n
/
following results were obtained at NOL20 with two dif- A
f /
ferent entry angles, at M = 7.2, ReD 2 X 106: 6
WITH MODEL y
/
5
" a SUP HUH 1 A
//
Diffuser entry angle = 7 12.9
Optimum \p =5.5 5.5 D
3
Length to throat from diffuser entry = 6.6D 3.6> >
NORMAL SHOCK (THEORY) ^ V '
Subsonic diffuser angle = 3 3
N0RMAL SHOCK AT Yma x
ft
/
Min. P o / i Y = 58 36 a ISEN TROPIC FLOVl v r
/ / >
In the same series of experiments it was found that 10
/y /
the effect of subsonic diffuser divergence (up to 5.5) 7
was negligible and that a constant-area throatt (3D fs
5 ff
Ay
length) had no effect on optimum contraction ratio and 4 //
minimum pressure ratio; at contraction ratios smaller f /
//
_,
WITH MODEL
/>
a SUPPORT ^
t As pointed out in reference 20, in hypersonic tunnels oper- 2 >V|
ating at high stagnation temperature, it is advantageous to have
a short diffuser throat in order to reduce regions of high heat-
transfer rates and thus ease cooling problems. 1 ^~

FIG. 16. Variable geometry diffusers:pressure ratio.

2 3 4 5 6 7 8 9 10 M 2 4 6 8 10 12 14 '
FIG. 15. Variable geometry diffusers:contraction ratio. Effect of angle of contraction on contraction ratio.
ReD from 0.5 to 3 X 106, a t M ~ 6.86 a n d ^ = 2.8. 75 and No. 7-8, pp. 734-778, 1935; Trans. No. 366, A M C ,
T h e N O L results, summarized below, indicate increase Dayton, Ohio, 1943.
2
Lukasiewicz, J., Supersonic Diffusers, R.A.E. (Farnborough)
with ReD in t h e m a x i m u m possible contraction but
Report Gas 8, 1946, and British A.R.C., R. & M. No. 2501,
small change in t h e pressure r a t i o : H.M.S.O., London, 1952.
3
M = 7.2 To = 15C. Naumann, Efficiency of Diffusers at High Subsonic Speeds,
Poat. 5 10 20 30 F.B. 1705, Germany, 1942, and British Trans. MAP-VG-RT321.
4
ReD X 10~6 1 2 4 6 Goethals, R., Contribution a Vetude des diffuseurs en vue de
i, 6.6 7.5 9.2 10.4 Vapplication aux souffleries aerodynamiques subsoniques, P. Sc.
Po/iY . 36 33.5 Tech. No. 243, Paris, 1950.
5
Squire, H. B., Experiments on Conical Diffusers, R.A.E.
(Farnborough) Report Aero 2216, 1947.
CONCLUSIONS 6
Castagna, A., Experimental Research on the Transformation
(1) A b o u t 88 per cent of theoretical pressure rise is of Energy of a Gas Flowing in a Pipe, Atti della Reale Accademia
delle Scienze di Torino, Vol. 70, pp. 284-317, 1935.
obtained in subsonic small angle (7 cone) diffusers a t 7
Eggink, H., Flow Structure and Pressure Recovery in Super-
M < 0.9. sonic Tunnels, F.B. 1756, Germany, 1943.
8
(2) Shock compression in d u c t s m a y involve com- Payman, W., and Shepherd, F., Explosion Waves and Shock
plete flow d e t a c h m e n t in divergent d u c t s or flow sepa- Waves, VI, The Disturbance Produced by Bursting Diaphragms
Downloaded by PENNSYLVANIA STATE UNIVERSITY on June 1, 2014 | http://arc.aiaa.org | DOI: 10.2514/8.2763

ration in c o n s t a n t - a r e a ducts with multiple shock sys- with Compressed Air, Proc. Roy. Soc. London, Series A, Vol. 186,
1946.
tems extending over several diameters of d u c t length. 9
Glass, I. I., The Design of a Wave Interaction Tube, Institute
T h e l a t t e r t y p e of shock compression can be observed of Aerophysics, University of Toronto, Report No. 6, 1950.
10
in wave-speed camera records of shock-tube flow. Lukasiewicz, J., Adiabatic Flow in Pipes, Aircraft Engineer-
Shock compression losses decrease with d u c t divergence ing, February and March, 1947.
11
decreasing a n d s t a t i c pressure ratios observed in con- Keenan, J. H., and Neumann, E. P., Friction in Pipes at
Supersonic and Subsonic Velocities, NACA T N No. 963, 1945.
stant-area tubes correspond (at least in one case) 12
Neumann, E. P., and Lustwerk, F., Supersonic Diffusers for
to theoretical normal shock values t a k e n a t m e a n or Wind Tunnels, J. Appl. Mech., p p . 195-202, June, 1949.
13
F a n n o line M a c h N u m b e r s . I n general, t h e p h e n o m - Neumann, E. P., and Lustwerk, F., High-Efficiency Super-
enon of shock compression in c o n s t a n t - a r e a d u c t s can sonic Diffusers, Journal of the Aeronautical Sciences, Vol. 18,
be expected t o depend on M a c h N u m b e r a n d R e y n o l d s No. 6, pp. 369-374, June, 1951.
14
Stewart, J. D., A Preliminary Investigation of the Effects of
N u m b e r s based on t u b e d i a m e t e r a n d b o u n d a r y - l a y e r
Condensation and Slip Flow in the UTIA 5" x 5" Supersonic Wind
thickness (for a given b o u n d a r y - l a y e r flow). Tunnel, Institute of Aerophysics, University of Toronto, Report
(3) Pressure recovery of diffusers w i t h o u t contraction No. 22, 1952.
15
is improved b y use of a long c o n s t a n t - a r e a e n t r y b u t is Puckett, A. E., Design and Operation of a 12 inch Supersonic
always smaller t h a n t h e theoretical normal shock (i.e., Wind Tunnel, Institute of the Aeronautical Sciences, Preprint
No. 160, July, 1948.
P i t o t t u b e ) recovery. 16
Simons, F. P., Investigations of Diffusers for Supersonic Wind
(4) A t M a c h N u m b e r s below 4, m a x i m u m contrac- Tunnels, F.B. 1738, Germany, 1943.
tion of c o n s t a n t geometry diffusers (or s t a r t i n g con- 17
Cohen, C. B., and Valerino, A. S., Investigation of Operating
traction of variable diffusers) is closely predicted b y t h e Pressure Ratio of a Supersonic Wind Tunnel Utilizing Distributed
simple t h e o r y . A t o p t i m u m contractions (which are Boundary-layer Suction in Test Section, NACA R M E50H04,
practically equal t o m a x i m u m contractions) normal 1950.
18
Chapman, Dean R., An Analysis of Base Pressure at Super-
shock pressure recoveries are achieved, a l t h o u g h a p -
sonic Velocities and Comparison with Experiment, NACA Report
preciably larger s t a r t i n g pressure ratios are usually re-
No. 1051,1951.
quired a t M > 2. Small angles of two-dimensional 19
Diggins, J. L., Diffuser Investigations in a Supersonic Wind
contractions, of t h e order of 3 t o 4 , are favorable a t Tunnel, NOL NAVORD Report 1570, 1951.
M < 4 a n d increase t o a b o u t 9 a t M = 7 t o 10. 20 Wegener, P. P., and Lobb, R. K., N.O.L. Hypersonic Tunnel
(5) W i t h variable diffusers, o p t i m u m contraction in- No. 4, Results IIDiffuser Investigation, NOL NAVORD Re-
port 2376, 1952.f Also An Experimental Study of a Hypersonic
creases with M a c h N u m b e r a n d is appreciably larger
Wind-Tunnel Diffuser, Journal of the Aeronautical Sciences,
t h a n s t a r t i n g or \pmax. contraction but smaller than Vol. 20, No. 2, pp. 105-110, February, 1953.
maximum possible contraction. C o n t r a c t i o n angles of 21
Heppe, R. R., Investigation of a Variable Geometry Supersonic
the order of 5 are best a t M < 3 a n d increase t o 15 Diffuser, GALCIT Thesis, 1947.
22
and more a t M = 7 t o 10. M i n i m u m pressure ratios, Bertram, M. H., Investigation of the Pressure Ratio Require-
appreciably smaller t h a n t h e n o r m a l shock values, are ments of the Langley 11 inch Hypersonic Tunnel with a Variable
Geometry Diffuser, NACA R M L50113, 1950.
obtained a t M > 2 a n d a m o u n t t o only half t h e n o r m a l 23
Eggink, H., The Improvement in Pressure Recovery in Super-
shock stagnation pressure r a t i o a t M ~ 7. Although
sonic Wind Tunnels, R.A.E. (Farnborough) Report Aero 2326,.
performance of variable diffusers deteriorates when S.D.34, 1949.
models are m o u n t e d in t h e working section, it is ad-
vantageous t o use t h i s diffuser t y p e in hypersonic t u n - f Hypersonic diffuser test data quoted from this reference
nels. is based on experiments carried out with air at room stagnation
temperature and therefore in presence of air condensation. Mach
Numbers were obtained from Pitot/stagnation pressure ratios
REFERENCES (insensitive to condensation) and it was ascertained t h a t at M =
1
7.2 diffuser performance was only little affected by condensation,
Croceo, L., Gallerie aerodinamiche per alter velocita {High the minimum operating pressure ratio being about 10 per cent,
Speed Wind Tunnels), L'Aerotechniea, Vol. 15, No. 3, pp. 2 3 7 - larger for condensation-free flow.

Вам также может понравиться