| |
By Thomas Scheidegger, Fluent Inc.
View the pdf of this article
Vortical structures visualized using isosurfaces
of the second invariant of the
deformation tensor, colored by velocity
magnitude
Landing gear noise is not the first thing that comes to mind when
thinking about noise pollution at a busy airport. The once dominant
jet engine noise has been reduced significantly over the past thirty
years, primarily through the introduction of high bypass turbofan engines.
As a result, airframe noise has emerged as a leading component of aircraft
noise during the final approach phase of a landing. Environmental concerns
and noise certification regulations are therefore causing aircraft manufacturers
to take a closer look at this phenomenon.
Surface mesh with 173,000 tri-elements
Complex surface pressure patterns are observed on the rear
wheels and rear diagonal strut; red contours indicate
stagnation pressure
The main contributors to airframe noise in a landing configuration are
high-lift devices, such as slats and deployed flaps, and surprisingly, the
landing gear. Measurements have shown that these components are not
equally important on all aircraft. While the high-lift devices are noisier on
medium size aircraft, the landing gear is becoming the dominant source
on large airplanes, such as the Boeing 777.
Landing gear systems have complex, non-streamlined geometries, and
generate highly turbulent wakes. Vortices shed from one component
impinge on other elements, generating noise. While very little is known
about this interaction, the resulting noise generally has a broad spectrum,
from a few hundred Hz to several kHz. Engineering tools that can predict
the complex flow features and estimate the generated noise levels are
highly desirable. If noise can be predicted, modifications such as fairings
and streamlining can be introduced during the design phase. Ideally, the
acoustics analysis tool should provide the expected far-field sound pressure
levels, directivities and spectra, and insight into the source mechanisms
and source distributions.
Predicting aeroacoustic noise is not a trivial matter. Only a minute fraction
of the kinetic energy present in the primary flow is converted into
acoustic energy and radiated. To correctly capture the acoustics, the turbulent
flow must be calculated with high fidelity. Since turbulence is an
inherently unsteady phenomenon, a time-consuming transient simulation
is required. Integral techniques that predict the far-field acoustic signal
using source data input from a near-field CFD simulation have emerged as
a promising and economic way to compute sound levels. The most universal
and complete integral method available today is based on the Ffowcs-
Williams and Hawkings (FW-H) equations.[1]
At Fluent, engineers have recently analyzed a 1/10th scale landing gear
model, representative of the gear used on a Boeing 757 aircraft. The same
configuration has been studied using CFD by researchers at Penn State [2]
and NASA Langley [3, 4], and will also be tested in a wind tunnel at the
Quiet Flow Facility at the Langley Research Center. The tests will be conducted
at a Reynolds number of 1.23 x 106, based on the wheel diameter,
and at a free stream Mach number of 0.2. The four-wheel landing gear
assembly contains all of the major components, including the oleo-strut,
axles, connecting blocks, diagonal struts, a door, and additional parts that
hold the configuration together. A flat plate simulates the wing surface.
For noise predictions in a sufficiently broad frequency range, large eddy
simulation (LES) and detached eddy simulation (DES) simulations have
shown more promise than unsteady RANS calculations. LES and DES have
a wider range of resolved unsteady scales, have higher levels of resolved
turbulent kinetic energy, and therefore predict higher sound pressure levels.
Only the resolved unsteady eddies can generate sound with the FW-H
methodology. Thus, LES (with the Smagorinsky subgrid scale model) was
used for the landing gear calculation.
The CAD (STEP) model provided by NASA Langley was cleaned up in
GAMBIT, and a computational grid was built using GAMBIT and TGrid. A
combination of GAMBIT’s fixed and curvature-based size functions was
used to generate a high quality surface mesh. Boundary layer prisms were
grown in TGrid, so that the prism cap surface mesh could be used to control
(using the mesh-based size function) the growth and continuity of the
tetrahedral elements away from the boundary layer. Additional size functions
were used to cluster elements in the vicinity and wake of the landing
gear, resulting in a 5.3 million cell mesh, suitable for an LES simulation.
The landing gear case was run incompressibly (a valid approximation
for compact sound sources) for nearly 10,000 time steps before the turbulence
statistics were sufficiently stabilized and the acoustic source data
sampling could be started. The total time for this portion of the calculation
corresponds to one flow-pass through the domain, or a distance of 18
wheel diameters. The acoustic source data was extracted directly on the
landing gear surface over approximately one additional flow-pass, and
then processed with the FW-H solver.
Dipole source strength, using contours of dp/dtRMS, shows a
high source intensity at the rear diagonal strut and behind
the oleo-strut
The FW-H tool is ideal for predicting far-field radiation in the absence of
external scattering surfaces. The necessary source data can be extracted
from permeable (interior) or solid (wall) surfaces, and the method is not
very sensitive to the actual source surface placement. The direct output
includes the far-field pressure signals at user-specified receiver locations.
Postprocessing tools are available to perform spectral analyses of these signals,
including overall sound pressure level (OASPL) outputs. Also available
is the local dipole source strength, which can be used to assess contributions
from different source locations.
Surprisingly, but in good agreement with other studies performed on
the same configuration [3, 4], flow visualization revealed that the two diagonal
struts shed nearly as much vorticity as the big wheels. The behavior is
much more complex than blunt body vortex shedding, and a very short
distance downstream of the landing gear, it is difficult to differentiate the
flow structures originating from different components. Persistent flow separation
due to an asymmetric flow was observed at the gear door leading
edge. Animations of unsteady surface pressure showed more complex patterns
on the rear wheels and rear strut, as expected.
The acoustic analysis indicated that the overall sound pressure levels at
a distance of 10 wheel diameters upstream and downstream of the landing
gear, compared to the two lateral directions, are lower by about 4 dB.
Differences were also noticed in the sound pressure spectra. The lateral
directions peak at around 700 Hz, and the same frequency was observed
to be dominant in the crossflow force response. The streamwise spectra
peak at considerably higher frequencies. A total of 18 surface pressure
probes were strategically placed in the rear of the wheels, struts, and along
the wheel door. The recorded pressure traces confirmed that the rear diagonal
strut is one of the dominant noise sources. Fluent engineers are anxiously
awaiting the experimental data expected from the wind tunnel to
confirm these findings.
Overall sound pressure levels
(OASPL) for five receivers located
1 m from the landing gear
Sound pressure level spectra (dB) for four of the receivers shown in the figure above
References:
- J.E. Ffowcs-Williams and D.L. Hawkings, Proceedings of the Royal Society of London
A264, p. 321-342 (1969).
- F.J. Souliez, L.N. Long, P.J. Morris and A. Sharma, International Journal of Aeroacoustics
1, No. 2, p. 115-135, 2002.
- F. Li, M.R. Khorrami and M.R. Malik, AIAA Paper 2002-2411, 8th AIAA/CEAS
Aeroacoustics Conference, Beckenridge, CO, June 17-19, 2002.
- D.P. Lockard, M.R. Khorrami and F. Li, AIAA Paper 2004-2887, 10th AIAA/CEAS
Aeroacoustics Conference, Manchester, UK, May 10-13, 2004.
|
|
|
