You are on page 1of 10

AIAA-2006-2427

Analysis of Control Concepts for Cavity Flows


Srinivasan Arunajatesan
1
, Chandrasekhar Kannepalli
2
and Neeraj Sinha
3
Combustion Research & Flow Technology, Inc. (CRAFT Tech)
6210 Kellers Church Rd. Pipersville, PA 18917
Phone: 215-766-1520/Fax: 215-766-1524
ajs@craft-tech.com
In this paper, we describe on-going research to identify physical mechanisms responsible
for the suppression of surface pressure loads on cavities when subjected to control. The
primary control concepts under examination are mass blowing concepts. One concept uses
an array of microjets across the leading edge of the cavity. The second concept uses slot
blowing, with several slot configurations. This study represents the first numerical analysis
of control using microjets and these are the main focus of this study. The results from the
slot blowing cases will be reported in a future report. The studies are carried out using
Large Eddy Simulations of the flow field. The calculations have been validated against
detailed experimental measurements of both surface pressures and PIV measurements of the
flow field. The calculations show that the microjets are very effective in suppressing shear
layer oscillations. The primary control mechanism appears to be an alteration of the shear
layer receptivity at the upstream end to disturbances. The near field shear layer structure is
altered and based on preliminary analysis of the turbulent kinetic energy budget significant
differences in the near field budget are seen. The shear layer flapping, very clearly evident in
the uncontrolled cavity is suppressed, resulting in lowered surface pressure fluctuation on
the cavity walls.
I. Introduction
An active program is underway at CRAFT Tech to develop a control system for the improvement of dynamic
surface loads on the walls of and, improve store separation envelope from, internal weapons bays in modern military
aircraft. The program, conducted in partnership with National Center for Physical Acoustics (NCPA), University of
Mississippi and Florida State University (FSU), consists of several components, employing experimental and
computational methods. One of the goals of this program is to develop an understanding of the physical mechanisms
responsible for control. Large Eddy Simulations, along with detailed surface pressure measurements and Particle
Image Velocimetry techniques are being used to achieve this goal. Here, we present some results from the
computational studies carried out as part of this program.
Control of cavity flows has been an area of active research in the past few years due to increased emphasis on
internal weapons carriage in air vehicles. Reducing the dynamic loads on the cavity surfaces can lead to life cycle
gains for air-vehicle and store structures. Further, control also has the potential to improve separation characteristics
for these stores during deployment. This is especially important due to deployment of increasingly smaller
munitions from internal bays. While a substantial amount of literature exits on the instances of successful control
(see the recent review in Ref. [1]), the amount of literature on the mechanisms involved has been relatively small.
This body is even smaller for supersonic regimes due to the limited number of facilities available. Furthermore,
detailed measurements of the flow fields have only recently become possible to aid such analysis. We aim to
supplement the body of work in this area by creating a very detailed numerical database of flow fields that can be
studied to understand these mechanisms. This is done through the use of very high resolution Large Eddy
Simulations of the flow fields. The accuracy of these calculations is verified by the comparisons with experimental
measurements.

1
Senior Research Scientist, 6210 Keller Church Rd., Pipersville, PA 18947, and AIAA Member.
2
Research Scientist, 6210 Keller Church Rd., Pipersville, PA 18947, and AIAA Member.
3
Vice-President and Technical Director, 6210 Kellers Church Road, AIAA Fellow.

American Institute of Aeronautics and Astronautics
1
AIAA-2006-2427
For the flow control concepts, we focus on mass blowing techniques. The potential to adaptively use mass
blowing to provide control over a wide variety of flow conditions makes these concepts attractive candidates. Our
focus here is on supersonic flows and steady mass blowing techniques, employing small mass flow rates. Several
attempts at using mass flow rate concepts for cavity flow control have been attempted. In many of these cases, even
at laboratory scales, mass flow rates of typically a few percent (1-15%) of the free stream mass flow rate have been
required. (see Ref.[1]). Furthermore, in most cases, these experiments have demonstrated control at a single flow
condition and their effectiveness at varying/different flow conditions is not well understood. One example of
successful control using reasonable mass flow rates at laboratory scales was the work of Ukeiley et al. [2]. A more
recent example of successful control for cavity flows and other flows such as impinging jets is the work of Alvi and
co-workers [3] using microjets. The mass flow rates used in that test were very low and it was seen that the jets were
very effective in reducing the dynamic loads in the cavity. A reduction in the turbulence levels in the cavity along
with reduced re-circulation velocity levels were identified as the dominant mechanisms for control. In our program,
preliminary experimental evidence suggests that similar levels of success using segmented slots can also be
obtained. This simulation effort has been undertaken to identify the respective mechanisms in slot and microjet
blowing control in order to better understand the physics of the problem.
In the following sections, we present an outline of the configurations, concepts and analysis methods. Care has
been taken at every step to carefully validate the calculations against experiments. The primary discussion here is
restricted to microjets, analysis of the slotted jet configurations will be presented in a future report.
II. Flow Conditions and Cavity Configuration
As mentioned above, we focus on the supersonic regime. Two mach numbers, Mach 2.0 and Mach 1.5 are being
studied. These Mach numbers represent conditions well within the flight Mach number regimes of existing military
aircraft, but fall beyond the store drop envelopes for these aircrafts. By focusing on this regime, we hope to
contribute to the expansion of these envelopes.
Two types of cavities are being studied in this work. A rectangular cavity of L/D=5.1 and a non-rectangular
cavity of L/D=5.6 (based on aft depth) are being studied. The non-rectangular cavity consists of spanwise and depth
variations to mimic similar variations likely to exist in practical configurations. Schematics of the cavities are shown
in Figure 1. The rectangular cavity study has focused on shear layer mechanisms and is discussed in detail here. The
analysis of the non-rectangular cavity, representative of practical configurations, has only been initiated and only
preliminary results are presented here.


(a). Rectangular Cavity L/D=5.1

(b). Non-Rectangular Cavity, L/D=5.6
Figure 1. Schematics of the cavity geometries being studied in the present work.

The microjets were tested originally under Mach 2.0 conditions at FSU [3], these conditions are used here. In
these tests, an array of 12 microjets were used, placed 1.5mm upstream of the leading edge of the cavity shown in
Figure 1 (a). The microjets were 400m in diameter and were operated under choked conditions. In the original
experimental work, several microjet pressures were used and it was observed that the effect saturated around 30 psig
total pressures (23.6psia static pressure). Hence, in our work we chose to analyze two operating pressures for the
microjets, 23psia and 30psia respectively, straddling the saturation point. The microjet configuration is
schematically shown in Figure 2 (a). The calculations uses a simplified model setup of this configuration and is
discussed in detail in Section III.
The slot jets were tested at the NCPA facility at Mach 1.5 conditions and hence these conditions are mimicked
here for the slot jet study. Several slot jets were tested in that study, varying from single slot spanning the cavity

American Institute of Aeronautics and Astronautics
2
AIAA-2006-2427
width to an array of seven segmented slots spanning that width. The conditions chosen to be modeled here mimick
that of the most effective configuration, three slots spanning the cavity width. The slot dimensions are shown
schematically in Figure 2(b). A detailed discussion of the slot configuration and the simulation setup is presented
next.


(a) Schematic of the microjets

(b) Schematic of the slot jet configuration.
Figure 2. Schematic of the control configurations studied in this paper.
III. Simulation Setup and Methodology
The calculations are performed using the CRAFT CFD

flow solver. The LES methods in the solver are well


validated for cavity flow applications and have been previously discussed in several papers [4]-[8]. The details are
omitted here for brevity. The simulation is setup as follows.
The characteristics of the cavity oscillations depend strongly on the upstream boundary layer. In order to model
this boundary layer correctly, the upstream inflow profiles to the LES calculation are extracted from a steady state
RANS calculation of the upstream nozzle and the test section without the cavity. These profiles are then imposed a
distance of approximately 1 cavity depth upstream of the leading edge of the cavity. In this manner, the costs
associated with modeling the upstream boundary layer are minimized and a full LES simulation can then be used in
the cavity region itself. This method has been successfully used in the simulations of backward facing step flows and
has shown to be a valid approach [9].
The baseline calculations to validate this approach used the entire cavity configuration shown in Figure 1(a).
Subsequent to this, preliminary calculations were carried out to identify the grid requirements to model the mass
blowing concepts. From these calculations, it was determined that the number of grid point required to fully resolve
the same configuration would exceed 9 million grid points. Hence, another simplification has been used to keep the
costs manageable in the present calculations. The spanwise extent of the cavities has been reduced and periodic
boundary conditions are imposed on the boundaries. The extent of the domain is about half the cavity width and
about 10 times the upstream boundary layer thickness. In the vertical direction, the domain extends to about half the
height of the tunnel test section in the experiments. The grid used in the present calculations is block structured and
contains about 3 million grid points.
IV. Baseline Rectangular Cavity: Preliminary Results
The rectangular cavity configuration was also studied experimentally by Alvi and Co-Workers [3]. To validate
the simulation methodology, the present calculations were validated against the measurements presented in that
paper. This corresponds to the Mach 2.0 flow condition. As discussed earlier, this simulation used the entire domain
shown in Figure 1(a).
A comparison of the wall pressures (both mean and rms) with the experimentally measured profiles along the
centerline on the floor of the cavity is shown in Figure 3. The agreement with the experimental measurements is
fairly good. The computed profiles show a lightly larger dip in the aft section of the cavity compared to the
measured profiles. The rms pressures are also seen to slightly deviate from the measurements at this location. The
reason for this discrepancy is not yet clear. Further investigations are underway to determine the reasons for this.

American Institute of Aeronautics and Astronautics
3
AIAA-2006-2427

(a). Mean Floor Pressure

(b). RMS Wall Pressure
Figure 3. Comparison of mean and rms weal pressures with measured profiles.

In Ref. [3] measurements were also carried out of the velocity field statistics using PIV technique on the
symmetry plane of the cavity. The comparison of the mean and rms streamwise velocities on this plane with the
statistics computed from the present calculations are shown in Figure 4. The overall agreement between the
computed and measured fields is good. From the mean velocity contours the size of the recirculation bubble in the
cavity is seen to be replicated fairly well in the calculations. The extent of spreading of the shear layer and the
intensities of the velocity fluctuations in the shear layer are also seen to be well captured. However, small
differences are seen in the aft region close to the floor of the cavity. This could be related to the differences in the
wall pressures. Further analysis is underway to better understand this.

(a) Experimental Mean U Contours (b) Computed Mean U Contours

(c) Experimental RMS U Contours (c) Computed RMS U Contours
Figure 4. Comparison of the mean and rms streamwise velocity contours with measured values.

Comparison of the profiles of mean velocities and turbulence intensities obtained from the computations with
those measured are shown in Figure 5. Comparisons are shown at two locations, x/L=0.3 and x/L=0.8. At the
upstream locations, all the profiles compare reasonably well with the measured profiles. At the downstream end, the
streamwise turbulence intensities are slightly under predicted. However, the overall agreement is fairly reasonable.

American Institute of Aeronautics and Astronautics
4
AIAA-2006-2427

(a). Mean streamwise velocity

(b) RMS streamwise velocity

(c) Mean vertical velocity

(d) RMS vertical velocity.
Figure 5. Comparison of profiles of the mean and rms velocities with measured values.
V. Baseline Non-Rectangular Cavity
The calculations of the non-rectangular cavity have been
carried out at Mach number of 2.0. The initial results of the
Mach 2.0 case were compared with wall pressures measured on
the cavity by Alvi and co-workers [10] and this comparison is
shown in Figure 6. Only two pressure taps were used in the
measurements and it is seen that the computed results agree
very well with the data. The main purpose of these calculations
was to validate the computational framework for the complex
geometry case. From this figure it is clear that the methods used
here are able to capture the dynamic loads, and hence, by
inference the flow characteristics.
Calculations of the baseline non-rectangular cavity shown
in Figure 1(b) were also carried out at a Mach number of 1.5.
Detailed surface pressure measurements on these geometries
with and without control have been carried out, however, no PIV data is yet available for these cases. A simulation
with a solid spoiler at the leading edge was also carried out for this geometry since this represents a classical
baseline control concept for such configurations. The spoiler height is equal to the boundary layer thickness at the
leading edge. A snapshot of the instantaneous Mach number field from the baseline case and the spoiler case is
shown in Figure 7. The presence of the spoiler is seen to shift the shear layer location farther out from the lip-line of
the cavity. Also, differences in the recirculation bubble are seen between the two cases. The upstream region inside
the cavity is seen to be quieter for the baseline case, whereas, for the spoiler case, more activity is visible in this
region.

Figure 6. RMS wall pressures compared
to measured valued for the non-
rectangular cavity.

American Institute of Aeronautics and Astronautics
5
AIAA-2006-2427

(a) Mach number contours for baseline non-
rectangular cavity case.

(b) Mach number contours for the non-rectangular
cavity with spoiler case.
Figure 7. Instantaneous contours of Mach number for the baseline non-rectangular cavity case and the case
with the spoiler at the leading edge.

A comparison of the mean and fluctuating streamwise velocity contours shown in Figure 8 manifests this clearly.
The mean velocities in the upstream portion of the cavity are significantly larger (larger negative) for the spoiler
case than for the baseline case. In the shear layer region, the mean velocity in the spoiler case is shifted outward into
the free stream and a lightly greater spread is also visible for this case, compared to the baseline. However, the larger
velocities in the cavity suggest that the rotation velocity of the re-circulation bubble is greater with the spoiler than
it is without the spoiler. The RMS of the streamwise velocities, shown in Figure 8 (c) and (d) also show that the
fluctuating intensities in this region are higher for the spoiler case than for the baseline case. This indicates that the
oscillatory pulsing of the bubble is stronger in the spoiler case than it is for the baseline case. This indicates that the
spoiler may not be a very good control concept for such cavities. It has been seen before [2] that in the case of
cavities where control has resulted in a suppression of the dynamic loads, the velocities in the recirculation bubble
are usually lower.

(a) Mean U Baseline
(b) Mean U Spoiler
(c) RMS U Baseline (d) RMS U - Spoiler
Figure 8. Comparison of the mean and fluctuating streamwise velocities for the non-rectangular baseline and
spoiler controlled cases.

American Institute of Aeronautics and Astronautics
6
AIAA-2006-2427
A comparison of the dynamic pressure loads on the walls of the cavity with and without the spoiler are shown in
Figure 9. It is clear that despite the larger re-circulation velocities in the upstream portion of the cavity, the overall
dynamics loads on the cavity walls are reduced. The region of high loading on the aft wall is significantly smaller
for the spoiler case than it is for the baseline case. Corresponding to the region of increased fluctuating velocities in
the upstream portion of the cavity, there is a slight increase in the dynamic loading, however, this increase is more
than offset by the decrease in the aft end. Thus it is clear that care must be exercised in drawing conclusions from
symmetry plane data, as is usually done in experiments.

(a) Baseline Case
(b) Spoiler Case
Figure 9. Comparison of the dynamic loads on the cavity walls for the baseline non-rectangular and spoiler
controlled cases.
VI. Controlled Rectangular Cavity
As mentioned above, two control concepts, microjet blowing and slotted jet blowing are being studied in this
work. Two microjet calculations, one with the microjets operating at a pressure of 20 psi and one with the microjets
operating at a pressure of 30 psi have been carried out. The microjets are 400 microns in diameter and are located
about 1.5mm upstream of the cavity leading edge. Both these calculations are at Mach 2.0 conditions.
The effect of the microjets on the wall pressures is
shown in Figure 10 where the rms wall pressures on the
floor of the cavity are plotted along with that for the
baseline case. It is seen that the 20psi jets reduce the
wall pressure loads whereas the benefits for the 30psi
cases are not as clear cut. It may be recalled that the
conditions for the microjet calculations were chosen to
straddle the experimentally observed saturation point
and this figure shows that the calculations are able to
capture this effect.
The effect of the microjets on the near field is
shown in Figure 11 and Figure 12, where the velocity
iso-surfaces corresponding to 0.8U

are shown for the
baseline and microjet simulations. The microjets induce
strong counter-rotating pairs of vortices that modify the
shear layer in the near field. The spanwise coherence
seen in the baseline case is completely destroyed in this case. Also, a strong redistribution of the vorticity is seen. In
the baseline case, the vorticity is primarily oriented in the spanwise direction, whereas, in the microjet cases, the
vorticity is oriented in the streamwise direction. Other observations of this type of re-orientation of vorticity have
shown that this has a strong stabilizing effect on the shear layer [11].

Figure 10. Comparison of the wall pressures for
the microjet controlled cavity cases.

American Institute of Aeronautics and Astronautics
7
AIAA-2006-2427

Figure 11. Isosurfaces of velocityfor the baseline case
showing the strong spanwise coherence in the shear
layer structures.

Figure 12. Isosurfaces of velocity for the microjet case
1 showing the presence of counter-rotating vortex pairs
in the near field generated by the microjets.

A comparison of the streamwise velocity profiles for the three cases is shown in Figure 13. Here, profiles at x/L=
0.02, 0.5, and 0.75 are shown for the baseline and the two microjet cases. The pressures described in the legends
correspond to the exit plane static pressures of the jets. In the near field, the presence of the microjets significantly
alters the profiles. The baseline case profile exhibits the classical shear layer type profile across the cavity lip. In the
case of the microjet calculations, a small peak corresponding to the microjets is seen, followed by a dip in the
profile, before it recovers to the freestream value. This modified mean velocity profile, results in a very different
near field stability characteristic of the separating flow over the cavity leading edge. Further downstream, at the
midway point, we see that the shear layer region profiles have recovered to roughly equivalent shapes for all the
three cases. However, for the baseline case, inside the cavity, the negative velocities are greater in magnitude
compared to the microjet cases. This observation also holds for the downstream location profiles. This indicates that
the recirculation velocity magnitudes in the cavity are smaller when the microjets are turned on.


(a) x/L=0.02

(b) x/L=0.5

(c)x/L=0.75
Figure 13. Mean streamwise velocity profiles for the baseline and the microjet cases.

The transverse velocity fluctuation intensity contours for the three cases are shown in Figure 14. There is a
substantial reduction in the magnitude of the intensities for the microjet cases when compared to the baseline case.
The reduced strength of the re-circulation bubble results in a reduced pumping motion seen in the cavity. This
results in a reduction of the velocity fluctuations in the aft portion of the cavity. This has also been observed to be
the case in the experimental measurements [3]. In the case of the 30psi microjet cases, we see that the reduction in
the rms values is not quite as significant as the 20 psi cases, this could explain the differences in the floor dynamic
load profiles seen in Figure 9. In the aft region, the reduction in the dynamic loads for the 20 psi case is greater than
for the 30 psi jet cases.


(a) baseline

(b) 20 psi microjets

(c) 30 psi microjets
Figure 14. Contours of fluctuation intensties of the y component of velocity for the three cases.

American Institute of Aeronautics and Astronautics
8
AIAA-2006-2427
The turbulent kinetic energy and Reynolds shear stress profiles at the same location as Figure 13 are presented in
Figure 15. Here, we see strong differences in the near field turbulence with and without the microjets. The baseline
case, at all the three axial stations shown, has a very classical shear layer type behavior. In the presence of the
microjets, there is a strong peak in both the Reynolds stress and the turbulent kinetic energy. The increased shear in
these cases, due to the injection of the jets contributes to an elevation in the Reynolds stress levels. This in turn
affects the turbulence production in this region and manifests itself in elevated levels of turbulent kinetic energy. By
the half way point again, we see that, analogous to profiles in Figure 13, the profiles of the mean kinetic energy and
the Reynolds stress are similar for the three cases. Further downstream, the differences seen in the fluctuation
intensities is also seen in the TKE profiles, with elevated levels for the baseline case and lower levels with the
microjets turned on. Thus, in terms of the energy budget, it appears that the presence of the microjets increases the
near field stresses and turbulence levels at the expense of the levels at the downstream locations. The increased
intensities in the near field do extend all the way downstream. In fact, the levels downstream are reduced, despite
this increase in the near field.


(a) x/L=0.02

(b) x/L=0.5

(c)x/L=0.75
Figure 15. Profiles of the turbulent kinetic energy and Reynolds stress for the three cases at three axial
stations. THe positive curves correspond to the TKE and the negative curves correspond to the Reynolds
stresses.

In order to better understand this, plots of the enstrophy (defined as
ij ij
= ) and the Q variable (defined as
) are shown. The enstrophy is the square of the vorticity, while the Q variable has the pure
strain rate removed it and hence signifies pure rotation. Thus, the Q values are representative of vortex regions, with
higher values signifying stronger vortices, while the enstrophy magnitude is representative of higher vorticity. Iso-
surfaces of the Q variable colored by the enstrophy is shown in
ij ij ij ij
Q S = S
Figure 16 for the baseline and 20spi microjet cases.
The size of the structures is seen to be significantly smaller for the microjet case than it is for the baseline case.
Further, the magnitude of the enstrophy variable, signifying vorticity, is greater for this case. This is because, due to
conservation of angular momentum, as the size of the structures reduces, the intensity of rotation increases. Thus, in
the microjet cases, the shear layer is broken-up in small scale structures, with very high vorticity in
concentrations. In contrast the shear layer in the baseline case is seen to contain much larger scale structures and
lower vorticity magnitudes. This implies that the increased turbulence in the near field is also accompanied by
increased dissipation in the shear layer due to a reduction in the structure sizes. Thus the increased kinetic energy
levels are dissipated out in the shear layer before it impinges on the aft end. This altered dynamic in the shear layer
coupled with the highly reduced coherence levels in the shear layer results in reductions in dynamic loads on the
cavity walls.

(a) Baseline Case

(b) 20 psi microjet case.
Figure 16. Isosurfaces of Q colored by enstrophy for the baseline and one of the microjet cases.

American Institute of Aeronautics and Astronautics
9
AIAA-2006-2427
VII. SUMMARY
Large Eddy Simulations of supersonic flow over rectangular and non-rectangular cavities has been presented.
The calculations have been validated against experimental measurements of surface pressures and PIV
measurements mean and fluctuating velocities. Results from calculations of un-controlled, baseline configurations
have been compared with the results from controlled configurations.
The effect of a 1-delta spoiler on the non-rectangular cavity shows that the shear layer over the cavity is
deflected away from the lip line of the cavity. This results in a modification of the shear layer interaction with the aft
corner of the cavity, resulting lowered dynamic loads on the cavity walls. Surprisingly, the mean re-circulation
velocities and fluctuations are seen to be slightly higher for the spoiler controlled case.
Analysis of the microjet blowing concepts shows that the shear layer, both in the near field and the far field are
strongly modified. The near field turbulence levels are seen to increase substantially for the microjet cases,
compared to the baseline case. However, this is accompanied by a reduction in the spanwise coherence of the shear
layer. The flapping motion of the shear layer, very clearly evident in the baseline cases is not seen in the microjet
cases. The reduced spanwise coherence is also accompanied by a reduction in the length scales of the structures in
the shear layer. Hence, the increase in the Reynolds stresses and turbulence levels is offset by the increase in the
dissipation levels. Thus, the energy levels of the impinging fluid on the aft wall are reduced, thereby reducing the
magnitude of the dynamic loads on the cavity walls.
Experimental evidence of similar effectiveness levels with slot blowing have also been observed. Our future
efforts are directed towards understanding these flow fields also and will be presented in the near future.
VIII. Acknowledgements
The authors would like to gratefully acknowledge the contributions of collaborators Dr. J ack Seiner, Dr.
Lawrence Ukeiley and Dr. Farrukh Alvi during this work. Our thanks to Dr. Alvi for providing us with the
experimental results.
IX. Reference
[1] Cattafesta, L., Williams, D., Rowley, C., and Alvi, F., Review of Active Control of Flow Induced Cavity
Resonance, AIAA-2003-3567.
[2] Ukeiley, L.S, Ponton, M.K., Seiner, J.M. and J ansen, B., Suppression of Pressure Loads in Cavity Flows,
AIAA 2002-0661, 2002.
[3] Zhuang, N., Alvi, F.S., Alkislar, M.B., Shih, C., Sahoo, D., Annaswamy, A.M., Aeroacoustic Properties of
Supersonic Cavity Flows and Their Control, AIAA-2003-3101.
[4] Arunajatesan, S. and Sinha, N., Hybrid RANS-LES Modeling for Cavity Aeroacoustics Predictions,
International Journal of Aeroacoustics, Vol. 2, No. 1, pp 65-91, 2003.
[5] Sinha, N., Arunajatesan, N. and Ukeiley, L., High Fidelity Simulations of Weapons Bay Aeroacoustics and
Active Flow Control, AIAA Paper 2000-1968, 2000.
[6] Arunajatesan, S., Shipman, J .D. and Sinha, N. Hybrid RANS-LES Simulation Of Cavity Flow Fields With
Control, AIAA 2002-1130, 2002.
[7] Arunajatesan, S., Shipman, J.D. and Sinha, N. Mechanisms in High Frequency Control of Cavity Flows,
AIAA 2003-0005, 2003.
[8] Arunajatesan, S., and Sinha, N., Large Eddy Simulations of Impinging J et Flow Fields, AIAA-2002-4287.
[9] Ayyalasomayajula, H., Arunajatesan, S., Kannepalli, C., and Sinha, N., Large Eddy Simulations of Supersonic
Flow over a Backward Facing Step For Aero-Optical Analysis, AIAA-2006-1416.
[10] Alvi, F., Personal Communications.
[11] Arunajatesan, S. and Sinha, N., Modeling Approach for Reducing Helmholtz Resonance In Submarine
Structures, AIAA-2005-2858.


American Institute of Aeronautics and Astronautics
10

You might also like