1

Analytical, Numerical and Experimental calculation of sound

transmission loss characteristics of single walled muffler shells

A thesis submitted to the

Division of Graduate Studies and Research

of the University of Cincinnati

in partial fulfillment of the

requirements for the degree of

MASTER OF SCIENCE

In the Department of Mechanical Engineering

of the College of Engineering

May 2007

by

John George

B. E., Regional Engineering College, Surat, India, 1995

M. E., Regional Engineering College, Trichy, India, 1997

Committee Chair: Dr. Jay Kim

2
 Abstract

Accurate prediction of sound radiation characteristics from muffler shells is of significant

importance in automotive exhaust system design. The most commonly used parameter to

evaluate the sound radiation characteristic of a structure is transmission loss (TL). Many

tools are available to simulate the transmission loss characteristic of structures and they

vary in terms of complexity and inherent assumptions. MATLAB based analytical

models are very valuable in the early part of the design cycle to estimate design

alternatives quickly, as they are very simple to use and could be used by the design

engineers themselves. However the analytical models are generally limited to simple

shapes and cannot handle more complex contours as the geometry evolves during the

design process. Numerical models based on Finite Element Method (FEM) and Boundary

Element Method (BEM) requires expert knowledge and is more suited to handle complex

muffler configurations in the latter part of the design phase. The subject of this study is to

simulate TL characteristics from muffler shells utilizing commercially available

FEM/BEM tools (NASTRAN and SYSNOISE, in this study) and MATLAB based

analytical model. The results obtained from analytical model and numerical model are

correlated with experimental data. The predicted results agreed reasonably well with the

experimental results. The effects of important design parameters are studied to provide

design guidelines.

3
4
 Acknowledgements

First and foremost, I would like to express my sincere gratitude to my advisor

Dr. Jay Kim for his valuable guidance and outstanding support during the course of this

work. His constructive comments and thoughtful insights have greatly improved the

quality of this work.

I would like to thank Dr. David Thompson and Dr. Yijun Liu for accepting my request to

serve on my masters’ thesis committee.

I am truly grateful to Jim Egan and Dr. Chulho Yang, my colleagues at ArvinMeritor, Inc

for their valuable help and suggestions. Their expert knowledge and assistance has helped

me a great deal to complete this work successfully.

My sincere thanks to Dave Leehaug, Chief Engineer at ArvinMeritor, Inc for appreciating

this work and providing me with whatever I requested.

Last but by no means least, I am forever indebted to my wife Vidhya for her unwavering

support and encouragement throughout the course of my studies.

5
 Table of contents
1. INTRODUCTION …………………………………………………………………………..9

2. THEOROTICAL BACKGROUND……………………………………………………….12
2.1. INSERSTION LOSS ……………………………………………………………………….12
2.2. TRANSMISSION LOSS …………………………………………………………………...12
2.3. LEVEL DIFFRENCE ………………………………………………………………….......12

3. FORMULATION AND SOLUTION OF TL FOR SINGLE WALLED CYLINDRICAL

SHELL BY ANALYTICAL METHOD…………………………………………………...13
3.1. FORMULATION AND SOLUTION OF TL FOR SINGLE WALLED CYLINDRICAL SHELL……...14
3.2. TRANSMISSION LOSS (TL) CACULATION FOR SINGLE WALLED CYLINDRICAL SHELL…...18
3.3. CORRECTIONS TO PAST WORK ON TL FOR SINGLE WALLED CYLINDRICAL SHELL…........20
3.3.1. Discussion of results….........................................................................................20

4. DESCRIPTION OF THE PROTOTYPE MODEL USED FOR CORRELATION……23

5. FINITE ELEMENT MODAL ANALYSIS………………………………………………..24

5.1. OVERVIEW OF ANALYSIS USING NASTRAN……………………………………………....24
5.2. OBTAINING NATURAL MODES USING NASTRAN…………………………………………25
5.3. RESULTS FROM FINITE ELEMENT MODAL ANALYSIS FOR PROTOTYPE MUFFLER..............26

6. ACOUSTIC BOUNDARY ELEMENT ANALYSIS AND EXPERIMENTAL

CORRELATION....................................................................................................................28
6.1. OVERVIEW OF CALCULATING RADIATED NOISE USING SYSNOISE....................................28
6.1.1. Jump boundary condition.....................................................................................30
6.1.2. Modeling of acoustic source.................................................................................30
6.2. CALCULATION OF ACOUSTIC MODES OF THE CAVITY ......................................................32
6.3. TL CALCULATION FROM EXPERIMENT.............................................................................. 35
6.3.1. Frequency range for TL calculation.................................................................... 35
6.3.2. Acoustic pressure measurement inside the cavity.................................................35
6.3.3. Acoustic pressure measurement outside the cavity...............................................37
6.3.4. Characterization of background noise................................................................. 39
6.3.5. TL from test……………………………………………………………………...41
6.4. TL CALCULATION USING BEA…………............................................................................43
6.5. COMPARISON OF RESULTS BETWEEN EXPERIMENT, BEA AND ANALYTICAL
FORMULATION..................................................................................................................46
6.5.1. Discussion of results.............................................................................................48

7. STUDY ON THE SENSITIVITY OF DESIGN PARAMETERS ON TL........................49

7.1. EFFECT OF THICKNESS.......................................................................................................49
7.2. EFFECT OF SHAPE..............................................................................................................52
7.3. DISCUSSIONS ON DESIGN SENSITIVITY STUDY..................................................................54

8. CONCLUIONS AND RECOMMENDATIONS FOR FUTURE WORK........................56

6
 List of figures

3.1 SCHEMATIC OF THE MODEL USED IN THE FORMULATION FOR SINGLE WALLED SHELL............15
3.2 COMPARISON OF TL BETWEEN ORIGINAL FORMULATION AND CORRECTED FORMULATION OF
SINGLE WALLED SHELL FOR INCIDENCE ANGLE =45 DEG. .......................................................21
4.1 GEOMETRIC DETAILS OF THE PROTOTYPE MUFFLER.................................................................23
5.1 FINITE ELEMENT MODEL OF THE CAVITY UNDER STUDY..........................................................26
5.2 MODE SHAPES FOR 0.6 MM SHELL (PROTOTYPE MODEL) .........................................................27
6.1 TEST/SYSNOISE MODEL OF THE SPEAKER..............................................................................31
6.2 MEASURED SPEAKER RESPONSE VS SYSNOISE SOURCE.........................................................31
6.3 MESH OF THE ACOUSTIC CAVITY AND APPLIED IMPEDANCE....................................................33
6.4 ACOUSTIC MODES OF THE CAVITY............................................................................................34
6.5 SCHEMATIC REPRESENTATION OF PRESSURE MEASUREMENT INSIDE THE CAVITY..................36
6.6 MEASURED VALUE AND AVERAGE PRESSURE INSIDE THE MUFFLER........................................37
6.7 TEST SET UP TO MEASURE ACOUSTIC PRESSURE OUTSIDE THE CAVITY....................................38
6.8 MICROPHONE LOCATIONS TO MEASURE RADIATED NOISE.......................................................38
6.9 AVERAGE PRESSURE VALUES MEASURED INSIDE AND OUTSIDE THE MUFFLER.......................39
6.10 TEST SET UP TO MEASURE BACKGROUND NOISE................................................................40
6.11 RESULTS FROM BACKGROUND NOISE MEASUREMENT.......................................................41
6.12 CORRECTED SOUND PRESSURE OUTSIDE THE CAVITY........................................................42
6.13 TL FROM TEST....................................................................................................................42
6.14 BE MODEL USED TO CALCULATE PRESSURE INSIDE THE CAVITY......................................43
6.15 CALCULATED VALUE AND AVERAGE PRESSURE INSIDE THE CAVITY FROM BEA..............44
6.16 BE MODEL USED TO CALCULATE PRESSURE OUTSIDE THE CAVITY...................................45
6.17 AVERAGE PRESSURE VALUE OUTSIDE THE CAVITY CALCULATED USING BEA..........45
6.18 TL-CALCULATED USING BEA............................................................................................46
6.19 COMPARISON OF AVERAGE ACOUSTIC PRESSURE INSIDE THE CAVITY..............................47
6.20 COMPARISON OF TL’S........................................................................................................47
7.1 MODEL WITH 1.2/1.8 MM SHELL...............................................................................................49
7.2 COMPARISON OF RADIATED NOISE FOR DIFFERENT THICKNESS...............................................50
7.3 COMPARISON OF TL FOR DIFFERENT THICKNESS FROM BEA..................................................51
7.4 COMPARISON OF TL FOR DIFFERENT THICKNESS FROM ANALYTICAL CALCULATION.............51
7.5 MODELS WITH ELLIPTICAL/STAMPED CROSS-SECTION.............................................................52
7.6 COMPARISON OF RADIATED NOISE FOR DIFFERENT CROSS-SECTIONS......................................53
7.7 COMPARISON OF TL FOR DIFFERENT CROSS-SECTIONS..........................................................54

7
 List of tables

3.1 PARAMETERS USED FOR TL CALCULATIONS ...........................................................................21

5.1 STRUCTURAL MODES FOR THE 0.6 MM MUFFLER SHELL..........................................................27
6.1 ACOUSTIC MODES OF THE CAVITY............................................................................................34
7.1 STRUCTURAL MODES OF THE 1.2/1.8 MM MUFFLER SHELL.......................................................50
7.2 STRUCTURAL MODES OF THE ELLIPTICAL/STAMPED MUFFLER SHELL.....................................53

8
1. Introduction
One of the main contributors of radiated noise in an automotive exhaust system is the

muffler shell. Restrictive environmental legislation and customer preferences call for

reliable predictive methods to calculate the sound transmission characteristics from

muffler shells. Analytical, numerical or experimental method could be used to estimate

the performance level with respect to sound transmission. Each method has its own

inherent advantages and disadvantages.

Lee and Kim established solution techniques to study sound transmission

characteristics of cylindrical side walls with single layer [4], double layer [5], double

layer with porous core [14] and a single layer with periodic stiffeners [15]. Generally the

system was idealized as an infinitely long cylinder subjected to a plane incident wave.

For all the cases solutions were obtained by solving the full shell vibration equations

using the mode superposition method. Relative performance comparisons of different

designs could be done very quickly with this method. But this method is limited to

cylindrical shapes only.

Various researchers have used coupled structural FEM-BEM models to effectively

predict the transmission loss (TL) of different structures [2, 3, 9, and 10]. In this method,

normal modes of the structure are calculated using finite element software. The modes

obtained are transferred to a boundary element software to launch a coupled structural-

boundary element analysis to calculate the sound pressure fields. Sound pressure levels

are calculated both inside and outside the structure to compute the transmission loss

characteristics. This method can be used very effectively to predict the TL for mufflers

9
with complex external contours and internal details. However this method requires CAE

specialist’s skills.

The objective of this study is to obtain TL from muffler shells utilizing commercially

available FEM/BEM tools (NASTRAN and SYSNOISE, in this study) and MATLAB

based analytical model. TL of a single shelled muffler was measured experimentally and

compared to both numerical and analytical solution. Once the simulation model is

validated with experimental results they could be used very effectively to evaluate

various design alternatives. The model used in the experimental, numerical and analytical

models had same dimensions and properties except for their finite lengths in experimental

and numerical model. Thick end caps were used to close both ends of the cylinders to

eliminate the effect of sound radiated from end plates. Experimental comparison helps to

find out whether the assumptions involved in the simulation models are reasonable as

compared to a real world situation. It also helps to understand if any fundamental

problem exists in the simulation strategy or in the theoretical formulation. Depending on

the geometry of the model and objective of the study, either of the predictive method

could be used.

The discussion of this thesis is as follows

Chapter 2 and 3: Brief overview of different sound transmission characteristic of a

structure like transmission loss (TL), insertion loss (IL) and level difference (LD) is

discussed. Formulation of TL for a single walled shell by analytical method based on past

work is reproduced. Proposed corrections to past formulations are done. TL for a typical

single walled muffler shell is obtained from the corrected formulations. The difference in

results between the past work and corrected formulation is discussed.

10
Chapter 4: The schematic representation of the prototype model used in this study is

presented.

Chapter 5: Brief overview of the finite element method (FEM) based modal analysis is

discussed. Natural modes and mode shapes for the prototype model are obtained using

NASTRAN, commonly used commercial finite element software.

Chapter 6: Brief overview of the Boundary Element Analysis (BEA) is discussed. TL

curves are obtained for a single walled shell (model with same geometry as in the

prototype model) from SYSNOISE, commonly used commercial boundary element

software. The test setup and the transmission loss obtained from the test are presented.

Comparison between experimental results, BEA results and results from the corrected

analytical work are discussed.

Chapter 7: Effect of important design parameters like thickness and shape are analyzed

and results from the analysis are presented.

Chapter 8: Conclusions from the study and recommendations for future work is

presented.

11
2. Theoretical background

Sound transmission characteristic of a structure can be measured in terms of the one of

the following parameters [1].

(a) Insertion loss, IL

(b) Transmission loss, TL

(c) Level difference LD, or noise reduction, NR

2.1 Insertion loss, IL

Insertion loss is defined as the difference between the acoustic power radiated without the

structure and that with the structure. Symbolically,

IL= 10 log(W1/W2) dB (2.1)

where W1 and W2 denote the acoustic power without the structure and with the structure.

2.2 Transmission loss, TL

Transmission loss is defined as the ratio of the incident power and transmitted power

from the structure. Symbolically,

TL= 10 log(Wi/Wt) dB (2.2)

where Wi and Wt denote the incident acoustic power and transmitted acoustic power. TL

is used in this work to evaluate the performance of the muffler wall.

2.3 Level difference, LD

Level difference LD, or noise reduction, NR is the difference in sound pressure levels at

two arbitrarily selected points inside the structure and outside the structure. Symbolically,

LD= 20log(Pi/Po) dB (2.3)

where Pi and Po denote the pressure inside the structure and outside the structure.

12
3. Formulation and solution of TL for single walled cylindrical

shell by analytical method

Lee and Kim developed analytical models to solve for TL of cylindrical muffler

side walls with various cross-sectional designs. They considered a cylindrical side wall of

single layer [4], double layer [5], double layer with porous core [14] and a single layer

with periodic stiffeners [15]. To enable analytic solution, some simplifications were

employed.

With reference to figure 3.1, the following assumptions were made.

1) The system is considered infinitely long, which enables both the acoustic media and

the shell to be described by traveling waves.

2) The incident wave to the system is a plane wave from the outside space.

3) Shell cavity is anechoic so that no reflected wave exists inside the cavity.

The last assumption is not derived from a physical consideration, but from the need to

make the problem solvable by an analytic procedure.

The assumptions make the system a rough approximation of the inverse of a muffler

running in an anechoic chamber, which has a diffusive field inside and anechoic

condition outside. Obviously the real muffler has neither an infinite length nor a plane

incident wave. Therefore, it is clear that the theoretical model is developed only to

represent main characteristics of the problem for relative comparisons of design

alternatives.

13
In the following, the analyses conducted by Lee and Kim [4, 8] are summarized.

Obviously, this is done only to make the discussions in this work somewhat more

complete and no credit is claimed.

3.1 Formulation and solution of TL for single walled cylindrical shell

The vibro-acoustic response of the system is obtained by solving the shell

equations and acoustic waves simultaneously. Love’s equation is used to describe the

shell motion and the acoustic wave equations describe the responses of acoustic media.

Figure 3.1 illustrates the schematic of the cavity that is used for the formulation of the

problem. The cylinder is assumed to have infinite length. γ1 represents incidence angle, Ri

represents radius of the cavity, hi represents wall thickness, ρi represents mass density

µi represents Poisson’s ratio and Ei represents Young’s modulus. The incident wave is a

plane wave whose rays are traveling on planes parallel to the x-z plane. The acoustic

media in the outside and inside of the shell are defined by density and speed of sound:

{ρ1,c1} inside and {ρ3,c3} outside.

14
 Figure 3.1: Schematic of the model used in the formulation for single walled shell

The wave equation in the external and internal spaces are given by [4]

∂ 2 ( p I + p1R )
c1∇ ( p + p ) +
2 I R
=0 (3.1)
∂t 2
1

∂ 2 p3T
c3 ∆2 p3T + =0 (3.2)
∂t 2

where, ∇ 2 is the Laplacian operator, p I , p IR p3T are the acoustic pressures of incident,

reflected and transmitted waves.

The equation of the shell is, letting {u10 , v10 , w10 } be the displacements of the shell at the

neutral surface in the axial, circumferential and radial directions respectively,

{
L1 u10 , v10 , w10 } = ρ i hi u&&10 (3.3)

L2 {u10 , v10 , w10 } = ρ i hi v&&10 (3.4)

L3 {u10 , v10 , w10 } + ( p I + p IR ) − p3T = ρ i hi w

&&10 (3.5)

15
The differential operators L1, L2 and L3 in equations (3.3), (3.4) and (3.5) are found in [4]

On the internal and external shell surfaces, the particle velocities of the acoustic media in

the normal direction have to be equal to the normal velocity of the shell, which results in

the following equations [4]:

∂ ( p I + p1R ) ∂ 2 w10
= − ρ1 at r=Ri (3.6)
∂r ∂t 2

∂p3T ∂ 2 w10
= ρ3 at r=Ri (3.7)
∂r ∂t 2

The harmonic wave pI incident from outside direction in the cylindrical coordinates as

shown in Figure 3.1 can be expressed as [4] :

∞
p I ( r, z, θ , t ) = p0 ∑ ε n ( − j ) n J n ( k1r r ) cos[nθ ]e j (ωt −k1 z z ) (3.8)
n =0

where p0 is the amplitude of the incident wave, j = − 1 n=0,1,2,3……,Jn is the Bessel

function of the first kind of order n; ω is the angular frequency, εn = 1 for n=0 and εn=2

for n=1,2,3……, k1z=k1sin(γ1) , k1r=k1cos(γ1) and k1=ω/c1. It is easily seen that

k1r = k12 − k12z . The phase speed of the wave to the direction of shell surface is defined

as :

ω c1
c p1 = = (3.9)
k1z sin(γ 1 )

The waves radiated from the shell to the outside and into the cavity, p1R and p3T can be

represented as:

16
 ∞
p ( r , z , θ , t ) = ∑ p1Rn H n2 (k1r r ) cos[nθ ]e j (ωt − k1 z z )
R
1 (3.10)
n =0

∞
p3T ( r , z , θ , t ) = ∑ p3Rn H n1 ( k 3r r ) cos[nθ ]e j (ωt − k3 z z ) (3.11)
n =0

where H n1 and H n2 are the Hankel functions of the first and second kind of order n,

respectively. The former represents the incoming wave and the latter the outgoing wave.

Because the traveling waves in the acoustic media and in the shell are driven by the

incident traveling wave, the wave numbers (or trace velocities) in the z direction should

match throughout the system therefore k32=k12. Hence, the three components of the shell

displacements can be expressed as [4] :

∞
w10 ( z, θ , t ) = ∑ w10n cos[nθ ]e j (ωt −k1 z z ) (3.12)
n =0
∞
u10 ( z , θ , t ) = ∑ u10n cos[nθ ]e j (ωt −k1 z z ) (3.13)
n =0
∞
v10 ( z , θ , t ) = ∑ v10n cos[nθ ]e j (ωt −k1 z z ) (3.14)
n =0

where, n=0,1,2,3….are the circumferential mode numbers

Substituting equations (3.8) and (3.10)-(3.14) into three shell equations and utilizing the

orthogonality between trigonometric functions, three equations are obtained for each

circumferential mode number n=0,1,2,3……………[4]

 K (1 − µ )  K (1 + µi ) K µ
u10n  ρ i hiω 2 − K i k12z − i 2 i n 2  − i nk1z v10n j − i i k1z wIn0 j = 0 (3.15)
 2 Ri  2 Ri Ri

17
 Ki (1 − µi ) K D K (1 − µi ) 2 K i 2 Di (1 − µi ) 2
u10n [ k1z nj + i µi k1z nj − 2i µi k12z n ] + vIn0 [ − i k1z − 2 n − k1z ]
2 Ri Ri Ri 2 Ri 2 Ri2
Di 2  K D (1 − µ ) D 
+ v In0 [ 4
n + ρ i hiω 2 ] + ω In0  − 2i n − i 2 i k12z n − 4i n 3  = 0 (3.16)
Ri  Ri Ri Ri 
D K D D (1 − µ ) D K 
u10n [ − Di k14z − 2i µi k12z n 2 + i µi k1z j ] + v10n  2i µi k12z n − i 2 i k12z n + 4i n 3 − 2i n 
Ri Ri  Ri Ri Ri Ri 
D 2 Di (1 − µi ) 2 2 Di 4 K i
+ ω1n0 [ − 2i µi k12z n − 2
k1z n − 4 n − 2 + ρ i hiω 2 ]
Ri Ri Ri Ri
+ p1Rn H n2 ( k1r Ri ) − p3Tn H n1 ( k 3r Ri ) = − p0ε n ( − j ) n J n ( k1r Ri ) (3.17)

Because two boundary conditions are available, we have 5 equations and 5 unknowns,

which can be arranged in the following form.

0 0 A B C   p1Rn  0 
0  T   
0 D E F  p3n  0 
   0   
G H I J K  u1n  =  P  (3.18)
   0  Q 
L 0 0 0 M 
v1n   
0 N 0 0 O   w10n  0 
 

Where, the details of coefficients A, B, C. etc can be found from [4].

Five unknown coefficients p1Rn , p3Tn , u10n , v10n , w10n are obtained in terms of po by

solving equation (3.18), which can be substituted back to equations (3.10)-(3.14),

providing a set of exact solutions in series expressions for the solution of the system [4].

3.2 Transmission loss (TL) calculation for single walled cylindrical shell

TL can be defined as the ratio of the incoming and transmitted power per unit

length of the cylinder [4].

18
 WI
TL = 10 log10 ∞
(3.19)
∑W
n =0
n
T

Where WnT is the transmitted power flow per unit length of the shell

WT =
1
2
{
0
2π
Re ∫ p3T . ∂ ( w10 ) * Ri dθ
∂t
} where r=R i (3.20)

Where Re{.} and the superscript * represent the real part and the complex conjugate of

the argument, respectively. Substitution of equations (3.11) and (3.12) for p3T , w10 into

above equation (3.20) yields an expression for the component of WnT [4].

2π
Re{p3Tn H n1 ( k 3r Ri ).( jωw10n )}× ∫ cos 2 [nθ ]. Ri dθ
1
WnT = where r=Ri
2 0

πRi
= × Re{p3Tn H n1 ( k 3r Ri ).( jωw10n )* } (3.21)
2ε n

where εn=1 for n=0 and εn=2 for n=1,2,3…..

W I , the incident power flow per unit length of the shell is

cos(γ 1 ) p02
WI = × 2 Ri (3.22)
ρ 1c1

Finally, the transmission loss can be obtained by substituting equations (3.21) and (3.22)

into (3.19).

Re{p3Tn × H n1 ( k 3r Ri ) × ( jωw10n )* }× ρ1c1π

∞
TL = −10 log10 ∑ (3.23)
n =0 4ε n cos(γ 1 ) p02

where εn=1 for n=0 and εn=2 for n=1,2,3…..

19
3.3 Corrections to past work on TL of single walled cylindrical shell

A sign mistake in the formulation of Lee and Kim [4] summarized above was

identified during this work. The original equation of motion is:

L3 {u10 , v10 , w10 } + ( p I + p IR ) − p3T = ρ i hi w

&&10 (3.24)

Because the outward normal direction should be taken positive in the shell equation, the

correct equation should have been:

L3 {u10 , v10 , w10 } − ( p I + p IR ) + p3T = ρ i hi w

&&10 (3.25)

Ensuing corrections in the solution are as follows,

Equation 3.17 becomes

Di K D D (1 − µ ) D K 
u10n [ − Di k14z −2
µi k12z n 2 + i µi k1z j ] + v10n  2i µi k12z n − i 2 i k12z n + 4i n 3 − 2i n 
Ri Ri  Ri Ri Ri Ri 
D 2 Di (1 − µi ) 2 2 Di 4 K i
+ ω10n [ − 2i µi k12z n − 2
k1z n − 4 n − 2 + ρ i hiω 2 ]
Ri Ri Ri Ri
− p1Rn H n2 (k1r Ri ) + p3Tn H n1 ( k 3r Ri ) = − p0ε n ( − j ) n J n ( k1r Ri ) (3.26)

Terms G, H and P from equation 3.18 becomes

G = − H n2 ( k1r Ri )

H = H n1 ( k 3r Ri ) (3.27)

P = − poε n ( − j ) n J n (k1r Ri )

3.3.1 Discussion of results

Figure 3.2 shows the comparison of the results from the original and corrected

versions of the equation of motion of the single walled cylindrical shell. The parameters

used for the analysis are given in table 3.1. Incidence angle of 45 deg was chosen for this

20
comparison and it was found that the results show similar trends for other values of

incidence angle.

100
original TL for incidence angle=45 deg
corrected TL for incidence angle=45 deg
80

60
TL (dB)

40

20

0
10 100 1000 10000
Frequency (Hz)

Figure 3.2: Comparison of TL between original formulation and corrected formulation of

single walled shell for incidence angle =45 deg.

Shell Cavity Ambient

Material Steel Air Air
Density (Kg/m3) 7,750 0.5243 1.21
Poisson's Ratio 0.3 - -
Young's Modulus (Pa) 1.90E+11 - -
Radius (m) 0.1 - -
Thickness (mm) 1 - -
Temperature (0C) - 400 20
Sound Speed 6,100 520 343
Incidence Angle (degree) 45

Table 3.1: Parameters used for TL calculations

21
 The sign mistake essentially has the effect of applying the force terms to the

opposite direction of the shell. This would not have made any error if the system is flat,

thus does not have any curvatures. As it is shown in figure 3.2, the error due to the sign

mistake is confined in the low frequency range. This may be explained by the fact that

the low frequency region is largely controlled by stiffness, whose effect is dependent on

the direction of the forces.

22
4. Description of the prototype model used for correlation

The prototype muffler used in this study consisted of typical dimensions used in

the automotive industry. The model was fabricated with single walled shell with 0.6 mm

thickness, 0.5 m length and 0.125 m radius. This model was used for experimental testing

and was considered as the baseline for the study on design parameters detailed in

chapter-7. The geometric details of the prototype muffler are shown in Figure 4.1.

The muffler had thick end caps to eliminate the effect of sound radiation from end plates

to make the model compatible with the theoretical model. The geometry of the muffler is

shown in figure 4.1. The pipe on the right side was used for outlet and pipe on left side

for inlet.

Figure 4.1: Geometric details of the prototype muffler

23
 5. Finite element modal analysis and experimental correlation

In order to calculate the TL of a structure using the coupled structural FEA-BEA

method, the modes/mode shapes of the structure has to determined first. In this chapter

natural modes of the test object under study are established. Natural modes are calculated

using finite element method, using readily available commercial finite element software,

NASTRAN.

5.1 Overview of analysis using NASTRAN

A typical finite element analysis in NASTRAN is conducted in the following

three steps.

• Pre-processing steps (building the model): A computer aided design (CAD)

geometry is transformed in to a finite element model. NASTRAN is essentially a

solver and does not have any preprocessing capability. Hypermesh is commonly

used as the pre-processor for NASTRAN. The following parameters are defined

in the preprocessor [21].

o Element type

o Element thickness and integration parameters (real constants)

o Material properties

o Mesh of the geometry

• Solution steps (loads and boundary conditions): The parameters defined in this

steps are :

o Analysis type and analysis options

o Application of loads

24
 o Application of boundary condition

o Specification of load step options

o Initiation of analysis

• Post-processing steps (review of results): Results obtained in the solution phase

are reviewed at this stage. Since NASTRAN does not have any post processing

capability, Hypermesh is used to review the results.

5.2 Obtaining natural modes using NASTRAN

The conditions and properties used in the FEA are as follows [12].

o Element Type : Quadrilateral Plate Element (CQUAD4)

o Material property used : Steel [22]

Young’s Modulus : 206.2 GPa

Poisson’s Ratio : 0.29

Density :7695.04 Kg/m3

Assumptions implicitly involved in finite element model are:

o Linear isotropic material properties

o No imperfections in the geometry (uniform thickness, perfectly

cylindrical shape)

o Integral connection between all components

Boundary condition: A free-free boundary condition.

The finite element model of the cavity under study is shown in figure 5.1.

25
 Figure 5.1: Finite element model of the cavity under study

5.3 Results from the finite element modal analysis for the prototype

muffler

The fourteen modes obtained for the prototype model (0.6 mm shell) are summarized in

table 5.1 and the corresponding mode shapes are presented in the figure 5.2.

Modes for 0.6 mm shell

Mode # Frequency (Hz) Description
1 339.4 Circumferential mode with five lobes
2 340.1 Repeated root of mode #1
3 348.1 Circumferential mode with four lobes
4 348.8 Repeated root of mode #3
5 396.5 Circumferential mode with six lobes
6 396.8 Repeated root of mode #5
7 462.5 Circumferential mode with three lobes
8 463.5 Repeated root of mode #7
9 494.7 Circumferential mode with seven lobes
10 494.8 Repeated root of mode #9
11 609.0 Circumferential mode with eight lobes
12 609.2 Repeated root of mode #11
13 786.2 Circumferential mode with two lobes
14 786.6 Repeated root of mode #13

Table 5.1: Structural modes for the 0.6 mm muffler shell

26
1st mode at 339.4 Hz 3rd mode at 348.1 Hz 5th mode at 396.5 Hz

7th mode at 462.5 Hz 9th mode at 494.7 Hz 11th mode at 609.0 Hz

13th mode at 786.2 Hz

Figure 5.2: Mode shapes for 0.6 mm shell (prototype model)

27
6. Acoustic boundary element analysis and experimental

correlation

The radiated noise from a structure could be numerically calculated using a

variety of commercially available codes [16, 17, and 19]. The four most popular codes

used in the industry are:

o SYSNOISE

o ABAQUS

o ANSYS

o NASTRAN

ABAQUS, ANSYS and NASTRAN are based on finite element algorithm and they do

not have the capability for performing any kind of boundary element analysis.

SYSNOISE can do both boundary element and finite element calculations. The main

disadvantage of finite element analysis is that the entire fluid medium surrounding the

object of study should be meshed and for a complex object like an automotive muffler

this could be quite tedious. For the boundary element calculations, one need not mesh the

surrounding fluid medium and hence the meshing stage of the analysis is quite simple.

But boundary element analyses are quite computationally intensive and could take much

longer solution time. For the geometry considered in this study, the model size in terms

of the number of nodes is small and hence the boundary element analysis using

SYSNOISE as solver was deemed to be more appropriate.

6.1 Overview of calculating radiated noise using SYSNOISE

A typical radiated noise calculation using SYSNOISE has the following steps

28
Step-1

Model-1 – Structural FEM

o Import the FE model to SYSNOISE

o Import the natural modes obtained from FEM analysis to

SYSNOISE

Model-2 – Acoustic BEM

o Import the FE model to SYSNOISE

o Model the acoustic source (for excitation)

o Apply jump boundary condition on free edges

o Point normal vectors of the shell elements towards the cavity

o Couple the FEM model (model-1) and BEA model (model-2)

to launch a coupled solution

o Sample the acoustic pressure inside the cavity

o Store the displacements of the shell in a separate file

Step-2

Model-3 – Acoustic BEM

o Import the FE model to SYSNOISE

o Import the displacements obtained in step-2

o Point normal vectors of the shell elements outwards

o Apply jump boundary condition on free edges

o Calculate the radiated noise

29
The step-2 is not necessarily applied to all cases. Radiated noise could be calculated in

step-1 itself. Step-2 is specifically done to avoid any influence of direct input from source

on sound pressure calculation outside the cavity.

6.1.1. Jump boundary condition

In BEM there cannot be a finite pressure jump along a free edge, because the fluid

on both sides of the surface is in direct contact and there is no mechanism to support a

pressure difference. This pressure continuity condition requires that a zero jump pressure

condition is defined along the free edges of the mesh [12].

6.1.2. Modeling of acoustic source

The first step of the correlation study between numerical simulation and

experiment is to establish the characteristics of the excitation source. For all tests done in

this work, a 12 inch speaker was used as the sound pressure source. The acoustic output

of the speaker was measured using a B&K microphone (type 2670) and this measured

response was utilized to represent the sound source inside SYSNOISE. B&K PULSE

labshop version 11.1 software was used for data acquisition and signal processing. The

data was collected at a sampling frequency of 4096 Hz, frequency resolution of 4 Hz and

average of 40 samples were taken. The speaker with the adapter and the equivalent

SYSNOISE model is shown in figure 6.1. The comparison between the measured SPL

values and the SYSNOISE source model is shown in figure 6.2. It is shown that the

BEM analysis provides a quality solution.

30
 0.1 m long and 0.085 mm dia
Source

Speaker Adapter pipe

Test setup with speaker and connector pipe BEA model with source and connector pipe

Figure 6.1 : Test/SYSNOISE model of the speaker

120

100

80

60
Pressure (dB)

speaker-measurement
40 BEA Source

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.2: Measured speaker response vs. SYSNOISE source

31
6.2. Calculation of acoustic modes of the cavity

It is a reasonable speculation that the frequency at which the radiated noise would

be maximum is either at the structural modes of the shell or at the acoustic modes of the

cavity [18]. The structural modes of the prototype model were established in chapter 5

using the finite element modal analysis. Structural modal analysis is an integral step

during the calculation of radiated noise using BEM. The acoustic modes of the cavity

could be obtained by performing a FLUID FEM analysis in SYSNOISE. Acoustic finite

element analysis is not a part the boundary element analysis, which is used to calculate

the radiated noise. This analysis is performed only to obtain a better understanding of the

system behavior. The acoustic FLUID finite element analysis consists of the following

steps.

• Mesh the acoustic cavity in HYPERMESH

• Import the FE model to SYSNOISE

• Assign the properties of air to the mesh

Mass density = 1.225 kg/m3

Sound velocity = 340 m/sec

• Apply impedance boundary condition on the open faces using the equation [7]

Z = ρc(0.24( ka ) 2 + j (0.56ka ))
(6.1)
k = 2πf
340
a= radius of pipe

• Solve the model to obtain acoustic modes

The mesh of the acoustic cavity and applied impedance are shown in figure 6.3.

32
 0.25 m dia
0.5 m long

Impedance applied
on this face

Figure 6.3: Mesh of the acoustic cavity and applied impedance

The twelve acoustic mode of the prototype model (0.5 m length and 0.25 m diameter)

obtained from FLUID finite element analysis is listed in table 6.1. The pressure

distributions of the first twelve modes are presented in figure 6.4.

Acoustic modes for cavity with

0.5 m long and 0.25 m dia.
Mode # Modal frequency
1 253
2 344
3 510
4 676
5 767
6 809
7 812
8 880
9 893
10 1017
11 1045
12 1059

Table 6.1: Acoustic modes of the cavity

33
mode #1 @ 253 Hz mode #2 @ 344 Hz mode #3 @ 510 Hz

mode #4 @ 676 Hz mode #5 @ 767 Hz mode #6 @ 809 Hz

mode #7 @ 812 Hz mode #8 @ 880 Hz mode #9 @ 893 Hz

mode #10 @ 1017 Hz mode #11 @ 1045 Hz mode #12 @ 1059 Hz

Figure 6.4: Acoustic modes of the cavity

34
6.3 TL calculation from experiment

As explained in section 2.2 the TL can be calculated by knowing the acoustic

power inside the muffler and outside the muffler. Equation 2.2 can be expressed in terms

of acoustic pressure as [20],

TL= 20log(Pain/Paout) (6.2)

where Pain is the acoustic pressure inside the muffler and Paout is the acoustic pressure

outside the muffler. In order to calculate the TL, acoustic pressure inside the cavity and

acoustic pressure outside the cavity has to be determined.

6.3.1. Frequency range for TL calculation

Most of the muffler radiated noise issues in automotive industry exist in the 2000-

6000 rpm range during the 2-12 order engine speed. This translates into the frequency

range of 65-1200 Hz. Therefore, all the analysis in this work uses the frequency range

between 50-1250 Hz.

6.3.2 Acoustic pressure measurement inside the cavity

The acoustic pressure inside the cavity varies from location to location. In order to

obtain a representative pressure value inside, average pressure at three locations inside

the cavity was calculated. The schematic representation for pressure measurement inside

the cavity is shown in figure 6.5. The pressure obtained at three microphone locations

where averaged to obtain the pressure value inside the cavity.

35
 Point-3 Point-2 Point-1
Adapter pipe

0.2625m

Connection pipe 0.325m

muffler
Speaker 0.45m

Figure 6.5: Schematic representation of pressure measurement inside the cavity

The speaker with the adapter was connected to the muffler with a connection pipe (54

mm diameter and 288 mm length). Acoustic pressure inside the cavity was measured at

three locations. The three measurements were averaged using equation 6.3 [11].

Pinsde-avg=10log[1/3{10(Pinside1/10) + 10(Pinside2/10) + 10(Pinside3/10)}] (6.3)

Pinside1, Pinside2 and Pinside3 are the pressure values inside the muffler. Sound pressure

measurements were done using B&K (type 2670) microphones. B&K PULSE labshop

version 11.1 software was used for data acquisition and signal processing. The data was

collected at a sampling frequency of 4096 Hz, frequency resolution of 4 Hz and average

of 40 samples were taken. The measured pressure values and the average value inside the

muffler are shown in figure 6.6.

36
 160

140
Point-2 :Nodal point for
acoustic mode at 676 Hz
120

100
Pressure (dB)

80

60
Point-3 :Nodal point for
point-1
acoustic mode at 344 Hz
40 point-2
point-3
20 average-pressure-inside

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.6: Measured value and average pressure inside the muffler

6.3.3 Acoustic pressure measurement outside the cavity

The test setup used to measure the acoustic pressure outside the cavity is shown in

figure 6.7. In order to avoid the influence of direct tail pipe noise on radiated noise

measurement, the tail pipe of the muffler was connected to a set up for anechoic

termination as shown in figure 6.7. Pressure outside the muffler was averaged at three

locations to calculate the radiated noise. All the microphones were placed at 0.625 m

from the muffler center. Schematic of the microphone locations are shown in figure 6.8.

The average pressure value outside the muffler is shown in figure 6.9.

37
Microphone location at 0.625 m
from the center of cavity Anechoic
termination

Muffler with 0.6 mm shell

Connector pipe

Figure 6.7: Test set up to measure acoustic pressure outside the cavity

Three microphones
at 0.625 m

0.625 m

muffler
Rigid floor

Figure 6.8: Microphone locations to measure radiated noise.

38
 160

average-pressure-inside
140
averaged-radaited-noise-outside
120

100
Pressure (dB)

80

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 6.9: Average pressure values measured inside and outside the muffler

6.3.4 Characterization of background noise

The sound pressure measured with the set up explained in the previous section

(section 6.3.3) consisted of both the background noise and radiated noise from the

muffler shell. The background noise could be due to the leakage from the joints,

reflection from the hard floor and wall, leaked noise from the speaker etc. To quantify the

radiated noise accurately the influence of background noise should be established. The

background noise was measured by isolating the contribution of radiated noise from the

measurement. The source of the radiated noise is the muffler shell. In order to isolate the

radiated noise, the muffler shell was wrapped with a thick sound insulating sheet. The

measurement set up to characterize the background noise is shown in figure 6.10. The

results from the measurement are shown in figure 6.11.

39
 Measurement taken at 0.625 m

adapter

Connection pipe Anechoic termination

Surface wrapped with sound absorbing material
Speaker

Figure 6.10: Test set up to measure background noise

40
 90
total sound (radiated+noise)
80
Background-noise

70

60

50
Pressure (dB)

40

30
Radiated noise very
low till 640 Hz
20

10

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 6.11: Results from background noise measurement

The measurement shows that the radiated noise between 50-640 Hz is very low and most

of the sound power in the 50-640 Hz range is due to background noise. Beyond 640 Hz

radiated noise is significantly higher as compared to background noise. Based on the data

shown, radiated noise in the 50-640 Hz region was set to zero and no correction was

applied to the measured SPL above 640 Hz. The corrected noise outside the cavity after

applying the corrections is shown in figure 6.12.

6.3.5 TL from test

Once the acoustic pressure values inside and outside the cavity were established,

the TL was calculated using equation 6.2. The calculated TL for the cylindrical muffler

with 0.6 mm thickness, 0.5m length and 0.25 m diameter is shown in figure 6.13.

41
 160

average-pressure-inside
140
corrected-average-pressure-outside
120

100
Pressure (dB)

80

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.12: Corrected sound pressure outside the cavity

160

140

TL-measured
120

100
Pressure (dB)

80

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 6.13: TL from test

42
6.4 TL calculation using BEA

TL for the muffler with 0.6 mm thickness, 0.5m length and 0.25 m diameter was

calculated in SYSNOISE following a similar procedure that was used in experimental

method . The outline of the procedure is listed below:

1. Calculate the average sound pressure inside cavity at same locations used

in the expermental method by follwing the procedure outlined in step-1 of

section 6.1.

2. Calculate the average sound pressure outside cavity at same locations

used in the expermental method by following the procedure outlined in

step-2 of section 6.1.

3. From the the average pressure inside and outside the cavity calculate TL

using equation 6.2.

The BE model used to calculate the sound pressure inside the cavity is shown in figure

6.14. The pressure values at three locations inside the cavity and the average value

obtained using BEA is shown figure 6.15.

Point-3 Point-2
adapter Point-1
source

Connector

Figure 6.14: BE model used to calculate pressure inside the cavity

43
 160

140
Point-2 :Nodal point for
acoustic mode at 676 Hz
120

100

80
Pressure (dB)

60
Point-3 :Nodal point for BEA-point-1
acoustic mode at 344 Hz BEA-point-2
40
BEA-point-3
BEA-Average
20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 6.15: Calculated value and average pressure inside the cavity from BEA

The BE model used to calculate the sound pressure outside the cavity is shown in figure

6.16. The calculated pressure values at four diametrically opposite locations were

averaged (each at 0.625 meter from muffler center). The average pressure value outside

(radiated noise) along with the average value inside the muffler, acoustic modes and

structural modes are shown in figure 6.17. From the pressure value inside and outside the

cavity, the TL for the muffler shell was calculated. The TL of the 0.6 mm shell obtained

from BEA is plotted in figure 6.18.

44
 Four locations each at
0.625 m from muffler
center

Figure 6.16: BE model used to calculate pressure outside the cavity

160
BEA-averaged-inside
Radiated noise averaged-outside
140
acoustic-modes
structural modes
120

100

80
Pressure (dB)

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.17: Average pressure value outside the cavity calculated using BEA

45
 160

140

TL-BEA
120

100
Pressure (dB)

80

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.18: TL-calculated using BEA

6.5 Comparison of results between experiment, BEA and analytical

formulation

The TL curve obtained from numerical simulation and analytical method were

compared with the TL curve from experiment to study the correlation. The average

pressure inside the cavity obtained from BEA and experiment are plotted in figure 6.19.

(Analytical formulation does not give the pressure distribution inside the cavity). Figure

6.20 shows the comparison of TL’s from numerical simulation, analytical method and

experiment. For the analytical calculation of TL, an incidence angle of 45 deg was used.

46
 160

140

120

100
Pressure (dB)

80

60
Measurement-averaged
40 BEA-averaged
acoustic modes
20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.19: Comparison of average acoustic pressure inside the cavity

160

140
Measurement
BEA
120
Analytical-Matlab

100
Pressure (dB)

80

60

40

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 6.20: Comparison of TL’s

47
6.5.1 Discussion of results

1) Numerical simulation and experiment

The results show a very good agreement between the numerical simulation and

experiment for sound pressure levels inside the cavity. TL comparison shows

reasonable agreement. Numerical simulation was able to accurately capture the

acoustic modes and structural modes of the test specimen. The differences in TL

could be due to the difference in the boundary conditions and surrounding

environment during experiment and simulation. BEA assumes perfectly anechoic

surroundings. In actual experiment reflected waves exists because of the presence

of hard floor and sidewalls. In experiment an anechoic termination was used. In

numerical simulation open termination was used. Over all it could be concluded

that numerical simulation predicts the acoustic behavior of the prototype model

accurately.

2) Analytical formulation and experiment

TL’s from analytical method and experiment compares fairly well in the

frequency range above 650 Hz. A relatively large difference in the low frequency

range could be attributed to the inherent assumptions involved in the formulation

of analytical model. Analytical model assumes infinite length and effect of this

boundary condition on the TL calculation of experimental model with finite

length is more significant in the lower modes which have longer wavelength. The

plane wave incidence condition as well as the modeling of the shell motion only

by bending wave also might have contributed to the differences between

experimental results and analytical model.

48
7. Study on the sensitivity of design parameters on TL

Different design parameters of automotive mufflers like shell thickness and shape

of the cross-section were studied to determine its sensitivity on TL calculations. Both

analytical calculation and numerical simulations were done by keeping all the other

parameters constant. For all the comparisons, the prototype model used in the previous

correlation study was used as baseline.

7.1 Effect of thickness

In order to quantify the effect of thickness on TL, muffler shell with three

different thickness, 0.6 mm (baseline model), 1.2 mm and 1.8 mm were analyzed. All

other parameters were kept the same as used in the correlation study. The 1.2/1.8 mm thk.

model is shown in figure 7.1. The first twelve structural modes of the 1.2 mm shell and

1.8 mm shell are listed in table 7.1. Figure 7.2 shows the comparison of radiated noise

obtained from numerical simulation and figure 7.3 shows the comparison of TL obtained

from numerical simulation. Figure 7.4 shows the comparison of TL obtained from

analytical method using an incidence angle of 45 deg.

1.2/1.8 mm thk.

Figure 7.1: Model with 1.2/1.8 mm shell

49
 Modes for 1.2 mm shell Modes for 1.8 mm shell
Mode # Frequency Description Mode # Frequency Description
Circumferential mode with Circumferential mode with
1 412.4 four lobes 1 486.9 three lobes
2 412.5 Repeated root of mode #1 2 487.3 Repeated root of mode #1
Circumferential mode with Circumferential mode with
3 461.0 three lobes 3 521.3 four lobes
4 461.7 Repeated root of mode #3 4 521.4 Repeated root of mode #3
Circumferential mode with Circumferential mode with
5 518.4 five lobes 5 733.2 five lobes
6 518.9 Repeated root of mode #5 6 733.4 Repeated root of mode #5
Circumferential mode with Circumferential mode with
7 705.9 six lobes 7 784.7 two lobes
8 706.2 Repeated root of mode #7 8 785.1 Repeated root of mode #7
Circumferential mode with Circumferential mode with
9 783.9 two lobes 9 1033.3 six lobes
10 784.4 Repated root of mode #9 10 1033.4 Repated root of mode #9
Circumferential mode with Circumferential mode with
11 937.3 seven lobes 11 1383.2 seven lobes
12 937.4 Repated root of mode #11 12 1383.4 Repated root of mode #11

Table 7.1: structural modes for the 1.2/1.8 mm muffler shell

70
BEA-0.6 mm shell
BEA-1.2 mm shell
60 BEA-1.8 mm shell

50

40
Pressure (dB)

30

20

10

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 7.2: Comparison of radiated noise for different thickness

50
 140

BEA TL 0.6 mm shell

120
BEA TL 1.2 mm shell
∆=3.9 dB BEA TL 1.8 mm shell
100

80
Pressure (dB)

∆=3.4 dB
∆=6.0 dB
60

40
∆=5.5 dB

20

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 7.3: Comparison of TL for different thickness from BEA

70
∆=3.7 dB

60

50
∆=6 dB

40

30

20

Analytical TL for 0.6 mm shell

10 Analytical TL for 1.2 mm shell
Analytical TL for 1.8 mm shell
0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Figure 7.4: Comparison of TL for different thickness from analytical calculation

51
7.3 Effect of geometry

In order to quantify the effect of shape on TL, muffler shell with three different

cross- sections, cylindrical (baseline), elliptical and stamped cross-section were analyzed.

All the three models had the same cross-sectional area (0.049m2). All remaining

parameters were kept the same as used in the correlation study. The model with elliptical

cross section and stamped cross-section is shown in figure 7.5. The first twelve structural

modes of the model with elliptical cross-section and stamped cross-section are listed in

table 7.2. Figure 7.6 shows the comparison of radiated noise obtained from numerical

simulation and figure 7.7 shows the comparison of TL obtained from numerical

simulation.

Figure 7.5: Model with elliptical/stamped cross-section

52
 Modes for 0.6 mm thk shell with Modes for 0.6 mm thk shell with
elliptical cross-section stamped cross-section
Modes # Frequency (HZ) Modes # Frequency (HZ)
1 172.6 1 65.8
2 176.5 2 68.2
3 257.4 3 90.4
4 261.3 4 91.0
5 299.3 5 124.2
6 306.1 6 124.6
7 385.8 7 144.8
8 397.5 8 167.3
9 398.9 9 167.6
10 415.3 10 173.9
11 418.3 11 174.5
12 420.5 12 198.5

Table 7.2: Structural modes for the elliptical/stamped muffler shell

90

80

70

60
Pressure (dB)

50

40

30

20 BEA-cylindrical section
BEA-elliptical section
10 BEA-stamped section

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250
Frequency (Hz)

Figure 7.5: Comparison of radiated noise for different cross-sections

53
 120

110 BEA TL cylindrical section

100 BEA TL elliptical section
BEA TL stamped section
90

80

70
Pressure (dB)

60

50

40

30

20

10

0
50 150 250 350 450 550 650 750 850 950 1050 1150 1250

Frequency (Hz)

Figure 7.6: Comparison of TL for different cross-sections

7.4 Discussions on design sensitivity study

From the study the following could be concluded

o Both the numerical simulation and analytical simulation predicts a TL increase of

about 6 dB as the thickness is doubled from 0.6 mm to 1.2 mm. This is well

anticipated and in accordance with mass law which states that the transmission

loss through a panel increases by 6 dB per doubling of panel thickness [20].

Numerical simulation as well as the analytical model predicts about 3.6 dB

increase in TL as the tickness is increased from 1.2 mm to 1.8 mm.

o It is shown that TL is very sensitive to the design of the muffler cross-section and

the muffler with the stamped cross-section has the least TL. It is also observed

54
 that both muffler with stamped cross-section and elliptical cross-section has high

modal density in the low frequency range as compared to the muffler with circular

cross-section. Both the decrease in TL and increase in modal density could be

attributed to the reduction in stiffness as the muffler section deviates away from

the circular shape.

o Due the inherent assumptions involved in the formulation of the problem, the

analytical method may not be able to predict the acoustic and the structural modes

correctly. But the relative difference in TL between two designs predicted by the

analytical method matched well with the numerical prediction. So the MATLAB

based analytical program could be very effectively used to predict the relative

changes in TL between different designs of cylindrical shape.

55
 8. Conclusions and recommendations for future work

Basically this work consisted of three major parts. The conclusions from each part

are summarized below:

In the first part, analytical formulation of TL for a single walled cylindrical shell

done by Lee and Kim was reviewed. A simple correction to the formulation was

identified and proposed corrections were presented. The results showed that the

modification affected the stiffness controlled low frequency range of the transmission

loss curve.

In the second part, the TL of a typical cavity used in automotive industry was

established experimentally. TL for the same cavity was calculated by numerical method

(Boundary Element Analysis) using the commercially available SYSNOISE software.

Results demonstrated that the numerical simulation was able to identify all the acoustic

modes and the structural modes accurately. The TL curves from the two methods showed

reasonable agreement. In this step the validity of the numerical method to predict the TL

was established.

In the third part, the effect of various design parameters on TL was studied. Both

numerical method and analytical method was used to calculate the TL for different

designs of cylindrical shape. The relative difference in TL between the various

cylindrical designs predicted by the analytical method and numerical method agreed very

56
well. This study demonstrated that analytical method could be used very effectively for

relative comparisons of different designs of cylindrical shape. Numerical simulation

showed that the shape of the cross-section of the muffler cavity is a critical design

parameter and TL is very sensitive to the design of the cross-section. Among the various

cross-sections that are typically used in the automotive industry (i.e. circular, elliptical

and stamped), the muffler with stamped cross-section had the highest modal density in

low frequency range and hence the least TL.

The most important contribution of this work is that it provided baseline data to

understand the accuracy of analytical formulation proposed by Lee and Kim and

numerical simulation using BEA. This work shows that the analytical method could be

used very effectively for relatively comparing different designs of cylindrical shape. This

work also shows that numerical simulation could be used to accurately model TL

predictions.

The comparisons showed that the analytical model, although has a large

discrepancy in the low frequency range, predicts the TL in the high frequency range, 650

Hz or higher. Also, the model can be used for parameters studies which the model

predicts the influence of design parameter variations quite accurately. The analytical

approach will be useful as a first-cut design tool. The analytical models of double shells

with and without porous core and stiffened shells that Lee and Kim developed will be

especially useful because compatible numerical analyses are very difficult.

57
 Future work could be found in developing a method to predict the TL for dual

shelled structures using numerical simulation. Most commercial codes impose the

limitation that the two sheets in the numerical model have to be separated by at least one

element edge length. This could lead to physical gap of about 4 mm between the two

shells. But in actual structures the two sheets are spot welded and the gap is less than

1mm. So the conventional methods couldn’t be used to model dual shelled structures in

BEA. Lee and Kim identified that most of the sound attenuation in dual shells occurs by

damping. Considering the facts above it could be recommended to develop a damping

factor that could be applied on a single shell to account for the effect of dual shells.

58
References

[1] M. L. Munjal, “Acoustics of Ducts and Mufflers with Application to Exhaust and

Ventilation System Design”, New York: John Wiley and Sons, 1987.

[2] Kexin Hu, Chul Lee, Emile Homsi and David Moenssen, “Acoustics Modeling and

Radiated Noise Prediction for Plastic Air-Intake Manifolds”, SAE 2003011448.

[3] Desmet W. and Sas P. , “Vibro-acoustic analysis procedures for the evaluation of

the sound insulation characteristics of agricultural machinery cabins”, Proceedings

for 25th ISMA conference, 2000, 1587-1598

[4] Joon Huyn Lee and J. Kim, “Study on sound transmission characteristics of a

cylindrical shell using analytical and experimental models”, Journal of applied

acoustics, 2003, 64, 611-622

[5] J.H. Lee and J. Kim, “Analysis and measurement of sound transmission through

double walled cylindrical shell”, Journal of sound and vibration, (2002), 251(4),

631-649

[6] L. E. Kinsler, A. R. Frey et al. “Fundamentals of Acoustics”, New York: John

Wiley and Sons, 2000.

[7] Jay Kim, University of Cincinnati, Acoustics I and Acoustics II class notes.

[8] Joon Hyun Lee, “ Development of new technique for damping identification and

sound transmission analysis through various structures”, Ph.D. dissertation,

University of Cincinnati, 2001.

[9] D.W. Herrin, Z. Tao, J. Liu, and A.F. Seybert, “ Using Boundary Element Analysis

to Analyze Multi-Component Exhaust Systems”, SAE 2007012182.

59
[10] Glen C. Steyer and Brian Campbell, “Simulation of radiated noise from a

transmission side cover ”, SAE 972022

[11] J.D. Irwin and E.R. Graph, “Industrial Noise and Vibration Control”, Englewood

Cliffs, NJ: Prentice Hall, 1980.

[12] MSC NASTRAN 2001 User’s Manual.

[13] SYSNOISE Rev5.5 User’s Manual.

[14] Joon Huyn Lee and J. Kim, “Simplified method to solve sound transmission through

structures lined with elastic porous material”, Journal of applied acoustics, 2003,

110, 2282-2294.

[15] J.H. Lee and J. Kim, “Sound transmission through periodically stiffened cylindrical

shells”, Journal of sound and vibration, (2002), 251(3), 431-456.

[16] Y. Charles Lu and Karl D’Souza “Acoustic analysis of isolated engine valve covers”

SAE 03NVC-77.

[17] Richard Eberhart, Fred W. Catron, Allen Fagerlund and Denis G. Karczub, “ Piping

noise transmission loss calculations using finite element analysis” , Journal of

applied acoustics, 2005, 110, 1918-1929.

[18] Christopher E. Shaw, David J. Moenssen and John D. Kostum,“ A Correlation

Study of Computational Techniques to Model Air Induction System Response

Including BEM, FEM and 1D Methods ”, SAE 200301644.

[19] Sang Huyn Jee, Jong Cheol Yi and Jae Keon Park, “The comparison of the BEM

and FEM techniques for the interior noise analysis of passenger car” ,Proceedings

for the 2nd MSC Worldwide Automotive Conference, 2000

60
[20] L.L.Beranek and I.L.Ver, “Noise and Vibration Control Engineering Engineering :

Principles and Applications”, 2005, New Jersey : John Wiley & Sons Inc,

[21] Altair Hyperworks Verion 7 User’s Manual.

[22] James M. Gere and Stephen P. Timoshenko, “Mechancis of Materials ”, Boston,

MA, : PWS Publishers, 2000.

61