DE VELO PMENT  Ac oustic s | N V H

Transient Simulation
of Flow Noise in Exhaust Mufflers

© Faurecia

Flow noise is an extremely annoying component of tailpipe FLOW NOISE IS A KEY CHALLENGE
noise. It is caused by turbulence and pressure fluctuations,
Due to stringent noise emission regula-
which can be resolved in space and time by transient computa­ tions and increased customer sensitive-
tional fluid dynamics or Detached Eddy Simulation. This direct ness toward disturbing noises, flow
noise has to be considered as one of the
simulation of the flow field simultaneously with the noise gene­ key issues in the development of exhaust
ration and sound propagation provides the necessary insight systems. Generated especially at high
mass flow rates, flow noise is caused by
and understanding for the optimization of muffler designs,
local production of turbulence and pres-
as Faurecia and Siemens demonstrate in the following. sure fluctuations in areas with perforated
walls or areas with flow separation, flow
70
 AUTHORS

Dr.-Ing. Rolf Kaiser Dr.-Ing. Christof Hinterberger Francisco Ezquerra Larodé, M. Sc.
is Senior Expert CFD at is Thermofluid Analyst at is Senior Engineer at Siemens
Faurecia Clean Mobility in Faurecia Clean Mobility in Digital Industries Software in
Augsburg (Germany). Augsburg (Germany). Nuremberg (Germany).

impingement or free stream jets with tra). The required relative accuracy is
strong shear forces. The flow noise can ± 2 dB for the assessment of different
further intensify if there are acoustic design variants, the absolute accuracy
system modes in resonance with the is ± 5 dB related to prototype measure-
pressure fluctuations. ments. Based on the simulation, it should
The so-called Improved Delayed be possible to localize flow noise sources
Detached Eddy Simulation (IDDES) and to visualize the phenomenon of noise
enables the resolution of turbulence and generation. The duration of the simulation
pressure fluctuations in space and time should not exceed five to ten days. Since
[1, 2]. This allows the direct simulation the com­putational effort for transient flow
of the flow field simultaneously with the noise simulations is huge, the main chal-
noise generation and sound propagation. lenge is the trade-off between accuracy
Results of transient flow noise simulation and simulation duration, which is mainly
with Simcenter Star-CCM+ from Siemens driven by the resolution in space and time.
Digital Industries Software are presented For the examination of flow noise gen-
and discussed. Simcenter Star-CCM+ eration and the propagation of the sound
is a multiphysics solution for the simula- waves, the local cell size of the simula-
tion of products and designs operating tion mesh is crucial. In the areas of flow
under real-world conditions, allowing noise generation the cell size has to be
to efficiently explore the entire design small enough to resolve the turbulent
space instead of focusing on single point structures and the related pressure fluc-
design scenarios. tuations. Therefore a local refinement
around critical features is necessary.
If the local refinement is insufficient,
OBJECTIVES AND REQUIREMENTS
a potential flow noise source will not
The aim of the transient flow noise simu- become evident in the simulation. Inside
lation is the prediction of Sound Pressure the muffler chambers and pipes the reso-
Levels (SPL) versus frequency (SPL spec- lution should be sufficient to resolve

FIGURE 1 SPL spectra (1/12 octave bands), comparison of measurement (blue line) and IDDES simulation (red line) (© Faurecia)

MTZ worldwide 11|2019    71
DE VELO PMENT  Ac oustic s | N V H

FIGURE 2 Snapshot of vorticity for muffler 1 (left) and corresponding pressure fluctuation (right) (© Faurecia)

acoustic modes of the system. Further- MUFFLER TEST CASES Measurements were conducted in an
more the wave propagation at low losses AND MEASUREMENTS anechoic chamber. A blower provided air
for frequencies up to 10 kHz is import- at ambient temperature and at different
ant. For that reason an appropriate mesh For the evaluation of the simulation mass flow rates. The microphones were
resolution is required. Also the environ- methodology three muffler types were positioned according to the standard,
mental area has to be meshed to ensure designed as test cases and built for meaning at a distance of 500 mm to the
free-field conditions in combination with measurements. The aim was the re­­ tailpipe outlet and at 45° inclination.
a low reflection rate at the outer bound- production of typical flow phenomena
ary. Based on these requirements the and the generation of significant levels
COMPARISON BETWEEN
total mesh size for a standard application of flow noise. Muffler 1 consists of one
SIMULATION AND MEASUREMENTS
can easily reach several millions of cells. chamber and a continuous pipe with a
The resolution in time shall be chosen perforated area. Muffler 2 is an expan- A comparison between measurement
to be appropriate for pressure waves up sion cham­­ber type and muffler 3 consists and simulation is shown in FIGURE 1
to 10 kHz. An interval of 200 ms is recom- of two chambers separated by a perfo- based on SPL spectra for all three muf-
mended for spectral analysis at a suffi- rated baffle. The outer diameter of the fler types. The SPLs, which are plotted
cient resolution in the frequency domain. wrapped mufflers is 200 mm at a length in 1/12 octave bands, refer to the micro-
The requirements of high spatial and tem- of 400 mm [3]. The diameter of the phone positions outside the muffler. The
poral resolution lead to an effort of around attached pipes is 55 mm. No muffler considered mass flow rate is 1000 kg/h
50,000 h of computing time. contains any wool. at ambient temperature. Overall, the

FIGURE 3 SPL spectrum (1/12 octave bands) for muffler 2 (left) and corresponding small-band SPL plot at 1700 Hz (right) (© Faurecia)

72
THE LOOK INTO THE FUTURE.

MTZ has successfully accompanied the development of engines and powertrains for the past
80 years. The internationally leading technical and scientific magazine for decision-makers in
powertrain development and production sees itself as an information platform on future power-
trains – conventional, electrified or purely electric. MTZ promotes the transfer of information
between manufacturers, suppliers, service providers and research and development centers all
over the world, in German and English, in print and digital editions.

GET YOUR FREE TRIAL NOW

© istockphoto.com/matdesoígn24

www.mtz-magazine.com
MTZ worldwide 11|2019    73
DE VELO PMENT  Ac oustic s | N V H

FIGURE 4 SPL spectrum (1/12 octave bands) for whistle case – measurement (blue line), IDDES simulation (red line) (left) – and
related snapshot of vorticity (right) (© Faurecia)

results match well for broad-band as well noise at 1700 Hz for muffler 2 and the For muffler 1 a snapshot of vorticity is
as small-band noise. A very dominant broad-band noise of muffler 3 were pre- shown in FIGURE 2, which clearly local-
whistle at 7940 Hz for muffler 1, a peak dicted accurately. izes the source of turbulence and hence

FIGURE 5 1/3 octave band SPL plots for whistle case (© Faurecia)

74
the source of noise in the per­forated which is the peak value of the SPL muffler. Overall the results match well
pipe area. By means of this energy spectrum at the microphone position. and especially the two dominant whistle
source, the acoustic chamber and The related SPL plot for a frequency peaks were predicted correctly by the
pipe mode are excited at 7940 Hz. The band of 1700 Hz ± 50 Hz clearly shows simulation. The related snapshot of vor-
related mode pattern can be seen in an acoustic resonance mode of the muf- ticity shows very well the root cause of
the snapshot of the local pressure fler chamber. Here the nodal areas as the whistling: In the joint area of the pipe
fluctuations, FIGURE 2 (right), where well as the SPL peak areas can be seen. and the bended tailpipe there is a sudden
the nodal areas and the pressure peak In addition the source of the noise can increase of pipe diameter, where vortex
areas can be identified. be identified in the shear layer of the shedding occurs. Downstream these vor-
For root cause analysis and design free jet entering the muffler chamber tices hit the leading edge of the tailpipe
optimization, it is very useful to get and the inflow area to the outlet pipe. inlet. This instability excites two acoustic
frequency dependent SPL values at modes of the system at 4250 and 8500 Hz.
each location of the simulated geometry 1/3 octave band SPL plots are shown
WHISTLE CASE STUDY
domain. Based on SPL plots, it is possi- in FIGURE 5 within a frequency range of
ble to identify local noise sources as An interesting whistle case based on 1 to 8 kHz. The center frequencies of
well as acoustic system modes excited a multi-chamber muffler was studied. the 1/3 octave bands are indicated on
by these sources for selected frequency The simulation was conducted at a each plot. This provides information
bands. At Faurecia a very efficient tool mass flow rate of 850 kg/h at ambient on acoustic system modes and noise
for the calculation of these SPL values temperature. sources inside the muffler dependent on
during the simulation run was devel- A comparison between measurement the frequency band. At 4 and 5 kHz a
oped and implemented. and simulation is shown in FIGURE 4 specific SPL pattern can be identified
A typical application is demon- based on SPL spectra. The SPLs, which in the joining area of pipe and tailpipe.
strated in FIGURE 3. The frequency are plotted in 1/12 octave bands, refer to At 8 kHz the pattern has changed com-
of interest in this case is 1700 Hz, the microphone positions outside the pletely in the same area.

RESEARCH. KNOWLEDGE. PROGRESS.

Developing the future – today‘s production facilities need to meet the requirements of tomorrow. This is one of the main challenges facing
vehicle manufacturers. Innovations in manufacturing processes, resource use, the effectiveness of value chains and increasing automation
and digitization are all helping to shape smart vehicle production. ATZproduction provides the latest information from the fields of research
and development and includes in-depth technical reports on all aspects of vehicle manufacturing. In addition to the print edition, we offer
an interactive e-magazine. You can also benefit from the extensive information in our unique online archive, which gives you the option of
downloading PDF files.

production W O R L D W I D E DAYS
FREE TR
IAL!
ISSN (Online)
2524-8812

02 May 2019 | Volume

06
production W O R L
D W I D E
©istockphoto.com/PhonlamiPhoto

ELECTRI C VEHICLE
PRODUC TION

The Challenge of
Mass Production
MACHINING TECHNOLOG
Y
Has to be Optimized PLASTIC FREEFORM
Materials and
for New ERS
Components Offer Flexible
Possibilities in AUTOMATIO N
Additive Manufactur Requires Reliable
ing Validation of
Human-robot
Collaboration

www.my-specialized-knowledge.com/ATZproduction-worldwide
/// INTERVIEW
Andreas Schick
[Schaeffler]
/// GUEST COMM
ENTARY Christophe
r Reuß [Edag]

MTZ worldwide 11|2019    75
DE VELO PMENT  Ac oustic s | N V H

GENERAL STUDIES design with circular holes was already The third case is a pattern of circular
discussed before. The corresponding holes like the reference design, but each
The impact of pipe perforation patterns  overall SPL is 108 dB(A). column of holes is rotated incrementally
is studied on the muffler 1 case at a By changing to longitudinal perfora- by 2°. Compared to the reference design
mass flow rate of 1000 kg/h at ambient tion holes, the pressure signal becomes the excitation of the muffler resonance
temperature. An overview of the study much stronger and the excitation pro- takes a bit longer. But finally the same
is given in FIGURE 6 and FIGURE 7. In the cess is very fast, FIGURE 6 (right col- acoustic mode is excited at a similar
first column of FIGURE 6, the perforation umn). The whistle frequency is shifted SPL magnitude.
pattern is shown, followed by pressure down to 3750 Hz and the related SPL It has to be emphasized that this eval-
time series at the microphone position peak is heavily increased, FIGURE 7. The uation of the perforation pattern is valid
during the excitation phase covering overall SPL reaches 119 dB(A). The only for the considered muffler design.
80 ms. In FIGURE 7 the SPL spectra and related pressure fluctuation plot indi- Obviously longitudinal holes cause a
the muffler mode are visualized by cates that a different acoustic muffler shift of the excitation frequency to lower
pressure fluctuation plots. The reference mode is excited. values. If there is no significant acoustic

FIGURE 6 Impact of perforation pattern: Design of perforation pattern, pressure time series (© Faurecia)

76
FIGURE 7 Impact of perforation pattern: 1/12 octave SPL spectra (base design: red line, design variants: black line), snap-shots of pressure fluctuation (from left to right)
(© Faurecia)

system mode in this frequency range, shown are very helpful to visualize REFERENCES
the result could be completely different. flow noise sources and acoustic systems [1] Shur, M. L.; Spalart, P. R.; Strelets, M. Kh.;
­Travin, A. K.: A hybrid RANS-LES approach with
modes. SPL plots to visualize flow noise delayed-DES and wall-modelled LES capabilities.
sources and acoustic systems modes were In: International Journal of Heat and Fluid Flow 29
CONCLUSIONS
shown. Essential in this context is an (2008), No. 6, pp. 1638-1649
In this paper, transient flow noise simu­ in-house developed tool, which provides [2] Gritskevich, M. S.; Garbaruk, A. V.; Schütze, J.;
Menter, F. R.: Development of DDES and IDDES
lation results derived from the IDDES frequency-dependent SPL values for each Formulations for the k- ω Shear Stress Transport
method of Simcenter STAR-CCM+ were cell of the mesh. Studies on the impact Model. In: Flow, Turbulence and Combustion 88
presented and discussed. Based on test of pipe per­foration patterns on the flow (2012), No. 3, pp. 431-449
[3] Kaiser, R.; Hinterberger, C.: Flow noise evalua­
case muffler designs covering typical phe- noise of the systems were discussed. The
tion of mufflers based on transient CFD. Aachen
nomena for flow-induced noise, a good high accuracy level of the results proves Acoustics Colloquium Proceedings, 2018, Aachen,
agreement between measurement and the capability of the approach for the vir- pp. 209-217
virtual prediction was demonstrated tual assessment of muffler designs in the
by means of SPL spectra. The SPL plots early development phase.
MTZ worldwide 11|2019    77