2006 01 1192
2006 01 1192
2006 01 1192
PAPER SERIES
2006-01-1192
The Engineering Meetings Board has approved this paper for publication. It has successfully completed
SAE's peer review process under the supervision of the session organizer. This process requires a
minimum of three (3) reviews by industry experts.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.
For permission and licensing requests contact:
SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Tel:
724-772-4028
Fax:
724-776-3036
2006-01-1192
ABSTRACT
Computational Fluid Dynamics (CFD) is an integral part
of product development at Visteon Climate Systems with
a validated set of CFD tools for airflow and thermal
management processes. As we increasingly build CAE
capabilities to design not only thermal comfort, but quiet
systems, developing noise prediction capabilities
becomes a high priority.
Two Broadband Noise Source (BNS) models will be
presented, namely Proudmans model for quadrupole
source and Curles boundary layer model for dipole
source. Both models are derived from Lighthills acoustic
analogy which is based on the Navier-Stokes equations.
BNS models provide aeroacoustic tools that are effective
in screening air handling systems with higher noise
levels and identifying components or surfaces that
generate most of the noise, hence providing
opportunities for early design changes.
In this paper, BNS models were used as aeroacoustic
design tools to redesign an automotive HVAC center
duct with high levels of NVH. The design direction
suggested by BNS tools were later supported by physical
test data. These models were found to be valuable and
cost effective tools in providing reliable design direction
early in the development cycle.
INTRODUCTION
The ability of an analysis tool to characterize a systems
level of noise and its spectral distribution at a receiver
location is a powerful design tool. It could be used as a
virtual test lab to measure whether a given system meets
its design requirements. Currently, such capability is not
available within a development cycle for complex
industrial applications. This is due to the fact that noise
information at receiver locations requires the solution of
both the flow field and the acoustic field. However,
during early stages of the product development, noise
information at receiver locations is not critical. Rather,
reliable system noise information at the source may be
sufficient to provide design direction during development.
AEROACOUSTIC MODELING
Aeroacoustic behavior can be completely characterized
by solving the compressible Navier-Stokes equations. In
other words, aeroacoustic phenomena can be explained
through the use of the principles of mass, momentum,
and energy conservation. For simple problems, Direct
Numerical Simulation (DNS) that solves the
compressible Navier-Stokes equations is able to solve
for both the aerodynamic flow field and the acoustic field.
However, for problems of industrial applications, this
approach becomes impossible due to the fact that the
acoustic energy is several levels of magnitude smaller
than the hydrodynamic energy. In addition, the acoustic
pressure perturbations are several levels of magnitude
smaller than the hydrodynamic pressure. Further, length
and time scales of the two fields are not compatible.
Therefore, the need to separate the two flow fields
becomes apparent.
Tij
a02 2 =
2
xi x j
t
2
(1)
With
(1.2)
Where
tensor,
ij
a0 is the speed of
+
t xi xi x j
t 2
Where
(2)
Q
is due to unsteady volume displacement of
t
Fi
is due to interaction of the flow with the rigid body
xi
and represents the contribution of the dipole noise.
2Tij
xi x j
COUPLED APPROACHES
A coupled approach utilizes the Lighthills analogy and
FWH equation implemented in its most general form.
Here, a CFD solver is used to characterize the
fluctuations in the unsteady flow field, and an acoustic
code is utilized to solve for the propagation equation.
Two separate mesh domains are used in this approach
with great potential in simulating industrial aeroacoustic
problems. Commercial acoustic solvers that can be
coupled with commercial CFD solvers are currently
available. Kim et al [7] and Ayar et al [8] used this
approach for automotive applications and found
reasonable correlation with experimental data.
This coupled approach can be simplified further if there
is a line of sight between the source and the receiver.
Under this assumption, the FWH equation is solved
analytically and no acoustic solver is required. This
approach is suited for external flows in industrial
applications with low to medium complexity.
Implementation of a coupled approach based on
Lighthills analogy has increased the role of aeroacoustic
modeling for industrial applications. However, it still
requires unsteady turbulence models that can capture a
detailed fluid flow structure and an acoustic solver for the
wave equation. These two requirements still present a
challenge in incorporating this approach into the
development process.
During early stages of product development, a measure
of a systems noise level at the source is sufficient to
drive the design optimization process. However, a
coupled aeroacoustic approach with detailed transient
CFD analysis and a noise propagation solver becomes a
burden to the early stages of a fast paced product
development cycle. Instead, approximate noise source
strength extracted from a steady state CFD solution
could provide a useful and practical aeroacoustic design
tool.
AP = 0 (
is a constant,
u2 =
2
k
3
as:
(4)
Where
Mt =
p ( x, t ) =
1
4a 0
( xi y i )ni p
( y, )dS ( y )
t
r2
(5)
p is the surface
pressure, n is the wall normal direction, p is the
acoustic pressure, and a0 the speed of sound.
Where,
p
Ac ( y) t dS ( y)
(6)
is the density.
p
t is the mean-square time derivative of the source
surface pressure, x is the receiver coordinate, y is the
coordinate of the source surface, and S is the source
surface.
AP = 0 M
(3)
Where
1
SAP =
12 0 a 03
u u
)
l a 05
2k
a0
Vent Registers
Cross section @ y = 60 mm
Airflow Outlet
Registers
Center Duct
Cross section @ y = 60 mm
Center Duct Mesh Close-up
Outboard
Airflow Inlet
Vent Registers
Mesh Close-up
RESULTS
Figure 5 shows that most of the dipole noise was
generated from the center duct with very little
contribution from the outboard ducts and registers. The
contour plot of the quadrupole noise generation in Figure
6 shows that the neck area of the center duct radiates
most of the noise. In addition, Figure 7 shows surface
acoustic power level iso-surfaces of radiated dipole noise
with significant generation at the neck.
3.0E-05
2.5E-05
SAP (Watts)
3.5E-05
2.0E-05
1.5E-05
1.0E-05
5.0E-06
0.0E+00
Center
Outboards
Registers
Component
Figure 5: Initial design component source noise
LHO
LHC
RHC
RHO
Total
CFD
36.85
34.79
36.67
34.61
142.91
Test
35.32
34.61
40.61
34.61
145.15
Change
-1.53
-0.18
+3.94
0.00
+2.24
Low Noise
High Noise
Noise Reduced
Center Duct
Cross-section @ y = 60 mm
3.5E-05
3.0E-05
Current
SAP (Watts)
2.5E-05
2.0E-05
1.5E-05
1.0E-05
Redesign
5.0E-06
0.0E+00
Center
Outboards
Registers
REFERENCES
Component
1.
Figure 11: Initial and Redesign component source noise
Initial
SAP
Source
(Watts)
4.34e-5
SAP Level
Source
(dB)
66.4
Redesign
17.32e-6
62.4
60 %
-4.0
Model
Change
Recirc.
(dB)
Fresh
(dB)
Initial
71.3
66.9
Redesign
66.4
62.5
Change
-4.9
-4.4
CONCLUSION
BNS models are found to be effective tools in screening
systems with higher noise levels and identifying