John

DESIGN & ANALYSIS OF ACOUSTICALLY
IMPROVED VEHICLE FLOOR CARPETS
A thesis submitted in fulfillment of the requirements for

the degree of Master of Engineering
THOMAS K JOHN
School of Aerospace, Mechanical and Manufacturing Engineering

RMIT University
December 2008
Dedicated to my parents, Prof. K C John and Treesa John, and all my teachers, without
whose selfless support and love every dream of mine would have been mere dreams.
II
ACKNOWLEDGEMENTS
Throughout my research program, I have gained valuable experience and advice from
many people I have interacted with and at this juncture I feel obliged to thank them all for
being so helpful to me. First and foremost, I would like to express my sincere gratitude to my
supervisor, Dr. Xu Wang from RMIT University, for his continuous support, valuable
guidance and fruitful discussions in every aspect of my research work. Without Xu’s
guidance, I wonder if I could have able to finish the research. Also I would like to thank my
second supervisor, Dr. Simon Watkins from RMIT University, an expert in the field of Fluid
Dynamics and Dr. John Davy from RMIT University for being my Acoustics teacher and
guide me develop the panel radiation prediction model. I would also like to use this
opportunity to thank the Futuris and AutoCRC team, especially Jason Miller, Garry White
and Kate Neely, for funding and executing the research project.
I would like to dedicate this thesis to my parents Prof. K C John and Treesa John.
Most importantly I am highly memorable of my grandfather, late Mr. K C Chummar who
was my primary source of inspiration and confidence in acoustics; my elder brother Simon K
John and sister Jesmin, engineers by profession, for their assurance and support leading me
to my overseas studies, my younger brother Jacob, my uncle Davis & family and all JY
friends, who have always shown love and encouragement.
A very special thanks to Dr Edsil Dilla, who can be very well be described as a
friend, philosopher and guide, for he is the first person whom I consult about both my
academic and personal matters. I would also like to acknowledge the RMIT technical team
members, particularly Peter Tkatchyk and Brett Vincent, Lina Bubic and other colleagues in
the RMIT research office for their assistance whenever required.
III
DECLARATION
I, Thomas K John, hereby submit the thesis entitled “Design and analysis of
Acoustically improved vehicle floor carpets” for the degree of Masters by Research and
certify that except where due acknowledgement has been made, the work is that of the author
alone; the work has not been submitted previously, in whole or in part, for any other
academic award and that the content of the thesis is the result of work that has been carried
out since the official commencement of the program.
Thomas K John
12th December 2008
IV
TABLE OF CONTENTS
1 Introduction
1 GENERAL DESCRIPTION ........................................................................................... 1

2 SCOPE AND OBJECTIVE............................................................................................. 2
3 LITERATURE REVIEW................................................................................................ 3
3.1 CARPET MATERIAL AND ACOUSTICS ............................................................................ 3
3.2 ACOUSTIC SIMULATION AND MEASUREMENT .............................................................. 7
4 RESEARCH QUESTIONS ........................................................................................... 15
5 OUTCOMES .................................................................................................................. 16
6 THESIS STRUCTURE.................................................................................................. 16
2 Benchmark analysis and target setting for vehicle cabin

noise
1 INTRODUCTION ..........................................................................................................18
2 VEHICLE NOISE ANALYSIS.....................................................................................19
2.1 HYUNDAI SONATA VEHICLE CABIN NOISE TEST ..........................................................20
2.1.1 Test procedure.....................................................................................................23
2.2 HYUNDAI SONATA - TEST RESULTS AND DISCUSSION .................................................23
2.3 AUDI A6 VEHICLE CABIN NOISE TEST ..........................................................................27
2.3.1 Test procedure.....................................................................................................28
2.4 AUDI A6 – COMPARATIVE TEST RESULTS AND DISCUSSION ........................................29
2.5 PSYCHO-ACOUSTIC AND SOUND QUALITY EVALUATION..............................................34
3 TARGET DEFINITION................................................................................................40
3.1 THE CARPET SYSTEM ..................................................................................................40
3.2 CARPET DESIGN OBJECTIVES AND CONSTRAINTS.........................................................40
3.2.1 Objectives - Physical Characteristics .................................................................40
3.2.2 Objectives - Functional Requirements ................................................................43
3.3 DESIGN VALIDATION...................................................................................................46
4 CONCLUSIONS.............................................................................................................47
APPENDIX ............................................................................................................................48
V
3 Forced vibration analysis of panel systems and sound
radiation directivity model
1 INTRODUCTION ..........................................................................................................51
2 SIMPLE HARMONIC ANALYSIS OF VEHICLE FLOOR TRIM PANEL..........52
3 FINITE SIZED PANEL RADIATION – THEORY ...................................................56
4 DERIVATION OF THE PROPOSED ANGULAR WEIGTHING FUNCTION
W(φ)........................................................................................................................................59
4.1 RADIATION EFFICIENCY FUNCTION DETAILS ...............................................................62
5 COMPARISON WITH PUBLISHED RESULTS ON DIRECTIVITY ...................63
6 CONCLUSIONS.............................................................................................................65
4 Design concepts for reducing noise from floor panel

sound radiation
1 INTRODUCTION ..........................................................................................................66
2 CURRENT VEHICLE CARPET SYSTEM................................................................67
2.1 FUNCTIONAL ANALYSIS ..............................................................................................68
2.2 BASELINE AND TARGETS FOR THE CARPET DESIGN......................................................70
3 THEORETICAL BASE FOR FUNCTIONAL ANALYSIS ......................................71
3.1 LAYERED UP MATERIAL INTERACTION ........................................................................73
4 PROPOSED CONCEPT DESIGNS .............................................................................73
4.1 MATERIAL SELECTION ........................................................................................76
4.2 EVALUATION TEST SETUP ...........................................................................................77
5 PRELIMINARY EVALUATION RESULTS .............................................................78
5.1 RESULT DISCUSSION ...................................................................................................83
6 CONCLUSIONS.............................................................................................................83
5 Defining the design process and acoustic tools
1 INTRODUCTION ..........................................................................................................84
2 REVIEW OF SAC MEASUREMENT METHODS ...................................................85
3 DESIGN CONCEPT ......................................................................................................90
3.1 THEORETICAL BASIS ...................................................................................................90
VI
4 ACOUSTICS SIMULATION AND VALIDATION TOOLS....................................93
4.1 SCOPE OF VEHICLE CARPET ACOUSTIC SIMULATION ....................................................93
4.2 SIMULATION PROCESS CYCLE .....................................................................................94
4.3 REVIEW OF ACOUSTIC AND VIBRATION ANALYSIS SIMULATION TOOLS .......................95
5 CONCLUSIONS...……………………………………………………………….....98
6 Material simulation, validation and theory
1 INTRODUCTION ..........................................................................................................99
2 MATERIAL LEVEL SEA SIMULATION PROCEDURE .....................................100
2.1 INPUT PARAMETERS FOR SIMULATION.......................................................................100
2.2 SIMULATION .............................................................................................................101
3 VALIDATION PROCEDURE....................................................................................102
3.1 ABSORPTION COEFFICIENT MEASUREMENT USING IMPEDANCE TUBE ........................102
3.2 NORMAL TO RANDOM INCIDENCE ABSORPTION CONVERSION ...................................102
4 RESULTS AND DISCUSSION...................................................................................103
4.1 NORMAL INCIDENCE VS. RANDOM INCIDENCE...........................................................103
4.2 SIMULATION VS. MEASURED RESULTS .......................................................................105
5 CONCLUSIONS...………………………………………………………………...108
7 Virtual modeling and validation, at component and

vehicle levels
1 INTRODUCTION ........................................................................................................109
2 COMPONENT LEVEL (ALPHA CABIN) MODELLING .....................................110
2.1 CAVITY MODELLING .................................................................................................112
2.2 NOISE SOURCES ........................................................................................................113
2.3 VIRTUAL MODELLING TEST RESULTS ........................................................................114
2.4 RESULT DISCUSSION .................................................................................................115
3 VEHICLE LEVEL MODELLING.............................................................................116
3.1 ENGINE COMPARTMENT ............................................................................................116
3.2 PASSENGER CABIN ....................................................................................................117
3.3 VEHICLE REAR COMPARTMENT .................................................................................118
3.4 ACOUSTIC CAVITY MODELLING.................................................................................120
3.4.1 Vehicle interior cavities ....................................................................................120
3.4.2 Exterior cavities ................................................................................................121
3.5 SOUND PACKAGES ....................................................................................................122
3.6 NOISE SOURCES ........................................................................................................125
3.7 SEA MODEL VALIDATION USING TRANSFER FUNCTION METHOD ...............................125
3.8 EVALUATION OF CARPETS USING VEHICLE SEA MODELLING ...................................128
VII
3.8.1 Thermo grams of engine and tyre noise paths ..................................................130
4 CONCLUSIONS...........................................................................................................132
8 Proto-type validation and evaluation of design concept
1 INTRODUCTION ........................................................................................................133
2 TEST SAMPLES..........................................................................................................134
2.1 STANDARD MAT (REFERENCE) .................................................................................134
2.2 PROTOTYPE MAT (CONCEPT 3 DESIGN) ....................................................................134
2.2.1 C3-Special_mat_Lab-made...............................................................................134
2.2.2 C3-Mat_Prototype ............................................................................................135
2.3 CONSTRAINTS AND LIMITATIONS ..............................................................................135
2.3.1 Tufting ...............................................................................................................135
2.3.2 Lamination ........................................................................................................136
3 PROTOTYPE TESTING ............................................................................................137
3.1 MATERIAL ACOUSTIC PROPERTY TESTING ...............................................................137
3.1.1 Impedance Tube Test ........................................................................................137
3.1.2 RMIT Alpha Cabin Test ....................................................................................138
3.1.3 Vehicle testing using in-situ methods................................................................139
4 TEST RESULTS ..........................................................................................................141
4.1 IMPEDANCE TUBE TESTS ..........................................................................................141
4.2 ALPHA CABIN TESTS ................................................................................................142
4.3 IN-SITU VEHICLE TESTS ............................................................................................143
4.3.1 Stationary Conditions .......................................................................................143
4.3.2 On-Road Constant Speed test results................................................................146
4.3.3 2nd Gear Slow Acceleration Test.......................................................................148
5 CONCLUSIONS...........................................................................................................149
APPENDIX .........................................................................................................................150
9 Design optimization (Taguchi method based), at

material, component and vehicle levels
1 ABSTRACT & INTRODUCTION.............................................................................154
1.1 OPTIMIZATION REQUIREMENT & TECHNIQUES ..........................................................156
1.2 CONCEPT 3 CONSTRAINTS AND PERFORMANCE TARGETS ..........................................157
2 TAGUCHI METHOD - THEORY.............................................................................158
3 MATERIAL LEVEL TAGUCHI METHOD OPTIMIZATION RESULTS.........161
3.1 ON/OFF TAGUCHI DE-COUPLING ANALYSIS .............................................................161
3.1.1 Cabin side absorption: ON/OFF Taguchi tables..............................................162
3.1.2 Sound transmission loss: ON/OFF Taguchi tables...........................................164
3.1.3 Parameter elimination discussion.....................................................................166
VIII
3.2 DETAILED TAGUCHI ANALYSIS .................................................................................166
3.2.1 Taguchi analysis: Cabin side absorption..........................................................167
3.2.2 Taguchi analysis: Transmission loss ................................................................179
3.2.3 Summary of detailed Taguchi analysis .............................................................174
4 ALPHA CABIN BASED CARPET SECTION TAGUCHI ANALYSIS................175
4.1 CARPET SECTION TAGUCHI ANALYSIS PROCEDURE ...................................................175
4.1.1 Carpet ON/OFF Taguchi sections ....................................................................176
4.2 ALPHA CABIN (COMPONENT LEVEL) TAGUCHI ANALYSIS RESULTS ..........................177
4.2.1 Result discussion ...............................................................................................179
5 VEHICLE SOUND PACKAGE TAGUCHI ANALYSIS ........................................180
5.1 SOUND PACKAGE TAGUCHI SECTIONS .......................................................................181
5.2 TAGUCHI ANALYSIS RESULTS ....................................................................................182
5.2.2 Taguchi analysis including window screens.....................................................187
5.3 VEHICLE CARPET SECTION TAGUCHI ANALYSIS ........................................................191
5.3.1 Taguchi analysis results....................................................................................192
6 CONCLUSIONS...........................................................................................................195
10 Conclusions and recommendations for further research
1 CONCLUSIONS...........................................................................................................196
1.1 THIS THESIS ..............................................................................................................196
1.2 RESEARCH QUESTIONS ..............................................................................................198
1.2.1 What is the role of vehicle floor carpet in eliminating engine and tyre noise and
how good is the current design in achieving it?.............................................................199
1.2.2 Is virtual modelling and validation an effective tool for designing and evaluating
a vehicle carpet acoustically at vehicle level? ...............................................................199
1.2.3 What could be the theoretical limits of noise reduction in the vehicle cabin by
acoustically improved vehicle carpets and where are we now? ....................................199
1.2.4 Are there simple and efficient methods for in-situ measuring the acoustic
performance of carpet design, non-destructively? .........................................................200
1.2.5 What are the future possibilities in introducing tuned acoustic layers for vehicle
carpets? ..........................................................................................................................200
2 RECOMMENDATIONS FOR FURTHER RESEARCH........................................200
2.1 SCOPE FOR AN IMPROVED DESIGN .............................................................................201
2.2 CONCEPT 4 CARPET DESIGN CONSTRAINTS AND TARGETS .........................................201
3 DOUBLE PANEL AND AIR-GAP THEORY ..........................................................202
4 RESULT ANALYSIS...................................................................................................203
5 EVALUATING AGAINST CONSTRAINTS: FOAM VS HONEY COMB ..........206
6 SPECIAL TUNED RESONANT ABSORBER LAYERS ........................................208
References……………………………………………………………………....... 213
IX
LIST OF ABBREVIATIONS
SAC Sound Absorption Coefficient

TL Transmission Loss
IL Insertion Loss
dB Decibel
dB(A) Decibel, A-weighted
B&K Bruel & Kjaer
SEA Statistical Energy Analysis
SPL Sound Pressure Level
HMS Head Measurement System
RPM Rotations Per Minute
AC Air Conditioner
Hz Hertz
AI Articulation Index
GSM Gram per Square Meter
T60 Reverberation Time
AFR Air Flow Resistance
NCT Noise Control Treatment
X
LIST OF FIGURES
Chapter 1
Figure 1.1– Acoustic and thermal package of a typical car (Rieter automotive) ………… ……….. 2
Chapter 2
Figure 2.1 – cabin noise mapping index for microphone placement ………………………………… 19
Figure 2.2 – cabin noise mapping index for microphone placement ………………………………… 21
Figure 2.3– B & K Pulse data acquisition system ……………………………………………………… 21
Figure 2.4– Microphones were placed in the front foot well area …………………………………… 22
Figure 2.5– A microphone was located at the Driver’s left Ear (DE) ……………………………….. 22
Figure 2.6– Hyundai Sonata testing vehicle was mounted on the chassis dynamometer …………. 23
Figure 2.7 – Sound pressure level water fall spectrum at DE for the 2nd gear slow
acceleration tests (a) on the road and (b) the chassis dynamometer …………………………………. 24
Figure 2.8- The 2nd order sound pressure level at DE for the 2nd gear slow acceleration
on the road (a) and the chassis dynamometer (b)………………………………………………………. 24
Figure 2.9– Sound pressure level at DE for constant speeds of 40, 60 and 80 km/h on
the road (a) and the chassis dynamometer (b)…………………………………………………………… 25
Figure 2.10 – Sound Pressure Auto-spectrum at DE for Idle in Drive_AC On & Off …………….. 25
Figure 2.11 – The SPL distribution on the car carpet; 2nd gear slow acceleration on-road test…. 26
Figure 2.12– The SPL distribution on the car carpet; 40 kph constant speed on-road test ………. 27
Figure 2.13 – The Audi A6 test vehicle …………………………………………………………………... 28
Figure 2.14– Sound pressure level waterfall spectrum at DE for the 2nd gear acceleration
on the road for (a) Audi A6 and (b) Hyundai Sonata.………………………………………………….. 29
Figure 2.15 – The 2nd order sound pressure level at DE for the 2nd gear slow acceleration
on the road (a) Audi A6 and (b) Hyundai Sonata ………………………………………………………. 30
Figure 2.16– Sound pressure level at DE for constant speeds of 40, 60, 80 and 100 km/h
on-road test for Audi……………………………………………………………………………….………… 31
Figure 2.17 – Sound Pressure Auto-spectrum at DE for Idle in drive and A/C off, with
and without mats …………………………………………………………………………………………….. 32
Figure 2.18 – Comparative low frequency (69 Hz) SPL distribution in the car carpet
during 80 kph constant speed on-road test ………………………………………………………………. 32
Figure 2.19– Comparative high frequency (5163 Hz) SPL distribution on the car
carpet during 80 kph constant speed on-road test……………………………………………………….. 33
Figure 2.20 – Comparative SPL distribution on the car carpet during slow acceleration
on-road test just after starting (low rpm; 840 rpm ≈ 28Hz)…………………………………………… 33
Figure 2.21 – Comparative SPL distribution on the car carpet during slow acceleration
on-road test at high engine rpm (3300 rpm ≈ 110Hz) …………………………………………………. 34
Figure 2.22 – Head Acoustics HMS III …………………………………………………………………… 35
Figure 2.23 – The 6th Octave Band Spectra ……………………………………………………………… 36
Figure 2.24 – The 6th Octave Band Spectra ……………………………………………………………… 36
Figure 2.25 – The Articulation Index comparison ………………………………………………………. 37
Figure 2.26 – The Loudness Index comparison ………………………………………………………….. 37
Figure 2.27– The Fluctuation Strength comparison …………………………………………………….. 38
Figure 2.28 – The Roughness comparison ……………………………………………………………….. 38
Figure 2.29– The Sharpness comparison ………………………………………………………………… 39
Figure 2.30 – The tonality comparison …………………………..………………………………………. 39
Figure 2.31 – Random incidence sound absorption coefficient design target ………………………. 44
Figure 2.32 – Sound transmission loss/insert loss design target ……………………………………… 44
Chapter 3
Figure 3.1– Sound transmission loss/insert loss design target ………………………………………… 52
Figure 3.2– Simple Harmonic Motion (SHM) oscillator equivalence ……………………………….. 53
Figure 3.3– Simple Harmonic Motion (SHM) oscillator equivalence ……………………………….. 55
Figure 3.4– Transmission Loss results of the current carpet design variants ………………………. 55
Figure 3.5– Panel or opening radiating sound due to forced excitation on the incident side…….. 56
Figure 3.6– Sound incident at an angle φ to the normal to a panel and radiated
at an angle of θ ……………………………………………………………………………………………… 58
Figure 3.7– Calculating the number of wall reflections before sound hits the panel
or opening at an angle of φ to the normal ……………………………………………………………… 61
XI
Figure 3.8 – The sound level at 90° relative to that at 0° as a function of Strouhal
number for 6 mm thick glass installed in the wall of a room…………………………………………… 64
Figure 3.9 – The sound level at 90° relative to that at 0° as a function of Strouhal
number for an unbaffled duct end opening………………………………………………………………. 64
Chapter 4
Figure 4.1– Current vehicle carpet design ………………………………………………………………. 67
Figure 4.2– Normal sound absorption coefficient of various carpet tufting (L to Q) ………………. 68
Figure 4.3 – Normal incidence Transmission Loss coefficient of various heavy layers …………… 69
Figure 4.4 – Absorption characteristics of a typical de-coupler layer ………………………………. 69
Figure 4.5– Absorption characteristics of a typical de-coupler layer ……………………………….. 70
Figure 4.6– Transmission loss of various carpet designs, with maximum and minimum defined … 71
Figure 4.7– Material acoustic properties in relevance order for modelling ………………………… 72
Figure 4.8 – Concept 1 carpet design …………………………………………………………………….. 74
Figure 4.9– Concept 2 carpet design ……………………………………………………………………… 74
Figure 4.10– Concept 2 carpet design ……………………………………………………………………. 75
Figure 4.11 – Concept 4 carpet design …………………………………………………………………… 75
Figure 4.12– Material layer selection matrix ……………………………………………………………. 77
Figure 4.13 – Cabin side absorption characteristics of Concept 1 carpet design ………………….. 79
Figure 4.14– Transmission loss characteristics of Concept 1 carpet design ………………………… 79
Figure 4.15 – Cabin side absorption characteristics of Concept 2 carpet design ………………….. 80
Figure 4.16– Transmission loss characteristics of Concept 2 carpet design ……………………….. 80
Figure 4.17– Cabin side absorption characteristics of Concept 3 carpet design …………………… 81
Figure 4.19– Cabin side absorption characteristics of Concept 4 carpet design …………………… 82
Chapter 5
Figure 5.1 – US Patent 4732039 ………………………………………………………………………….. 87
Figure 5.2 – US Patent 4537630 ………………………………………………………………………….. 87
Figure 5.3– US Patent 6134968 …………………………………………………………………………… 88
Figure 5.4– Set-up for transfer function measurement in front of an absorbing surface ………….. 88
Figure 5.5– Set-up for transfer function measurement in front of an absorbing surface ………….. 89
Figure 5.6 – Diffuse field measurement concept (B&K) ……………………………………………….. 90
Figure 5.7– SAC device working principle ………………………………………………………………. 94
Figure 5.8– Simulation process cycle …………………………………………………………………….. 95
Figure 5.9– Simulation process cycle …………………………………………………………………….. 96
Figure 5.10– BEM modelling vehicle interior (Image courtesy – LMS & ESI group) ……………… 96
Figure 5.11 – IBEM model of engine (Image courtesy – University of Kentucky) ………………….. 96
Figure 5.12 – IBEM model of engine (Image courtesy – University of Kentucky)………………….. 97
Chapter 6
Figure 6.1– Illustrating Tortuosity Left: Tortuosity = 1 Right: Tortuosity > 1 ……………………… 100
Figure 6.2– Material simulation using AutoSEA…………………………………………………………. 101
Figure 6.3– Sample B foam measured normal vs. measured random (Paris method) ……………… 104
Figure 6.4– MJA316 SAC measured normal vs. measured random (Paris method) ……………….. 104
Figure 6.5– Foam (open cell) sample B foam simulation vs. measured SAC………………………… 105
Figure 6.6– Sample MJA316 fibre simulation vs. measured SAC …………………………………….. 105
Figure 6.7– Sample S3D AFR simulation vs. measured SAC ………………………………………….. 106
Figure 6.8– Sample S3G AFR simulation vs. measured SAC ………………………………………….. 106
Figure 6.9– Sample S3M AFR simulation vs. measured SAC …………………………………………. 107
Figure 6.10– Sample S3M AFR simulation vs. measured SAC ………………………………………… 107
Chapter 7
Figure 7.1– Alpha Cabin CAD design drawing………………………………………………………….. 110
Figure 7.2– SEA model of Alpha Cabin ………………………………………………………………….. 111
Figure 7.3– Current production (M380) carpet …………………………………………………………. 112
Figure 7.4– Carpet model labelled into 18 sections……………………………………………………… 112
Figure 7.5– Cavity sections in the SEA model of the Alpha Cabin……………………………………. 112
Figure 7.6– Complete virtual Alpha Cabin SEA model, with noise sources and test carpet……….. 113
Figure 7.7– Test results of the virtual Alpha Cabin……………………………………………………… 114
Figure 7.8– SEA model of engine compartment………………………………………………………….. 117
XII
Figure 7.9– Passenger cabin simulation SEA model …………………………………………………… 118
Figure 7.10– Rear compartment of vehicle SEA model ………………………………………………… 118
Figure 7.11– Complete vehicle, sedan car, SEA virtual model………………………………………… 119
Figure 7.12– Vehicle interior acoustic cavities………………………………………………………….. 120
Figure 7.13– Vehicle exterior acoustic cavities………………………………………………………….. 121
Figure 7.14– Carpet section of sound package ………………………………………………………….. 123
Figure 7.15– Headliner section of sound package ……………………………………………………… 123
Figure 7.16– Seats section of sound package …………………………………………………………… 123
Figure 7.17– Trunk cover section of sound package …………………………………………………… 123
Figure 7.18– Pillars section of sound package …………………………………………………………. 124
Figure 7.19– Control dash board section of sound package…………………………………………… 124
Figure 7.20– Doors section of sound package…………………………………………………………… 124
Figure 7.21– Fire wall section of sound package ……………………………………………………….. 124
Figure 7.22– Power train (engine) noise source ……………………………………………………….. 125
Figure 7.23– Tyre noise sources ………………………………………………………………………….. 125
Figure 7.24– Transfer function test setup ………………………………………………………………… 126
Figure 7.25– Omni directional loudspeaker at the driver’s ear ………………………………………. 126
Figure 7.26– Microphone at wheel arch ………………………………………………………………… 126
Figure 7.27– Wheel arch to the driver’s ear transfer function ……………………………………….. 127
Figure 7.28– Engine to the driver’s ear transfer function ……………………………………………... 127
Figure 7.29– Noise spectrum at the driver’s ear, for engine noise source ………………………….. 129
Figure 7.30– Noise spectrum at the driver’s ear, for tyre noise ………………………………………. 130
Figure 7.31– Thermo gram for engine noise and tyre noise, at various frequencies……………….. 131
Chapter 8
Figure 8.1– Impedance tube setup ………………………………………………………………………… 137
Figure 8.2– RMIT Alpha Cabin …………………………………………………………………………… 138
Figure 8.3a– Aachen HMS III and KMT Tachometer set-up ………………………………………….. 139
Figure 8.3b– Test run way specifics ………………………………………………………………………. 139
Figure 8.4– PULSE set-up (left); Microphone position on floor carpet (right) …………………….. 140
Figure 8.5– Impedance tube SAC of mat specimens ……………………………………………………. 141
Figure 8.6– Effect of tufting pile length on SAC ………………………………………………………… 142
Figure 8.7– Alpha cabin SAC of mat specimen …………………………………………………………. 142
Figure 8.8– Total SPL at Driver's Ear level during stationary conditions at 2000 engine
rpm (PULSE data)…………………………………………………………………………………………… 143
Figure 8.9– Summary of total SPL at driver's ear level for various engine rpm's during
stationary tests (PULSE data)……………………………………………………………………………… 144
Figure 8.10– Summary of in-cabin average SPL at various engine rpm's during stationary
tests (HMS III data) ………………………………………………………………………………………… 145
Figure 8.11– Summary of in-cabin articulation index (AI) during stationary tests (HMS III data).. 145
Figure 8.12– Summary of total SPL at driver's ear level during on-road constant
speed tests (PULSE data) …………………………………………………………………………………... 146
Figure 8.13– Summary of in-cabin average SPL during on-road constant speed
tests (HMS III data) ………………………………………………………………………………………… 147
Figure 8.14– Measured AI at 3000 rpm, stationary conditions (HMS data)…………………………. 147
Figure 8.15– Summary of in-cabin Articulation Index (AI) during on-road constant
speed tests (HMS III data) ………………………………………………………………………………….. 148
Figure 8.16– 2nd order SPL at driver's ear level for run-up test (PULSE data)……………………… 148
Chapter 9
Figure 9.1– Concept 3 ………………………………………………………………………….................. 155
Figure 9.2– Concept 3 absorption measurement results……………………………………………….. 156
Figure 9.3– Taguchi loss function ………………………………………………………………………… 159
Figure 9.4– Taguchi orthogonal parameters for carpet concept design……………………………… 162
Figure 9.7– Carpet zones for Alpha cabin based ON/OFF Taguchi analysis……………………….. 177
Figure 9.8– Alpha Cabin - Carpet section analysis (Low freq) Taguchi loss function……………… 179
Figure 9.9– Alpha Cabin - Carpet section analysis (Mid freq) Taguchi loss function……………… 180
Figure 9.10– Alpha Cabin - Carpet section analysis (High freq) Taguchi loss function…………… 180
Figure 9.11– Sound package sections/categories for ON/OFF Taguchi analysis…………………… 182
Figure 9.12– Vehicle sound package analysis (Engine noise – Low freq)
XIII
Taguchi loss function ……………………………………………………………………………………….. 186
Figure 9.13– Vehicle sound package analysis (Tyre noise – Mid freq)
Taguchi loss function ……………………………………………………………………………………….. 187
Figure 9.14– Vehicle sound package & window analysis (Engine noise – Low freq)……………… 190
Figure 9.15– Vehicle sound package & window analysis (Tyre noise – Mid freq)…………………. 190
Figure 9.16– Vehicle level carpet section ON/OFF Taguchi zones…………………………………… 191
Figure 9.17– Carpet section (Vehicle level) analysis (Engine noise – Low freq)…………………… 194
Figure 9.18– Carpet section (Vehicle level) analysis (Tyre noise – Mid freq)………………………. 194
Chapter 10
Figure 10.1– Double panel sound barrier ……………………………………………………………….. 202
Figure 10.2– Concept 4 carpet design ……………………………………………………..…………….. 203
Figure 10.3– Transmission Loss measurements of Concept 4 design models……………………….. 204
Figure 10.4– Sound Absorption Coefficient measurements of Concept 4 design models………….. 206
Figure 10.5– Concept 3 absorber/de-coupler layer replaced by honey-comb………………………. 207
Figure 10.6– Comparison of concept 4 and concept 3, without new weight
or thickness overheads………………………………………………………………………………………. 208
Figure 10.7– Tuned Helmholtz resonant absorber layer included in the Honey Comb
dual panel structure …………………………………………………………………………………………. 209
Figure 10.8– SAC result of perforated facing foam (25mm) Helmholtz resonator………………….. 210
Figure 10.9– SAC results of corrugated foil facing foam resonant membrane absorber…………... 211
XIV
LIST OF TABLES
Chapter 4
Table 4.1- Nomenclature of layered material combinations ………………………………………… 78
Chapter 5
Table 5.1 - nomenclature of layered material combinations ………………………………………… 98
Chapter 7
Table 7.1- Statistic of vehicle SEA model ………………………………………………………………. 119
Chapter 8
Table 8.1- Sample Specimen Properties ……………………………………………………………….. 136
Chapter 9
Table 9.1 - Taguchi loss function table …………………………………………………………………. 160
Table 9.2- Concept 3 initial ON/OFF Taguchi based test run results ……………………………… 162
Table 9.3- ON/OFF Taguchi loss function table ………………………………………………………. 163
Table 9.4- Concept 3 ON/OFF Taguchi based test run results ……………………………………… 164
Table 9.5- ON/OFF Taguchi loss function table ………………………………………………………. 165
Table 9.6- Carpet parameters used in Taguchi analysis ……………………………………………. 167
Table 9.7- Concept 3 Taguchi based test run results ……………………………………………….. 168
Table 9.8- Taguchi loss function table …………………………………………………………………. 169
Table 9.9- Carpet parameters used in Taguchi analysis (TL) ……………………………………... 171
Table 9.10- Concept 3 Taguchi based test run results (TL) ………………………………………… 172
Table 9.11- Taguchi loss function table (TL)…………………………………………………………. 173
Table 9.12- Taguchi loss function table (TL)…………………………………………………………. 176
Table 9.13- Taguchi Parameter table …………………………………………………………………. 177
Table 9.14 - Taguchi Parameter table …………………………………………………………………. 178
Table 9.15- Taguchi loss function table ………………………………………………………………. 178
Table 9.16- Taguchi Parameter table ………………………………………………………………. 182
Table 9.17- Taguchi test table ………………………………………………………………. 183
Table 9.19- Taguchi test table ………………………………………………………………. 184
Chapter 10
Table 10.1 - Nomenclature of layered material combinations ………………………………………. 204
XV
Masters By Research SUMMARY
SUMMARY
The main objective of this research project is to design, develop and validate an
innovative vehicle floor carpet system with improved acoustic performance, and thus reduce
noise levels inside the vehicle cabins. The proposed solutions are expected to improve
vehicle floor carpet product in areas of acoustic performance, cost, weight and waste
reduction, to be environmentally friendly and sustainable in manufacturing.
The following are main research outcomes of the project.
• Acoustically improved vehicle floor carpet with higher sound transmission loss
and in-cabin sound absorption coefficient, compared to current production carpet
designs.
• Vehicle floor carpet designs that introduce minimum weight and cost penalty for
the acoustic performance improvement obtained.
• Material database, i.e. measured acoustic parameters for mathematical modelling,

of different vehicle carpet layers.
• Virtual modelling and validation method for design evaluation at component and
vehicle levels.
• In-situ vehicle on-road validation test methods for carpet designs
• Optimization methods for further improving the design
In the course of the research, the following questions are also addressed and discussed in the
thesis:
• What is the role of vehicle floor carpet in eliminating engine and tyre noise and
how good is the current design in achieving it?
• Is virtual modelling and validation an effective tool for designing and evaluating
a vehicle carpet acoustically at vehicle level?
• What could be the theoretical limits of noise reduction in the vehicle cabin by
acoustically improved vehicle carpets and where are we now?
• Are there simple and efficient methods for in-situ measuring the acoustic
performance of carpet design, non-destructively?
XVI
• What are the future possibilities in introducing tuned acoustic layers for vehicle
carpets?
Initial benchmark analyses and target settings for vehicle cabin noise are conducted
to get a clear understanding of the design requirements and constraints. The following
observations are worthwhile to be noted from the vehicle tests. Low frequency power train
noise has higher sound pressure level (SPL) distribution in the front part of the carpet (front
foot well area), which may need more sound barriers. Meanwhile, high frequency tyre-road
interaction noise has higher sound pressure level (SPL) distribution in the rear part of the
carpet (rear wheel arch and the boot areas), which may need more sound absorption
measures.
Based on the benchmark analysis results and review of current carpets systems, the
key acoustic parameters for new designs are identified to be Transmission Loss (TL) and in-
cabin Sound Absorption Coefficient (SAC), and the various carpet design objectives and
constraints are hence defined. This includes design objectives for physical characteristics and
functional requirements. A clear understanding of the validation process required for the
design evaluation, both virtually and on road, is achieved.
Theoretical analysis of forced vibration of the carpet system is studied by developing

mathematical models based on the simple harmonic response and sound radiation directivity
of the floor panel and carpet system. The Simple Harmonic Motion oscillator model of
analyses of the current carpet design system shows that there is a resonance around the
frequency of 323Hz which is in the frequency range of engine or power train noise. This is
verified from the sound Transmission Loss measurements of the carpet variants, using the
impedance tube method. Further, the theoretical model presented for panel radiation
directivity prediction can be used to successfully predict the sound level radiated at a
particular angle to the normal of a panel or opening, relative to the sound level radiated in the
direction of the normal.
With a basic theoretical understanding of the current carpet design model, four new
concept carpet designs are proposed and evaluated in laboratory for acoustic performance. A
concept design for both good cabin side sound absorption coefficients and Sound
Transmission Loss, identified as Concept 3 design, is developed.
From the above mentioned benchmarking and target setting stage, a systematic
design and simulation process flow diagram is formulated. Based on the acoustic targets and
the resolution requirement of the results, a suitable simulation tool (Statistical Energy
XVII
Analysis) is selected for design and evaluation. A simulation process is defined for carpet
systems, which includes material level, component level and vehicle level. Based on this
simulation process, the current carpet systems are evaluated, and further new design concepts
are identified and proposed based on the simulation results.
Material level simulation validation is conducted for various acoustic samples and
measured material sound absorption coefficients have a similar trend to simulated ones and
have a variation of only about 10-15% for various samples. The simulation results do not
include the low frequency resonance peaks, i.e. below 300Hz, observed in normal incidence
absorption measurements. Further, since SEA tools use random incidence Sound absorption
Coefficient (SAC) values for simulation inputs, Paris method [53] of converting normal
incidence SAC to random incidence is used and validated.
The proposed carpet designs are further simulated and evaluated at component and
vehicle levels. In the Alpha Cabin simulation (component level) tests, compared with the
current production carpet (MITSUBISHI 380 carpet), Concept 3 carpet consistently reduces
about 4.2 dB SPL above 2kHz and reduces up to average of 1 dB below 1 kHz. Concept 3
special mats reduce the noise level by 3 dB above 2 kHz and up to an average of 1 dB below
1kHz, in comparison with the current production carpet (MITSUBISHI 380 carpet).
The vehicle virtual SEA model was evaluated and validated using acoustic transfer
function method and well matching results were achieved for engine and tyre noises, except
for the coincidence frequency ranges of the backing sheet metal floor. Concept 3 carpet
evaluation at the vehicle level shows the reduction of the in-cabin noise level further by up to
3-5 dB, compared to the current production carpet design, for power train noise. The
Concept 3 carpet simulation at the vehicle level shows the reduction of the in-cabin noise
level further by 2 dB at high frequencies, compared to the current production carpet design,
for the tyre-road noise. The Concept 3 mat padded with the current production carpet reduces
noise level by about 1dB SPL at high frequencies, compared with the carpet only case for the
power train and tyre-road noise. It is illustrated that the proposed the concept 3 carpet has
nearly reached the noise reduction limit.
As the component and vehicle simulation results are giving promising results, the
proposed Concept 3 mat design is proto-typed for on-road evaluation. A proof of concept
mat based on the proposed Concept 3 design (C3-Mat_Prototype) was factory moulded,
albeit slight modifications from the original specifications due to material, time and
technical constraints. In the impedance tube SAC test, the C3-Mat_Prototype performed
XVIII
better than the standard mat in the low to mid frequency ranges (up to 3000Hz). Meanwhile,
RMIT Alpha Cabin SAC test showed that the C3-Mat_Prototype outperformed the standard
mat by an average of 14.2 % for in-cabin side SAC, with the highest improvement of about
40 % of SAC at 800 Hz. On-road vehicle tests show that the C3-Mat_Prototype has better
sound absorption at the low to mid frequency range than Standard mat. On road tests using
both B&K Pulse and binaural head acoustics were conducted and the data indicates a
reduction of in-cabin total SPL by the prototype mat is about 2.1 dB(A) during stationary
tests and 0.625 dB(A) during constant driving tests. It is noted that these improvements are
measured for the proof of concept factory moulded mat, which points to the fact that a
Concept 3 carpet system as a whole can give promising results, as predicted by the
simulation model results.
Sensitivity and optimization of the proposed carpet design (Concept 3) are conducted
by the Taguchi method. Simulation based Taguchi optimization gives a clear indication for
further optimization directions. It is preferred that separate Taguchi loss function calculation
be done for various frequency ranges, like Low (315Hz-1kHz), Mid(1kHz-4kHz) and
High(4kHz-8kHz) frequency ranges. In Concept 3 carpet layers the AFR and substrate layers
have the largest air flow resistivity and therefore they have the largest influence on the sound
absorption coefficients in the low frequency range (315 Hz – 1000 Hz). Meanwhile, the foam
de-coupler has the largest influence on the sound transmission loss in the low and mid
frequency ranges. ON/OFF Taguchi analysis in Alpha Cabin shows that the front foot rest
areas consistently dominate in the low, mid and high frequencies and ON/OFF Taguchi
analysis for vehicle sound package shows that for engine noise (low frequencies 315Hz to 1
kHz), the headliner and then carpet are the dominating components. Meanwhile, in the case
of tyre-road noise, the seats have the highest loss function in all the frequency ranges of
interest, followed by carpet treatment, at the mid frequencies (1kHz - 4kHz). An interesting
fact identified from the Taguchi analysis results is that the window glasses play a very
significant role in the in-cabin sound pressure levels over the low frequency ranges, for both
the engine and tyre noises.
The following are identified as recommendations for future design work in improving
vehicle carpet acoustics.
• Introducing air-gaps in the range of 10mm in between the heavy layers by use of a
honey-comb structure improves the transmission loss by up to 20dB in the frequency
range of 1kHz – 2kHz, and up to 10dB in the frequency range of 500Hz – 1kHz, and
achieves the best sound transmission loss in these frequency ranges.
XIX
• The introduction of honey-comb structure as an air-gap structure does not add any
over-head in terms of weight or thickness, compared to foam, nor affects the total
absorption of the carpet system.
• Special tuned absorber layers like the perforated facing foam Helmholtz resonator
and the corrugated foil faced foam membrane absorbers are excellent in extending the
noise reduction frequency ranges to specific low frequency ranges of interest.
XX
CHAPTER 1
INTRODUCTION
1 GENERAL DESCRIPTION
The study and prediction of sound insulation and absorption of various materials,
which includes carpets, trims, heavy sound insulation layers and films, has always been of
great significance and interest for acoustic consultants, involved in design of sound
packages, especially for vehicle acoustic package designs.
Research has shown that the material acoustic properties can be measured to a
desired level of accuracy, which can be used for calculating and predicting the independent
acoustical response in a controlled and isolated enclosure with the material samples. Various
acoustic evaluation methods like reverberation room measurements, alpha cabin method,
Johnson Control method for Transmission loss, etc have been researched and applied for
these independent evaluations.
Meanwhile, in the case of an acoustical system which has a combination of various

acoustical materials, including structurally linked bodies like in a vehicle cabin, current
research shows that a direct addition of the measured acoustical parameters of separate
materials may not fully apply. In that case, Statistical Energy Analysis is one of the
recommended approaches to predict such a complex acoustical system, as current research
shows that it can predict acoustic responses for a wider range of frequencies, even
comparable to Finite Element Analysis method.
Masters By Research CHAPTER: 1
Acoustic and
Thermal package
of a typical car
Figure 1.1: Acoustic and thermal package of a typical car

(Courtesy: Rieter automotive, http://www.rieter.com/en/general/products)
A typical car acoustic and thermal package is illustrated in the above Figure 1.1,
where this research will be focussed on the improvement of the acoustic insulation and in-
cabin absorption of the vehicle floor and carpet, i.e. the blue coloured section of Figure 1.1.
Based on the current research and developments on Statistical Energy Analysis, this
research hereby attempts to design acoustically improved vehicle carpets, validate them at
component and vehicle levels, and further optimize the design for given constraints, like
defined industry standards and targets.
2 SCOPE AND OBJECTIVE
As per current research, Statistical Energy Analysis works well with prediction for
reasonably complex acoustic systems, with given acoustic parameters of materials, measured
using methods like Reverberation room method, Alpha cabin method (preferred for higher
frequencies), which are mainly based on random incidence measurements. However, this
modelling method has the following drawbacks,
1. Acoustical properties measured using methods like the impedance tube doesn't seem to
work well with this modelling method. This requires a study of the influence of angle of
sound incidence, as impedance tubes can measure only normal incidence acoustic
properties.
2
2. Material properties like flow resistivity, porosity, tortuosity, etc are required for
Statistical Energy Analysis modelling of the material, which are measured generally
using various different other methods, which itself brings a dependency on different test
setups.
This research also studies the possibilities of providing theoretical explanations for
the reasons behind the issues with regards to using impedance tube measurement results
(normal incidence) for Statistical Energy Analysis modelling.
There had been large amount of research and development in the area of acoustic
insulation and noise reduction for different types of vehicles. There are several set rules
applied which have been successful in meeting the current noise standards of the industry.
But these methods are generally man power intensive and involve high precision tools and
methodologies, which eventually take up time and money. So there comes a need for a quick
and portable method for predicting and evaluating an acoustic solution or sound package, for
vehicles.
This project contributes significantly to existing technology aiming at dramatically

reducing cost and production lead times whilst delivering improved function and
performance of an automotive carpet system.
Improved functionality and performance of an automotive carpet system with the

opportunity to reduce cost, waste and manufacturing lead-times is one of the focuses of this
research. Outcomes of this research project are expected to provide competitive know-how
in design and development of automotive carpet systems.
3 LITERATURE REVIEW
3.1 Carpet Material and Acoustics
Hilyard, et al., [7] described procedures being used to rank the acoustic performance
of all diphenylmethane di-isocyanate (MDI) foam backed carpet systems. The procedures
were based on the airborne noise Insert Loss (IL) measurement of the carpet composite under
laboratory conditions and well established criteria used to assess human response to noisy
environments, such as annoyance and speech perception. Data were presented which
demonstrated how the interior noise spectrum might be influenced by the unit mass of the
heavy layer, and the thickness of the foam backing layer. It was shown that low modulus
isolating systems are superior in Articulation Index performance. Other criteria indicated
3
that for incident airborne noise the best overall performance should be achieved with a
system having a resonance frequency within the 1/3 octave bands centred at 260 Hz and 315
Hz. The influence of system design parameters and foam properties on the resonance
frequency was discussed.
Procedures had been described for ranking foam [7] based automotive noise control
carpet systems with an assumed airborne noise input, Insert Loss data measured in the
laboratory and “noisiness” criteria. The criteria examined were the Articulation Index (AI),
the Noise Criteria (NC) and discrete noise component band sound pressure level. Data were
presented for the airborne noise IL behaviour of all-MDI polyurethane foam based carpet
systems with filled EVA heavy layers. They showed that the resonance frequency trough was
shifted to lower frequencies and the IL at resonance enhanced by increasing the unit mass of
heavy layer. When the ranking procedures were applied to environmental data it was
concluded that the NC approach did not have sufficient sensitivity to differentiate between
carpet systems. However, the use of AI, to assess speech perception qualities, and low
frequency octave band level, to assess annoyance, showed promise.
Six foam backed carpet systems had been analysed and ranked in terms of potential
AI performance [7]. It was found that performance ranked in this way was strongly related to
the resonance frequency of the system, systems having low resonance frequency being
predicted to perform better than systems with high resonance frequency. System resonance
was governed primarily by the effective stiffness and thickness of the foam isolating layer
and the unit mass of the steel substrate and heavy layer. Good agreement between predicted
and measured resonance frequencies was obtained.
It was demonstrated that systems with good AI performance, which was related to the
high frequency IL behaviour, did not perform well at low frequencies. It was concluded that
there was potential for optimising the design of a carpet system for a particular vehicle type
[7]. This involved designing the system such that the resonance frequency fell within a
certain range. For the situation analysed here the optimum range was within the 1/3 octave
frequency bands centred at 250 Hz and 315 Hz. Further refinement to performance could be
achieved through the control of the damping and airflow resistivity of the MDI foam
insulating layer.
One half of interior noise energy penetrates through the floor pan and bulkhead. The
two main requirements for the development of improved carpet systems were: (i) criteria for
assessing performance and (ii) knowledge of how system design parameters affect their
4
behaviour. The engine or drive train noise has discrete frequency and controls the noise
annoyance, the random broad band noise from the road surface was the major factor
contributing to the loudness, and hence interference with speech perception. The main
discrete components occurred at low frequencies, below 200 Hz and make no contribution to
the AI value [7].
A systematic benchmarking study was performed by Balte, et al., [8] to investigate

the acoustic performance of production floor coverings of vehicles. A large number of
passenger cars including compact, mid-size, full size, and a truck were selected. The floor
coverings were removed from these vehicles and evaluated both on absorption and sound
transmission loss performances. It was discovered that the design of the carpet was more
important than the materials used. In addition, a carpet with highest absorption did not
necessarily have the best sound transmission loss and vice versa. However, an optimum
design could achieve high performance in both categories.
Felt-Barrier-Felt-Barrier carpet, a thick carpet provided the best STL. The carpet with
nearly 100% coverage of the insulation with the lowest cut-outs provided the best design and
acoustic potential. Absorption, transmission loss, homogeneity of material, coverage of
insulation, cut-outs and weight determined the acoustic performance of carpets.
Geometric and physical data of selected benchmarking carpets consisted of

Dimensions (mm), Area (m2), Cut-outs(%), Absolute Coverage (%), Weight (kg),
Draw(mm), Heel Pad, Thickness (mm), Density (kg/m3) and Squared Density (kg/m2).
The use of renewable materials instead of manmade fibres as reinforcement in

composite parts provided interesting alternatives for the production of ecologically friendly
products [9]. Especially in Europe, renewable materials were of increasing interest because
of a directive of the European Parliament that dictated an automotive material recycling rate
greater than 80% by the year 2006. This directive was a big opportunity to introduce
materials from renewable resources. Therefore, new materials for various interior parts of
cars were of interest.
All interior parts of vehicles had different requirements to fulfil regarding acoustics
and mechanical performance. Some parts were damping, some parts were absorbing, and
some parts fulfilled requirements somewhere in between. In the automotive industry, there
were various materials used for acoustic parts and four interesting material compositions
were chosen to present in more detail: a standard cotton material, polyester micro-fibre felts,
5
Lyocell material and natural fibre felts. Nevertheless, it was very difficult to find an
optimum material that both fulfilled all requirements and improved passenger comfort.
In general, interior vehicle noise was influenced by various sources; e.g., the engine,
exhaust system, gears, wheels, and wind which were so-called primary sound sources.
Secondary sound sources were glass and metal sheets. So there was a collection of noise
factors that influenced each other.
The use of textiles, especially felts, was based on two major advantages of these
materials: low production costs and excellent noise absorption capacity. Felts based on
polyester micro-fibres and Lyocell [5] were compared to standard material felts and natural
fibre felts. The standard material was a mixture of cotton, polyester bio-component fibres,
and polypropylene fibres and represented an ecological solution for good noise absorption.
Natural fibre felts were a mixture of natural fibres (flax or hemp) and polypropylene fibres
(each 50%). Lyocell felts contained Lyocell (50%), polyester bio-component fibres, and
polypropylene fibres. Variations in material weight, thickness, and fibre diameter were
considered.
Lyocell is a cellulose fibre based on renewable resources, e.g., wood, and is obtained
by an organic solvent spinning process. It is strong, absorbent, wrinkle resistant,
biodegradable, and can be fibrillated during processing and has good drapability.
Comparing standard material to Lyocell and to polyester micro-fibre felt showed

different advantages of the materials in different frequency ranges. The standard material
performed very well in the frequency range above 4000 Hz, whereas it showed some deficits
between 1200 Hz and 4000 Hz. Polyester micro-fibre felts showed an improved noise
absorption in the middle frequency range of 1200 Hz up to 4000 Hz, which was important
for the articulation inside a car. Below 1200 Hz, micro-fibre felts were not as good as
standard material because of the homogeneous fibre thickness. Felts made of Lyocell fibres
showed the best absorption behaviour over the whole frequency range. Especially below
1600 Hz, the absorption was significantly improved. This range was very important for
diesel motors. The comfort for passengers inside a car could be improved by increasing noise
absorption with an intelligent application of renewable materials like Lyocell.
Skinners, et al.[10] described trends in the automotive market and then key features
determining acoustic absorption of material. They then highlighted the development of a
novel technology named ACOUSTIFLEX from Huntsman Polyurethanes delivering
6
acoustically active foams straight from production. Physical and acoustic data demonstrated
that this technology was suitable for a number of application areas within the automotive
market and that the balance of acoustic performance and weight outperformed any other
materials currently available on the market. The technology therefore offered producers the
opportunity to enhance the acoustic comfort of their vehicles whilst simultaneously lowering
the overall weight to achieve this.
3.2 Acoustic Simulation and Measurement

Biot’s theory provided a framework for the numerical modelling of propagating stress
waves in elastic porous materials. Sureshkumar, S., et al. [11] described a finite element
method technique based on the adaptation of Biot’s theory to acoustic porous material that
was applicable for the solution of complex systems consisting of porous, fluid and structural
media. Acoustic indicators such as absorption coefficient and transmission loss were
calculated for flat samples and these results were compared to known solutions. Finally the
transmission loss of a complex dash system was computed and contrasted with the
corresponding planar multi-layer results.
Computational simulation tools based on the adaptation of Biot’s theory for the
modelling served to accurately and effectively predict the acoustical behaviour of porous
materials. Prediction of the acoustic behaviour of arbitrary noise control configurations was
not feasible using analytical methods and therefore a finite element method based solution
technique was used. The validity and applicability of the approach and resulting
implementation were demonstrated by solving and verifying example problems.
Duval, et al. [12] first characterized the in-situ diffuse field absorption coefficients
using micro-flown p-u probes, showing encouraging results compared to measurements
using the reverberation time technique or to poro-elastic simulation.
Obtaining reliable diffuse field absorption coefficients was a key issue for the
development of noise control treatment using simulation tools like S.E.A or Ray-Tracing
methods [12]. The energy based method called “Vehicle Acoustic Synthesis Method”,
calculating the sound Pressure Level at ear points from the combination of sound power
measurements and acoustic transfer function panel/ear measured or simulated -with Ray-
Tracing Methods- for the middle and high frequency range [12].
The 2nd generation of the “Vehicle Acoustic Synthesis Method” (VASM 2) using
micro-flown pressure -particle velocity (p-u) [1], for both sound intensity and transfer
7
functions measurements (without moving the probes), was used to speed up the time to build
a validated model of a fully trimmed vehicle in the middle and high frequency range, while
increasing accuracy both in terms of source localization and source quantification and while
addressing unsteady operating conditions like run-ups [2].
When the transfer functions were measured with a monopole source positioned at ear
point applying the reciprocity principle, only the pressure channel of the p-u probe was used.
The equation was if it would be possible to get the reflection coefficient of the materials
constituting the panels using both pressure and particle velocity channels (from the direct
impedance or from the “absorbed” intensity). Unfortunately, an acoustic field hypothesis was
needed to answer that question: free-field, standing waves or diffuse field [3].
The reality of a car interior cavity was a semi-statistical field in the middle and high
frequency range (315 - 10000 Hz). Nevertheless, upper than a given frequency (typically
1000 Hz in a car), the hypothesis of local diffuse fields could be made like S.E.A. users did.
In order to validate this procedure, the absorption coefficients (respectively reflection
coefficient α=1-|R|2) of poro-elastic materials in reverberant rooms of different sizes were
measured.
Compared to the 1st generation of the “Vehicle Acoustic Synthesis Method” where
the source strength characterization was performed using the classical p-p sound intensity
technique, the 2nd generation validated the implementation of the micro-flown p-u pressure-
particle velocity probes for both sound intensity and transfer functions measurements
without moving the probes and without any specific measurement environment [2].
The new measurement technique allowed speeding up the measurement protocol,

performing the “Vehicle Acoustic Synthesis Method” on a complete car with all power train
operating sources then required only two weeks. Moreover, intensity maps were more
refined and unsteady conditions like run-ups became feasible.
The originality of this approach consisted in the simulation of the acoustic package
by introducing the insulators in the model as a modification of the power injected in the car
compartment (simulated for example by simple Transfer Matrix codes or SEA models or
even measured), and as a modification of the cavity transfer functions according to the
modification of absorption properties. Using Ray-Tracing simulation, this method became
100% numerical on the acoustic treatment for the middle and high frequency range. The
8
“Vehicle Acoustic Synthesis Method” was based on an optimized trade-off between good
insulation and broadband absorption.
The very good simulation-experiment results had led to the design of an optimized
acoustic package reaching the challenging target of a weight reduction of 10%, while
maintaining the acoustic performance and the cost.
An important automation effort had been done leading to a comprehensive MATLAB

program supporting the modelling effort from the meshing, to the measurements and to the
poro-elastic and Ray-Tracing simulation up to 3D colour maps post-treatments.
Three methodical approaches to aid decision making about noise reduction during
design stage, using SEA, is being presented in [21].
In order to simulate successively the absorption coefficients of materials for normal

and random incidence with Maine 3A V1.3, the Biot parameters of these materials had been
determined using an inverse technique based on impedance tube measurements and direct air
flow resistivity measurements [4].
In the automotive industry, random incidence sound absorption tests were conducted
on flat material samples as well as on finished components such as headliners, seats, and
floor carpet systems. Veen, et al. [14] discussed a feasibility study that was being pursued by
an SAE task force, under the direction of the Acoustical Materials Committee, to develop a
small volume reverberation room test method for conducting random incidence sound
absorption tests. This method had the potential to be suitable for the flat material and
component testing. A round robin test program was being conducted to determine variability
due to test procedures, room size differences and laboratory differences.
The round robin study conducted thus far showed that a properly designed small
volume reverberation room, which was significantly smaller than a “full size” room, had the
potential to generate comparable results to that of the industry standard full size
reverberation room for appropriate frequency ranges. However, the study done thus far also
revealed that tests conducted in large rooms had high variations at high frequencies (at and
above the frequency where the performance peaks). This variation was larger for higher
performing samples than it was for lower performing samples. Also, small size rooms had
high variations in the lower frequency range. Therefore, the reason for high variations in test
results on similarly designed test rooms needed to be thoroughly understood so that it could
9
be minimized. Otherwise, there was the potential for larger than acceptable variation in test
results from small reverberation room test facilities.
Design Sensitivity analysis (for final stages, to meet targets) – DSA
This was an automated analysis (by computer), where the effect of doubling each
small components (like even screw hole treatments) was used to measure sensitivity of
design changes (for each frequency), which was taken as input and automatically found
optimum combination, considering price also.
Once an SEA model was ready, the 3 key parameters, i.e. acoustic performance, cost
and weight, could be used by optimisation software to iteratively find the best range of
package configuration. A methodology for finding a range of solutions for finding the best
sound package relative to cost and weight was considered [22]. Since iterations on an SEA
model, by varying properties like absorption, damping and thickness were easy and fast, it
was feasible to run an optimiser to work on optimising for cost and weight criteria.
Different cost functions, placing varying emphasis on acoustic performance at target

locations and frequencies, plus total cost and weight criteria were used to analyse the SEA
model and find an optimum solution. As a first step, a baseline SEA model with a baseline
sound package configuration was done. Further the major contributing subsystems were
identified. Next, several iterations were run by varying absorption, damping and thickness
for these subsystems, to measure the sensitivities of each variation, which was used as input
of the optimiser. The 3 parameters mentioned could form a 3D space where the design
converges to the optimised package point, visually. A cycle of successive steps would thus
converge toward configurations that optimised for the given constraints and emphasis
(relative weights) of acoustic performance, cost and weight [22].
With reference to case studies, when a vibration source was known to create a
problem at a given frequency, the optimisation was tailored to weight the response at the
trouble frequency to be more important than other "non-critical" frequencies. This
methodology was also powerful in its flexibility, since the acoustic performance metric could
be weighted to stress problem frequencies.
A focused study on the development of treatments to control airborne noise through

the dash panels was done [23]. Interior sound levels in trucks continued to be reduced;
however, airborne sound transmission through the floor and dash panel (firewall) remained
dominant noise paths. There were actually two types of design criteria that must be
10
established in the sound package development. One was the criterion describing the sound
pressure level inside the cab and the other criterion described the quiet sound package
treatment.
When establishing the criteria for the interior sound level there were a number of
metrics that could be used. These typically included the A-weighted sound level, 1/3 octave
band sound pressure level, speech interference level (SIL), preferred speech interference
level (PSIL), loudness and articulation index (AI).
The SIL was the arithmetic average of the linear-weighted sound levels in the 500,
1000, 2000 and 4000Hz octave bands. The PSIL was similar to the SIL except that it only
averaged the sound level in three octave bands (500, 1000, and 2000Hz). The AI was a
measure of the intelligibility of speech with background noise present.
It was important to understand that the noise reduction obtained by the material
depended on the substrate to which the material was applied. The truck manufacturers
utilized many different substrate materials and constructions so a package developed for one
truck might not be optimum for another truck.
Some restrictions included resistance to heat, thickness and weight constraints, limits
on how much moisture the de-coupler might absorb and aesthetics.Depending on the
substrate and the design of the dash panel, the “optimum” barrier surface weight might be
between 3.8 and 9.5 kg/m2.
Component testing had been found to be invaluable as a tool for developing noise
control treatments. With full truck on road testing, it was difficult to test actual parts due to
space constraints. In addition, variability in the diesel engine sound levels can be of the same
order of magnitude as the changes expected due to slight treatment modifications.
Loudspeaker tests, involving placing a loudspeaker near the engine block, can also be
difficult due to space constraints and flanking paths.
A case study based on the acoustic tools like SEA, insertion loss and sound intensity
measurements, was done [24]. A SEA model was constructed to predict the primary paths
(panels or area) contributing to the overall interior sound field. Insertion loss measurements
were used to verify the primary contributing paths identified using SEA. To provide further
details of the primary paths, intensity maps of identified panels were measured allowing
detailed construction of the contributory panels.
11
It was found that SEA predicted the correct airborne acoustic paths in terms of sound
transmission loss, and could also provide direction in material changes to improve the TL
where necessary. The experimental tools of insertion loss and intensity reinforced the
prediction by SEA, and also found highly directive acoustic paths that SEA modelling was
unable to consider.
The interior noise of a vehicle usually originated from three major sources: power
train/drive line, road/tire interactions, and wind.
SEA was primarily used for noise and vibration analysis in the high frequency range,
considering diffuse energy transmission. Therefore this approach was appropriate for
reduction of air-borne road noise in the high frequency range. The key concept in SEA was
the averaging of energy. The spatial and frequency band averaged energy was used as the
primary variables in the governing equations.
The SEA model could contain basic parts, Exterior Subsystems -> Body panels ->
Interior subsystems. A SEA model could be an analytical or test-based model. The model in
this case was a test-based model. So most of the model parameters were obtained from
measurements.
The following two points were to be noted; 1. The exterior subsystems were
constrained to have the same energy level as a set of exterior vehicle measurements. The
measured SPLs were applied to the exterior cavities as energy constraints in the SEA model.
2. For body panels, the in-vehicle measured STL were used in all area junctions in the
model. The coupling areas in those junctions were also from measurements of the size of the
panels.
The Damping Loss Factor (DLF) was obtained by using an in-vehicle measured T60
and by following a redistribution procedure.
The concept behind panel contribution was to remove the area junction from the
model for the panel under consideration.
1. Insertion loss concept:
To determine the relative contribution of each path, the insertion loss of a barrier
material covering each successive partition was measured with reference to the occupant
headspace locations.
12
Limitations: One such limitation was the reduction in cabin absorption by removing
the seats; the removal was a necessary consequence for producing reliable insertion loss
measurements of all the vehicle partitions. A second limitation was the restriction of the
method to airborne acoustic energy transmission only. With the addition of heavy barrier
material to the partition, the structural response was obviously changed from the intended in-
situ performance that this method tried to encapsulate.
2. Insertion loss measurement procedure:
The following partitions were used: 1 Floor partition 2 Rear gate partition 3 Rear
quarter partition 4 Rear door partition 5 Front door partition 6 Front partition. The acoustic
excitation was provided by operation on a chassis dynamometer with a coarse road shell
installed on the outer roll surface. The resulting sound pressure level for each of the occupant
head-space locations before and after the sound barrier layer was installed were subtracted,
by insertion loss rule.
3. Sound intensity measurement:
The approach: In order to perform accurate intensity measurements, one must

consider the acoustic field in which the measurements are to be performed. An indication of
the field might be made from the Reactivity Index (sound pressure level Lp – sound intensity
level Li). If RI was zero, the field was perfectly anechoic. A value greater than zero
indicated the presence of sound waves in directions other than the incident wave of interest.
On analysis, above 2.5kHz, leaks in the door panel and the coincidence frequency of
the glass appeared to be the major paths through the rear door partition. Since SEA did not
account for leakage through seals, the path was under predicted in the panel contribution
analysis. This result further underscored the need for a reliable method to account for seal
and body leakage in SEA.
The ability to utilize existing infrastructure i.e. tools, equipment and machinery to
generate a unique acoustic product offering with minimal changes was quite important [25].
New materials, manufacturing processes and concepts regarding the treatment of automotive
noise problems had provided the opportunity for enhanced tuning of multi-layered systems.
For carpet and dash systems, it was claimed that a process was developed to control
the density over the part, vary the fibre blend and had complete control of the thickness from
2-3mm to in excess of 50mm. With the addition of other acoustic layers, a complete acoustic
answer to the challenge of meeting both the absorption and STL requirements was provided.
13
This along with meeting all of the other customer requirements, for weight, re-cyclability
and design parameters, while being cost effective.
Thus the key features of the concept are, 1. The ability to tune the multi-layered
system to meet the needs of a specific platform. 2. The ability to vary the density of the de-
coupler over the carpet and be able to maintain high lofts. 3. A flexible manufacturing
process gives the ability to manipulate the layers to achieve optimum performance 4. Once
the influence of each layer was understood, it was possible to manipulate in the poro-elastic
properties of each material to then adjust its performance.
A methodology for solving a full vehicle optimization problem for the shape and
damping layout of structural panels, by means of FE models, had been developed to meet
NVH and weight requirement early in a vehicle program [15]. Numerical results indicated
that tackling at the same time the structural shape modifications of the body panels, together
with geometry and the materials configuration of the damping viscoelastic layers could lead
to improved NVH performance.
Gansen, et al. [16] presented acoustical and physical comparisons among viscoelastic
foam, HR foam, and fibre pad. A mass backed carpet in junction with a low density underlay
transported by the floor was used to reduce air and structure borne noise which was induced
by the road, exhaust, and engine. Commonly used underlay included slab polyurethane (PU)
foam (HR foam), cast-in place PU foam, and cotton fibre pad. Fibre pad was pre-dominantly
used in the U.S. While, in Western Europe cast-in-place MDI-based viscoelastic PU foam
had been a preferred material for carpet underlay. Foam-backed carpets with mass-backing
had been produced in Europe for approximately twenty years; only a few North American
car makers currently utilized cast-in-place PU foam.
Some of the advantages of the use of foam-in-place over fibre pad for carpet underlay
were:
• Good dimensional stability, since the foam was fit to the contour of the floor structure.
• Good moisture resistance
• Easy assembly within the car, which lowered fitting costs
• Foam thickness could be easily varied in order to achieve optimum damping in areas
which needed effective damping.
14
• A variation of sound absorbing characteristics could be achieved by altering the foam

formulation.
For air-borne noise absorption, testing method- normal incident sound absorption
measurement was the impedance tube method. Viscoelastic and HR foam seemed to parallel
each other in sound absorption performance, while they both exhibited better sound
absorption performance than fibre pad at all frequencies except 3150 Hz and greater.
For structure borne noise damping, the Geiger Plate Test - attached the PU foam
sample to a metal sheet of specified thickness. The metal plate was excited in its fundamental
frequency mode and the dampening performance of the foam material was determined by
measuring the vibrational decay when the excitation was suddenly removed.
The Complex Modulus Test - A PU foam sample was attached to a thin metal bar
(Oberst bar) which was then excited using a random noise signal, FRF was measured and
half power band method to determine damping.
The viscoelastic PU foam performed as well as, or better than, the fibre pad and HR
foam at all modes of vibration except for the fifth. Viscoelastic PU foam exhibited very high
dampening performance around the third and fourth modes of vibrations, where fibre pad
had high dampening only at the third mode of vibration within the frequency range of 100 -
1000 Hz. Viscoelastic foam performed well as a vibration dampening material. It was a very
effective medium to use as carpet underlay in order to dampen structural-borne noise.
At the same density, viscoelastic PU foam across the board exhibited better physical
properties than fibre pad. The most significant differences in physical properties were tensile
strength, elongation and compression set which were of importance for a carpet underlay
application [16].
4 RESEARCH QUESTIONS
The specific research questions that are addressed in this thesis are the following:
• What is the role of vehicle carpet in eliminating engine and tyre noise and how
good is the current design in achieving it?
• Is virtual modelling and validation an effective tool for designing and evaluating
a vehicle carpet acoustically at vehicle level?
15
acoustically improved vehicle carpets and what are the current achieved levels?
• Are there simple and efficient methods of in-situ measuring the performance of a
design, non-destructively?
carpets?
5 OUTCOMES
The basic goal of this research project is to design, develop and validate an innovative
car carpet system with improved acoustic performance. The proposed solutions must
promote competitive advantage in areas of acoustic performance, cost effectiveness, product
weight and waste reduction, and propose environmentally friendly and sustainable
manufacturing solutions. The following outcomes are expected as the research yield in the
field.
• Acoustically improved vehicle carpet with higher sound transmission loss and
in-cabin sound absorption coefficient.
• Vehicle carpet designs that introduce minimum weight and cost penalty for the
acoustic performance improvement obtained.
• Material database, i.e. measured acoustic parameters for mathematical

modelling, of different vehicle carpet layers.
• Virtual modelling and validation method for design evaluation at component and
vehicle levels
• In-situ vehicle on-road validation test methods for carpet designs
• Optimization process methods for further improving the design
A detailed definition of the design target is discussed in section 3 of Chapter 2.
6 THESIS STRUCTURE
The development of acoustically optimized vehicle carpets is introduced and further
analysis is conducted on vehicles for identifying design targets, in Chapter 2. Chapter 3 is
dedicated for mathematical model analysis of the vehicle carpet and for sound radiation
model derivation. In Chapter 4 the new concept designs for acoustically improved carpets are
16
proposed and an initial level of evaluation is being conducted. A study of the process and
tools for optimizing and validating the proposed designs, that include new measurement
methods and software simulation, is conducted for the project, in Chapter 5. Virtual
modelling and validation of the acoustic material layers using acoustic modelling software is
the focus of Chapter 6. Meanwhile component level and vehicle level modelling and
evaluation of the carpet designs are described in Chapter 7, in detail. The selected concept
design is modelled and evaluated at vehicle level using Statistical Energy Analysis, and is
tested on-road; the results are analysed and discussed in Chapter 8. Optimization possibilities
and sensitivity analysis are conducted on the simulation models for further improvement of
the design, in Chapter 9. Chapter 10 discusses about few other possibilities of improving
acoustic performance of vehicle carpets, with further research in mind.
17
CHAPTER 2
BENCHMARK ANALYSIS AND TARGET SETTING FOR

VEHICLE CABIN NOISE
1 INTRODUCTION
This chapter outlines the initial vehicle benchmarking tests and discusses the vehicle
cabin noise test results. Further it is followed with a section for defining the design targets of
the improved vehicle carpet system, in a systematic approach. The vehicle benchmarking
tests measure and compare near field vehicle floor noise for various car models to analyse
variable factors and identify the areas to be improved.
Meanwhile, the results from the vehicle tests display the possibilities and potentials
of improving the vehicle carpet system to meet tangible targets for both physical and
functional requirements. A clear understanding of the listed manufacturing requirements and
the similar ones, apart from mere acoustic performance, helps in designing a relevant and
marketable solution for vehicle applications.
It is worth noting that emphasis is given to psychoacoustic and subjective evaluation

results, while analysing vehicle carpet systems. Hence, subjective analysis of the carpet
systems are included well enough in the vehicle cabin noise test procedures.
18
2 VEHICLE NOISE ANALYSIS
Initial benchmark testing for vehicle cabin noise in current vehicle models,
specifically Sedan models, is a good start in collecting data required for defining design
targets. A multi-point near field data acquisition of the noise distribution just above the
vehicle carpet surface is suggested to get a clear picture of the noise transmission paths and
noise isolation characteristics of the various vehicle carpets being tested.
The below figure depicts the various points that are appropriate for measuring the in-
cabin noise level, following a typical vehicle noise measurement procedure template. Note
that the number of measurement points can be optimized and grouped as required, and the
irrelevant measurement points (e.g. 1a, 1b and 1c) are not marked in the below diagram. The
specific measurement points used for each vehicle are listed in the corresponding test
procedure.
FRONT
Figure 2.1: Vehicle cabin noise mapping index for microphone placement
19
Two test vehicles of Sedan category are selected for vehicle cabin noise analysis 1) Audi A6
and 2) Hyundai Sonata. Audi A6 fits into the luxury class sedan series and while Hyundai
Sonata is from the low price class.
2.1 Hyundai Sonata vehicle cabin noise test
The test vehicle details are listed as below:
Hyundai Sonata 2005, Auto Sedan
2.4 littre, 4 cylinders, 16 Valves, TARE 1500
Reg. No. TLH 705,
Engine No. G4JS4138682
VIN No. KMHEM41DR5A162022
Tyre Pressures:
Spec. 30- 33 psi or 230 kPa
Actual Tyre Pressures:
Left Front: 32 psi
Left Rear: 33 psi
Right Front: 31 psi
Right Rear: 30 psi
A PCB accelerometer is glued on the top of engine cover to pick up tacho signals (Figure
2.2). Three G.R.A.S. microphones are at the respective locations (2-3-4 as in Figure 2.1) as
shown in Figures 4-5. The distance of microphone tips to the floor carpet surface will be 50
mm. One microphone was placed at the driver’s left ear as shown in Figure 2.5. The Bruel &
Kjaer Pulse laptop& the front end – the Bruel & Kjaer Pulse intelligent data acquisition
system is shown in Figure 2.3. The Hyundai Sonata testing vehicle was mounted on the
chassis dynamometer as shown in Figure 2.6.
20
Figure 2.2: RPM tacho sensor on top of engine block
Figure 2.3: B & K Pulse data acquisition system.
21
Figure 2.4: Microphones were placed in the front foot well area
Figure 2.5: A microphone was located at the Driver’s left Ear (DE).
22
Figure 2.6: Hyundai Sonata testing vehicle was mounted on the chassis dynamometer.
2.1.1 Test procedure
The tests were conducted for both the on-road and chassis dynamometer conditions.
Sound pressure data at idle in drive and neutral, and at constant driving speeds of 40 kph, 60
kph and 80 kph and at the 2nd gear slow acceleration were recorded twice with Air
Conditioner (A/C) off by the Bruel & Kjaer Pulse intelligent data acquisition system.
The same tests were repeated according to the following microphone placement
clusters as shown in Figure 2.4:
• Microphone locations 2, 3 & 4
• Microphone locations 2a, 3b & 4c
• Microphone locations 3d, 4f and at driver ear (DE) level
2.2 Hyundai Sonata - Test results and discussion
The sound pressure waterfall spectra measured from the microphone at DE (Figure 2.7)
shows that the on-road second gear slow acceleration test data correlates well with the
chassis roll second gear slow acceleration test data above 2500 rpm. Tacho buffer size needs
to be increased to catch the right data below 2500 rpm or the testing driver needs to be better
trained for the next test runs.
23
Working : Input : Run-Up : FFT Analyzer Working : Input : Run-Up : FFT Analyzer
60.0
4k
54.0
48.0
3.5k 42.0 3k
36.0
3k 30.0
24.0
2k
18.0
2.5k 12.0
6.00
2k 0.00
1k
0 400 800 1.2k 1.6k 0 400 800 1.2k 1.6k
[Hz] [Hz]
(a) On road 2nd gear slow acceleration (b) Chassis dynamometer, 2nd gear slow acceleration
Figure 2.7: Sound pressure level water fall spectrum at DE for the 2nd gear
slow acceleration tests (a) on the road and (b) the chassis dynamometer
The previous observation is confirmed by comparison of the order tracking results of

the on-road and chassis dynamometer tests for the second gear slow acceleration, focussing
on the engine combustion order of the 2nd as shown in Figure 2.8, where the engine is a four
stroke – four cylinder engine.
(a)
(b)
Figure 2.8: The 2nd order sound pressure level at DE for the 2nd gear
slow acceleration on the road (a) and the chassis dynamometer (b)
24
Figure 2.9 shows 1/12th octave band SPL readings at the Driver’s Ear (DE). Sound
pressure levels during the chassis dynamometer test are lower because the tyre-roller
interaction is limited on the front wheels.
[dB(A)/20.0u Pa] Autospectrum(mic2) - Mark 3 [dB(A)/20.0u Pa] Autospectrum_40(mic2) - Mark 2

Working : RT-S80-ACOff_3_set3 : Input : CPB Analyzer Working : 40 AC off : Input : CPB Analyzer
60 60
40 40
20
▬ 40 kph 20
▬ 40 kph
▬ 60 kph ▬ 60 kph
▬ 80 kph ▬ 80 kph
0 0
63 250 1k 4k 63 250 1k 4k
[Hz] [Hz]
(a) (b)
Figure 2.9: Sound pressure level at DE for constant speeds of 40, 60 and 80
km/h on the road (a) and the chassis dynamometer (b).
As shown in Figure 2.10, the air flow noise from the car air conditioning unit is in the
frequency range less than 300 Hz.
[dB(A)/20.0u Pa] Autospectrum_ID_ACOn(mic2) - Mark 1

Working : ID AC on : Input : CPB Analyzer
60
40
20
30 100 300 1k 3k 10k

[Hz]
Figure 2.10: Sound Pressure Auto-spectrum at DE for Idle in Drive_AC On & Off
25
During the on-road tests, the front portion of the car floor carpet (eg. front foot well
area) is mainly subjected to high SPL, low frequency power train noise as shown in Figure
2.11. whilst higher SPL of high frequency noise (tyre-road interaction noise) is concentrated
towards the rear part of the floor carpet (e.g. wheel arch and boot areas) as shown in Figure
2.12. This noise distribution trend applies to all the on-road tests with the SPL increasing
with the prescribed driving speeds.
The above findings imply that sound barriers should be applied to the front part of the
carpet to block low frequency noise and high frequency absorbing materials should be
incorporated in the rear areas of the car carpet assembly.
0.00759
0.00506
0.00253
0.0
Figure 2.11: The SPL distribution on the car carpet; 2nd gear slow acceleration on-road
test.
26
Figure 2.12: The SPL distribution on the car carpet; 40 kph constant speed on-road
test.
2.3 Audi A6 vehicle cabin noise test
The second test vehicle for benchmarking and noise mapping is Audi A6, which
relatively falls into the luxury class of sedan category. The test vehicle details are listed as
below and Figure 2.13 is a picture of the sedan:
Audi A6 2007, Auto Sedan
2.0 litre, 4 cylinders, FSI Turbo
Reg. No. UMF 487,
VIN No. WAUZZZ4F17N055308
Odometer Reading: 8749 km
Tyre Pressure Specs: 29 psi (front & rear) under normal conditions.
27
Figure 2.13: The Audi A6 test vehicle
For each set of measurements, three (3) G.R.A.S. microphones were used at the
locations as shown previously in Figure 2.1. The distance of microphone tips to the floor was
50 mm as shown in Figure 2.3. One microphone was placed at the driver’s left ear level as
shown in Figure 2.5. The KMT RPM-8000 was used to monitor engine rpm. The Bruel &
Kjaer Pulse laptop & the front end – the Bruel & Kjaer Pulse intelligent data acquisition
system is shown in Figure 2.3.
2.3.1 Test procedure
In-car Sound Pressure Level (SPL) measurements using the Bruel & Kjaer Pulse intelligent
data acquisition system were conducted during on-road tests for various driving conditions,
to wit:
• Constant speed 40 kph on-road (with and without mats)
• The second gear slow acceleration on-road (with and without mats)
• Idle in-drive AC on and AC off, (with and without mats)
28
The KMT RPM-8000 was used to tacho the engine rpm during the tests. The output of the
RPM-8000 was recorded and integrated in the Bruel & Kjaer Pulse intelligent data
acquisition system.
To create an SPL contour map of the floor carpet surface, the above tests were
repeated according to the following microphone placement clusters as shown in Figure 2.1:
• Set 1: microphone locations 2, 3 & 4
• Set 2: microphone locations 2a, 3b & 4c
• Set 3: microphone locations 3d, 4f and at driver ear (DE) level
The ME’scopeVES software was then used to generate the SPL contour map of the
floor carpet surface using the microphone SPL data from Pulse. Visualization techniques
through the use of movie clips from ME’scopeVES were used for comparative analysis of
the carpet’s acoustic performance.
2.4 Audi A6 – Comparative test results and discussion
Comparative sound pressure waterfall spectra (Figure 2.14) measured from the
microphone at DE shows that the on-road second gear slow acceleration test data from the
Audi A6 correlates well with that of the Hyundai Sonata. This run-up test represents
transient conditions occurring during typical driving situations.
Working : Input : Run-Up : FFT Analyzer Working : Input : Run-Up : FFT Analyzer
60.0
4k
4k 54.0
48.0
3.5k 42.0 3.5k
rpm
36.0
rpm
30.0 3k
3k
24.0
18.0
2.5k 12.0 2.5k
6.00
2k 2k
0.00
0 400 800 1.2k 1.6k 0 400 800 1.2k 1.6k
[Hz] [Hz]
(a) Audi A6 (b) Hyundai Sonata

Figure 2.14: Sound pressure level waterfall spectrum at DE for the 2nd gear
acceleration on the road for (a) Audi A6 and (b) Hyundai Sonata.
29
Further comparison of the order tracking results of the on-road second gear slow
acceleration tests for the engine combustion signature (the 2nd order) as shown in Figure
2.15, suggests that the Audi A6 has a slightly higher total SPL than the Hyundai Sonata. This
can be attributed to the turbo charged engine of the Audi A6. The small inconsistency in 2nd
order tracking graph for the Audi A6 (Figure 2.15a) below 2500 rpm may be the result of
synchronization lapses from the KMT RPM-8000.
(a)
(b)
Figure 2.15: The 2nd order sound pressure level at DE for the 2nd gear slow
acceleration on the road (a) Audi A6 and (b) Hyundai Sonata
Figure 2.16 shows the comparative 1/12th octave band SPL readings at the Driver’s
Ear (DE) level for the Audi A6 during the constant speed on-road tests. The trend of the SPL
increasing with driving speed holds true. Audi A6 registered a slightly higher total SPL
compared to the Hyundai Sonata (see Table 1). However, the discrepancy is only on the
order of 1-2 dB(A).
30
[dB(A)/20.0u Pa] Autospectrum(mic2) - Mark 3

Working : CS_100 : Input : CPB Analyzer
60
40
20
▬ 40 kph
▬ 60 kph
0 ▬ 80 kph
▬ 100 kph
-20
63 250 1k 4k
[Hz]
Figure 2.16: Sound pressure level at DE for constant speeds of 40, 60, 80 and
100 km/h on-road test for Audi A6.
On Road Constant Speed Test
Speed Total Sound Pressure Level [dB(A)] at DE
Hyundai Sonata Audi A6
40 kph 63.0 64.7
60 kph 65.6 66.8
80 kph 67.6 69.5
100 kph 70.9 70.3
Table 1. Comparative total SPL’s during on-road tests
Figure 2.17 shows the comparative SPL perceptible at DE for the idling conditions. It
should be noted that the introduction of the rubber foot mat have no significant effect on the
perceived SPL. This would indicate that the acoustic performance of the Audi A6 carpet is
very good.
31

Working : IdleTest_noMat : Input : CPB Analyzer
60
with mat
40
20
no mat
-20
63 250 1k 4k
[Hz]
Figure 2.17: Sound Pressure Auto-spectrum at DE for Idle in drive and A/C off, with
and without mats.
During the on-road tests, the front portion of the car floor carpet (eg. front foot well
area) is mainly subjected to high SPL of low frequency (power train) noise as shown in
Figure 2.18, whilst high SPL of high frequency noise (tyre-road interaction noise) is
concentrated towards the rear part of the floor carpet (e.g. wheel arch and boot areas) as
shown in Figure 2.19. This noise distribution trend applies to all the constant speed tests with
the SPL magnitudes increasing with the prescribed driving speeds.
FRONT
Hyundai Sonata
Audi A6
Figure 2.18: Comparative low frequency (69 Hz) SPL distribution in the car carpet
during 80 kph constant speed on-road test.
32
FRONT
Hyundai Sonata
Audi A6
Figure 2.19: Comparative high frequency (5163 Hz) SPL distribution on the car carpet
during 80 kph constant speed on-road test.
The same carpet SPL mapping trend applies during the acceleration run up as shown
in the comparative Figures 20 & 21. It should be mentioned that on run-up (slow
acceleration) tests, individual car responses (e.g. spectral peak responses) are unique such
that point-to-point comparisons are not always possible. Nevertheless, the carpet surface SPL
mapping trends illustrated in Figures 20 & 21 are typical.
FRONT
Hyundai Sonata
Audi A6
Figure 2.20: Comparative SPL distribution on the car carpet during slow acceleration
on-road test just after starting (low rpm; 840 rpm ≈ 28Hz ).
33
FRONT
Hyundai Sonata
Audi A6
Figure 2.21: Comparative SPL distribution on the car carpet during slow acceleration
on-road test at high engine rpm (3300 rpm ≈ 110Hz).
The above results indicate that at lower frequencies, where engine noise is mainly
concentrated on the frontal areas, the performances of the floor carpets of both test cars are
comparable. At 110 Hz, the Audi A6 floor carpet performed better than the Hyundai in
reduction of power-train noise in the front part of the floor carpet as evidenced by the
reduced ‘hot spots’ (incidence of high SPL) as shown in Figure 2.21.
2.5 Psycho-acoustic and sound quality evaluation
The Head Acoustics HMS III (Figure 2.22) mounted on the front passenger seat of the
vehicle was used to measure and record the various sound quality parameters discussed in
this report.
34
Figure 2.22: Head Acoustics HMS III.
Six (6) sound quality parameters, directly recorded and analysed using the Head
Acoustics HMS III, are listed and discussed herein. The parameters are Articulation Index,
Loudness, Fluctuation strength, Roughness, Sharpness and Tonality. Of these parameters, the
first three (3) on the list are considered to be of major importance for purposes of our
evaluation. Basic description of these sound quality and psychoacoustic parameters are found
in Appendix 1 and the comparative parameter plots are included in Appendix 2. The sound
quality evaluation is indicative of the overall performance of the car cabin acoustic package.
The 6th octave band sound pressure spectra were measured from the right ear
microphone of the HMS III as shown in Figure 2.23. Figure 2.24 shows that the on-road
constant speed test results of the Audi A6 correlate well with those of the Hyundai Sonata.
Total SPL variation is in the order of 1-2 dB(A) and the total SPL increasing with higher
driving speeds. The introduction of the foot mats to the Audi A6 has no significant effect on
the perceived SPL in the car interior (Figure 2.24).
35
Figure 2.23 The 6th Octave Band Spectra
6th Octave Band Spectra Comparison
80
70
dB (A)
60
50
40
40 60 80 100
Driving Speed (kph)
Audi A6-No Mat Audi A6-With Mat Hyundai
Figure 2.24: The 6th Octave Band Spectra comparison.
The articulation index (AI) denotes the ease of holding conversation within an
acoustic enclosure. The articulation index is affected by noise, interference, and distortion.
The higher the articulation index, the more intelligible is the normal conversation. Figure
2.25 show that the Audi A6 has higher AI values than the Hyundai Sonata inside the vehicle
cabins. The AI deteriorates with increasing driving speed due to increased noise levels.
The low AI result for the Hyundai Sonata at 40kph constant speed may be due to
inherent flaws in the suspension or chassis design. A poorly responsive suspension for
example, can generate resonance resulting to added structural borne noise. The higher AI
36
exhibited by the Audi A6 can not be explicitly linked to its floor carpet assembly, but it can
be surmised that the Audi A6 has a superior overall interior acoustics package.
Articulation Index Comparison
100
Articulation Index (%)
80
60
40
20
0
40 60 80 100
Driving Speed (kph)
Figure 2.25: The Articulation Index comparison.
In terms of loudness index, the Audi A6 cabin interior proved to be better than that of
the Hyundai Sonata. Figure 2.26 shows that the comparative loudness index indicated about
10 sonesGF difference between the test cars. The lower loudness index of the Audi A6
(hence quieter) provided a positive impact on the AI as previously discussed.
Loudness Comparison
40
Loudness (SonesGF)
30
20
10
0
40 60 80 100
Driving Speed (kph)
Figure 2.26: The Loudness Index comparison.
37
The fluctuation strength is related to low frequency modulation and Figure 2.27
shows that the Audi A6 registered a slightly higher fluctuation strength index than the
Hyundai Sonata. In addition, the variation in the Audi’s fluctuation strength indices for the
“with mat” and “without mat” cases were not significant.
Fluctuation Strenght Comparison
0.2
Fluctuation Strenght
0.15
(vacil)
0.1
0.05
0
40 60 80 100
Driving Speed (kph)
Figure 2.27: The Fluctuation Strength comparison.
The comparative sound roughness indices inside the test car cabins are shown in
Figure 2.28. Apart from the extreme case for the 40kph constant speed, the Hyundai Sonata
had a higher roughness than the Audi A6. The Hyundai Sonata exhibited a trend of elevated
roughness index with increasing driving speed which impact negatively on the AI. Since
roughness is mostly associated with low frequency sounds, the comparative results indicate
that the Audi A6 acoustic package handles engine noise better than that of the Hyundai
Sonata.
Roughness Comparison
4
Roughness (asper)
0
40 60 80 100
Driving Speed (kph)
Figure 2.28: The Roughness comparison.
38
The same trends and elevated results of the sharpness index are found for both the
Audi A6 and the Hyundai Sonata as shown in Figure 2.29. A better “balanced” sound
envelop contributes well to the overall perception of a pleasant acoustic environment. The
high reading for the Hyundai Sonata at 40kph may be the result of suspension and chassis
design flaws as mentioned earlier.
Sharpness Comparison
3
Sharpness (acum)
0
40 60 80 100
Driving Speed (kph)
Figure 2.29: The Sharpness comparison.
There are no appreciable differences in the comparative tonality indices between the
two test cars as shown Figure 2.30. The high measurement value for the Hyundai Sonata at
40kph may be attributed to reasons earlier mentioned. The tonalities for the Audi A6 with
and without the “mats” are almost identical.
Tonality Comparison
0.15
Tonality (tu)
0.1
0.05
0
40 60 80 100
Driving Speed (kph)
Figure 2.30: The tonality comparison.
39
3 TARGET DEFINITION
3.1 The carpet system

The vehicle floor carpet assembly refers to the whole floor covering attached to the
dash partition and floor pan of the vehicle. The basic function of the floor carpet is to provide
suitable cover (e.g. to the bare metal car floor) and comfort to the occupant as well as offer
shock, thermal, and acoustic insulation. The carpeting is an integral component of the
aesthetics of the car interior. It can enhance the perception of a luxurious ambiance to the
car cabin interior.
The typical floor carpet is constructed with multiple layers of materials. The major
carpet components are the tufting and the carpet backing. The former is usually made from
woven natural or synthetic materials while the later can be comprised of one or more layers
of suitable materials such as felt, foam or latex depending on the intended application. Floor
carpet components are mostly bonded together by thermoforming and pressing. The finished
product is then cut and finished to size. For standardized applications, the carpet can be
moulded to shape and form to fit the surface application.
3.2 Carpet design objectives and constraints

The goal of this R&D project is to design, develop and validate an innovative car
carpet system with improved acoustic performance. The proposed solutions must promote
competitive advantage in areas of acoustic performance, cost effectiveness, product weight
and waste reduction, and propose environmentally friendly and sustainable manufacturing
solutions. The objectives of the vehicle carpet system design can be categorized into a)
Physical characteristics and b) Functional requirements, as given below.
3.2.1 Objectives - Physical Characteristics
The new carpet system shall feature improvements or an alternative design to the
conventional vehicle floor carpet’s heavy layer construction and under layer material. The
new design includes a novel composite layered structure with improved noise reduction
performance tailored to be compatible with the Futuris’ “footprint” concept. The introduction
of new heavy layer components, layouts, and structural arrangements will be considered in
the design.
40
Materials
New materials or combination of materials that will constitute the new product design
will be explored whilst focusing on improving acoustic performance as the primary design
consideration.
Surface layer, Tufting / Velour Needle Punching
The look, feel and wear durability are strong drivers in the selecting the carpet
surface. These may override acoustic performance criteria. CSIRO does not have any
tufting or velour needle punching equipment. Making experimental samples would need to
be carried out with a manufacturer. Thus it would be CSIRO’s preference to concentrate on
the under layers where it has the capability to produce pilot scale sized samples.
Underlay
Improvements to noise reduction potential of the carpet underlay shall be investigated

through the use of new materials and component arrangements (layering). Particular interest
shall be on the construction of improved barrier for low frequency noise control and its
selective application to identified critical floor areas. In addition to conventional materials
(PU, PEP, felt, latex…), new materials (scrims, microfibre) and alternately methods of
construction shall be explored.
Use of microfibres is known to increase the performance of sound absorption

particularly at higher frequencies, primarily due to a higher surface area and the potential for
a higher tortuosity to be engineered into the construction. There are a number of approaches
to creating micro fibre layers and blends worth considering. Namely electro spun and split
able micro fibres. There are potential novel properties to be exploited with the use of micro
fibres, for example it may be possible to have fibres with a range of mechanical resonances
in the lower frequency range (100-500Hz). The resonant frequencies are dependent on fibre
material properties such as Young’s Modulus, fibre diameter and effective length. These
desirable material properties have the potential to be engineered. It is likely that a
multilayer system will be required for the underlay.
Material thermoforming property
In order to facilitate manufacturing and/or production trials to set thermoforming

machine temperatures, the new material melting points should be defined.
41
Carpet Design
Carpet design will have to meet vehicle floor pan packaging requirement (assuming
Mitsubishi 380 as baseline for example). Design will have to provide acoustic, comfort and
aesthetic characteristics. It must be durable, safe and versatile. Preferably, it must be
recyclable and low cost.
Carpet manufacturing process
Carpet manufacturing starts from tufting, latex/PE coating, heavy layer lamination,
heating, moulding, foot pad bonding, foaming and water-jet trimming.
The new design will probably involve additional processes for lamination of air flow
resistance layers and microfibre/scrim before thermoforming. The heavy layer may be
replaced by light weight heavy layers or air flow resistance layers. The material melting
points will have to be considered during heating. After heating, standard processes for
trimming and sizing will follow.
Both RMIT and CSIRO do not have production facility. Production trials will have to
be conducted in Futuris’ manufacturing facility.
Quality
Material or component quality must conform to Futuris and OEM clientele

specifications as well as all applicable standards:
• Australian / New Zealand Standards
1. AS/NZS 2111.0-16:1996 (Textile floor coverings)
2. AS/NZS 2455.2:1996 (Textile floor coverings – Installation practice)
• British Standards Institute / International Standards
1. BS ISO 11859:1999, BS EN 1307:2005, DD CEN/TS 14159:2004, BS EN

14499:2004
2. BS 5229:1975
3. ISO 11859:1999, ISO 1763:1986, ISO 2550:1972
4. ASTM D3676-01, ASTM D6004-04, ASTM D6005-03, ASTM D6325-98(2004)
• European Standards
1. IS EN 1307:2005, IS CEN TS 14159:2005, IS EN 14499:2004
42
2. DIN EN 1307 (2005-05), DIN EN 14499 (2005-02)
Weight Reduction
Selection of new materials for the underlay (heavy layer), new design, and the use of
alternative methods of manufacture is intended to provide improvements towards overall
weight reduction. Should new and lighter materials (eg. microfibre components) with
favourable acoustic performance can be utilized, it is likely to lead to a substantial mass
reduction of the carpet system.
Design constraint
Current Futuris carpet product has a non-permeable heavy layer barrier under the
tufting textile which insulates the cabin from exterior noise. Due to a lack of the tufting
facilities, working only on the carpet under-layer development may limit carpet acoustic
performance enhancement if the carpet heavy layer backing design is not changed. However,
the removal of the heavy layer backing from the current carpet design may be constrained by
other negative commercial effects on Futuris.
3.2.2 Objectives - Functional Requirements
Acoustic Performance
The new vehicle floor carpet system is designed to improve acoustics performance
inside the vehicle cabin by offering targeted solutions to problematic acoustic areas, to wit:
• Low frequency airborne noise reduction solutions through the use of light weight
heavy layers/air flow resistance layers plus air gap in between the component layers,
or barriers integrated with mat application/inner dash insulator or as a separate and
specific structural treatment – selectively applied to the structure.
• High frequency airborne noise control through better absorption capability by the
porous carpet such as moulded cavity absorber or air gap between air flow resistant
components inside the moulded floor carpet.
• Improved vibration damping at low frequency and thus reduced low frequency noise
infiltration or propagation through the use of porous control layers in place of heavy
layer barriers.
43
The consolidated acoustic performance targets on random incidence absorption coefficient

(Figure 2.31) and sound transmission loss/insert loss (Figure 2.32), adopted from the Project
Objectives and Constraints Report (0407M2) are presented below.
Figure 2.31: Random incidence sound absorption coefficient design target
Figure 2.32: Sound transmission loss/insert loss design target
44
Comfort and Aesthetic Appeal
The look and feel of the new carpet surface must conform or exceed the typical
characteristics of current Futuris carpet products or OEM conventional vehicle floor carpets.
The appearance of the conventional tufting will be preserved whilst improving the overall
acoustic performance of the carpet assembly.
Safety
The product must conform to various safety standards concerning flammability,

slip/non-slip, volatile organic compounds emissions, moisture retention, anti-microbial
activity, cleaning, and electrical insulation, to wit:
1. HB 37.5-1995 (fire safety)
2. ASTM D7339-07 (volatile organic compounds emission)
3. ASTM E2471-05 (anti-microbial activity)
4. AS/NZS 3733:1995, BS 4088-1.2 -4.0:1994-95, BS 7460:1991, ASTM F1284-04,

ASTM F608-03, ASTM F655-06 (cleaning)
5. CSA C22.2.222:1986 (under-carpet wiring)
Durability
The new carpet system shall have a product lifespan of 30 years.
Versatility
The overall design for the new carpet system shall be adaptable to most car flooring
layouts. The integration of the “footprint” concept must not limit the products applicability to
specific vehicle cabin design and shall entail minor adjustments in manufacturing methods.
Service Environment
Vehicle cabin normal operating conditions:
Temperature - -40°C - +50°C
Relative Humidity - 40 to 85%
45
Environmental Impact
The use of natural fibre and/or recyclable material for the underlay will be pursued.
Waste minimization will be addressed through the use of recycled materials and minimizing
the number and type of material components of the composite layer. Sustainable
manufacturing will be a prime consideration by means of utilizing locally available and/or
selected natural material components whenever possible.
Environmental considerations will be factored-in during material selection. There is

potential for natural fibres and also for constructing multilayer systems from polymers of
compatible chemistry which will allow for the easy reclamation and recycling at the end of
product life cycle.
Cost and Manufacturability
Cost considerations will be tackled through the combination of simplified or integrated

manufacturing techniques (fewer steps, simplified/integrated assembly schemes), use of new
materials (resorting to fewer components or material quantity which translate to product
weight reduction), and local raw material procurement whenever feasible.
3.3 Design validation
The above new carpet design and development will have to be validated through carpet
product APQP process following the test verification matrix. RMIT has the capability to
validate the designed carpet acoustic performance using AUTOSEA 2 software simulation of
material component performance coupled with in-vehicle testing.
Impedance tube test and AUTO SEA 2 material simulation will support CSIRO
material development to gain an understanding of the materials acoustic performance and the
interaction, and the importance of a variety of material parameters which can be adjusted to
engineer and optimise acoustic performance of the material. AUTO SEA 2 parts or
component simulation will be used to validate and optimise component design. RMIT
intends to develop a simple, portable and reliable methodology or procedure for carpet
acoustic evaluations and measurements.
For other carpet design performance such as self-ignition, durability, etc., Futuris
development engineers will have to send the parts to relevant NATA certified test labs for
testing validations.
46
4 CONCLUSIONS
Hyundai Sonata cabin noise analysis
• Low frequency power train noise has higher sound pressure level (SPL) distribution
in the front part of the carpet (front foot well area) which may need more sound
barriers.
• High frequency tyre-road interaction noise has higher sound pressure level (SPL)
distribution in the rear part of the carpet (rear wheel arch and the boot areas) which
may need more sound absorption measures.
• The on-road second gear slow acceleration test data correlates well with the chassis
roll 2nd gear slow acceleration test data above 2500 rpm. Testing driver needs to be
better trained or tacho buffer size needs to be increased for the improved data quality
for the next test runs.
• AC noise is in the frequency range below 300 Hz.
Audi A6 cabin noise analysis and sound quality
• The introduction of foot mats over the floor carpet has no significant effect on the
overall acoustic performance of the Audi A6 floor carpet assembly.
• The Audi A6 test car registered a slightly higher total SPL (1-2 dB(A) variation) than
the Hyundai Sonata test car under various constant speed driving conditions
• The Audi A6 acoustic package performed better than the Hyundai’s in terms of other
indices like articulation, loudness, roughness, and sharpness. For fluctuation strength,
the Hyundai returned a marginally better performance. The tonality index results are
almost identical for the two test cars
• Overall, the acoustic package of the Audi A6 performed better than that of the
Hyundai Sonata
47
APPENDIX
SOUND QUALITY PARAMETERS [42]
1. Articulation Index
The Articulation Index indicates the extent to which noise reduces

intelligibility of speech. Intelligibility depends on level and the frequency of the
background noise. The noise level in the vehicle cabin is thus the factor for
intelligibility between vehicle occupants.
Language and the audible range of human hearing can be represented as an

area in the vehicle spectrum, limited on one hand by sound pressure level (whispering
to shouting), and, on the other hand by sound frequency (200 Hz to 6300 Hz).
2. Loudness
A way to quantify loudness is to relate the sensation stimulus to a known

standard sound by asking subjects how much louder or softer a test sound is. This
method allows subjective loudness to be placed on a linear scale (as opposed to the
logarithmic scale of the deciBel).
The standard for this is a 1kHz tone with a sound pressure of 40dB. This
reference loudness value is made equal to one (1) sone. Sones GF is the loudness
index measured in a free field and sones GD is measured in a diffuse field. For a 1
sone, 1kHz tone to sound twice as loud, its sound pressure must be increased by
~10dB which corresponds to a loudness of 2 sones. So that approximation is used in
the definition of the phon: 0.5 sone = 30 phon, 1 sone = 40 phon, 2 sone = 50 phon,
4 sone = 60 phon, etc, where phon is another unit of loudness.
Wouldn't it be great to be able to convert from dB to sones (which

approximate loudness as perceived by people)? As a crude approximation, it can be
said that the A weighting curve approximates the human frequency response at low to
moderate sound levels, so dB(A) is very roughly the same as phons. The logarithmic
relation between sones and phons is shown below.
48
3. Fluctuation Strength
Fluctuation Strength measures the sensation of “slow moving” modulation.

Sounds modulating at 20 Hz and below can be perceived as fluctuating over time. A
good example is a pair of closely spaced tones which cause beating.
The rate of fluctuation, the sound pressure level, and degree of modulation
influence fluctuation strength. The reference sound for fluctuation strength: 1kHz
tone at 60 dB SPL 100% modulation at 4 Hz gives the reference fluctuation strength
= 1 vacil.
4. Roughness
Rough sounds are rapidly and repetitively fluctuating noise that contains tones
spaced within a critical band amplitude. Sounds which contain modulation between
about 20Hz and 200 Hz are considered to be rough-sounding. The sensation of
roughness is not limited to true modulating sounds. Noises (broad-band & narrow-
band) may also be perceived as rough due to a modulated amplitude envelope.
Parameters important to roughness include the degree of modulation (AM),

frequency modulation index (FM) and modulation frequency. Sensitivity to
roughness peaks at approximately 70 Hz modulation rate. The modulation frequency
range for roughness is ~ 20 Hz - 200 Hz. For centre frequencies at and above 1kHz,
the peak roughness sensation occurs at 70 Hz. For centre frequencies below 1kHz,
the peak roughness depends upon the width of the critical band.
The reference sound for roughness is: 1 kHz tone at 60 dB, 100% AM
modulated at 70 Hz gives the reference roughness equal to 1 asper.
5. Sharpness
The basic description of sharpness is the ratio of high frequency level to

overall sound frequency level. “Centre of gravity” on frequency scale of spectral
envelope describes sharpness: the higher the “centre of gravity”, the sharper the
sound.
Integration of specific loudness multiplied by a weighting function, divided

by total loudness (hence, sharpness is level-independent). Normalized to a reference
sound, a narrow band of noise centred at 1kHz at a level of 60dB and a bandwidth of
160Hz, has an agreed sharpness value of 1 acum.
49
6. Tonality
Sounds can be perceived as tonal when they contain pure tones or noise with
bandwidths less than 1 critical band. Tonal sounds exhibit voiced component(s) and
periodicity. Non-tonal sounds are noise-like and non-periodic.
Tonality is a measure which can be used to evaluate the overall tonal

characteristics of a sound. Takes into account: bandwidth of tonal components,
frequency, loudness level with and without tonal components. The reference sound
for Tonality: 1kHz tone at 60dB equals 1 tu (tonality unit).
50
CHAPTER 3
FORCED VIBRATION ANALYSIS OF PANEL SYSTEMS AND

SOUND RADIATION DIRECTIVITY MODEL
1 INTRODUCTION
In this chapter the vehicle carpet system is analysed acoustically using simple, but
effective, mathematical models like the Simple Harmonic Motion model. The study helps to
quickly identify the acoustic weaknesses like resonant frequencies of the system, where the
noise isolation is the minimum, and where could be the focus of development or research.
Further, the study also describes a theoretical method for predicting the directivity of
the sound radiated from a panel or opening excited by sound incident on the other side. This
directivity needs to be known when predicting the sound level at a particular position due to
sound radiation from various kinds of panels like a vehicle floor panel. This is equally
applicable to panels like factory roof, wall, ventilating duct or chimney flues. There is
surprisingly little information on how to predict this directivity in the scientific literature.
Most of this information is based on limited experimental data or its basis cannot be
determined.
51
2 SIMPLE HARMONIC ANALYSIS OF VEHICLE FLOOR TRIM

PANEL
The current vehicle carpet system in production is the best starting point for both
conceptual design and functional analysis studies. The current production carpet we have is
based on a three material layer design.
The three layers of the current carpet system are from top to bottom, carpet tufting
layer, heavy sound barrier layer and the vibration de-coupler/absorber layer.
Cabin side noise Tufting layer

Heavy layer (barrier)
Noise path
(from engine) De-coupler/absorber layer
Figure 3.1: Current vehicle carpet design
The materials applied in the current production are,
Carpet tufting layer - 585 gsm carpet fabric & substrate
Heavy layer - 1.7 kg/m2 heavy layer
De-coupler/absorber - 1.69 kg/m2 Poly Urethane Foam
The carpet system can be acoustically analyzed by considering it as a spring-mass

system in a Simple Harmonic Oscillator (SHM) case, where the de-coupler resonance
frequencies are derived from the well established SHM mathematical solution.
52
r s
Figure 3.2: Simple Harmonic Motion (SHM) oscillator equivalence
x 1 m
= 2 (3.1)
F s - mω + jr
jηωω N
Where, “s” it the stiffness of unit area of the carpet panel or the de-coupler layer and
η is the damping loss factor of the carpet panel.
The resonant frequency is when the denominator is minimum, i.e. when
s
ω = ω0 = rad/s (3.2)
m
In the case of the carpet system,
m = GSM de-coupler /2 + GSM tufting + GSM heavy layer (3.3)
GSM de-coupler = ρ de-coupler * t de-coupler (3.4)
= 49.96 kg/m3 * 0.034m = 1.6988 kg/m2
53
GSM tufting = ρ tufting&substrate * t tufting (3.5)
= 90 kg/m3 * 0.0065 = 0.585 kg/m2
GSM heavy layer = ρ heavy layer * t heavy layer = 1.7 kgsm (3.6)
Where, GSM is equivalent surface density in gsm unit, ρ is density and t is the thickness.
Therefore,
m = 3.1344 kg/m2
s = E de-coupler / t de-coupler (3.7)
= 439.12 * 103 / 0.034 = 12.953 * 106 N/m3
where, E de-coupler is the Young’s modulus (unit area) of the de-coupler layer
and t de-coupler is the thickness of the de-coupler layer.
ω0 = √ s/m (3.8)
= √ (12.953 * 106 / 3.1344 kg/m2 )= 2032.86 radians/second
f0 = ω0 / 2π = 323 Hz (3.9)
Hence, from Simple Harmonic Motion analysis method, the current carpet system is
expected to show a normal incidence resonance frequency at 323 Hz, which is definitely in
the power train or engine noise source frequency range. The carpet panel parameters like
Young’s modulus, density, damping loss factor and GSM were measured in CSIRO material
laboratory.
54
The below graph is a log frequency - amplitude of the x/F transfer function variable
values of the current carpet. The resonant peak at 323Hz is clear on the transfer function plot,
which is the expected weak link of the current carpet acoustic design.
Vehicle floor trim SHM response (damped)

s
ω = ω0 = rad/s
10 m
0 323 Hz x 1 m
= 2
-10 F s - mω + jr
jηωω N
Amplitude response (x/F)
-20
-30
Carpet res pons e (SHM dam ped)
-40
-50
-60
-70
-80
100.0 1000.0 10000.0
fo Frequency (Hz)
Figure 3.3: Simple Harmonic Motion (SHM) oscillator equivalence
Transmission Loss measurements are conducted on the variations of the current

carpet design using the B&K impedance tube method, explained in Chapter 5. The results are
plotted in the below graph for samples B to K of the current production carpet design.
Figure 3.4: Transmission Loss results of the current carpet design variants
55
It is clear from the Transmission Loss results that the carpet design has an evident
resonance dip or acoustic weakness at approximately 323 Hz frequency. This is the same
case for majority of the samples, with small deviations introduced by factors like the varying
GSM of the heavy layer and the fact that the simple spring-mass model is not exact.
3 FINITE SIZED PANEL RADIATION – THEORY
A two dimensional method for calculating the directivity of the air-borne radiation of
sound from a panel or opening whose vibration is excited by the incidence of sound from the
other side, like power train or tyre noise sources, is being presented. The directivity of the
radiation depends on the angular distribution of the incident sound energy (w(φ)) or angular
weighting function. The proposed method for deriving the angular weighting (w(φ)) is
generic and so a variety of cases are studied. For panels or openings like in the wall of a
cabin or room, the angular distribution of the incident sound energy is predicted using a
physical model which depends on the sound absorption coefficient of the cabin/room
reflective surfaces. For an opening, like at the end of a duct, the sound absorption coefficient
model is used in conjunction with a two dimensional model for the directivity of the sound
source in the duct. The finite size of the panel is taken into account by using a two
dimensional model for the real part of the radiation efficiency of the finite size panel or
opening.
Transmitted sound
Finite
opening or
Zwft (wave fluid transmit side) panel
θ
Zwp (wave panel)
Infinite baffle
Zwfi (wave fluid incidence side)
Incident sound Normal
Figure 3.5: Panel or opening radiating sound due to forced excitation on the incident side
56
The directivity of the radiation depends strongly on the length of the radiating object
in the direction of the observer and only slightly on the width of the object at right angles to
the direction of the observer. For panels, the plate wave impedance of the panel is used.
Above its critical frequency or normal incidence resonance frequency a single panel radiates
strongly at the angle at which coincidence occurs, i.e. when the angular wavelength
coincides with the longitudinal modal frequencies of the panel. The method is compared with
published experimental results
The effective impedance Z e (φ ) of a finite panel in an infinite baffle to a plane sound

wave incident at an angle of φ to the normal to the panel is [45]
Ze(φ) (effective) = Zwft(φ) wave fluid transmit side + Zwfi(φ) wave fluid incidence side + Zwp(φ)wave panel (3.10)
where
Z wfi (φ ) is the wave impedance of the fluid as experienced by the finite panel in an
infinite baffle, whose vibration is due a plane sound wave incident at an angle of φ to the
normal to the panel, on the side from which the plane sound wave is incident (this is the fluid
loading on the incident side),
Z wft (φ ) is the wave impedance of the fluid as experienced by the finite panel in an
infinite baffle, whose vibration is due a plane sound wave incident at an angle of φ to the
normal to the panel, on the side opposite to which the sound is incident (this is the fluid
loading on the non-incident or transmitted side) and
Z wp (φ ) is the wave impedance of the finite panel in an infinite baffle to a plane sound
wave incident at an angle of φ to the normal to the panel, ignoring fluid loading.
It will be assumed that the fluid wave impedances on both sides are the same and the
imaginary part of the fluid wave impedance will be ignored [43]. That is
Z wfi (φ ) = Z wft (φ ) = ρ cσ (φ ) (3.11)
57
where ρ is the density of the fluid, c is the speed of sound in the fluid and σ (φ ) is the
radiation efficiency into the fluid of one side of the finite panel in an infinite baffle, whose
vibration is due a plane sound wave incident at an angle of φ to the normal to the panel.
Transmitted sound
Finite
opening or
panel θ
σ(φ)(radiation efficiency)
Infinite baffle
2a
Incident sound
Normal
Figure 3.6: Sound incident at an angle φ to the normal to a panel and radiated at an angle of θ
Reflections at the panel edges are ignored [43]. The rms normal velocity vrms (φ ) of the
panel due to a plane sound wave incident at an angle of φ to the normal to the panel which
exerts an rms pressure pirms (φ ) is
pirms (φ )
vrms (φ ) = . (3.12)
2 ρ cσ (φ ) + Z wp (φ )
Ze(φ) (effective)
The transmitted rms sound pressure ptrms (θ , φ ) which is radiated by the panel on the
non-incident side to a receiving point, which is at an angle of θ to the normal to the centre of
the panel and a large distance from the panel, is given [44]
58
sin  ka ( sin θ − sin φ ) 

ptrms (θ , φ ) ∝ vrms (φ ) (3.13)
ka ( sin θ − sin φ )
where k is the wave number of the sound and 2a is the length of the panel in the
plane of the source and observation points. Thus, substituting for vrms (φ )
pirms (φ ) sin  ka ( sin θ − sin φ ) 

ptrms (θ , φ ) ∝ . (3.14)
2 ρ cσ (φ ) + Z wp (φ ) ka ( sin θ − sin φ )
4 DERIVATION OF THE PROPOSED ANGULAR WEIGTHING

FUNCTION w(φ
φ)
The case where the incident sound is generated by a sound source in an enclosure, like
a room, cabin or duct is now considered. We assume that the sound pressure waves are
incident at different angles φ with random phases and mean squared sound pressures which
are proportional to a weighting function w(φ ) .
2
pirms (φ ) ∝ w (φ ) . (3.15)
The weighting function is to account for the fact that sound waves at grazing angles of
incidence will have had to suffer more wall collisions and therefore be more attenuated
before reaching the panel. The total mean square sound pressure | pTrms (θ ) |2 at the receiving
point is
2
2 π /2 w (φ )  sin  ka ( sin θ − sin φ )  
pTrms (θ ) ∝ ∫ 2   dφ . (3.16)
−π / 2
2 ρ cσ (φ ) + Z wp (φ )  ka ( sin θ − sin φ ) 
The case when sound is incident from a source in a free field at an angle θ to the
normal to the panel and the panel radiates at all angles φ into a room or duct is also of
59
interest. In this case the weighting function w(φ ) is to account for the fact that sound waves
radiated at grazing angles will have had more wall collisions and therefore be more
attenuated before reaching the receiving position which is assumed to be a reasonable
distance from the panel or opening which is radiating the sound. In this second case, we have
to integrate over all angles of radiation φ because of the reverberant nature of the sound. For
this case, the impedance terms in the integral are functions of θ rather than φ and can be
taken outside the integral. However in this study both cases are calculated using the formula
for the first case which is shown above. This is because both cases should be the same by the
principle of reciprocity and it is difficult to determine which form of formula is more
accurate.
For large values of ka , the formula will be similar for the two cases. If ka is much
greater than 1, the function
2
 sin  ka ( sin θ − sin φ )  
  (3.17)
 ka ( sin θ − sin φ ) 
has a sharp maximum at φ = θ and is symmetrical in both θ and φ about the point
φ = θ . We can exploit these facts by evaluating the impedance terms for the first case at
φ = θ and taking them out side the integral. This gives the formula for the second case.
The relative sound pressure level L(θ ) in the direction θ is
( ) (
L (θ ) = 20 log10 pTrms (θ ) − 20 log10 pTrms ( 0 ) . ) (3.18)
60
a meters
Sound source
b meters
b tan φ m
Figure 3.7: Calculating the number of wall reflections before sound hits the panel or opening at an
angle of φ to the normal.
Assume that the sound source is distance b from the surface of the room containing
the panel or opening and that the room width is a in the plane containing the incident sound
ray. If the sound ray is incident at an angle of φ to the normal to the panel or opening, it
travels a minimum distance of b tan φ parallel to wall containing the panel or opening before
hitting the wall. This is similar to the image source method of predicting reflections in a
room. The sound which travels this minimum distance hits the wall the following “n”
number of times,
b
n= tan φ (3.19)
a
before reaching the panel or opening. If the sound absorption coefficient of the walls of
the room is α , the sound intensity incident at an angle of φ to the normal is proportional to
b
n tan φ
w (φ ) = (1 − α ) = (1 − α ) a . (3.20)
61
Equation (3.20) gives us the weighting function w (φ ) . Uniform diffuse field weighting
is obtained when α is zero.
4.1 Radiation efficiency function details
In this study we use the radiation efficiency of a strip of width 2a , which we

approximate with the following equation [44].
 1
 if φ ≤ φl
π
 + cos φ
 2k 2 a 2
σ (φ ) =  (3.21)
 1 π
if φl < φ ≤
 π φ
3cos l − cos φ 2
 2k 2 a 2 + 2
where
 π 
φl = arccos   (3.22)
 2ka 
and k is the wave number of the sound and 2a is the length of the panel in the
direction of the source.
For an opening with no panel in an infinite baffle we put Z wp (φ ) = 0 . For a finite panel
in an infinite baffle we use the infinite panel result for Z wp (φ ) . This result is expected to be
the correct result when averaged over frequency, because this approach gives the correct
result for point impedances when averaged over frequency and position on a finite panel
[45].
   ω 2  ω 
2

Z wp (φ ) = mω  j 1 −   sin (φ )  + η   sin 4 (φ ) 
4
(3.23)
   ωc    ωc  
62
where m is the surface density (mass per unit area) of the panel, η is the damping loss
factor of the panel, ωc is the angular critical frequency of the panel and ω is the angular
frequency of the sound.
In the case of a rectangular enclosure like a duct etc, the directivity of the sound source
is also included. The sound source is modelled as a line source of length 2r where r is the
radius of the sound source. The directivity of the sound source is proportional to
2
 sin ( kr sin φ ) 
  (3.24)
 kr sin φ 
where k rad/m is the wave number.
5 COMPARISON WITH PUBLISHED RESULTS ON DIRECTIVITY
In this section, the panel radiation directivity prediction method described in the
previous section is compared with experimental results and prediction methods for finite size
panels and finite size openings from the literature, which are mainly based on previous glass
panel and duct studies, as previous published results are only available in the field of
building acoustics. Results are presented on a logarithmic scale of Strouhal number. The
Strouhal number is defined as the ratio of the distance across the finite flat panel or finite
opening in the direction of the receiver to the wavelength of the sound in the air.
Stead [47] measured the sound insulation of a window installed in one wall of a room.
The sound was incident at an angle to the normal to the window from outside the room. This
is the opposite direction to the calculation method used in this paper, but is expected to give
similar results because of the principle of reciprocity. The window was 1450 mm wide by
2120 mm high. The glass was 6 mm thick. The wall of the room containing the window was
part of the external wall of a larger building which served as a baffle. The internal
dimensions of the room were 2880 mm wide by 3000 mm high by 5120 mm deep. The
loudspeaker was 20 m from the middle of the widow. The edge of the building in the
direction of the measurements was 11 m from the centre of the window. Thus the baffle
length was set to twice this distance, namely 22 m. To show the result of the diffraction
correction, Stead’s results at an angle of 90° are compared with the theory presented, in
Figure 3.8.
63
15
10
Relative Sound Level (dB)
Theory
5
Experiment
-5
-10
-15
0.1 1 10 100
Strouhal Number
Figure 3.8: The sound level at 90° relative to that at 0° as a function of Strouhal number for 6 mm
thick glass installed in the wall of a room.
-5
Relative Sound Level (dB)
-10 Theory
Experiment
-15
-20
-25
-30
-35
0.1 1 10
Strouhal Number
Figure 3.9: The sound level at 90° relative to that at 0° as a function of Strouhal number for an
unbaffled duct end opening.
64
Stead’s measured reverberation times were used to calculate the average wall
absorption coefficients of the room for use in the calculation of the weighting function. The
difference between the experimental and theoretical results at coincidence does not occur at
low angles of incidence.
Sutton [48] measured the directivity of eight unbaffled duct end openings in an
anechoic room. Figure 3.9 shows a comparison of his measurements at 90° for an 80 mm
square cross section duct of length 750 mm. Sutton used two sound sources. The sound
source directivity was modelled by assuming a source diameter of 300 mm for Strouhal
numbers less than one and a source diameter of 30 mm for higher Strouhal numbers. Note
that the lower frequency sound source would not have actually been 300 mm in diameter.
The 300 mm diameter is used to model the directivity of the sound radiation into the duct. If
only plane waves had been excited in the duct, a sound source diameter of infinity would
have been used. An absorption coefficient of 0.05 was assumed for the internal walls of the
duct.
6 CONCLUSIONS
The Simple Harmonic Motion oscillator model of analyses of the current carpet
design system shows that there exists a resonance in the 323Hz frequency range which is in
the engine or power train noise frequency range. This is verified from the sound
Transmission Loss measurements of the carpet variants, using impedance tube method.
Further, the theoretical model presented for panel radiation directivity prediction can
be used to successfully predict the sound level radiated at a particular angle to the normal of
a panel or opening, relative to the sound level radiated in the direction of the normal. The
theory depends on the length of the radiating object in the direction of the observer divided
by the wavelength of the sound in air, and is independent of the width of the object at right
angles to the direction of the observer. The relative sound level radiated from a panel is
relatively independent of the Strouhal number and the angle of radiation apart from a strong
peak at coincidence. The relative sound level radiated from an opening decreases as the both
the Strouhal number and the angle of radiation increase.
65
CHAPTER 4
DESIGN CONCEPTS FOR REDUCING NOISE FROM FLOOR

PANEL SOUND RADIATION
1 INTRODUCTION
A set of layered carpet concept designs is proposed for vehicle carpet applications,
which are theoretically expected to meet the required acoustic performance targets, and at the
same time expected to meet the constraints on various other factors like cost, weight,
manufacturability, aesthetics, recyclable etc. Through the design process of deriving a
hypothetical concept vehicle carpet having defined layer types, but with variable options for
materials and corresponding properties, a set of realizable concept designs are proposed and
evaluated with the available materials and infrastructure. The preliminary evaluation is
conducted and discussed for the acoustic performance of the layered concept designs.
This chapter outlines the design concepts underlying the vehicle carpet systems and
gives a functional analysis report of the designs with respect to the acoustical performance
requirements. By explaining the current vehicle carpet design, a baseline is made for further
comparison and evaluation. Targets are set with reference to the current production carpet,
against which each concept design will undergo analysis for acoustic performance.
Since theoretical explanation and modelling are efficient tools for validating and
perfecting the designs, a brief outline of the modelling methods is explained in the context of
the layered carpet design techniques.
The expectation is to lay-out key layered concept designs for the future vehicle carpet
systems, which can perform superior to the existing layered carpet design, by improving
certain acoustic parameters of interest. Further to this stage of conceptual design, a detailed
66
optimization and functional evaluation stage of the designs is expected, to realize the concept
design.
2 CURRENT VEHICLE CARPET SYSTEM
The current vehicle carpet system in production is the best starting point for both
conceptual design and functional analysis studies. The current production vehicle carpet we
have is based on a three material layer design.
The three layers of the current carpet system are from top to bottom, carpet tufting
layer, heavy sound barrier layer and the vibration de-coupler/absorber layer.
Tufting layer
Cabin side (top) Heavy layer (barrier)
Engine side (bottom) De-coupler/absorber layer
Figure 4.1 – Current vehicle carpet design
The materials applied in the current production are,
Carpet tufting layer - Face 320gm/m2 carpet fabric
Heavy layer - 5 kg/m2 heavy layer
De-coupler/absorber - 1.65 kg/m2 Poly Urethane Foam
Total area density - 7.5 kg/m2
67
2.1 Functional analysis
The top carpet tufting layer functionally helps to give the looks, touch & feel and
keeps the carpet surface clean and durable to wear. At the same time, in an acoustic point of
view, the tufting layer helps, to an extent, for absorbing the sound in the cabin side.
The normal incidence sound absorption coefficient of the various carpet tufting layers
are shown below.
1.00E+00
9.00E-01
8.00E-01
7.00E-01
6.00E-01
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
50
302
554
806
1058
1310
1562
1814
2066
2318
2570
2822
3074
3326
3578
3830
4082
4334
4586
4838
5090
5342
5594
5846
6098
6350
L M N O P Q
Figure 4.2 – Normal sound absorption coefficient of

various carpet tufting (L to Q)
The average sound absorption is relatively less, of only about 30% for the tufting layer.
Meanwhile, the heavy layer functions as a simple sound barrier, where the denser and
thicker the layer, the more sound is stopped from entering the vehicle cabin. So the heavy
layer contributes to the total weight of the carpet, and so the main parameter would be the
surface density of the layer, which is typically between 1.5kg/m2 and 5kg/m2.
The below graph shows the normal incidence Transmission Loss of various heavy
layer carpet systems used in current carpet designs. The measured transmission loss ranges
from 10dB to 35 dB.
68
Transmission Loss - Set 2
70.0
60.0
50.0 L
20Log(Inc/Trans)
40.0 M
TL (dB) -
30.0
N
20.0
10.0 O
0.0 P
-10.0 Q
.0
.0
.0
.0
.0
.0
-20.0
5.
0.
5.
0.
0.
50
80
50
00
50
00
12
20
31
50
80
12
20
31
50
-30.0
Frequency
Figure 4.3 – Normal incidence Transmission Loss

coefficient of various heavy layers
The bottom final de-coupler layer acts as both a vibration de-coupler/absorber and
also helps to fit the carpet bottom well onto the bottom metal floor.
The absorption coefficient for the de-coupler (bottom of carpet) is comparatively

high as the below measured result shows. The de-coupler layer can absorb about 80% of the
sound incident on it, but mainly in the frequency above 750Hz.
SAC - L foam de-coupler (carpet bottom layer)
1.20
1.00
0.80
SAC
0.60
0.40
0.20
0.00
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Fre que ncy
Lfoam
Figure 4.4 – Absorption characteristics of a typical

de-coupler layer
69
2.2 Baseline and targets for the carpet design
The concept design for the vehicle carpet is focussed on improving the acoustic
performance of the carpet, both as a power train noise barrier and a cabin noise absorber,
while giving highest priority to aesthetics and comfort.
The acoustic evaluation results for current production design can be taken as the
baseline for further acoustic performance improvement, at various stages of the conceptual
design. For normal incidence absorption coefficient, the best performance curve fit (green
line) is taken as the baseline curve, to which the future results will be compared for
evaluating improvement. Meanwhile, for Transmission Loss, the minimum curve (heavy
layer of 1.7 kgms) and maximum curve (heavy layer of 5 kgms) are considered for defining
the baseline and target range. The carpet samples are assumed to have thickness limited to
40mm, and in the pre-moulded stage.
The baseline for power train noise transmission loss and cabin noise absorption are
shown as the green lines passing through the best TL and absorption coefficient peaks of the
current carpet design.
Absorption coefficient (Cabin/tufting side)
40 mm pre-moulded
1.00E+00
9.00E-01
8.00E-01
7.00E-01
6.00E-01
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
50
302
554
806
1058
1310
1562
1814
2066
2318
2570
2822
3074
3326
3578
3830
4082
4334
4586
4838
5090
5342
5594
5846
6098
6350
L M N O P Q
Figure 4.5 – Cabin side absorption coefficient of

current carpet designs
70
Transmission Loss
Curren production, 40mm pre-moulded
60.0
50.0
TL (dB) - 20Log(Inc/Trans)
40.0
30.0
20.0
10.0
0.0
0.0 500.0 1000.0 1500.0 2000.0 2500.0 3000.0 3500.0 4000.0 4500.0 5000.0
Frequency
L N O P Q Min HL1.7kgms Max HL5.0kgms
Figure 4.6 – Transmission loss of various carpet

designs, with maximum and minimum defined
3 THEORETICAL BASE FOR FUNCTIONAL ANALYSIS
The acoustic properties, i.e. absorption coefficient and transmission loss, of various
layers at the material level can be predicted using existing mathematical models. Meanwhile,
the inter-layer interaction and effects are predicted using a mathematical approach based on
transfer functions [53].
Fibre & Foam:
Four different models are available to represent foam and fibrous materials [4, 49, 50,
52]:
1. The elastic porous (foam) model,
2. The limp porous (fiber) model,
3. The rigid (fiber) porous model and
4. The Delany-Bazley (fiber) model [52].
The elastic porous model is used for foam materials where the stiffness of the frame
is important in vibro-acoustic response of the noise control material. The energy exchange
71
between structural energy and acoustical energy within a foam material typically provides
much of the desired energy absorption. The full elastic porous model requires all the fluid
properties and the elastic bulk properties.
The below diagram gives a basic view of the order of development and application of
the mathematical models.
Figure 4.7 – Material acoustic properties in relevance

order for modelling [53]
For materials in which the frame waves do not carry a significant amount of energy,
such as many fiber-based materials, the frame waves may be disregarded altogether and the
noise control layer modeled with a single type of acoustic compression wave. In this case
one of the fiber models should be used to represent such a material.
When the frame of the porous material can be considered limp, the fluid properties
and the bulk density are required to predict the vibro-acoustical behavior. Similarly when the
frame of the porous material can be considered rigid, only fluid related properties are
required.
The Delany-Bazley model is the simplest fiber model and only requires the fluid
density, fluid speed of sound and the flow resistivity of the acoustic material to characterize
the performance of a noise control treatment. The Delany-Bazley model is activated when
both the viscous and thermal characteristic lengths are zero.
72
AutoSEA2 acoustic simulation software based on Statistical Energy Analysis method

incorporates these mathematical models. Hence, given the required material properties for
modelling, any foam or fibre material response in an acoustic field can be modelled and
evaluated easily, than working out the tough mathematics.
3.1 Layered up material interaction
The layered interaction between various material types, foam, fibre, panel, air gap
etc, is solved mathematically using a method called Transfer matrix [4, 30, 35, 39, 57, 59],
where the transfer function for the acoustic variable pressures and velocities are
parameterised into a 2x2 complex matrix for each layer. The methodology is based on the
representation of plane wave propagation in layered media, in terms of transfer matrices.
Each layer is assumed to be constructed of a homogeneous and transversely isotropic
material. The elastic and fluid layer models are based on classical plane wave propagation.
Modelling of the poro-elastic layer is based on the extension of Biot theory to acoustics [4].
Meanwhile, various acoustically known methods and knowledge can be used as a

direction for bringing up designs. Adding of new layers for specific requirements, using the
double panel separated by a fluid technique, etc are few of them that have relevance. So,
theoretically validated or explainable directions have been taken for each concept design.
4 PROPOSED CONCEPT DESIGNS

The process of arriving at concept designs is based on the design requirements and by
moving in design directions that are theoretically acceptable and that may result in satisfying
results on implementation. Hence, the conceptual designs may remain to be hypothetical, to
an extent.
73
Layered concept 1:
Introducing minor changes to the existing production vehicle carpet system can improve the
cabin side absorption sufficiently, i.e. by adding an absorber layer or air gap (honey comb)
just below the tufting layer. This concept design will be more manufacturing friendly.

Heavy layer (barrier)
Absorber/air-gap
40mm
Noise path De-coupler/absorber layer

(from engine)
Figure 4.8 – Concept 1 carpet design
Layered concept 2:
Replacing the heavy layer with a double panel of Air Flow Resistance (AFR) layers, filled
with sound absorbing material or air gap (honey comb).

AFR layer 1
AFR layer 2
40mm
Absorption layer
Noise path
74
Layered concept 3:
By replacing the thick heavy layer with a double panel system, where the top panel is an
AFR layer and the bottom panel is a thin heavy layer. This concept actually tries to make a
compromise result of concepts 1 and 2.

AFR layer 1
Heavy layer
40mm
Absorption layer

(from engine)
Layered concept 4:
Introducing absorber layer just below carpet tufting and also having a double panel system
using AFR panels, separated by an air-gap (honey comb).
Tufting layer
Cabin side noise

Absorber
Thin heavy layer
1Thin heavy layer
40m2
Noise path
(from De-coupler/absorber
layer
75
4.1 MATERIAL SELECTION
For the purpose of selecting the best material suiting each concept design layer, a
category wise material table is maintained. Referring to the four concept designs laid above,
the basic material categories involved are,
1. Carpet tufting type
2. Air Flow Resistant (AFR) type
3. Absorber type
4. Heavy Layer type
5. Air-gap structure type and
6. Scrim/micro-fibre film type, to increase absorption, which is included in later

design stage.
As shown below, a simple and direct approach was followed for selecting the best
material in each category for building up the concept layers. The carpet or material samples
(e.g. Set1 – K) are listed from top to bottom and the various layer types of these samples are
labelled into one of the material type columns (e.g. Foam). For particular design proposals,
coloured lines are drawn through the preferred material label boxes, giving a complete view
of the materials selected. The material cost, weight, durability, whether its recyclable, etc are
expected to be handled after the concept design stage.
76
Figure 4.12 – Material layer selection matrix
4.2 Evaluation test setup
Absorption coefficient measurement:
The normal incidence absorption coefficients of the samples were measured using
B&K impedance tube, following the broad-band 2 microphone method.
100mm and 29mm diameter samples were cut and measurements were done inside
the large tube set-up (low frequency 50-1.6kHz) and small tube set-up (high frequency 500-
6.4kHz).
The microphone and the tube correction factors were considered and compensated
throughout the measurements.
At least 2 co-related results were picked up for averaging and combining low and
high frequency plots, with the cross-over in 500-6.4 kHz range.
77
Transmission loss measurement:
Impedance tube method is used to measure the normal incidence Transmission Loss
of the material sample. This is a newly developed 4 microphone technique, where 2
microphones each measure the wave field on either sides of the sample. This is explained in
chapter 6 for reference.
Similar to the absorption coefficient testing, the samples are cut in 100mm and 29mm
diameters, and tested for TL in the large tube (50-1.6kHz) and small tube (500-6.4kHz)
setups, and the cross over frequency range values are averaged to get a continuous curve.
5 PRELIMINARY EVALUATION RESULTS
Acoustic measurements are conducted on the proposed design concepts using

various selected material layer samples detailed in the below table.
Table 1 – Nomenclature of layered material combinations
Layer Code Material & Specs
1. Tufting Ktuft Type K, brown tufting, 320gsm
2. Air Flow Resistant layers (AFR) S3D AFR layer
S3F
3. Absorber MJA316 CSIRO fibre absorber sample
Nfoam 1.5 kgsm, close cell polyester foam
S2M Poly-felt absorber samples
S3M
S3N
4. Heavy layer HL1.7 2.5 mm Heavy Layer (1700gsm)
HL5.0 Heavy Layer (5000gsm)
5. Air gap HC10mm 10 mm Honeycomb structure (grey)
Gap10mm 10mm air gap
78
Layered concept 1: Introducing minor changes to the existing production vehicle carpet
design, i.e. by adding an absorber layer or air gap (honey comb) just below the tufting layer.
Heavy layer
(barrier)
Absorber/air-gap
40mm
De-coupler/absorber layer
Ktuft_MJA316[1p75mm]_HL1p7_MJA316[1p75mm]_1
Noise path
(from engine)
1.00
0.90
0.80
Absorption coefficient
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 1000 2000 3000 4000 5000 6000
Ktuft_MJA316[1p75mm]_HL1p7_MJA316[1p75mm]_1
Figure 4.13 – Cabin side absorption characteristics of Concept 1 carpet design
Transmission Loss Cabin side noise Tufting layer
Heavy layer
(barrier)
Absorber/air-gap
TL - Concept 1 40mm
60.0 Noise path
(from engine)
50.0
40.0
TL (dB ref. 1)
30.0
20.0
10.0
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Freq (Hz)
Ktuft_HL1.7_MJA316
Ktuft_MJA3161p75cm_HL1.7_MJA3161p75cm
Min HL1.7kgms
Max HL5.0kgms
Figure 4.14 – Transmission loss characteristics of Concept 1 carpet design
79
Layered concept 2: Replacing the heavy layer with a double panel of Air Flow Resistance
(AFR) layers, filled with sound absorbing material. Cabin side noise Tufting layer
AFR layer 1
AFR layer 2
40mm
Absorption coefficient Absorption layer
1.00
Noise path
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 1000 2000 3000 4000 5000 6000
Ktuft_S3D_Nfoam20mm_S3D_Nfoam20mm Ktuft_S3D_S3N_S3D_S2M
Ktuft_S3D_MJA316-30mm_S3D_S3M Ktuft_S3F_MJA316-20mm_S3F_S3N

AFR layer 1
AFR layer 2
40mm
Transmission loss Absorption layer
TL - Concept 2 Noise path

(from engine)
60.0
50.0
40.0
TL (dB ref. 1)
30.0
20.0
10.0
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Freq (Hz)
Ktuft_S3D_MJA31630mm_S3D_S3M Ktuft_S3F_MJA31620mm_S3F_S3N
Min HL1.7kgms Max HL5.0kgms
80
Layered concept 3: Replacing the thick heavy layer with a double panel system, where the
top panel is an AFR layer and the bottom panel is a thin heavy layer.
AFR layer 1
Heavy layer
40mm
Absorption coefficient Absorption layer
SAC Noise path

(from engine)
1.00E+00
9.00E-01
8.00E-01
7.00E-01
6.00E-01
SAC
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
0 1000 2000 3000 4000 5000 6000 7000
Frequency
Ktuft_S3F_MJA316_17p5mm_HL1p7_MJA316_17p5mm
Ktuft_S3D_MJA316_17p5mm_HL1p7_MJA316_17p5mm

AFR layer 1
TL small - Combinations Heavy layer
40mm
60.0
Absorption layer

(from engine)
50.0
40.0
TL (dB ref. 1)
30.0
20.0
10.0
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Freq (Hz)
Min HL1.7kgms
Max HL5.0kgms
Ktuft_S3F_MJA316_17p5mm_HL5_MJA316_17p5mm
Ktuft_S3D_MJA316_17p5mm_HL5_MJA316_17p5mm
Ktuft_S3F_MJA316_17p5mm_HL1p7_MJA316_17p5mm
Ktuft_S3D_MJA316_17p5mm_HL1p7_MJA316_17p5mm
81
Layered concept 4: Introducing absorber layer just below carpet tufting and also having a
double panel system using thin heavy layer panels, separated by an air-gap (using honey
comb or foam structure). Cabin side noise
Tufting layer
Absorber
AFR layer 1
SAC - Concept 4 AFR layer 2
40mm
1.20
Noise path
1.00
0.80
SAC
0.60
0.40
0.20
0.00
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Fre que ncy
Ktuft_Nfoam10mm_HL1p7_Gap10mm_HL1p7_Nfoam20mm
Ktuft_Nfoam10mm_HL1p7_HC10mm_HL1p7_Nfoam20mm
Tufting layer
Cabin side noise
Absorber
AFR layer 1
TL small - Combinations AFR layer 2
40mm
70.0
Noise path
60.0
50.0
TL(dBref. 1)
40.0
30.0
20.0
10.0
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Freq (Hz)
Min HL1.7kgms Max HL5.0kgms C4_HL1p7 C4_HCHL1p7
82
5.1 Result discussion
The absorption coefficient measurement results of the new carpet concept designs
clearly illustrates that introducing a layer of absorption material between the tufting
layer and the heavy noise isolation layer improves the vehicle cabin side sound
absorption coefficient to a great extent. Concept 4 carpet design shows the minimal
sound absorption coefficient improvement, compared to other design concepts.
Meanwhile, when Concept 1 and Concept2 design Sound Transmission Loss results
are not that promising for noise isolation improvement compared to the Heavy Layer
(1.7kgsm) noise isolation method, Concept 3 and Concept 4 has a reasonable
improvement in Sound Transmission Loss, especially above 2 kHz frequency range. It is
to be noted that in the case Concept 4, the measured Sound Transmission Loss is even
better than the 1.5kgsm Heavy Layer carpet design, which is a significant improvement
in noise isolation.
6 CONCLUSION
• Air flow resistance layer in place of heavy layer largely improves the sound
absorption coefficients on the cabin side.
• Foam/Fibre Insertion between carpet tufting and heavy layer (Concept 1) improve s
largely the sound absorption coefficients on the cabin side, especially in mid-high
frequency.
• Concept 2 carpet design, i.e. double air flow resistance layers with absorber layer
sandwiched inside instead of heavy layer, is best for SAC at low frequencies.
• Heavy Layer (HL) in the middle could achieve the effect of airflow resistance layer
panels separated by MJA 316 (20 mm), for sound transmission loss.
• HL in the middle (Concept 1) will also achieve both good SAC and TL, which is also
a development direction.
• The compromised concept design for both cabin side sound absorption and
Transmission Loss will be Concept 3, i.e. replacing the thick heavy layer with a
double sandwiched panel system, where the top panel is an AFR layer and the bottom
panel is a thin heavy layer.
83
CHAPTER 5
DEFINING THE DESIGN PROCESS AND ACOUSTIC TOOLS
1 INTRODUCTION
Development of a better acoustic structure involves the introduction of new types of

materials or a combination of various materials types to suit the application. The acoustic
properties provide important criteria for material selection. The acoustic performance is
evaluated by the materials’ sound absorption and transmission loss coefficients.
The use of computer simulation to evaluate component layers of materials is a

convenient way of optimising acoustic performance. Computer programs like AUTOSEA,
require random incident sound absorption and transmission loss coefficients and/or intrinsic
material property data to run the simulations. Measuring intrinsic property data such as
tortuosity, viscous and thermal lengths, among others, are time consuming and require
specialized equipment. On the other hand, the evaluation of sound absorption coefficient
using a large reverberant chamber requires a large space and facilities [28] and is limited by
lower frequency limits. Likewise, sound absorption and transmission loss coefficient
measurements using an impedance tube are limited to normal incidence conditions and can
not be directly inputted to the simulation programs.
Other issues such as the cost of the facility, portability and ease of operation pose
considerable challenges and thereby create a need to explore alternative means to evaluate
acoustic properties. It is in this context that a multi-function device is proposed. The
proposed 45° SAC device is designed for quick and simple measurement of αθ,f and TLθ – in
situ or in laboratory conditions. It can also be used as a portable listening device for
subjective analysis.
84
The proposed V or Y-shaped acoustic wave guide, with which the upper branches forms
a 45° angle with respect to the horizontal, is designed to perform the functions of (1)
measurement of the sound absorption coefficient (SAC) (αθ,f) and the transmission loss
coefficient (TLθ) of carpet materials, and (2) act as a listening device for subjective
evaluation. An acoustic source injects directive sound waves into the duct, hits the sample
material at the bottom of the Y-shaped junction, and establishes a standing wave condition.
A two-microphone set-up will measure the acoustic pressure and derive the transfer
functions needed for the calculation of αθ,f and TLθ. The proposed device is portable and can
be used for quick in situ measurements.
Further, an overview about the various acoustic simulation tools, especially AUTOSEA
2 acoustic simulation tool, based on the well established Statistical Energy Analysis method,
is provided as a platform on which all the design validation is based on.
2 REVIEW OF SAC MEASUREMENT METHODS

Literature review for sound absorption measuring devices reveal interesting designs
based on the type of measurement methods categorized [29] as follows:
Reverberation room
The Sabine equation for estimating the reverberation time, T60 is
55.25V
T60 = ………………………………………(1)
cSα
where V is the volume of the room, c=343 m/s is the speed of sound in air, S is the total
surface area of the room. is the average absorption coefficient for the room, calculated
from
α 1 S1 + α 2 S 2 + ... + α n S n
α = ………………………..(2)
S
where αn are the individual absorption coefficients for each surface Sn and S is the total
surface area of the whole room.
85
From Sabine’s formula and the definition of absorption, the absorption of the sample can
be expressed as:
0.16V 1 1
α=  −  …………………………………(3)
S  Ts Te 
where:
α is the absorption coefficient of the sample
S is the area of the sample material
V is the volume of the chamber
Ts is the reverberation time, with the sample
Te is the reverberation time of the empty chamber
There are drawbacks in this method where it is possible to get measurements of α

exceeding 100% using the reverberation chamber. This is due to the inherent definition of α
hence the uncertainty is at best ±10%. Should we take the physical definition of the
absorption coefficient: α = (absorbed energy)/(energy impinging the surface), the
reverberation room cannot measure the absorption according to the above definition as there
is no real control of either the energy absorbed or impinged in the surface[28]. In addition,
large pieces of sample material are needed and the chamber itself is a large and expensive
facility.
Impedance Tube Apparatus
The impedance tube, or its modern equivalent, is based on the Kundt’s tube design and
applies the standing wave theory. The Bruel & Kjaer Pulse Impedance Tube – type 4206 and
its predecessors are widely used for evaluating both normal sound absorption coefficient (α0)
and sound transmission loss properties. Depending on the tube size, the applicable frequency
range is 50-6.4kHz. Normal sound absorption (α0) measurements use the two-microphone
set-up whist the four-microphone set-up and the associated transfer functions are needed for
sound transmission loss evaluation [30].
Other Devices
Microflown PU intensity probe
The Microflown PU intensity probe is an acoustic particle velocity sensor developed in

the University of Twente in 1994 and commercialized in 1997[57]. While traditional method
86
for measuring sound intensity is to use a PP probe (two pressure microphones) and
determines particle velocity to acquire sound intensity, Microflown Technologies introduced
an improved sensor using a PU probe (microphone and Microflown) which measures particle
velocity instantly. The PU sensor can be used to measure acoustic impedance. The reflection
coefficient can then be determined and the absorption coefficient calculated
accordingly[57,58,59]. Reliability issues are currently being addressed in an attempt to
improve the technology.
Devices under patent
Patent search uncovered some types of devices for SAC measurements similar to the
impedance tube. Examples are given below.
1. US Patent 4732039 (Figure 5.1) is a flexible U-shaped channel abutted against a

surface to be measured, thereby forming an acoustic duct, with the sample surface forming
one wall of the duct. The acoustic wave injected into the duct travels parallel to the surface
of the sample and operates on the standing wave theory. The device is claimed for use in
acoustic impedance measurement.
Figure 1: US Patent 4732039

2. US Patent 4537630 (Figure 5.2) is an open-ended acoustic impedance tube abutted
against the sample material. It requires a material of known acoustic property for proper
87
calibration and the unknown sample material measurements is compared to the base
readings. This device uses the two-microphone set-up and ascribes to the standing wave
theory. An enhanced version of a similar device is shown in Figure 5.3.

(2000)
Diffuse field measurement techniques
In-situ, diffuse field measurement methods are used particularly in acoustic evaluation of
road surfaces[29]. The diagram in Figure 5.4 illustrates the concept.
Figure 4: Set-up for transfer function measurement

in front of an absorbing surface [61].
A similar, though simpler, idea was earlier mentioned in the B&K manual. Figure 5.5
shows a similar concept but without the source-receiver leakage pathway.
88
Figure 5. Diffuse field measurement concept (B&K).
The B&K tone burst method[35] can be applied using the following formulation:
Absorption coefficient:
α θ , f = 1 − rθ , f ......................(4)
rθ,f = the reflection coefficient

(L p , d − L p , r )
−
10
rθ , f = 10 ……….(5)
Lp,d = SPL of direct sound
Lp,r = SPL of the reflected sound
The SPL can be expressed as[37]:
P2
SPL = 10 log 2 ……..(6)
Pref
If we assign local SPL readings as:
Lp,d → PA and Lp,d → PB, and substituting (6) into (5), then the sound absorption
coefficient can be expressed as:
89
2
P
αθ , f = 1 − B .......................................(7)
PA
3 DESIGN CONCEPT
3.1 Theoretical basis

The proposed wave guide operates on the basis of the planar, standing wave theory and
assumes that the attenuation in the duct walls is minimal. Acoustic pressure wave
measurements are taken from two locations using a two-microphone set-up. Pressure field
measurements will be employed to derive the transfer functions (complex, incidence and
reflected), impedance, and the reflection factor needed to calculate αθ and TLθ.
Figure 5.6 depicts the planar wave propagation within the proposed device. The sound
waves are projected downward at an angle of 45° from the left side (PA), strikes the sample at
the bottom junction of the ‘V’, and resulting in the specular reflection (PB) upwards to the
right side. The dominant reflection pathways are also shown as PAr and PBr. Through the use
of a semi-anechoic termination cap, it can be further assumed that the magnitude of PBr is
small and that the potential leakage back to the source (i.e. similar to the PAr pathway) is
negligible.
Loc Loc
A B
4
1 s
3 PBi
2
d
PBr
PAi
PAr
Figure 6. SAC device working principle [62]
The standing wave theory behind the working principle of the proposed SAC device is
outlined herein following established methods [30,38,39,40,41].
90
The sound intensity, at any position inside the waveguide, is the product of the
instantaneous pressure and particle velocity at that location, viz:
I (t ) =< P(t ) u ( t ) > ………………………………………(8)
Assuming that the sound field is sinusoidal, employing time averaging and using the
complex notation, the sound intensity can be written as:
1
I =
2
( )
Re Pu ∗ …………………………………….(9)
Time averaging is implicitly given by the factor ½ for rms values of sound pressure and
particle velocity, and u* is the complex conjugate of u. The term Re[…] means only the real
part of the quantity inside the brackets is considered. Referring to Figure 5.6, the absorption
coefficient, αθ,f , can be expressed as the net intensity falling and leaving the sample surface
divided by the incoming intensity. The points of reference are at locations A and B.
1 1
IA − IB 2
( ) 2
(
Re PA u ∗A − Re PB u B∗ ) (
Re PB u B∗)
αφ, f = = = 1− ……………………(10)
IA 1
( ∗
Re PA u A ) (
Re PA u ∗A)
2
The characteristic impedance (in the direction of propagation) in a single plane wave is:
Z 0 = ρ 0 c0 …………………………………………..(11)
The particle velocity can be calculated from the pressure using the analogy of Ohm’s
Law.
91
u = P / ρ 0 c0 = P Z 0 ………………………………………..(12)
The field impedance at any point x within the sound field is:
Z ( x ) = P( x ) / u ( x ) → 1 Z ( x ) = u ( x ) P( x ) ……………..(13)
Using the relation u ∗ P ∗ ≡ u P , (13), and further simplifying (10) yields:
Re u B  
2 Re
1 
P B
2
 PB  P  Z B 
αφ, f = 1− = 1 − B2 …………………………………….(14)
P Re u A 
2
PA Re 1 
A
 P   Z A 
 A 
The field impedance at location A can be quantified from equations (12) & (13), viz:
PA PAi + PAr PAi + PAr 1 + (PAr PAi ) 

ZA = = = = Z0   …………………(15)
u A u Ai − u Ar (PAi Z 0 ) − (PAr Z 0 ) 1 − (PAr PAi ) 
The reflected components of the sound waves at locations A & B can be expressed as:
PAr = PAi R A and PBr = PBi R B …………………..(16)
Then, the expression for the complex reflection coefficients are:
R A = PAr PAi and R B = PBr PBi …………………..(17)
92
Substituting (17) into (15), the expression for the field impedance simplifies to:
1 + R A 
At location A, Z A = Z 0   …………………………………(18a)
1 − R A 
1 + RB 
Similarly for location B, Z B = Z 0   ………………………(18b)
1 − R B 
Combining (18) and (14) yields:
1 − R B 
2 Re  
αφ, f = 1 −
PB 1 + RB  ………………………………………(19)
PA 1 − R A 
Re  
1 + R A 
Employing a perfect anechoic termination cap on the right hand side of the device will
render RB ≈ 0 and leaves only the complex reflection function near the source (i.e. RA at
location A) which can be evaluated by the two-microphone set-up and FFT analyser. Sound
pressure fields at locations A & B can be measured using standard pressure microphones and
the sound absorption coefficient can be calculated using (19).
If all the reflected sound components are neglected (i.e. RA & RB ≈ 0), then equation (19)
will simplify to the form similar to (7).
A working prototype model of the proposed measurement device is developed and

calibrated, but due to IP issues on pending patents for the design, further detailed
specifications are excluded from the document.
4 ACOUSTICS SIMULATION AND VALIDATION TOOLS
4.1 Scope of vehicle carpet acoustic simulation
Vehicle noise sources are basically categorised as engine/power-train noise, tyre noise,
HVAC noise, wind noise, exhaust noise and other external noises. The noise transmission
paths to the driver ear include the floor trim, which hence has a role in blocking the noise or
absorbing the noise already inside the vehicle cabin. Compared to the other sound package
93
components, floor trims cover most of the noise-exposed areas of the vehicle, and hence
plays a relatively significant role in noise reduction. For a given distribution of noise sources
and transmission paths, the acoustic properties of the carpet sections can be locally optimized
and thus achieve efficient noise isolation for given constraints of weight and cost.
Virtual modelling and testing is a widely accepted tool for improving product design and
predicting performance at early stages of vehicle development, even before prototype build,
which saves tooling cost and time caused by design changes. In order to evaluate vehicle
trim components in a sound package, intensive vehicle tests are normally conducted, which
leads finally to compromised solutions by trial and error methods. New acoustic modelling
methods can be used to simulate the component and vehicle tests during early sound package
design stages and optimize these trim components, and also provide a highly repeatable
standard platform for design comparison and benchmarking.
4.2 Simulation process cycle
An outline of simulation process in vehicle carpet design is given in Figure 5.1. Each
block also represents the set of resources and sub-tasks involved at each stage. Referring to
the process cycle, the key loops that iterate are the simulation validation and design
optimization loops.
Figure 7: Simulation process cycle
94
The selection of the right simulation tool is an important decision, which largely affects
the completion time line of the various processes or stages. Statistical Energy Analysis
(SEA) is considered here to provide sufficient resolution (in the frequency range 315Hz -
8kHz) in terms of vehicle noise simulation, and the acoustic modelling software AutoSEA2
from ESI group is used for material (Foam & Fibre modules), component and vehicle
(Template Modeller module) level simulations. Re-usability of various simulation resources
contribute to the efficiency of the whole modelling process. A detailed discussion is
conducted in the following sections regarding the selection of the acoustic simulation tool.
4.3 Review of acoustic and vibration analysis simulation tools
Based on the mathematical formulation used for modelling a system there are a variety
of options in selecting the appropriate simulation tool. Listing the established methods,
• Finite Element Method/Analysis (FEM or FEA),
• Boundary Element Method (BEM),
• Inverse Boundary Element Method (IBEM),
• Statistical Energy Analysis (SEA),
• Hybrid FEA & SEA method,
• Couple FEM & BEM method and
• Acoustic ray-tracing method.
FEM analysis is an established tool for modelling sheet metal bodies, engines,
suspensions and even complete vehicles, but requiring high computation resources, in
particular for high frequency modelling, which may take even weeks to evaluate a complete
vehicle on a normal personal computer. The accuracy of the results are highly dependent on
variation of the model details , and hence could result in erroneous results.
Figure 8: FEM model of a Sedan (Image courtesy – LMS)
95
Meanwhile, Boundary Element Method (BEM) is useful simulation tool for modelling
vehicle interior body cavities, interior trim, power train noise, exterior radiation and tyre
noise.
Figure 9: BEM modelling vehicle interior

(Image courtesy – LMS & ESI group)
Inverse Boundary Element Method (IBEM) is a modelling tool preferred for predicting
surface velocities in applications like power train and tyre acoustic modelling.
Figure 5.10: IBEM model of engine

(Image courtesy – University of Kentucky)
Statistical Energy Analysis (SEA) is a proven technology initially formulated for

applications in NASA projects. It subdivides the complete model into Systems and sub-
systems which interact each other based on energy flow principles. The system and
subsystem parameters are derived from experimental measurements. This acoustic modelling
method is best suitable for modelling vehicle components like doors, cockpits and also for
complete vehicle simulation, both interior and exterior.
Figure 5.11: SEA model of a sedan (Image courtesy – ESI

group)
96
Hybrid FEM & SEA is a new method of combining the strengths of both the simulation
techniques, where rigid junctions and structures are modelled as FE sections and cavities &
related high modal density sections are modelled using SEA method.
Figure 5.12: Hybrid FEM & SEA method (Image courtesy – ESI group)
The important factors for selecting a suitable modelling tool includes the fact that
whether it applies to the nature of the acoustic field and its excitation, which are basically
categorised as,
• Airborne
OR
• Structure-borne
Further, the complexity of simulating and validating the complete design model, at
system or vehicle level, can be selection criteria for defining the scalability of the simulation
tool. For the purpose of stage wise simulation, the following are the three levels of modelling
and validation to be handled accurately by the single simulation tool.
• Layered material simulation,
• Component or prototype design simulation,
• Complete system or vehicle level simulation
The frequency range of interest, i.e. low frequency, mid frequency or high frequency, is
also a primary selection criterion for finalizing the choice of simulation tool. The below table
97
gives a clear picture of the various simulation tools available and the factors to be considered
and compared against.
Table 1: Comparison of various acoustic simulation methods
Thus, referring to the above table for comparison, it is clear that for applications of
simulating air-borne vehicle noise (500 Hz and above) and for low computational resources,
Statistical Energy Analysis (SEA) is most suitable accommodate modelling method to go
forward. AutoSEA from ESI group is one of the recommended SEA modelling softwares in
the market, which is used throughout the project.
5 CONCLUSIONS
• The proposed 45o SAC measurement device provides non-destructive SAC

measurement for sound package samples.
• Defining the simulation process cycle and identifying the iterating execution loops is
important for project efficiency and result accuracy.
• Statistical Energy Analysis (SEA) is identified as the well suited simulation tool for
the vehicle noise frequency range of interest and in terms of computational
requirements.
• A stage wise simulation and validation process, i.e. material level, component level
and vehicle level, is required for a scalable simulation and validation of the designs.
98
CHAPTER 6
MATERIAL SIMULATION, VALIDATION AND THEORY
1 INTRODUCTION
Acoustic materials range from a variety of foams, fibre, poly-vinyl, fabric to other
structural material like steel, compressed paper or glass, used to hold these acoustic
materials. The visible variables in the layered designs of sound packages are usually
thickness, density, number of layers, special surface contours etc. Deriving out the best
combination of these variables by experience to solve a particular acoustic problem is
normally a time consuming job of trial and error.
For a given homogeneous acoustic material sample, if there is a possibility of

simulating these physical variations using a virtual material modelling tool, a huge advantage
lies ahead for optimization and quick design validation, compared to conducting
experiments.
Acoustic/physical properties of materials are measured and a database of various

acoustic package materials has been maintained for the purpose of evaluation. CSIRO
provided test data for material properties of the layers of samples, which is given as input to
SEA software to simulate the absorption coefficient (random). The conversion of measured
normal incidence sound absorption coefficient to random incidence sound coefficient and
comparison of simulated and measured results have been conducted in this chapter.
99
2 MATERIAL LEVEL SEA SIMULATION PROCEDURE
2.1 Input parameters for simulation

Material properties like air flow resistivity, porosity, tortuosity and Young’s modulus
were provided by CSIRO for each possible layer of the carpet sample, so that a material
property database was established for calculation of sound absorption coefficient and sound
transmission loss. The measured key material properties are explained below [66].
Air flow resistivity
Air flow resistivity is defined as the ratio of the pressure drop to the volume flow rate
of air through a porous material. We have measured the air permeability of the samples using
AS 2001.2.34-90 with an orifice size of 5 cm2 and a differential pressure of 98Pa. The
average of 5 repeats and the Coefficient of Variation (CV%) is reported. The Air flow
Resistivity is derived from the Air permeability test results. Unit is Ns/m3 or Ns/m4 for unit
thickness.
Porosity
Porosity φ is defined as φ = Va / Vt . Where Vt is the total volume occupied by the

sample and Va is the volume of the free air within the sample. Va is calculated as (Vt – Vs).
Unit dimensionless number in the range 0-1. Vs is s the volume occupied by the test sample
material excluding the air. A sample was cut for test its area and thickness are measured to
give the total volume Vt, the total volume of the sample in its free state has a unit of ml. The
sample was then immersed into a fluid (water and a surfactant to assist wetting) and the
displaced volume of liquid is recorded. Unit is ml.
Tortuosity
Tortuosity is a measure of the tortuous path length of a material see Figure 6.1
Figure 6.1: Illustrating Tortuosity Left: Tortuosity = 1 Right: Tortuosity > 1
100
Tortuosity has a dimensionless unit, typically larger than one and in the range of 1 - 2.
Tortuosity is a property of a porous material it has no meaning if the sample is not
permeable.
Poisson's ratio
Poisson's ratio is the ratio of the relative contraction strain, or transverse strain
(normal to the applied load), divided by the relative extension strain, or axial strain (in the
direction of the applied load). Poisson’s ratio for the materials concerned are normally very
small in value.
Dynamic modulus (Young’s modulus and Loss factor)
Young’s modulus is calculated using the equation, E = k x d / A, where d is the

sample thickness, A is its area and k is the stiffness calculated from the resonant frequency
of the sample. Similarly, the damping is characterised by the loss factor, η, which is given by
the 3dB bandwidth divided by the resonant frequency: η = ∆f/fo.
2.2 Simulation
Further, noise treatment layers were formulated in basis of the materials and their
property database and applied to a circular metal plate, for which a script was run to simulate
random incidence absorption coefficient values, i.e. inside AutoSEA2 software, select pull
down menu sequence as Scripts->UtilitiesQuery->VTL. Note that the noise treatment layer,
i.e. like sample B foam, is assumed to have a homogeneous property. Similar material layers
are used to simulate multi-layered noise control treatments.
Figure 6.2 – Material simulation using AutoSEA2
101
3 VALIDATION PROCEDURE
3.1 Absorption coefficient measurement using impedance tube
The normal incidence absorption coefficients of the samples were measured using
B&K impedance tube, following the broad-band 2 microphone method.
100mm and 29mm diameter samples were cut and measurements were done inside
the large tube set-up (low frequency 50-1.6kHz) and small tube set-up (high frequency 500-
6.4kHz). The microphone and the tube correction factors were considered and compensated
throughout the measurements. At least 2 co-related results were picked up for averaging and
combining low and high frequency plots, with the cross-over frequency in 500-1.6 kHz
range. Refer Chapter 5 for specific details on normal absorption measurement procedure.
3.2 Normal to random incidence absorption conversion
According to AutoSEA software manual, simulated sound absorption is random

incidence sound absorption coefficient. In order to compare the simulated results with the
measured results, the measured normal incident sound absorption coefficients have to be
converted to the random incident sound absorption coefficients.
Oblique incident sound absorption coefficient is given by,
where, Z is the impedance to the sample material, ρ is sound medium density, c is velocity of
sound in medium and θ is the angle of incidence with reference to normal to sample.
Normalized impedance ratio of the sample is also given as,
where, r and x are the real and imaginary variables of the normalized impedance ratio, given
from the measured impedance ratio.
Thus, after substituting for impedance ratio, we have
102
Since we are interested in random incidence absorption coefficient, integration for 180
degrees is to be done.
Refer [53]
Hence, the random incidence absorption coefficient by Paris method [53] is
PARIS method, Refer [63]
Albert method of calculating random incidence sound absorption coefficient is given below.
ALBERT method
Refer [53]
Detailed and lengthy derivation and explanation of Albert method is given in [63].
4 RESULTS AND DISCUSSION
4.1 Normal incidence vs. random incidence
Random incidence absorption coefficient calculation was tried using two different but
related methods, Paris and Albert methods. Comparing with the simulated results and as per
the support documents supplied with the AutoSEA material software package, it seems that
the calculated measurement results using Paris method matches well with the AutoSEA
simulation results. So a statistical angular averaging of sound absorption, following Paris
103
method (AS/NZS 1935.1:1998/Appendix) can be used to convert measured normal incidence

absorption coefficient. (Ref: E. T. Paris, Proc. Roy. Soc. 115, 407 (1927) )
B foam SAC : Normal vs random incidence (Paris method)
1.00E+00
9.00E-01
8.00E-01
7.00E-01
6.00E-01
SAC
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
B foam measured normal B foam measured random
Figure 6.3: Sample B foam measured normal vs. measured random (Paris method)
MJA316 SAC : Normal vs random incidence (Paris method)
1.00E+00
9.00E-01
8.00E-01
7.00E-01
6.00E-01
SAC
5.00E-01
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
MJA316 measured normal MJA316 measured random
Figure 6.4: MJA316 SAC measured normal vs. measured random (Paris method)
104
4.2 Simulation vs. measured results
From the results obtained for both simulation and calculated measurements, they
show similar trend for various samples, including foam, fibre, felt and Air Flow Resistance
(AFR) layers, with a variation of about 10-15%. The simulation results do not include the
low frequency resonance peaks, observed in normal incidence absorption measurements.
This could be because AutoSEA uses statistical averaging across frequencies, which
averages out the peaks.
B foam SAC validation
0.9
0.8
0.7
0.6
0.5
SAC
0.4
0.3
0.2
0.1
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
B foam simulation B foam measured random
Figure 6.5: Foam (open cell) sample B foam simulation vs. measured SAC
MJA316 SAC validation
1.2
0.8
SAC
0.6
0.4
0.2
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
MJA316 simulation MJA316 measured random
Figure 6.6: Sample MJA316 fibre simulation vs. measured SAC
105
Air Flow Resistant - Fibre (AFR) sample
S3D AFR SAC validation
0.9
0.8
0.7
0.6
0.5
SAC
0.4
0.3
0.2
0.1
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
S3D AFR simulation S3D AFR measured random
Figure 6.7: Sample S3D AFR simulation vs. measured SAC
Air Flow Resistant – Compressed poly-felt (AFR) sample
S3G AFR SAC validation
0.9
0.8
0.7
0.6
0.5
SAC
0.4
0.3
0.2
0.1
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
S3G AFR simulation S3G AFR measured random
Figure 6.8: Sample S3G AFR simulation vs. measured SAC
106
Poly-felt absorber sample
S3M AFR SAC validation
1
0.9
0.8
0.7
0.6
SAC
0.5
0.4
0.3
0.2
0.1
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
S3M AFR simulation S3M AFR measured random
Figure 6.9: Sample S3M AFR simulation vs. measured SAC
Foam (closed cell) sample
Q foam SAC validation
1.2
0.8
SAC
0.6
0.4
0.2
0
0 1000 2000 3000 4000 5000 6000 7000
Frequency (Hz)
S3M AFR simulation S3M AFR measured random
Figure 6.10: Sample S3M AFR simulation vs. measured SAC
107
It should be noted that the comparison of homogeneous materials is considered, i.e.

multi-layered or mixed samples are avoided, because AutoSEA assumes the bulk properties
of the material to be continuous throughout the sample.
CONCLUSIONS
• Simulated and measured results of material sound absorption show the similar trend
for various samples, with a variation of only about 10-15%.
• The simulation results do not include the low frequency resonance peaks, observed in
normal incidence absorption measurements.
• Material samples should be simulated for homogeneous samples only. Multi-layered

or non-homogeneous samples will give erroneous results, if simulated as a single
material.
• A statistical angular averaging of absorption, following Paris method (AS/NZS

1935.1:1998/Appendix) can be applied to convert measured normal incidence
absorption coefficient into random incidence absorption coefficient. (Ref: E. T. Paris,
Proc. Roy. Soc. 115, 407 (1927) )
108
CHAPTER 7
VIRTUAL MODELLING AND VALIDATION, AT COMPONENT

AND VEHICLE LEVELS
1 INTRODUCTION
This chapter outlines the various aspects about virtual modelling of component and
vehicle level acoustics using the Statistical Energy Analysis models for carpet design
evaluation. Alpha Cabin simulation is used for component level comparative study of the
carpets and the results are discussed. A brief explanation of the various stages of the
simulation process is provided for a better understanding of the virtual modelling and
evaluation cycle. The vehicle virtual SEA model is validated by a transfer function based
model evaluation process and finally the vehicle level evaluation of various carpets is
conducted and the comparative results are analysed.
Virtual modelling and testing is a widely accepted CAE (Computer Aided

Engineering) tool for improving the product design and predicting performance at early
stages of vehicle development. This tool allows for early identification of design problems,
solving the problems before prototype build, which saves the tooling cost and time, caused
by the design changes at the later stages. Vehicle sound package includes floor trims, head
liners, door trims and other acoustic components, which are usually tested at component
level in acoustic labs, according to various international & national automotive design
standards. For example, the reverberation room or Alpha Cabin for measuring sound
absorption coefficient and anechoic room/reverberation room test for measuring sound
transmission loss. However, this requests tooling for moulding these components for each
measurement. In order to evaluate these trim components in the sound package, intensive
109
vehicle level tests will have to be conducted, which leads to the compromised solutions by
trial and error. Further more, any design changes recommended by the test data will induce
tooling changes, which are very time and cost consuming.
Virtual modelling tools such as AUTOSEA 2 can be used to simulate the component
and vehicle level tests during early sound package design stages and optimize these trim
components and thus the sound package design, which saves the time and cost caused by the
later changes and provides a highly repeatable standard platform for design comparison and
benchmarking.
Following sections also illustrates how to use the software AUTOSEA 2 for carpet
design evaluation at the component and vehicle level for design concepts developed in the
previous sections.
2 COMPONENT LEVEL (ALPHA CABIN) MODELLING
The random incidence absorption coefficient of automobile trims is preferably measured

inside smaller versions of the Reverberation chambers, called Alpha Cabin. Normally it is
approximately 1/3rd the size of a standard Reverberation chamber. A typical Alpha Cabin has
a sample size area of 1.2 m2, which is sufficient for testing vehicle trim parts individually.
As a part of the project, an Alpha Cabin has been designed for physical tests [67], and its
CAD isometric view is shown below.
Figure 7.1: Alpha Cabin CAD design drawing

[67]
110
The following are the peculiarities of the Alpha Cabin design, which are also
considered while developing the SEA model.
• None of the cabin walls are parallel to each other, in order to avoid standing waves
• The cabin wall absorption is kept low by using sheet metal surface.
• The noise sources are small speakers, placed in different points, preferably corners, to
generate a diffused sound field
A SEA model of the Alpha Cabin, having similar dimensions as the real Alpha Cabin
design and the cabin walls made of sheet metals, is developed. The nodal points and
structural components were first designed in CAD and then imported to AutoSEA acoustic
software. This was opted, as AutoSEA geometrical modeling interface is not as efficient as
CAD.
Figure 7.2: SEA model of Alpha Cabin
The Alpha Cabin is modeled in such a way that it encloses a single fluid (air) cavity
inside, into which the carpet or sound package will be introduced. The below figure is the
modeled Alpha Cabin, with the cabin walls made semi-transparent, for through visibility.
Further, a SEA model of the current production carpet, i.e. the Mitsubishi 380 carpet
as shown below, is developed, having about 18 zones/sections in total. These are sheet metal
sections, applied with Noise Control Treatment, are made such a way that they correspond to
the sections on the real carpet, for which the section wise absorption coefficient and
transmission loss were measured separately.
111
1
Front 15 2 Front
16 3
4
12 5 11
6 10
18 17 14
9
7 8
13
Figure 7.3: Current production (M380) Figure 7.4: Carpet model labelled into 18
carpet sections
2.1 Cavity modelling

The modelling of the fluid (air) cavities properly plays a key role in getting the right
SEA model of a system. Each cavity is a fluid subsystem, which connects with the other
Main cabin cavity
Carpet cavities
Carpet cavities
Figure 7.5: Cavity sections in the SEA model of the Alpha Cabin
112
subsystems through the area junctions of the cavity faces.
Similar to the single big cavity that the Alpha Cabin encloses, there are a few cavities
that fill the bottom of the carpet structure. These form the connection between the main cabin
cavity and the bottom wall of the cabin. These cavities are required, as the carpet bottom is
not assumed to be flat, but curved as the real carpet. The grey semi-transparent sections are
the cavities, as shown in Figure 7.5.
2.2 Noise sources

Similar to the physical Alpha Cabin, three noise sources of equal power (1V
reference) are placed on three different faces of the cabin cavity walls. Compared to one
noise source, three sources improve the diffused field of sound in the cabin. The absolute
powers of the noise sources are not important as we are evaluating for comparative results
only.
Figure 7.6 illustrates the noise source locations and gives an idea of the virtual Alpha
Cabin test setup. The procedure involved further is to solve for the SEA parameters, which
takes a minute or two, and then to select the main cabin cavity for plotting the sound pressure
level results, under Engineering Units heading.
Source 1 (top)
Source 2 (front)
Source 3 (side)
Figure 7.6: Complete virtual Alpha Cabin SEA model, with noise sources and test carpet
113
2.3 Virtual modelling test results
Virtual Alpha Cabin tests were conducted for 4 different cases, which would aid a
comparative study of the noise reduction by the carpet. The reference for the noise level in
the comparative study is the case where there is no NCT (Noise Control Treatment) on the
sheet metal sections of the carpet structure. This is equivalent to no sound package inside the
cabin.
The current production carpet of Mitsubishi 380 is tested in two forms, one with the
best absorption coefficient of all the sections (i.e. section 5) applied throughout the carpet
Current production – M380 (best SAC)
Figure 7.7: Test results of the virtual Alpha

Cabin
and the second form where each section (1 to 18) is assigned a user defined NCT data, which
are the impedance tube measured absorption coefficient and transmission loss values.
114
Acoustic parameters of the Concept 3 carpet material were measured using

impedance tube and applied as a user defined NCT data for modelling and testing. It is to be
noted that, since we are using the experimentally measured sound absorption and
transmission loss data as user defined NCT data, the layered material level acoustic
predictions are expected to match perfectly. The resulting plots are compared in the figure
below.
2.4 Result discussion

The graph plot in the Figure 7.7 shows the sound pressure levels measured in the
Alpha Cabin cavity for various carpet materials, which are compared with those without
carpet/noise control treatment.
There are two current production (MITSUBISHI 380) carpet SPL results for
comparison, the dark green curve is the actual current production carpet, with each section
(1-18) having impedance tube measured absorption coefficient and TL values; meanwhile
the light green curve represents the noise level when the best sound absorption coefficient of
all the 1-18 sections, i.e. section 5 SAC coefficient, is used for all the other sections also.
These two differ by only about 1.5dB in the low frequencies and about 1dB at higher
frequencies.
The current production carpet seems to reduce the SPL in the cabin by 4-5 dB at both
the low and high frequencies, in comparison with the cabin without the carpet. An SPL
reduction of this range, i.e. greater than 3dB, will be reasonably an audible noise reduction
for an average listener. At 400 Hz, highlighted by the grey circle, there is a SPL dip as
expected, because the best sample of current production carpet (section 5) has a peak
absorption coefficient at this same frequency.
It is shown by the curve in blue that the Concept 3 carpet consistently reduces further
the SPL by about 4.2 dB above 2 KHz and reduces only 1.5 - 2 dB SPL below 2 KHz, in
comparison with the section wise simulated current production carpet. At low frequencies,
i.e. less than 1 kHz, the sound absorption performance of both current production and
concept 3 carpets are in the similar range.
Meanwhile, for the Concept 3 special mats padded with current production
MITSUBISHI 380 carpet, SPL curve is shown in Figure 7.7 as the bluish green line. The
Concept 3 special mats reduce the noise levels by 3dB SPL above 2 kHz and by 1 dB SPL at
1 kHz.
115
Thus, the random incidence Alpha Cabin simulation tests indicate that concept 3
carpet design can reduce the noise level (SPL) inside a vehicle cabin by up to 4.2 dB at high
frequencies ( above 2 kHz) and up to an average of 1 dB at low frequency range (below
1kHz), in comparison with the current production carpet.
3 VEHICLE LEVEL MODELLING

Virtual simulation and validation for acoustics is a developing field, and is well
applied in building acoustics and component level design. Compared to these applications,
vehicle acoustic prediction and design based on virtual validation demands highly accurate
simulation software, as there are variety of acoustic package components and structural
components, which interact in the acoustic field.
The AutoSEA modelling and acoustic simulation software is a well suited tool for the
purpose. Based on Statistical Energy Analysis models for each component of the vehicle,
AutoSEA generates a subsystem matrix of SEA parameters, which is solved for arriving at
noise levels at a particular location in the vehicle. The frequency range of SEA based
acoustic simulation is 500Hz-20kHz, as the modal density of air-borne sound is less below
500Hz, and hence can give less accurate results for lower frequencies. But this frequency
range (500Hz-20kHz) is sufficient for air-borne acoustic simulations of vehicles. AutoSEA
acoustic simulation is becoming a standard acoustic simulation tool of vehicle designers.
The vehicle model used for simulation is a generic sedan model, which would be the
preferred popular vehicle model. Geometric details of the car model can be directly imported
from CAD or CATIA model of a car. AutoSEA has provided with the Template Modeller
module, which is a powerful tool to morph the generic SEA model to the required imported
car geometry. So, for simulating the acoustics of sound packages like floor trim, seats etc,
which are not vehicle specific, it is preferred to do simulation on a generic SEA model.
This document will try to outline the key sections of the modelling process and will
keep aside SEA software specific details. A brief explanation about each vehicle model
section is given below.
3.1 Engine compartment

Figure 7.8 will help to give a picture about the engine compartment model of the
vehicle. For clarity, the vehicle sections, other than the engine compartment, have been
replaced by a semi-transparent grey section.
116
Engine cavity
Foot rest
Fire wall
Figure 7.8: SEA model of engine

compartment
This vehicle SEA model is for air-borne acoustic simulation, and hence a noise
source in the location of the engine cavity is sufficient to run the virtual validation of engine
noise. The engine compartment has followed a box structure in general, as the actual shape
doesn’t make a significant difference in the generated acoustic field, other than its
directivity.
The firewall, in orange shade, is made up of a curved multi plate structure, which
matches a normal car fire wall structure. The firewall is applied with specific noise control
treatment, which is explained below. Meanwhile, the red shaded section forms the foot rest
area for the front passengers, and is normally covered with the carpet section.
3.2 Passenger cabin

The passenger cabin model includes almost all the components, i.e. the front and
back seats, control dash, chassis, roof, doors with loud speaker vents, floor carpet, pillars,
glass windows (not shown) etc, which are found in a usual sedan car, except for specific
details like the steering wheel, air conditioning vents, mirrors etc, which are not important in
an acoustic point of view.
The acoustic modelling has been done inside the passenger cabin, especially the
driver side, and so compared to the engine and vehicle rear compartments, the passenger
cabin is expected to be closer in detail to the real car cabin. A transfer function analysis is
conducted as a part of the validation process for the vehicle model, where the measured
117
acoustic path transfer functions from the engine and tyre locations to the driver ear are
compared with the simulated ones from the SEA model. The results are provided in the
following sections.
Figure 7.9: Passenger cabin simulation SEA model
3.3 Vehicle rear compartment

Similar to the engine compartment, the rear compartment is a simple box shaped
structure, which includes the trunk, the spare tyre cabin and the wheel arches. The red
coloured section is the section beneath the back seats, which will be covered with the floor
carpet.
Spare wheel cabin
Wheel arch
Under seat
Figure 7.10: Rear compartment of vehicle SEA

model
118
The complete vehicle SEA model is illustrated in Figure 7.11, one model with
opaque surfaces and the other with semi-transparent surfaces. It is difficult to give a detailed
picture of the full model, as it includes a lot of components or sub-systems. A basic statistics
of the components involved is given in Table 1.
Table 1: Statistic of vehicle SEA model

Statistics of vehicle SEA model
Nodes 802
Materials 91
Physical Properties 46
Noise Control Treatments 83
Sound Packages 11
Groups 14
Subsystems 468
Junctions 1619
Loads & Constraints 5
Load Cases 0
Spectra 100
Figure 7.11: Complete vehicle, sedan car, SEA virtual

model
119
3.4 Acoustic cavity modelling
The acoustic cavity modelling is a key to the accuracy of the SEA modelling process,
as air-borne transmission paths between sub-systems depend on how well placed and
connected are the acoustic cavities and the vehicle body components.
The vehicle cavities can be separated into interior and exterior cavities, for
convenience, as explained below.
3.4.1 Vehicle interior cavities
As shown in Figure 7.12, the vehicle interior cavities are expected to fill up all the
empty spaces between the vehicle components and make the air-borne connection between
the SEA sub-systems. The cavities are modelled by specifying the closed nodes surface
which encloses the cavity.
Trunk cavity
Back passenger
ear level cavity
Engine cavity
Inside seat cavity
Wheel arch cavity
Driver ear level
Figure 7.12: Vehicle interior acoustic cavities
120
The cabin cavity has been divided into totally 8 zones, i.e. there are upper and lower
cavities for each of the four passenger spaces. The number of cavities can be increased for
more accuracy, but in our evaluation process this number is sufficient for measuring average
SPL levels at the driver’s ear.
3.4.2 Exterior cavities
Engine and tyre noise that escapes into the air or space surrounding the vehicle can
enter into the vehicle cabin by various acoustic phenomena like diffraction or by indirect
coupling of the vehicle component with the cavity of interest. So the vehicle is made
surrounded by cavities similar to the vehicle interior cavities, as shown in Figure 7.13. Each
exterior cavity is connected directly to the external surface of the vehicle body
Figure 7.13: Vehicle exterior acoustic

cavities
121
3.5 Sound Packages
The vehicle sound package of interest is a collection of the passenger cabin interior
trim components. Each section is treated using different material combination for noise
reduction, and hence has different sound absorption and sound transmission loss coefficients.
The interior vehicle sound package can be divided into 8 sets, for ease of modelling, as
follows.
1. Vehicle floor carpet – This covers the complete vehicle floor, below the seats and a
portion of the fire wall as foot rest area, and normally acts as a sound barrier.
2. Headliner – This covers the vehicle roof area and is normally absorptive.
3. Passenger seats – These are expected to be the most absorptive of all.
4. Trunk cover – This normally has a similar absorption as the floor trim.
5. Pillars – They are usually treated with plastics or similar synthetic material.
6. Control board – This area is mostly made of plastics or leather sections.
7. Doors – The doors are normally made in a double panel structure, with a leather or
synthetic leather skin, of low absorption.
8. Firewall – This is a more barrier like treatment, than an absorptive layer, which
separates the passenger from the engine compartment.
Refer the figures in the following two pages, to get a figurative idea of the sound package
sections and the measured sound absorption coefficient curves used for the inputs of each
section.
122
Figure 7.14: Carpet section of sound Figure 7.15: Headliner section of sound
package package
Figure 7.16: Seats section of sound packageFigure 7.17: Trunk cover section of sound package
123
Figure 7.18: Pillars section of sound Figure 7.19: Control dash board
package section of sound package
Figure 7.20: Doors section of sound package Figure 7.21: Fire wall section of
sound package
124
It is notable that the last 4 sound package sections are having almost zero absorption
coefficients, as these sections are normally covered with plastics or leather like synthetic
skins.
3.6 Noise sources
Vehicle noise sources consist of mainly power train (engine) noise, tyre/road noise,
wind noise, external traffic noise, vehicle exhaust noise and wiper noise. Out of these, we are
interested in the primary noises only, i.e. engine noise and tyre noise.
Figure 7.22 and 7.23 illustrate the engine and tyre noise source locations in the
vehicle simulation model. Since we are conducting comparative analysis and transfer
function model validation, instead of the actual noise source spectrum, the diffused field
noise sources are assigned to be white noise of 1V reference power.
Figure 7.22: Power train (engine) noise Figure 7.23: Tyre noise sources
source
3.7 SEA model validation using transfer function method
Virtual model validation is an important stage in the modelling process and there are
various techniques used for the same purpose. Transfer function method is a method used in
acoustics to find the spectral ratio of the source to receiver noise path. Once the transfer
function of a noise path is known, it can be applied to predict the noise spectrum at the
receiver for any other noise source type at the same source position. The transfer function is
normally an array of complex numbers, having both magnitude and phase, but here we are
125
interested only in the magnitude ratio and represented in dB, where the reference is the noise
power at the source location.
Figure 7.24: Transfer function test

setup
Figure 7.25: Omni directional loudspeaker Figure 7.26: Microphone at wheel arch
at the driver’s ear
In the test setup, since linear acoustic systems obey the reciprocity of source and
listener position, the Omni directional loudspeaker source (by Bruel & Kjaer) is placed at the
driver ear position, as shown in Figure 7.25. This also helps to avoid moving around the
loudspeaker across various parts of the vehicle. The microphones were placed at the engine
bay and four wheel arches, for measuring the noise level at those locations, as shown. A
microphone was also placed at the driver’s ear position to record the reference SPL at the
source location.
126
Measurements were done using both sine sweep and pink noise sources, as sine
sweep helps to improve the signal to noise ratio and pink noise helps to keep the power
focused in the low frequency range of interest.
The resulting ratio of the noise spectra, at the engine bay and four wheel arches, over
the noise spectra at the source location of the driver’s ear, are calculated and compared with
the virtual SEA model results, as shown in the Figures 27 and 28.
Sine Sweep Waveform (Wheel to Cabin)
20
0
Transfer Function (dB)
-20
-40 SEA Simulation

-60
-80
-100
10 1010 2010 3010 4010 5010 6010 7010 8010 9010
Frequency (Hz)
| P Wheel Right Back / P Cabin |

| P Wheel Right Front / P Cabin |
| P Wheel Left Back / P Cabin |
| P Wheel Left Front / P Cabin |
SEA simulation model - | P Wheel Right Front / P Cabin |
Figure 7.27: Wheel arch to driver’s ear, transfer function
Sine Sweep Waveform (Engine to Cabin)
20
10
0
Transfer Function (dB)
-10
-20
-30
-40
-50 SEA Simulation

-60
-70
-80
10 1010 2010 3010 4010 5010 6010 7010 8010 9010
Frequency (Hz )
| P Engine / P Cabin | SEA simulation model - | P Engine / P Cabin |
Figure 7.28: Engine to the driver’s ear transfer function
127
The transfer function results for both the engine to the ear and the wheel arch to the
ear show a good level of match between the real experimental tests on the vehicle and the
simulation. As shown by the light blue shaded area in Figure 7.27, there seems to be a
deviation for the simulation of right front wheel arch to the driver’s ear transfer function at
about 2 - 3 kHz range. This could be because of the coincidence frequencies of the sheet
metal panels used in simulating the vehicle. Averaging out the transfer function results for all
the 4 wheel arches to driver ear simulation is expected to reduce this deviation from the
actual measured transfer function for tyre/road noise.
3.8 Evaluation of carpets using Vehicle SEA modelling
Once the vehicle model is validated to show matching results with the real car test
results, we are at a good position to run evaluation tests for the various carpet designs and to
see the effects of introducing each carpet. Since we are interested in a comparative study of
the carpet contribution in noise reduction, we start with noise level without the carpet as the
reference noise level, against which other carpets are compared with. This is shown in Figure
7.29 as a black dot-dashed line, which shows the highest dB noise level.
The ideal carpet is the one when floor noise control treatment having 100% sound
absorption at in-cabin side and having infinite (450dB) transmission loss. The SPL results of
this carpet help to know the maximum possible noise reduction potential offered by the floor
trims.
128
Vehicle cabin noise level - Engine noise source
50
45
40
Noise level (dB)
35
30
25
20
15
10
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Freq (Hz)
No carpet
M380 carpet
Concept 3 : Ktuft + S3D+PT1foam10_scrim + M-380 carpet 25mm
C3 mat + M380-25mm (Ktuft + S3D+PT1foam10_scrim + M-380 carpet)
Ideal carpet
Figure 7.29: Noise spectrum at the driver’s ear, for engine noise source
Referring to Figure 7.29 for engine noise reduction by carpets, it is clear that
the Concept 3 design reduces the noise level by about 3-5 dB compared to the current
production MITSUBISHI 380 carpet design. It is worth noting that the Concept 3 results are
very close to the ideal carpet results. The Concept 3 mat with current production carpet
seems to give only about 1dB of improvement at high frequencies, compared to the carpet
only.
129
Vehicle cabin noise level - Tyre noise (4 no)noise sources)

(Four
50
48
46
44
Noise level (dB)
42
40
38
36
34
32
30
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Freq
No carpet (Hz)
M380 carpet - area wise SAC/TL
Concept 3 : Ktuft + S3D+PT1foam10_scrim + M-380 carpet 25mm
C3 mat + M380-25mm (Ktuft + S3D+PT1foam10_scrim + M-380 carpet)
Ideal carpet
Figure 7.30: Noise spectrum at the driver’s ear, for tyre noise
Figure 7.30 shows the tyre noise results for various carpets. The noise reduction
improvement from carpets is not significant, and the difference in the driver’s ear SPL for
the no carpet case and the ideal carpet is just 2-4dB.
So these virtual vehicle model tests indicate that floor trim can reduce the sound
pressure level inside the vehicle cabin caused by engine or power train noise, but not much
caused by tyre noise.
3.8.1 Thermo grams of engine and tyre noise paths
Figure 7.31 shows the thermo gram of the air-borne noise paths/levels for 500Hz,
1kHz and 2 kHz frequencies. The colour contour extends from red (high-94dB) to blue (low-
10dB) SPL noise levels. A close analysis can help in identifying noise zones, silent zones
and key transmission paths.
130
Engine noise @ 500Hz Engine noise @ 1kHz Engine noise @ 2kHz
Tyre noise @ 500Hz Tyre noise @ 2 kHz Tyre noise @ 1 kHz
Figure 7.31: Thermogram for engine and tyre noise, at various frequencies
131
4 CONCLUSIONS
Alpha Cabin evaluation
• In comparison with the current production carpet (MITSUBISHI 380 carpet),

Concept 3 carpet consistently reduces about 4.2 dB SPL above 2kHz and reduces up
to average of 1 dB below 1 kHz.
• Concept 3 special mats reduce the noise level by 3 dB above 2 kHz and up to an
average of 1 dB below 1kHz, in comparison with the current production carpet
(MITSUBISHI 380 carpet).
Vehicle level modelling and carpet evaluation
• The vehicle virtual SEA model was evaluated/validated using transfer function
method and well matching results were achieved for engine and tyre noises, except
for the coincidence frequency ranges of the backing sheet metal floor.
• Concept 3 carpet evaluation at the vehicle level shows the reduction of the in-cabin
noise level further by up to 3-5 dB, compared to the current production MITSUBISHI
380 carpet design, for power train noise.
• The Concept 3 carpet evaluation at the vehicle level shows the reduction of the in-
cabin noise level further by 2 dB at high frequencies, compared to the current
production MITSUBISHI 380 carpet design, for the tyre-road noise.
• The Concept 3 mat padded with the current production carpet (MITSUBISHI 380)
reduces noise level by about 1dB SPL at high frequencies, compared with the carpet
only case for the power train and tyre-road noise.
The concept 3 carpet SPL results are very close to the ideal carpet SPL results, which mean
that the concept 3 carpet has nearly reached the noise reduction limit.
132
CHAPTER 8
PROTO-TYPE VALIDATION AND EVALUATION OF DESIGN

CONCEPT
1 INTRODUCTION
The design concepts proposed for the acoustically improved vehicle carpet system
has been proto-type built factory moulded carpet samples, following the standard
manufacturing procedures, and validated by the test results of both the laboratory made
samples and the proto-type samples. This is required to verify the manufacturing process
related factors that can alter or degrade the acoustic performance.
In addition to verifying manufacturing process effects, it is required to test the

moulded carpet system in a real life test environment, and compare the psycho-acoustical
measurement results with the benchmark carpets. In case the complete carpet could not be
moulded following the standard manufacturing process, it is proposed to build proto-type
special mats, having the required additional top layers only, which can be placed on top of
existing carpet, to form a carpet system similar to the proposed concept design.
Two construction methods were undertaken to produce samples for the C3-Special
mat, designated as C3-Special_mat_Lab-made and C3-Mat_Prototype, respectively. The
physical testing program consisted of material acoustic property and in-situ vehicle tests. The
former was designed to evaluate the acoustic effectiveness of the prototypes through the
measurement of the sound absorption coefficient (SAC), whilst the latter aims to provide
direct comparison of noise reduction performance among various mat types fitted to the
vehicle floor carpet.
133
2 TEST SAMPLES
Based on practical considerations, the development and construction of the C3-

Special mat was pursued because of limitations in the production of a full C3 type vehicle
floor carpet. Concept 3 (C3) carpet design is explained in previous chapters. The prototype
mat (C3-Mat_Prototype), having similar material specifications as the proposed C3 carpet,
was intended to also demonstrate improved acoustic performance over other
contemporary/standard mat products.
Specimens were cut to size per requirement of the testing procedure. For the
impedance tube, a 29mm and 100mm diameter samples were needed. The alpha cabin test
require a 1x1m sample and the vehicle test demands that specimens be cut to the size and
shape of conventional car mats.
2.1 Standard Mat (reference)

Physical characteristics of the standard mat are summarized in Table 1. Standard mat
specimens were provided as reference to compare the acoustic performance of the new
design concepts. These were cut to size to conform to the requirements of the testing regimes
outlined in the following section.
2.2 Prototype Mat (Concept 3 design)

The proof of concept for an acoustically improved carpet is embodied on the
proposed C3-Special mat. The C3-Mat_Prototype is a simplified version of the Carpet
Design Concept 3 floor carpet, i.e. without the thin heavy layer and de-coupler. In addition,
absorber layer thickness was reduced (e.g. 7mm) to conform to conventional mat overall
thickness of 15mm.
The development of the C3-Mat_Prototype follows the material and construction

specification previously described in the concept design explained in previous chapters. The
following two versions of the product concept were realized and were subjected to the testing
protocols.
2.2.1 C3-Special_mat_Lab-made
The C3-Special_mat_Lab-made specimen was constructed under laboratory

conditions. The material specifications and order of layered construction (see C1-07 RMIT
M06_addendum report) was followed but the components were bonded by glue instead of
134
thermal lamination. The 3M spray adhesive was used for this purpose. Due to practical
considerations, only specimens for use in the impedance tube tests were made and tested.
Description of the C3-Special_mat_Lab-made is given in Table 1.
2.2.2 C3-Mat_Prototype
The C3-Mat_Prototype specimen was constructed by employing the expertise of

professional laminator recommended by Futuris. Material components and construction
details were followed from the C1-07 RMIT M06Addendum report recommendations.
Various material components were either sourced locally or provided by Futuris.
A technical problem encountered during the lamination process was that when the
AFR layer (Air Flow Resistance layer; material P830N is a high performance mouldable
acoustic material sourced from a third party supplier in Melbourne) was heated, it does not
develop a “gummy” consistency to facilitate bonding with the other layers (namely the
tufting and the absorber layer, i.e. 7mm thick open cell foam with scrim on top (labelled
PT1withScrim); sourced from a supplier in Melbourne). Nevertheless a C3-Mat_Prototype
Model 1 was successfully laminated with the use of polyethylene (PE) dusting. The second
attempt using only thermal bonding was not successful hence the only usable specimen was
designated as C3-Mat_Prototype. Relevant physical properties are listed in Table 1.
2.3 Constraints and Limitations

Discrepancies in the material properties between the proposed and actual materials
used in the construction of the prototype mat were noted. Shrinkage to the final product due
to heat treatment during the lamination was also observed. The specified overall thickness of
the C3-Mat_Prototype was 15mm but due to shrinkage during lamination the specimen
thickness was measured at 11.94 mm (see Table1).
The thin layer of PE as backing for the tufting and the use of slight PE dusting during
the lamination process for the C3-Mat_Prototype is duly noted. The presence of added PE
may degrade the performance of the AFR resulting to reduced sound absorption potential.
2.3.1 Tufting
The K type tufting without backing was originally proposed for the mat prototype.
However, the tufting provided by Futuris for the construction of the C3-Mat_Prototype had a
shorter pile length (see Table 1) and had a thin layer of PE as backing. Due to time
135
constraints, it was decided to use this tufting material for the construction of the C3-
Mat_Prototype. The C3-Special_mat_Lab-made specimen used the K tufting (with the
longer pile length) without backing.
2.3.2 Lamination
Since thermal bonding/lamination trials failed, the use of polyethylene (PE) powder
to facilitate the lamination of C3-Mat_Prototype was resorted to. A second attempt to
construct the mat prototype by thermal lamination was not successful. The use of scrim/web
could be a future alternative to help the lamination and was suggested by the laminator and
the supplier of the AFR material.
Table 1 - Sample Specimen Properties
Mat Specimen Properties

Average Pile
Sample Particulars gsm Thickness Length
2
(g/m ) (mm) (mm)
Standard Mats Production Carpet Mats 2997.28 8.71 4.5
K-Tufting plus AFR layer
plus open cell foam
(label:PT1) with scrim, 15.00 5.1
bonded by spray
C3-Special_mat_Lab-made adhesive
Tufting plus AFR layer
plus open cell foam
(label:PT1) with Scrim, 2218.07 11.94 2.4
laminated with PE
C3-Mat_Prototype (Model 1) dusting
Tufting plus AFR plus
open cell foam
(label:PT1) with scrim,
C3-Mat_Prototype (Model 2) thermally bonded **** Final product "unusable"
136
3 PROTOTYPE TESTING
The testing protocols employed were designed to directly compare the acoustic
performance of the prototype mat with that of the standard mat. The testing regimes
consisted of material property tests and performance testing inside a vehicle cabin.
The metric of primary concern is the sound absorption coefficient (SAC). This metric is
decidedly more important than transmission loss (TL) when evaluating the relative
performance of the mats, as variable in the tests was the top layer which is the absorptive
side. In addition, it was expected that the floor carpet of the Mitsubishi 380 (supplied by
Futuris) had been optimised for good TL performance.
3.1 Material Acoustic Property Testing
3.1.1 Impedance Tube Test
Normal incidence sound absorption coefficients (SAC) were measured using the
B&K impedance tube (PULSE™ Acoustic Material Testing in a Tube Type 7758).
The mat specimen samples were subjected to impedance tube testing namely: C3-
Special_mat_Lab-made, C3-Mat_Prototype and the standard mat. The impedance tube set-up
is shown in Figure 8.1. For more details about the test setup, please refer to the B&K manual
and Chapter 5.
Figure 8.1: Impedance tube setup
137
3.1.2 RMIT Alpha Cabin Test
The newly built RMIT Alpha Cabin testing chamber [67], shown in Figure 8.2, was
used to measure the random incidence sound absorption coefficient of the sample specimens.
Direct comparisons will be possible because of identical measurement conditions and
standard specimen sizes employed. The C3-Mat_Prototype and the standard mat specimens
were tested in the alpha cabin. The C3-Special_mat_Lab-made specimen was excluded from
this test because of lack of appropriate sample size. The alpha cabin test required a 1x1m
material specimen.
The procedure developed during the alpha cabin tests employed the B&K PULSE set-
up to measure T60 decay time measurements, done similar to a reverberation room
measurement and following the strict guidelines contained in AS ISO 354-2006 “Australian
Standards: Acoustics – Measurement of sound absorption in a reverberation room”. The
RMIT Alpha Cabin has three 50-W co-axial speakers for sound source and three
microphones were used to monitor the SPL inside the chamber. The reverberation time
measurements were taken in three (3) replications for each sample. The method also required
three (3) different configurations, i.e. the microphone and sound source placement variations
inside the Alpha Cabin. Overall, for each sample specimen, the T60 decay time
measurements were repeated 9 times, for statistical averaging. The random incidence SAC
was then calculated from the averaged T60 decay times, using Sabine’s equations.
1m x 1m sample
Figure 8.2: RMIT Alpha Cabin
138
3.1.3 Vehicle testing using in-situ methods
In addition to the material acoustic property tests, in-situ vehicle cabin performance
testing was conducted on the C3-Mat_Prototype and the standard mats. The “mats-on/mats-
off” technique was employed, which was previously done in carpet vehicle benchmarking.
This involved changing the car mats and taking repeated acoustic measurements, to obtain
relative performance measures of the different car mats for direct comparison.
The vehicle tests were conducted under stationary and on-road conditions. The
former will focus mainly on the effects of the engine noise while the later will take into
account the contributions of tyre-road interaction noise also. In addition, 2nd gear slow
acceleration tests were also conducted specifically to monitor the distinctive engine noise i.e.
combustion cycle noise. The stationary tests were conducted in a quiet and open spot at
RMIT Bundoora campus itself, away from the reflective buildings. The on-road tests were
conducted in a relatively quiet, well paved suburban road, on sunny and fair weather days.
The road surface is representative of normal smooth bitumen paved arterial roads in Victoria
with a roughness level of <4.2 IRI (International Roughness Index, taken from VIC Roads
Annual Report 2007).
Two test set-ups were also employed – Aachen HMS (Head Measurement System)
III psychoacoustic measurements and the PULSE Acoustic measurement system using
Figure 8.3a: Aachen HMS III and KMT Tachometer set-up
139
calibrated pressure condenser microphones (2.5”, 200V polarization). The HMS set-up is
shown in Figure 8.3a and the test run way specifics are shown in Figure 8.3b, while the
PULSE set-up is illustrated in Figure 8.4. The HEAD acoustics measurement system details
and its requirement are explained in [55].
Run A
Mic
Run B
T = 5s
Figure 8.3b: Test run way specifics
Figure 8.4: PULSE set-up (left); Microphone position on floor carpet (right)
The car used during the vehicle testing is a Mitsubishi 380 Series III, 2007 Model SX
with a 3.8L V6 engine. The mileage is 26765 km. Before the testing, the car tyres were
inflated to the recommended normal operating conditions of 32psi (220kPa) for all tyres.
This particular test vehicle was chosen because its floor carpet is reportedly optimized for
dealing with vehicular noise in specific problem frequencies, and hence considered as the
reference design.
140
4 TEST RESULTS
4.1 Impedance Tube Tests
The normal incidence SAC measured for the three mat specimens are shown in Figure
8.5. The C3-Mat_Prototype gave better SAC than the standards mat at the low to middle
frequency ranges. This can be an advantage in managing low frequency noise, such as engine
or power-train noise.
Impedance Tube SAC Measurements

0.9
0.8
Sound Absorption
0.7
Coefficient
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 1000 2000 3000 4000 5000 6000
Frequency (Hz)
C3-Special_Mat_Lab-made C3-Mat_Prototype Standard Mat
Figure 8.5: Impedance tube SAC of mat specimens
The lab made Concept 3mat specimen (C3-Special_mat_Lab-made) had the better
SAC at high frequency range compared to either the C3-Mat_Prototype mat or the standard
mat. This can be due to the fact that the lab made specimen uses K-tufting having longer pile
thickness, which eventually can improve sound absorption. Figure 8.6 show that the pile
length influences the SAC of tufting materials. In addition, some form of degradation
occurred in the C3-Mat_Prototype during lamination, e.g. shrinkage as evidenced by the
reduced mat thickness, which might have a negative impact on its sound absorption
capability. The shrinkage of the absorber layer, in this case the foam with scrim component,
can degrade the SAC potential of the prototype mat.
141
Tufting Comparison
1.0
Absorption Coefficient (%) 0.8
0.6
Pile thickness = 4.5mm
0.4
0.2
Pile thickness = 2.4mm
0.0
0 1000 2000 3000 4000 5000 6000
Frequency (Hz)
Lab made C3-Special

tufting (K)Tufting Proto-type tufting (K)
BlkTuft(Futuris)
Figure 8.6: Effect of tufting pile length on SAC
4.2 Alpha Cabin Tests
SAC measurements from the RMIT Alpha Cabin for the C3-Mat_Prototype and
standard mat specimens are shown in Figure 8.7. The C3-Mat_Prototype SAC showed an
average improvement of about 14.2 % over that of the standard mat. Highest improvement is
about 40 % in SAC at 800 Hz.
Alpha Cabin SAC Measurements

1.0
0.9
0.8
Sound Absorption Coefficient
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 1000 2000 3000 4000 5000
Frequency (Hz)
Standard Mat C3-Mat_Prototype
Figure 8.7: Alpha cabin SAC of mat specimen
142
The advantage of the standard mat having a longer tufting pile length, which
enhances SAC performance, is to be offset while comparing the constructions; whereas the
C3-Mat_Prototype has a distinct absorber layer beneath the AFR. The standard mat
construction consists of the tufting backed by a thin heavy layer on a rough underlay. This
construction limits the penetration of sound waves into the interior of the carpet/mat mass for
further absorption or dissipation, and hence the less SAC performance on cabin side.
4.3 In-situ Vehicle Tests
Sound pressure levels (SPL) near the driver’s ear were monitored as an indication of
noise levels inside the car cabin. Repeated measurements, while different mats were fitted
into the floor carpet, will indicate the relative performance levels of the test mats.
4.3.1 Stationary Conditions
Figure 8.8 shows the typical SPL levels at the driver’s ear level during stationary
conditions, with the engine revved up at various rpm’s as measured using PULSE Acoustic
measurement system. The C3-Mat_Prototype showed improved noise reduction capability
compared to the standard mat. It is particularly effective in the low frequency ranges (e.g. < 1
kHz) where the engine noise is prevalent.

Proto_Mat : N,2Krpm_Prototype-Mat_Set3 : Input : CPB Analyzer
70
60
50
40
30
20
10
Standard Mat (Total SPL = 77.4 dB(A))
0 Prototype Mat (Total SPL = 74.9 dB(A))

31.5 63 125 250 500 1k 2k 4k 8k
[Hz]
Figure 8.8: Total SPL at Driver's Ear level during stationary conditions at 2000
engine rpm (PULSE data)
143
This trend is applicable to other engine rpm levels where the C3-Mat_Prototype
outperforms the standard mat. Please refer Appendix 1 for more results. Figure 8.9 shows the
consolidated SPL data for the stationary tests at various engine rpm’s. The C3-
Mat_Prototype consistently reduces the noise level inside the car cabin for various engine
rpm’s. The average reduction in the total SPL is about 2.1 dB (A) compared to that of the
standard mat.
It is notable that the vehicle level simulation results of concept 3 mat on M380
standard carpet, for engine noise, showed similar possible improvement of around 1dB,
especially at the higher frequency ranges. These predictions are compared to standard carpet
system (M380), rather than standard mat. However, the simulation results are quite practical
in predicting the SPL range.
SPL at Driver's Ear; Stationary Conditions

(PULSE Data)
80
Total SPL (dB(A))
78
76
74
72
70
1000 2000 3000
Engine RPM
Standard Mat Prototype Mat
Figure 8.9: Summary of total SPL at driver's ear level for various
engine rpm's during stationary tests (PULSE data)
Analyses of data for stationary conditions using HMS III followed the trend observed
in PULSE data although at a generally lower SPL level. Note that the ArtemiS 4.00 Head
Analyser software, used on the HMS III data, calculates the average SPL, while the PULSE
system calculates total (peak or sum) SPL, and hence the different numerical values for SPL
level quantification. Figure 8.10 shows the consolidated HMS III data for various engine
rpm’s during stationary tests. The reduction in the average SPL, obtained from HMS III data,
attributed to the use of C3-Mat_Prototype as compared to the standard mat is 0.46 dB(A), for
all engine rpm’s considered.
144
M-380 Cabin Noise at Stationary Conditions

(HMS III Data)
60
Average SPL (dB[A])
50
40
Engine on Idle
30
0 1000 2000 3000
Engine RPM
Figure 8.10: Summary of in-cabin average SPL at various engine rpm's

during stationary tests (HMS III data)
This improvement can be further illustrated when looking at the Articulation Index
(AI) report shown in Figure 8.11. Use of the C3-Mat_Prototype consistently improves the in-
cabin AI compared to that of using the standard mat. Other HMS data for the stationary tests
are provided in Appendix 2.
M-380 Cabin AI at Stationary Conditions

(HMS III Data)
100
Engine on Idle
95
90
0 1000 2000 3000
Engine RPM
Figure 8.11: Summary of in-cabin articulation index (AI) during

stationary tests (HMS III data)
145
4.3.2 On-Road Constant Speed test results
A summary of in-cabin noise levels from PULSE measurements are shown in Figure
8.12. The C3-Mat_Prototype performed better than the standard mat for all range of constant
speeds. The averaged improvement in noise reduction, in the form of total SPL, is about
0.625 dB (A). Individual total SPL plots for the driving speed tests are provided in Appendix
3.
M-380 Cabin Noise at Constant Speed Test

(Pulse Data)
78
Total SPL at DE (dB(A))
76
74
72
40 60 80 100
Constant Speed (kph)
Figure 8.12: Summary of total SPL at driver's ear level during on-road
constant speed tests (PULSE data)
Results from the analyses of HMS III for constant speed tests were not conclusive.
Figure 8.13 shows the in-cabin averaged SPL plotted against the driving speeds considered.
It appears that the differences in the performance of C3-Mat_Prototype and the standard mat
are marginal. The measured AI at 3000 rpm in Figure 8.14 shows a marginal advantage of
using the C3-Mat_Prototype over standard mat. However, overall analyses of AI results
indicate that the performances of the mats are comparable, as shown in Figure 8.15. Relevant
HMS III data during constant speed tests are shown in Appendix 4.
146
M-380 Cabin Noise at Various Driving Speeds

(HMS III Data)
70
Average SPL (dB[A])
65
60
40 60 80 100
Figure 8.13: Summary of in-cabin average SPL during on-road

constant speed tests (HMS III data)
C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s).Articulation Index vs t (300.0ms). AI/%

100
Std_Mat_Neutral_3000rpm-3 ( 0.00-10.67 s) 98.75

Articulation Index vs t (300.0ms)
Right : 93.9 %
C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s) 97.5
Right : 94.4 %
96.25
95
93.75
92.5
91.25
90
0 1 2 3 4 t/s 5 6 7 8
Figure 8.14: Measured AI at 3000 rpm, stationary conditions (HMS data)
147
M-380 Cabin AI at Various Driving Speeds

(HMS III Data)
100

90
80
70
60
40 60 80 100
Figure 8.15: Summary of in-cabin Articulation Index (AI) during on-road constant
speed tests (HMS III data)
4.3.3 2nd Gear Slow Acceleration Test
Figure 8.16 shows the 2nd order SPL inside the vehicle cabin, measured near the
driver’s ear level, during the 2nd gear slow acceleration on-road tests. The 2nd order SPL
isolates the specific engine noise frequencies, thus making it possible to analyse its effect to
the overall noise reduction strategy. The graph in Figure 8.16 indicates that the C3-
Mat_Prototype provided a distinct advantage over the standard mat at the frequency range of
2.1-3.0k rpm (100~156 Hz). This can be useful in managing low frequency engine noise
entering the vehicle cabin. Such insight is important for developing future improvements in
mat/carpet acoustic capability.
[dB(A)/20.0u Pa] Autospectrum (micro 4) - M ark 2 - slice

W orking : ProtoMat_set3_A : Run-Up : FFT Analyzer
60
56
52
48
44
40
36
32
28
Standard Mat
24
Prototype Mat
20
1.2k 1.6k 2k 2.4k 2.8k 3.2k 3.6k 4k
[RPM] (Speed tachor pm )
Figure 8.16: 2nd order SPL at driver's ear level for run-up test (PULSE data)
148
5 CONCLUSIONS
• A proof of concept mat (C3-Mat_Prototype) was constructed albeit slight

modifications from the original specifications due to material, time and technical
constraints. These deviations pertaining to tufting pile length, presence of thin PE at
the back of the tufting, and shrinkage during lamination were noted.
• In the impedance tube SAC test, the C3-Mat_Prototype performed better than the
standard mat in the low to mid frequency ranges (up to 3000Hz).
• RMIT Alpha Cabin SAC test showed that the C3-Mat_Prototype outperformed the
standard mat by an average of 14.2 percentage points. Highest improvement is about
40 percentage points in SAC at 800 Hz.
• On-road vehicle tests show that the C3-Mat_Prototype has better sound absorption at
the low to mid frequency range than Standard mat. PULSE data indicates an
improvement in the reduction of in-cabin total SPL by the prototype mat by about
2.1 dB(A) during stationary tests and 0.625 dB(A) during constant driving tests.
• HMS III results are consistent with PULSE results during stationary vehicle tests.
149
Appendix 1. Stationary Test – PULSE Results

Comparative plots of in-cabin total SPL, from PULSE™ data measured in the
vicinity of the driver’s ear, during stationary tests on the vehicle fitted with a standard mat
and C3-Mat_Prototype. A separate binaural test is conducted using Head Acoustics (HMS
III), which will takes into consideration of subjective and directivity parameters of the
vehicle noise perceived by an average human listener.
[dB(A)/20.0u Pa] Autospectrum(mic4) - Current [dB(A)/20.0u Pa] Autospectrum(mic4) - Mark 3

No_Mat : Idle,in_Drive_No_Mat_Set3 : Input : CPB Analyzer Rubber_Mat : Idle,N_Rubber-Mats_Set3 : Input : CPB Analyzer
70 70
60 60
50 50
40 40
30 30
No Mat 20 No Mat
20
Noise Reduction Rubber Mat Rubber Mat
10 10 Standard Mat
Standard Mat
Prototype Mat Prototype Mat
0 0
31.5 63 125 250 500 1k 2k 4k 8k
31.5 63 125 250 500 1k 2k 4k 8k [Hz]
[Hz]
[dB(A)/20.0u Pa] Autospectrum(mic4) - Mark 2 [dB(A)/20.0u Pa] Autospectrum(mic4) - Mark 2

Proto_Mat : N,1Krpm_Prototype-Mat_Set3 : Input : CPB Analyzer Proto_Mat : N,3Krpm_Prototype-Mat_Set3 : Input : CPB Analyzer
70 70
60 60
50 50
40 40
30 30
20 No Mat 20 No Mat
Rubber Mat Rubber Mat
10 Standard Mat 10 Standard Mat
Prototype Mat Prototype Mat
0 0
31.5 63 125 250 500 1k 2k 4k 8k 31.5 63 125 250 500 1k 2k 4k 8k
[Hz] [Hz]
150
Appendix 2. Stationary Test – HMS III Results

Average SPL and AI results for standard mat and C3-Mat_prototype at stationary conditions.
C3_Mat_Idle_Neutral-3 ( 0.00- 9.73 s ).6thOctave (4096,66.6%). L/dB(A)[SPL] C3_Mat_Neutral_1000rpm-3 ( 0.00- 8.96 s).6thOctave (4096,66.6%). L/dB(A)[SPL]
40
30
30
20
10 20
0 10
Std_Mat_Idle_Neutral-2 ( 0.00- 9.64 s )
6thOctave (4096,66.6%) Std_Mat_Neutral_1000rpm-2 ( 0.00-10.15 s)
Right : 38.4 dB(A)[SPL] 6thOctave (4096,66.6%)
C3_Mat_Idle_Neutral-3 ( 0.00- 9.73 s ) -10 Right : 40.2 dB(A)[SPL]
6thOctave (4096,66.6%) C3_Mat_Neutral_1000rpm-3 ( 0.00- 8.96 s) 0
Right : 37.5 dB(A)[SPL] 6thOctave (4096,66.6%)
Right : 41.2 dB(A)[SPL]
-20
-10
20 50 100 200 f/Hz 1000 2000 5000 10k 20k 20 50 100 200 f/Hz 1000 2000 5000 10k 20k
C3_Mat_Neutral_2000rpm-2 ( 0.00-11.01 s).6thOctave (4096,66.6%). L/dB(A)[SPL] C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s).6thOctave (4096,66.6%). L/dB(A)[SPL]
60
40
50
30
40
20 30
10 20
Std_Mat_Neutral_2000rpm-3 ( 0.00- 8.53 s)
6thOctave (4096,66.6%) Std_Mat_Neutral_3000rpm-3 ( 0.00-10.67 s)
10
Right : 48.9 dB(A)[SPL] 0 6thOctave (4096,66.6%)
C3_Mat_Neutral_2000rpm-2 ( 0.00-11.01 s) Right : 55.5 dB(A)[SPL]
6thOctave (4096,66.6%) C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s)
Right : 46.3 dB(A)[SPL] 0
6thOctave (4096,66.6%)
-10 Right : 55.7 dB(A)[SPL]
-10
20 50 100 200 f/Hz 1000 2000 5000 10k 20k 20 50 100 200 f/Hz 1000 2000 5000 10k 20k
C3_Mat_Idle_Neutral-3 ( 0.00- 9.73 s).Articulation Index vs t (300.0ms ). AI/% C3_Mat_Neutral_1000rpm-3 ( 0.00- 8.96 s ).Articulation Index vs t (300.0ms). AI/%
100 100
99 99
98 No_Mat_Neutral_1000rpm-1 ( 0.00- 8.02 s ) 98

Right : 99.6 %
No_Mat_Idle_in-Drive-3 ( 0.00- 9.22 s ) Std_Mat_Neutral_1000rpm-2 ( 0.00-10.15 s)
Articulation Index vs t (300.0ms ) 97 Articulation Index vs t (300.0ms) 97
Right : 100.0 % Right : 99.8 %
Std_Mat_Idle_Neutral-2 ( 0.00- 9.64 s ) C3_Mat_Neutral_1000rpm-3 ( 0.00- 8.96 s )
Articulation Index vs t (300.0ms ) Articulation Index vs t (300.0ms)
Right : 99.8 % Right : 99.9 %
96 96
C3_Mat_Idle_Neutral-3 ( 0.00- 9.73 s)
Articulation Index vs t (300.0ms )
Right : 100.0 %
95 95
0 1 2 3 t/s 4 5 6 7 0 1 2 3 t/s 4 5 6 7
C3_Mat_Neutral_2000rpm-2 ( 0.00-11.01 s).Articulation Index vs t (300.0ms). AI/% C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s).Articulation Index vs t (300.0ms). AI/%
100 100
No_Mat_Neutral_3000rpm-3 ( 0.00-10.92 s)
Articulation Index vs t (300.0ms )
98.75
Right : 92.4 %
99 Std_Mat_Neutral_3000rpm-3 ( 0.00-10.67 s)
Articulation Index vs t (300.0ms ) 97.5
Right : 93.9 %
C3_Mat_Neutral_3000rpm-1 ( 0.00- 8.79 s )
Articulation Index vs t (300.0ms ) 96.25
98
Right : 94.4 %
95
No_Mat_Neutral_2000rpm-3 ( 0.00- 9.64 s)
Right : 98.8 % 97 93.75
Std_Mat_Neutral_2000rpm-3 ( 0.00- 8.53 s)
Right : 98.3 % 92.5
C3_Mat_Neutral_2000rpm-2 ( 0.00-11.01 s) 96
Articulation Index vs t (300.0ms) 91.25
Right : 99.3 %
95 90
2 3 4 t/s 6 7 8 0 1 2 3 4 t/s 5 6 7 8
151
Appendix 3. Constant Speed Test – PULSE Results
Comparative plots of in-cabin total SPL, from PULSE™ data measured in the
vicinity of the driver’s ear, during constant speed tests on the vehicle fitted with a standard
mat and C3-Mat_Prototype.

Working : ProtoMat_set3_60_A : Input : CPB Analyzer Template : CS_ProtoMat_40_set3_A : Input : CPB Analyzer
70 70
60 60
50 50
40 40
30 30
20 20
Standard Mat (Total SPL = 75.8 dB(A)) Standard Mat (Total SPL = 75.3 dB(A))
Prototype Mat (Total SPL = 74.9 dB(A)) Prototype Mat (Total SPL = 74.7 dB(A))
10 10
31.5 63 125 250 500 1k 2k 4k 8k 31.5 63 125 250 500 1k 2k 4k 8k
[Hz] [Hz]

Working : ProtoMat_set3_80_B : Input : CPB Analyzer Working : ProtoMat_set3_100_A : Input : CPB Analyzer
70 70
60 60
50 50
40 40
30 30
20 20
Standard Mat (Total SPL = 76.0 dB(A)) Standard Mat (Total SPL = 76.2 dB(A))
Prototype Mat (Total SPL = 75.4 dB(A)) 10

Prototype Mat (Total SPL = 75.8 dB(A))
10
31.5 63 125 250 500 1k 2k 4k 8k 31.5 63 125 250 500 1k 2k 4k 8k
[Hz] [Hz]
152
Appendix 4. Constant Speed Test – HMS III Results

Average SPL and AI results, for standard mat and C3-Mat_prototype samples, at
various driving speeds.
CS_C3_Mat_60 ( 0.00- 8.70 s).6thOctave (4096,66.6%). L/dB(A)[SPL] CS_C3_Mat_40 ( 0.00- 8.02 s).6thOctave (4096,66.6%). L/dB(A)[SPL]
50 50
40 40
30 30
20 20
CS_StdMat_60 ( 0.00-10.50 s) CS_StdMat_40 ( 0.00-10.50 s)
6thOctave (4096,66.6%) 6thOctave (4096,66.6%)
Right : 63.4 dB(A)[SPL] 10 Right : 61.5 dB(A)[SPL] 10
CS_C3_Mat_60 ( 0.00- 8.70 s) CS_C3_Mat_40 ( 0.00- 8.02 s)
Right : 63.5 dB(A)[SPL] Right : 60.9 dB(A)[SPL]
0 0
20 50 100 200 f/Hz 1000 2000 5000 10k 20k 20 50 100 200 f/Hz 1000 2000 5000 10k 20k
CS_C3_Mat_80 ( 0.00- 8.19 s).6thOctave (4096,66.6%). L/dB(A)[SPL] CS_C3_Mat_100 ( 0.00-11.09 s).6thOctave (4096,66.6%). L/dB(A)[SPL]
60 60
50 50
40 40
30 30
CS_StdMat_80 ( 0.00-10.50 s) 20 CS_StdMat_100 ( 0.00-10.67 s) 20

CS_C3_Mat_80 ( 0.00- 8.19 s) CS_C3_Mat_100 ( 0.00-11.09 s)
10 6thOctave (4096,66.6%) 10
6thOctave (4096,66.6%)
0 0
20 50 100 200 f/Hz 1000 2000 5000 10k 20k 20 50 100 200 f/Hz 1000 2000 5000 10k 20k
CS_C3_Mat_40 ( 0.00- 8.02 s).Articulation Index vs t (300.0ms). AI/% CS_C3_Mat_60 ( 0.00- 8.70 s).Articulation Index vs t (300.0ms). AI/%
100 CS_NoMat_60 ( 0.00-11.09 s) 100
CS_NoMat_40 ( 0.00-11.09 s)
Articulation Index vs t (300.0ms) Articulation Index vs t (300.0ms )
Right : 94.3 % 98.75 Right : 89.3 %
CS_StdMat_60 ( 0.00-10.50 s) 97.5
CS_StdMat_40 ( 0.00-10.50 s )
Articulation Index vs t (300.0ms) Articulation Index vs t (300.0ms )
97.5 Right : 91.2 %
Right : 94.0 %
CS_C3_Mat_60 ( 0.00- 8.70 s) 95
CS_C3_Mat_40 ( 0.00- 8.02 s)
96.25 Articulation Index vs t (300.0ms )
Right : 90.1 %
Right : 94.3 %
92.5
95
93.75 90
92.5 87.5
91.25
85
90
0 1 2 3 t/s 4 5 6 7
0 1 2 3 t/s 4 5 6 7
CS_C3_Mat_80 ( 0.00- 8.19 s).Articulation Index vs t (300.0ms). AI/% CS_C3_Mat_100 ( 0.00-11.09 s).Articulation Index vs t (300.0ms ). AI/%
100 100
CS_NoMat_80 ( 0.00-11.01 s )
Articulation Index vs t (300.0ms ) CS_NoMat_100 ( 0.00-11.09 s ) 95
Right : 79.6 % 95 Articulation Index vs t (300.0ms )
CS_StdMat_80 ( 0.00-10.50 s ) Right : 65.5 %
90
Articulation Index vs t (300.0ms ) CS_StdMat_100 ( 0.00-10.67 s )
Right : 78.3 % Articulation Index vs t (300.0ms )
90 85
CS_C3_Mat_80 ( 0.00- 8.19 s) Right : 64.6 %
Articulation Index vs t (300.0ms ) CS_C3_Mat_100 ( 0.00-11.09 s)
Right : 80.5 % Articulation Index vs t (300.0ms ) 80
85 Right : 64.4 %
75
80
70
65
75
60
70
0 1 2 3 t/s 4 5 6 7 0 1 2 3 t/s 4 5 6 7
153
CHAPTER 9
DESIGN OPTIMIZATION (TAGUCHI METHOD BASED), AT

MATERIAL, COMPONENT AND VEHICLE LEVELS
1 ABSTRACT & INTRODUCTION
This chapter outlines the design constraints for the proposed concept 3 carpet design
and discusses about how to go forward improving the acoustic performance. The need for
efficient optimization techniques is introduced and Taguchi loss function method is
explained with brief theoretical background. A detailed Taguchi method analysis is
conducted on concept 3 design for cabin side sound absorption performance and
Transmission loss performance separately, using a simulated model from Statistical Energy
Analysis tools. The results are analysed to reach at conclusions regarding the optimization
directions for future design improvement or customization for specific noise control issues in
vehicles.
Optimization of the carpet concept design against various constraints is a key to an

acoustically efficient design. Taguchi analysis method is considered as the suited statistical
optimization tool and applied for component level and sound package level analysis also.
The sound package and carpet section wise Taguchi analysis procedure and results are
outlined, which can provide directions for further optimization of the concept design. The
154
analysis is conducted using Statistical Energy Analysis models of Alpha Cabin for
component level analysis and generic Sedan model for vehicle level analysis.
Outlining the performance requirements and constraints for the design and
conducting an analysis using optimization techniques gains insight into the key parameters
that contribute to meet the performance requirements, while adhering to the constraints. The
basic design concept labelled Concept 3, which has 2 additional layers, i.e. Air Flow
Resistant (AFR) layer and Absorber layer, added to the current production carpet design, is
the basis for going forward with the detailed analysis and optimization of the carpet design.
The Concept 3 carpet design, chosen from the earlier proposed four concepts, is shown
below.
Tufting
Cabin side noise AFR layer
Absorption layer
Heavy layer
De-coupler layer
40mm
Noise path
(from engine)
Figure 9.1 – Concept 3
Refreshing briefly, Concept 3 design introduces an Air Flow Resistant (AFR) layer
and Absorber layer just below the carpet tufting, compared to the current production carpet
design. This design approach was taken because the current production carpet design has a
relatively low sound absorption coefficient on the cabin side of the carpet. The measurement
results of the proposed Concept 3 design (40mm thick) in Figure 9.2 below show evidently
the improved sound absorption coefficient in cabin side, compared to the current production
(40mm thick) carpet shown in green line. The two concept-3 combinations of interest,
having different AFR layers, are
155
Sample 1
Tufting layer (sample K) + AFR layer (sample S3D) + Absorber layer (sample SA_sx
- Visco Elastic foam 20mm with scrim) + SL - ( Heavy Layer 1.7kgsm and De-
coupler layer N foam of 20mm)
Sample 2
Tufting layer (sample K) + AFR layer (sample S3E) + Absorber layer (sample SA_sx
- Visco Elastic foam 20mm with scrim) + SL - ( Heavy Layer 1.7kgsm and De-
coupler layer N foam of 20mm)
Sound Absorption Coefficient - Concept 3
1.2
1.0
0.8
SAC
0.6
0.4
0.2
0.0
0 1000 2000 3000 4000 5000 6000 7000
Fre que ncy Hz
Sample 1
Ktuff+S3D+SA_sx+SL Sample 2
Ktuff+S3E+SA_sx+SL
Figure 9.2 – Concept 3 absorption measurement

results
1.1 Optimization requirement & techniques

Once the basic carpet design concept is formulated, with the given constraints like
weight, thickness, cost and performance targets, we need to reach the best layered material
combination that provides the most competitive performance from the design. This special
combination of materials with appropriate parameters is expected to perform the best, while
meeting the defined constraints.
Various approaches can be taken to discover this right combination of layers, but
time and resources are always natural constraints. So, we try to rely on existing and proven
efficient methods or models, to get directions regarding optimizing for performance.
156
A Statistical Energy Analysis (SEA) model has been developed for the proposed
Concept 3, to conduct simulated performance measurements. The Concept 3 simulation
model is initially tuned up and matched with the corresponding experimentally measured
performance results. Thus, further optimizing techniques can be worked out using this
simulation model, at a faster pace.
Performance sensitivity analysis for various parameters of the layered materials is

one way of studying the critical parameters to be optimized. Taguchi Loss Function method
is one such optimal sensitivity analysis technique, based on statistical mathematical
observations for quality improvement. Taguchi method involves running a few minimum test
cases and arriving at a weighted table for the various parameters controlling the quality or
performance of a system. A theoretical background is also given in the next sections.
Meanwhile, a possible optimization procedure for sound-package level is to identify

the acoustically significant sections of the carpet which contribute in reducing the cabin
noise and to apply the treatment in those carpet sections in an extensive manner, compared to
the less significant sections. Special noise reduction structures or materials can be applied to
these significant sections, to attain a better noise isolation while maintaining an optimum
trade-off with various constraints.
1.2 Concept 3 constraints and performance targets

The key constraints for the concept design are
1. Weight
2. Thickness
3. Cost of manufacturing
At the current stage of design, only the weight and thickness are considered as
variables of concern as we are conducting the optimizing process on a generic Concept 3
carpet design. Though there has not been a defined constraint for weight, the current
production carpet having a heavy layer of 1.7kgsm, which would measure totally
4.56kgsm for the whole carpet, is taken as the reference weight constraint. The thickness
constraint has been defined as 40mm (+5mm). The optimization techniques will be
worked out around these design constraints.
Meanwhile, Taguchi Analysis for both Alpha cabin and vehicle level analysis are
conducted on the carpet sections for the cavity Sound Pressure Level (SPL) as the output. In
157
the carpet section ON/OFF Taguchi analysis, the carpet is divided into 7 different zones of
approximately equal area and each carpet section zone noise control treatment is switched
ON or OFF according to the setting in the Taguchi table. Meanwhile, for Sound Package
analysis, the complete vehicle cabin trim is divided into 7 categories, and each category
treatment is switched ON or OFF. Separate Taguchi analyses are conducted for engine noise
and tyre noise sources, at low (315 Hz - 1 kHz), mid (1 kHz - 4 kHz) and high (4 kHz - 8
kHz) frequency ranges.
2 TAGUCHI METHOD - THEORY
Based on the principles of experimental design and statistical methods, and from his
years of research experience, Prof. Genechi Taguchi [60, 61] introduced an approach for
1. Designing products or processes so that they are robust to environmental conditions
2. Designing and developing products so that they are robust to component variation
3. Minimizing variation around a target value
These three goals of Taguchi method are referred to as “parameter design” goals.
Designing of experiments to meet the goals are sometimes made unnecessarily complicated,
inefficient and sometimes ineffective. Taguchi’s method advocates some novel methods of
statistical data analysis and approaches to the design of experiments.
Taguchi considers three stages in a product’s development to meet a target or goal:
1. System design (basic configuration)
2. Parameter design (specific values for system parameters)
3. Tolerance design (determine best tolerances)
Taguchi recommends that statistical experimental design methods be employed to

assist in quality improvement, particularly during parameter design and tolerance design.
Taguchi method is already being applied in a variety of applications by corporate firms,
including areas like automobile engineering, web design services etc [60, 61].
Generally, each product or process performance characteristics will have a target or

nominal value that delivers the customer with maximum value. The objective is to study the
variance or sensitivity around this target value. Taguchi method tries to model the departure
that may occur from this target value with a “loss function”. Taguchi method imposes a
158
quadratic loss function of the form as shown below, which also quantifies the customer
response to the variance from the target,
L(y) = k(y - T)2,
where, L(y) is the Taguchi loss function of variable parameter y, k is a statistically derived
constant that controls the slope of the quadratic curve and T is the target for the design
parameter y that bring maximum customer satisfaction.
L(y)
L(y) = k(y - T)2
Low Target Upper Spec.
Figure 9.3 – Taguchi loss function
The Taguchi loss function is evaluated by running a set of test runs on the product
being evaluated, by which the variance of the product’s performance is measured for a
defined variation of each parameter in the product. The parameters are selected in such a way
that they are orthogonal in nature, in the context of the tests undertaken. This process is made
efficient by running only the necessary tests, following a Taguchi table.
Below is a “7 parameters - 8 runs” Taguchi table. Each parameter (P-1 to P-7) is

considered to have two possible values (A and B). For example, if P-1 is the testing
temperature, then 18 degree Celsius (A) and 20 degree Celsius (B) are the two temperature
values for running the tests. So a test run can be designed by a combination of these 7
parameters, each having the A or B value. The below Taguchi table has 8 such different test
runs. Note that though the total permutation of 7 parameters (A & B) involves 27 test runs,
the below 8 test runs are selected for running the orthogonal set of Taguchi tests.
159
Table 1 – Taguchi loss function table
Input variables Performance
P-1 P-2 P-3 P-4 P-5 P-6 P-7
Run 1 A A A A A A A Output 1
Run 2 A A A B B B B Output 2
Run 3 A B B A A B B Output 3
Run 4 A B B B B A A Output 4
Run 5 B A B A B A B Output 5
Run 6 B A B B A B A Output 6
Run 7 B B A A B B A Output 7
Run 8 B B A B A A B Output 8
Avg. of level 1 (A’s) X1 X2 X3 X4 X5 X6 X7
Avg. of level 2 (B’s) Y1 Y2 Y3 Y4 Y5 Y6 Y7
Diff. Xn - Yn(effect) W1 W2 W3 W4 W5 W6 W7
Y1 = (o/p5+o/p6+ o/p 7+ o/p 8)/4 o/p = Output
X1 = (o/p 1+ o/p 2+ o/p 3+ o/p 4)/4
The resulting weighting factors W1 to W7 indicate the prominence of a parameter in

the output performance, by varying the parameters P1 to P7 respectively. So, if the variance
ratio (B-A)/A of the parameters are the same, for example 20%, then the parameter which
has the largest weight (W) is the most sensitive one. Hence, in an optimization point of view,
this parameter will be the one to first work on, in priority order, because a small addition to
this parameter should give the maximum output compared to others.
Thus, Taguchi Method helps to find out easily, within a few test cases, the sensitivity
order of the parameters for a given product or process being analysed or optimized.
160
3 MATERIAL LEVEL TAGUCHI METHOD OPTIMIZATION

RESULTS
In the case of Concept 3 design Taguchi method analysis, the sound absorption
coefficient and transmission loss are done separately, in two analysis tables. Further, the
frequencies have been subdivided into 3 groups, for which separate Taguchi analysis is done.
Note that the analysis is done on a slightly modified version or special case of concept 3
design, where there are two tufting/substrate layers, which can be considered as a mat placed
over a carpet, together forming a concept 3 carpet.
Low frequencies - 315 Hz to 1 KHz
Mid frequencies - 1 KHz to 4 KHz
High frequencies - 4 KHz to 8 KHz
For the case where there are more than 7 parameters that cannot fit into the Taguchi
table, one option is to increase the table size or the other option is to identify the less
significant parameters by first conducting an initial run of Taguchi tests with the first set of
parameters, using an ON/OFF Taguchi analysis, where instead of varying the parameter by a
percentage, the material layer itself is included (ON) or not included (OFF).
So we have included an ON/OFF Taguchi de-coupling analysis for both cabin side
sound absorption and transmission loss, which is followed by a detailed Taguchi analysis.
3.1 ON/OFF Taguchi de-coupling analysis
The selected orthogonal parameters, for initial ON/OFF analysis of the proposed
concept carpet design, are as shown below. The parameters are selected such a way that they
can be used for both cabin side sound absorption and transmission loss analysis of the carpet.
P1 - De-coupler
P2 - Substrate (tufting base)
P3 - Tufting
P4 - Absorption layer
161
P5 - Air Flow Resistant (AFR)
P6 - Substrate (top tufting base)
P7 - Tufting (top)
P3 - Tufting thickness
P4 - Absorption thickness
P2 – Substrate (tuft) thickness P5-AFR thickness
40mm
P7 Tufting thickness
P6 Substrate thickness
P1-De-coupler thickness
Figure 9.4: Taguchi orthogonal parameters for carpet concept design
3.1.1 Cabin side absorption: ON/OFF Taguchi tables
Tables 2 & 3 list the ON/OFF de-coupling analysis results for Concept 3 cabin side
sound absorption, where the target output is the sound absorption coefficient having
frequency average values between 0 and 1. If the material layers are not included (OFF), the
material layers are replaced by the same thickness of heavy layer (septum).
Table 2 – Concept 3 initial ON/OFF Taguchi based test run results

1. N foam (bottom decoupler)
4. B foam (top absorber)

2. Substrate (bottom)
3. Tufting (bottom)
6. Substrate (top)
7. Tufting (top)
5. S3D AFR
Input variables Averages
1 2 3 4 5 6 7 Low Mid High
Run 1 1 1 1 1 1 1 1 0.47 0.86 0.84
162
Run 2 1 1 1 2 2 2 2 0.08 0.00 0.00
Run 3 1 2 2 1 1 2 2 0.09 0.00 0.00
Run 4 1 2 2 2 2 1 1 0.08 0.11 0.20
Run 5 2 1 2 1 2 1 2 0.16 0.11 0.01
Run 6 2 1 2 2 1 2 1 0.18 0.11 0.14
Run 7 2 2 1 1 2 2 1 0.09 0.09 0.14
Run 8 2 2 1 2 1 1 2 0.22 0.03 0.00
Table 3 – ON/OFF Taguchi loss function table


2. Substrate (bottom)
3. Tufting (bottom)
6. Substrate (top)
7. Tufting (top)
5. S3D AFR
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 0.18 0.22 0.21 0.20 0.24 0.23 0.20
Avg. level2 0.16 0.12 0.13 0.14 0.10 0.11 0.14
Diff. (effect) 0.02 0.10 0.09 0.06 0.14 0.12 0.07
Mid (1KHz-4KHz)
Avg. level1 0.24 0.27 0.25 0.27 0.25 0.28 0.29
Avg. level2 0.08 0.06 0.08 0.06 0.08 0.05 0.04
Diff. (effect) 0.16 0.21 0.17 0.20 0.17 0.23 0.26
High (4KHz-8KHz)
Avg. level1 0.26 0.25 0.20 0.25 0.25 0.26 0.33
Avg. level2 0.08 0.09 0.09 0.09 0.09 0.07 0.00
Diff. (effect) 0.18 0.16 0.11 0.16 0.16 0.19 0.33
163
Note: The red numbers indicate the most sensitive parameters having high or close to high
weighting or Taguchi loss function value. The blue numbers indicate the second most
sensitive parameter.
3.1.2 Sound transmission loss: ON/OFF Taguchi tables
Tables 4 and 5 list the ON/OFF Taguchi concept 3 sound transmission loss de-
coupling analysis, where the target output is sound transmissions loss coefficient. If the
material layers themselves are not included (OFF), the material layers are replaced with the
same thickness of air gap.
Table 4 – Concept 3 ON/OFF Taguchi based test run results

2. Substrate (bottom, 5mm)

3. Tufting (bottom)
6. Substrate (top)
7. Tufting (top)
5. S3D AFR
1 2 3 4 5 6 7 Low freq Mid High
Run 1 1 1 1 1 1 1 1 25.94 47.27 62.45
Run 2 1 1 1 2 2 2 2 24.12 43.86 55.70
Run 3 1 2 2 1 1 2 2 25.50 46.45 61.09
Run 4 1 2 2 2 2 1 1 24.20 42.19 54.07
Run 5 2 1 2 1 2 1 2 20.31 40.87 70.50
Run 6 2 1 2 2 1 2 1 22.08 37.02 66.41
Run 7 2 2 1 1 2 2 1 20.95 38.65 69.24
Run 8 2 2 1 2 1 1 2 22.08 36.94 66.52
164
Table 5 – ON/OFF Taguchi loss function table
2. Substrate (bottom, 5mm)

3. Tufting (bottom)
6. Substrate (top)
7. Tufting (top)
5. S3D AFR
Low (315Hz-1KHz)
Avg. level1 24.94 23.11 23.27 23.18 23.90 23.29 23.29
Avg. level2 21.36 23.18 23.02 23.12 22.40 23.16 23.01
Diff. (effect) 3.59 -0.07 0.25 0.05 1.50 0.13 0.29
Mid (1.25KHz-
3.15KHz)
Avg. level1 44.94 42.26 41.68 43.31 41.92 41.82 41.28
Avg. level2 38.37 41.06 41.63 40.00 41.39 41.50 42.03
Diff. (effect) 6.58 1.20 0.05 3.31 0.53 0.32 0.75
High (4KHz-8KHz)
Avg. level1 58.33 63.77 50.78 65.82 64.12 63.39 63.04
Avg. level2 68.17 62.73 63.02 60.68 62.38 63.11 63.45
Diff. (effect) 9.84 1.03 12.24 5.14 1.74 0.28 0.41
165
3.1.3 Parameter elimination discussion
The above ON/OFF Taguchi tables for cabin side absorption and transmission loss
helps to identify the parameters that does not play a key role in improving the acoustic
performance of the carpet. There are 2 parameters that clearly show low profile in low and
mid frequencies, i.e. not coloured red or blue in the low and mid frequency tables, in terms
of the loss function, which are
a) Substrate (bottom),
b) Tufting (bottom) and
So, any of these parameters can be replaced for introducing new parameters, for a
detailed Taguchi analysis.
Hence, for going ahead with the detailed Taguchi analysis, the remaining parameters
included are,
P6) Heavy layer thickness and
P7) Heavy layer surface density (kgsm)
3.2 Detailed Taguchi analysis
The selected orthogonal parameters, for analysis of the proposed concept carpet
design, are as shown below. The parameters are selected such a way that they can be used for
both cabin side absorption and transmission loss analysis of the carpet.
P1 - De-coupler thickness
P2 - Substrate (top) thickness
P3 - Tufting (top) thickness
P4 - Absorption layer thickness
P5 - Air Flow Resistant (AFR) thickness
P6 - Heavy layer thickness
P7 - Heavy layer surface density (in kgsm)
166
P3 - Tufting thickness
P4 - Absorption thickness
P2 – Substrate (tuft) thickness P5-AFR thickness
40mm Heavy layer
P6-thickness
P7-GSM
P1-De-coupler thickness
Figure 9.5: Taguchi orthogonal parameters for carpet concept design
3.2.1 Taguchi analysis: Cabin side absorption
The below tables detail the Taguchi loss function calculations done for the carpet
concept in an evident manner.
Taguchi parameter table
Table 6 – Carpet parameters used in Taguchi analysis

Input variables
Level 1 Level 2
1. N foam (bottom de-coupler) thickness (m) 0.015 0.018
2. Substrate (top) thickness (m) 0.0005 0.0006
3. Upper (top) tufting thickness (m) 0.003 0.0036
4. B foam (top absorber) thickness (m) 0.015 0.018
5. S3D AFR thickness (m) 0.0032 0.00384
6. Septum (Heavy Layer) thickness (m) 0.0015 0.0018
7. Septum (Heavy Layer) kgsm 1.7 2.04
167
Test run table
Table 7 – Concept 3 Taguchi based test run results

1. N foam (bottom decoupler) thickness
6. Septum (Heavy Layer) thickness

4. B foam (top absorber) thickness
3. Upper (top) tufting thickness
7. Septum (Heavy Layer) gsm

2. Substrate (top) thickness
5. S3D AFR thickness
Input variables Average absorption coefficients (freq)
1 2 3 4 5 6 7 Low Mid High
Run 1 1 1 1 1 1 1 1 0.46633 0.860225 0.841579
Run 2 1 1 1 2 2 2 2 0.534223 0.871294 0.83339
Run 3 1 2 2 1 1 2 2 0.46769 0.865318 0.8451
Run 4 1 2 2 2 2 1 1 0.537464 0.868971 0.837163
Run 5 2 1 2 1 2 1 2 0.482397 0.881431 0.835819
Run 6 2 1 2 2 1 2 1 0.520675 0.881431 0.845255
Run 7 2 2 1 1 2 2 1 0.486145 0.859206 0.834327
Run 8 2 2 1 2 1 1 2 0.512789 0.883313 0.843565
168
1. N foam (bottom de-coupler) thickness



1 2 3 4 5 6 7
Low 315Hz-1KHz
Avg. level1 0.501427 0.500906 0.499872 0.475641 0.077106 0.502654 0.502654
Avg. level2 0.500502 0.501022 0.502057 0.526288 0.080424 0.502183 0.499275
Diff. (effect) 0.000925 0.000116 0.002185 0.050647 0.003318 0.00047 0.003379
Mid 1KHz-4KHz
Avg. level1 0.866452 0.873595 0.86851 0.866545 0.872572 0.873485 0.867458
Avg. level2 0.876345 0.869202 0.874287 0.876252 0.870226 0.869312 0.875339
Diff. (effect) 0.009893 0.004393 0.005778 0.009707 0.002346 0.004173 0.007881
High 4KHz-8KHz
Avg. level1 0.839308 0.839011 0.670572 0.839206 0.843875 0.839531 0.839581
Avg. level2 0.839741 0.840039 0.840834 0.839843 0.835175 0.839518 0.839469
Diff. (effect) 0.000433 0.001028 0.170262 0.000637 0.0087 1.35E-05 0.000112
169
3.2.1.1 Result discussion
Referring to the Taguchi loss function table for Sound Absorption Coefficient, the
loss function weighting results indicate that if there is need for improving the sound
absorption in low frequencies, we need to first increase the thickness of the absorber layer,
just below the carpet top tufting layer, and then if possible try increasing the AFR layer
thickness.
It is slightly surprising to note that Taguchi results show that increasing the heavy
layer GSM may increase the cabin side sound absorption for low frequency range. This
might be because the sound is transmitted through the heavy layer at near natural resonance
frequencies of the material, to be absorbed by the de-coupler layer just below the heavy
layer.
At mid frequency range, both the absorber layer and de-coupler layer thickness show
almost equal sensitivity. The concept 3 carpet sound absorption is quite good at mid and high
frequency ranges and the sensitivity of output is only about 0.95%. Similar is the case for the
high frequency range, where carpet tufting thickness and AFR layer thickness are the two
sensitive areas of improvement, with a higher sensitivity of about 17% for the upper tufting
layer.
Referring to the initial ON/OFF Taguchi tables, we can see that few parameters
which showed high sensitivity to ON/OFF loss function may show less sensitivity in the
detailed Taguchi analysis where a slight variation is applied to the parameters. This can be
expected because, like shown in the below Figure 9.6, in ON/OFF Taguchi it is the absolute
sensitivity or rate of output variation that comes into picture; meanwhile in detailed Taguchi
analysis it will be the local sensitivity or rate of output variation around the current design
parameter value. So, with orthogonally designed parameters, the ON/OFF de-coupling
Taguchi analysis gives a sensible for the effective sensitivity of the parameter.
run 3
170
run 3
Output run 2
Local δy/δx
run 1
ON
Absolute
OFF
δy/δx
Parameter
∆
Where,
ON - Contribution of parameter to target (output) value is ON
OFF - Contribution of parameter to target (output) value is OFF (zero)
∆ - Small (delta) deviation for parameter
Figure 9.6:analysis:
3.2.2 Taguchi Taguchi orthogonal parameters
Transmission lossfor carpet concept design
Taguchi parameter table
Table 9 – Carpet parameters used in Taguchi analysis (TL)

Input variables
Level 1 Level 2
1. N foam (bottom de-coupler) thickness (m) 0.015 0.018
2. Substrate (top) thickness (m) 0.0005 0.0006
3. Upper (top) tufting thickness (m) 0.003 0.0036
4. B foam (top absorber) thickness (m) 0.015 0.018
5. S3D AFR thickness (m) 0.0032 0.00384
6. Septum (Heavy Layer) thickness (m) 0.0015 0.0018
7. Septum (Heavy Layer) kgsm 1.7 2.04
171
Table 10 – Concept 3 Taguchi based test run results (TL)



Input variables Transmission Loss averages (freq)
1 2 3 4 5 6 7 Low Mid High
Run 1 1 1 1 1 1 1 1 25.94 47.27 62.45
Run 2 1 1 1 2 2 2 2 26.39 50.28 65.26
Run 3 1 2 2 1 1 2 2 25.88 49.18 64.08
Run 4 1 2 2 2 2 1 1 26.53 48.51 63.86
Run 5 2 1 2 1 2 1 2 26.01 49.86 67.49
Run 6 2 1 2 2 1 2 1 25.86 48.23 66.08
Run 7 2 2 1 1 2 2 1 25.82 48.23 65.88
Run 8 2 2 1 2 1 1 2 26.03 49.82 67.62
172
Taguchi loss function table
Table 11 – Taguchi loss function table (TL)



Low 315Hz-1KHz
Avg. level1 26.19 26.05 26.05 25.91 25.93 26.13 26.04
Avg. level2 25.93 26.06 26.07 26.20 26.19 25.99 26.08
Diff. (effect) 0.25 0.01 0.02 0.29 0.26 0.14 0.04
Mid 1.25KHz-3.15KHz
Avg. level1 48.81 48.91 48.90 48.64 48.63 48.87 48.06
Avg. level2 49.04 48.94 48.94 49.21 49.22 48.98 49.79
Diff. (effect) 0.23 0.02 0.04 0.57 0.60 0.11 1.73
High 4KHz-8KHz
Avg. level1 63.91 65.32 52.24 64.97 65.06 65.35 64.57
Avg. level2 66.77 65.36 65.38 65.70 65.62 65.33 66.11
Diff. (effect) 2.85 0.04 13.14 0.73 0.56 0.03 1.55
weighting or Taguchi loss function value. The blue number indicates the second most
sensitive.
173
3.2.2.1 Result discussion
From the above loss function table it is clear that at low frequencies (315Hz-1kHz),
the absorber thickness, AFR layer thickness and de-coupler layer play the key role, with the
heavy layer down in the line of sensitivity. This might be because the sound is transmitted
through the heavy layer at near natural resonance frequencies, to be absorbed by the absorber
layer just above the heavy layer. Further, the heavy layer and the AFR layer forms like a
double panel system where the absorber comes in between them, for which the transmission
loss increases by the gap between the AFR and heavy layers.
At mid frequency range the heavy layer comes back to picture, with the AFR layer
thickness and top absorber layers seconding in improving the transmission loss. The high
frequency sound insulation seems to be very sensitive to the increase of carpet tufting layer
thickness, seconded by the de-coupler thickness.
3.2.3 Summary of detailed Taguchi analysis
• The below table summarises the optimization weighting to be given while trying to
improve acoustic performance in each frequency range. Note that the Taguchi loss
function values are shown in brackets, beside the layer names placed in priority
order.
TL SAC (cabin side)
1. Absorber (0.29) 1. Absorber (0.05)
Low (315Hz - 1kHz) 2. AFR (0.26) 2. AFR (0.003)
3. De-coupler (0.25) 3. HL gsm(0.003)
1. HL gsm(1.73) 1. De-coupler (0.01)
Mid (1kHz - 4kHz) 2. AFR (0.6) 2. Absorber (0.01)
3. Absorber (0.57) 3. Tufting (0.006)
1. Tufting (13.14) 1. Tufting (0.17)

High (4kHz - 8kHz)
2. De-coupler (2.85) 2. AFR (0.01)
174
• At low frequency range (315Hz-1kHz), the heavy layer is expected to transmit sound
to the next layer and so the de-coupler below the heavy layer might absorb low
frequency noise from the cabin side.
4 Alpha Cabin based carpet section Taguchi analysis
4.1 Carpet section Taguchi analysis procedure

A seven parameter and eight run Taguchi analysis table is used for the carpet
analysis. The carpet zones are divided in such a way that it matches physically with the
separate areas of the carpet, like the foot rest zone, under seat zone etc, for separate analysis.
The orthogonal nature of the Taguchi parameters is not an issue in carpet sections, as
physically they are already orthogonal.
The acoustic significance of each carpet section is identified by verifying the loss
function results obtained for the ON/OFF Taguchi analysis. ON level is considered as the
case where there is carpet or NCT (Noise Control Treatment) on that section and OFF level
is the case where there is no NCT, i.e. with sheet metal facing. Following the 7 parameter – 8
runs Taguchi table as shown below, each run test is executed on the simulation model and
solved for obtaining the SPL at the cavity of measurement. In the case of Alpha Cabin
simulation model, the main cavity is the measurement cavity.
As shown in the sample Taguchi table, for each test run (row) the output, i.e. SPL of
the measurement cavity, is noted. After running the 8 test runs and obtaining the
corresponding outputs, each parameter loss function is calculated (column) by finding the
difference between the average of all the test run outputs with the parameter/carpet section
NCT ON and the average of outputs when the parameter/carpet section NCT is OFF. The
sample Taguchi table illustrates the loss function calculation procedure. Further, since the
carpet zones are not exactly equal in area, while calculating the loss function it is being
normalized to the loss function of 1m2 carpet sections, by dividing it by the section area.
175
Input variables Performance
P-1 P-2 P-3 P-4 P-5 P-6 P-7
Run 1 ON ON ON ON ON ON ON Output 1
Run 2 ON ON ON OFF OFF OFF OFF Output 2
Run 3 ON OFF OFF ON ON OFF OFF Output 3
Run 4 ON OFF OFF OFF OFF ON ON Output 4
Run 5 OFF ON OFF ON OFF ON OFF Output 5
Run 6 OFF ON OFF OFF ON OFF ON Output 6
Run 7 OFF OFF ON ON OFF OFF ON Output 7
Run 8 OFF OFF ON OFF ON ON OFF Output 8
Avg. of ON cases X1 X2 X3 X4 X5 X6 X7
Avg. of OFF cases Y1 Y2 Y3 Y4 Y5 Y6 Y7
Diff. Xn - Yn(effect) W1 W2 W3 W4 W5 W6 W7
o/p = Output
Y1 = (o/p5+o/p6+ o/p 7+ o/p 8)/4
X1 = (o/p 1+ o/p 2+ o/p 3+ o/p 4)/4
4.1.1 Carpet ON/OFF Taguchi sections
The carpet is divided into 7 approximately equal area zones for ON/OFF Taguchi
analysis. As shown below, the sections that come into the same zone have the same zone
number. Note that, a few similar or symmetric sections may not be included in the Taguchi
zones, for equalling the surface areas.
176
Front
5
3
3
6
5
4
1 4
6
1
7 2
Figure 9.7: Carpet zones for Alpha cabin based ON/OFF Taguchi analysis
4.2 Alpha Cabin (component level) Taguchi analysis results
The ON/OFF Taguchi analysis test results in the Alpha Cabin model are given in the
tables below.
Table 13: Taguchi Parameter table
Input variables
Level Level
Carpet zones 1 2
1. Zone1 ON OFF
2. Zone2 ON OFF
3. Zone3 ON OFF
4. Zone4 ON OFF
5. Zone5 ON OFF
6. Zone6 ON OFF
7. Zone7 ON OFF
177
Table 14: Taguchi test table
1. Zone1
2. Zone2
3. Zone3
4. Zone4
5. Zone5
6. Zone6
7. Zone7
1 2 3 4 5 6 7 Low Mid High
Run 1 ON ON ON ON ON ON ON 101.35 102.74 103.73
Run 2 ON ON ON OFF OFF OFF OFF 102.09 103.96 105.20
Run 3 ON OFF OFF ON ON OFF OFF 102.69 103.96 105.16
Run 4 ON OFF OFF OFF OFF ON ON 103.05 104.55 105.67
Run 5 OFF ON OFF ON OFF ON OFF 102.49 104.24 105.27
Run 6 OFF ON OFF OFF ON OFF ON 102.63 104.24 105.28
Run 7 OFF OFF ON ON OFF OFF ON 102.37 103.81 105.20
Run 8 OFF OFF ON OFF ON ON OFF 103.07 104.42 105.39
Since the carpet zones are not exactly equal in area, while calculating the loss
function it is being normalized to the loss function of 1m2 carpet sections, by dividing it by
the section area.
Table 15: Taguchi loss function table

1. Zone1
2. Zone2
3. Zone3
4. Zone4
5. Zone5
6. Zone6
7. Zone7
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 102.29 102.14 102.22 102.22 102.43 102.49 102.35
Avg. level2 102.64 102.79 102.71 102.71 102.50 102.44 102.58
Diff. (effect) 0.35 0.66 0.49 0.49 0.06 0.04 0.23
Area normalized (effect) 1.61 3.03 2.01 1.98 0.27 0.19 1.52
Mid (1KHz-4KHz)
178
Avg. level1 103.80 103.79 103.73 103.68 103.84 103.98 103.83
Avg. level2 104.17 104.18 104.24 104.29 104.14 103.99 104.14
Diff. (effect) 0.37 0.39 0.52 0.60 0.30 0.00 0.31
High (4KHz-8KHz)
Avg. level1 104.94 104.87 104.88 104.84 104.89 105.02 104.97
Avg. level2 105.29 105.36 105.35 105.39 105.34 105.21 105.26
Diff. (effect) 0.35 0.48 0.47 0.55 0.44 0.19 0.29
4.2.1 Result discussion
ON/OFF Taguchi analysis in Alpha Cabin shows that zones 2, 3 and 4 dominate in
the low and high frequencies and zones 3 and 4 dominate in the mid frequencies, in reducing
the average SPL inside the Alpha cabin. It is to be noted that the Alpha Cabin simulations are
conducted using multiple point sources, to avoid sensitivity to the source location.
Alpha cabin: Carpet section analysis (Low frequency)

2. Zone2
3.5
Taguchi loss function
3. Zone3
4. Zone4
7. Zone7
3
1. Zone1
2.5
2
1.5
6. Zone6
5. Zone5
1
0.5
0
1
Param eters (Descending)
Figure 9.8: Alpha Cabin - Carpet section analysis (Low freq) Taguchi loss function
179
Alpha cabin: Carpet section analysis (Mid frequency)
4. Zone4
3. Zone3
7. Zone7
1. Zone1
2. Zone2
2.5
5. Zone5
2
1.5
6. Zone6
0.5
0
1
Figure 9.9: Alpha Cabin - Carpet section analysis (Mid freq) Taguchi loss function
Alpha cabin: Carpet section analysis (High frequency)

2. Zone2
4. Zone4
7. Zone7
5. Zone5
3. Zone3
2.5
1. Zone1
2
6. Zone6
1.5
0.5
0
1
Figure 9.10: Alpha Cabin - Carpet section analysis (High freq) Taguchi loss function
5 Vehicle sound package Taguchi analysis

The Taguchi ON/OFF analysis on the vehicle sound package helps to identify the
order of significance of each noise treatment in reducing the noise level inside the vehicle
180
cabin, and also to identify the percentage of contribution in reducing the noise level, in
consideration of variable coupling effects.
Compared to the carpet section Taguchi analysis, where the loss functions were
normalized with respect to surface area, the sound package Taguchi analysis does not have
area normalization.
The in vehicle sound package is divided into 7 sections/categories as explained

below.
5.1 Sound package Taguchi sections
The following are the sound package sections or parameters (1-7), for which the
ON/OFF Taguchi analysis is conducted for both engine noise and tyre noise sources. Note
that a separate analysis is conducted for the study of window contribution in cabin noise
level.
1 2 3
Seats Headliner Door trim
4 5 6
Carpet Dash board & firewall Trunk trim
181
OR
7 7
Windows
Chasis pillars
Figure 9.11: Sound package sections/categories for ON/OFF Taguchi analysis
5.2 Taguchi analysis results
The ON/OFF Taguchi analysis test results for vehicle sound package are given in the
below tables. The variable and level settings for the vehicle sound package ON/OFF Taguchi
analysis are listed in Table 16. The Taguchi test run settings and output results are listed in
Table 17.
Table 16: Taguchi Parameter table
Input variables
Level Level
1 2
1. Seats ON OFF
2. Head liner ON OFF
3. Door trim ON OFF
4. Carpet ON OFF
5. Dash board + Fire-wall ON OFF
6. Trunk ON OFF
7. ABC pillars ON OFF
182
Engine noise
5. Dash board+FireWall
7. ABC pillars
2. Head liner
3. Door trim
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7 Low Mid High
Since power train or engine noise stays in low-mid frequency ranges, the Taguchi test
run results are averaged in the low (315Hz - 1kHz) and mid (1kHz - 4kHz) frequency ranges.
183
7. ABC pillars
2. Head liner
3. Door trim
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 31.75 31.62 32.96 31.70 31.94 32.98 31.84
Avg. level2 33.23 33.36 32.03 33.28 33.04 32.00 33.14
Diff. (effect) 1.47 1.75 0.93 1.58 1.10 0.97 1.30
Mid (1KHz-4KHz)
Avg. level1 25.73 25.76 28.26 26.34 26.48 28.12 26.31
Avg. level2 28.50 28.47 25.97 27.89 27.75 26.10 27.92
Diff. (effect) 2.77 2.71 2.29 1.54 1.27 2.02 1.61
Tyre noise
7. ABC pillars
2. Head liner
3. Door trim
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7 Low Mid High
184
7. ABC pillars
2. Head liner
3. Door trim
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 47.25 47.43 47.85 47.25 48.03 47.85 47.87
Avg. level2 48.65 48.47 48.05 48.65 47.87 48.05 48.03
Diff. (effect) 1.40 1.05 0.21 1.40 0.15 0.20 0.16
Mid (1KHz-4KHz)
Avg. level1 42.69 44.38 43.94 43.88 44.17 44.55 44.25
Avg. level2 45.86 44.17 44.61 44.67 44.37 44.00 44.30
Diff. (effect) 3.17 0.21 0.67 0.79 0.20 0.55 0.04
High (4KHz-8KHz)
Avg. level1 32.04 33.51 34.81 33.13 34.62 34.27 36.41
Avg. level2 36.57 35.10 33.80 35.48 33.99 34.34 32.20
Diff. (effect) 4.53 1.59 1.00 2.36 0.64 0.07 4.21
185
Engine noise
At low frequencies, the headliner and the carpet seem to be the dominating sound
package components. Meanwhile, in the mid frequency range of engine noise, the seats,
headliner and door trim follow the dominating order.
Sound Package analysis (Engine noise- Low Hz)
2. Head liner
4. Carpet
7. ABC pillars
1. Seats
1.8
1.6
3. Door trim
6. Trunk
1.4
1.2
0.8
0.6
0.4
0.2
0
1
Figure 9.12: Vehicle sound package analysis (Engine noise – Low freq) Taguchi loss function
Tyre noise
As expected the seats have the highest loss function in all the frequency ranges of
interest. Meanwhile the carpet and headliner are the next two treatments that dominate in the
low frequency range. Next to seats, at the mid frequency range, the carpet and the door trim
dominate, and the chassis pillars and carpet dominate at the high frequency ranges.
186
Sound Package analysis (Tyre noise - Mid Hz)
1. Seats
3.5
Taguchi loss function 3
2.5
4. Carpet
3. Door trim
6. Trunk
2. Head liner
1.5
7. ABC pillars
1
0.5
0
1
Figure 9.13: Vehicle sound package analysis (Tyre noise – Mid freq) Taguchi loss function
5.2.2 Taguchi analysis including window screens
A separate Taguchi analysis is conducted by including the vehicle window glasses, in

order to see the contribution of glass absorption and transmission loss. It is to be noted that,
similar to the other sound package section, the OFF case for a window is considered as sheet
metal (1mm thick) without noise control treatment.
5.2.2.1 Taguchi analysis results
The ON/OFF Taguchi analysis test results (engine noise and tyre / road noise) for
window contribution study are given in the below tables.
187
Engine noise
2. Head liner
3. Door trim
7. Windows
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 33.02 32.76 33.90 33.09 33.12 34.04 31.84
Avg. level2 34.29 34.55 33.41 34.22 34.19 33.27 35.47
Diff. (effect) 1.27 1.79 -0.48 1.13 1.06 -0.77 3.63
Mid (1KHz-4KHz)
Avg. level1 25.65 25.65 28.18 26.23 26.40 28.01 26.31
Avg. level2 28.38 28.39 25.85 27.81 27.64 26.03 27.73
Diff. (effect) 2.73 2.74 -2.33 1.58 1.24 -1.98 1.42
High (4KHz-8KHz)
Avg. level1 16.66 15.46 20.15 17.78 16.89 19.81 16.15
Avg. level2 19.78 20.99 16.30 18.67 19.56 16.64 20.30
Diff. (effect) 3.12 5.53 -3.85 0.89 2.67 -3.18 4.15
188
Tyre noise
2. Head liner
3. Door trim
7. Windows
4. Carpet
1. Seats
6. Trunk
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 48.62 48.85 49.07 48.69 49.26 49.14 47.87
Avg. level2 49.94 49.71 49.49 49.87 49.30 49.42 50.69
Diff. (effect) 1.32 0.86 0.43 1.17 0.03 0.28 2.82
Mid (1KHz-4KHz)
Avg. level1 41.65 43.90 43.18 43.31 43.32 44.26 44.25
Avg. level2 45.57 43.32 44.04 43.91 43.90 42.96 42.97
Diff. (effect) 3.93 -0.59 0.87 0.60 0.58 -1.30 -1.28
High (4KHz-8KHz)
Avg. level1 33.71 35.24 36.32 35.09 36.36 36.07 36.41
Avg. level2 38.37 36.84 35.77 36.99 35.72 36.02 35.67
Diff. (effect) 4.65 1.60 -0.55 1.91 -0.65 -0.05 -0.73
The Taguchi analysis results show that the windows play a very significant role in the
low frequency ranges for both the engine and tyre noises.
189
Sound Package & windows analysis - Engine noise (Low Hz)
7. Windows
4
3.5
3
2. Head liner
2.5
4. Carpet
1. Seats
2
6. Trunk
3. Door trim
1.5
0.5
0
1
Param e te rs (De s ce nding)
Figure 9.14: Vehicle sound package & window analysis (Engine noise – Low freq)
Sound Package & windows analysis - Tyre noise (Mid Hz)

1. Seats
3.5
2.5
7. Windows
6. Trunk
3. Door trim
4. Carpet
2
2. Head liner
1.5
0.5
0
1
Figure 9.15: Vehicle sound package & window analysis (Tyre noise – Mid freq)
190
5.3 Vehicle carpet section Taguchi analysis
Carpet sections
The carpet is divided into 7 approximately equal area zones for ON/OFF Taguchi
analysis. As shown below, the sections that come into the same zone have the same zone
number.
6
6
6
6
5
5
7 7
5
5
4
7 7 4
3
3
2
2
1
1 7 2
7 2
1
1
Front
Figure 9.16: Vehicle level carpet section ON/OFF Taguchi zones
Referring to Figure 9.10, the red carpet sections are the areas where normally mats
are placed on the carpet. These sections are carefully placed as separate Taguchi
zones/parameters, for analysing the contribution of such mat sections. Similar to the Alpha
191
Cabin Taguchi analysis, the carpet section wise loss function results are normalized to 1m2
as various carpet zones have different surface areas.
Compared to the component level Alpha Cabin Taguchi analysis, where only sound
absorption of the carpet is considered, in vehicle level carpet section analysis the Sound
Transmission Loss of the carpet also comes into picture. Further, the engine and tyre noise
sources are transmitted both as air-borne and structure borne through the vehicle components
and cavity subsystems.
5.3.1 Taguchi analysis results
The ON/OFF Taguchi analysis test results for vehicle carpet sections are given in the
below tables.
Engine noise

1. Zone1
2. Zone2
3. Zone3
4. Zone4
5. Zone5
6. Zone6
7. Zone7
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 33.48 33.14 33.53 33.09 33.23 33.50 33.17
Avg. level2 33.74 34.07 33.68 34.12 33.99 33.71 34.05
Diff. (effect) 0.26 0.93 0.15 1.03 0.76 0.21 0.88
Mid (1KHz-4KHz)
Avg. level1 30.37 29.85 30.33 29.76 30.05 30.32 30.02
Avg. level2 30.54 31.06 30.58 31.16 30.86 30.60 30.90
Diff. (effect) 0.17 1.21 0.25 1.40 0.81 0.29 0.88
High (4KHz-8KHz)
192
Avg. level1 21.43 20.70 21.38 20.86 21.11 21.37 20.99
Avg. level2 21.63 22.36 21.68 22.20 21.95 21.69 22.07
Diff. (effect) 0.20 1.67 0.31 1.33 0.83 0.31 1.08
Tyre noise
1. Zone1
2. Zone2
3. Zone3
4. Zone4
5. Zone5
6. Zone6
7. Zone7
1 2 3 4 5 6 7
Low (315Hz-1KHz)
Avg. level1 47.55 47.70 47.53 47.69 47.56 47.62 47.58
Avg. level2 47.94 47.78 47.96 47.80 47.92 47.86 47.91
Diff. (effect) 0.38 0.08 0.42 0.11 0.36 0.24 0.33
Mid (1KHz-4KHz)
Avg. level1 44.82 44.85 44.77 44.82 44.78 44.78 44.82
Avg. level2 45.04 45.01 45.09 45.04 45.08 45.08 45.04
Diff. (effect) 0.23 0.16 0.33 0.22 0.30 0.30 0.21
High (4KHz-8KHz)
Avg. level1 36.90 36.98 36.88 36.97 36.86 36.94 36.92
Avg. level2 37.22 37.13 37.23 37.15 37.26 37.17 37.19
Diff. (effect) 0.32 0.14 0.35 0.18 0.40 0.23 0.27
193
The results clearly indicate that in the case of engine noise, the front two carpet
sections nearer to the engine, i.e. zones 2 and 4, play the key role in noise reduction, which is
expected. Meanwhile in the case of tyre noise the front carpet sections nearer to the driver
ear, i.e. zones 1 and 3 play significantly in reducing the SPL at the driver ear.
Carpet section analysis - Engine noise (Low Hz)

2. Z one2
4
4. Z one4
3.5
3
5. Z one5
1. Z one1
2.5
7. Z one7
2 4
6. Z one6
3. Z one3
1.5
1
0.5
2
0
1
Parameters (Descending)
Figure 9.17: Carpet section (Vehicle level) analysis (Engine noise – Low freq)
Carpet section analysis - Tyre noise (Mid Hz)

1. Z one1
3. Z one3
2. Z one2
6. Z one6
1
4. Z one4
0.9
0.8
5. Z one5
0.7
7. Z one7
0.6
0.5
0.4
0.3
3
0.2
0.1
0
1
1
Parameters (Descending)
Figure 9.18: Carpet section (Vehicle level) analysis (Tyre noise – Mid freq)
194
6 CONCLUSIONS
• Simulation based Taguchi optimization can give a clear indication for further
optimization
• It is preferred that separate Taguchi loss function calculation be done for various
frequency ranges, like Low (315Hz-1kHz), Mid(1kHz-4kHz) and High(4kHz-8kHz)
frequency ranges
• In Concept 3 carpet layers the AFR and substrate layers have the largest air flow
resistivity and there fore they have the largest influence on the sound absorption
coefficients in the low frequency range (315 Hz – 1000 Hz). In the mid-high
frequencies the tufting & substrate have the largest influence on the sound absorption
coefficients (refer Table 3)
• In Concept 3 carpet layers, the foam de-coupler has the largest influence on the sound
transmission loss in the low and mid frequency ranges (refer Table 5)
• ON/OFF Taguchi analysis in Alpha Cabin shows that the front foot rest areas (zones
3 and 4) consistently dominate in the low, mid and high frequencies.
• ON/OFF Taguchi analysis for Vehicle sound package shows that for engine noise
(low frequencies 315Hz to 1 kHz), the headliner and then carpet seem to be the
dominating components.
• ON/OFF Taguchi analysis for tyre noise shows that the seats have the highest loss
function in all the frequency ranges of interest, followed by carpet treatment, at the
mid frequencies (1kHz - 4kHz)
• The Taguchi analysis results show that the windows play a very significant role in the
low frequency ranges, for both the engine and tyre noises.
• The results for carpet section Taguchi analysis indicate that in the case of engine
noise, the front two carpet sections nearer to the engine, i.e. zones 2 and 4, play the
key role in noise reduction. Meanwhile, in the case of tyre noise, the front carpet
sections nearer to the driver ear, i.e. zones 1 and 3, play significantly in reducing the
SPL at the driver ear.
195
CHAPTER 10
CONCLUSIONS AND RECOMMENDATIONS FOR FURTHER

RESEARCH
1 CONCLUSIONS
1.1 This thesis

Initial benchmark analyses and target settings for vehicle cabin noise were conducted
to get a clear understanding of the design requirements and constraints. The following
observations are worthwhile to be noted from the vehicle tests. Low frequency power train
noise has higher sound pressure level (SPL) distribution in the front part of the carpet (front
foot well area), which may need more sound barriers. Meanwhile, high frequency tyre-road
interaction noise has higher sound pressure level (SPL) distribution in the rear part of the
carpet (rear wheel arch and the boot areas), which may need more sound absorption
measures.
Theoretical analysis of forced vibration of the carpet system is studied by developing

mathematical models based on the simple harmonic response and sound radiation directivity
of the floor panel and carpet system. The Simple Harmonic Motion oscillator model of
analyses of the current carpet design system shows that there is a resonance around the
frequency of 323Hz which is in the frequency range of engine or power train noise. This is
verified from the sound Transmission Loss measurements of the carpet variants, using the
impedance tube method. Further, the theoretical model presented for panel radiation
directivity prediction can be used to successfully predict the sound level radiated at a
particular angle to the normal of a panel or opening, relative to the sound level radiated in the
direction of the normal.
196
With a basic theoretical understanding of the current carpet design model, four new
concept carpet designs are proposed and evaluated in laboratory for acoustic performance. A
concept design for both good cabin side sound absorption coefficients and Sound
Transmission Loss, identified as Concept 3 design, is developed.
Material level simulation validation is conducted for various acoustic samples and
measured material sound absorption coefficients have a similar trend to simulated ones and
have a variation of only about 10-15% for various samples. The simulation results do not
include the low frequency resonance peaks, i.e. below 300Hz, observed in normal incidence
absorption measurements. Further, since SEA tools use random incidence Sound absorption
Coefficient (SAC) values for simulation inputs, Paris method [53] of converting normal
incidence SAC to random incidence is used and validated.
The proposed carpet designs are further simulated and evaluated at component and
vehicle levels. In the Alpha Cabin simulation (component level) tests, compared with the
current production carpet (MITSUBISHI 380 carpet), Concept 3 carpet consistently reduces
about 4.2 dB SPL above 2kHz and reduces up to average of 1 dB below 1 kHz. Concept 3
special mats reduce the noise level by 3 dB above 2 kHz and up to an average of 1 dB below
1kHz, in comparison with the current production carpet (MITSUBISHI 380 carpet).
The vehicle virtual SEA model was evaluated and validated using acoustic transfer
function method and well matching results were achieved for engine and tyre noises, except
for the coincidence frequency ranges of the backing sheet metal floor. Concept 3 carpet
evaluation at the vehicle level shows the reduction of the in-cabin noise level further by up to
3-5 dB, compared to the current production carpet design, for power train noise. The
Concept 3 carpet simulation at the vehicle level shows the reduction of the in-cabin noise
level further by 2 dB at high frequencies, compared to the current production carpet design,
for the tyre-road noise. The Concept 3 mat padded with the current production carpet reduces
noise level by about 1dB SPL at high frequencies, compared with the carpet only case for the
power train and tyre-road noise. It is illustrated that the proposed the concept 3 carpet has
nearly reached the noise reduction limit.
As the component and vehicle simulation results are giving promising results, the
proposed Concept 3 mat design is proto-typed for on-road evaluation. A proof of concept
mat based on the proposed Concept 3 design (C3-Mat_Prototype) was factory moulded,
albeit slight modifications from the original specifications due to material, time and
197
technical constraints. In the impedance tube SAC test, the C3-Mat_Prototype performed
better than the standard mat in the low to mid frequency ranges (up to 3000Hz). Meanwhile,
RMIT Alpha Cabin SAC test showed that the C3-Mat_Prototype outperformed the standard
mat by an average of 14.2 % for in-cabin side SAC, with the highest improvement of about
40 % of SAC at 800 Hz. On-road vehicle tests show that the C3-Mat_Prototype has better
sound absorption at the low to mid frequency range than Standard mat. On road tests using
both B&K Pulse and binaural head acoustics were conducted and the data indicates a
reduction of in-cabin total SPL by the prototype mat is about 2.1 dB(A) during stationary
tests and 0.625 dB(A) during constant driving tests. It is noted that these improvements are
measured for the proof of concept factory moulded mat, which points to the fact that a
Concept 3 carpet system as a whole can give promising results, as predicted by the
simulation model results.
Sensitivity and optimization of the proposed carpet design (Concept 3) are conducted
by the Taguchi method. Simulation based Taguchi optimization gives a clear indication for
further optimization directions.
In Concept 3 carpet layers the AFR and substrate layers have the largest air flow
resistivity and therefore they have the largest influence on the sound absorption coefficients
in the low frequency range (315 Hz – 1000 Hz). Meanwhile, the foam de-coupler has the
largest influence on the sound transmission loss in the low and mid frequency ranges.
Based on the sensitivity analysis studies and prototype evaluation of the Concept 3
carpet design, acoustically improved carpet designs are being proposed for further research,
in the following sections.
1.2 Research questions

The thesis has addressed the following research questions:
• What is the role of vehicle floor carpet in eliminating engine and tyre noise and how
good is the current design in achieving it?
• Is virtual modelling and validation an effective tool for designing and evaluating a
vehicle carpet acoustically at vehicle level?
acoustically improved vehicle carpets and where are we now?
198
• Are there simple and efficient methods for in-situ measuring the acoustic
performance of carpet design, non-destructively?
carpets?
The conclusions drawn regarding the above questions are summarised below.
1.2.1 What is the role of vehicle floor carpet in eliminating engine and
tyre noise and how good is the current design in achieving it?
ON/OFF Taguchi sensitivity analysis for Vehicle sound package shows that for
engine noise (low frequencies 315Hz to 1 kHz), the headliner and then carpet seem to be the
dominating components. Meanwhile, analysis for tyre noise shows that the seats have the
highest loss function in all the frequency ranges of interest, followed by carpet treatment, at
the mid frequencies (1kHz - 4kHz).
The average in-cabin sound absorption is relatively less, of only about 30% for the
tufting layer. The measured transmission loss ranges from 10dB to 35 dB. The underlying
de-coupler layer absorbs about 80% of the sound incident on it, but mainly in the frequency
above 750Hz.
Referring to the vehicle simulations results, the no carpet noise level is 4-6 dB higher
compared to the current production MITSUBISHI 380 carpet design, for engine noise. The
tyre noise reduction improvement from carpets is not significant, and the difference in the
driver’s ear SPL for the no carpet case and the ideal carpet is just 2-4dB.
1.2.2 Is virtual modelling and validation an effective tool for designing

and evaluating a vehicle carpet acoustically at vehicle level?
It is notable that the vehicle level simulation results of concept 3 mat on M380
standard carpet, for engine noise, showed similar possible improvement of around 1dB,
especially at the higher frequency ranges.
1.2.3 What could be the theoretical limits of noise reduction in the

vehicle cabin by acoustically improved vehicle carpets and where
are we now?
It is clear that the Concept 3 design reduces the engine noise level inside the vehicle
cabin by about 7-10 dB compared to the no carpet case. It is worth noting that the Concept 3
results are very close to the ideal carpet results.
199
The ideal carpet is the one when floor noise control treatment having 100% sound
absorption at in-cabin side and having infinite (450dB) transmission loss. The SPL results of
this carpet help to know the maximum possible noise reduction potential offered by the floor
trims.
From the simulation results for tyre noise reduction by vehicle floor carpets, the noise
reduction improvement from carpets is not significant, and the difference in the driver’s ear
SPL for the no carpet case and the ideal carpet is just 2-4dB.
1.2.4 Are there simple and efficient methods for in-situ measuring the
acoustic performance of carpet design, non-destructively?
The existing principles and devices for measuring the acoustic properties of materials
are reviewed in detail. The proposed 45o SAC measurement device provides non-destructive
SAC measurement for sound package samples. A working prototype model of the proposed
measurement device is developed and calibrated.
1.2.5 What are the future possibilities in introducing tuned acoustic

layers for vehicle carpets?
A dedicated study is being carried out on tuned absorbers and double panel noise
isolation layers in the following sections, as recommendations for further research.
2 RECOMMENDATIONS FOR FURTHER RESEARCH
This document outlines the continuing developments and possibilities of improving

the vehicle carpet design, in an acoustic point of view, by incorporating existing and proven
acoustic techniques or by introducing novel design concepts. This design process is not
restricted to the normal production constraints, which includes cost, manufacturability,
material availability etc. This helps to keep acoustic performance as the key focus, while
raising the concept design platform to a higher level. It is hoped that all of the results can be
kept aside for future production proposals or as applicable to specific sections of the current
carpet system.
200
2.1 Scope for an improved design
Improving or trying to improve a design ahead of an existing proven carpet design

should be driven by a reason-based analysis of the current design and its issues. A few of the
key reasons behind working on the current design are,
- Sound isolation using heavy layer is a simple acoustic design.
- Increasing area density (GSM) of the heavy layer can improve sound isolation. But
the weight penalty is high for the sound isolation obtained.
- More layers helps more energy dissipation
- Utilizing air-gap or double panel techniques have almost zero weight over-heads
- Application of existing aerospace noise reduction techniques
- There are various other novel acoustic structures that can be tried in a carpet design.
All the above reasons should justify going forward with a concept design for
continuing research.
2.2 Concept 4 carpet design constraints and targets
The regular constraints for vehicle carpet design such as manufacturing cost,
aesthetics, durability, strength etc are not kept as constraints for coming up with an
acoustically superior vehicle carpet design. But at the same time, the weight and total
thickness are expected to be maintained the same as the current production carpet.
Total carpet thickness constraint - 40mm
Total carpet weight (GSM) constraint - 4.6 kgsm
Further, a reasonable acoustical improvement is considered, if it shows at least a 3dB

improvement compared to the current production carpet transmission loss, especially the low
frequencies (25Hz – 4kHz).
201
3 DOUBLE PANEL AND AIR-GAP THEORY
There are various acoustic techniques applied in building and aerospace industry,
which are yet not tapped in vehicle acoustics, which could be due to various design
constraints, implementation practicalities or mere process issues. Double panelling is a
popular technique used in building acoustics, which involves placing two sound barrier
panels close to each other with a fixed air gap in between. If a single barrier layer has a
transmission loss of X dB, then this double layer system with air gap is expected to reduce
sound transmission by 2X dB, compared to the transmission loss by placing the two barriers
on each other which will be only X + 10*log 22 = X + 6dB. So, as long as X > 6dB, then it
becomes practical to introduce the air gap between the barrier layers.
N dB X dB X dB N - 2X dB
Air gap
Figure 10.1 – Double panel sound barrier
In the case of carpet applications, the heavy layer acts as the heavy and thin sound
isolation layer. Since vehicle floor carpets undergo step-on pressures from passengers and
the edges are not fixed to the walls like in building acoustics, an air-gap structure is required
to hold the two sound barrier layers at a fixed favourable distance. Honey-combs are popular
structures in industry, normally used to increase the structural strength while keeping the
weight low. So, honey-comb structures made of light materials like polycarbonate, polyester
or polypropylene etc can be used for the same purpose.
202
In the experimental concept 4 case, we use a poly-carbonate honey-comb which is

quite rigid and also has a comparable weight with foam. Below is the proposed concept 4
diagram.
Tufting layer
Concept4 AFR layer
Absorber
Cabin side noise Thin heavy layer 1
Air gap structure
40mm
Thin heavy layer 2

Noise path
(from engine) De-coupler layer
In concept 4, instead of the single heavy layer in concept 3 which is up to 5kgsm, we can
split the heavy layer into two separate lighter layers, i.e. two heavy layers of 1.7kgsm each.
4 RESULT ANALYSIS
The below graph plots show the impedance tube test results for Transmission Loss,
which also includes the transmission loss of the air-gap plus heavy layer alone (dashed line),
without the foam or tufting layers. The layer nomenclatures and abbreviations are briefly
explained in Table 1. For example, 2HL&HCg10 means two heavy layers (2HL) separated
by the grey (g) honey comb (HC) structure of 10mm.
From the below plots in Figure 10.3, the best four layered combinations are
highlighted for detail analysis. It is clear from the graph that there is a steadily increasing
transmission loss from 500Hz to approximately 1.5 kHz, at a rate of 20dB per octave, with
two local minima at about 400Hz and 2.5 kHz. Note that even the current production carpet
having 5kgsm heavy layer is quite low in Transmission Loss through out the frequency
ranges and even up to 50dB less at mid frequencies. This clearly illustrates that Concept 4
203
carpet design using a honey comb air-gap structure is quite superior in acoustic performance
of Transmission Loss.
Transmission Loss: Concept 4 (best set)
90
80
Transmission Loss (dB)
70
60
50
40
30
20
10
0
is the table
1100
Below
explaining the 2100
100
material configurations
3100
and the nomenclatures
4100
or
code names used. Frequency Hz
Ktuft+S3D+SB+2HL&HCg10+SB Ktuft+SE+SA+2HL&HCg10+SA Ktuft+SF+SA+2HL&HCg10+SA

Ktuft+SE+SC+2HL&HCg10+SC 2HL1p7&HCg10 Current production 5kgsm HL
Figure 10.3 – Transmission Loss measurements of Concept 4 design

models
Table 1 – nomenclature of layered material combinations
Layer Code Material & Specs Construction
1. Tufting Ktuft Type K, brown tufting, 320gsm N/A
2. Air Flow S3D 1237gsm AFR layer N/A

Resistant layers S3E 1146gsm AFR layer
(AFR)
3. Absorber SA_scrim Visco-elastic foam with scrim,

Scrim
20mm
SB_scrim Scrim Absorber
Visco-elastic foam with scrim, Scrim

SC_scrim
20mm SA & SC SB & SD
SD_scrim
Fibre with scrim, 20mm
Scrim = ks
Fibre with scrim, 20mm
Scrim = sx
204
VE10_ks 100 gsm scrim
PT1f10_scrim 44 gsm scrim
Visco-elastic foam + scrim,

10mm
PT1foam+scrim, 10mm
4. Heavy layer SL 2.5 mm Heavy Layer HL
(HL)& decoupler (1700gsm) +1.5 kgsm close Foam

cell polyester foam (Nfoam),
20mm
Nf20
Nfoam, 1.5 kgsm, close cell
polyester foam, 20mm
5. Air gap 2HL&HCg10 2.5 mm Heavy Layer HL
(1700gsm) + 10 mm HC
Honeycomb structure (grey) + HL
2.5 mm Heavy Layer

(1700gsm)
6. Heavy layers HL x kgsm X kgsm Heavy layer

HL
It is to be noted that it should be verified whether the cabin side sound absorption is
affected in any form by the air-gap introduced between heavy layers. So absorption tests
were conducted on the concept 4 designs to verify this. The below graph shows that
absorption coefficient above 1kHz has gone down by about 10% compared to the concept 3
design with single heavy layer. As the tests were done by just replacing the Concept 3 heavy
layer with new Concept 4 heavy layer, it can be expected that some sound is normally
transmitted through the heavy layers and absorbed by the de-coupler foam behind, i.e. in the
reverse direction of engine noise path into the cabin, and since the dual heavy layer structure
of Concept 4 gives higher transmission loss, this high TL makes sound energy be absorbed
by the layers above the dual heavy layer only; the de-coupler play little role in absorption.
205
SAC of C4 Model variants
1.0
0.9
0.8
Sound Absorption Coefficient
Concept 3
0.7 10 %
Concept 4
0.6
0.5
0.4
0.3 Current carpet design
0.2
0.1
0.0
100 1100 2100 3100 4100 5100 6100 7100
Frequency (Hz)
Current production - M-380_carpet Ktuft+S3D+Nf20+2HL&HCg10+Nf20

Ktuft+S3D+SA+2HL&HCg10+SA Ktuft+S3D+SC+2HL&HCg10+SC
Concept 3 - Ktuff+S3D+SC_ks+SL_top
Figure 10.4 – Sound Absorption Coefficient measurements of Concept 4 design models
5 EVALUATING AGAINST CONSTRAINTS: Foam VS Honey comb
The design constraints we have in hand for concept 3 design are weight and total
carpet thickness, against which we could compare concept 4 design weight penalty. So a
brief study is done in comparing between concept 3 and concept 4, where 10mm of absorber
or de-coupler foams in concept 3 have been replaced by the honeycomb structure 10mm
thick, as shown below in Figure 10.5.
206
It is to be noted that with regards to weight constraints both 10mm honeycomb

structure and 10mm foam showed about the same gsm or weight. So the honey-comb is
expected not to introduce any weight penalty, other than manufacturability. The heavy layers
used are 1.7kgsm of area density.
10mm
HL combined
Concept 4
Concept 3
10mm
40mm
10mm
Figure 10.5 – Concept 3 absorber/de-coupler layer replaced by honey-comb

Further, transmission loss measurements were conducted to have direct comparison
of introducing the honey-comb into concept 4 without any constraint overheads, and the
results are as shown in below figure
It can be seen that concept 4 gives about 20dB of improvement in TL, compared to
concept 3, at mid frequency ranges. It is notable that this increment in acoustic performance
is without any weight or thickness increase introduced by the honeycomb layer of 10mm.
Concep4 and concept 3 are close in transmission loss across the high frequency range. Thus
this quantifies that the new concept 4 is superior to concept3 in terms of acoustic
performance (TL) for the given weight and thickness constraints, and hence is a design for
future acoustically improved vehicle carpets.
207
Transmission Loss: HoneyComb VS Foam
80.0 Concept 4
70.0
Transmission Loss (dB)
60.0
Concept 3
50.0
40.0
Current production
30.0
20.0
10.0
500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency Hz
Ktuft+S3D+VE10_ks+2HL&HCg10+SB_scrim (Concept 4)
Ktuft+S3D+VE10_ks+VE10_ks+HL1p7+HL1p7+SB_scrim (Concept 3)
Ktuft+S3D+VE10_ks+HL1p7+HL1p7+SB_scrim+VE10_ks (Concept 3)
Current production: N (HL 5kgsm)
Current production: L (HL 1.7kgsm)
Figure 10.6 – Comparison of concept 4 and concept 3, without new weight or thickness
overheads
6 SPECIAL TUNED RESONANT ABSORBER LAYERS
Acoustic absorber structures have been widely used in applications like building
acoustics, where specific frequencies are the target of absorption, like the room modal
frequencies. Literature for resonant absorber structures point to mainly two successfully
applied resonant absorbers, the Helmholtz resonator panels and the membrane (bass-trap)
absorbers. These special structures and their performance at low frequency ranges is worth
investigating.
The Helmholtz absorption principle is well known for its results and hence the
realization of its application to a vehicle floor trim depends on how well it can be included
into the current carpet layers, meanwhile adhering to the weight and cost constraints.
A possible way of introducing a Helmholtz resonator structure tuned for

compensating for the weak transmission loss of the heavy layer material, i.e. measured to be
around 323 Hz, is given below.
208
Tufting layer
Cabin side noise AFR layer
Absorber
Thin heavy layer 1
Thin heavy layer 2
40mm
Air gap structure

Noise path
Air cavity Honeycomb

depth 5.5mm Absorber
Panel 2.5mm
Hole radius Hole spacing

80mm
Figure 10.7 – Tuned Helmholtz resonant absorber layer included in the Honey Comb
dual panel structure
Example calculation of the Helmholtz resonator parameters :
Frequency of resonant absorption f = c/2π * √ (π*a2/(t*D2*d)),
Where,
C = Velocity of sound (343m/s)
a = hole radius
t = Panel thickness
D = Hole spacing
d = cavity depth
209
Frequency tuned for – 250Hz
Panel thickness 2.5mm
Cavity depth 5.5mm
Hole spacing 80mm
Hole radius 0.8mm
Absorber thickness 2mm
Total thickness 10mm
From the above calculation of the Helmholtz resonator layer parameters for a
frequency range around 250 Hz, it is clear that the dimension like panel thickness (2.5mm),
cavity depth (5.5mm) and absorber thickness (2mm) all can fit into the dimensions of the
current vehicle carpet structure.
Construction of such a structure is quite involved and sophisticated equipment is

required to get the matching hole radius. Hence, perforated material samples were out-
sourced from third party building acoustic material providers. Such a sample (V white micro
25mm), open cell foam with a perforated synthetic facing, was tested to verify it. The
specifications and details of the sample is not presented due to IP concerns.
Absorbers - Sorber series (RMIT - Mar 08) 25mm
1.0
0.9 Foam with perforated facing
Normal Absorption Coefficient
0.8 (Helmholtz resonator layer)

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
100 1000 10000
365 Hz Frequency (Hz)
Sorber_S8_5 (Vwhite micro)
Figure 10.8 – SAC result of perforated facing foam (25mm) Helmholtz resonator
210
Similarly, the possibility of introducing resonant membrane absorber layers was

investigated briefly. The below graph illustrates the improved low frequency absorption by a
25mm thick open cell foam having a thin corrugated foil membrane facing.
Absorbers - Sorber series (RMIT - Mar 08) 25mm
1.0
0.9 Foam with foil membrane
Normal Absorption Coefficient
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
100 1000 10000
546 Hz Frequency (Hz)
Sorber_S8_3 (25 ALR)
Figure 10.9 – SAC results of corrugated foil facing foam resonant membrane absorber
The SAC results for the special resonant absorbers indicate that introducing these
layers into the vehicle carpet system or even other noise prone areas of the vehicle, can
extend the noise reduction capability of the noise treatment to specific low frequency ranges.
Further research in the direction of making these structures manufacturing friendly and cost
effective can definitely give promising results.
211
Summarizing the recommendations for further improvement of vehicle carpet designs,
• Introducing air-gap in between the heavy layers improves the transmission loss by up
to 20dB in the frequency range of 1kHz – 2kHz, and up to 10dB in the frequency
range of 500Hz – 1kHz, and so is the best for these frequency ranges.
• The introduction of honey-comb structure as an air-gap structure doesn’t add any

over-head in terms of weight or thickness, compared to foam.
• Replacing a part of absorber or de-coupler layer by a honey-comb structure doesn’t

affect much the sound absorption.
• Special tuned absorber layers like the perforated facing foam Helmholtz resonator
and the corrugated foil faced foam membrane absorbers are excellent in extending the
noise reduction into specific low frequency ranges of interest.
212
Masters By Research
REFERENCES
[1] Name, Initial and name, initial (year), The title of document. Journal, serial page section
[2] De Bree, H.-E., (1996), The microflown, a noval device measuring acoustical flows,
sensors and actuators. A. Physical, SNA054/1-3:552-557.
[3] Duval, A. and al.(2005), Vehicle acoustic synthesis method 2nd generation: an effective
hybrid simulation tool to implement acoustic lightweight strategies. Journee
SFA/Renault/SNCF/-Guyancourt.
[4] Lannoye, R. and al. (2004), A practical device to determine the reflection coefficient of
acoustic materials in-situ based on a microflown and microphone sensor, I.S.M.A. -
Leuven.
[5] Allard, J.-F. (1993), Propagation of Sound in Porous Media, Applied Science.
[6] Kampl, R. and Leitner, J. (1999, Sept. 15-17), 38th International Man-made Fibres
Congress, Dombim/Austria.
[7] Buck, K. A. and Wellman, R. A. (1997). ICI Polyurethanes, Importance of Durability
on Acoustical Performance for Carpet Underlay, SAE-970148.
[8] Hilyard, N.C. and Cunningham, A. (1991), Design Optimisation of MDI-Based
Polyurethane Foam Backed Automotive Carpet Systems, SAE 911087.
[9] Balte, R., Choe, J. and Cherng, J.G. (2003), Investigation and Benchmarking for Vehicle
Floor Coverings, SAE 2003-1-1575.
[10] Nick, A., Becker, U. and Thomas, W., (2002, July), Improved Behaviour of Interior
Parts of Renewable Resources in the Automotive Industry, Journal of Polymers and the
Environment, Vol 10, No.3.
[11] Skinner, C., Peters, J., Vandenbroeck, J. (2006, Mar), Acoustic Absorbers: A third way
for the management of sound in automobiles, Paper for UTECH Europe.
[12] Sureshkumar, S. and Raveendra, S.T. (2003). Modelling Noise Control Materials, SAE
2003-01-1580.
[13] Duval, A., Rondean, J.-F., Bishoff, L. Dejaeger, L. (2006, 30th Nov), In Situ Impedance
and Absorption Coefficient Measurement Compared to Poro-elastic Simulation in Free,
Diffuse or Semi-statistical Fields Using Micro-flown p-u Probes, Journee
213
Masters By Research
SFA/Renault/SNCF.
[14] Trevor, J. Cox and Peter, D’Antonio, Acoustic Absorbers and Diffusers - Theory,
design and application, Taylor & Francis Group.
[15] Veen, J. R. and Saha, P., (2003), Feasibility of a Standardized Test Procedure for
Random Incidence Sound Absorption Tests Using a Small Size Reverberation Room,
SAE 2003-01-1572.
[16] Caprioli, D., Bertolini, A. and Gaudino, C. (2005), Improved NVH Performance Via
Genetic Optimization of Damping and Shape of Vehicle Panels, SAE 2005-1-2329.
[17] Gansen, P. and McCullough, D. (1990), Viscoelastic MDI Based Polyurethane Foam
for Sound and Vibration Dampening in Automobiles, SAE 900093.
[18] MacFarland, D. R. (1994), Cost Effective Moulded Polyurethane Foam for Automotive
Acoustical Carpet Underlay and Dash Insulators, SAE 940700.
[19] Pendergast, P. L. (1989), General Motor Corp., Use of Taguchi Techniques as a Process
Development Tool for Carpet-Covered Door Panel Production, SAE-890238.
[20] Dempsey, M. P. and Hurley, M. F. (1991), Mobay Corp., Advances in Urethane Foam
Composites for Interior Trim, SAE 910520.
[21] Bohm, R. (1989). Foam Backed Automotive Carpets in Europe, SAE 890237.
[22] John, T. Long (2003), SEA for Design: A Case Study, 2003-01-1565
[23] Chadwyck, T. Musser (2003), Cambridge Collaborative, Inc., Sound Package
Performance, Weight, and Cost Optimization Using SEA Analysis, 2003-01-1571
[24] Development of Quiet Sound Package Treatments for Class 8 Trucks, SAE 2001-01-
1541
[25] Case Study on Airborne Road Noise Reduction of a Passenger Vehicle, SAE 2003-01-
1407
[26] Tuning of Multi-layered Acoustic Systems, SAE 2003-01-1437
[27] Object- Oriented SEA Modelling, SAE 2003-1-1551
[28] Acoustical Optimization of Perforated Laminate Material and its Application to
Vehicles, SAE 2003-01-1567
[29] Bruel, PV and Mattia, G. M. (1998), Bruel Acoustics Technical Review 97-01. (web-
link www.bruel-ac.com/tr/tr9701.html)
[30] Garai, M. (1993), Measurement of the Sound-Absorption Coefficient In Situ: The
Reflection Method Using Periodic Pseudorandom Sequences of Maximum Length.
214
Masters By Research
Applied Acoustics Vol. 39.

[31] User Manual -. B&K Impedance/Transmission Loss Measurement Tubes-Type 4206.
[32] De Bree, H-E, Lanoye, R ,de Cock, S and Van Heck, J. (2005), In situ, broad band
method to determine the normal and oblique reflection coefficient of acoustic materials.
SAE International 2005-01-2443.
[33] Lanoye, R, Vermeir, G, Lauriks, W, Kruse, R and Mellert, V. (2006), Measuring the
free field acoustic impedance and absorption coefficient of sound absorbing materials
with a combined particle velocity-pressure sensor. J. Acoustical Society of America
119(5): 2826-2831
[34] Lanoye, R, de Bree, H-E, Lauriks, W & Vermeir, G. (2004), A practical device to
determine the reflection coefficient of acoustic materials in situ based on a Microflown
and microphone sensor. (web-link : www.microflown.com/data/ISMA_2004_10177 )
[35] Garai, M, Berengier, M, Guidorzi, P and L’Hermite, PH. (1998, 4th – 7th Oct),
Procedure for measuring the sound absorption of road surfaces in situ. Proc. Euro
Noise’98 Conference. Munchen.
[36] Volker, M and Nocke (2007), C. Applications of in-situ measurement techniques of
absorption.(web-link: www.akustikbuero.info)
[37] Bruel & Kjaer (1998), Application note: Measurements in building acoustics. (web-
link: www.bk.de)
[38] Blake, W.K. (1986), Mechanics of Flow-Induced Sound and Vibration: Vol. I.
Academic Press.
[39] Fisher, K. (2004), Broadband acoustic impedance tube comparison of two microphone
broadband method with conventional discrete frequency traversing technique and
development of broadband analysis software. MS Thesis. Applied Physics. RMIT
University.
[40] Schultz, T, Cattafesta, LN and Sheplak, M. (2006), Comparison of the two-microphone
method and a modal decomposition method for acoustic impedance testing. ©
University of Florida, Published by American Institute of Aeronautics and Astronautics.
[41] Cooke, M. (2001), Frequency range of the acoustic impedance tube. MS Thesis,
University of Queensland.
[42] AS/NZS 1935.1:1998 (ISO 10534-1:1996), Acoustics – Determination of sound
absorption coefficient and impedance in impedance tubes. Part 1: Method using
215
Masters By Research
standing wave ratio.

[43] Internet resource web-link: http://www.phys.unsw.edu.au/music/.
[44] Rindel, J.H. (1975), Transmission of traffic noise through windows – Influence of
incident angle on sound insulation in theory and experiment. Report No. 9, The
Acoustics Laboratory, Technical University of Denmark, Lyngby, Denmark.
[45] Davy, J.L. (2004, 3-5 November), The radiation efficiency of finite size flat panels.
Acoustics 2004, Transportation Noise and Vibration – The New Millennium,
Proceedings of the Annual Conference of the Australian Acoustical Society, Gold
Coast, Australia, Publisher Australian Acoustical Society, Castlemaine, Victoria,
Australia, ISBN 909882-21-5 pp 555-560.
[46] Cremer, L. and Heckl, M. (1973), Structure-Borne Sound, Springer-Verlag Company,

New York.
[47] Davy, J.L. (2007, 28-31 August), A model for predicting diffraction on a finite flat
surface as a function of angle of incidence and surface size. Inter-noise 2007, Istanbul,
Turkey.
[48] Stead, M. (2001), Sound Reduction for Reverberant, Direct and Diffracted Sound
through Single Isotropic Glass Panels of Finite Size, Master of Engineering Science
Thesis, Department of Mechanical Engineering, Monash University, Melbourne,
Australia.
[49] Sutton, M. (1990), Noise Directivity of Exhaust Stacks. Final year thesis for the
Honours Degree of Bachelor of Engineering, University of Adelaide, Adelaide,
Australia.
[50] Bolton, J. S. et al. (1996), Sound Transmission Through Multi-Panel Structures Lined
with Elastic Porous Materials. Journal of Sound and Vibration 191(3), pp. 317- 347.
[51] Bolton, J. S. and Kang, Y. J (1997)., Elastic Porous Materials for Sound Absorption
and Transmission Control, Proceedings, SAE Noise and Vibration Conference, 971878,
pp. 77-91.
[52] Lawrence, E. Kinsler, Austin, R. Frey, Alan, B. Coppens and James, V. Sanders
(2000), Fundamentals of Acoustics, Fourth edition - John Wiley & Sons.
[53] Delany M. E., Bazley E. N. (1970). Acoustical properties of fibrous absorbent

materials. Appl. Acoust. 3, 105-116.
216
Masters By Research
[54] AutoSEA2 Foam 2005, User’s guide. Theory and Q/A.
[55] Practical Designed Experiments, Alan Long Quality Pty. Ltd.,1999.

[56] HEAD Acoustics, Application Note: Binaural Measurement, analysis and playback,
(web-link: Application-notes@HEAD-acoustics.de).
[57] AS ISO 354-2006 “Australian Standards: Acoustics – Measurement of sound

absorption in a reverberation room”
[58] De Bree, H-E, Lanoye, R ,de Cock, S & van Heck, J. (2005), In situ, broad band
method to determine the normal and oblique reflection coefficient of acoustic materials.
SEA International 2005-01-2443.
[59] Lanoye, R, Vermeir, G, Lauriks, W, Kruse, R and Mellert, V. (2006), Measuring the
free field acoustic impedance and absorption coefficient of sound absorbing materials
with a combined particle velocity-pressure sensor. J. Acoustical Society of America
119(5): 2826-2831.
[60] Lanoye, R, de Bree, H-E, Lauriks, W and Vermeir, G (2004), A practical device to
determine the reflection coefficient of acoustic materials in situ based on a Microflown
and microphone sensor. (web-link:www.microflown.com/data/ISMA_2004_10177 )
[61] Genichi Taguchi (1993), Taguchi on Robust Technology Development, ASME Press,
New York.
[62] Dr. Genichi Taguchi and Yuin Wu (1989), Taguchi methods: Case Studies from the
U.S. and Europe, Japaneese Standards Association, ASI Press.
[63] Edsil, Dilla and Xu, Wang (2007), 45o SAC device concept paper (Report C1-07/M5D),
AutoCRC Carpet Acoustics Project Report.
[64] Cordioli, J. A. andM. T. Jr., et al. (2004). Applications of the Statistical Energy
Analysis to Vibro-Acoustic Modeling of Vehicles, SAE.
[65] Paris, E. T. (1927), Proc. Roy. Soc. 115, 407.
[66] Albert, London (1950), The Determination of Reverberant Sound Absorption

Coefficients from Acoustic Impedance Measurements, JASA vol. 22, no.2.
[67] Dr. Stuart, Lucas (2007), CSIRO Material Test Results, Samples B-K, Project Report.
[68] Chris, Murphy and Ross (2007), Final Year Thesis, SAMME RMIT Melbourne.
217

John

Uploaded by

Copyright:

Available Formats

John

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

John

Uploaded by

Copyright:

Available Formats

DESIGN & ANALYSIS OF ACOUSTICALLY

IMPROVED VEHICLE FLOOR CARPETS

A thesis submitted in fulfillment of the requirements for

School of Aerospace, Mechanical and Manufacturing Engineering

12th December 2008

1 GENERAL DESCRIPTION ........................................................................................... 1

2 Benchmark analysis and target setting for vehicle cabin

4 Design concepts for reducing noise from floor panel

5 Defining the design process and acoustic tools

6 Material simulation, validation and theory

7 Virtual modeling and validation, at component and

8 Proto-type validation and evaluation of design concept

9 Design optimization (Taguchi method based), at

10 Conclusions and recommendations for further research

SAC Sound Absorption Coefficient

The following are main research outcomes of the project.

• Material database, i.e. measured acoustic parameters for mathematical modelling,

• In-situ vehicle on-road validation test methods for carpet designs

• Optimization methods for further improving the design

Theoretical analysis of forced vibration of the carpet system is studied by developing

Meanwhile, in the case of an acoustical system which has a combination of various

Figure 1.1: Acoustic and thermal package of a typical car

2 SCOPE AND OBJECTIVE

This project contributes significantly to existing technology aiming at dramatically

Improved functionality and performance of an automotive carpet system with the

3.1 Carpet Material and Acoustics

A systematic benchmarking study was performed by Balte, et al., [8] to investigate

Geometric and physical data of selected benchmarking carpets consisted of

The use of renewable materials instead of manmade fibres as reinforcement in

Comparing standard material to Lyocell and to polyester micro-fibre felt showed

3.2 Acoustic Simulation and Measurement

The new measurement technique allowed speeding up the measurement protocol,

An important automation effort had been done leading to a comprehensive MATLAB

In order to simulate successively the absorption coefficients of materials for normal

Design Sensitivity analysis (for final stages, to meet targets) – DSA

Different cost functions, placing varying emphasis on acoustic performance at target

A focused study on the development of treatments to control airborne noise through

1. Insertion loss concept:

2. Insertion loss measurement procedure:

3. Sound intensity measurement:

The approach: In order to perform accurate intensity measurements, one must

• Good moisture resistance

• Easy assembly within the car, which lowered fitting costs

• A variation of sound absorbing characteristics could be achieved by altering the foam

• Material database, i.e. measured acoustic parameters for mathematical

• In-situ vehicle on-road validation test methods for carpet designs

• Optimization process methods for further improving the design

A detailed definition of the design target is discussed in section 3 of Chapter 2.

BENCHMARK ANALYSIS AND TARGET SETTING FOR

It is worth noting that emphasis is given to psychoacoustic and subjective evaluation

2 VEHICLE NOISE ANALYSIS

2.1 Hyundai Sonata vehicle cabin noise test

The test vehicle details are listed as below:

Hyundai Sonata 2005, Auto Sedan

2.4 littre, 4 cylinders, 16 Valves, TARE 1500

Reg. No. TLH 705,

Engine No. G4JS4138682

VIN No. KMHEM41DR5A162022

Spec. 30- 33 psi or 230 kPa

Actual Tyre Pressures: