USE OF SUPERSONIC MICROJETS FOR ACTIVE SEPARATION CONTROL IN DIFFUSERS

Vikas Kumar* and Farrukh S. Alvi†

Department of Mechanical Engineering Florida A & M University and Florida State University

2525 Pottsdamer Street Tallahassee, FL-32310

Inlets to aircraft propulsion systems must supply flow to the compressor with minimal pressure loss, distortion or unsteadiness. Flow separation in inlets and ducts can reduce the overall system performance. The present paper describes an experimental investigation carried out to study the feasibility of using supersonic microjets to control boundary layer separation in an adverse pressure gradient. The geometry used for this study is a simple diverging “Stratford ramp” equipped with arrays of 400µm supersonic microjets. Measurements include detailed surface flow visualizations, surface pressure distributions, flow visualizations and velocity field measurements using Particle Image Velocimetry. The results clearly indicate that by activating these microjets, the separated flow was eliminated. This was achieved with minimal mass flux, lower than 1.5 % of the primary flow. The activation of microjets and the resulting elimination of separated flow led to a significant increase in the momentum of the flow near the surface. The increase in momentum was at least an order of magnitude higher than the momentum injected by the microjets. As such, supersonic microjets appear to be very effective actuators for separation control.

1. Introduction Flow separation is generally accepted to be the

breakaway or detachment of fluid from a solid surface. Whether caused by a severe adverse pressure gradient, a geometrical aberration or by any other means, separation is usually accompanied by significant thickening of the rotational flow region adjacent to the surface with a marked increase in the velocity component normal to the surface. This separation is almost always associated with losses of some kind, including loss of lift, drag increase, pressure recovery losses and in some cases, it is also the cause for aerodynamic stall10,11. While the inlet length required to avoid separation and its associated losses may not be a significant design driver for some vehicles (such as uninhabited air vehicles), the inlet may drive the size of the overall vehicle1. For many military applications, the inlet design is also constrained by low observability requirements. To reduce the radar signature from the compressor face, a serpentine inlet is typically used (Fig. 1) to block the line of sight2,3. Similar buried propulsion system installation has also been considered for Blended Wing Body (BWB) design. In case of BWB4, engines are located at the aft end of the aircraft that requires the ingestion of a thick boundary layer developed over the aircraft surface.

* Graduate Research assistant, FSU, Student Member AIAA †Associate Professor, FSU, Senior Member AIAA

The degraded condition of this thick boundary layer makes it much more susceptible to separate when it encounters the pressure gradients of a diffusing inlet duct. The pressure loss due to this separation reduces the overall system efficiency. Moreover, flow distortion and unsteadiness created due to this separation can result in aerodynamic stall and surge of the compressor and the fan blades5,6. Henceforth, it is highly desirable to avoid boundary layer separation as it can significantly diminish the engine performance.

Not surprisingly, there has been a tremendous amount of research and development to control this boundary layer separation7,8. Traditionally, beyond tripping a laminar boundary layer, four major approaches were used for separation control: 1. Tangential blowing (in various forms including slotted flaps, and moving wall) to directly energize the low momentum region near the wall, 2. Wall suction to remove the low momentum region, 3. Vortex generators (v.g’s) in form of vanes, bumps, 4. Forced excitation devices e.g. synthetic jets. Tangential blowing and suction are very effective in controlling the separation. However, they have the parasitic cost of very high-pressure requirement involved and are infrequently used. Vortex generators are among the most widely examined flow control methods, where v.g.’s of various shapes and sizes have been used to control the boundary layer separation9. Although the mechanism is still not understood very well, it has

1 American Institute of Aeronautics and Astronautics

33rd AIAA Fluid Dynamics Conference and Exhibit23-26 June 2003, Orlando, Florida

AIAA 2003-4160

Copyright © 2003 by Vikas Kumar. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

2 American Institute of Aero

powered by a radial blower, which passes through a settling chamber before entering the test section. Honeycomb screens present in the settling chamber ensure the flow uniformity. The contraction then accelerates the flow from the settling chamber into the test section, further reducing the percentage variations in the velocity. A Pitot static probe was used for measuring the inlet flow speed of the wind tunnel.

nautics and Astronautics

parasitic drag when they are not in use.

, such as synt

2. Experimental Details

been suggested that the v.g.’s produce strong vortices, which enhance the mixing between the high momentum core flow and low momentum boundary layer flow, thus energizing the boundary layer fluid10. However, the performance of these v.g.’s which are passive in nature has been varied; usually there is a need to optimize their location, size and other parameters to achieve optimal performance. Also, they have an associated

Some active flow control deviceshetic jets11 have also been examined for

separation control applications. However, studies have shown that these synthetic jets do not have sufficient velocity/momentum output to achieve effective control12. In this study, we plan to investigate the efficacy of supersonic microjets to control the flow separation. Supersonic microjets are believed to achieve this flow control by high momentum/low mass injection, creating strong streamwise vortices13, thereby enhancing mixing and still retaining the essence of active flow control devices.

2.1 Facility and Model

The experiments were conducted in a subsonic, closed return, wind tunnel with a maximum freestream velocity of 60m/s in the 48”x24” test section. The diffusing duct flow is

he geometry used for the test model is a simple adve milar to model used at the NASA Langley Research Center

iveness of various control tech

Trse pressure gradient ramp, si

for examining the effectniques7. This geometry produces a Stratford

like pressure gradient14 in the test section. A picture of this model, mounted in the wind tunnel, is shown in Fig. 2. The pressure distribution in the test section is determined from the surface pressure measurements made along the ramp. This ramp is equipped with approximately 50 surface static pressure ports to obtain mean pressure distributions. The pressure ports were placed along the model centerline, as well as in transverse rows at selected locations. The pressure coefficient, Cp, is then determined using the conventional definition

25.0)(

∞

−=

UPP

C staticsurfacep ρ

iFig.2- Ramp model, mounted n the wind tunnel

Fig.1- Serpentine Inlet

A schematic of the ramp is shown in Fig. 3. As shown in the figure, the ramp begins at X=0″ and is preceded by a flat region of 21.5″. The figure also indicates the region where the PIV measurements, discussed later in this paper, were obtained. As shown in this figure, PIV data was obtained at two longitudinal planes, along the centerline and 0.1S away form the centerline, where S is the span of the ramp. This off-centerline plane is referred to as the “0.4S” location. Also, shown in the figure are the locations of microjets, where only the 1st and 3rd array of microjets were used for control purposes. Each row consists of about 60 microjets,

d vertically, i.e. 90° with respect to the free-stream. The placement of these microjets with respect to

400µm in diameter. The microjets were oriente

separation location is expected to be a critical parameter for the control scheme. It was also anticipated that the actuators should be placed upstream of the separation zone for maximum effect. Since the exact location of the separation was not known, the microjets were placed upstream of the region where separation was observed in the NASA study7. The supply to these microjets was made using compressed nitrogen tanks. Nitrogen was used because it is easily available in pure form and has essentially the same gas dynamic properties as air.

2.2 Data Acquisition

Both qualitative and quantitative measurements were taken for a better understanding of the flow field. Mean surface static pressure measurements were first obtained by sequentially scanning a series of pressure taps, suitably distributed over the ramp surface. The pressures were scanned using a 48-channel scanivalve unit, which in turn was connected to a pressure transducer. The pressure transducers were calibrated before each data acquisition. The PC based data acquisition system, using National Instruments Data acquisition hardware and software, was used to obtain the data. An average of 4000 samples was acquired to obtain an accurate measurement of mean surface pressures.

The topographic nature of the surface flow field was then studied using surface flow visualization. This was achieved by applying water based paints onto the surface of the ramp. The ‘painted’ model was then exposed to the flow to obtain the surface flow pattern.

The flow field above the surface was also qualitatively visualized using the Planar Laser Scattering (PLS) technique. The test section was seeded with smoke particles approx 5µm in diameter using a Rosco fog generator. The flow was then illuminated in a given plane using a thin

es were then acquired using a S-

field. Each PIV image pair was then acquired using Kodak ES1.0 high-resolution CCD camera capable of recording 10-bit digital image pairs in separate frames at a rate of 15-image pairs/second. Further details of this PIV technique can be found in Lourenco15 et al. One of the main advantages of this PIV technique is a novel processing scheme with high spatial resolution that uses image matching to extract the particle displacements, hence the velocities, from particle image pairs16.

2.3 Test conditions

laser sheet. These imagVHS Panasonic video camera. Although both

pulsed and CW lasers were used to obtain instantaneous and mean flow visualization images, the results presented here are limited to the instantaneous visualizations

To obtain whole- field velocity data in this flow, quantitative measurements were obtained using 2D Particle Image Velocimetry (PIV) technique. A New Wave Nd-YAG pulsed laser with a repetition rate of 15Hz was used to illuminate the particles introduced into the flow

Experiments were conducted over an extended range of freestream velocities of 10m/s, 30m/s, 40m/s and 50m/s. For all of these freestream velocities, various combinations of

Ramp L.E., X= 0″

microjet loca

will primarily be limi

tion and pressure were studied. However, for the sake of brevity, unless otherwise stated, results discussed here are mainly limited to 40m/s as it illustrates the principal flow features and its response to microjet control. Also, the effect of microjet control presented here

ted to microjet array MJ3. The Reynolds number at the ramp leading

edge at 40m/s is 1.256 × 106. At this velocity, the boundary layer thickness, δ, measured at the centerline, 3” before the leading edge of the ramp and was found to be 0.74 inches. The boundary-layer profile was in close agreement with seventh power law profile indicating that the incoming boundary layer is nominally turbulent.

3. Results and Discussion

3.1 Baseline case: No control

Fig. 4 shows comparison between measured Cp distributions for the baseline, uncontrolled flow at 40m/s and with microjet control where the MJ1 array has been activated at 25psig. As can be noted, Cp values have been plotted on reverse scale on y-axis. For reference, the ramp geometry is also included, where the ramp height is indicated on the y-axis on the right and the location of first and last array of microjets is shown in dashed lines. The

Supersonic Microjets

X= 1.7″ MJ1

X= 2.7″ MJ3

Axial extent of PIV region ~2.3″ to 6″ from L.E.

Y ZBase X

Fig.3- Schematic of the ramp model

3 American Institute of Aeronautics and Astronautics

repeatability of these experiments was checked over a period of four months and results were found to be repeatable within 1%. The nature of this curve is very similar to the Cp distribution obtained for NASA model12 indicating the similarity between the two models. Further, it ca

tic

iefapT

F a

n be observed that upstream of the microjet location,

a similar pressure distribution. Dow

skin friction lines spiral around a single point. Saddle points are typically bound between adjacent nodes. They are characterized by two perpendicular lines passing through a point and rest of the lines converging to these two lines. For a node, all skin friction lines are tangential to a common line. For a nodal point of attachment, the flow direction is

gion. However, three-dimensional effects can be

both the cases show nstream of the microjet location, however, the

pressure distributions differ, the pressure recovery being higher in case of MJ1 at 25psig. However, based on Cp distribution alone, it is difficult to determine if the effect of microjets is beneficial or not?

The surface flow field was then visualized to characterize the nature of the separation region on the model. The surface flow streamlines are skin friction lines as they are formed under the influence of shear stress rather than pressure gradients. However, this is not true for reverse flow regions as the surface shear stress is too small and the motion is a result of both shear stress and pressure gradient. As such, the lines of separation indicated by the flow pattern are normally ahead of the actual separation location17. Fig. 5a shows the surface streamline pattern for flow without control.

Also drawn on Fig. 5a are the streamlines indicating the flow direction. The flow pattern obtained is very similar to “Owl Face” pattern of first kind18. It consists of two foci and two saddle points, which are marked in Fig. 5a as F1, F2, S1 and S2 respectively. A focus exists where all the

outbound and for a nodal point of separation, it is inbound.

The horizontal streamlines in the upstream, left half, of Fig. 5a indicate that the flow field is fairly uniform and relatively two-dimensional in this re

observed right after the leading edge. The ‘secondary flow’ near the ramp edges also (see Fig. 5a) appears just downstream of the location where the ramp begins. The nature of this secondary flow and its effect, if any, on flow separation is not clear at this point. However, the lack of sidewalls in the present tests may be partly responsible for this flow, a matter that will be explored further in ongoing experiments.

MJ3 MJ1

Fig. 4- Cp r e at /s; with and without microjets.

dist ibution along the centerlin40m

F2

F1 Separated flow

Secondary flowa)

S1 S2

b)

4 American Institute of Aeronau

ig. 5- Surface flow trace at 40m/s; ) No control b) MJ3@25psig

s and Astronautics

The reverse or separated flow region is also ndicated in Fig. 5a. The su s manating from the f agnation point (S1) orm a separation bubble, which ultimately spirals round two focal points F1 and F2. These focal oints are al he point of local ssure minima. he streamlines then reattach at S2. In a nutshell,

rface streamlineront st

so t pre

the surface flow pattern shows a ‘trapped’ separation bubble, generated due to the ramp curvature. As the velocity measurements, presented late

Figure 6a shows the flow field, along the model centerline, obtained using the PLS technique. Smoke was introduced from the

t generated using an Nd-YAG laser. In the

r confirm, this is indeed a region of low reverse velocities, which is consistent with a zone of separated flow. The surface slow visualization with control (Fig. 5b) is discussed later on in this paper. Next we examine the flow above the surface using qualitative and quantitative optical/visualization methods.

pressure taps in the upstream section of the model. When the smoke is injected directly into the vorticity carrying fluid, such as boundary layer, it marks the progress of that vortical fluid and formation of vortices, if that occurs. These entrained particles were visualized using a thin laser shee

upstream region, the flow seems to be turbulent and attached to the surface, however, as we go further downstream, the structures appear to be detaching from the wall, indicating separation. Also shown in the figure is location of microjet 1. A discussion of the flowfield with microjets activated at 25 psig, shown in Fig 6b is delayed until later in the paper. Although PLS images yield some insight into the flow properties, they were not sufficient to understand the details of separated flow, especially the extent of the separation zone and the magnitude of the reverse velocities. To answer some of these questions, quantitative velocity-field measurements were obtained using the Particle Image Velocimetry (PIV) technique, results of which are discussed next.

RAMP Surface se flow

PIV region

RAMP Surface

RAMP Surface

a)

b)

c)

d)

Rever

Fig. 6- Flow Visualization (PLS) at 40m/s o Control b) MJ1@25psig

b)

a) N

a)

Ram p MJ1 @ 25psi

5 American Institute of Aeronautic

c) Vertical velocity, V d) Spanwise Vorticity, ω .

Fig. 7- Velocity and Vorticity data for 40m/s; No Cont a) Streamwise velocity component, U b) Velocity vector for region indicated in (a)

rol.

y

s and Astronautics

Figure 7a and 8a show the mean streamwise velocity, U field at 40m/s and 50m/s, respectively

id velocity, whi

flow is from left to right and microjets are issuing from the top to the bottom of the photograph. The inset in fig. 7a shows the ramp geometry and the region where the PIV data was obtained. In these plots, all the dimensions are in inches where the y-axis is the vertical distance from the base of the ramp and the x-axis represents the streamwise distance from the ramp leading edge.

A closer look at both plots shows that as one proceeds downstream in the vicinity of the surface, there is a rapid deceleration in the flu

RAMP Surface

RAMP Surface

RAMP Surface

a)

b)

c)

ch eventually leads to a region of reverse flow. This reverse flow zone, which exists for both cases, corresponds to the dark blue velocity contours and starts between x = 3.5 and 4 inches and ends between x = 5.5 and 6 inches. A small sub-region of this flow field, roughly indicated by the box in Fig. 8a, has been magnified and shown in Fig. 8b. Velocity vectors have also been plotted in this figure to provide a better feel for the local flow direction. The presence of reverse flow, close to the ramp surface is clearly visible in Fig. 7b, indicating that the flow has separated locally and a separation bubble, with recirculating flow, is visibly present. It was also observed that increasing the freestream velocity causes the flow to separate further upstream while the flow attachment location does not change appreciably. This can be confirmed by comparing Fig. 7a and Fig. 8a, where the separation location moves from x ≈ 3.9″ in. at 40m/s (Fig. 7a) to x ≈ 3.6″ for the 50m/s case (Fig. 8a). The presence of reverse flow is further established from the contour plots of vertical velocity (in the z-direction, perpendicular to freestream), shown in Fig. 7c for 40m/s case and Fig. 8b for 50m/s case. Low magnitude velocity moving away from the boundary can be observed in the same region where reverse flow field is seen in the streamwise velocity plots. The presence of a strong vorticity field close to the surface, as seen in Fig. 7d for 40m/s case and Fig. 8c for 50m/s case, emphasizes the presence of significant spanwise vorticity, ωy (see Fig. 3 for coordinate system) due to large velocity gradients. These large velocity gradients can be either due to development of a thick boundary layer or due to separated flow, both of which are detrimental for the engine inlets. In accordance with the velocity field data, it can be argued that the vorticity plot indicates a region of separated/reverse flow.

FCbc

ig. 8- Velocity and Vortcicity data at 50m/s; No ontrol. a) Streamwise velocity component, U. ) Vertical Velocity component, V ) Spanwise Vorticity, ω

in the centerline plane. Note that in these and all subsequent PIV frames included in this paper, the ramp model is on the top of the plot, the freestream

r, where the size of the sepa

To summarize, the adverse pressure gradient along the ramp leads to local separation of the incoming boundary laye

y

ration bubble increases by approximately 15%(from 1.9″ to 2.2″) with increase in freestream velocity from 40m/s to 50m/s. In the following

6 American Institute of Aeronautics and Astronautics

section, we examine effect of microjets on this separated flow.

3.2 Effect of Microjet control

It was observed that when the microjets are turned on at the appropriate pressure, the reverse

gion is eliminated. The

be observed, no regions of reverse velocity are

ar effect. A comparison with Fig. 6a shows that

line plane at 40m/s, where the 3rd arra

um coefficient, Cµ were then defi

where, δ is the boundary layer thickness at the ram

he mass input, m* due to supersonic

es.

flow or the separated flow rese observations were confirmed with the PIV

results, discussed later in this paper. This effect of microjet control was observed for all the conditions where separated flow was present for the baseline case.

The surface flow visualization (Fig. 5b) shows the effect of microjet on the surface flow pattern. As can

visible. The flow pattern, however, is not completely two-dimensional and three-dimensional features, such as diverging streamlines, can be noticed. As discussed later, it is anticipated that one of the mechanism at work in the present control technique is the introduction of streamwise vorticity by the microjets. As such, one can argue that some of the features in Fig. 5b are similar to footprints of streamwise vortices. However, at present the exact nature of these features is not clear. What is clear is the fact that the microjets eliminate the large separation bubble seen in Fig. 5a.

The PLS image, shown in Fig. 6b with the 1st array of microjets at 25 psig appears to show a simil

with microjets turned on, the flow near the surface becomes fuller and excursions of large-scale structures from the near-surface region into the outer flow is reduced. This again conveys the idea that the microjets have a desirable effect on the separation. However, based on only the surface pressure distributions, surface flow visualization and PLS images, it is difficult to make this assertion with confidence. Also, it is very difficult, if not impossible, to determine the exact location of flow separation and probable reattachment due to microjets with the PLS images alone. In the following, we further examine the PIV results to better quantify the effect of microjets on the separated flows.

The first parameter considered was the freestream velocity. Fig. 9a presents the velocity data for the center

y of microjets is operating at 25psi. A comparison of Fig. 9a with Fig. 7a shows that the activation of microjets completely eliminates the reverse flow region. Although not shown here for the sake of brevity, similar effects were observed for the 50m/s case also. It was also noticed that at 40m/s, the velocity measured with control near the

surface was somewhat higher than that in 50m/s case. This suggests that the ratio of the microjet momentum relative to the freestream momentum may be an important parameter. This effect of the momentum flux ratio is of course well known and has been discussed by numerous other investigators19.

To quantify the effect of mass and momentum flux input, the mass flux coefficient, M*, and the steady moment

ned as M*= (Mass Input)/(Mass deficit based on δ)

p leading edge. T

microjets is estimated by assuming choked flow through micro-nozzl

Then,

δρ zUmM

∞∞

=*

*

where, ρ∞ is the freestream density, U∞ is the freestream velocity& z is the width of the model.

The the steady momentu iven as

,

conventional definition of m coefficient is used19 and is g

δρ zU 25.0 ∞∞µ

UNmC j

*

=

where, N is the number of microjets,

& Uj is the jet velocity.

ious microjet operat lotted in Fig. 10. As expected, C and M* have a parabolic relation, so that

Values of Cµ and M* for varing pressure have been p

µ

small increments in M* yield progressively larger Cµs. It was observed that an increase in Cµ results in higher velocities closer to the ramp surface, indicating the presence of higher momentum nearer to the wall. The reason behind this is that with an increase in Cµ, a) Momentum is directly injected into the boundary layer, b) Strong streamwise vortices are generated which enhance mixing with the outer, high momentum fluid and c) The microjet jet momentum and penetration depth20,21 increases, enhancing the transfer of momentum from the mainstream fluid to boundary layer fluid.

7 American Institute of Aeronautics and Astronautics

Cµ(%) Vs M*(%) 2nd Order Fit

RAMP Surface

RAMP Surface

RAMP Surface

b)

a)

c)

d)

tpw(svccidc

M

Fig. 9- Velocity and Vorticity data for 40m/s; MJ3@25psig. a) Streamwise velocity component, U

city, V d) Spanwise Vorticity, ωy. b) Velocity vector for region indicated in (a) c) Vertical velo

8 American Institute of Aeronautic

Fig. 10- Steady momentum coefficient, Cµ Vs.ass flux coefficient, M*.

the ong with the velocity

vectors is also shown in Figure 9b. A comparison

A magnified view of a selected region for40m/s case (Fig. 9a), al

with Fig. 7a shows that with the activation of microjets, velocity vectors in the reverse flow zone change their direction and now possess significant momentum in the forward direction. At 25 psig, the mass flux supplied by the microjets is approximately 1.75 % of the mass flux across the ramp, based on boundary layer thickness, i.e. M* = 0.0175. At this pressure, the Cµ value was observed o be 39.7%. Also shown are the vertical velocity rofiles (Fig. 9c) and vorticity profiles (Fig. 9d) ith microjet control. A comparison of Fig. 9c

controlled case) with Fig. 7c (uncontrolled case) hows that there is a substantial increase in vertical elocity component towards the surface for the ontrolled case (Fig. 9c). The spanwise, vorticity ontours in Fig. 9d for the same region clearly ndicate a reduction in this component of vorticity, ue to microjets, when compared to uncontrolled ase (Fig, 7d). This reduction in the ωy may partly

be due to the redirection of the spanwise vorticity in the streamwise direction when the microjets are activated. As mentioned earlier, we believe that the generation of significant streamwise vorticity is one of the mechanisms responsible for separation control. Direct measurements of the streamwise vorticity, presently underway, should resolve this issue.

Although the microjets clearly eliminate the separation zone along the centerline, since this separation is three-dimensional, as observed in the surface flow visualizations, the effect of microjets away from the centerline was also examined. Ideally, one would like to obtain velocity measurements in planes perpendicular to the axial direction and we plan to make such measurements in the future. However, PIV measurements were obtained at two off-centerline axial planes to provide some insight into the three-dimensional

s and Astronautics

RAMP Surface RAMP Surface

a) a)

RAMP Surface RAMP Surface

b) b)

Fig. 11-Off centerline ‘0.4S’; No Control a)Streamwise velocity, U b)Vertical velocity, V

Fig. 12-Off centerline ‘0.4S’; MJ3@25psig a)Streamwise velocity,U b)Vertical velocity, V

effects. The results are shown here for one off-axis plane only, marked as the ‘0.4S’ plane in Fig. 3. Figs. 11a-b and 12a-b shows the velocity contour plots for the uncontrolled and controlled case, respectively along this plane. The freestream velocity in the test section is 40m/s for these figures and the 3rd array of microjet is activated at 25psig for the control case shown in Fig. 12a-b. It is observed that the size of the bubble in the 0.4S, off-centerline case (Fig.11a), is smaller than that in the centerline case (Fig. 7a). It was observed in general, that the size of the bubble decreases as we move away from the centerline, which agrees with the shape of the separation zone seen in Fig. 5a. Comparison of Fig. 11a and Fig. 12a also shows that the activation of microjets was successful in eliminating the separated flow at off-centerline locations as well. However, it should be noted that the effect of control was not as pronounced as in the centerline case (Fig. 9a). For example, the magnitude of the U component of the velocity, near the ramp surface (Fig. 12a) is less than the velocities measured along the centerline (Fig. 9a). Furthermore, Fig. 12b, shows the presence of

negative velocities near the ramp, downstream of x ~5.6”. This may suggest that a smaller separation zone is present at this location. This is consistent with the surface flow pattern (Fig. 5b), where a small ‘blob’ can be observed off the centerline. The reasons for this are attributed to the three dimensional nature of the separation bubble. Although the details of the effect of microjets at off-centerline locations are not clear at present, there is no uncertainty that overall, it has positive effect and that the microjets significantly reduce the size of the separation region.

In general, any separation control input must be applied at or close to the separation point. To investigate the effect of actuator location, experiments were conducted with microjets at different axial locations along the ramp. Result can be seen in Fig. 13, where the 1st array of microjet has been activated at 25psig. As seen in Fig. 13, it is clear that the microjets were successful in eliminating the reverse flow region. Comparing this effect with the results shown in Fig. 9a, for the 3rd array of microjets (MJ3), it appears that, although both microjet arrays were able to

9 American Institute of Aeronautics and Astronautics

eliminate separation, higher velocities closer to the surface were obtained with the activation of MJ3. It was also noticed that separation was eliminated with a lower momentum flux input using MJ3 compared to MJ1.

In order to determine whether the significant increase in the momentum near the ramp surface is simply due to the direct injection of momentum by the microjets, a momentum gain ratio was defined. This is the ratio of the increase in momentum due to microjets (= momentum of controlled flow – momentum of uncontrolled flow) to the momentum injected by the microjets. A comparison of the momentum gain achieved with application of different microjets at various operating pressures is listed in Table 1. As seen in this table, for the same Cµ values the gain achieved with MJ3 is significantly more than that achieved with MJ1. In particular, at 25psig, the momentum gain with MJ1 is 6, while the momentum gain achieved with MJ3 is 21.6. In general, it can be inferred that, closer one is to the upstream of separation location; more effective control can be achieved.

Although the physical mechanism behind microjet control still needs to be explored further, our results strongly suggest that the streamwise vorticity due to the microjets plays a primary role in this control approach. The generation of streamwise vorticity can be due to number of mechanisms. First, the microjets may behave as fluidic ‘tabs’ much like the micro-v.g.’s used in earlier work9,10. Second, the vorticity in the microjets is redirected in the streamwise direction by the mean flow, and finally the microjets may also redirect the spanwise vorticity in the base flow in the streamwise direction by vorticity ‘tilting’. These mechanisms may be similar to those discussed by Alvi22 in the context of impinging

jets. Additional, more detailed measurements are planned to shed more light on the physics behind this control mechanism. The effect of microjet location, microjet angle as well an examination of the flowfield in the cross-plane are some areas that require further

exploration in order to better understand the mechanism for flow control. We are presently conducting experiments to address some of these issues.

Table 1: Momentum injected versus Momentumgain achieved for various microjet pressure andlocation at 40m/s.

RAMP Surface Microjet activated

Microjet pressure (psig)

Momentum Injected (kg.m/s2)

Momentum gain achieved (kg.m/s2)

Gain Ratio

1 10 0.7 7.4 11.3 1 15 1.1 8.4 7.7 1 20 1.6 10.1 6.2 1 25 2.2 13.2 6 3 10 0.7 25.8 39.4 3 15 1.1 34.7 31.8 3 20 1.6 41.8 25.8 3 25 2.2 47.8 21.6 Fig. 13- Streamwise velocity, U ; MJ1@25psig

4. Concluding remarks

A study of separation control in an adverse pressure gradient region over a Stratford ramp has been performed. Separation control was implemented using strategically placed arrays of supersonic microjets. Initial results, which consist of surface pressure distributions, surface flow visualizations, PLS and PIV measurements, indicate that the microjets have an appreciable and desirable effect on flow separation. Over the range of conditions examined, the reverse or separated flow regions are either completely eliminated or drastically reduced in size. The critical parameters studied include Reynolds number, steady momentum coefficient, Cµ and the microjet location. As a representative case, for Reynolds number of 2.3×106/m, the separation bubble was approximately 16% of the ramp length and was completely eliminated with mass flux input lower than 1.5% using the microjets. The steady momentum coefficient Cµ for this microjet pressure (20psig) was 29.1%. The control is achieved by injection of high momentum fluid and by generating strong streamwise vortices, responsible for mixing. As such, supersonic microjets show considerable promise as a simple, adaptable and effective ‘actuators’ for controlling flow separation and flow distortion.

10 American Institute of Aeronautics and Astronautics

5. Acknowledgements

This study is supported by the NASA Langley Research Center under grants monitored by Ms. S. Gorton; the authors are grateful for this support. We thank Mr. B. DePriest for his considerable assistance in designing and fabricating the hardware and models. The authors also appreciate the contributions of Dr. B. Alkislar in conducting the experiments.

4. References

1. MacMartin, D.G., Verma A., Murray R.M., Paduano J.D., “ Active control of integrated inlet/ compression systems” FEDSM2001-18275, June 2001.

2. Mayer, D.W., Anderson, B.H., Johnson T.A., “3D subsonic diffuser design and Analysis”, AIAA98-3418.

3. Wellborn, S. R., Reichert, B.A., Okiishi, T.H., “Study of compressible flow in a diffusing S-duct.”, Journal of Propulsion and Power, Vol. 10, No. 5, Sept-Oct 1994, pp. 668-675.

4. Anabtawi,A.J., Blackwelder,R.F., Lissaman,P.B.S., Liebeck,R.H., “An experimental investigation of boundary layer ingestion in a diffusing S-duct with and without passive flow control”, AIAA 99-0739.

5. Rabe, D., Boles, A. And Russler, P. “ Influence of Inlet distortion on transonic Compressor blade loading”, AIAA 95-2461.

6. Wygnanski, Seifert A., “Delay of airfoil stall by periodic excitation” AIAA J Aircraft 1996;33(4):691–8.

7. Lachmann, G.V. “Boundary layer and Flow control, Its principles and applications, Vol. 1, Permagon press, New York, USA, 1961, pp.1, 2.

8. Chang, P.K., Control of Flow separation, Hemisphere publishing corporation and McGraw Hill, New York, USA, 1970, pp. 154-177.

9. Storms B. L, Ross JC. “Experimental study of lift-enhancing tabs on a two-element airfoil.” AIAA J Aircraft 1995; 32: 1072–8

10. Oster D., Wygnanski, I., Dziomba B., Feidler H., “The effect of Initial conditions on the two dimensional turbulent mixing layer. In: Fiedler H., editor. Structure and mechanics of turbulence. Lecture notes in Physics, Vol. 75. Berlin, Springer, 1978. p. 48-64.

11. Amitay, M., Smith, B. L., and Glezer, A. “Aerodynamic Flow Control using Synthetic Jet Technology,” AIAA Paper 98-0208.

12. Jenkins, L.N., Gorton, S.A., Anders S.G., “Flow control device evaluation for an internal

flow with an adverse pressure gradient” AIAA, Reno 2002.

13. Lou, H., Alvi, F.S., Shih, C., Choi, J., and Annaswamy, A.,, “ Active Control of Supersonic Impinging Jets: Flowfield properties and Closed loop strategies” AIAA Paper 2002-2728, 1st AIAA Flow Control Conference and Exhibit, June, St. Luis, MO.

14. Stratford, B.S., “The prediction of separation of turbulent boundary layer”, J. Fluid Mech. 5, 1-16, 1959b.

15. Lourenco, L.M., “True Resolution PIV: A mesh free second order accurate algorithm” International Conference in application of lasers in Fluid mechanics, Lisbon, Portugal, July 2000.

16. Lourenco, L.M., Krothapalli A., “Mesh free second order algorithm for PIV processing”, Proceedings of the International conference on Optical technology and Image Processing in flui, thermal and combustion flows, Yokohoma, Japan, Dec. 1998, pp. 24.

17. Barberpoulos A.A, Garry K.P., “ The effect of skewing on the vorticity produced by an airjet vortex generator”, The Aeronautical Journal, March 1998.

18. Alvi, F. S., Elavarsan R., Shih, C., Garg G., and Krothapalli, A., “ Control of supersonic impinging jet flows using Microjets,” AIAA Paper 2000-2236, to appear in the AIAA Journal, 2003.

19. Greenblatt David, Wygnanski, I.J., “The control of flow separation by periodic excitation” Progress in Aerospace Sciences 36(2000) 487-545.

20. Norster E.R., “Jet penetration and mixing studies”. Unpublished work, College of Aeronautics, Cranfield.

21. Lefebvre, A.H., “Gas turbine combustion” Mcgraw Hill, New York, 1983.

22. Alvi, F.S., “The role of Streamwise vorticity in the control of Impinging Jets”, FEDSM2003-45062.

11 American Institute of Aeronautics and Astronautics