Computational Aeroacoustics - Stanford University


Feb 22, 2014 (3 years and 3 months ago)


•OASPL, Comparison of experiment (symbol) and LES (symbol/line)
j=1.95and HeatedM
j=0.97,Bodonyand Lele(2005)
•With same exit velocity, cold jet is louder in both LES & experiment

Dilatation(black/white) -Acoustic pressure fluctuation
•Vorticity(color) -Hydrodynamic pressure fluctuation
•Cold M
j=2.2Underexpandedjet, Bodony, Ryuand Lele(2005)
•Upstream propagating shock-associated noiseand downstream
propagating turbulent mixing noiseare observed
Future work
Aircraft noiselimits the future growth of aviation.
Increasingly stringent community noise regulations
are in effect.
Long term goals are to limit significant aircraft noise
to airport perimeter.
Environmentally friendly solutions are needed.
(Noise and emissions need to be reduced)
Computational Aeroacoustics
2, William Wolf
1, Mohammad Shoeybi2, ParvizMoin
and SanjivaK. Lele
1, 2
Stanford University, Departments of Aeronautics and Astronautics1
,Mechanical Engineering
Sponsored by Boeing, NASA, Fulbrightand CAPES
K. Viswanathan, Boeing
Aircraft noiseconsists of
Airframe noise
(wings, slats/flaps, landing gear)
Engine noise
(combustion, turbo-machinery)
Jet noise
Fan noise

Jet noise
is dominant at take-off.
is dominant at landing.
Modeling must capture unsteady processesthat
generate aerodynamic noise.
Flows of interest are complexand have broad range
of spatial and temporal scales.
Only a tiny fraction of flow energy radiates as sound.
Accurate far-field sound calculations are
computationally expensive.
Need to include the nozzle geometry in order to
capture high frequency sound sources.
Direct Computation of Near-fields
(Direct Numerical Simulation)
Navier-Stokes equationsare numerically solved
without any turbulence model
Whole range of spatial and temporal scales of the
turbulence must be resolved
With high order finite difference scheme and careful
boundary treatment, can compute noise directly
(Large Eddy Simulation)
Navier-Stokes equations with sub-grid scale model
are solved
Can predict instantaneous turbulent flow structures
and noise like DNS, but more efficient
Hybrid Method
Unsteady nonlinear flow (noise sources) captured
Acoustic analogy
(Lighthill, FfowcsWilliams-
Hawkings) used to predict far-field sound.
surface extrapolation for far-field sound
Boundary Element Method (BEM)
used to
predict far-field sound. Computationally expensive
for realistic configurations with several DOF.

Fast BEM
used to predict far-field sound in large-
scale problems. BEM is accelerated by a multi-level,
adaptive Fast MultipoleMethod (FMM).
High speed jet aeroacoustics

Understand the sources of jet noise using LES data
Increase the bandwidth of predictions:
Subgridscale noise models
Include nozzle geometry in RANS/LES simulation
High resolution in the near-nozzle region
Low speed airframe noise

Develop new capability for high-fidelity, physics-based
aerodynamics noise predictions for airframe noise
Improve understanding of airframe noise sources and
explore the strategies for their mitigation including
novel application of flow control.
•Axial velocity (main figure) and vorticitymagnitude (inset) are
shown. Mach number locally gets as high as 1.25.
j=0.9jet simulation with ARN converging nozzle with
•Acoustic scattering around a multi-element wing due to two monopole
sources for kc~5.0, Wolf and Lele(2008)
•Fast 2-D and 3-D scattering code with a multi-level adaptive FMM,
Wolf and Lele(2008)
•Trailing edge scattering due to a monopole source for kc~40.0 (left)
•CPU time for 3-D Fast BEM and Direct BEM multi-element wing
simulation (right). FMM-BEM is 9.7 times faster than Direct BEM
•Vorticity(color) and dilatation (gray) contours for M=0.3 flow over
NACA0012 airfoil at Re=10,000 at angle of attack of 5 deg.
•Noise due to vortex shedding observed.