AIAA JOURNAL

Vol. 49, No. 3, March 2011

Low-Speed Static Stall Suppression Using Steady

and Pulsed Air-Jet Vortex Generators

Simon A. Prince∗ and Vahik Khodagolian†

City University London, London, England EC1V 0HB, United Kingdom
DOI: 10.2514/1.J050754
An experimental investigation was undertaken to assess the ability of low-momentum steady and unsteady air
injection, using pitched and skewed air-jet vortex generators on a Royal Aircraft Establishment 9645 rotor section in
a nominally two-dimensional flow, to suppress quasi-steady flow separation at low speeds (0:07 < M < 0:12). Results
from this study show that unsteady blowing using air-jet vortex generators at a jet-to-freestream velocity ratio of 2.0
and pulsing frequencies equal to or above 50 Hz led to an
20% increase in CN max , while steady blowing with an
equal amount of total mass flow as the unsteady case led only to an
5% increase. Similar results were confirmed with
respect to the delay in drag and pitching-moment divergence. A correlation was found between aerofoil trailing-edge
shedding frequency and maximum normal force enhancement, where a maximum normal force coefficient was
obtained at a reduced pulsing frequency, F 
0:6, corresponding with the natural shedding frequency of the
aerofoil in stall. Tufts applied to the upper surface of the aerofoil highlighted interesting periodic wavelike motions of
the separated area downstream of the air-jet orifices at low frequencies below
10 Hz, and they showed that the
advance and retreat of the separated region was in synchronization with the activation and deactivation of the air jets
during each pulse. The oscillations increased with increasing jet pulsing frequency until no wavelike motion could be
detected.

Nomenclature Subscripts
c = chord length, m c = based on chord
CD = drag coefficient based on chord j = jet
CM = pitching-moment (about quarter chord) coefficient l = local condition
based on chord max = maximum magnitude
CN = normal force coefficient based on chord min = minimum magnitude
CT = tangential force coefficient based on chord t = trailing edge
Cp = pressure coefficient 1 = freestream
C
= jet blowing momentum coefficient based on planform
area (mean rms value for pulsed jet case)
DC = duty cycle, % I. Introduction
f = pulsing frequency, Hz
fw
F

=
=
Jones wake function
nondimensional pulsing frequency, fL=U
T HE suppression of upper-surface boundary-layer separation
from wings and rotor blades and the resulting suppression, or
delay to higher angle of attack, of subsequent stall has long been a
H = total pressure, Pa goal of aerodynamics research. Flow control methods using geom-
L = reference length (between air-jet orifices and aerofoil etrical modifications, such as leading-edge slats and slots, leading-
trailing edge), m edge droop, leading-edge rotation, and dynamically deforming
p = static pressure, Pa or psi geometries have been investigated in the past. Fluidic active flow
Re = Reynolds number control has also been investigated, including boundary-layer suction,
U = tunnel (freestream) flow speed, m=s steady air-jet or tangential slot blowing, and boundary-layer
v = jet exit velocity, m=s transition control.
x = chordwise coordinate from leading edge, m However, unsteady (oscillatory and pulsed) blowing through
y = chord normal coordinate, m narrow slots or orifices located on the top surface of a wing or blade
z = spanwise coordinate, m has been a recent primary focus of active flow control research and
 = angle of attack,  has been shown to be an effective method for delaying incompres-
 = boundary-layer thickness, m sible static and dynamic stall, primarily due to the reduced mass flow
 = jet orifice skew angle with reference to local freestream requirements (compared with steady blowing) and more effective
flow,  unsteady interaction with the separated flow region.
= jet orifice pitch angle with reference to local surface Most of the previous work on unsteady blowing to suppress stall
tangent,  has been conducted using oscillatory air injection through narrow
slots. Oscillatory or pulsed air injection through a spanwise narrow
hole or inclined slot induces lateral shear layer roll up and the
formation of either a series of discrete spanwise vortices (as shown in
Received 6 August 2010; revision received 6 October 2010; accepted for Fig. 1), which reenergize the boundary-layer and delay flow
publication 12 October 2010. Copyright © 2010 by the American Institute of separation, or a series of large-scale turbulent coherent structures,
Aeronautics and Astronautics, Inc. All rights reserved. Copies of this paper which can deflect and reattach a fully separated flow.
may be made for personal or internal use, on condition that the copier pay the
The focus of this study is the investigation of the effectiveness of
$10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood
Drive, Danvers, MA 01923; include the code 0001-1452/11 and $10.00 in
pulsed injection through an array of pitched and skewed orifices, or
correspondence with the CCC. air-jet vortex generators (AJVGs), that are known to be effective in
∗
Senior Lecturer, School of Engineering and Mathematical Sciences. delaying and suppressing stall from an aerofoil section when
Member AIAA. operated under steady blowing conditions. Pulsed activation of such
†
Research Assistant, School of Engineering and Mathematical Sciences. a system is likely to induce the formation of longitudinal vortices or
642
 PRINCE AND KHODAGOLIAN 643

turbulent eddies
propagating downstream

pulsed air jets

undisturbed boundary-layer
displacement thickness

Fig. 1 Conceptual representation of the unsteady flowfield generated by leading-edge pulsed air-jet injection.

discrete turbulent eddies, depending on the pulsing frequency, which with respect to the surface tangent and skewed (angle in Fig. 2)
should reenergize the boundary layer in both spanwise and with respect to the flow velocity vector over the surface. AJVGs do
chordwise directions. not induce as large a drag penalty as vane vortex generators and they
can be actively operated and controlled, depending on the flow
characteristics over a surface.
II. Review of Steady and Unsteady Air-Jet Flow The early studies [3–9] identified the range of optimum jet
Control for Separation Suppression orientations for maximum vorticity generation with steady blowing.
A. Steady Vortex-Generating Jet Flow Control Pearcey [3] and Freestone [4] showed that the optimum pitch angle
for the strongest streamwise vortices is around   30 . The
By artificially increasing the rate of fluid mixing within the literature shows that the optimum skew angle is very sensitive, but it
boundary layer, one can increase the kinetic energy of the low- broadly lies in the range of  45 to 90 . Selby et al. [7] identified an
momentum near-wall fluid and thereby delay, or to some extent optimal skew angle between 60 and 90 , whereas Compton and
prevent, the onset of flow separation. Various flow control tech- Johnston [8] suggested a value between 45 and 90 . Henry and
niques, to reenergize boundary layers and suppress flow separation, Pearcey [9] computed the turbulent flow over a flat plate with a single
have been identified and successfully tested, such as slot blowing, rectangular air jet, solving the Navier–Stokes equations with a simple
tangential blowing, and synthetic jets. The method of increasing fluid algebraic turbulence model. A parametric study was performed,
mixing rates by the artificial generation of near-surface longitudinal varying the slot aspect ratio between 1.0 and 5.0, the skew angle
vortices has been found to be a particularly powerful technique. The between 0 and 90, the pitch angle between 15 and 90 , and the jet
vortices act to entrain high-energy flow from the undisturbed outer velocity ratio (VR) between 0.5 and 2.0. The authors concluded that,
airstream and transport it into the low-momentum near-wall region in terms of the computed vortex strength and surface skin friction,
deep inside the boundary layer. Mechanical, passive, vane vortex where an increase in skin friction is desirable in this context, as it
generators, first devised by Taylor and Hoadley [1], are the most implies a thinning of the boundary layer and, hence, a healthier
common and widely used streamwise fluid vortex generators, and boundary layer to withstand separation, the optimum jet setting was a
they commonly consist of thin solid strips fixed to the surface, pitch angle of   30 and a skew angle of  60 . Zhang and
usually located ahead of a separated flow region, at an angle to the Collins [10,11] also showed that the optimum skew setting is in the
oncoming flow. However, it has been shown that mechanical vane- range of 30 and 60 for effective flow control. Singh et al. [12,13]
type vortex generators impose an increase in drag caused by both the showed that the 30 pitch, 60 skew setting was, marginally, the most
local pressure increase from flow blockage and by an increase in effective for stall suppression in experiments at low speeds on single-
surface skin friction downstream of the device. element aerofoils, and it was this orientation that was chosen for
An alternative to passive solid vane-type vortex generators is an implementation in this study.
active fluid vortex-generating device, which was proposed by Wallis
[2]. The idea is to use fluid injection via inclined and skewed (relative
to the freestream flow vector) wall-bounded jets to induce B. Oscillatory and Pulsed Air-Jet Flow Control
longitudinal vortices for flow control instead of solid vane-type It is now understood that oscillatory or pulsed blowing for the
vortex generators. AJVGs usually consist of an array of small orifices suppression of boundary-layer separation is much more effective
embedded on a surface and supplied by a pressurized fluid source, than pure steady blowing, and it has the additional benefit that much
wherein longitudinal vortices are induced by the interaction between lower total mass flow levels are needed. The first exploratory study of
the jets issuing from each orifice and the fluid flowing along the unsteady blowing flow control on a wing section was by Oyler and
surface, as shown in Fig. 2. The orifices are pitched (angle  in Fig. 2) Palmer [14], who developed an unsteady tangential blowing system

Fig. 2 Typical geometry of a leading-edge array of AJVGs with illustration of the physics of longitudinal vortex formation (after Pearcey [3]).
644 PRINCE AND KHODAGOLIAN

on a trailing-edge flap. Their experiments demonstrated a significant stall from 12 to 16 angle of attack, and increasing VR and mass flow
improvement of stall suppression and corresponding maximum lift rate improved AJVG effectiveness. The results also showed that the
capability, compared with steady blowing of equivalent total mass optimum pulsing frequency is highly dependent on VR. When
flow. VR  2, the optimum reduced pulsing frequency was F
 0:5, and
Oster et al. [15], Oster and Wygnanski [16], Wygnanski and for VR > 2, optimum F
reduced to 0:4. The study also pointed to
Pedersen [17], and Ho and Huerre [18] conducted extensive the important result that the optimum pulsing frequency for
fundamental studies of free turbulent mixing layers using artificial separation control is achieved when it is the same as the natural
unsteady flow excitation (by means of mechanical oscillators), and shedding frequency of the eddies in the separated flow.
they highlighted the complex flow physics and fluid dynamic McManus and Magill [26,27] undertook an experimental study of
benefits of introducing periodic disturbances to control the mixing in pulsed AJVGs on a more realistic configuration: a two-element blade
a fluid flow. Katz et al. [19] showed that a wall-bounded separated section comprising a NACA 4412 aerofoil main section and a
turbulent mixing layer could be attached to a curved surface by leading-edge slat of 20% overall chord. The tests were performed
unsteady flow excitation, using a small vibrating ribbon to induce across a Mach number range between 0.1 and 0.5. This study, again,
flow oscillations. highlighted the importance of jet VR. Optimum aerodynamic
Seifert et al. [20] studied unsteady slot blowing on a multielement performance was found with VR  7:4 (maximum) and a pulsing
wing section comprising a NACA 0012 main wing with a trailing- frequency of F
 0:6. The difference in F
between this study and
edge flap. The experiments were performed under quasi-steady pitch McManus et al.’s earlier study [24] indicates that the optimum
conditions at Reynolds numbers, based on chord, between 105 –106 . pulsing frequency is geometry dependent.
The results showed that, for an order of magnitude less power Greenblatt and Wygnanski [28] and Greenblatt et al. [29,30]
consumption than required for steady blowing, unsteady blowing performed a number of experiments on pulsed blowing, specifically
provided a significant increase in lift and a reduction of drag for for the control of dynamic stall. These experiments have
angles of attack above which the unblown model was found to stall. demonstrated several consistent results, including the following:
McManus et al. [21,22] successfully implemented unsteady 1) The most effective location for unsteady forcing is near the point
oscillatory blowing AJVGs to suppress turbulent boundary-layer of separation.
separation from a 20 ramp diffuser. The low-speed wind-tunnel tests 2) The optimum reduced frequency for the oscillations is about
were conducted at a Reynolds number based on the boundary-layer F
 0:5–1:0.
thickness at the jet exit j of 2300. The AJVG array, located at 8j 3) The amplitude of the oscillations required for effective
from the ramp shoulder, consisted of three circular jet orifices of separation control is about two orders of magnitude lower than that
diameter 0:67j, spaced 4j apart, at a 45 pitch angle and a 90 skew for steady blowing.
angle relative to the oncoming freestream flow. Flow visualization When oscillatory separation control is applied to an aerofoil that is
using acetone vapor seeding and pulsed laser sheet illumination not separated, the effects are not detrimental to lift. For dynamic
demonstrated that each pulse generated a large-scale vortical pitching, prior work has shown that higher frequencies, typically
structure, which then convected downstream, keeping close to the 200 Hz, are necessary, so that many pulses are produced during the
ramp surface. The results of McManus et al. clearly demonstrated pitchup. The effectiveness of the jets was found to generally increase
that, for a given total mass flow rate, oscillatory AJVGs were more with VR until a maximum effect was reached. Further increase in VR
effective in suppressing turbulent separation than steady blowing had no effect. Lift enhancement generally saturates in this way when
AJVGs. It also became clear that the mechanism of unsteady the VR is increased beyond VR  4. Prior investigations also
oscillatory or pulsed AJVG flow control is completely different to showed that duty cycles of 25% produce the same effect as those of
that of steady AJVG blowing. Under pulsed conditions, the jets do 50% for a given VR. Thus, 25% is typically chosen, as it uses less air.
not generate the strong, continuous, streamwise vortices embedded Eroglu and Breidental [31] investigated the flow structure when a
within the turbulent boundary layer that are observed with steady pulsed jet of water was injected into a water flow, particularly
AJVG air injection. focusing on the effect of pulsing frequency. It was shown that high-
Unsteady fluid dynamic excitation is believed to induce the frequency pulsed blowing of jets in crossflow results in a flow very
generation of large-scale turbulent coherent structures, particularly similar to steadily blown jets at the same blowing momentum
when the mean flow is unstable (as in a separated flow). The turbulent coefficient. The time period between each injection of fluid mass was
coherent structures act to entrain high-momentum freestream air and too short to allow the formation of coherent turbulent structures, and
transport it into the low-momentum inner layers of the boundary the individual pulses seemed to merge into a continuous stream,
layer, thereby enhancing turbulent mixing and suppressing allowing the roll up of the streamwise boundary-layer-embedded
separation. vortices seen with steady AJVG blowing. Johari [32] made use of
Seifert et al. [23] undertook further studies on unsteady slot Eroglu and Breidental’s [31] flow visualization data to propose a
blowing from the upper surface of a trailing-edge flap linked with classification scheme for fully pulsed jets in crossflow based on the
four different aerofoil section main elements. These low-speed stroke ratio of the jet pulse and the duty cycle of the pulse train.
experiments were performed at Reynolds numbers, based on chord, Scholz et al. [33] tested an array of pulsed air jets near the leading
of between 0:15–1:2  106 and involved the parametric variation of edge of a specially developed single-element aerofoil model
the blowing frequency f. The results showed that the most effective designed to stall with an abrupt leading-edge separation. The
blowing frequency occurred when reduced pulsing frequency was experiments were performed in low-speed flow with a Reynolds
near unity. number based on chord limited to 1:3  106 . The spanwise array of
McManus et al. [24,25] studied pulsed AJVG stall suppression on 80 rectangular air jets, each with a length of l=c  0:125 and an
a stylized aerofoil section wing, where the 15.24 cm chord aerofoil aspect ratio of 25, were skewed at 45 to the freestream flow. The
comprised a leading-edge flat-plate flap and a flat-plate main section. array was tested at a chordwise location of 1% chord: first on the
The leading-edge section had a length of 21% of the total chord of suction side, whereby the air jets were issuing directly into
the model and was deflected down at 15 to the flat-plate main the separated region, and then at the same chordwise location on the
section. A zero angle of attack was thus defined as when the main pressure side, with the jets issuing into the turbulent boundary layer
section was at zero incidence to the freestream flow. The model in between the attachment line and the separation line. It was found
incorporated a single circular orifice AJVG, of diameter 0:016c, that with the jets positioned on the suction side, in the region of the
pitched at 45 and skewed at 90 to the local freestream flow. Low- separation line, they could not prevent separation. However, they
speed wind-tunnel tests were performed at a Reynolds number, based were found to be able to increase the normal force coefficient above
on total chord, of 5:0  105 . Jet-to-freestream VR (1.5–6.0), mass the prestall value by as much as 8%. The most sensitive parameter in
flow rate, and pulsing frequency (10–100 Hz) were varied to gain a this case was found to be the pulsed jet duty cycle, with the best
better insight into their effect on stall suppression. The results of this results obtained with duty cycle in the range 12–25%. Pulsing
study showed that the pulsed AJVG successfully delayed aerofoil frequency was not found to have a big effect in this case.
 PRINCE AND KHODAGOLIAN 645

Ortmanns et al. [34] presented a study of the interaction of the frequency attenuation. Since oscillatory or pulsed excitation of an
turbulent coherent structures generated by a vortex-generating jet internal plenum chamber, used in a previous study [13], led to major
with a flat-plate turbulent boundary layer in a water tunnel with a pressure loss and frequency attenuation, it was decided to use
maximum freestream speed of 2 m=s. The rectangular water jet individually controlled flow actuators placed as close to the jet exit as
orifice, with a dimension of 10  0:32 mm, was pitched at 90 and possible. For application to aircraft wings or helicopter rotor systems,
skewed at 45 to the freestream water flow. The authors used phase- the individually controlled air injectors would be beneficial, as it was
locked stereoscopic particle image velocimetry to track the passage expected that the optimum actuation frequency would be different at
of the turbulent vortical structures as they passed downstream, different spanwise locations and for different azimuthal blade angles.
varying the jet VR and the pulsing duty cycle. The pulsing frequency A pulsed injection (rather than oscillatory) strategy was chosen, as it
was fixed at 1 Hz. It was found that the velocity overshoot that was deemed to be the best approach to achieve accurate control of the
occurred at the beginning of each pulse, which causes an enhanced generation of sharply defined individual vortical structures. The final
mixing effect, is the most important factor in developing larger design of the pulsed blowing system on the RAE 9645 wing model,
vortical structures with greater downstream impact. This startup shown in Fig. 3, included a common internal pressure-regulated
effect was found to be sensitive to jet-to-freestream VR but generally plenum tube (component 1) feeding the air injectors (component 2),
independent of duty cycle. The downstream decay of the turbulent with the air exiting at the aerofoil surface through the air-jet orifice
structures was also found to be independent of duty cycle. (component 4) via connecting tubes (component 3). Components 5
In summary, the past work on oscillatory or pulsed AJVGs and 6 are cavities designed for future flow control experiments at the
indicates that they represent a considerable improvement over steady trailing edge.
AJVGs for the application of aeronautical flow control and, The ideal aerodynamic design would be to have the air-jet injectors
especially, stall suppression. Separation control using oscillatory flush-mounted with the aerofoil surface; this would reduce the
blowing involves two different aspects: first is the delay of flow pressure losses and frequency attenuation. The limiting factor to
separation, and second is the reattachment of a fully separated flow, achieve this design was the internal geometric space restrictions of
which may require two different approaches, since the receptivity of the wind tunnel, thereby placing a limit on the maximum size of the
the unforced flow to unsteady forcing is different. air-jet actuator. However, the length and geometry of the tubes
connecting the injector to the jet exit and the pressurized air supply
was designed to avoid flow resonance, reduce the internal pressure
III. Experimental Details loss, and keep flow frequency distortions to a minimum.
A. Wind-Tunnel Facility The RAE 9645 model was configured with a single spanwise array
The experiments were conducted in the City University London of AJVGs on the upper surface of the aerofoil, consisting of 15
T2 low-speed closed-circuit wind tunnel. The Royal Aircraft pitched and skewed AJVGs with circular orifice geometries (Fig. 2)
Establishment (RAE) 9645 model was mounted vertically in its located at 12% chord, since past studies [1–13] have shown that
octagonal working section of the width 1.12 m, the height 0.81 m, AJVG orifices need to be well upstream to provide enough length for
and the length 1.68 m. The rear of the working section was vented to fully developed boundary-layer embedded vortices to form. The jet
atmosphere, and flow quality improvement was achieved by two orifices had a diameter of 4.8 mm, compared with the local
fine-meshed turbulence screens mounted before the settling undisturbed boundary-layer thickness of 5 mm at   0 . The
chamber. The T2 wind tunnel had a Reynolds number capability of spacing between orifice centers was 45 mm, which is in the range for
3:1 million=m, with a maximum speed limit of 45 m=s. The tunnel an optimum spacing-to-orifice-diameter ratio of 6 to 10 [3,4]. They
was also equipped with a compressed air feed system, with regulator were pitched at 30 from the surface tangent and skewed at 60 with
valves to feed four inlet pipes. The fluctuation of the freestream respect to the freestream direction. The pitch and skew angles used at
dynamic pressure in the tunnel working section was measured to be the jet exits were arranged to induce corotating streamwise vortices
within 0.5% of the nominal value for a flow velocity of 35 m=s. with steady blowing, in which each individual jet formed a
The experiments were performed using the computer-controlled streamwise vortical flow structure over the aerofoil surface. The
model turntable, allowing continuous variation of pitch. Angle-of- British Standard mass flow rate of jet air entering the model was
attack changes were performed at a pitching rate of 1 =s, and all measured using a BS orifice plate rig, as described in [13].
pressure data were taken 30 s after each 1 incremental change in
order to ensure that there would be negligible effect of pitch rate. The
maximum blockage in the tests was 15%, but automatic
compensation of fan speed was used to maintain constant tunnel flow
velocity in the working section as the model was pitched. Correction
of surface pressure measurements was not deemed to be necessary, as
the study was comparative.

B. Model Design Details

An RAE 9645 aerofoil section model of 480 mm chord was
developed, of the same basic design as those used in the study of
Singh [13]. The model, designed to be a representative spanwise
section of a helicopter rotor blade, was made of Tufnol, manufac-
tured using layers of laminates bonded together with high-quality
thermosetting resins. The model was then assembled by integrating
individual spanwise aerofoil sections and held together by three
spanwise rods. Endplates were also fitted at the two sides of the
model to maintain a nominally two-dimensional flow. These
endplates, while preventing the formation of strong tip vortices, also
worked to isolate the model from the tunnel wall boundary layer. A
thin 3 mm strip of sand roughness was glued to the leading edge of the
model to promote turbulent boundary-layer development from the
leading edge.
An important design objective for the pulsed blowing flow control
system was to achieve a sufficient amount of jet exit momentum for a
range of input frequencies, with minimum pressure loss and Fig. 3 RAE 9645 pulsed AJVG model and experimental setup.
646 PRINCE AND KHODAGOLIAN

Table 1 Performance of the electronically 30 m=s: the highest speed for the model given the force limits on the
controlled high-speed jet actuator balance. It was shown that the pressure integration technique, in this
Parameter Value case, provided normal force values 5% less than the measured
value at 0 incidence; at the highest incidence of 18 , this discrepancy
Mass flow, g=s 0:0  m_ j  3:665 was only 2%. For tangential force, this discrepancy was 10 and 4%,
Supply pressure, psi 0 < p  30:0 and for pitching moment, it was 10 and 5%. The accuracy of the
Pulsing frequency, Hz 0:0  f  300:0
pressure integration method in this study is expected to by slightly
Duty cycle (DC), % 20:0  DC  100:0
better for the standard freestream velocity of 35 m=s. It is important
to remember, however, that this is a comparative study, and these
errors will follow the same trend and should be equivalent at the same
The model was instrumented with a total of 44 surface pressure incidence where data are compared at the same freestream velocity.
tappings, positioned at the center span along the chord line, A wake rake was used to measure the aerofoil wake momentum
connected via short tubes to a 64-channel high-frequency electronic deficit one chord length downstream of the aerofoil trailing edge. The
pressure scanner. An additional 18 surface pressure tappings were wake rake consisted of 40 stainless-steel pitot probes and five static
located downstream of one AJVG orifice. probes (outer diameter of 1.05 mm, and inner diameter of 0.81 mm).
The pitot pressure probes were spaced at 7 mm intervals in the center
C. Pulsed Air Injector and at 15 mm intervals toward each spanwise extremity, giving a total
span of 350 mm. The static pressure tubes were used to measure any
The pulsed air injector developed by Synerject, Inc., based on
static pressure gradient across the wake. The angle of attack was
internal combustion engine solenoid fuel injector technology, was
measured using the electronic computer-controlled balance turntable
chosen as the pulsed air-jet actuator, since it was found to satisfy the
with an accuracy of 0:2 . Zero angle of attack was set using a laser
required mass flow, pulsing characteristics, and geometrical size
mounted on the model chord line, for which the beam was centered
requirements. Table 1 provides details of the operating characteristics
on a vertical wire mounted on the turning vanes 5 m down the tunnel
of the injector.
from the trailing edge. Measuring the pitot and static pressures across
An electronic injector drive control unit was developed to enable
the wake, via the Chell high-speed pressure scanning system,
individual or simultaneous activation and frequency control of each
permitted the calculation of the profile drag coefficient using Jones’s
air injector and AJVG. A linear integrated circuit was designed to
method [35].
control and reduce the total input current required to open and hold
Before the tests, the velocity of the air jets were calibrated against
the air injector valve. By holding the injector current at one-fourth of
the plenum pressure for the case of zero freestream tunnel flow. The
the peak current, power dissipation in the solenoid could be reduced
effect of freestream flow would be to increase the jet velocity by a few
by at least the same factor. Preliminary tests were conducted to assess
percent. The average jet velocity was measured using a small pitot-
the flow characteristics, including jet exit pressure, frequency, and
static tube with a 2 mm outer diameter (0.4 times the jet orifice
the velocity of the jet. The jet flow characteristics were measured
diameter) mounted to traverse the air jet.
using a dynamic pressure transducer (Kulite CTQH-187, type B)
The model is shown in Fig. 3, mounted in the T2 wind tunnel,
installed at the jet exit (in the core region) to measure the pressure
together with the endplates and the wake rake, mounted from a
signal and frequency content of the air jet. The Kulite sensor was
manual three-axis arm-traverse system. Surface ultraviolet (UV)
traversed laterally through the pulsing air jet to identify the jet core
fluorescent tufts were also mounted in a grid system on the upper
location, and dynamic pressure traces were recorded across the
surface of the model during the last series of experiments. This
operating frequency range. The pressure traces were found to follow
simple surface flow technique allowed for the detection of surface
the sharp square-wave input signals very well up to 100 Hz. Above
flow features, such as separation lines, stall cells, and longitudinal
10 Hz, there appeared an overshoot or spike in the pressure signal at
vortices, and it could also resolve, for low frequency, their movement
the start of the pulse, similar to that described by Ortmanns et al. [34].
over time.
The air injectors could be operated in a steady or pulsed mode.
Compressed air was supplied to a pressure-regulating valve and fed
into the air injector intake. The pulsed output from the injector was
fed through short connecting brass tubes to the jet exit. This bench IV. Results and Discussion
test arrangement was identical to the one designed for the final A. Comparison of Steady and Pulsed Air-Jet Blowing
RAE 9645 aerofoil model. The tests confirmed that the pressure
The first series of experiments involved the measurement of the
signals measured at the air-jet orifices matched the sharp input
surface pressure distributions on the RAE 9645 model with
electrical square-wave signals up to a pulsing frequency of 100 Hz.
continuous angle-of-attack variation ( 10 <  < 26 ) for a
Beyond 100 Hz, the output pressure signals exhibited progressively
constant U  35 m=s wind speed for the cases of 1) a clean surface
more noise and less sharpness.
model with the air-jet orifices smoothly sealed; 2) air-jet orifices open
but inactive air jets; 3) steady air-jet blowing at VR  0:5, 1.0, 1.5,
D. Measurement Details and 2.0; 4) pulsed air-jet blowing at VR  1:0, with pulsing
Surface and wake pressures were recorded using a piezoelectric frequency set at f  10, 50, and 100 Hz; and 5) pulsed air-jet blowing
high-frequency pressure scanning system from Chell Systems, with a at VR  2:0, with pulsing frequency set at f  10, 50, and 100 Hz.
quoted accuracy of 0:1% in the pressure range measured in this Comparison of the surface pressure distributions for both the sealed
study. The surface pressure was measured via the chordwise array of and open inactive air-jet cases showed that there was no detectable
44 pressure tappings located on the centerline, and two other parallel effect of open air-jet cavities on the surface pressures.
lines of 18 pressure taps located 10 and 20 mm downspan on the Examples of the surface pressure distributions measured for cases
upper surface, downstream of the AJVG array. These were 1, 3, and 5 are plotted in Fig. 4 for   10, 16, and 22 , respectively.
introduced in order to measure any spanwise surface pressure At   10 , before the occurrence of trailing-edge separation (Cpt
variation induced by upwash and downwash flows of the longitudinal still positive), there is no discernible difference, within the accuracy
vortical structure. The pressures from these stations were used to of pressure scanning equipment, between the six sets of data. At
obtain average pressure distributions with which to gain a better   16 , the trailing-edge pressure coefficient is seen to be negative
calculation, by integration of the measured pressure distributions, of for the inactive blowing case, indicating that the boundary layer has
the aerodynamic forces and moments. A previous study with a separated over the rear upper surface. For all of the blowing cases,
NACA 23012 model of 20% less chord length than the current however, Cpt is seen to remain positive, indicating that all blowing
model, but with similar density of pressure tappings, also involved cases are effective in suppressing trailing-edge separation at this
the direct measurement forces and moments using a six-component angle of attack. Interestingly, both steady and pulsed air-jet blowing
force balance. This study was undertaken at a freestream velocity of is seen to increase the level of suction over the forward 50% of the
 PRINCE AND KHODAGOLIAN 647

-3.5

-7.0

-3.0 -3.0 -7.0

-6.0

Cp

Cp
-6.0 -5.0
-2.5

-2.0 -5.0 -4.0

-2.0
0 0.02 0.04 0.06 0.08 0.1
x/c -4.0 -3.0
0 0.02 0.04 0.06 0.08 0.1
x/c
Cp

Cp
-1.0 -3.0

-2.0

0.0 -1.0

0.0

1.0 1.0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/c x/c
a) α = 10º b) α = 16º

-9.0
-9.0
-8.0

-8.0 -7.0

Cp -6.0
-7.0
-5.0

-6.0 -4.0

-3.0
-5.0
-2.0
0 0.02 0.04 0.06 0.08 0.1
x/c
Cp

-4.0
Inactive air jets
-3.0 Steady air jets, VR=1.0, C µ = 0.0014
Steady air jets, VR=2.0, C µ = 0.0062
Pulsed air jets, VR=2.0, f=10Hz, Cµ = 0.0033
Pulsed air jets, VR=2.0, f=50Hz, Cµ = 0.0033
-2.0 Pulsed air jets, VR=2.0, f=100Hz, Cµ = 0.0033

-1.0

0.0

1.0
0 0.2 0.4 0.6 0.8 1
x/c
c) α = 22º
Fig. 4 Variation of centerline surface pressure distribution with steady and pulsed air-jet blowing at angles of attack of a) 10, b) 16, and c) 22
(U  35 m=s and ReC  1:1  106 ).

aerofoil upper surface, with the pulsed air-jet blowing at 100 Hz steady blowing at VR  1:0 does not appear to differ significantly
pulsing frequency clearly giving the greatest enhancement. from the unblown case until beyond   18 . At higher incidences,
At   22 , fully separated flow beyond 10% chord is indicated steady blowing at VR  1:0 is seen to prevent the sudden drop in CN
for the inactive air-jet case by the pressure plateau. While the rear- associated with the collapse of leading-edge suction and the onset of
most portions of the upper surface are seen to experience separated full stall all the way up to   25 . At   26 , the level drops down
flow for all of the cases, air-jet blowing is clearly demonstrating a to the unblown level. Once the pitchdown began, CN levels were
capability to defend the leading-edge suction against the upstream initially found to plunge to a level some 5% below those seen with
advance of the separation front, thereby delaying stall to higher angle no blowing before (between   21 –20 ), a significant recovery
of attack. Steady blowing at VR  2:0 can clearly be seen to be was seen bringing the CN level back to its pitchup level for the same
outperforming VR  1:0 steady blowing in maintaining high levels angle of attack. This pitch hysteresis in CN is not observed in the data
of leading-edge suction. Pulsed blowing at VR  2:0 at the lowest for the unblown model case. The addition of energy into the
f  10 Hz pulsing frequency is seen to match the steady VR  1:0 separated flow region by steady air-jet blowing clearly promotes the
results, but an increase in pulsing frequency to 50 Hz results in early recovery of attached flow on the leading edge. Close scrutiny of
performance almost equivalent to steady blowing at VR  2:0. Given Fig. 5d shows that steady blowing at VR  1:0 does not suppress the
that the duty cycle for all of these experiments was set at a constant occurrence of incipient trailing-edge separation (Cpt drops to zero),
50%, this is achieved with half the overall mass flow of the equivalent which is seen to occur at between   11 and 12 .
steady case. A further increase in pulsing frequency to f  100 Hz Increasing the blowing VR to 2.0 vastly improved the
is not seen to result in any further significant improvement. effectiveness of steady AJVG blowing. CN max was increased from
Figure 5 provides a comparison of the effects of steady and pulsed 1:42 without air-jet blowing, to 1:5 with steady blowing at
AJVG blowing on the variation of aerodynamic characteristics with VR  1:0, up to 1:76 with blowing at VR  2:0. The increase in
continuous variation angle of attack, starting from   10 , blowing jet VR is also seen to delay the onset of trailing-edge
increasing up to a maximum   26 (pitchup), and followed by a separation to about a 3.5 greater angle of attack. A small loss in CN ,
progressive reduction back down to   10 in steps of
  1 . resulting in a kink at about   15 , with a further small increase in
Figure 5a compares the results for the calculated variation of a angle of attack, is then almost certainly due to the loss of surface
normal force coefficient with the angle of attack. The CN curve for suction on the rearmost upper surface as the separation front
648 PRINCE AND KHODAGOLIAN

1.80 0.30

1.60
0.25 Estimated maximum error range ( α < 22 deg)

1.40
0.20
1.20

1.00 Estimated maximum error range 0.15

CN

CD
0.80
Inactive air jets
0.10
Steady air jets, VR=1.0, C µ = 0.0014
0.60 Steady air jets, VR=2.0, C µ = 0.0062
Pulsed air jets, VR=2.0, f=10Hz, C µ = 0.0033

0.40
Pulsed air jets, VR=2.0, f=50Hz, C µ = 0.0033 0.05

0.20 0.00

0.00
0 4 8 12 16 20 24 0 4 8 12 16 20
α (deg) α (deg)
a) Normal force coefficient b) Drag coefficient

0.25 0.2
Estimated maximum error range
Incipient trailing-edge
0.0 separation criterion
0.20

-0.2
0.15
Cpt
-CM

-0.4
Estimated maximum error range
0.10
-0.6

0.05
-0.8

0.00 -1.0
0 4 8 12 16 20 24 0 4 8 12 16 20 24 28
α (deg) α (deg)
c) Quarter-chord pitching-moment coefficient d) Trailing-edge pressure coefficient
Fig. 5 Variation of CN , CM , CD , and Cpt with angle of attack for steady and pulsed air-jet blowing (U  35 m=s and ReC  1:1  106 ).

advances rapidly upstream. Steady blowing at VR  2:0 is seen to effective in delaying trailing-edge separation as steady blowing.
deliver considerably higher normal force levels, beyond CN max , However, significantly improved levels of CN are seen at the
compared with steady VR  1:0 blowing up to   25 . A further beginning of the pitchdown with 50 Hz pulsing, which are nearly
pitchup to the maximum 26 angle of attack, however, results in a 10% higher than those calculated for either the unblown model or the
complete collapse of leading-edge suction, as with the VR  1:0 10 Hz pulsed blowing at VR  2:0. Recovery to CN max occurs at the
case, but recovery on the pitchdown occurs at about a 1 angle of same angle of attack on the pitchdown as was seen with the lower
attack earlier, resulting in a return to CN max . Hysteresis is also evident 10 Hz pulsing. Clearly, 50 Hz pulsing provides improved normal
in the occurrence of trailing-edge reattachment on the pitchdown, force characteristics over 10 Hz pulsing. The results for f  100 Hz
which takes place at about a 1 lower angle of attack. were found to be practically identical with those for 50 Hz and were
Pulsed blowing at VR  2:0 and f  10 Hz resulted in a CN therefore omitted for clarity.
characteristic on the pitchup very similar to that obtained with steady For the case of VR  2:0, the occurrence of pitch hysteresis
blowing at VR  1:0 but with marginally increased levels (3%) of between pitchup and pitchdown aerodynamic characteristics was
normal force at a given angle of attack above the incipient trailing- observed with pulsed blowing at the same angle-of-attack ranges as
edge separation angle. On the downstroke, however, pulsed blowing seen with steady blowing. This is the case for all three pulsing
at VR  2:0 and f  10 Hz resulted in an initial drop in CN to the frequencies of 10, 50, and 100 Hz.
same level as seen without any air-jet blowing and a recovery at the Since the pulsed jet duty cycle is 50%, steady blowing at VR  1:0
same angle of attack as observed with steady VR  2:0 blowing. nominally delivers the same total mass flow rate as the pulsed jets
This indicates that pulsed air-jet blowing, even at a low 10 Hz operating at VR  2:0, whatever pulsing frequency is employed. The
frequency, is much more effective in enhancing mixing in, and results show that pulsed AJVGs are capable of delivering improved
reenergization, of a separated shear layer than steady air-jet blowing aerodynamic performance over steady AJVG blowing for the same
of equivalent jet-to-freestream VR. total mass flow rate. This is, in part, due to the higher VR jet obtained
An increase in the pulsing frequency from 10 to 50 Hz resulted in for the same total mass flow of blown air that is enabled by pulsed
CN characteristics almost identical to those seen with steady blowing blowing. Pulsed blowing is seen to give a CN max of approximately
at the same VR. Close examination reveals, however, that this 50 Hz 1.8, compared with a figure of approximately 1.5 for the
pulsing case has a marginally higher CN max of 1:8, which the corresponding steady blowing case: an improvement of 20%.
surface Cp measurements reveal is associated with slightly higher Figure 5b presents the variation of drag coefficient CD (calculated
suctions on the front portions of the upper surface. Incipient trailing- by integration of the Jones wake function) with angle of attack for
edge separation with VR  2:0 and 50 Hz pulsing is seen to occur at U  35 m=s and jet blowing at VR  2:0. Above   21 , the
around   13 , which is some 2 lower than with steady blowing at viscous wake was seen to become too wide to be completely resolved
the same VR, indicating that pulsed air jets may not be quite as by the wake rake, and data above this angle of attack have therefore
 PRINCE AND KHODAGOLIAN 649

been omitted. In addition, separated flow and flow reversal induced Similar behavior can be seen in the variation of the quarter-chord
an angular momentum component that could not be measured using pitching moment with angle of attack (Fig. 5c). In contrast with the
the linear momentum deficit method. As angle of attack increases, evidence from the steady blowing cases, the pulsed AJVG cases are
the level of the underprediction of CD will increase. The conclusions seen to delay, by
  1, the onset of the rise in pitching moment
can, therefore, only be accurately drawn from the data below following trailing-edge separation. The subsequent magnitude of CM
  21 . for a given  up to   21 is seen to be consistently lower for the
The experimental data presented in Fig. 5b show that the rapid pulsed blowing cases than for the inactive jet case. The hysteresis in
divergence in CD (beginning at   16 ) soon after trailing-edge CM , between pitchup and pitchdown, for   22–26 is observed,
separation for the inactive jet case is considerably suppressed by where the 50 Hz pulsing case is observed to cause a significantly
steady AJVG blowing at VR  2:0 and pulsed AJVG blowing at higher value of CM before the pitchdown. This is due to the
VR  2:0 for pulsing frequencies f of 10, 50, and 100 Hz. Pulsing at maintenance of slightly higher levels of leading-edge suction for the
10 Hz is seen to considerably reduce the drag coefficient for a given 50 Hz pulsing case.
angle of attack between 17 and 21 , but the effect is perhaps only half It is important to note that these results were found, within the
as effective as steady blowing. Pulsing at 50 and 100 Hz, which limits of experimental and integration accuracy, to be repeatable,
seems to deliver almost identical behavior, provides almost the same with three complete pitch sweeps having been performed. The clear
level of suppression in drag divergence as steady blowing. For a differences between the results at f  10 and 50 Hz are important
given angle of attack, between 17 and 20 , the magnitude of CD for clues, showing that whatever physical effect is occurring with pulsed
f  50 and 100 Hz is almost identical to the steady blowing result. AJVG blowing (an important change in physics) occurs between 10
This reveals that, for the same jet-to-freestream velocity, pulsed and 50 Hz pulsing frequency.
AJVG blowing at 50 and 100 Hz can provide a similar suppression of
drag divergence as steady blowing but at only half the total blowing B. Effect of Freestream Reynolds Number, VR  1:75
mass flow. and f  50 Hz
The data show the typical variation of CD versus  for a turbulent The results of the experimental investigation of the effect of the
case exhibiting no sign of any laminar drag bucket. One interesting freestream Reynolds number (based on model chord) is presented in
feature is a sudden drop in the magnitude of CD by as much as 30%, Fig. 6 for pulsed AJVGs at VR  1:75 and f  50 Hz. The
seen to occur soon after trailing-edge separation, before the results show that, for a given angle of attack, there is very little
beginning of a steady rise with angle of attack. discernible difference up to   20 within the limits of experimental

2.0 0.26
0.24
1.8
0.22
1.6
0.20
1.4 0.18

1.2 0.16
0.14
-CM
CN

1.0
0.12
0.8 U
U
= 35 m/s, Inactive air jets
= 40 m/s, VR=1.75, f=50Hz 0.10
U = 35 m/s, VR=1.75, f=50Hz
0.6 U = 30 m/s, VR=1.75, f=50Hz 0.08
U = 25 m/s, VR=1.75, f=50Hz

0.06
0.4
0.04
0.2
0.02
0.0 0.00
0 4 8 12 16 20 24 0 4 8 12 16 20 24
α (deg) α (deg)
a) Normal force coefficient b) Quarter-chord pitching-moment coefficient

0.2

Incipient trailing-edge
0.0 separation criterion

-0.2
Cpt

-0.4

-0.6

-0.8

-1.0
0 4 8 12 16 20 24
α (deg)
c) Trailing-edge pressure coefficient
Fig. 6 Variation of CN , CM , and Cpt with angle of attack for pulsed air-jet blowing, with varying freestream velocity at constant air-jet-to-freestream VR
of 1.75 and pulsing frequency of f  50 Hz.
650 PRINCE AND KHODAGOLIAN

16
accuracy in the values of CN , CM , and Cpt within the Reynolds

α at incipient trailing-edge separation

number range:
15
Rec  0:79  106 U  25 m=s –Rec  1:27  106 U  40 m=s

Above   20 , there are small differences between the results on the 14
pitchup, with the highest CN and lowest CM measured at the lowest
freestream Reynolds number. No Reynolds number effects were 13
discerned in the data for the inactive jet cases at different freestream
velocities. These results are, of course, only valid across the limited
Reynolds number range of 0:79  106 < Rec < 1:27  106 . A 12
further study is required, employing a wind tunnel, ideally, with a
much higher speed capability to characterize the effect of the
11
Reynolds number across a much greater and physically meaningful
Reynolds number range.
10
0.0 0.5 1.0 1.5 2.0 2.5
C. Effect of Jet-to-Freestream-Velocity Ratio, U  25 m=s and f  Jet VR
50 Hz
Fig. 8 The effect of jet-to-freestream VR on the angle of attack at
Figure 7 presents the results of the experiments to highlight the incipient separation at the trailing edge (U  25 m=s) and pulsing
effect of varying the jet-to-freestream VR (between 1.0 and 2.8) for a frequency of f  50 Hz.
constant tunnel speed of U  25 m=s and a constant pulsing
frequency of 50 Hz. Incipient trailing-edge separation can clearly be
seen (from Fig. 7c) to be strongly effected by an increase in jet VR. action of the pulsed AJVGs to a value estimated to be about 15.5
The results are summarized in Fig. 8, which presents the estimated with VR  2:8. The data for VR  1:0 would suggest that, at this low
(from Cpt versus ) variation of the angle of attack for incipient jet VR, the effect at the trailing edge is minimal. A progressive
trailing-edge separation with jet-to-freestream VR. increase in jet-to-freestream VR above 1.0 results in a progressive
The angle of attack for incipient trailing-edge separation, which is increase in the angle of attack required to cause trailing-edge
about 12 for the inactive jet case, is seen to be suppressed by the separation.

2.00 0.26
0.24
1.80
0.22
1.60
0.20
1.40 0.18

1.20 0.16
0.14
-CM
CN

1.00
0.12
0.80 0.10
U = 25 m/s, Inactive air jets
U = 25 m/s, VR=1.00, f=50Hz, Cµ = 0.0012
0.60 U = 25 m/s, VR=1.75, f=50Hz, Cµ = 0.0037 0.08
U = 25 m/s, VR=2.00, f=50Hz, Cµ = 0.0049
= 25 m/s, VR=2.80, f=50Hz, Cµ = 0.0100
U
0.06
0.40
0.04
0.20
0.02
0.00 0.00
0 4 8 12 16 20 24 0 4 8 12 16 20 24
α (deg) α (deg)
a) Normal force coefficient b) Quarter-chord pitching-moment coefficient

0.2

Incipient trailing-edge
0.0 separation criterion

-0.2
Cpt

-0.4

-0.6

-0.8

-1.0
0 4 8 12 16 20 24
α (deg)
c) Trailing-edge pressure coefficient
Fig. 7 Variation of CN , CM , and Cpt with angle of attack for pulsed air-jet blowing, with varying air-jet-to-freestream VR at constant freestream velocity
of U  25 m=s and pulsing frequency of f  50 Hz.
 PRINCE AND KHODAGOLIAN 651

Figure 7a, which plots the variation of CN with  for the inactive jet and a jet-to-freestream VR of 2.0. These data are obtained from the
case and compares this with data for four selected pulsed jet cases, various pitch sweep experiments undertaken during the study, where
presents a number of interesting findings. First, scrutiny of the curves the angle of attack is varied steadily while the tunnel freestream
in the vicinity of   16 reveals that the pitch hysteresis, between dynamic pressure is maintained constant.
pitchup and pitchdown, associated with trailing-edge separation is It was found that there is a significant effect of pulsing frequency
not present in the data for VR  2:8. This phenomenon, which is not on the maximum achievable normal and tangential pressure forces.
evident in any of the inactive jet or clean aerofoil data, has been found The optimum pulsing frequency for both CN and CT was found to be
in all of the data for steady and pulsed AJVG blowing at VR  1:0 in the region of F
 0:6–0:8, which agrees with the findings of
and 2.0, so far investigated. The data from this experiment confirm Greenblatt and Wygnanski [28] and Greenblatt et al. [29,30].
that hysteresis between the pitchup and pitchdown data occurs for jet- Figure 10 presents the results of the tests performed to study the
to-freestream VRs of 1.0, 1.5, 1.75, 2.0, and 2.5. For VR  2:8, effect of varying the pulsing frequency with a constant jet VR of 2.0
however, both of the experiments performed for this case reveal no and a freestream tunnel flow speed of 35 m=s at a fixed angle of
similar hysteresis phenomenon. attack,   22 . Four tests were performed. Tests 1 and 3 started with
Interestingly, the same finding holds true for the second hysteresis a high frequency (250 Hz) and an attached leading-edge boundary
event, associated with the collapse and recovery of boundary-layer layer, and they progressed by reducing the pulsing frequency, taking
attachment on the forward 10% of the upper surface. For surface pressure readings at appropriate frequency increments.
VR  1:0–2:5, clear differences in the CN and CM data for pitchup Tests 2 and 4 were similar, except that they began with a low
and pitchdown are evident. For VR  2:8, however, boundary-layer frequency (2 Hz) and a fully separated leading-edge boundary layer,
separation from the leading edge and the associated sudden loss of and they progressed by incrementally increasing the pulsing
leading-edge suction occur at the same angle of attack (  22–23 ) frequency. For each increment in pulsing frequency, the measured
on the pitchup as the recovery of the boundary-layer attachment and pressure distribution was integrated over the surface to yield the
the leading-edge suction on the upper surface occurs on the corresponding magnitude of normal force coefficient. Interestingly,
pitchdown. It is expected that for VR > 2:8, the suppression of any there was a small difference between the data of Figs. 9 and 10, which
hysteresis in aerodynamic characteristics between pitchup and indicate a sensitivity to the procedure used in the experiment. In the
pitchdown will persist. Much more in-depth investigation is required data for Fig. 10, where the angle of attack was constant, the only
to reveal the physical mechanism for this hysteresis suppression variable was pulsing frequency. Surface pressure distributions were
phenomenon at high VR. carefully monitored to ensure that the correct physical starting
The present results, however, clearly show that increasing jet VR conditions for the given test were achieved. With no changes in , the
results in a higher value of achievable CN max but does not disturbances during this experiment were significantly less than
significantly effect the value of CM max (Fig. 7b). As with all of the those with pitch variation.
previous experiments (for both pulsed and steady AJVG blowing), The data for the constant  experiment (Fig. 10) clearly show that,
however, application of AJVG blowing results in a considerably despite a high degree of experimental scatter, a strong correlation
higher pitching moment just before, and following, full stall on the between CN and pulsing frequency is observed. For zero pulsing
pitchup, compared with the inactive jet or clean aerofoil cases. frequency (inactive jets), CN is 1:5, which is the value measured in
In summary, it is suggested that a practical pulsed AJVG system previous tests of the inactive jet/clean surface model at   22 .
should be designed to operate at a VR of at least 2.0. The results of Under this condition, the upper-surface separation line was seen in
this set of experiments confirm that increasing the jet-to-freestream the UV-tuft experiments to undergo transient switching between the
VR, and thereby the blowing momentum coefficient C
, improves fully separated flow case and the case where attachment was
the performance of an AJVG system in effectively suppressing recovered on the forward 10% of the upper surface (see Fig. 11a).
boundary-layer separation and delaying stall. It is therefore more The periodicity in this switching between the two flow conditions
important to maximize the jet velocity than the blowing mass flow was observed to be between 5 and 10 s. It is suggested that, at
rate.   22 , the leading-edge boundary layer is extremely sensitive to
external disturbances, and the flow switches between two bistable
states in response to the level of freestream turbulence.
D. Effect of Pulsing Frequency, U  35 m=s and VR  2:0 Switching on the pulsed air jets at a frequency of 2 Hz resulted in a
Figure 9 presents the variation of CN max and CT max with air-jet significant change in the transients observed during the UV-tuft
pulsing frequency for the case of a freestream wind speed of 35 m=s visualization experiments. The forward 10% of the upper surface was

+ +
Reduced air-jet pulsing frequency, F Reduced air-jet pulsing frequency, F
0 0.4 0.8 1.2 1.6 2 2.4 0 0.4 0.8 1.2 1.6 2 2.4
0.60

1.80

1.75 0.55

1.70
C Nmax

C Tmax

0.50
1.65

1.60
0.45

1.55

1.50 0.40
0 50 100 150 200 0 50 100 150 200
Air-jet pulsing frequency (Hz) Air-jet pulsing frequency (Hz)
a) Maximum CN b) Maximum C T
Fig. 9 The effect of jet pulsing frequency on the maximum achievable normal and tangential force coefficient (U  35 m=s and VR  2:0).
652 PRINCE AND KHODAGOLIAN

+
Reduced air - jet pulsing frequency: F dominant pressure fluctuation at about 45 Hz when the dynamic
2.0
0 0.5 1 1.5 2 2.5 3 pressure sensor entered into the wake region, which was not present
in the freestream. The assumption can confidently be made that this is
associated with natural vortex shedding from the aerofoil. A similar
analysis was performed for U  30 m=s, where vortex shedding
Fully attached flow ?

1.9
appeared to take place at a frequency of about 35 Hz. For both cases
CN at α = 22 º

1.8 Approximate position of

Trailing separation line
edge

1.7

1.6
Fully separated
flow

1.5
0 50 100 150 200 250
Air - jet pulsing frequency (Hz)
Fig. 10 Variation of normal force coefficient with AJVG pulsing
frequency for the RAE 9645 section model for an angle of attack of 22 ,
VR  2:0, and U  35 m=s.
Air jet Leading
orifice edge
seen to become steadily attached, indicating that leading-edge
suction had been permanently reestablished. However, downstream
of the jets, the tufts were seen to respond in a wavelike manner to the a) Inactive air jets
activation of the AJVGs (Fig. 11c). When the jets were off, the
separation line appeared to advance to a position just downstream of
the jet array. When the jets were activated, the blowing appeared to
Approximate position of
force the separation line to retreat back downstream to a position of separation li
x=c  0:5. The separation line was, in this way, clearly seen to
advance and retreat in synchronization with the activation of the
pulsed air jets. Analysis of the integrated CN measurements
undertaken at this condition indicate that, with the reestablishment of
leading-edge suction, the normal force increased to between 1.57–
1.61 by the activation of pulsing at 2 Hz. Further increases in pulsing
frequency were observed in the UV-tuft experiments to increase the
frequency at which the wavelike advance and retreat of the separation
line occurred. The tuft technique was not able to resolve any small
movements in the location of farthest extent, downstream or
upstream, of the advancing–retreating separation line. However, the
movement was clearly seen to remain in synchronization with the
activation of the air jets.
At between f  40–50 Hz, however, a different situation was
observed in the response of the UV tufts. The rapid wavelike
advance–retreat of the upper-surface separation line was no longer b) Steady air jets, VR=2.0
observed, and the separation line appeared to remain fixed at the
downstream position. The maximum frame rate of the video camera
prevented meaningful analysis at these high pulsing frequencies.
Analysis of the calculated CN (Fig. 10) appears to show that a Approximate maximum (upstream)
position of separation line
maximum value of CN  1:85–1:9 is achieved at around f 
50–70 Hz (F
 0:6–0:8). For frequencies above 100 Hz, CN was
found to stabilize at an average value of 1.82. It is important to
remember that, with the exception of the inactive jet case, the mass
flow rates and the jet VR were constant for all pulsing frequencies,
and the observed variation in CN was purely frequency dependent.
The two tests (1 and 3) performed, progressing from high jet pulsing
frequency to low, further confirmed these findings, with the data
following the same broad trends (Fig. 10 presented the combined
data from all four of these tests).
As part of this set of experiments, dynamic pressure measurements
at the trailing edge of the model, at   22 , were recorded for the
case of the clean aerofoil (sealed orifices) in an attempt to measure the Approximate minimum (downstream)
natural shedding frequency of the RAE 9645 model at this angle of position of separation line
attack. This was done in order to investigate any link between the
pulsing frequency, f  50–70 Hz, at which maximum CN is
measured, and a change in flow physics is inferred in the c) Pulsed air jets, VR=2.0, f = 30Hz
experimental data. Analysis of the power density spectrum obtained Fig. 11 Comparison of instantaneous surface flow visualization images
during a vertical traverse through the wake, 10 mm offset of UV tufts on the upper surface of the RAE 9645 section model for
downstream from the trailing edge, clearly identified a very strong
 22 and U  35 m=s.
 PRINCE AND KHODAGOLIAN 653

of U  35 m=s and U  30 m=s, the Strouhal number associated 40–50 Hz, close to the measured natural shedding frequency, and the
with the shedding frequency was found to be about 0.6 for both cases. separation front appeared to remain almost fixed, although this
This corresponds with the value for the Strouhal number of 0.5–0.7 observation has to be fully verified with high-speed cameras in future
found in literature for aerofoils of low to moderate thickness-to- studies.
chord ratios.
Analysis of Fig. 10 shows that maximum CN is achieved with an Acknowledgments
F
of 0:6. This suggests that there is a lock-in effect between
natural shedding frequency and the driving frequency of the AJVGs. This work was funded and supported by the U.K. Ministry of
However, experiments have only been performed for U  35 m=s Defence and Department of Trade and Industry, as well as Augusta–
and VR  2:0 at   22 , where the flow is known to be particularly Westland, Ltd., and QinetiQ, Ltd. The work was supported at City
sensitive to pulsing frequency. To gain a more in-depth under- University by Chris Barber, Mike Smith, and Tim Barnes.
standing of the physics behind pulsed AJVG blowing, and
specifically the effect of pulsing frequency on the flow physics, a References
more in-depth set of experiments are required, covering different [1] Taylor, D. H., and Hoadley, H. H., “Application of Vortex Generator
tunnel flow speeds and jet VRs, as well as the generation of more data Mixing Principle to Diffusers,” United Aircraft Corp. Rept. R-15064-5,
for each condition. East Hartford, CT, 1948.
In summary, there appears to be a clear correlation between jet [2] Wallis, R. A., “The Use of Air Jets for Boundary Layer Control,”
pulsing frequency and the stall suppression effectiveness for a given Aeronautical Research Labs., Aero. Note 110, Melbourne, Australia,
jet VR and freestream tunnel speed. The optimum frequency also 1952.
appears to be equivalent to the local natural shedding frequency of [3] Pearcey, H. H., “Shock Induced Separation and its Prevention by
the aerofoil. A practical pulsed AJVG flow control system can Design and Boundary Layer Control,” Boundary Layer and Flow
therefore be effectively tuned to the local flow conditions to achieve Control, edited by G. V. Lachmann, Pergamon, New York, 1961,
pp. 1166–1134.
the optimum stall suppression effect. [4] Freestone, M., “Preliminary Tests at Low Speeds on the Vorticity
Produced by Air-Jet Vortex Generators,” City University, Department
of Aeronautics Research Memo. 85/1, London, Feb. 1985.
V. Conclusions [5] Pauley, W. R., and Eaton, J. K., “Experimental Study of the
This experimental investigation aimed to conduct extensive tests, Development of Longitudinal Vortex Pairs Embedded in a Turbulent
applying low-momentum steady and unsteady air injection on a Boundary Layer,” AIAA Journal, Vol. 26, No. 7, 1988, pp. 816–823.
RAE 9645 rotor blade section in a nominally two-dimensional flow, doi:10.2514/3.9974
[6] Johnston, J. P., and Nishi, M., “Vortex Generator Jets: A Means of Flow
using pitched and skewed AJVGs to suppress quasi-steady flow
Separation Control,” AIAA Paper 1989-0564, 1989.
separation and/or promote flow reattachment at low speeds. Results [7] Selby, G. V., Lin, J. C., and Howard, F. G., “Control of Low-Speed
from this investigation have led to the following conclusions: Turbulent Separated Flow Using Jet Vortex Generators,” Experiments in
1) Hysteresis between the pitchup (continuous increase in angle of Fluids, Vol. 12, No. 6, 1992, pp. 394–400.
attack from   10 to a maximum of 26 ) and the pitchdown doi:10.1007/BF00193886
(continuous decrease in angle of attack) data was evident in the [8] Compton, D. A., and Johnston, P., “Streamwise Vortex Production by
normal and pitching-moment coefficients for jet-to-freestream VRs Pitched and Skewed Jets in a Turbulent Boundary Layer,” AIAA
of 1.0, 1.5, 1.75, 2.0, and 2.5. No similar hysteresis phenomenon was Journal, Vol. 30, No. 3, 1992, pp. 640–647.
observed for the highest jet velocity case, VR  2:8, or in the data for doi:10.2514/3.10967
the clean model data. [9] Henry, F. S., and Pearcey, H. H., “Numerical Model of Boundary Layer
Control Using Air-Jet Generated Vortices,” AIAA Journal, Vol. 32,
2) Steady AJVG blowing at a VR of 1.0, equal to the local No. 12, 1994, pp. 2415–2424.
freestream velocity, was ineffective in delaying the onset of trailing- doi:10.2514/3.12308
edge separation, while a higher jet blowing velocity with VR  2:0 [10] Zhang, X., and Collins, M. W., “Measurements of a Longitudinal
delayed the onset of trailing-edge separation to a 2 to 3 higher angle Vortex Generated by a Rectangular Jet in a Turbulent Boundary Layer,”
of attack. Physics of Fluids, Vol. 9, No. 6, 1997, pp. 1665–1673.
3) Unsteady AJVG blowing at a VR  2:0, at pulsing frequencies doi:10.1063/1.869286
equal to or above 45–50 Hz, led to a 20% increase in the [11] Zhang, X., and Collins, M. W., “Nearfield Evolution of a Longitudinal
maximum normal force over the clean aerofoil case, while steady Vortex Generated By Inclined Jet in a Turbulent Boundary Layer,”
Journal of Fluids Engineering, Vol. 119, No. 4, Dec. 1997, pp. 934–
AJVG blowing with an equivalent amount of total mass flow led only
939.
to a 5% increase in the maximum normal force. doi:10.1115/1.2819520
4) Unsteady AJVG blowing at or above 45–50 Hz showed [12] Singh, C., Peake, D. J., Coton, F., Kokkalis, A., Khodagolian, V., Coton,
similar suppression of drag divergence (about 2 compared with F., and Galbraith, R. A., “Parametric Study of an Air-Jet Vortex
inactive jets) as steady blowing but using half the total blowing mass Generator Configuration to Control Rotorcraft Retreating Blade Stall,”
flow, while low-frequency pulsing at 10 Hz was much less effective. 43rd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, AIAA
5) A correlation was found between aerofoil shedding frequency Paper 2005-1366, 10–13 Jan. 2005.
and maximum normal force, where a maximum normal force [13] Singh, C., “Application of Airjet Vortex Generators To Control
coefficient was obtained at about 45–50 Hz, corresponding to the Helicopter Retreating Blade Stall,” Ph.D. Dissertation, City University
measured natural shedding frequency. It is suggested that there is a London, 2006.
[14] Oyler, T. E., and Palmer, W. E., “Exploratory Investigation of Pulsed
tuning effect between natural shedding frequency and the driving Blowing for Boundary Layer Control,” North American Rockwell
frequency of the AJVGs. Although more extensive dynamic Corp. Rept. NR 72H-12, 1972.
measurements of aerofoil shedding frequency are necessary in order [15] Oster, D., Wygnanski, I., Dziomba, B., and Fiedler, H., “On the Effect of
to completely verify this. Initial Conditions on the Two Dimensional Turbulent Mixing Layer,”
6) The tufts detected periodic transient switching between fully Lecture Notes in Physics, Vol. 75, 1978, pp. 48–64.
separated and semiattached flow on the aerofoil upper surface at doi:10.1007/3-540-08765-6_5
stalled flow conditions. This was believed to be low- frequency [16] Oster, D., and Wygnanski, I., “The Forced Mixing Layer Between
oscillations induced by unsteadiness in the freestream wind. The Parallel Streams,” Journal of Fluid Mechanics, Vol. 123, 1982, pp. 91–
130.
separation front was seen to advance upstream and retreat back
doi:10.1017/S0022112082002973
downstream in a wavelike manner, in synchronization with the [17] Wygnanski, I., and Pedersen, R. A., “Coherent Motion in Excited Free
activation of the pulsed AJVGs at low frequencies (below 5 Hz). Shear Flows,” AIAA Journal, Vol. 25, No. 2, 1987, pp. 201–213.
The oscillations increased with increasing jet pulsing frequency. doi:10.2514/3.9610
7) The rapid wavelike advance–retreat of the upper-surface [18] Ho, C. M., and Huerre, P., “Perturbed Free Shear Layers,” Annual
separation front was no longer observed at the higher frequencies of Review of Fluid Mechanics, Vol. 16, No. 1, 1984, pp. 365–422.
654 PRINCE AND KHODAGOLIAN

doi:10.1146/annurev.fl.16.010184.002053 Exhibit, Reno, NV, AIAA Paper 1998-0675, 1998.

[19] Katz, Y., Nishri, B., and Wygnanski, I., “The Delay of Turbulent [28] Greenblatt, D., and Wygnanski, I., “Dynamic Stall Control by
Boundary Layer Separation by Oscillatory Active Control,” Physics of Oscillatory Addition of Momentum. Part 1: NACA 0015 Parametric
Fluids, Vol. 1, No. 2, 1989, pp. 179–181. Study,” Journal of Aircraft, Vol. 38, No. 3, 2001, pp. 430–438.
doi:10.1063/1.857555 doi:10.2514/2.2810
[20] Seifert, A., Bachar, T., Koss, D., Shepshelovich, M., and Wygnanski, I., [29] Greenblatt, D., Nishri, B., Darabi, A., and Wygnanski, I., “Dynamic
“Oscillatory Blowing: A Tool to Delay Boundary Layer Separation,” Stall Control by Oscillatory Addition of Momentum. Part 2:
AIAA Journal, Vol. 31, No. 11, 1993, pp. 2052–2060. Mechanisms,” Journal of Aircraft, Vol. 38, No. 3, 2001, pp. 439–447.
doi:10.2514/3.49121 doi:10.2514/2.2811
[21] McManus, K. R., Ducharme, A., Goldley, C., and Magill, J., “Pulsed Jet [30] Greenblatt, D., “Dual Location Separation Control on a Semispan
Actuators for Suppressing Flow Separation,” 34th AIAA Aerospace Wing,” AIAA Journal, Vol. 45, No. 8, 2007, pp. 1848–1860.
Sciences Meeting and Exhibit, Reno, NV, AIAA Paper 1996-0442, doi:10.2514/1.27757
Jan. 1996. [31] Eroglu, A., and Breidental, R. E., “Structure, Penetration and Mixing of
[22] McManus, K. R., Legner, H. H., and Davis, S. J., “Pulsed Vortex Pulsed Jets in Crossflow,” AIAA Journal, Vol. 39, No. 3, 2001, pp. 417–
Generator Jets for Active Control of Flow Separation,” 25th AIAA Fluid 423.
Dynamics Conference, Colorado Springs, CO, AIAA Paper 1994- doi:10.2514/2.1351
2218, June 1994. [32] Johari, H., “Scaling of Fully Pulsed Jets in Crossflow,” AIAA Journal,
[23] Seifert, A., Darabi, A., and Wygnanski, I., “Delay of Airfoil Stall by Vol. 44, No. 11, 2006, pp. 2719–2725.
Periodic Excitation,” Journal of Aircraft, Vol. 33, No. 4, 1996, pp. 691– doi:10.2514/1.18929
698. [33] Scholz, P., Casper, M., Ortmanns, J., Kähler, C., and Radespiel, R.,
doi:10.2514/3.47003 “Leading-Edge Separation Control by Means of Pulsed Vortex
[24] McManus, K. R., Joshi, P. B., Legner, H. H., and Davis, S. J., “Active Generator Jets,” AIAA Journal, Vol. 46, No. 4, April 2008, pp. 837–846.
Control of Aerodynamic Stall Using Pulsed Jet Actuators,” 26th AIAA doi:10.2514/1.26176
Fluid Dynamics Conference, AIAA Paper 1995-2187, June 1995. [34] Ortmanns, J., Bitter, M., and Kahler, C. J., “Dynamic Vortex Structures
[25] McManus, K. R., Ducharme, A., Goldey, C., and Magill, J., “Pulsed Jet for Flow-Control Applications,” Experiments in Fluids, Vol. 44, No. 3,
Actuators for Suppressing Flow Separation,” 34th Aerospace Sciences 2008, pp. 397–408.
Meeting and Exhibit, Reno NV, AIAA Paper 1996-0442, Jan. 1996. doi:10.1007/s00348-007-0442-8
[26] McManus, K. R., and Magill, J., “Separation Control in Incompressible [35] Jones, B. M., “The Measurement of Profile Drag by the Pitot Traverse
and Compressible Flow Using Pulsed Jets,” AIAA Paper 1996-1948, Method,” Aeronautical Research Council Rept. 1688, 1936.
June 1996.
[27] Magill, J. C., and McManus, K. R., “Control of Dynamic Stall Using F. Coton
Pulsed Vortex Generator Jets,” 36th Aerospace Sciences Meeting and Associate Editor