Full text of DOI:10.2514/2.2708

JOURNAL OF AIRCRAFT
Vol. 37, No. 6, November December 2000
–
Flow Stabilization in a Transonic Wind
Tunnel with a Second Throat
Thombi Layukallo¤ and Yoshiaki Nakamura†
Nagoya University, Nagoya 464-8603, Japan
A second throat has been used to stabilizethe  ow in the test section of an induction-driventransonicwind tunnel.
Different second throat conŽ gurations were attempted by employing simple wall-Ž xed  aps and a centerline-Ž xed
strut, and their efŽ ciency based on the duration of stable test section  ow was examined. Generally, a longer
duration of stable  ow could be achieved when the tunnel was operated with a second throat than without one.
Moreover, a longer run time was achieved with the wall-Ž xed  aps than with the center strut. The performance
of the center strut was so poor that it raised concerns about the installation of the sting support, which is usually
installed in the same manner as the center strut. It was also found that when the  aps were installed vertically
on the sidewalls the low-momentum reentry  ow from the plenum chamber reduced the effectiveness of the  aps
in choking the  ow. The performance was greatly improved by repositioning the  aps horizontally and directly
behind the reentry region. However, there was an upper limit to the effectiveness of the  aps in choking the  ow.
Beyond a certain injector total pressure the effective cross-sectional area of the diffuser at the injector’s location
was reduced by the jet, turning it into the effective throat area. The cross-sectional area at the second throat is
rendered ineffective, causing the Mach number in the test section to temporarily drop.
Introduction
is usually less than one. Installing a second throat upstream of the
injector seems likely to jeopardize the performance of the injector
as a result of losses caused by shock waves and/or  ow separa-
tions occurring at the second throat location. As a consequence,
the efŽ ciency advantageover blowdown tunnels may be lost. How-
ever, because the second throat providesa simple yet quite effective
means to stabilize the test section  ow, its use even in this type
of tunnel remains attractive and worth considering. In this study
the use of a second throat in stabilizing the  ow in the test section
of an induction transonic wind tunnel is reported, and several fac-
tors pertaining to its efŽ ciency such as shape and positioning are
discussed.
ECAUSE of the large amount of power required to produce
transonic  ows in wind tunnels, intermittently operated tran-
B
sonic wind tunnels are still the preferred choice over continuous
ones. However, becauseintermittentwind tunnelsare operatedwith
a limitedamountof airsupply,theshortdurationofstable ow is one
of the biggestconcernsin the operationof these tunnels. Therefore,
the tunnel has to be operated such that the required Mach number
can be obtained quickly and held constant for as long as possible.
One of  ow control devices that are widely used in transonic
wind tunnels is the second throat, although it may be referred to
by different names in different facilities.1¡3 The objective of using
the second throat is to choke the  ow at a location downstream of
the test section to effectively “freeze” the  ow in the test section.
As a  ow control device, second throats have been very effective
in blowdown wind tunnels, such as the tunnel of Ref. 1. They have
also found wide use in continuouswind tunnelsto suppressthe test-
section turbulence level, such as in the tunnel described in Ref. 4.
However, the use of second throats is not commonly heard of in
induction-typewind tunnels.
In inductiontunnels ow in the test sectionis inducedby injecting
high-pressureair into the tunnelat a locationdownstreamof the test
section. The most common way to control the  ow is by changing
the injector total pressure and/or the injector slot width.2,5¡7 The
induction tunnel of Ref. 2 does employ a second throat of a novel
design as part of the  ow control system. However in that particular
tunnel the injector is located in the return leg of the tunnel circuit,
considerablyfar enoughdownstreamto be unaffectedby the second
throat.
Mass EfŽ ciency of Induction Wind Tunnel
g
m
Mass efŽ ciency
of an induction-driventransonic wind tunnel
is deŽ ned in Ref. 6 as the ratio of the mass of useful air induced in
the test section to the total mass of air blown through the injectoror
gm
m t m t
= Ç TS Ç
/
t j f
m
m
where Ç TS is the induced test-sectionmass  ow, Ç j is the jet mass
t
t
 ow, is the useful test time, and is the total time required to
t
f
discharge air into the tunnel.
Rewriting this in terms of Mach number yields
p
¡
¢
p
c +
c
1)/ 2(
¡(
]M2
}
|
+
P A
t
Tt
c R M
[ c
¡ 1)tt
(
)
{1
(
¡ 1) 2
/
/
gm
=
m t
Ç
j
f
P
T
where t and t aretotalpressureandtotaltemperature,respectively,
R
In more conventional induction tunnels the injector is located
a short distance behind the test section in order to achieve high-
injectorefŽ ciencyby taking advantageof the high-speed ow com-
ing out of the test section.Tunnel mass efŽ ciency, which is the ratio
of the mass  ow of air through the test section to that through the
injector, in such a tunnel has been observed to be between two to
seven,6 in contrast to the efŽ ciency of blowdown tunnels, which
and is the gas constant for air.
In the present experimentthe amount of total mass of air used is
calculatedfrom the initial and Ž nal pressuresin the air reservoirand
is given by
m j
V RT
/
P
init
P
Ž nal
=
(
¡
)
r
m
V
is the total mass blown by the jet, is the volume of the
where
reservoir,
j
T
P
init
is the reservoir temperature,
is the initial reser-
r
P
voir pressure, and
is the Ž nal reservoir pressure. Using this
Ž nal
Received 23 August 1999; revision received 25 June 2000; accepted for
deŽ nition, the efŽ ciency becomes
c
publication 12 July 2000. Copyright ° 2000 by the American Institute of
p
¡
¢
p
Aeronautics and Astronautics, Inc. All rights reserved.
c +
c
1)/ 2(
¡(
]M2
}
|
+
P A
t
Tt
c R M
[ c
(
¡1)tt
(
)
{1
¡ 1) 2
/
/
¤Graduate Student, School of Engineering. Student Member AIAA.
†Professor, Department of Aerospace Engineering. Member AIAA.
gm
=
V RT
P
init
P
¡
(
r )(
)
/
Ž nal
1033

1034
LAYUKALLO AND NAKAMURA
Fig. 1 Schematic of Nagoya University transonic tunnel.
Wind Tunnel
The presentexperimentwas conductedin the transonicwind tun-
nel at Nagoya University. The test section has a rectangular cross
section, which is 30 cm wide by 40 cm high. The top and bot-
tom walls are slotted with a 7% total open area ratio. Typical two-
dimensionalmodelscan be mountedon two sidewallturntables,and
three-dimensionalmodels can be accommodated by a circular-arc
sting support.
The  ow in the test section is induced by injectinghigh-pressure
air through an annular injector located downstream of the test sec-
tion, as schematicallyshown in Fig. 1. The injector has a Ž xed slot,
and the test section-to-injector area ratio is approximately 19. A
breather is located in the return leg whose function is to keep the
totalpressureand totaltemperaturein the test sectionat atmospheric
levels. As the energy source to drive the tunnel, high-pressureair is
stored in a 12 m3 reservoir, which is capableof storingup to 40 atm
of air.
Fig. 2 Schematics of test section and diffuser transition.
Usually the Mach number in the test section is set by manually
controlling the injector total pressure through the adjustment of a
pneumatic valve located between the air reservoir and the tunnel
circuit. However, because of the inherent delay in the response of
the tunnel to the opening/closing of the valve, controlling the test-
section Mach number has proved to be a very difŽ cult task. As
a result, much air is wasted in the process. More importantly, the
Flat  aps
Convex  aps
Fig. 3 Flap proŽ les.
–
achievable stable  ow duration is very short: 1 1.5 s at best.
Because of the slow response of the tunnel, the use of full-
automaticcontrolin this tunnelis consideredbeneŽ cial only if done
in conjunctionwith a variable-slotinjector.Such a system,however,
would require major structural changes to the tunnel. Therefore, as
(
)
an alternative and a less expensive solution, an attempt was made
to gaina quickerand easier controlof the Mach number by resorting
to the use of a second throat.
Second Throat
The second throat was mainly devised by using simple  aps,
which were manufactured from readily available materials. These
 aps were initially installed vertically at the diffuser entrance on
(
)
Fig. 4 Schematic of center strut positioning top view .
(
)
bothsidewallsandlessthan1 mupstreamoftheinjector Fig.2 .The
cross-sectionalarea of the diffuser from the entrance to a location
0.4 m downstream is constant at 0.1385 m2. However, the cross-
sectionalshape changes from rectangular to fully circular.
minimizeinterferencesto the  ow alongthe walls. If successful,this
conŽ gurationcould be integratedwith the stingsupportmechanism.
The  aps were Ž xed to the wall by means of simple brackets and
couldbe de ected at differentanglesto vary the cross-sectionalarea
at this location. To evaluate the effects of different ap wakes at the
same Mach number, two different  ap proŽ les were selected:  at
Experiment
In the experimentthe tunnel was mostly operated with the injec-
tor blown at the maximum rate, and the observation was made that
choking condition at Mach numbers greater than 0.7 could be ob-
tained in all test cases of interest if the reservoirpressure was above
16 atm absolute. The experiment was therefore conducted mainly
with an arbitrarily chosen initial reservoir pressure of 24 atm abso-
lute. For calculationof the total mass used,a Ž nal pressureof 11 atm
was used in all cases. This ensures that the same total mass of air
was used and allows a comparisonbetween different conŽ gurations
to be made directly from the duration of stable  ow.
Thetest-sectionMachnumberwasdeterminedbyusingtheempty
tunnel total and static pressure data, which were measured in the
settlingchamberandalongthe test sectionsidewalls,respectively.In
addition,thepressureinsidetheplenumchamberwasalsomeasured.
This plenum pressure was used as the reference pressure in tests
involving a model inside the test section.
(
)
 aps and convex  aps Fig. 3 . Both had a span length of 45 cm and
a chord length of 6 cm. The convex  aps were manufactured from
a commercially available steel pipe, which was cut to give a 16%
thick symmetrical proŽ le. Later in the course of the experimentthe
 aps were resizedand repositionedonto the upper and bottomwalls
of the diffuser for reasons that will be discussed later.
An attempt was also made to devise a second throat by installing
a single strut at the same downstream location as the wall  aps
(
)
but at the center of the diffuser with respect to the width Fig. 4 .
When installed, it spanned almost the entire height of the cross
section. The strut had an elliptical cross section whose dimensions
were 4 cm along its major axis and 2 cm along its minor axis.
Because the annular injector covers the whole circumference of
the diffuser wall, the main intention in using the center strut was to

LAYUKALLO AND NAKAMURA
1035
Results and Discussion
The Mach-numberhistory typicallyresulting from the use of the
second throat is shown in Fig. 5, where the results from the verti-
cal/convex  ap case are given as an example. As seen in the Ž gure,
a plateau of constant Mach number, which indicates the choking
period, is achieved for each  ap de ection. For this conŽ guration
Fig. 7 Pressure distribu-
tion along the midspan of
the convex  ap.
–
a stable test-section  ow was achieved 3 3.5 s after the tunnel was
started, depending on the Mach number.
At Mach number 0.86, which corresponds to 0-deg  ap de ec-
tion and with an initial reservoir pressure of 22 atm, the stable  ow
lasted for about 5 s. This time lapse became shorter as the  ap inci-
dence was increased,but still longer than what had been achievable
without a second throat. Althoughnot shown here, similar response
characteristicswere also observed for the cases of the  at  aps and
the center strut. Provided there is enough pressure in the air reser-
voir, running the tunnel without a second throat could also produce
a signiŽ cant amount of stable  ow time in the test section. This
phenomenon was a result of the aerodynamic choking caused by
the injector jet and is similar to the behavior observed in the tunnel
of Ref. 7. In this case choking occurs only at a certain supersonic
Mach number and cannot be changed at will.
Figure 6 summarizes the resulting stable  ow periods achieved
with the different conŽ gurations used in the experiment. The ma-
jor results obtained in the experiment are briey described in the
following sections.
Fig. 8 Static pressures in the test section and on the convex  aps.
Center Strut ConŽ guration
Vertical Flap ConŽ guration
As seen in Fig. 6, in the case of the verticalconŽ gurationthe more
aerodynamicallyshaped convex  ap proŽ le yielded only marginal
improvementsin the stable  ow durationwhen compared to the  at
one. One reason for this is believed to be the existence of shock-
induced separations, which were observed to occur on the surface
of the convex  aps. This is clearly seen in the pressure distribution
over the  ap, as depicted in Fig. 7, at three different time instants.
At each instant shown the static-to-total-pressure ratio behind the
The center strut yielded the poorest performanceamong the con-
Ž gurationstested.Becauseofthepoorresult,nodetailedexplanation
was pursued. However, it is strongly believed that the strut might
have generated shock waves that are strong enough to separate the
boundarylayers not only on the surfaceof the strut itself but also on
the diffuser sidewalls. Furthermore, like the vertical  ap conŽ gura-
tion the strut was also exposed to the reentry  ow effects. The com-
bined effect of all of these factorsseems to have resulted in the poor
performance observed here. Therefore, despite the advantage of
being able to integrate it with the sting support mechanism, the
center strut appeared to offer no other advantages that would make
it worth investigatingfurther.However, this raisedconcernsoverthe
installationof the sting support inside the tunnel, which is normally
installed at approximatelythe same location as the center strut. Al-
though the results are not shown here, during the investigation the
existenceof the sting support strut in the tunnel also proved to have
a similar degrading effect as the center strut. In most wind tunnels
the presence of the sting support inside the tunnel is almost always
inevitable. If choices are available, the degrading effect can proba-
bly be minimized by using a half-span, instead of a full-span, sting
support strut.
(
)
shock was below the sonic critical value, indicating that the  ow
behind the shock was separated. The other reason is believed to be
the effects of the reentry  ow, which will be discussed later.
Effects of Reentry Flow
(
Fig. 5 Tunnel response caused by vertical/convex  aps Ptank
22 atm .
=
)
Figure 8 shows the time history of static pressures in the test
section and on the  ap surface of the vertical  ap conŽ guration.
The data from the  ap were taken at two spanwise locations: at the
 ap midspan section and at a section that was directly exposed to
the reentry  ow. From these two locations only the data from the
(
)
= 0.5 are plotted on the Ž gure. At an
x c
/
 ap’s midchordlocation
instantofapproximately8 s intothe run, distinguishedpressurerises
are observed inside the test section and behind the reentry region,
whereas only slight changes are observed at the center section of
the  ap. This indicates that there is a close association between the
test section and the reentry  ows.
Measurementsof totalpressureby using a pitottubeat the second
throat location revealed much reduced total pressure levels in the
(
)
y
vicinityofthereentryregion = 20and22cmonFig. 9 .Moreover,
oil- ow visualization also revealed that there was a high degree of
(
)
three-dimensionalityin thisregion Fig. 10 . The effectof thisthree-
dimensionality is thought to be analogous to the effect of sweep
angle on an airplane wing, namely the delay in the onset of sonic
( )
Fig. 6 Duration of stabletest-section  ow; indicates initialreservoir
pressure.

1036
LAYUKALLO AND NAKAMURA
which continue to converge behind the  aps. These walls would
promote reattachment of the  ap wakes, hence preventing them
from reaching the injector.
As depicted in Fig. 6, the stable time yielded by this conŽ g-
uration is up to two times longer than the vertical conŽ guration.
–
Stable  ow was achieved 2 3 s into the run, which is slightly faster
than the vertical conŽ guration. Unlike the vertical conŽ guration,
within a certain Mach-number range the duration of stable  ow
(
)
actually increases as the Mach number throat area is reduced.
The stable  ow duration stretches from about 11.4 s at the maxi-
(
)
M
M
mum Mach number tested
= 0.867 to 12.6 s at
= 0.77 be-
= 0.7. The time increase from
= 0.867 to 0.77 is considered to be a direct beneŽt from having
M
fore dropping to about 10 s at
M
Fig. 9 Total pressure development at the second throat; y is vertical
distance from tunnel centerline.
the wakes reattach to the walls, as shown schematically in Fig. 12.
Beyond a certain  ap angle the wakes do not reattach and, conse-
quently, are able to interfere with the  ow entrainment process at
the injector. This behavior was conŽ rmed by oil- ow visualization
results.
When positioned horizontally, the differences between the  at
 apsand theconvex apsaremuchmorepronouncedthaninthecase
of the vertical conŽ gurations. As expected, the convex  aps show
far superior performance than the  at  aps, most likely because of
the weaker wake interferencesto the injector jet.
Area-Mach-Number Relation
Figure 13 compares the area-Mach-numbercurve resulting from
the experiment and that of the one-dimensionaltheory as given by
the equation
³
³
´
´
³
´
c +
c
2
(
1)/ ( ¡ 1)
A
A¤
c
1
M2
2
+
¡ 1
M2
+
1
=
c
1
2
Fig. 10 Oil- ow pattern on  ap in the vertical/convex conŽ guration.
A
A¤
is the second
where is the test-section cross-sectional area,
M
throat cross-sectionalarea, and is the test-sectionMach number.
For a given arearatio the experimentaltime-averagedMach number
is repeatable within a 0.2% band.
As depictedin the Ž gure,horizontalinstallationof the aps brings
the data curve much closer to the theory, suggestingthat the  ow is
Fig. 11 Horizontal  ap installation.
Fig. 12 Wake reattachment behind the horizontal  ap.
 ow. This means that for a given test section Mach number the  ow
in this vicinity cannot maintain sonic condition for as long as the
 ow along the centerline. The combined effects of the low total
pressure and the high degree of  ow angularityrendered the  ow at
the second throat location highly two- or even three-dimensional.
This ultimately led to the early termination of choking, hence a
stable test-section ow.
Horizontal Flap ConŽ guration
To get around the problem of the reentry  ow, it was decided
to reposition the  aps horizontally and directly behind the reentry
area, as shown in Fig. 11. To do this, the  aps actually had to be
shortened.
Two advantageswere soughtby employingthe horizontalconŽ g-
uration. First, the effects of the reentry  ow, as seen in the vertical
 ap cases, should not be as pronounced because the reentry  ow
should now be able to return to the mainstream more quickly. Sec-
ond, the  aps can take advantage of the upper and bottom walls,
Fig. 13 Area-Mach-number relation.

LAYUKALLO AND NAKAMURA
1037
more uniformly distributed throughout the second throat area than
in the vertical conŽ guration case.
caused by the jet occurs and renders the second throat ineffective.
As the jet width continuesto widen, the now-effectivethroat area at
the injector location gets even smaller, and the test-section Mach
number actually starts to drop. This continues until the injector
pressure reaches a maximum. Beyond this point the jet width starts
to narrow down, and the effective throat area at the injectorlocation
starts to widen again, causing the test-sectionMach number to start
increasing again. The Mach number continues to increase until the
cross-sectionalareaat the secondthroatagain becomesthe effective
throat area. When this condition is reached, the test-section Mach
number stabilizesagain to its normal value. Therefore, in operating
the tunnel it is important that the injector total pressure be kept
below the limiting value.
Effect of Model
The effectof thepresenceof a modelin the test sectionwas brie y
assessed by using a NACA 0012 two-dimensional model, which
had a chord length of 0.1 m. The tunnel was run with a  ap setting
that corresponded to the empty tunnel Mach number of 0.867. The
a
a
model angle of attack was varied from 0 to 5.5 deg. At = 0 deg
the resulting uncorrected Mach number was 0.863, which is only
slightly lower than the empty tunnel value. The largest effect was
a
observed at the highest , where the Mach number was observed
a
to fall to 0.848 or a 1.7% change from the = 0 deg value. How-
ever, no signiŽcant changes were observed in the duration of stable
 ow.
Conclusions
From the study of the efŽ ciencyof the secondthroat in stabilizing
the  ow in the test section of a transonicwind tunnel, the following
were found:
Mass EfŽ ciency
)
1 The use of second throats in induction transonic wind tunnels
Higher mass efŽ ciency was achieved with the horizontal/convex
 ap conŽ guration than with the other conŽ gurations. Within this
conŽ guration, the maximum efŽ ciency of 2 was achieved at Mach
0.77. Although this is much lower than the efŽ ciency of the mass-
 ow-controllabletunnelin Ref. 6, it is stillhigherthanthatof typical
blowdown tunnels. The efŽ ciency was calculated with a constant
totalmass of air. In most casesthismeansthatthereis a considerable
time lapse between the termination of stable  ow and the actual
shutting down of the tunnel. Because there is no need to run the
tunnel after stable  ow is terminated, higher efŽ ciency should be
obtainableif the amountof totalair used is limited by shuttingdown
the tunnel immediately after choking is terminated.
is not strictly prohibited and, in fact, can be beneŽcial to increase
tunnel efŽ ciency.
)
2 Installingthe secondthroat aps verticallyon the diffuserside-
wallswasnot veryeffectivebecauseofthe effectsofthereentry ow.
The affectingfactorsincludethe lowtotalpressureand the high ow
angularityin thereentryregion.The overalleffectwasthe premature
termination of the stable test-section ow.
)
3 These effects of the reentry  ow can be minimized by reposi-
tioningthe apshorizontallybehindthereentryregion.Thepresence
of the  aps makes the reentry ow return faster into the mainstream
to make the  ow more uniform. As a result, the second throat efŽ -
ciency is increased by up to 100%.
)
4 There exists an upper limit on the effectivenessof the second
Limiting Ejector Total Pressure
throat as a result of the aerodynamic choking of the injector jet,
which occurs beyond a certain injector total pressure. This causes a
temporary drop in the test-sectionMach number.
With the  aps presentthe durationof stable  ow simply becomes
a function of the initial reservoirpressure, and consequentlylonger
stable  ow can be achieved with higher tank pressures. However,
further testing showed that there is an upper limit to the choking by
the  aps, as shown in Fig. 14 for the case of 0-deg  ap de ection.
In this case the tunnel was run with an initial reservoir pressure of
35 atm absolute and a constant 60% valve opening. As the valve
was opened, the Mach number rose until it reached what would
normally be the choked Mach number. However, the  ow did not
stay long at this Mach number and started to dip slowly to a lower
Mach number, before rising again to its normal choking value. The
explanationto this phenomenonis given as follows. As the valve is
opened to its maximum, the Mach number rises until it reaches the
choking value as determinedby the cross-sectionalarea at the  aps
location. However, the injector total pressure continues to rise until
the injectorjet width has becomeso large thataerodynamicchoking
)
5 The use of a center strut is very detrimental to the tunnel efŽ -
ciency, probably because of the boundary-layerseparations on the
walls induced by the strong shock waves generated by the strut.
Similar results were observed in the case where the sting support
was present in the tunnel. To minimize the harm, whenever possi-
ble the use of a half-span, instead of a full-span, sting support is
recommended.
)
6 The performanceof the second throat was not signiŽ cantly af-
fectedby the presenceof a two-dimensionalmodelup to a moderate
angle of attack.
)
7 Even with a second throat installed,the efŽ ciencyof induction
tunnels can be maintained above that of typical blowdown tunnels,
the tunnel is immediately shut down at the end of the
especially if
stable period.
References
1Sewall, W. G., “Description of Recent Changes in the Langley 6- by
28-Inch Transonic Tunnel,” NASA TM 81947, May 1981.
2Torngren, L., Grunnet, J., Nelson, D., and Kamis, D., “The New FFA
T1500 Transonic Wind Tunnel Initial Operation, Calibration and Test Re-
sults,” AIAA 90-1420, June 1990.
3McDevitt, J. B., Polek, T. E., and Hand, L. A., “A New Facility and
Technique for Two Dimensional Aerodynamic Testing,” Journal of Aircraft,
–
Vol. 20, No. 6, 1983, pp. 543 555.
4McKinney, M. O., and Scheiman, J., “Evaluation of Turbulence Reduc-
tion Devices for the Langley 8-Foot Transonic Pressure Tunnel,” NASA
TM-81792, June 1981.
5Muhlstein, L., Petroff, D., and Jillie, D., “Experimental Evaluation of
an Injector System for Powering a High Reynolds Number Transonic Wind
Tunnel,” AIAA Paper 74-632, July 1974.
6Braha, J., Salomon, M., Seginer, A., and Rom, J., “Consideration for the
Design of a Second Generation Induction Driven Transonic Wind Tunnel,”
–
Journal of Aircraft, Vol. 11, No. 12, 1974, pp. 729 735.
7Lindsey, W. F., “Choking of a Subsonic Induction Tunnel by the Flow
from an Induction Nozzle,” NACA TN2730, July 1952.
Fig. 14 Mach-number dip associated with high-injector pressure.