SIMULATION, COMPUTING, AND MODELING

The jet exhaust nozzle geometries used in this

Simulation of Supersonic Military
research are representative of realistic engine nozzles
Aircraft Jet Noise in use on military aircraft and were designed by GE
Aircraft Engines. These nozzles do not have smoothly
K. Kailasanath, J. Liu, and R. Ramamurti varying contours. Instead, they are convergent-diver-
Laboratory for Computational Physics gent nozzles: they typically have a conical converging
and Fluid Dynamics section, a sharp throat, and a conical diverging section,
a design which allows the area ratio (ratio of the area
Background: There is a growing need to reduce at the throat to the area at the exit of the nozzle) to
significantly the noise generated by high-performance, be changed in flight to adapt to local conditions and
supersonic military aircraft. The noise generated during thrust requirements.
takeoff and landing on aircraft carriers has direct
impact on shipboard health and safety issues. Also, Numerical Simulations of the Flow Field
noise complaints are increasing as communities move and Noise: Monotonically integrated large-eddy
closer to military bases or when there are changes due simulations of imperfectly expanded jet flows from
to base closures and realignment. There is a significant a convergent-divergent nozzle with a design Mach
amount of literature dealing with noise reduction number of 1.5 were carried out. Total pressure ratios
in civilian, subsonic aircraft; some of the techniques ranging from over-expanded to under-expanded jet
found effective for those aircraft might be applicable conditions were investigated. A typical computed flow
for supersonic military jets. A distinct difference field from the nozzle exhaust is shown in Fig. 4. The
between civilian aircraft engines and advanced military quantity depicted in the figure is the non-dimensional
aircraft engines is that military engines tend to have low density (non-dimensionalized using the background
bypass ratios and high velocities. During certain flight ambient density). Key features of the flow field are also
conditions — such as during takeoff or landing — the identified. Results showed that spacing of the shock
exhaust from these engines tends to be non-ideally cells and the length of the potential core (an indica-
(under/over) expanded. Non-ideally expanded exhaust tor of jet mixing) increased as the total pressure ratio
flows contain shock cells in the jet exhaust, causing increased, and these results were in good agreement
high-amplitude screech tones and broadband shock- with experimental data from UC. The good agreement
associated noise, components that increase the overall suggests that the computations are resolving the details
noise level. Therefore, our current research is focused observed in the laboratory experiments.
on understanding these non-ideally expanded exhaust In principle, no shocks or other pressure waves
flow conditions and characterizing the noise sources, are expected at the design condition. Unexpectedly,
so that noise reduction techniques may be successfully weak shock cells were observed at the design condition.
pursued. Experiments confirmed this observation. Results from
the simulations have identified the cause of these waves
Approach: A monotonically integrated large-eddy to be the sharp contraction at the nozzle throat. Hence,
simulation (MILES)1 approach has been developed at these studies suggest that the flow fields from realistic
NRL to simulate computationally the flow dynamics military engine nozzles are not likely to be shock-free
and near-field acoustics of supersonic jet exhaust flows. under any operating condition.
A finite-element flow solver2 using unstructured grids
allows us to model the jet exhaust nozzle geometry Comparison of the Noise Spectra: Near-field
accurately, and the MILES approach directly computes sound pressure level (SPL) spectra at various locations
the large-scale turbulent flow structures. No explicit in the exhaust flow were calculated and compared with
subgrid scale model is used and the modeling of the the experimental measurements at UC. A screech tone
subgrid scales is implicitly provided by the embedded was observed in both experiments and simulations,
flux limiter in this approach. and both the intensity and frequency were in good
Numerical simulations can play a significant role in agreement between the numerical predictions and the
the test and evaluation of various noise reduction con- experimental measurements (see Fig. 5). The good
cepts, but for the results of the simulations to be cred- agreement highlights the complementary nature of the
ible, they need to be compared to and evaluated against two approaches. Since it is difficult to measure sound
relevant experimental data. The experimental condi- pressure level inside the jet experimentally without
tions should include geometries and flow conditions modifying the jet, measured data from inside the jet
that are representative of realistic engine configurations is not available. In the simulation, the fine grid (used
and operating conditions. The simulations in this study for resolving the details of the flow) is restricted to a
were compared to experimental data obtained at the region close to the nozzle exit due to the limitations of
University of Cincinnati (UC).3 current computer resources (it would be prohibitively

2009 NRL REVIEW 239

 SIMULATION, COMPUTING, AND MODELING

2 R. Lohner, “FEM-FCT: Combining Unstructured Grids with

expensive to grid the entire flow flow field using the
fine grid). Therefore, the direct overlap between the High Resolution,” Communications in Applied Numerical
Methods 4, 717–729 (1988).
computed region and the region where detailed mea- 3
J. Liu, K. Kailasanath, D. Munday, and E. Gutmark,
surements were taken is small. However, the transition “Investigation of Near-Field Acoustic Properties of Imperfectly
between the results from the simulations and those Expanded Jet Flows Using LES,” AIAA Paper 2009-0015,
from the measurements is almost flawless (as seen in presented at the 47th AIAA Aerospace Sciences Meeting,
Fig. 5) and shows that the two approaches together Orlando, Florida, Jan. 5–8, 2009.
can produce a more complete picture of the noise field
than can be obtained by either approach alone. Further
analysis of the results has provided new insight into the
sources of jet noise and has been reported elsewhere.3

Concluding Remarks: Our simulations were able

to accurately describe the flow field and noise from
supersonic military aircraft jets. The excellent agree-
ment shown between the results of our simulations and
the measurements made at the University of Cincinnati
indicates that such experiments and simulations can
play complementary roles in the investigation of noise
generation from supersonic jet flows. This was the first
step in a multi-year effort. After this successful initial
step to validate the computational methodology and
characterize the noise sources, work on specific noise
reduction techniques has begun.

Acknowledgments: This research was sponsored

by the Strategic Environmental Research and Develop-
ment Program (SERDP) and the NRL 6.1 Computa-
tional Physics Task Area. The authors are grateful to
Prof. Ephraim Gutmark from the University of Cin-
cinnati for sharing the experimental data and general
insights into nozzle flow and acoustics. We also thank
Prof. Rainald Lohner from the George Mason Univer-
sity for help with the FEFLO computer code.
[Sponsored by SERDP and NRL]

References
1
F.F. Grinstein and C. Fureby, “On Monotonically Integrated
Large Eddy Simulation of Turbulent Flows Based on FCT
Algorithms,” Ch. 3 in Flux-Corrected Transport: Principles,
Algorithms, and Applications, eds. D. Kuzmin, R. Löhner, and S.
Turek (Springer, 2005), pp. 79–104.

4 3.8
3.5
2 3.2
2.9
2.6
Y/D

0 2.3
2.0
-2 Shock Cells 1.7
1.4
1.1
-4
0.8
0 5 10 15 20 25
X/D
FIGURE 4
Normalized density distribution showing the key features of the computed flow field from the exhaust nozzle of
a supersonic jet.

240 2009 NRL REVIEW

 SIMULATION, COMPUTING, AND MODELING

7
6
170
5 150
4 130

Y/D
110
3 90
70
2 50
1
0
0 2 4 6 8 10 12 14 16 18 20 22
X/D
FIGURE 5
Sound pressure levels from the exhaust nozzle of a supersonic jet: comparison of simulations and
experimental data. The axial (X/D) and radial (Y/D) distances are non-dimensionalized by the jet
diameter (D) at the nozzle exit.

2009 NRL REVIEW 241