Downloaded from SAE International by University of British Columbia, Sunday, July 29, 2018

Methods for Determining Muffler Transmission Loss in 2016-01-1317

Octave Bands Published 04/05/2016

Peng Wang
University of Kentucky

Xin Hua
Faurecia Emissions Control Technologies

T. Wu and David W. Herrin

University of Kentucky

CITATION: Wang, P., Hua, X., Wu, T., and Herrin, D., "Methods for Determining Muffler Transmission Loss in Octave Bands," SAE
Technical Paper 2016-01-1317, 2016, doi:10.4271/2016-01-1317.
Copyright © 2016 SAE International

Abstract inlet duct at all frequencies. While the first two methods still rely on
the plane-wave assumption in the inlet and outlet ducts, the direct
Insertion loss in one-third or octave bands is widely used in industry
conversion method can be used for large silencers at higher
to assess the performance of large silencers and mufflers. However,
frequencies when higher-order cross modes propagate along with the
there is no standard procedure for determining the transmission loss
plane waves in the inlet and outlet ducts.
in one-third or octave bands using measured data or simulation. In
this paper, assuming that the source is broadband, three different
Two test cases are provided in this paper to demonstrate the three
approaches to convert the narrowband transmission loss data into
methods. The first test case is a simple expansion chamber at
one-third and octave bands are investigated. Each method is
frequencies below the plane-wave cutoff of the inlet and outlet ducts.
described in detail. To validate the three different approaches,
Below the cutoff, all three methods should agree with one another if
narrowband transmission loss data of a simple expansion chamber
the incident wave amplitude is assumed to be a constant across the
and a large bar silencer is converted into one-third and octave bands,
frequency range. The second test case is a large bar silencer for
and results obtained from the three approaches are demonstrated to
broadband noise reduction. The plane-wave cutoff of this particular
agree well with one another.
bar silencer example is at 612 Hz only, while the TL analysis is
carried out up to 8000 Hz. The direct conversion method is still
Introduction applicable above the plane-wave cutoff.
Measurements of insertion loss (IL) and noise reduction (NR) for
large mufflers and silencers are normally performed in one-third or Octave and One-Third Octave Bands
octave bands. On the other hand, transmission loss (TL) must be first
Each octave band filter has a fixed center frequency and is twice as
determined in narrow bands because TL computation requires
wide as the one before it, as shown in Fig. 1. In other words, the
additional post-processing of the measured or computed sound
bands are related by the following relationship
pressure data. In many cases, it is still desirable to convert the
narrow-band TL into one-third or octave bands in order to compare to
the measured IL or NR. However, there has been no standard
procedure for determining the TL in one-third or octave bands using
(1)
measured data or simulation. In this paper, we compare three
different methods for determining TL in one-third or octave bands. where fu and fu are the upper limit and lower limit frequency of the
They are (i) the wave decomposition method, (ii) the equivalent IL band, respectively.
method, and (iii) the direct conversion method, respectively. All three
methods assume a constant amplitude incident sound wave in the
 Downloaded from SAE International by University of British Columbia, Sunday, July 29, 2018

Determination of Muffler Transmission Loss in

Octave Bands
In this section, three different methods are presented to convert the
narrow-band TL into octave or one-third octave bands. All three
methods are based on the assumption of a constant amplitude incident
wave across the frequency range. In reality, however, the incident
sound power may be different at each frequency.

Method 1: Wave Decomposition Method

The wave-decomposition method is often used in numerical
Figure 1. Octave band filters
simulation [2] to calculate the narrow-band TL. The method is not
Each octave band filter may be divided into three one-third octave used as often in measurements because a true anechoic termination is
band filters for greater frequency resolution. The standard center, difficulty to achieve in the lab. As shown in Fig. 2, the sound
lower and upper limit frequencies for octave and one-third octave pressures at the three microphone locations are calculated by either
bands below 2900 Hz are shown in Table. 1 [1]. the boundary element method (BEM) or the finite element method
(FEM) with a unit-amplitude velocity boundary condition applied at
To convert the narrow bands to octave bands (octave bands or the inlet and an anechoic termination boundary condition applied at
one-third octave bands), it is necessary to calculate the overall sound the outlet.
level in each band filter. Since the sound level scales are logarithmic,
they cannot be combined algebraically. Instead, the energy addition
theorem is applied to combine the narrow-band sound power levels in
each band filter to obtain the overall sound power level for a given
band. This is essentially the numerical integration of sound power
over each band divided by the bandwidth. In other words, the end
result is the average sound power in each band. In practice, only a
Figure 2. Demonstration of three-point method
simple summation or integration procedure of the sound power is
performed without doing the average if the same frequency increment Below the plane-wave cutoff of the inlet and outlet ducts, sound
is used at every measurement point. pressures at the three microphone locations are

Table 1. Octave and One-Third Octave Bands Comparison [1]

(2)

(3)

(4)

Solve Eqs. (2) and (3) to extract the incident wave

(5)

The narrow-band TL is defined by the ratio between A and C. In the

narrow-band calculation, the velocity boundary condition applied at
the inlet can be arbitrary because both A and C are proportional to the
velocity applied. However, in octave bands, it is assumed that A is
constant at all frequencies for the TL calculation that follows.
Accordingly, we make A = 1 and scale C accordingly. The incident
sound power Wi and the transmitted sound power Wt can be
calculated by

(6)
 Downloaded from SAE International by University of British Columbia, Sunday, July 29, 2018

(7)

where S1 and S2 are the cross-sectional areas of the inlet and outlet,
respectively. (b).

Wi and Wt in narrow bands are then integrated (or simply summed) Figure 4. (a) Typical duct silencing system; (b) Electric analog
over each band to get the TL in octave bands:
po can also be expressed in terms of [T], ZS, ZT and ps by solving the
equations above,

(8)

(13)
Method 2: Equivalent Insertion Loss Method
This method assumes that the four-pole matrix of the muffler, as Thus, insertion loss (IL) can be calculated by
shown in Fig. 3, has been either measured or calculated in narrow
bands. However, it should be noted that the four-pole matrix does not
exist above the plane-wave cutoff of the inlet and outlet ducts.
(14)

where [D] is the four-pole matrix of the straight replacement pipe. It

is noted that when the source and termination are anechoic, Eq. (14)
can be reduced to Eq. (9). In other words, the IL is the same as TL if
both the source and termination are anechoic.

Figure 3. Four-pole matrix of the muffler To compute the TL in octave bands, the sound pressures at the outlets
of the two cases shown in Fig. 5 are computed. Since both the source
The narrow-band TL can be calculated by the four-pole parameters,
and termination are anechoic, the computed IL is the same as the TL.

(9)

Figure 4(a) shows the basic components of a typical duct silencing

system and Fig. 4(b) shows the corresponding circuit analog of the
acoustical system. In Fig. 4(b), ZS is the source impedance, ZT is the
termination impedance, pi and ui are the sound pressure and particle
velocity at inlet, and po and uo are the sound pressure and particle
velocity at outlet, respectively. Relations among these quantities are
shown in Eqs. (10), (11) and (12),
Figure 5. Two pseudo cases

According to Eq. (13), if the source and termination are both

(10)
anechoic, the sound pressures p1 and p2 at the termination of the
straight pipe and the muffler, respectively, can be calculated
separately by using the following equations:
(11)

(12)
(15)

(16)

(a).
 Downloaded from SAE International by University of British Columbia, Sunday, July 29, 2018
Assume ps is constant across the frequency range, and then convert p1
and p2 to octave-band sound pressure levels SPL1 and SPL2,
respectively. It can be proved that a constant ps with an anechoic
source is equivalent to a constant incident wave. Then, the equivalent
TL can be calculated by
(17)
Method 3: Direct Conversion Method
If the narrow-band TL is available already, the direct conversion
method can be easily applied by following the simple steps below:
1. Set the incident sound power Wi=1 in Eq. (8) at all frequencies.
2. Calculate the corresponding transmitted sound power Wt at each
frequency from the narrow-band TL.
3. Convert both Wi and Wt into octave bands by doing simple
integration (or summation).
4. Calculate the TL from the octave-band Wi and Wt.
Results and Discussion
The first test case is the simple expansion shown in Fig. 6. One-third
octave band TL curves using the three different methods presented in
this paper are compared to the narrow-band TL in Fig. 7. The
frequency stepping used in the narrow-band calculation is 10 Hz. It is
seen from Fig. 7 that all three methods produce the same TL.
Figure 6. Simple expansion chamber muffler test case (L=200 mm, d=35 mm,
D=150 mm).
Figure 7. Transmission loss comparison
The second test case is a large round bar silencer shown in Fig. 8.
First, the narrow-band TL is calculated by the impedance-to-
scattering method [3] in conjunction with the substructuring BEM
[4]. The bar silencer is made of polyester (flow resistivity=16,000
Rayl/m) and is covered by a 30% open perforated facing sheet and
two rigid end panels.
Figure 8. Dimensions of the bar silencer; L=6 m.
As demonstrated in Fig. 9, only three small sections need to be
modeled in the BEM: the beginning section (substructure 1), the
middle-section template (substructure 2 template), and the tail section
(substructure 3). The majority of the middle section does not need to
be meshed because the impedance matrix of the template can be
successively repeated along the bar silencer. An impedance matrix
synthesis procedure [4] is then performed to combine all the
substructure impedance matrices into a resultant impedance matrix.
Figure 9. A bar silencer model with three substructures
Due to the large cross section of the silencer, the plane-wave
assumption is not valid anymore at the inlet and outlet. Only the
direct conversion is used to convert the narrow-band TL into octave
bands. Fig. 10 shows the comparison between the narrow-band TL,
the 1/3-octave band TL, and the octave-band TL. The frequency
stepping used in the narrow-band TL calculation is 20 Hz.
Figure 10. Transmission loss of the bar silencer
 Downloaded from SAE International by University of British Columbia, Sunday, July 29, 2018

Conclusions
In this paper, three different methods for converting the narrow-band
TL into octave bands are presented. All three methods are based on
the same assumption that the incident wave amplitude is constant
across the frequency range. The direct conversion method is the
easiest to implement and can be extended to high frequencies when
the plane-wave assumption is not valid anymore. Although the direct
conversion approach may have been used by many noise control
engineers, a standard is still lacking at this point.

References
1. Ver, I. and Beranek, L., "Noise and Vibration Control
Engineering." Principles and Applications, Second Edition,
ISBN 471449423 (2008): 15-18.
2. Wu, T. W., and Wan, G. C. "Muffler performance studies using
a direct mixed-body boundary element method and a three-point
method for evaluating transmission loss." Journal of Vibration
and Acoustics 118, no. 3 (1996): 479-484.
3. Wang, P and Wu, T. W. "Impedance-to-scattering matrix method
for silencer analysis." in INTER-NOISE and NOISE-CON
Congress and Conference Proceedings, 248, 453-460, (2014).
4. Lou, G, Wu, T. W., and Cheng, C. Y. R. “Boundary element
analysis of packed silencers with a substructuring technique,”
Engineering Analysis with Boundary Elements, 27, 643-653
(2003).

Contact Information
T. W. Wu
Department of Mechanical Engineering, University of Kentucky
Lexington, KY 40506, USA

Acknowledgment
The work was partially supported by the Vibro-Acoustics
Consortium.

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. The 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 International.

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the content of the paper.

ISSN 0148-7191

http://papers.sae.org/2016-01-1317