&EPA
United State*
Environmental Protection
Agency
Office of Nolte
Abatement ft Control
Washington O.C. 20460
EPA 660/9-78-208
Proceedings
Surface Transportation
Exhaust System Noise
Symposium
October 11-13,1977
-------
JUNE 1978
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF NOISE ABATEMENT AND CONTROL
Washington, D.C. 20460
SURFACE TRANSPORTATION
EXHAUST SYSTEM NOISE SYMPOSIUM
OCTOBER 11, 12, 13, 1977
Howard Johnson's — O'Hare
Chicago, Illinois
-------
Table of Contents
ANNOUNCEMENT OF
SURFACE TRANSPORTATION EXHAUST SYSTEM NOISE SYMPOSIUM iv
AGENDA vi
INTRODUCTORY ADDRESS FOR THE SURFACE TRANSPORTATION
EXHAUST SYSTEM NOISE SYMPOSIUM 1
William E. Roper
U. S. Environmental Protection Agency
BENCH TEST PROCEDURES AND EXHAUST SYSTEM PERFORMANCE
PREDICTION 5
P.O.A.L. Davies
University of Southampton
Southampton, England
AUTOMOTIVE EXHAUST SYSTEM EVALUATION 49
D. A. Blaser
Fluid Dynamics Research Department
General Motors Research Laboratories
Warren, Michigan
THE METHOD OF MEASURING EXHAUST SYSTEM NOISE
A STUDY ON THE REDUCTION OF THE EXHAUST NOISE OF LARGE TRUCKS 79
Mineichi Inagav/a
Component Testing Section, Testing Dept.
Mitsubishi Motor Company
Kawasaki City, Nanagawa, Japan
METHOD AND APPARATUS FOR MEASURING MUFFLER PERFORMANCE 109
Peter Cheng
Stemco Manufacturing Co.
Longviev/, Texas
OPTIMUM DESIGN OF MUFFLERS 115
Dr. Donald Baxa
University of Wisconsin, Extension Dept.
Madison, Wisconsin
BENCH TEST AND ANALOG SIMULATION TECHNIQUES FOR ENGINE
MUFFLER EVALUATION 143
Cecil R. Sparks
Applied Physics Division
Southv/est Research Institute
San Antonio, Texas
-------
COMMENTS ON EVALUATION TECHNIQUES OF EXHAUST SYSTEM
NOISE CONTROL CHARACTERISTICS 161
D. W. Rowley
Donaldson Co.
Minneapolis, Minn.
A BENCH TEST FOR RAPID EVALUATION OF MUFFLER PERFORMANCE 181
A. F. Seybert
Dept. of Mechanical Engineering
University of Kentucky
Lexington, Kentucky
ANALYTICAL AND EXPERIMENTAL TESTING PROCEDURES FOR QUIETING
TWO-STROKE ENGINES 203
Donald L. Margolis
Dept. of Mechanical Engineering
University of California
Davis, California
POWER OR PRESSURE - A DISCUSSION OF CURRENT ALTERNATIVES IN
EXHAUST SYSTEM ACOUSTIC EVALUATION 233
Larry J. Eriksson
Nelson Industries, Inc.
Stoughton, Wisconsin
A COMPUTER-AIDED APPROACH TOWARD PERFORMANCE PREDICTIONS FOR
ENGINE EXHAUST MUFFLER 267
John E. Sneckenberger
West Virginia University
College of Engineering
Morgantown, W. Virginia
REVIEW OF INTERNAL COMBUSTION ENGINE EXHAUST MUFFLING 295
Malcolm J. Crocker
Ray W. Herrick Laboratories
Purdue University
West Lafayette, Indiana
SHOCK-TUBE METHODS FOR SIMULATING EXHAUST PRESSURE PULSES
OF SMALL HIGH-PERFORMANCE ENGINES 359
B. Sturtevant
California Institute of Technology
Pasadena, California
CORRELATION OR NOT BETWEEN BENCH'TESTS AND OUTSIDE
MEASUREMENTS FOR SNOWMOBILES 381
Jean Nichols
Bombardier
Research Center
Valcourt, Quebec
Canada
-------
MEASUREMENT OF ENGINE EXHAUST NOISE IN DYNAMOMETER ROOMS 397
James H. Moore
John Deere Horicon Horks
Horicon, Wisconsin
THE APPLICATION OF THE FINITE ELEMENT METHOD TO STUDYING
THE PERFORMANCE OF REACTIVE & DISSIPATIVE MUFFLERS WITH
ZERO MEAN FLOW 401
A. Craggs
Dept. of Mechanical Engineering
University of Alberta, Edmonton
Alberta, Canada
A COMPARISON OF STATIC VS. DYNAMIC TESTING PROCEDURES
FOR MUFFLER EVALUATION 417
H. L. Ronci
Walker Manufacturing
Grass Lake, Michigan
DISCUSSION OF PROPOSED SAE RECOMMENDED PRACTICE XJ1207,
MEASUREMENT PROCEDURE FOR DETERMINATION OF SILENCER
EFFECTIVENESS IN REDUCING ENGINE INTAKE OR EXHAUST
SOUND LEVEL 435
Larry J. Eriksson
Nelson Industries, Inc.
Stoughton, Wisconsin
A THEORETICAL EXAMINATION OF THE RELEVANT PARAMETERS FOR
DYNAMOMETER TESTING OF THE 2-CYCLE ENGINE MUFFLERS 449
Professor G. P. Blair
Department of Mechanical and Industrial Engineering
The Oueen's University of Belfast
PANEL DISCUSSION 497
LIST OF ATTENDEES 547
1X1
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
August 16, 1977
SURFACE TRANSPORTATION EXHAUST SYSTEM NOISE SYMPOSIUM
Sponsored by the U.S. Environmental Protection Agency
Conducted by the Environmental Protection Agency
and McDonnell Douglas Astronautics Company at
Howard Johnson's - O'Hare, Chicago, Illinois
on October 11,- 12, 13, 1977
The U.S. Environmental Protection Agency/Office of Noise Abatement
and Control (EPA/ONAC) has initiated studies pursuant to requirements
established under Section 8 of the Noise Control Act of 1972 which may
lead to Federal requirements for the labeling of surface transportation
vehicles and mufflers with respect to noise.
One study is designed to assess the methodologies available to measure
and communicate the noise reduction characteristics of surface transpor-
tation vehicle exhaust systems. The information communicated may be
actual sound levels or information relative to sound levels (i.e., veri-
fication that a vehicle with a particular aftermarket muffler installed
will meet an applicable standard), or other information such as warranty
claims, proper maintenance and operator instructions, etc. The informa-
tion would be used by dealers, repair facilities, enforcement personnel
and the general public.
The other study is to explore'avenues available to communicate to con-
sumers the noise characteristics of surface transportation vehicles (e.g.
total vehicle noise, interior noise, etc.). This second study, however,
is not the subject of this symposium.
In support of the exhaust system program the EPA desires information
on possible testing procedures which could be used in a Federal muffler
labeling requirement. EPA needs to know whether standardized procedures
exist or can be developed that can be used to characterize muffler per-
formance without having to test exhaust systems installed on the vehicles
for which they are intended.
To gain the necessary information, EPA is sponsoring a three day symposium
scheduled for October 11, 12, 13, 1977 in Chicago, Illinois. Inputs from
industry, research organizations and other interested parties are solicited
to provide information to the government on appropriate procedures.
XV
-------
Papers submitted for presentation should be directed primarily to bench test
procedures and their relationship to total vehicle sound level methodologies
for use in a Federal regulatory requirement. The methods discussed may
include the following:
o System testing using a standard sound source,
o analytical simulation techniques, and
o combination of testing and analytical methods.
Information that must be developed on vehicle or vehicle engine sound
characteristics (other than total vehicle noise) to make muffler labeling
useful should also be addressed.
While the primary purpose of the symposium is to assess "bench test
methodologies" and their use in a Federal regulatory requirement, it may
be necessary to address other testing methodologies, in the event that
a suitable bench test methodology does not appear to be available. In
this light a limited number of papers will be accepted on stationary (near
field) and dynamometer test methods, results and their relationship to
moving vehicle noise test methods.
Six sessions of in-depth papers are planned to cover all aspects of exhaust
system bench testing. Three plenary sessions will be held emphasizing the
application of various exhaust system bench test methods.
More information may be obtained from:
Environmental Protection Agency McDonnell-Douglas Astronautics Co.
John Thomas E. T. Oddo
Office of Noise Abatement McDonnell-Douglas Astronautics Co.
and Control (AW-471) 5301 Bolsa Avenue
Environmental Protection Agency Huntington Beach, CA 92467
Washington, D.C. 20460 Tel: (714) 896-4412
Tel: (703) 557-7666
Abstract of papers should be submitted to E. T. Oddo, MDAC no later than
September 19, 1977.
Room accommodations can be arranged at:
Howard Johnson's - O'Hare
10249 West Irving Park Road
Schiller Park
Chicago, Illinois 60176
Tel: (312) 671-6000
v
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
AGENDA
TUESDAY 11 OCTOBFR
8:30 - 9:30 am Registration
9:30 Opening Address
EPA, Washington, D.C.
SOUND GENERATION BY AN INTERNAL COMBUSTION ENGINE EXHAUST
A. J. Bramaer, National Research Council of Canada,
Ottawa, Canada (Paper not available)
TEST PROCEDURES AND EXHAUST SYSTEM PERFORMANCE PREDICTIONS
P.O.A.L. Davies I.S.V.R., University of Southampton,
Southampton, England
2:00 pm AUTOMOTIVE EXHAUST SILENCER EVALUATION
Dwight Blaser, General Motors Technical Center, Warren, Mich.
THE METHOD OF MEASUREMENT FOR EXHAUST SYSTEM NOISE
Mineichi Inagawa, Mitsubishi Motor Co., Nanagawa, Japan
METHOD AND APPARATUS FOR MEASURING MUFFLER PERFORMANCE
Peter Cheng, Stemco Mfg. Co., Longview, Texas
COMPUTER PROCEDURE FOR ASSESSING MUFFLER PERFORMANCE
Donald E. Baxa, University of Wisconsin, Madison, Wise.
WEDNESDAY 12 OCTOBER
8:30 - 9:30 am Registration
BENCH TESTS AND ANALOG SIMULATION TECHNIQUES FOR MUFFLER
EVALUATION
Cecil Sparks, Southwest Research Inst., San Antonio, Texas
COMMENTS ON EVALUATION TECHNIQUES OF EXHAUST SYSTEM NOISE
CONTROL CHARACTERISTICS
D. W. Rowley, Donaldson Co., Minneapolis, Minn.
vi
-------
BENCH TEST FOR RAPID EVALUATION OF MUFFLER PERFORMANCE
Andrew S. Seybert, University of Kentucky, Kentucky
ANALYSTICAL AMD EXPERIMENTAL TESTING PROCEDURES FOR QUIETING
TWO-STROKE ENGINES
D, Margolis, University of Calif, at Davis, Davis, Calif.
2:00 pm POWER OR PRESSURE - A DISCUSSION OF CURRENT ALTERNATIVES IN
EXHAUST SYSTEM ACOUSTIC EVALUATION
Larry J. Eriksson, Nelson Industries, Inc., Stoughton, Wise.
A COMPUTER-AIDED APPROACH TOWARD PERFORMANCE PREDICTIONS FOR
ENGINE EXHAUST MUFFLER
John E. Sneckenberger, West Virginia University, Morgantov/n, VA
REVIEW OF INTERNAL COMBUSTION ENGINE EXHAUST MUFFLING
Malcolm J. Crocker, Herrick Laboratories, Purdue University,
West Lafayette, Ind.
SHOCK TUBE METHODS FOR SIMULATING EXHAUST PRESSURE PULSES
OF SMALL HIGH PERFORMANCE ENGINES
B. Sturdevant, California Institute of Technology, Pasadena,
Calif.
THURSDAY 13 OCTOBER
8:30 - 9:30 am Registration
9:30 am CORRELATION OR NO, BETWEEN BENCH TESTS AND OUTSIDE MEASUREMENTS
FOR SNOWMOBILE EXHAUST SYSTEMS
Jean Nichols, Bombardier Research Center, Valcourt, Quebec
A METHOD OF MEASURING ENGINE EXHAUST NOISE IN A DYNAMOMETER
ROOM
James W. Moore, John Deere, Horicon Works, lloricon, Wisconsin
THE APPLICATION OF THE FINITE ELEMENT METHOD TO STUDY THE
PERFORMANCE OF REACTIVE & DISSIPATIVE MUFFLERS WITH ZERO MEAN FLOW
A. Craggs, University of Alberta, Alberta, Canada
COMPARISON OF STATIC VS. DYNAMIC TEST PROCEDURES FOR MUFFLER
EVALUATIONS
W. Ronci, Walker Manufacturing Co., Grass Lake, Mich.
DISCUSSION OF PROPOSED S.A.E. RECOMMENDED PRACTICE SJ1207
MEASUREMENT PROCEDURE FOR DETERMINATION OF SILENCER EFFECTIVE-
NESS IN REDUCING ENGINE INTAKE OR EXHAUST NOISE
Larry J. Eriksson, Nelson Industries, Inc., Stoughton, Wise.
vii
-------
2:00 - 4:00 pm PANEL DISCUSSION
Contributed Paper - Unable to Attend
A THEORETICAL EXAMINATION OF THE RELEVANT PARAMETERS FOR DYNA-
MOMETER TESTING OF 2-CYCLE ENGINE MUFFLERS
Professor G. P. Blair, Queens University of Belfast,
Belfast Ireland
viii
-------
OPENING ADDRESS SURFACE TRANSPORTATION EXHAUST SYSTEMS
NOISE SYMPOSIUM
Hi Hi am E. Roper
U. S. Environmental Protection Agency
It is my pleasure to welcome you to EPA's Surface Transportation
Exhaust Systems Noise Symposium here in Chicago. This is the first
major action EPA has undertaken through the labeling related respon-
sibilities of the Agency with regard to systems and components used to
a large degree in the surface transportation vehicles. In the past,
EPA has set legal noise standards for medium and heavy trucks and has
recently proposed noise*emission standards for buses, truck-mounted
solid waste compactors, and truck-mounted refrigeration units; in
addition to a number of other standards applicable to non-surface
transportation type vehicles. On all these vehicles, the exhaust
system is one of the important noise sources and in some cases the
principal source of noise. Throughout the life of a vehicle, compo-
nents of the exhaust system, particularly the muffler and portions of
the exhaust tubing are replaced as a routine maintenance practice on a
cyclic basis throughout the useful life of the vehicle. Because of these
characteristics, vehicle exhaust systems appear to be a good candi-
date for consideration in a Federal labeling program.
-------
EPA has already iMpler.icnted its general policy on noise labeling
and recently published a notice of proposed rulemaking laying the criteria
for such action. The specific objectives of EPA's labeling program
in the noise area include:
(1) Providing accurate and understandable information to product
purchasers and users regarding the acoustical performance of designated
products so that meaningful comparisons could be made concerning the
acoustical performance of the product as part of the purchase or use
decision.
(2) Providing accurate and understandable information on product
noise emission performance to consumers with minimal Federal involvement.
(3) Promoting public awareness and understanding of environmental
noise and the associated terms and concepts.
(4) Encouraging e/fective voluntary noise reduction and noise
labeling efforts on the part of product manufacturers and suppliers.
At this time, our study efforts are directed primarily at the assess-
ment of available measurement methodology techniques to adequately
define exhaust system noise performance. Clearly, the development of an
exceptable measurement methodology to be used to determine the appro-
priate acoustic performance information is central to being able to
properly label an exhaust system or exhaust system component. To assist
the Agency in carrying out this task, we have contracted tiwh McDonnell
Douglas Astronautics Company to provide technical support in this specific
area. A portion of their contract calls for the assessment of existing
-------
and proposed total vehicle sound testing methodologies to report on the
status of current muffler labeling required by Federal, State, or local
regulation and voluntary labeling programs, development of a general
description of the current aftermarket muffler industry and to organize
and assist in conducting this symposium of acknowledged muffling system
experts on the feasibility of using methodologies other than base-line
total vehicle sound procedures for evaluating exhaust system noise per-
formance.
We recognize -that the area we are about to embark on is one of many
technical complications and has equally sizable communication cnni'lica-
tions in order to effectively provide simplistic information to a consumer
or user. The initial step however, remains the development of an accept-
able measurement methodology to identify the acoustic performance of
exhaust systems. The symposium for the next three days is designed to
specifically focus on this issue with particular emphasis on assessment
of bench test procedures and their relationship to total vehicle sound
level methodologies. The methods that will be presented and reviewed in
the following three days will include but not be limited to: system
testing using a standard sound source, analytical simulation techniques,
and combination of testing and analytical methods.
For the next three days, we will likely have assembled in this room
some of the best expertise available on this subject. I hope that through
a constructive and objective interchange of ideas, we as a group will be
able to focus on the issues and develop specific recommendations for
testing of exhaust systems that can be related to total vehicle sound
levels and have potential use in a Federal regulatory labeling program.
-------
BENCH TEST PROCEDURES AND EXHAUST SYSTEM PERFORMANCE
PREDICTION
by P.O.A.L. DAVIES
SUMMARY
This contribution reviews the present state of development of
a rational approach to exhaust system performance evaluation based
on static test bed measurements. This depends primarily on a
quantitative understanding of the generation and propagation of
sound energy in ducts which are carrying a hot, high velocity gas
flow.
Elements of the approach are described which include methods
for characterising the sources, analytic or experimental methods
for adequately modelling the acoustic behaviour of system components,
appropriate precautions for assessing inter-component interactions
and a scheme for identifying those situations where source system
interactions can be important.
Component models are expressed in terms of transfer matrices,
or their equivalent, relating the pressure and volume velocity at
input to output. A useful range of linear analytic models for
reactive system components is described. Examples are presented
comparing bench measurements with predictions for a representative
set of practical systems including the U.K. Quiet Heavy Vehicle Project.
-------
BENCH TEST PROCEDURES AND EXHAUST SYSTEM PERFORMANCE PREDICTION
1. INTRODUCTION
A systematic and rational approach to the control of piston engine
intake and exhaust noise requires a quantitative specification of the
silencing requirements, with a procedure for the quantitative evaluation
of system acoustic and mechanical performance. This contribution reviews
the present state of development of such an approach which is based on
bench testing. Such tests concern primarily test bed measurements with
a running engine, but some of the details required for modelling system
elements and their behaviour have been provided with special cold flow
rigs.
The prediction of system performance usually concerns the calculation
of the transport of acoustic energy through the system from the source
to the outlet where.it \s radiated, |l|. For this.one requires a set
of models which describe the acoustic transfer characteristics of each
system element in quantitative (erms \2\ t with an analytical procedure
for combining the elements together to describe the overall transport
of energy through the complete system |2, 3|. An element may be
described as any part of the duct system that has an effect on the
propagation of acoustic waves (or energy) through it. Thus, in this
connection, the engine, sections of connecting pipe, the open end of the
system and any duct discontinuity or muffler component are all acoustic
elements.
Silencing requirements are normally determined by first performing
open pipe noise measurements, covering the full operational load and
speed conditions of the engine. This information can then be compared
with the statutory or specified noise limits to provide a quantitative
description of silencing requirements. If the open pipe data are
properly evaluated, they can also be used to describe the acoustic source
characteristics of the engine. This information provides a starting
point for the quantitative evaluation of the inlet or exhaust system
acoustic performance. Thus open pipe measurements with a loaded engine
represent one essential part of the test procedure.
7
-------
Acoustic performance is generally described in terms of insertion
loss. This can be defined as the difference in sound pressure level,
measured at a fixed reference point, between the noise emitted by an
open pipe and the noise emitted by "the silenced intake or exhaust.
Note that this definition assumes that the observed difference is due
to the presence of a muffler unit in the system and that the source
remains unchanged.
When the system i modified, it is well established |41that the
observed performance can be strongly influenced by the relative
positioning of the muffler unit along the exhaust or inlet duct.
That this should happen is well understood, since the sections of pipe
connecting components of the system each have a clearly identifiable
acoustic behaviour, depending on their length. This then forms part
of the installed response of the muffler unit. For this reason trans-
mission loss alone is not an appropriate practical method for describing
the acoustic performance of intake or exhaust system components.
Mechanical performance can be assessed in terms of the effect of
the intake and exhaust system on engine power and efficiency. Other
mechanical factors include' the packaging of the system components to
minimise flanking transmission, cost and weight, to provide adequate
durability and to fit in with dimensional or other installation constraints.
Some of these considerations have a direct effect on acoustic performance
and must be included in the noise control analysis.
The intake and exhaust gas is normally flowing sufficiently rapidly for
this to have a significant effect on acoustic performance. Furthermore,
the exhaust gas is hot^so significant temperature gradients exist which
change with engine (or vehicle) speed and load. Due allowance for these
operational and gas flow factors must be made during the performance
predictions and sufficient data for this purpose assembled during the
measurements. The mean kinetic energy of the gas flow may also be
converted to new sourc-es of acoustic energy within the intake or exhaust
system, appearing either as broadband flow noise, or as regenerated pure
tone components. Finally, there is good evidence \5\ that changes in
system acoustic characteristics may also modify the engine breathing
characteristics and consequently the acoustic source strength of the
engine.
-------
In summary, the procedures for bench testing and system performance
prediction for inlet and exhaust noise control can usefully be subdivided
into a set of study areas, namely:
a) Methods for measuring and characterising the acoustic source.
b) The specification of silencing requirements.
c) The assessment of operational factors with their relative
significance. For example, gas flow and gas temperatures,
mechanical performance, space constraints and flow or internal
noise generation.
d) Methods for modelling the acoustic transfer characteristics of
the system elements based on performance measurements or an analysis.
e) A procedure for assembling the elements together to provide an
appropriate description of the system, including all the inter-
actions between elements.
f) An appropriate procedure for predicting or determining overall
system performance including techniques to identify problems
arising from source system interaction.
Each of these factors will be considered in the light of current
knowledge and practical experience, indicating the level of confidence
with which the evaluation can be performed at the present time.
2. ACOUSTIC ENERGY PROPAGATION IN FLOW DUCTS
Sound propagation in flow ducts can be described by linear transmission
line equations. These are based on conservation of mass, energy and
momentum and describe the variation of acoustic pressure and particle
velocity associated with the wave motion in terms of position in the duct.
In their simplest and perhaps most practical form the flows and the wave
motion are both assumed to be one-dimensional. With these restrictions
exact solutions can be obtained for a comprehensive range of duct geometry
and boundary conditions. However, if the solution is to remain realistic
in terms of observed behaviour, special considerations may be necessary
to soecify acoustic conditions at discontinuities, as will be shown later.
-------
Empirical descriptions of acoustic performance become necessary
where a system element exhibits a strongly non-linear behaviour. Such
can be the case, for example, with acoustic transmission through orifices
with normal or grazing flow, or with sound transmission along passages
lined with absorbing materials. Other examples include flow-acoustic
coupling and amplification associated with flow-separation or edge-tones
as well as flow noise. Some examples of such behaviour are also
considered later.
2.1 Plane wave propagation in flow ducts
Acoustic energy propagation is by a wave mechanism, the energy
being provided by a source which excites the wave motion. At each
duct discontinuity some of the energy is transmitted as a new wave the
remainder being reflected, both waves travelling with a phase velocity
c relative to the gas. With one-dimensional wave propagation in ducts
one can describe the pressure p+ and particle velocity v+ in the positive
going (incident) wave by
•
+ "+ i(u)t-k+x) -ax
p^ = p+e e ,
v+ =
Z
s
2. Kb)
where p and v are the pressure and velocity amplitudes, oj the radian
frequency, k+ the wave number u)/(c+U), U the mean flow velocity and
a a coefficient which represents the decay of wave energy as it propagates
along the duct. Similarly the reflected wave is described by
2.2(a)
f >
s
where k = w/(c-U) .
An alternative description is to express the pressure etc by p+elaJte~YX,
where y - a+ ig. With hard walled ducts a-K> arid g^k+ while the duct
impedance Z^=pc, the characteristic acoustic impedance of the gas.
The sound pressure and particle velocity at any point are then given by
10
-------
P = P+ + p" 2.3(a)
v = v+ + v~ 2.3(b)
To represent discontinuities, one first notes that some of the
incident wave energy will be reflected and some transmitted. The
ratio(usually complex) of the reflected to incident wave amplitude,
termed the reflection coefficient r,is expressed by
r -El ->!_!§. " Re1*, 2.4
;+ "vzs
when the boundary conditions at the discontinuity are specified as
an impedance Z For an open end, the phase angle can be obtained
from the solution given in J6| for zero flow. The appropriate value
of R for various flow Mach numbers U/c can be found in |l|. Similar
relations for a baffled opening can be found in |?|.
describing conditions at x , then the pressure amplitude p at any
Neglecting for simplicity the attenuation along the duct with 2.4
ibing conditions at x , then the pres
other point x in a plain duct is given by
. . + . . .. -
~1K X X
- Po(ei(k" - ) e + Ree > 2.5
where k* = i (k+ + k~) . = u/c(l-M2). This shows that the distance
between the nodes of the standing waves is reduced by the factor- (1-M2)
with flow present, compared to the zero flow case. Thus the existence
of flow modifies the frequencies at which lengths of duct (and other
elements) resonate.
2.2 Acoustic Conservation relationships for flow ducts
With plane waves in a uniform flow duct, conservation of mass is
satisfied 2 if
A j(l+M)p+ - (l-M)p^]= a constant > 2.6
where A is the duct cross-section area,
11
-------
Similarly it can be shown that, for isentrop'ic conditions,
conservation of energy is satisfied if
(l+M)p+ + (l-M)p~ = a. constant. 2.7
Given a uniform duct of length £ with a steady flow of Mach number M,
one can show that conservation of acoustic energy and of mass flow for
non-decaying waves is satisfied by the simple transfer relationships
"+ ~+ ~ik+£ -- ~- ik £ 9 o
P^ = P0e and P£ = po e '
The termination conditions are often defined by
pc(po + PO} PC(P£ + PO)
Z = — and Z = — . 2.9
o ~+ £ *•+
po - po p£ ~ p£
This result indicates tl!at it is necessary to include measurements of
flow temperature and mean mass flow, to evaluate k , k and M. If the
duct wall pressure p is measured or determined, one also requires a
knowledge of Z before p can be decomposed into its two components
p and p . However, given Z , Z can then be evaluated, and so on.
Since the open pipe discharge impedance Z can be specified from
established data, the modelling of system characteristics can conveniently
begin here. The decay of the wave amplitude in ducts of significant
length can be included'by multiplying the right-hand side of 2.8 by a
a£
factor e , with a negative and dependent both on frequency and Mach No.
At discontinuities, however, the assumption that the flow is
isentropic is hardly realistic, particularly at the rapid changes in
duct cross section thac occur in expansion chambers etc. The transfer
characteristics can be established, however, along the lines set out in
reference |2|. Flow lossess and the consequent entropy changes can
be represented by a loss factor 6. (but see |s|). Describing acoustic
and flow properties before the discontinuity by the subscript 1 and those
well downstream by the subscript 2 and neglecting changes in mean density,
one can set out the conditions for conservation of mass flow, energy and
12
-------
momentum flux across the discontinuity.
Conservation of mass is expressed by
A2{p2(l+M2) - p2(l-M2)J = AjpiCl+Mi) - p^Cl-Mj) + 6Mj], 2.10
while conservation of energy is satisfied if
p£(l+M2) + p2(l-M2) = ptd+Mi) + pld-Mj) - 6/(Y-D, 2.11
where y is tne ratio of the specific heats. Momentum is conserved
if
n
P£[AI + A2(2M2 + M22)] + p2[Aj + A2(M2 - 2M2)]
o
= ptfAjU+M!)2] + ^TfAid-Mj) ] + 6A1M12 . 2.12
For one-dimensional flow, and known geometry, the incident and reflected
+ —**...
waves p2 and p2 after the discontinuity can be found in terms of the
known incident and reflected waves before it, after the unknown loss
factor 6 has been eliminated from the three equations. Thus these
three equations can be used to define a transfer relationship for any
area discontinuity. Other types of discontinuity can be treated using
a similar approach. One should note that the phase changes occurring
across the discontinuity can be determined from a non-propagating higher
order mode analysis for zero flowsthat satisfies the boundary conditions,
The mean acoustic energy flux per unit area of duct, or the
acoustic intensity, is expressed as
pv
where p and v are the r.nus. pressure and velocities respectively and
the overbar represents a time average. In terms of the wave components
this becomes, using 2.6 and 2.7,
(l-M)2<(p")2>] 2.13
13
-------
where the symbol < > represents taking the time mean value. The first
term in the brackets can be interpreted as an energy flux with the flow
or the incident wave motion, while the second represents energy flux
against the flow, or energy carried by the reflected waves.
The level of the sound radiated by the exhaust outlet can be
obtained by equating the nett energy in the tailpipe to that of a
spherically diverging wave. This gives for a tailpipe of radius a
2.14
o o
where pr "is the'r.m.s. acoustic pressure measured at a distance r
from the outlet. Equation 2.14 can be employed to determine the
fluctuating pressure level in the tailpipe from free field measurements,
provided the Mach number and radiation impedance are known.
The analysis presented above is restricted to situations where the
behaviour can be characterised by linear acoustic theory. Examples
are presented which indicates that this assumption is not restrictive
for many practical applications. The analysis presented is not the
only effective way of describing system characteristics since an alter-
native approach using transfer matrices has been described elsewhere J3|,|4|.
Though omitted for simplicity, the analysis can be extended to
the decay of the waves as they propagate. Axial temperature gradients
may also be accommodated by sub-dividing elements into smaller sections
where the temperature can be regarded as substantially constant.
2.3 Some examples of sound transmission across discontinuities
To complete this review of acoustic energy propagation in ducts,
some examples are presented comparing the measured characteristics of
some typical discontinuities obtained with flow rigs with predictions
based on the analysis presented here. A further series of comparisons
based on test bed or field measurements with silencer components and
systems can be found in references 1 1 1 , j 2 | , ] 5 | 'and J9J.
The first example concerns acoustic energy transport across a
14
-------
contraction which includes a sidebranch. The measurements were
performed with a special cold flow rig provided with a high intensity
acoustic source. Figure l(a) presents the measurements made at three
flow Mach numbers, the predictions assuming plane wave motion throughout
.and a higher order mode (exact) analysis for zero flow. The plane
wave analysis, represented by equations 2.10 to 2.12, cannot model the
zero acoustic particle velocity boundary condition on the wall at the
annulus between the inner and outer pipes forming the contraction.
The zero particle velocity condition here can be closely approximated by
including the first five radial modes, and this calculation provides the
exact result shown in the figure.
Comparison with the measurements shows that the plane wave analysis,
which includes a small decay factor for the waves in the sidebranch,
correctly predicts the amplitude of the transmitted waves, as can be seen
in Figure 1 (b) , but there is a constant frequency error. The exact
analysis for zero flow does however predict the frequency correctly.
Thus a combination of both methods of analysis provides an adequate
» *
description of the transfer characteristics of the discontinuity, with
plane wave analysis defining amplitude characteristics and higher order
mode analysis the phase change.
A second example concerns an area expansion with a sidebranch and
the results are illustrated in Figure 2(a) and 2(b). In this case the
boundary conditions at the discontinuity must also include the fact that
the flow separates at the end of the pipe, forming a jet. A detailed
analysis of this problem has been presented by Cummings |10| who shows
that amplitude characteristics are correctly predicted if the pressure
waves are assumed plane, but that the flow retains a top hat velocity
profile. Again comparison with measurements shows that amplitude
characteristics are adequately modelled by plane wave theory and that the
correct phase change can be predicted by higher order analysis.
The higher order mode analysis in laborious and a systematic
investigation |11| showed that the phase change can be calculated by
an appropriate end correction. This is analogous to the well known end
correction of just over 0.6 of the pipe radius that is applied for
15
-------
predicting the acoustic resonance of organ pipes to account for
fluid inertia effects at the discontinuity. The end corrections
appropriate to expansions or contractions in flow ducts are illustrated
in Figure 3. The dotted line indicates the lower frequency limit for
propagating higher order modes when plane wave analysis breaks down.
It can be seen that the corrections tend to the open pipe limit at
large area ratios. Furthermore, as a percentage of the duct length^
they become small for long connecting pipes and could be neglected
in practical prediction calculations.
A third example concerns the performance of folded chambers.
Effectively these can be regarded either as a Helmholtz resonator,
or a sidebranch, which for convenience of packaging is wrapped around
the expansion section. This geometry has the added advantage of
avoiding high velocity cross flow at the resonator neck, avoiding problems
with flow excitation. A detailed analysis including higher order modes
to match boundary conditions at the three connecting annuli has been
reported by Cummings |l2J. The predictions with an alternative and
simpler approach based on end corrections etc. by Adams |ll| is compared
with flow rig measurements in Figure 4. This illustrates the way that
the system resonance can be modified by changing the area of the neck,
a useful feature for tailoring acoustic characteristics within spacial
constraints. The good agreement between predictions and observations
illustrates the effectiveness of the modelling techniques described above.
2.4 Acoustic sources in intake and exhaust systems
An account of acoustic energy propagation in flow ducts would be
incomplete without some consideration of the sources. The primary sound
source provided by the unsteady flow processes at the valve. The
amplitude of these pressure fluctuations can be as high as 0.5 bar,
while the frequency spectrum consists of the first 100 or more harmonics
of the fundamental firing frequency for one cylinder. One can show,
by dimensional reasoning, that the source strength at any fixed frequency
varies as N , where N is the engine rotational speed. Broadband noise
at higher frequencies is also provided by broadband flow noise generated
at the valve, and at discontinuities where the flow can separate. This
16
-------
spectrum exhibits a flat peak at a characteristic Strouhal number FL/U
of around unity, where L is a characteristic scale of the source region
and U the phase velocity of the disturbances acting as sources. Such
noise will vary, at fixed frequencies, as N . Noise generated by flow
turbulence at the duct walls, bends etc. may also represent a significant
source of high frequency sound. Its strength will vary as V , where V
is the mean duct flow velocity.
Turbocharging modifies the exhaust noise signature since it tends
to reduce the amplitude of the low frequency components arising from
gas release processes. It may add new sources of noise generated by
unsteady flow interactions in the turbine or blower, by wake noise from
the blades or nozzles and so on. The strength of such sources tends
to vary as V where V is the mean turbine outlet flow velocity. The
o o
characteristic frequencies of such sources may be high, of the order of
the turbine blade passing frequency and its harmonics.
The strength of the- sources associated with the engine breathing
or the turbocharger can be studied and evaluated on the test bed. Flow
noise and acoustic regeneration within the silencer system represents
a different problem that can better be studied with special rigs. These
latter are generally lower in intensity than those associated with the
engine but are of practical significance since they set an upper limit
to the maximum attenuation that can be obtained unless care is taken
to minimise them.
Flow noise is broad band, generated by flow separations at valve
lips, bends, expansions, contractions and by turbulent boundary l-ayer flow.
It is of most significance when amplified by cavity resonances which
provide feed back to intensify the source. Noise generation by the
impingement of the jet formed at the chamber entrance on the lip of the
exit pipe in a steady flow rig is illustrated in Figures 5 and 6. The
broad band spectrum in Figure 5 has been modulated by tailpipe (peaks)
and chamber ^troughs) resonances. Figure 6 illustrates the way source
strength varies with pipe separation x/d and with flow velocity.
Practical separations lie close to x/d = 2, whare the strength is
greatest. Scaling the measurements to correspond to a 75 mm diameter
tailpipe with a flow Mach number of 0.26 at 600 C yields a sound pressure
17
-------
level of 85 dBA at 7.5 metres. This represents a minimum level for
tailpipe self-excitation unless this noise producing mechanism can
be suppressed.
Figure 7 indicates how this source of noise can be controlled
or reduced in strength by bridging the gap between the inlet and outlet
with a perforated pipe. The acoustic behaviour of the expansion
chamber is not significantly changed if the perforated pipe has about
20% open area, that is the hole pitch is of the order of twice the hole
diameter. Perforate C had stabbed holes 1.9mm across at 3.8mm pitch
giving an open area of 20%, while perforate D had holes 4.5mm diameter
at 7.5mm pitch giving an open area of 27%. The details of the hole
formation can be critical if high frequency discrete tone generation
(singing) by the perforate is to be avoided. Figure 8 shows that
perforates are of value in reducing back pressure and indicates the
'magnitude of the back pressure penalty that must be accepted, when sharp
changes in flow direction are employed in a silencer system.
The measurements in Figures 5 to 8 correspond to steady flow rigs
with a specially acoustically treated quiet supply system. Other
experiments were performed with single tone high level (up to 160 dB)
acoustic excitation. Some typical results are illustrated in Figure 9.
The solid lines on the figure represent the amplitude transfer charact-
eristics calculated by the linear acoustic methods described earlier.
The behaviour of an acoustically excited jet has been studied in connection
with jet noise and is fairly well understood |l3|, but the mechanisms
are non-linear and have been difficult to quantify. The results
illustrate (a) a relative- increase in transmitted sound at low forcing
levels due to high amplification by the shear layer, (b) a close approach
to predicted transmission at high levels of excitation due to saturation
when the shear layer amplification becomes negligible, and (c) a complex
interaction between the travelling vortex potential field, the sound
field and the tailpipe resonance at intermediate levels of forcing.
Fortunately, this complex non-linear behaviour can be effectively suppressed
by fitting perforated bridges as illustrated by the measurements in
Figure 10.
18
-------
An example which illustrates this noise generation mechanism in
more detail is provided by the acoustically syncronised vortex shedding
that is found at an expansion. The observations and analysis are
illustrated in Figure 11. The acoustic standing wave with zero flow
that is predicted by linear acoustic analysis with 130 dB excitation in
the upstream duct is shown in Figure 11 (a). This wave can be described
by
/ ,_s A . ,
p .(.x,t; = p sink^xe
a Q.
The f,onn of the travelling potential field associated with the shed
vortices in Figure ll(b) has been developed from a number df observations
of excited jet flows |ll|. It is an estimate rather than a prediction
but can be closely described by
, ,. -akox i
P (x,t) = p e 2 e
v v
The combined pressure distribution is the sum of the potential and
acoustic fields. The mean square value of the sum has been calculated
and then plotted for comparison with observations made with a travelling
probe microphone in Figure 11(c). The agreement is within the accuracy
of the measurements.
Though not efficient radiators in free space, the travelling
potential field of the vortices can interact with nearby surfaces
which then may radiate strongly. This is what appears to happen within
the expansion chamber, the effect being amplified by resonance in the
chamber and tailpipe. The role of the perforate bridge is thus to
suppress the vortex formation while leaving the other acoustic properties
unaffected.
It is tempting to speculate whether, perhaps, many of the non-linear
acoustic characteristics found with silencer elements or silencer systems
may not be the result of similar mechanisms that include vortex shedding
at discontinuities.
19
-------
3. MEASUREMENT AND PREDICTION OF EXHAUST SYSTEM ACOUSTIC PERFORMANCE
As outlined earlier, evaluation of exhaust system acoustic performance
is based on insertion loss measurements or predictions. For this
purpose open pipe (unsilenced) system measurements on a test bed are
required as well as measurements with the muffler units included. The
prediction of insertion loss involves, ideally first calculating the
transfer characteristics of the open pipe and then, starting at the
tailpipe, calculating the transfer characteristics of the system with
silencer units included. The insertion loss can then be calculated
from the ratio of the two transfer characteristics.
For simplicity (see for example |9|) the open pipe transfer chara-
cteristics may be taken as unity, and the predicted attenuation of the
system is then taken as the insertion loss. This can be acceptable
in situations where the run of exhaust pipe between engine and muffler
is at least two or more wavelengths long at the lowest exhaust frequency,
with the muffler situated near the exhaust discharge. Predicted
attenuation may not correlate well with measurements of insertion loss
* »
when the exhaust pipe is relatively short and the system contains two
or more distributed muffler units set well apart.
As mentioned earlier, the transfer characteristic can be calculated
working with the incident and reflected waves as described here, or by
using transfer matrices representing the relation between input and
output pressure and volume velocity for each element. The two methods
should give precisely the same results, if based on the same assumptions
and boundary conditions, as long as the input to each element in turn
is taken as the output of the preceding one.
This procedure is valid so long as the source characterisitcs
remain invariant at each of the prescribed engine running conditions.
Uncertainties will also arise due to flow noise generation within the
system unless due allowance for this can be included in the model for
each element. From what has been shown already, it is clear that the
appropriate flow and temperature conditions must always be included
in the analysis.
20
-------
3.1 Source characteristics
The acoustic source characteristics of the engine can be deduced
from open pipe measurements. To illustrate how this can be carried
out we must first set out a model of the system which identifies the
source as an element.
The primary sources are provided by the unsteady flow through the
valves which can be represented acoustically by a fluctuating volume
velocity. To complete the description of the source one must also
specify the effective source impedance. Each valve flow provides an
individual contribution to the total source strength which combine in
the manifold. A convenient reference plane for definition of source
characteristics is therefore the manifold or turbocharger outlet flange.
The source strength can be specified at this reference plane as a
fluctuating volume velocity U with an effective source impedance Z ,
both being quantified as complex variables. The exhaust system (or
inlet system) represents an acoustic load applied to the source. This
can be specified as-a fluctuating volume velocity U with an effective
With these definitions the acoustic model of the source
and system appears as shown in Figure 12.
impedance Z .
U
me
Uc
Figure 12 Acoustic model of engine and exhaust system
The driving pressure at the manifold or turbocharger outlet flange
p can be expressed as
s
p = U Z = (U -U )Z
rs s s m s m
3.1
Acoustic measurements obtained with microphones or transducers are
usually expressed as sound pressure levels. This information is usually
21
-------
reported as the root mean square value of the pressure. Such
information, e.g. P which is an r.m.s. pressure say, is not directly
s
comparable with the sound pressure p defined by equation 3.1, since
s
phase information has been discarded in the signal processing.
3. 2 Open pipe measurements
Open pipe noise measurements are usually sound pressure level
recordings made under effectively free field conditions. The results
are normally presented as a narrow band spectrum of the radiated noise
and represents the radiated sound energy. Thus the record represents
the spectrum of the signal P in equation 2-. 14. Provided the necessary
flow data have been recorded at the same time, this information,
together with a definition for tailpipe impedance Z , can be used with
equation 2.14 to evaluate the amplitude spectrum of the tailpipe
incident wave p+.
o
With a straight open pipe of length £- , the fluctuating volume
velocity U , < at the_ source plane can then be calculated as
pc
3.2
where A is the cross section area of the pipe. The pressure at the
driving plane, p can be found from
S
3.3
Repeating the observations, with a different acoustic load (i.e.
change of &) provides a second estimate of U and p . Provided U
s s m
and Zm Ere unaffected by changes in Z , this information can be used to
solve equation 3.1 for these two variables which characterise the source.
Some evidence exists I 111 that U and Z remain unaffected with a turbo-
' ' m m
charged engine, but will alter with a change in Z for a single cylinder
S
naturally aspirated engine.
Alternatively, one can predict the insertion loss or predict the
sound radiated by a silenced system from the open pipe measurements.
22
-------
The observed radiated spectra are adjusted assuming U remains
unaltered, but rather than p+ changes as the system is altered, in
accordance with the known changes to Z . This procedure is illustrated
by the results in Figure 13. The measurements were taken from reference
14, and were obtained on a special flow rig where the volume velocity of
the source was maintained constant. The results show that the silenced
system performance can be closely predicted from the open pipe
measurements using linear plane wave theory even though the pressure
wave amplitude was in excess of 0.5 Bar.
The calculations for the comparisons in Figure 13 were fairly
straightforward, since the flow temperature, mass flow and source
frequency were all constant. Test bed measurements on an engine
involve covering a wide range of speed and load conditions, which result
in large changes in flow temperature and temperature gradients, mass
flow velocity, source strength and so on. In noise control analysis for
the engine, the predicted system performance must provide a specified
though perhaps different minimum insertion loss for each operating
condition.
The problem can be simplified somewhat, by first assembling the
measured data in the most general way. One method of doing so is
illustrated in Figure 14. The upper figure is a carpet plot of a
narrow band analysis of the open pipe radiated noise for five engine
speeds at full load torque. Each record has been normalised in sound
pressure level by dividing by the corresponding mean flow dynamic head
at the manifold exit plane. This provides a plot where the increase
in radiated sound pressure due to increased engine speed has Keen
normalised.
A second normalization has been carried out on the data in Figure 14(b)
The data from each run have been replotted on a basis of k*£ (see
equation 2.5). The modulation of the radiated noise amplitude is
clearly in step with the .open pipe load impedance changes. Figure 14(b)
then represents the presentation of open pipe data for which insertion
loss comparisons are most likely to correspond to'predicted system
performance.
23
-------
3.3 Comparisons between -predicted and measured system performance
One example of a comparison between the measured performance of
an exhaust system and that predicted with linear acoustic theory from.
open pipe measurements was presented in Figure 13. A further example
of a similar comparison based on measurements with an engine on a test
bed is presented in Figure 15 and 16. These results formed part of
the systematic exhaust system bench test and design studies for the
quiet heavy vehicle project sponsored by the Department of the
environment in the United Kingdom.
The unsilenced noise of this turbocharged engine was 105 dBA at
7.5 metres under full load with an open pipe exhaust sytem. The design
specification for the system required exhaust levels below 70 dBA at
7.5 metres for any speed or load condition with a back pressure limit
of 45 mm of mercury. Open pipe measurements were performed and analysed
using the linear acoustic methods already described in this report.
The resulting open pipe noise 1/3 octave spectrum is shown by the full
line in Figure 15. . Included within this figure are two further sets
of spectral measurements with a silenced exhaust. The two silenced
systems were of the same design which is also sketched in Figure 15,
but the perforated bridges were omitted in one of them. The acoustic
performance predicted for the design, neglecting flow noise, is also
plotted in the figure.
The results for the two silenced systems demonstrate the value of
perforate bridges for suppressive flow noise. They also confirm that
flow noise levels of around 85 dBA can be expected if the bridges are
omitted as implied by the results in Figure 7. The performance of both
systems predicted by linear acoustic theory is the same if the flow noise
excitation by vortex shedding at the expansions is ignored. However
only in the case of the system with the perforate bridges can good
agreement be found with the predicted performance, since only with these
present have the non-linear acoustic regeneration effects been suppressed.
There is a significant discrepancy between predicted and measured
performance around 1600 Hz. The reason has not been established but
tailpipe resonance might be responsible, while it is worth noting that
this is just above the frequency when the first of the higher order
propagating modes will become cut on.
24
-------
The measured insertion loss for the system with bridges is
recorded in Figure 16. The information plotted here is raw data
with no attempt to account for modifications to the open pipe
measurements to allow for changes in flow conditions (e.g. temperature).
The cross-hatching indicates the range of variation that this can
produce, since careful measurements with difference acoustic loads
had already indicated that this engine behaved acoustically as a
constant volume velocity source at each mechanical load and speed.
The predicted insertion loss was in excess of 35 to 40 dB above 200 Hz,
while the measurements indicate two pronounced dips in the neighbourhood
of the 1250 Hz and the 4000 Hz, 1/3 octave bands. This result reinforces
the suggestion that acoustic energy propagation in the higher order
modes might be responsible for the discrepancy. Calculations show
that the first circumferential mode corresponds to about 1200 Hz in
the expansion chamber and 3000 Hz in the pipe with the flow conditions
in these components. These observations indicate that the predictions
of linear plane wave theory will only be reliable at frequencies below
those at which acoustic'energy will propagate in the higher order modes.
A proper understanding of higher order mode propagation with flow present
lies to the future.
3.4 Some observations concerning pressure measurements
The measurements of the sound energy radiated from the exhaust outlet
is a well established technique and should present few problems. The
interpretation of the results is also straightforward provided free field
conditions obtain for the experiments. The measurement of pressure
within the duct poses more severe experimental problems, since both incident
and reflected wave systems exist together producing standing waves.
A traditional approach to standing wave measurements is to employ
a traversing probe microphone. This is a laborious procedure and requires
great care if reliable observations are to be obtained. A full account
of the experimental problems appears in reference 1. Special care is
also required in the interpretation of the results of pressure traverses
near expansions, due ,to the non-acoustic potential fields that can
exist there, see for example Figure 11. Except for special research
situations this does not appear a satisfactory or practical technique for
25
-------
normal production test bed measurements for component evaluation.
An alternative is to employ wall pressure measurements, but these
may involve practical problems in the.evaluation of the information
obtained. There is no serious problem if there are no standing waves
or disturbed flow in the pipe, a condition that obtains with some trans-
mission loss measurements. However, with any practical system strong
standing waves will always be present. The problems created by their
existence can be overcome, if simultaneous records are obtained with
two pressure gauges. These should be sited on the wall with a
separation that is less than the wavelength of sound at the highest
frequency for which pressure measurements are required. Fast Fourier
transform technqiues can be employed to extract the amplitudes of the
positive and negative travelling components of the standing wave system
from these two signals. This procedure relies oh the assumption that
the waves are plane and that the pressure signals are wholly acoustic,
so may not be appropriate downstream of bends or other discontinuities
which introduce strong disturbances in the flow.
A third possibility is to make simultaneous observations of wall
pressure and particle velocity at the same duct position. Simultaneous
pressure and particle velocity measurements are particularly suitable
for direct application in matrix methods of system performance evaluation.
Velocity measurements in a hot gas flow are difficult but the new optical
techniques may offer practical possibilities. Intake system performance
evaluations have already been undertaken |l5| using wall pressure
measurements and velocity measurements made with hot wires. In this
case, though the temperature changes are relatively modest, they were
large enough to introduce difficulties with the hot wire calibration
and signal interpretation.. Though much more expensive, optical techniques
should be free of such difficulties.
4. DISCUSSION
The analysis and results presented here represent one approach to
improving the understanding of the acoustic behaviour of exhaust and inlet
system elements and how they interact. It has been shown that concepts
26
-------
based on linear acoustic modelling are applicable to the control of
intake and exhaust noise provided they are employed with an adequate
understanding of their limitations. Current knowledge and practical
experience confirms that linear acoustic modelling can define the
relationships that govern the interactions between system elements and
provide useful predictions of system performance. These facts have
been appreciated'by intake and exhaust system designers and manufacturers
for some time, see for example references [4) and |s|. At the same
time shortcomings with the approach have been experienced in cases where
there has been a failure to achieve the predicted insertion loss by a
substantial margin.
In reviewing progress in the study areas (a) to (f) listed in the
introduction, it has become clear that successful application of linear
plane wave acoustic techniques depends on
1) Taking due account of flow conditions, including temperature
gradients.
2) Employing appropriate boundary conditions, including end
corrections where required.
3) Recognising that the plane wave analysis is limited to those
frequencies below which significant acoustic energy propagation
can take place in higher order acoustic modes.
4) Appreciating the importance of correct packaging, in particular
the measures needed to maintain linear acoustic behaviour in
system elements and avoid excessive flow noise generation.
5) Taking care that pressure measurements are correctly performed,
processed and interpreted.
6) Recognising that insertion loss not transmission loss is
required for practical performance predictions.
In the light of all the existing evidence it appears unlikely that,
with the pressure amplitudes normally experienced in intake or exhaust
systems, any significant errors are introduced by employing linear
acoustic theory for noise control analysis. Non-linear behaviour is
however likely whenever uncontrolled flow separations occur, and also
appears to be exhibited by system elements employing absorbing materials.
The two approaches to linear system analysis that have been discussed
here are, in principle, equivalent to each other. The one described in
27
-------
detail presents the analysis in terms of incident and reflected pressure
waves or transmission line equations. (e.g. references |l|,|2|,|5|,|7|,
|8|,|9|, |lO|,|ll|,|l2|,|l5| ). An alternative is to present the
analysis in terms of transfer matrices relating to input and output
particle velocities and pressures for each element (e.g. references
|3|, |4|, ). It is recognised that other methods of analysis (e.g. |l4|),
have also been developed which can provide useful alternative approaches
for noise control analysis. These may be particularly relevant (e.g.
finite element methods) for providing new insight into the acoustic
behaviour of components for which linear acoustic analysis has so far
proved inadequate. A valid criterion by which each of the methods may
be judged is that they should be flexible and readily applicable to
practical situations and must provide reliable predictions of acoustic
performance.
Finally, an outstanding problem in the noise control analysis of
engine intake and exhaust systems lies in characterising the source.
Some results have been reported here, but these have been restricted to
• •
examples where the source characteristics appear to be independent
of the acoustic load. There is clear evidence that many other examples
exist where this is not the case. So that new developments in measure-
mentment and source analysis techniques are required to provide reliable
noise control predictions in such situations.
28
-------
REFERENCES
1. R.J. ALFREDSON and P.O.A.L. DAVIES 1970 Journal Sound and Vibration
Vol.13, 389 - 408. The radiation of sound from an engine exhaust.
2. R.J. ALFREDSON and P.O.A.L. DAVIES 1971 Journal Sound and Vibration
Vol.15, 175-197. Performance of exhaust silencer components.
3. M.L. MUNJAL 1975 Journal Sound and Vibration, Vol.39, 105 - 119.
Velocity ratio-cum-transfer matrix method for the evaluation of a
muffler with mean flow.
4. V.C. BYRNE and J.E. HART 1973 S.A.E. Paper No.730429. Systems approach
for the control of intake and exhaust noise.
5. M. AMANO; S. KAJIYA; T. NAKAKUBO 1977 I.Mech.E..London Conference
Paper No. C16/77- Performance predictions of'silencers for the internal
combustion engine.
6. H. LEVINE and J. SCHWINGER 1948 J.Phys.Rev. 73, 383. On the radiation
of sound from an unflanged circular pipe.
7. E. MYER and E.G. Neumann 1972 Physical Acoustics, Chapter 11.
(Academic Press).
8. P. HUNGER and G.M.L.. GLADWELL 1969 Journal Sound and Vibration, Vol.9
28 ~ 48. Acoustic wave propagation in a sheared fluid in a duct.
9. P.O.A.L. DAVIES 1973 Proc.I.M.A.S., London, Section 4, 59 - 61.
Exhaust system silencing.
10. A.J. CUMMINGS 1975 Journal Sound and Vibration, Vol.38, 149 - 155.
Sound transmission at sudden area expansions in circular ducts with
superimposed mean flow.
11. W.J. ADAMS 1975 I.S.V.R. Internal and Contract Reports, University
of Southampton.
12. A.J. CUMMINGS 1975 Journal of Sound and Vibration, Vol.41, 375 - 379.
Sound transmission in a folded annular duct.
13. C.J. MOORE 1977 Journal Fluid Mech. Vol.80, 321 - 368. The role of
shear layer instability waves in jet exhaust noise.
14. S.W. COATES and G.P. BLAIR 1974 S.A.E. Trans.84, 740173. Further
studies of noise characteristics of internal combustion engines.
15. A.T. HARCOMBE 1977 University of Southampton Honours Thesis. A study
of pressure waves in the intake duct of an internal combustion engine
using acoustic methods.
29
-------
x
D
E
OJ
o
CL.
UO
D
E
O)
o
o
o
(N
40
30
20-
M = 0.01
M = 0 .00
Predicted
plane waves
240
760
780 800
Frequency Hz
820
840
860
880
FIG. la. ACOUSTIC ENERGY TRANSPORT AT A CONTRACTION WITH SIDEBRANCH,
-------
50
C)
o
o
CM
40
30
20
10
Prediction
O Measurement
M
.1
Mach number in 3
.2
.3
FIG. l(b) ACOUSTIC ENERGY TRANSPORT AT A CONTRACTION WITH SIDEBRANCH.
-------
M = 0.010
0.034
0.08
0.126
Predicted plane waves
30\-
740
760
780
800 820
Frequency Hz
840
860
880
900
FIG. 2a.
-------
60
50
R 30
20
ro
o
o
og
10
Prediction
Measurement
_L
.1 .2
Mach number in |_U
M
FIG. 2(o}. ACOUSTIC ENERGY TRANSPORT AT AN EXPANSION WITH SIDE BRANCH
-------
.6
.5
.4
O)
I .2
o
0)
L-
. 1_
o
o
.1
10.0
4.0
.1 .2
Frequency parameter
\
.5
1
2al
t
2
i
i
i
i
a2
1.0
FIG. 3. END CORRECTION FOR EXPANSION OR CONTRACTION.
-------
40
x30
20
U)
Ln
CD
O
o
CN
• 10
L
1 1 7
l 1 . /
/ o
D3
•5 0
. 9
/I O
Measured
£
cf
Calculated
_L
J I III |
100
200
300
Frequency Hz
500 600 700 800 1000
18.4
47.6
D, =2.54
D 2 - 2 .85
D4 = 5.1
D = 7.55
FIG. 4. PERFORMANCE OF FOLDED CHAMBERS.
-------
60
50
o> 40
4>
E
o 30
CO
"D
20
Contour of constant
energy
x/d = 7
Velocity 160 ft/sec
(60 Hz bandwidth)
_L
JL
_L
40
100 200 400 1000
Frequency Hz
2000
4000
10,000 20,000
FIG. 5. SPECTRAL CONTENT OF FLOW NOISE
-------
-------
70
L/
/
/
60
00
oo
Q-
OO
50 h
v 285 ft/sec in line
A 285 ft/sec perforate c
• 200 ft/sec • c
A 285 ft/sec d
O 200 ft/sec '' d
\
1 2 3
Separation x/d
FIG. 7. FLOW NOISE REDUCTION BY PERFORATES.
-------
1.0
Q
.6
o
-------
150
140
130
CO
-D
D
£ 120
110
623 Hz
680 Hz
A 800 Hz
_L
_L
110
120 130
S.P.L. max 1
140
150
M
FIG. 9. FLOW ACOUSTIC COUPLING AT AN AREA EXPANSION.
-------
n
140
D
/ ,/
130
o/ A'
120
CD
"D
110
I
n
/
D 623 Hz
A 800 Hz
* 678 Hz
D'
= 2.5
M = 0.11
100
I
100
no
120
S.P.L.
130
1 dB
140
150
160
FIG. 10. SUPPRESSION OF FLOW-ACOUSTIC COUPLING BY A BRIDGE PERFORATE.
-------
I
3a
M
+
lOa
I
20a
iwt
Acoustic standing wave
Vortex travelling wave
—»-x
iwt
(b)
k 2 = 20/a ; 0 =-Tf/2 ; *= 0.04, x > 4a .
120
o
_D
X
CN
100
Observed
ooooo Predicted
lOa
FIG. 11 SOUND PRESSURE LEVEL AFTER A DUCT EXPANSION M =0.1 ,
2a = 25mm ; Uc = 0.63 Mco
- 1250Hz
42
-------
Ur
Fiqure
43
-------
120
100
80
60 -
20
measured open pipe
measured
predicted
silenced
100
frequency Hz
1000
28 .6mm dla .
J
T
7 L
.83m
open pipe
28.6mm dia.
J_
76mm dia .
7 L
.83r
305mm
expansion chamber
FIG. 13
SINGLE EXPANSION CHAMBER PERFORMANCE CONSTANT VOLUME
VELOCITY SOURCE LINEAR ACOUSTIC ANALYSIS
'••—52mm
-------
-40
-50
-60
U -70
>
_OJ
0)
CN
> ¥
a °-
< §
CO ^
CL CD
U
-3 -40
0)
-50
-60
-70
100 200 500
narrow band frequency Hz
1000
2000
I I I I I I L_
0.5 1 5
reduced wave number k(/tr
10
(b)
FIG. 14 OPEN PIPE MEASUREMENTS, FULL LOAD.
FOUR CYLINDER FOUR CYCLE PETROL ENGINE
2000 < N < 5250 rpm
45
-------
100
80 -
u
o
60 -
40
100
400 1600 6300
third octave filter band (Hz)
System with perforate bridges
omitting perforate bridges
(flow noise)
silenced, peak measured
* x, peak predicted
FIG. 15 AVERAGED MEASURED PERFORMANCE. TURBOCHARGED DIESEL AT 7.5m
FULL LOAD 8 SPEEDS 1000 < N < 2350
-------
40 -
30
CD
~D
o 20
D
c
10 -
0
J L_
100 400 1600
third octave filter band (Hz)
6300
FIG. 16 AVERAGE MEASURED INSERTION LOSS, FULL LOAD TORQUE 8 SPEEDS,
1000 < N < 2350 rpm . Hatching indicates range of variation.
-------
AUTOMOTIVE L'.XllAUSI' SYS'IKM EVALUATION
by
I). A. B laser, .1. V. Cliuiir,, and H. HLcklinj'.
Kill id Dvii.uiii c:-> Ki'.'iea ivh Department
Ci'iieral Motors Ki.':,e,irch Laboratories
Warn.'!!, Michigan 48090
To be1 presented to:
Surface Transportation Exhaust System Noise Syrnposium
Sponsored by the U.S. EnvironmenLal Protection Agency
and
(Conducted by the iCnvironiiK'ntal Protection Agency and
McDonnell Douglas AsLronaulirs Company at Chicago, Illinois
To be pub ILshed in:
Proceedings of Symposium
ABSTRACT
The results of several exhaust noise studies that have been performed at
the General Motors Research Laboratories are presented. The principal
contribution is a new transfer-function method of measuring the acoustic
characteristics of exhaust systems with flow. The method appears to provide,
for the first time, a means of making routine, test measurements over the
frequency range of interest without being too time consuming and without
the .need to use a computer system other than a laboratory type analyzer.
Other results presented in this paper relate the acoustical pressure in the
tail pipe to the radiated sound and indicate how exhaust noise is determined
by engine type and operating condition.
49
-------
INTRODUCTION
A few years ago, a program of exhaust noise research was initiated at the
CM Research laboratories that had as its goal an increased understanding of
exhaust system performance and of the mechanisms of noise generation in
exhaust systems. At the outset of the program it became apparent that,
.ill hough the acoustical theory ol silencer elements such as expansion
chambers, resonators and acoustically-absorbing linings was reasonably well
understood, the effect of gas flow, temperature, high-amplitude waves and
other important features of real exhaust systems was not. Also, it seemed
that there was little basic experimental information on exhaust system noise
and that suitable test methods were Jacking. It was decided, therefore,
to concentrate initially on experimental tests and test methodology before
proceeding to theoretical models and design methods.
This paper presents some of the results of this work. Several topics are
covered. First there is a short discussion of exhaust system noise as
determined by engine type and operating condition. Next some data are
presented on noise as it is radiated from the tail pipe. Finally a new
transfer function method of measuring the acoustic characteristics of
exhaust systems (such as reflection coefficients and transmission losses)
is described.
ENGINE KXHAUST SYSTEMS
Exhaust noise is determined by a complete system comprising the engine and
various exhaust components such as shown in Figure 1 which depicts a typical
automotive exhaust system. The components shown in Figure 1 include the
manifold, downpipes, catalytic convertor, silencers, resonators and tail
pipes. Exhaust system noise comprises tail pipe-radia-ted noise and shell-
radiated noise from the structural vibrations of the various components of
the exhaust. Both aspects have to be considered since occasionally they
are comparable in magnitude.
Exhaust noise is caused by the pressure pulsations emanating from the
exhaust valves of the engine. These pulsations are affected by the con-
figuration and the mode of operation of the valves as. well as by the operating
condition of the engine. To a gre.it extent the pulsations, which can
typically be of the order of 175 dB, are reflected back from the silencer
and resonator so that they are retained within the exhaust system and
attenuated through various dissipative mechanisms. Typically the pulsa-
tions are reduced by about 20 dB at the downstream side of the silencer
and^resonntor. These pulsations Interact with the structure of the exhaust
system and usually arc the primary cause of the shell-radiated noise from
the system.
Various types and sizes of engines are used to power ground transportation
vehicles, ranging from small ''(-cylinder spark ignition engines for compact
cars to large 20-cylinder Diesel engines for locomotives. Although the
exhaust system requirements may vary considerably over this range of
vehicles, the noise generated at the exhaust ports of the various engines
have several features in common.
50
-------
Figure 2 presents unsilenced exhaust noise data for a V-8 spark ignition
engine and an 8V-71 Diesel engine. Although the load and speed conditions
are not the same, both engines exhibit tonal noise below 1 kHz composed of
harmonics of the engine firing frequency, and broadband noise above 1 kHz
composed primarily of flow noise generated during the initial opening of
the exhaust valve. Since all engines create exhaust noise spectra similar
to those appearing in Figure 2, design of engine exhaust systems would
appear to be a.relatively straightforward task. However, complexities are
introduced by stringent space limitations in a vehicle, the extensive range
of operating conditions over which the designer has to limit the noise, and
the back-pressure requirements which are different for different engines.
A diesel engine operates unthrottlcd continuously and, hence, the -back
pressure at part load has a greater effect on engine performance than in a
spark ignition engine which is throttled at part load. A Diesel-engine
exhaust system must,- therefore, In general be designed to have a smaller
back pressure.
The effect of engine operating conditions on A-weighted exhaust noise is
shown in Figure 3. These data represent noise radiated from the tail pipe
of an unsilenced V-8 spark-igniIion engine throughout its complete
range of. operation. Although exhaust noise is known to increase in level
with increasing engine speed and with increasing load, these data show that
the level of exhaust noise is governed principally by the exhaust gas mass
flow rate. This is not too surprising since pressure pulsations are created
by the exhaust gas blow-down process during exhaust valve opening. It is
recognized that other engine parameters such as exhaust valve timing and
cam shape, exhaust manifolding, etc., can also affect exhaust noise; however,
once the engine is designed these parameters are fixed, thus the exhaust
noise level is set by the exhaust gas mass flow rate.
TAIL PIPE
The tail pipe opening plays an important role in the acoustic performance
of the engine exhaust system since it is at the tail pipe that a major
portion of the acoustic energy of the exhaust pressure pulses is radiated
as sound. There was considerable confusion concerning the details of this
radiation process until 1948, when Levine and Schwinger [I]* developed the
theory of the reflected wave from an unflanged circular pipe without flow.
Since then, several experimental studies .have been performed to determine
the effect of flow on the reflection process [2,3]. In this more recent
work the most significant result is probably that obtained by Alfredson and
Davies [2] who, by assuming monopole radiation from the pipe (i.e. equating
the energy of the plane wave in the pipe to the energy in the spherical
spreading wave outside the pipe), developed the following relation between
the amplitudes of the pressure p. of the plane wave inside the pipe to the
spherical wave pressure p outside the pipe,
* Numbers in brackets f] refer to References at the end of the report.
51
-------
Al. = 20 log --— = 20 Jog (^ ) + 10 log -j—y-
I'
o
- JO Jog [(J+M)2 - K2 d-M) ] (D
whore r if the radial distance from the end of the pipe, d is the pipe diameter,
(pc)i and (f>c)0 are the characteristic impedances inside and outside the pipe,
respectively, M is the Mach number of the flow in the pipe and R is the tail
pipe reflection coefficient. The three terms in equation (1) represent the
effects of area divergence, fluid properties or temperature, and acoustic energy
reflection and .convert ion, respectively on the radiation of sound from the tail
pipe. Apart from a few experiments in the original paper by Alfredson and
Uavies [2], little or no data has appeared in the literature to confirm the
validity of this equation. However, it appears to be a useful formulation and
it has been used in investigations at the CM Research Labs to study the un-
silenced radiation of acoustic energy from the tail pipes of different engine.
exhaust systems. Some of these data are discussed here.
Narrow band spectra of the exhaust jioise radiated from the pipe compared with
similar spectra for pressures at two locations within the pipe, one close to
the end and the other 1.385 m upstream, are shown in Figure 4. Far up the
pipe, the spectrum is seen to be dominated by low-frequency energy composed
primarily of harmonics of the engine firing frequency. Near the. end of the
pipe and in the outside noise, the dominance at lower frequencies is
somewhat reduced. Reflection at the end of the tail pipe and the nature
of the radiation process in the external sound field are responsible for
this chance.
An interesting observation from Figure 4 is that the shape of the spectrum
near the end of the tail pi
-------
Only frequencies for which the wave length is smaller than the jet region
will be strongly refracted, thus, low frequency sound is radiated rather
uniformly while high frequency sound is directed off the tail pipe axis.
This frequency splitting effect is illustrated by the frequency spectra in
Figure 6. Below 1 kHz the two spectra are very similar; however, above this
frequency, the sound pressure is nearly 10 dB higher off the tail pipe 'axis
(at 45°) than on the axis (at 0°). Since A-weighting makes-the level more
sensitive to higher frequencies, the A-weighted sound pressure level of
Figure 5 reflects this shift of high frequency sound off the tail pipe axis
while the linear level does not.
Radiation directivity patterns similar to the laboratory measurements of
Figure 5 have also been observed in noise tests of vehicles. For example,
the directivity of .sound measured 15 m (50') from the rear of a transit
coach is .shown in Figure 7. Although these latter data also contain
directivity peaks due to other sources of noise (such as engine block
noise, fan noise, etc.), the quiet region at the rear of the coach and the
secordary directivity lobes at + 45" are essentially caused by refraction of
the T oise radiated from the tail pipe opening.
A TRANSFER-FUNCTION TECHNIQUE FOR MEASURING THE
ACOUSTIC CHARACTERISTICS OF EXHAUST SYSTEMS WITH FLOW
For exhaust systems, it is important to have an efficient method of measuring
normal incidence acoustic properties, such as reflection coefficients,
transmission coefficients, acoustic impedances and transmission losses.
The sound has to be separated into incident and reflected components and
this can be a relatively difficult problem when the sound is being generated
continuously and standing waves are being formed in the exhaust system. Once
the separation has been achieved, however, into so-called right-running and
left-running waves, as depicted in Figure 8, all of the normal-incidence
acoustic properties ip an exhaust system can be determined.
The classical method of decomposing standing wave systems in ducts is the
standing-wave-ratio (SWR) method [4] in which a small microphone or
microphone probe is moved axially along the duct to measure the amplitude
and location of the acoustic pressure maxima and minima. From this informa-
tion, the reflection coefficient can be determined. The SWR method has
several disadvantages:
a.
The method requires acoustic excitation of the duct system at discrete
frequencies and, hence, is time consuming.
b. The microphone position must be known quite accurately to resolve the
phase of the reflected wave. This causes difficulty at high frequencies
c. The microphone must be moved at least a half-wavelength at each
frequency so that the microphone system has to be quite cumbersome
in order to make measurements at lower frequencies.
d. Measurements that have to bo made within a long duct section are
affected by dissipation at the duct walls.
53
-------
e. When there is flow in the duct, the flow noise generated by the
microphone system can completely mask the acoustic waves being
measured.
The'se disadvantages virtually eliminate the SWR method as a practical tool
for the routine testing of exhaust systems with flow.
Other less-cumbersome methods of separating incident and reflected waves
have been tried to avoid some of the difficulties.just cited. A direct
separation of incident and reflected sound can be achieved with the use of
broadband short-duration excitation pulses in a relatively long section of
duct [5,6]. Because of the length of duct needed, dissipation ptoblems
occur at the walls as mentioned in (d.) above. Also there is difficulty
in creating sufficient high-frequency content in the short-duration pulses
to overcome flow and/or background noise in the upper frequency range.
Another method of separating incident and reflected sound uses correlation
techniques with a discrete frequency excitation [7]. Two wall-mounted
microphones measure the standing-wave-amplitude and phase relative to a
common reference voltage and the cross-correlation between these measure-
ments is used to decompose the standing wave into incident and reflected
waves. Although the wall-mounted microphones reduce flow noise, the method
is essentially as time consuming as the SWR method
A broadband method is to be preferred, therefore, for practical testing
since, in general, discrete frequency methods appear to be too time
consuming. As we have seen, short pulses do not seem to work too well for
broadband excitation in a duct. This leaves random-noise excitation.
Random-noise excitation methods are now widely used in conjunction wit'h
Fourier analysis equipment, particularly in vibration analysis, and it would
obviously be beneficial if such powerful procedures could be applied to
exhaust noise testing. During the. past two years, a practical transfer-
function technique of this kind has, in fact, been developed at the GM Research
Laboratories for acoustic measurements in duct systems with flow. We would
like to present here a derivation of the method together with test data
relative to some known no-flow theoretical solutions. It should be noted
that this is not the only random-noise technique that has been proposed for
exhaust-noise testing. Seybert and Ross recently proposed such .a method [8].
However their procedure involves a mathematical formulation, based on
auto- and cross-spectra, rather than transfer functions, that cannot easily
be used in practical testing. The transfer-function method that we describe
here can provide an instantaneous readout of quantities such as reflection
coefficients and transmission losses over a reasonably broad frequency range-
using a two-channel laboratory analyzer.
Referring to the schematic diagram shown in Figure 8, consider two
arbitrary microphone locations 1 and 2 at the duct wall with a separation
distance s, in a uniform duct of finite length with flow from left to right
as indicated. The acoustic pressures measured by the microphones at these
locations may be expressed as the summation of right- and left-running
components as follows:
54
-------
PI = PI + PI. (2)
r £
and
Po = P9 + P, , (3.)
r £
where the subscripts 1 and 2 indicate the locations, and r and .? denote the
right- and left-running x-omponents of the pressure. The reflection co-
efficients R) and R2 at the two locations are defined as,
RX = F (PI }/F{PI } (4)
and
R7 = F {p }/F{p } , (5)
(U)
55
-------
liquations 2 to 11 are valid either for deterministic or random signals,
provided Fourier transforms exist* in the case of the random signal.
Generally., for a random signal, the frequency spectra, rather than the
Fourier transforms are estimated. In order for equations 2 to 11 to be
valid, it can be shown that the following requirement has to be satisfied,
i.e.,
q
F{P
F*(p
(12)
m = 1, 2
n = 1, 2
P = r, H
q = r, £
where the bar denotes an average value, and the asterisk indicates a complex
conjugate. liquation 12 is satisfied as long as the data segments among the
different sample records in the Finite Fourier Transform are mutually un-
correlated. This condition can be achieved by appropriately separating the
sample records.
The transfer functions associated with the right- and left-running pressures
can be expressed as,
12
-ik s
(13)
+ik£s
(14)
where s is the distance between the two microphones and
(15)
(16)
are the wave numbers corresponding to the right- and left-running wave com-
ponent's. In equations L5 and 16, the wave number k is defined as the
frequency divided by the speed of sound, while the Mach number M is the mean
flay velocity V divided by the speed of sound.
Strictly, the Fourier transform of a random signal does ,not exist because
a random' time-function is not absolutely integratable. The Fourier
transform referred to here is the finite Fourier transform used in
numerical computations.
56
-------
The values kr and k^ can be determined from the correlation function between
p and p,.. Thus H and H can be determined using k and k. together with
r ^£ r *
the known distance s. However, H^ 'n Aquation 11 is obtained directly from
the ratio of the cross-spectrum between p , p and the auto spectrum of p,,
i.e.,
H12 = (;1J/C;11 (17)
Using the quantities provided by equations J3 to J7, the reflection coefficient
R^ can be determined from' equation (1L). This computation is relatively
simple and can- be readily programmed into the analyzer to provide a direct
readout of the reflection coefficient.
Measurement Accuracy
The accuracy of the acoustic properties measured in a duct system with flow
by the transfer-function method is governed by many factors. The most
important of these are discussed briefly in this section.
As with all other acoustic measuremunts, the signal-to-noise ratio of the
acoustic signals with respect to the flow or background noise must be
sufficiently high. Also, for the frequency range of the measurements, the
dynamic range of the acoustic signals must be kept within^ the appropriate
ranges of the instrumentation to avoid excessive interference from instrument
noise.
The spacing of the mic^jphones must be chosen with several considerations
in mind. Microphones too cjosely spaced will create error due to the finite
size of the microphone's diaphragm since, theoretically, each microphone is
assumed to measure acoustic pressure at a point. Microphones spaced too far
apart will introduce excessive wall dissipation effects. At frequencies for
which the spacing is a half-wavelength of the sound, the two microphones
measure redundant portions of the standing wave and the reflection coefficient
calculated from equation 11 becomes indeterminant. Near these frequencies,
wall dissipation and statistical errors will become dominant in the reflection
coefficient calculation and large errors result. To reach a compromise among
these different factors the spacing should be at least a few diameters of the
microphone, no greater than a half-wavelength of sound at the maximum
frequency, and equal to an integral multiple of the speed of sound times
the time domain resolution of the ADC* unit. This latter requirement,
coupled with adequate temperature and flow velocity information, should
assure reasonably accurate computation of the functions 11 and H „ .
r t
The frequency range in which accurate measurements can be obtained has to
occur in the range where only plane waves propagate in the duct. For a
circular duct of diameter d, this implies that the frequency f has to he
less than 0.586 (c/d) while for a square duct of side d, it implies that
f has to be Jess than c/2d where c is the speed of sound.
AUC is the Analog-to-Digital Convertor unit of the Fourier Analyzer.
57
-------
The statistical error of the measurement is dependent on the coherence
between the two microphone signals and on the number of averages used in
the evaluation of the transfer function. To achieve equivalent accuracy,
a high coherence requires fewer averages than a low coherence. However,
the best approach is to repeat the tests using progressively more averages
until the final :esult is essentially unaffected by the number of averages.
C:i 1 Lbrat ion
The calibration of the microphone systems is accomplished by mounting two
microphones at a time in a plate that can be rigidly attached to the open
end of the duct. The two microphones can then be assumed to be exposed to
the same noise field,, and the transfer function measured in this configura-
tion represents the response (both in amplitude and phase) of one microphone
system relative to the other system.
If microphone ill (see Figure 8) is chosen as a reference, successive com-
parisons of each additional microphone system with the system of microphone //I
will result in measurement of the set of transfer functions [H,., , H._ , ... ]
c c
where the subscript c refers to the calibration configuration of the
microphones. This set of transfer functions is then used to correct measured
auto-spectra and transfer functions for microphone system response .according
to the following formulae:
'G
llcorrected
~ „]
22 corrected
[G33]corrected
[H
12corrected
[H13Jcorrected
t
_
22 12
13
"l2/H12
l3/H13
(18)
(19)
(20)
(21)
(22)
These corrected forms of the auto-spectra and transfer functions are used in
the c.ilculation of the reflection coefficients, transmission losses, and other
normal-incidence acoustic properties of a duct system.
Because microphone //I is chosen as the reference system,
58
-------
10
Instrumentation and Associated' Measurement Procedures
The instrumentation required to perform in-duct acoustic measurements using
the transfer function technique is shown schematically in Figure 9. A randora-
nuise generator is coupled through a power amplifier to an acoustic driver
unit and generates acoustic signals in the pipe. The inside diameter of the
pipe used in the experiments was 51 mm, and 6.35 mm (I/A") diameter Bruel and
Kjaer condenser microphones were mounted flush with the inside wall of this
pipe. For this pipe diameter, the upper limit of the frequency range in
which only plane waves propagate is 4 kHz.
For the reflection coefficient measurements, an axial spacing of 27 mm was
used between the two upstream microphones. Thus, according to equation 11,
the first indeterminant frequency occurs at 6.4 kHz which is above the frequency
range of interest in the m-easurements. A third microphone, mounted downstream
of the silencer, and an anechoic pipe termination are used in the transmission-
loss measurements. The anechoic termination, which consists of a long wedge
of acoustic fiberglas within a 51 mm diameter pipe, prevents the formation
of downstream reflected waves and thus permits measurement of transmitted
waves with only one microphone. If such a termination were not used, two
downstream microphones could be used in conjunction with the transfer-function
technique to decompose the downstream standing wave to determine the
transmission loss.
Amplified microphone signals were fed to an HP Merlin (Model //5420) Fourier
Analyzer for measurement of auto-spectra and transfer functions. These
measurements are stored on the digital tape unit built into the analyzer
and recalled for subsequent computations. The calibration transfer functions
were measured using pairs of microphones as described in the calibration
section and used to modify the auto-spectra and transfer functions according
to equations 18 through 22.
The function H , and H were computed by feeding Gaussian white noise
r i
voltages simultaneously to both input channels of the analyzer, time delaying
one channel by -krS/LC and k^s/w, respectively, (according to equations 13
and 14) and computing the transfer functions. The microphone spacing of
27 mm was chosen so that these time delays (both equal to 78 ys) were equal
to the time domain resolution of the analyzer's ADC unit for the 3.2 kHz
frequency range.
The computation of equation 11 was performed completely within the analyzer
unit. Therefore, spectra of acoustic parameters such as reflection co-
efficients, transmission losses, etc., could be displayed directly on the
analyzer's oscilloscope and/or an x-y plotter. The simplicity of the form
of equation 11 is a key feature of this technique since it' permits immediate
display of the measured acoustic parameter in the laboratory without resorting
to a pre-programmed digital computer.
59
-------
11
jF.xperimcntaI Results
Two experiments were conducted, during the two-week period that the HP
Merlin analyzer was available. Choice of the experiments was based on
available hardware and on ability to predict the results from known theory.
Open 1'ipe Termination: Reflection coefficients from an unflanged open
pipe termination were measured without flow using the transfer function
technique and are compared in Figure 10 with the theory of Levine and'
Schwinger [1]. As shown, the microphones were placed only 30. mm and 57 mm
from the end of the pipe to minimize wall dissipation effects. Since the
quantity calculated from equation 11 is complex, Figure 10 presents both the
magnitude and the phase angle of the reflection coefficient.
The agreement between experiment and theory' is seen to be. quite good through-
out the measured frequency range. At high frequencies, the experimentally
measured reflection coefficients tend to be lower in magnitude than the
theoretical values. This effect has been observed in previous measurements
using the correlation technique [7]. It is probably due to wall-dissipation
effects and the loss of acoustic energy through the walls of the pipe.
Inaccuracies also tend to be greater at higher frequencies due to errors
caused by the finite size of the microphones and to errors in the functions
H „ and H „ caused by the approximate values used for the speed of sound
r £
and microphone spacing. Typically, such errors vary linearly with frequency
and thus are more apparent at high frequencies. The excellent agreement
between theoretical and experimental reflection coefficient phase angles is
somewhat surprising. Usually errors arising from inaccuracies in spatial
resolution and the speed of sound create larger variations in phase than
in magnitude.
Expansion Chamber Silencer: Reflection coefficient and transmission loss
measurements were performed using the transfer function technique for the
expansion chamber silencer shown schematically in Figure 11. The inlet and
outlet pipes have a diameter of 51 mm and the chamber diameter, is 152 mm
giving an area expansion ratio of 9 to 1. The outlet pipe protrudes a
distance of 54 mm into the chamber. Tests were conducted with an anechoic
termination downstream of the silencer, as shown in Figure 9 and discussed
above in the section on instrumentation and associated measurement procedures.
Reflection coefficients measured for this silencer are shown In Figure 12.
Also shown arc theoretical calculations for the silencer using the methods
of Alfredson and Davies [9,10]. The magnitude of the measured reflection
coefficient is quite Low at frequencies for which the chamber length is a
multiple of half-wavelengths of sound. The greatest differences between
theory and experiment occur at these frequencies due to resonant energy
dissipation within the silencer. Similar losses at the entrance and exit
regions of the silencer prevent the reflection coefficient from being unity
at the off-resonance frequencies. It appears that these losses are under-
estimated in the theoretical calculations.
60
-------
12
Above 3 kHz, the experimental results fall far below the theoretical
prediction. This is believed to be due to the occurrence of the first
radial cross-mode within the silencer. The lowest frequency at which this
mode will propagate unattenuated in the chamber is given by [11]
f =
1.22 c = (1.22)(344 m/s)
d .152 m
= 2760 Hz.
(23)
The theoretical calculations do not account for the higher order modes. A
similar difference between prediction and experimental results lias been found
using the SWR method [12].
At low frequencies, the phase angle agrees very well with theory. As thex
frequency increases the measured phase angles gradually lead the theoretical
values more and more. This effect might be attributed to wave action
occurring at the entrance to the chamber which, at high frequencies, is
similar to a flanged open pipe termination. For the infinite flanged pipe,
a small end correction H' = 0.42 d must be added to the pipe length to predict
the phase of the reflected wave [13], and such an extension of the inlet pipe
length would greatly improve phase agreement between theory and experiment in
the present situation. In fact, the phase correction, AO, would approach the
value,
A9 = 2k£' = 0.04 f
(24)
at high frequencies. To illustrate this effect, a modified theoretical curve
for phase is shown in Figure 12 between 1.1 kHz and 2.7 kHz. As expected,
the infinite flanged pipe correction slightly overestimates the correction;
however, it does result in a better match with the measurements. Thus, the
comparison of measured results to theory not only serves to verify the
experimental technique but can be used to check and possibly to improve the
accuracy of the theory.
Transmission loss (TL) data for the expansion chamber silencer are presented
in Figure 13. These data are computed using measured reflection coefficients
;\ auto-spectra upstream and downstream of the silencer in the expression
TL = 10 log
10
C
11
33
(in dB)
(25)
where Cjj is the upstream auto-spectrum at the point of measurement of the
reflection coefficient R^, and 033 is the downstream auto-spectrum. This
expression assumes use of an anechoic termination downstream of the silencer.
The measurements are compared to theoretical predictions of transmission
Loss also using the methods of references 9 and 10. Quarter-wave resonances
over the length of the expansion chamber are responsible for the lobe structure
in the TL spectra that repeats approximately every 600 Hz. The large peak
near 1200 Hz is due to a quarter-wave resonance in the annular chamber region
61
-------
13
formed by the protrusion of the exit pipe into the chamber. The decrease
in the experimental TL data above 3 kHz is due to the occurrence of the
first radial cross-mode within the' chamber, as discussed earlier for the
reflection coefficient data.
The overall trend of the experimental TL data follows that of the theoretical
prediction. However, the measured data exhibit fluctuations throughout the
frequency range which are not accounted for hy the theory. Although the
origin of these fluctuations is not known, reflected waves from the anechoic
termination are suspected. If reflections were present downstream of the
silencer, then 033 used in equation 25 would be in error due to the standing
wave patterns. The associated error in TL would be of the fluctuating nature
similar to the data of Figure 13 due to the presence of pressure nodes and
antinodes at the downstream microphone location. A study of the acoustic
characteristics of the anechoic termination section and of other silencers
will be conducted in future tests.
CONCLUDING COMMENTS
In this paper we have presented results that we hoped would be of particular
interest at this Symposium. The mechanism of the radiation of sound from the
end of a tail pipe is an important topic in exhaust noise studies and the
possibility that the acoustic pressure in the pipe may be directly related to
the radiated sound should be further investigated. The transfer-function
technique developed by CM Research Laboratories appears to provide, for the
first time,- the means of making routine measurements of the acoustic charac-
teristics of exhaust systems with flow. We feel that this capability should
be of considerable -use both in exhaust system development and for possible
exhaust system evaluation purposes.
Whether transmission loss data obtained in bench tests with the transfer-
function technique described here "can be used to predict the performance of
silencers as installed in vehicles has not been investigated yet at the
GM Research Laboratories. Such an investigation should involve consideration
of the following -effects in order to determine whether or not the effects are
accounted for and, if not, what corrections are required:
1. Mean flow
2. Temperature and temperature gradients
3. Finite amplitude waves
4. Engine source impedance and tail pipe radiation impedance.
Since the transfer function technique can be used with mean flow, that effect
could be accounted for directly in any bench test using the technique. As far
as the two-part temperature effect is concerned, bench testing at room tem-
perature would introduce a reduction in the speed of sound from that for the
actual higher temperatures, and this effect could be accounted for by a
relatively straightforward frequency correction of the transmission loss data.
The effect of temperature gradients in the exhaust system, however is not
currently understood and thus the effect on silencer performance Is -not
62
-------
14
predictable by ;m_y presently known method. As indicated In references 2,
9, 10 and 14., nonlinear effects due to finite amplitude waves in expansion-
chamber .silencers occur near resonant ^frequencies and, hence, can usually
be neglected for design purposes. Il-ecause bench-test transmission loss
data do not include the effect of the engine and tail pipe impedances,
they cannot be used directly to predict either the level of noise radiated
from the tail pip'e or the decrease in noise level due to the insertion of
the silencer. This Jailer measurement of silencer performance, which is
termed the insertion loss, can be related to transmission loss if the engine
and tail pipe impedances are known. Several workers have attempted to specify
these impedances using experimental measurements [2,15]. However, their
results were not sufficiently general to cover the complete range of conditions
that exist in vehicle exhaust syste ;.
In summary, therefore, sufficient data are not yet available to correlate
bench test transmission loss of silencers with noise reductions obtained
when these silencers are installed in vehicle exhaust systems. Thus before
bench tests can be used tp develop silencer ratings, a series of silencers
should be bench tested and also should be evaluated on vehicles to determine
the degree of correlation. If such a correlation can be established, a
frequency-dependent criterion (similar in nature to noise criteria curves
used in architectural design) could perhaps be developed to determine a
silencer rating from transmission loss data obtained in bench tests.
63
-------
15
REFERENCES
1. Eevine, H. and Schwinger, J., "On the Radiation of Sound From an
Unflanged Circular Pipe," Physics Review, Vol.,73, No. 4, pp. 383-406,
February 15, -1946.
2. Alfredson, R. J., and Uavies, P.O.A.L., "The Radiation of Sound From an
Engine Exhaust," J. Sound and Vibration, Vol. 13, No. 4, December 1970,
pp. 389-408.
3. Ingard, K. U., and Singhal, V. K. , "Effect of Flow on the Acoustic
Resonances of an Open-Ended Duct," J. Acoustical Society of America,
Vol. 58, No. 4, pp. 788-793, October 1975.
4. Beranek, L., Acoustic Measurements, chap. 7, pp. 302-361, J. Wiley,
1949.
5. Catley, W. S. and Cohen, R., "Methods for Evaluating the Performance of
Small Acoustic Filters," J. Acoustical Society of America, 4_6_, pp. 6-16,
1969.
6. Singh, R. and Katra, T., "On the Dynamic Analysis and Evaluation of
Compressor Mufflers," Proceedings 1976 Purdue Compressor Technology
Conference," July 6-9, 1976, Purdue University, Wes't Lafayette, Indiana,
1976.
7- Schmidt, W. E.-, and Johnston, J. P., "Measurement of Acoustic Reflections
from Obstructions in a Pipe with Flow," NSF Report PD-20, March, 1975.
8. Seybert, A. F., and Ross, D. F., "Experimental Determination of Acoustic
Properties Using a Two-M\crophone Random-Excitation Technique,"
J. Acoust. Soc. Am., Vol. 61, No. 5, May 1977, pp. 1362-1370.
9. Alfredson, R. J., "The Design and Optimization of Exhaust Silencers,"
Ph.D. Thesis, Institute of Sound and Vibration Research, University of
Southampton, July, 1970.
10. Alfredson, R. J., and Davies, P.O.A.L., "Performance of Exhaust Silencer
Components," J. Sound and Vibration, Vol. 15, No. 2, March, 1971,
pp. 175-196.
11. Harris, C. M., Handbook of Noise Control, McGraw-Hill Book company,
1957, p. 21-16."
12. Davis, 1). I)., Stokes, C. M. , Moore, I)., and Sevens, G. L. Jr.,
"Theoretical and Experimental Investigation of Mufflers with Comments
on Engine-Exhaust Muffler Design," NACA Report 1192, 1954.
13. Beranek, L. L., Acoust ics_, McGraw-Hill Book Company. ,1954, p. 132.
14. Sacks, M. P., and Allen, I). L., "Effects of High Intensity Sound on
Muffler Element Performance," J. Acoustical Society of. America, Vol. 52,
No. } (part 1), 1972, pp. 725-731.
15,
Galaitsis, A. G., and Bender, E. K., "Measurement of the Acoustic Impedance
of an Internal Combustion Engine," J. Acoustical Societyof America,
Vol. 58 (supplement no. 1), Kali 1975.
64
-------
Resonators
Tailpipes—K
Manifold
Catalytic
Converter
Silencers
Figure 1. Components of an Engine Exhaust System.
-------
120
1
2 3
Frequency, kHz
4
4000 rpm \ c .
( o-i-
3000 rpm > 350-V8
2000 rpm ) 50% Load
8V-71 Diesel
1000 rpm (No Load)
5
Figure 2. Radiated Exhaust Noise fron Two Unsilenced Engines (measured
1.5 m from the end and 45° off the axis of the tailpipe).
-------
18
llOr
100
CQ
•O
90
ZD
tO
tn
80
Q
ID
o
70
D
ENGINE
RPM
olOOO
Q1500
^2000
02500
Q3500
04000
r
0
100 200 300 400 500
EXHAUST GAS MASS FLOW RATE, kg/h
600
Figure 3. Radiated Exhaust Noise Versus Exhaust Gas Mass
Flow Rate for-a V-8 Engine.
67
-------
19
180
160
a.
Lf)
b
*140
0)
CD
T3
*
15
I
o>
3
v>
(0
CD
TJ
C
120
100
i In-Pipe Pressure
( 1.385 m from End
I In-Pipe Pressure
( 19 mm from End
Radiated Noise
\ at 1.5 m and
' 45°
2 3
Frequency, kHz
Figure 4. Conparison of Pressure Spectra in the Exhaust Pipe
to Radiated Noise Spectra.
68
-------
20
m
Q.
LD
i
O
CM
0)
l_
co
105
100
20 40 60 80
Angular Position, Degrees
100
Figure 5. Linear and A-Weighted Exhaust Noise Directivity Pleasured
1.5 m frogi the End of the Tailpipe.
69
-------
21
2 3
Frequency, kHz
•Figure 6. Exhaust Noise Spectra at 0° and 45° from the Tailpipe Axis
and 1.5 m from the End of the Tailpipe.
70
-------
RPM
02000
O1750
A1500
01250
O1000
ENGINE
COMPARTMENT
to
to
85
Figure 7. Directivity of Noise from a Transit Coach.
-------
23
Microphone
Location
#1
u
/ Flow
/
P-
P
?
V
h
S
c-
/ Right-Running \
\ Waves >
1 Left-Running 1
( Waves f
/
P
P
/
2r
2e
X^ Microphone
Location
#2
/
P
Figure 8. Microphone Configuration and Notation.
72
-------
Acoustic
Driver
Anechoic Termination
-Oscilloscope
HP-5420
Fourier
Analyzer
I, i
Digital
/ Tape
Cartridge
X-Y Plotter
ho
-P-
Figure 9. Test Arrangement and Instrumentation.
-------
25
800
1600
2400
3200
180
90
O)
-------
26
r51
Dia.
Sound
Source
' U
Mic. V
=r1
u
\Mic.
if2
•• ouu - ....... - •-
i
152
i
— H 54
_ ,-61
Dia.
t
Dia
D
Anechoic
Termination
Mic.
#3
Figiore 11. Expansion Qiaittoer Silencer (dimensions in ran) .
75
-------
27
1.0
.75
.50
.25
800
1600
2400
3200
180
-180
800 1600 2400
Frequency, Hz
3200
Figure 12. Magnitude and Phase Angle of the Reflection Coefficient for the Expansion
Chamber Silencer; theory, modified theory, experiment.
76
-------
28
CO
T3
c
o
800
1600
Frequency, Hz
2400
3200
Figure 13. Transmission Loss for the Expansion Chamber Silencer;
theory, experiment.
77
-------
A METHOD OF MEASURING EXHAUST SYSTEM NOISE
Mineichi Inagawa
Trucks & Buses Engineering Center
Mitsubishi Motors Co.
In Japan, noise regulation for motor vehicles is on the verge of
becoming the strictest in the world. The noise level of heavy duty
trucks and buses will be limited to under 86 dB(A) from the present
89 dB(A) by the ISO method by 1979.
Figure 1 shows the contribution of each sound source to the total
noise level of Japanese heavy duty trucks and buses measured by ISO R362
method. The engines of the illustrated vehicles have from 250 to 300
horsepower outputs. Engine noise is responsible for the greatest
percentage of exterior noise. Exhaust system noise, and cooling fan
noise come next in order.
Our bench test on mufflers can be classified into four types.
(1) Measurement of Acoustic Attenuation of a Muffler,
(2) Measurement of Flow-Generated Noise of a fluffier,
(3) Exhaust Noise Test on a Stationary Vehicle, and
(4) Exhaust Noise Test on an Engine Bench.
(1) Measurement of Acoustic Attenuation of a Muffler
The setup, of the measur.ing system is shown in Figure 2, The
output noise is measured in a cubic anechoic test chamber. Its dimensions
are 2.5 meters or 7.5 feet on all sides.
Input sound pressure to a muffler is controlled constant at
110 dB(A), and as a noise source, sinusoidal wave, v/hite noise and
taped spectrum from the exhaust of an engine are used.
Obtained data is recorded and, post-processed by a computer.
Figure 3 shows our way of expressing "Acoustic Attenuation".
The difference of noise level between the reference straight pipe which
is referred to as the "Base Mode-1", and the tested muffler is designated
as "Acoustic Attenuation".
An example.of frequency response of the "Base Model" is shown in
Figure 4 in order to compare the fundamental elements of mufflers.
In this case, the equivalent length is 175 millimeters or 6.9 inches.
79
-------
The fundamental elements configuration and their parameters are
illustrated here (Ref. Figure 5). Though the expansion chamber type
and the resonator type muffler seem to be the most popular, the multi-
hole type is v/idely used and reveals an interesting feature which I
will mention later.
Figure 6 shows acoustic attenuation which I mentioned earlier
in relation to sinusoidal wave. He have shown an expansion chamber
type here as an example. This example is a very simple one-chamber
model. In this case, it is meaningless to illustrate the measurements
and calculations of frequencies above 2000 Hertz.
Figure 7 shows one response of the resonator type muffler. As
the number of holes is increased, its features begin to resemble
those of the expansion chamber type.
Next is shown an example of a response using white noise to
compare with that of sinusoidal wave. This comparison is made with
the multi-hole type muffler (Ref. Figure 8).
The attenuation characteristics using sinusoidal wave are
represented by the dotted line and those of the 1/3 octave band using
white noise are shown by the dots. In such simple models as this one,
the 1/3 octave band noise is sufficient to illustrate the acoustic
features of the muffler. When white noise is the input, an attenuation
at frequencies beyond 2 kHz, and overall, are obtained.
Figure 9 shows the attenuation characteristics of an actual muffler
for a vehicle. All the mufflers have a diameter of 280 mm. and are
1 meter in length. Frequencies of above 2000 Hz are best attenuated
by type C. The A-scale level also shows the best results. Figure 10
shows an example of acoustic attenuation with respect to a twin muffler.
When the actual exhaust noise of the engine is used instead of white
noise, the spectrum poses a problem. Figure 11 is the spectrum of
the exhaust noise from a V8 14.8 liter diesel engine without a muffler.
This 2400 rpm spectrum resembles that of white noise and this was
used as the sound source.
The acoustic attenuation of the noise of a muffler with white
noise input and the noise of a muffler with actual engine exhaust noise
input were compared using overall dB(A). The difference in acoustic
attenuation due to the difference in input spectrum was slight and
good correlation was seen. Accordingly, we decided to use white noise
input for acoustic attenuation studies.
(2) Measurement of Flow-Generated Noise of a Muffler
As a flow source, we used a rotary blower and a normal air flow
was supplied to the test muffler through a silencer. The flow-
generated noise was measured using a cubic anechoic test chamber.
The rotary blower used, had a flow volume of 54 m /min at 200 mmHq
in order to simulate the exhaust gas flow at full load of a 300 horse-
power class diesel engine which we manufacture. (Ref. Figure 13)
80
-------
Using this equipment, we tested various mufflers to obtain
their flow-generated noise levels. (Ref. Figure 14)
The change in noise levels according to the differences in flow
speed were as follows;
When flow speed is less than 50 m/s, noise level is proportional
to the value of V to the fourth power, where V represents the flow
speed.
When flow speed is less than 100 m/s, noise level is proportional
to the value of V to the sixth power, and when flow speed is more than
100 m/s, noise level is proportional to the value of V to the eighth
power or more.
We discovered the following tendency when testing the fundamental
elements of the muffler (Ref. Figure 15). The flow-generated noise
shov/ed a tendency to be higher in the expansion chamber type and the
multi-hole type muffler.
I would like to show typical examples of the spectra. Two tendencies
were observed. (Ref. Figure 16). First, as the amount of flow increases,
the dominant frequency was seen to rise to the higher range and at
the same time noise level is increased.
In the case of multi-hole type mufflers the noise level gradually
increased, and as you can see in the figure the dominant frequency
is above 2 kHz.
The flow-generated noise level was evaluated the same as acoustic
attenuation using differences of the levels of the test mufflers
based on the straight pipe. (Ref. Figure 17)
We tested a typical muffler and found that in mufflers which
do not produce a whistling noise the flow-generated noise level
remained constant when the amount of flow exceeded a certain limit.
(Ref. Figure 18)
Next, the correlation between the data obtained using the flow-
generating equipment and exhaust noise of the actual vehicle depends
on the correspondence of air-flow. The effect of engine rpm and the
temperature of the exhaust system was studied using testing equipment
for the exhaust noise of stationary vehicles. I will mention this
later. (Ref. Figure 19)
The difference in temperature betv/een the inlet and outlet of
the exhaust system is from 200 to 300 degrees centigrade, and when
the back pressure of this flow-generated noise and that of the actual
vehicle are compared, it, was found that better correlation is seen
when the rate of flow is converted at the outlet temperature of the
tail pipe.
81
-------
From this result, engine rpm and the exhaust flow rate can be
approximately related as shown in Figure 20.
When correspondence is made at the outlet temperature of the
exhaust system, the actual exhaust noise and the flow-generated noise
of the vehicle, when compared in the same muffler, is as shown in
Figure 21. And in this case, the flow-generated noise accounts for
only a small percentage of overall exhaust noise.
And also from our experience, if the muffler is normal and does
not produce any whistling noise, it can be said at present that
flow-generated noise contributes only slightly to overall exhaust
system noise.
(3) Exhaust Noise Test on a Stationary Vehicle
Figure 22 is the layout of the testing equipment.
The base of this testing equipment is a heavy-duty truck of a
maximum payload of 11 tons, equipped with a 305 horse power V-8
diesel engine.
An Eddy Dynamometer v/as mounted on the rear body of the truck
and connected to the engine through a transfer to absorb the engine
output and also for automatic speed control of the engine.
For this test, the exhaust system was mounted at the side of
the vehicle and a sound insulating wall was set to avoid the influence
of engine noise and other noise from the vehicle. By using this
apparatus, radiated noise from the exhaust system can also be easily
evaluated.
Figure 23 shows the changes in the exhaust noise with respect
to its temperature. The engine was operated at the speed of its
maximum output, and the level of exhaust noise which is represented
by "NL3" in this figure, goes up as the temperature rises while the
level of radiated noise goes down.
The change in the spectrum is shown in Figure 24. For the exhaust
noise, the spectrum below 2000 Hertz tends to rise as the
temperature rises. And for radiated noise, the spectrum above
1 kHz tends to decrease as the temperature rises.
The Figure 25 shows a muffler which was shown earlier. This
figure shows the relationship between the exhaust noise and the
back pressure when different arrangements of pipes, tail pipes, and
sub-muffler were applied to the muffler shown earlier. From this
result, you can see that a difference of a few dB(A) is seen when
the exhaust pipe and tail pipe are arranged differently.
82
-------
The relationship between the back pressure and the attenuation
is inversely proportional. When one increases the other decreases,
and the quickest (fastest) way to achieve sufficient attenuation
without raising the back pressure is to carefully add another
muffler.
The relationship between the attenuation of stationary vehicles
and acoustic attenuation which was mentioned before is shown in
Figure 26.
The solid line shows a one-to-one correlation ratio and as
you can see, there is bad correlation between acoustic attenuation
by white noise and the attenuation by using the engine of the vehicle.
The attenuation on the vehicle is much greater.
When this is compared using the spectrum it can be expressed
as the following. (Ref. Figure 27) In the attenuation spectrum
obtained from the engine, attenuation above 2 kHz tends to increase
compared to the acoustic test and on the contrary, the attenuation
of the spectrum near 500 Hz tend to be much lower. At present,
we have not been able to explain the causes for these phenomena.
And this will be the object of further study.
(4) Exhaust Noise Test on an Engine Bench,
The method of measuring the exhaust noise in engine bench test
is specified by the Japan Industrial Standard D1616. (Ref. Figure 28).
This standard specifies only the microphone location and the running
conditions of the engine, but we have also considered the length of
the exhaust pipe and the tail pipe. Also some measures should be
taken to avo'id the influence of radiated noise from the exhaust system.
"La" must be equal to the length of the exhaust pipe of the actual
vehicle, and also "Lb" must be equal to the length of the tail pipe
of the actual vehicle.
The microphone is set at an angle of 45 degrees and a position
of 50 centimeters with respect to the exhaust pipe axis.
The engine bench test, in essence, is the same as the bench
test of the stationary vehicle which was mentioned before so the
correlation between these two tests were not checked.
Figure 29 shows the relationship between the attenuation and
the back pressure of different engines and a variation of mufflers
on the bench test. The figure on the right shows the amount of noise
attenuation and the figure on the left shows the back pressure.
They both show good correlation.
83
-------
The engines compared here are the V8 pre-combustion chamber
type with a volume of 13.27 liters and maximum output of 265 horse-
power and the V8 direct-injection type with a volume of 14.8 liters
and maximum output of 305 horse-power. Me regret that we did not
make any comparison with the in-line 6 cylinder type.
Next, Figure 20 shows the relationship between the exhaust
noise of the engine bench test and that of the actual vehicle.
And in this case, the relationship changes greatly depending upon the
ratio of the exhaust noise to the various other noise of the vehicle.
For this vehicle the amount of exhaust noise on the right side of the
vehicle is about 30 percent. The upper line shows the acceleration
noise measured by ISO method when the microphone was set at 3 meters
from the center of the vehicle, and the lower line when the microphone
was set at 7.5 m from the center of the vehicle.
We have drawn the conclusion that the most practical method of
measuring the noise from the exhaust system is to use the engine bench.
However, sufficient consideration must be given to the length of the
exhaust pipe and the tail pipe, also it is necessary to consider the
influence of radiated noise, and to estimate the level of back pressure.
84
-------
o
"2
DUMP
THUuK
.m p
TRACTOR
CARGO
TRUCK
BUS
:IG-I lt£_.CGNn
-------
F I G"5 Fundamental elements configaratioiis'and their parameters of exhaust system
Type
Resonator type
Multi-holes type
Exhaust pipe
Tail pipe
r type
e
Parameter
L
1 -,
i-oi
fl
;
n
Dp
It
L
HF
n
SepaiLitor
Shape
=±-- — —...., -T'-° - '
; , = - - '- -
1 i__
— _ — I. — •. _ _
J IT!-'. L
1 ''!
- 'i- ni Number of
il Number of holea
1- .-'--, ..
'installed ; HolB '-;—
or not Jianu-tcr - -
Ellipse
type °
Square j— )
type
e-
( )j Expansion ratio
ic"'
Frequency ,'HI)
FIG"6 Attanuation of the expansion
chamber type mufflers
Number of resonant hoJes
.n o
20
0
10
ro
o
ro
10- 10 10'
Frequency mi)
FlG~"7 Effect of number of
resonant hole?
86
-------
t'.o liole
Sinuso idal
-i-i-.i,— Djnd noise .
20 SO ICC 200 bOO . > 21 5k !0k :0« A Lin
Frequency m>
Attenuation. (dB)
DO
60
40
10
0
1 f.o hole '
_^U_,^r- -~ Sinusoidal
: *_';i-.-* ;'.; ^ 1/3 Oct.
| U'j ^ p md noise .
-v '
ro 50 100 -'oo '.oc i.
Frequency IH;I
p|(3~8 Effect of noiao source to the
attenuation characteriatica
of mufflers
1 3oci lljml nuisc
20 50" K-) :;// 530 i. :
l-'rcqucncy (h:)
A Lin
60
20
'. ?ul'_ __) V/V^"
Hund noise I ^ ,
20 50 100 200"" iOO Ik 2k Ok 10k ,20k
frequency (Hz)
ALm
_ • _ L— _ • _ . _ , _
20 50 100 200 500 Ik 2k ik 10k 20k
< Frequency (Hz)
FIG>~9 Acoustic attenuation of
various test muffler^
A Lin
SUB-MUFFLER MAIN-MUFFLER
240^) (280^;
^$5 50 TOO 200 500 ik 2K"
y± OCTAVE BAND CENTER FREQ. (HZj
FIG-10 EXAMPLE OF ACOUSTiC ATTENUATION Of MUFFLER
87
-------
10" A L
OCTAVE BAND CENTER FTCEQ. (Hz)
FIG-11 THE SPECTRUM OF EXHAUST NOiSE EMITTED FROM ENGINE MANIFOLD
FK3-I2 ACOUSTIC ATTENUATION OF MUFFLER
25
£20
10
. 0/9
< 0 5 -10 15 20 25
ATT. (dSAj-WHITE NOiSt INPUT SPECTRUW
Bypass valve Silencer T(...
Roury ^Jj. Swirl Tc5t
'
Microphon ,. ,
" Back pressure
controller
r
V r Late \ jKu
duct
FIG "13 txpcriincnljl layout for flow ycneraleci
130r
•120
110
90
80
70
60 -
in u f Her
straii'lit
pipe
5 10 20 30 =0
Q (m3 mm)
FIG-14
30 50 70 "iOO , 200
i (m s)
Flow pern-rated noise level muter various
luul lU'rs
88
-------
"50 100 " i50
Back pressure (ninillt;)
F I G ~1 5 Relation of flow generated noise and back
pressure of typical muffler elements
?0 20Q
Frequency {HZ
«'*,
Frequency ;„.)
(l) Spectrum of expansion ':"sP"trum of Multi-Holes
chamber type ^P6 with separator
r \(j~ ID Change of flow generated noise spectra
by air flow rate
FiG-1? FLOW GENERATED NOiSE
(1) BASE MODEL
FLOW „ L
Q IT
,2i MEASURE
NL
NL
(3) RESULT
(B)-(A)
MlC
DATA
260'
iB)-(A)
50
< 40
o
1
1
1
^
c
734
•~^£-ENE-
^x3lSE..
RACK
PRESS..
llOO
5 10 15 20 25 30
FLOW RATE (1^niB)
FIG-18 EXAMPLE .OF FLOW GENERATED NOISE
1000
-N
- 800
c 600
D
C
J 400
L
U
300
EXHAUST
S MANIFOLD
J OUTLET
TAIL PIPE
OUTLET
ENGINE: V8, 14.886
FULL LOAD
500 1000 1500 2000 2500
ENGINE SPEED
89
-------
soo'c
/ 500*C /
1 . 9
/TOO C ,r
/ x -400" *C
TAIL PIPE
QUTLEI
20 t
0 500 TOOO 1500 2000 2500
ENGINE SPEED
FI&-20 ..RELATION BETWEEN ENGINE SPEED
_AND EXHAUST GAS FLOW
130
120
, 110
[
i
' 100
! -90
I
I
' 70
6O(-
50
EXHAUST NOISE FROM VEHICLE
FLOW GENERATED NQISE
\ STRAIGHT PiPE
MUFFLER
MUFFLER.
.A STRAIGHT PIPE
0 5OO 10OO 15OO 2000 2500
ENGINE SPEED
FIG-21 COMPARISON OF EXHAUST NOISE
AND FLOW":"GENERATED'NOISE,
CONTROLLER
MUFFLER
.SOUND .. ^_
i INSULATING WALL,
GOOLJNG
WATER
\ TRUCK.
FIG-22IUAGRAM OF E>HAUST NOiSE MEASURIN
-iNSJHJMENT..ON A STATiONARY
80
Nl.
prc- 7'(Tfmp.)
niufflcr QM:iin m-ifflcr
Engine o 0 0 'D.vt,
•Vi, J/' ,\i,
Engine 2500 rpm full load
a
«
c
00
•V
O
30 3
' 'C7~
0 100 JOO 300 4CC bOu SCO *»
Exhausi gas temp. (°c )
Influence of cxluust gas i(;ai|ieratnre on
exhaust system noise
90.
-------
no
too
90
eo
(Ij Lxliaust noise spectra
j I
High-temp. (
O 80
(2) Premuffler radiated noise
^ .o-«-\
Low-temp, ' t
(3) Maun muffler i;idiali:d'noi«:
/V
To 3 fT
Frequency fllz)
FIG ~2^- Influence of c\!i:iusl ^a.1. temperature on
cxliuust noise spectra
Straight pipe
I
Resonator |
n--r_:zk--
Multi-holes
Expansion
chum.ber
o
(Drum can) '
Premufller .\!am !
nl 1~ mufller-
-(2 2 K.ldsas
much as
0 vt ix :iO .00 ?so
Bjck pressure (inmilL:)
FIG~25 Relation of cxhausl noise re duel ion and
back pressure of various exhaust system
arrangements
ENGINE I 8DC 4
(265 PS)
ENGINE : 80C 8
C305PS)
10 15 20 25 30 35
ACOUSTIC ATTENUATION CdBA)
(BY WHITE NOiSE)
FULL LOAD AT I
20 50 1CO 200 500 1000 2K
FREQUENCY ;HZ)
FIG-27 CCAI11--.\.SCN ?.F
1GK 2CK A
SPECI-.M
RELATION SHiP BETWEEN ACOUSTiC ATTENUATION q:
AND ATTENUATION OF STATIONARY VEHICLE *
-------
U : EQUAL LENGTH TO VEHICLE
EXHAUST PiPE
Lb : EQUAL LENGTH TO VEHICLE
TAiL PIPE
\
SOUND
INSULATING
WALL
FIG-28 DIAGRAM OF EXHAUST NOISE TEST
.ON A ENGINE BENCH.
£ BACK PRESSURE
o 150
Q
CO
e
1/1
tf
Q_
5
100
50
50
o
-20
z
o
s
• 101—
100 10
NOISE ATTENUATION
8DC2 I V8 13.271
8DC8 : V8 14.881
20
30
BACK PRESS', ("""^l
-------
A Study on the Reduction of the Exhausi
Noise of Lap'e Trucks
By Tomoyuki 111 RANG"
KaisuTOlDA1*
Toslniiiiisii SAlTO ***
Mineichi JNAGA\VA""
KooNAKAMURA'•'••*
Summary
Traffic noise in urban areas is posing a serious problem in many countries of the world
and the reduction of the vehicle noise of large trucks is now a social problem requiring
immediate solution in our country.
To cope with the social circumstances, four major large truck manufacturers have been
conducting a joint research on the reduction of the noise of large trucks under the leudcrilurj
of the Ministry of International Trade and Industry us a three-year project. Mitsubishi Heavy
Industries is in charge of the reduction of exhaust noise which is one of the main sources of
vehicle noise.
The exhaust noise of trucks can be divided'into discharge noise emitted from the exhaust
outlet and radiated noise emanated from ihe surfaces of the exhaust pipes and mut'fleis.
This paper reports on the results of our experiments made on the reduction of the exliausi
noise of actual trucks on the basis of the results of-our basic studies including acoustic study
and studies on air flow noise and radiated noise.
1. INTRODUCTION
The worldwide problem of reducing city traffic noise has increasingly drawn the attention of many
countries. We, in Japan, are also deeply concerned about the urgent problem of reducing vehicle noise.
The effect that large-scale trucks and buses have on traffic noise varies somewhat depending on such
factors as vehicle speed, traffic volume and the ratio of large-scale vehicles to other vehicles in a ceiiain
area. However, it is a fact that they do contribute a great deal to traffic noise and furthermore, the
general public also point to large-scale trucks and buses as being noisier than other vehicles.
Consequently, the administrative authorities of countries all over the world are sucessively
establishing noise control laws mainly for large-scale trucks and busses. Japan was one of the first ones to
realize such lasvs, for in September, I 975, the Japanese Ministry of Transportation set strict regulations
of lowering -3dBA for large-scale trucks and -2dBA for passenger cars. Moreover, the Central Council
lor Public Nuisance Measures proposed a draft for further restricting noise another —3dBA which will be
put into effect in 1979.
Under these circumstances, the Ministry of Internatinal trade and Industry started in 1974 a major
technical research and development project on noise reduction of large-scale trucks, and a joint research
program was begun based on a 3-year plan by four large-scale truck manufacturers (Isuzu, Nissan Diesel,
Hino. and Mitsubishi).
* Truck/Bus Testing Department Maruwer, Technical Center. Mitsubishi Motors Corporation
** Component Testing Section Manauer, Truck/Hub Testini! Department, Technical Center
*** Component Testing Section, Truck/Bus l-ixperimcnt Department, Technical Center
93
-------
During the first two yeras, research work was divided and each of the four companies was put in
charge of studying different subsystems such as engine noise, cooling system noise, exhaust system noise,
etc. On the third year, the four companies exchanged the results of their two-year-studies and then, each
company began working on developing its own low-noise proto-type turck.
In this project, the research area that Mitsubishi was in charge of noise reduction of the exhaust
system. We were able to obtain substantial results during this two year period. Therefore, we would like
to present a brief summary of our results.
2. CONTENTS OF INVESTIGATION
In studying the exhaust system noise reduction, feasibility of large-scale truck exhaust system was
taking into consideration in determining the target and conditions. Test and research were conducted
accordingly.
2.1 Target of Study and Conditions
(I) Reduction target: 8 dBA in exhaust noise reduction (at the maxim output of the
engine)
(2) Muffler back pressure: Less than 60 mmHg in pressure losses at the muffler
(3) Muffler size: 1,000 mm in cavity length, outside diameter less than 300 mm
(4) Type of muffler: Reactance type without using any sound absorbing material
To systematically investigate noise from the exhaust system, a lot of fundamental elements of the
exhaust pipe and tail pipe composing the muffler are fabricated as prototype exhaust systems with the
basic and mountable shapes on the vehicle. The following items are tested for study.
2.2 Investigation Items
(1) Acoustic investigation
Investigation of the acoustic attenuation characteristics of the exhaust systems using a speaker as
the sound source
(2) Investigation of draft noise (flow generated noise)
Investigation of noise which is produced due to a draft corresponding to an exhaust gas stream
flowing through the exhaust system of the vehicle
(3) Investigation of radiated noise from the exhaust system
Investigation to obtain .correlation between vibration and noise which are produced by vibrating
the exhaust system, also to grasp the radiated noise in the vehicle.
(4) Investigation of the exhaust noise in vehicle
Investigation of the exhaust system fabricated for trial based on the investigation results of items
(1) and (2) on the vehicle
3. ELEMENTS TESTED
The fundamental elements of the exhaust system which are currently used for trucks are
provided as test elements. To facilitate a variety of combinations of these fundamental elements, the
outer shell of the muffler and separator are constructed to permit splitting and coupling. Typical
examples of the test elements are shown in Table 1. The premuffler, main muffler, tail- pipe submuffler,
exhaust pipe and tail pipe are provided as test elements for the vehicle.
(1) The premuffler is fabricated for trial based on the resonance and expansion type fundamental
elements.
(2) The main muffler is fabricated for trial based on combination of the perforated-pipe gas dispersion
94
-------
Table 1 Fundamental elements configurations and their parameters of exhaust system
Type
Expansion chamber typo
Resonator type
Perforated-pipe Gas Dispersion
(Multi-holes type)
Exhaust pipe
Tail pipe
Parameter
Li
L
l-o
L,!
La\
B
Lf
n
Df
li
L
D,,
n
Separator
installed
or not
R
e
Ellipse
type ^
Square Q
type
Shape
_1 , _ ts. _ 1
L,, ~{ ^- ' a
1 L-- J
L M
1 i n<-p
— — —
I !"
I j ^-4+./J(|
^ ^ n: Number of holes
n: Number of holes
/
-U^^J,.
Hn,P / L Separator
diameter -t
aj /L
^-n-iPhl ^
and expansion type elements taking into consideration the acoustic and draft characteristics and back
pressure.
(3) The tail pipe submuffler is constructed with easy mounting and demounting mainly based on the
resonance type in trial fabrication to secure attenuation of a characteristic frequency.
4. SOUND TESTS
4.1 Calculation of Muffler Sound Attenuation
In calculating the acoustic attenuation characteristics of the exhaust system, there are the Davies and
Hirata methods which take into consideration the mean Air flow of exhaust gases. However in this
paper, the calcuation were performed based on the analysis method of Fukuda and Ohters.
The following hypothetic conditions are provided in ejaculating the noise attenuation of the exhaust
system.
(1) Sound pressure is much lower than the mean pressure in the pipe.
(2) The density and sound speed of the medium in the pipe are uniform.
(3) Influences and energy losses due to the viscosity of the medium are neglected.
(4) The wall surface is not vibrated and acoustic energy does not transmit the wall.
(5) Influences of draft are neglected.
(6) The sound wave in the pipe is a plane wave which travels in an axial direction.
Underthese conditions, let us assume that without the muffler installed, radiation power at the outlet is
represented by W2 , a volume velocity of wave motion at the outlet opening by U2 and radiation resistance
95
-------
at the outlet opening by K/ , and that with the muffler, these factors are respectively represented
by H',, £', and R^ . Acoustic attenuation of the muffler can be expressed as:
i * r i r r i D I
A11 = 10 log^-^- = 20 logu,-^-- + 10 log 10 — 11)
Assuming (lint /' shows an mis value of sound pressure, U an mis value of the volume velocity of
wave motion, suffix 1 the mlei opening and suffix 2 the outlet opening, the matrix of the exhaust pipe
without the muffler (pipe length: /') can be expressed as:
t/,'J LC' Z)'JLl/2'J (2)
The matrix of the whole exhaust pipe system with the muffler is represented as:
\uHc X'J (3)
Let us consider the case that when the sound source has a constant sound pressure, its sound pressure
does not vary regardless of installation of the muffler (P\ = P\} and that radiation resistance at the out-
let opening has also an expression of (Kt'=R2) as an assumption. Equation (1) will be:
(4)
Hence if values D and B' are found by substituting an electric circuit for the matrix of the whole
exhaust system, attenuation can be obtained.
Fundamentally speaking, when p, c, S and / respectively represent the density of a medium, each
mean value of sound velocity, the sectional area of the pipe, and pipe length with the pipe opened at
both openings, the following Equation is given.
rA, B,1 [cos*/, J5ink!'
= c
1C, D,± l'j±!s[nkl, cos*/,
PC
where k=2nflc
When the pipe closes at one opening, the equation is represented as follows.
A' B'l_[1 °'
•Attenuation is calculated by obtaining value B substituting equation (5) and (6) for equation (3)
and using equation (4) based on B'= j(pc/S')sinkl' given from equation (6).
4.2 Test Method
In the acoustic test, differnce between noise levels of the exhaust pipe without the muffler (/' = 175
mm) and that with the muffler is measured, to indicate attenuation. The sound pressure level measuring
point is fixed at a given position from the exhaust system outlet.
A pure tone, white noise and exhaust noise from the vehicle are selected as sound sources, and
investigation is performed including the evaluation (weighting) method for the acoustic attenuation-
distance characteristics.
4.3 Test Results
4.3.1 Pure Tone Test and Band Noise Test
The acoustic attenuation-distance characteristics of the 1/3,-octave band noise, using white noise as a
noise source, matches well with the characteristics of a pure tone up to approx. SOOHz, when compared
96
-------
in the simple models shown in Fig. 1.
Over a band to approx. 8kHz. distribution of actual exhaust noise spectra is close to that of white
noise having considerable power. In the case of indication of the said characteristics of sinusoidal, its
evaluation is difficult over a band exceeding 2kli/ but if the characteristics of band noise is used, the
evaluation guideline of the band can be obtained.
But spectral indication based on sinusoidal is required to accurately weigh the characteristics over a
band of lower than 2kllz. It is desirable to choose the noise source considering its merits.
:o hole
Sinusoidal
Band noise
I • , yr'\ />i| u"; j.
'-.-.--,. / ' i' '! 'i i! '[t •'•••'•'
'
20 50 100 200 500 Ik 2k 5k 10k 20k A Lin
Frequency (HZ)
-20
rfio hole
, 400 '
L-l^r-. i .' . 1 _^ Sinusoidal |
1 : Separator (4; f 1/3
' ' !'( -j Ban'
'! Y : * '
Oct.
noise .
20 50 100 200 500 Ik 2k 5k 10k 20k ALm
Center frequency of 1/3 octave (HZ)
Fig. 1 Effect of test signals to attenuation character
characteristic of mufflers
HO
120
100
80
S 60
40
—Calculated values
Measured values
Frequency (Hz)
Fig. 2 Comparison of the measured and the cal-
culated muffler attenuations
Frequency (HI]
-40'
10' 10'
Frequency (HZ)
Fig. 3 Attenuation of muffler expansion chamber
type
97
-------
4.3.2 The relationship between Calculated Values and Measured Values
The relationship between calculated values and measured values of the fundamental elements shows
almost satisfactory approximation. Combination of the expansion and resonance types is exemplified in
.Fig. 2 as a combination of the fundamental elements.
This shows that also in the combined, models, coveration is excellent and that estimation of the
attenuation characteristics is possible.
4.3.3 Characteristics of Fundamental Elements
(1) Expansion chamber type
From calculation of equation (4), the practical approximate equation to check a qualitative
tendency in 'the expansion chamber type is as follows.
S sinkL sinfcZ.0
S CQS&L,'! COS&Loi
where
S: Sectional area of the cavity
s: Sectional areas of the inlet and outlet pipes
In contrast, a qualitative tendency in parameter variations using the actual models is given as
follows, and the typical examples are shown in Fig. 3.
D (cavity diameter): The maximum attenuation is proportional to 20 logs(S/s).
L (cavity length): The number of passing frequencies increases as L is lengthened.
L0 (tail pipe length): It shows the same tendency as variation of L
Z.,, (insertion pipe): The characteristics of the resonance type can be superimposed on those
of the expansion type whenZ.OIand L;iare lengthened.
(2) Resonator type
Where the volume of the resonance chamber .is represented by V and the area of the resonance hole
by Sp , resonant frequency (f^ of the resonator type is given as follows.
(c : Sound speedl (8)
Variations of the parameters with these .factors are given below, and the typical example is shown
in Fig. 4.
Lf (cavity length): f\ decreases with an increase of V if L increases, and the
number of pass frequencies which depends upon /. also
increases.
I> (resonant hole length): /idecreases with an increase of L/, but attenuation does not
vary.
Dp (resonant hole diameter): The same tendency as in the expansion type is shown as Dp
when Dp Dp is increases to some extent:
'i (position): No influence
"(the number of resonant holes): /, changes by V/I-folds as n increases, and when it is further
increased, the tendency becomes close to the expansion type
(see Fig. 4).
(3) Perforated-pipe gas dispersion type (Multi-holes type)
This type has the same tendency as the expansion type with respect to the acoustic characteristics.
98
-------
Number of resonant holes
Frequency (HZ)
Fig. 4 Effect of number of resonant holes
1 3 oci Band noise
i ' i
20 50 100 200 500 Ik 2k 5k 10k 20k" A Lin
Frequency (HZ)
co 40
20 50 100 200 500 Ik 2k 5k 10k 20k A Lin
Frequency (Hz)
« 20 50 100 200 500 Ik 2k 5« 10k 2Ck A Lin
< Frequency !ri:)
Fig. 5 Attenuation of various test mufflers for
vehicles
With or without separator: No affectation upon the acoustic characteristics (see Fig. 1).
Dp and n; Same as stated above.
The characteristics of. the fundamental elements were mentioned above. Seeing the band noise
characteristics, there is a tendency that the perforated-pipe gas dispersion type is larger than the expan-
sion type in the attenuation characteristics over a band of higher than 2kHz.
4.3.4 Characteristics of Mufflers for Vehicle
The acoustic attenuation-frequency characteristics of a prototype muffler is shown in Fig. 5. It is
found from the attenuation characteristics that the muffler showing extreme decrease at a particular
frequency is disadvantageous. But the damping effect of the exhaust system includes complicated factors
such as variation of acoustic attenuation due to the influence of the exhaust gas stream, so it cannot
absolutely be weighed.
99
-------
5. DRAFT NOISE TEST (FLOW GENERATED NOISE TEST)
5.1 Test Method
A rotary blower is used as a draft source, and steady air current is supplied to the exhaust system to
be tested via the silencer. To measure draft noise, an nnechoie box is used, a microphone is installed at
an angle of -15° and a positon of 50 cm from the exhaust poit and a stragiht pipe is used 'for the
evaluation standard. The test system block diagram is shown in Fig. 6.
5.2 Test Results
When steady air flow is sent to the exhaust system, power of draft noise which will be produced
from the exhaust port can approximate to flow velocity as follows from data of types of muffler shown
in Fig. 7.
With v < 50 m/s.PTKLo: 100 mls,
5.2.1 Features of Fundamental Elements
(1 ) Expansion type
(a) Draft noise level is 10 to 20 dBA higher than that of the straight pipe.
(b) If a gap of the input/output insertion pipes is reduced, whistling close to the spectrum of a
pure tone tends to be produced (see Fig. 8).
(c) When the outlet insertion pipe is lengthened, draft noise increases (see Fig. 9).
Bypass valve
Rotary U. Swirl
blower ft Tflo\vrnetcr
Microphone
Fig. 6 Experimental layout for flow generated noise
130
120
110
100
90 -
70-
601-
muffler
straight
pipe
5 10 20 "30 '" 50
Q (nrVmm)
L—I 1 i : i
30 50 70 100 200
» (m/s)
Fig. 7 Flow generated noise level under various
mufflers
100
-------
120
110
100
80
70
o
E
60
5 10 15 20 25
Flow rate (m3/min)
50
5 10 15 20
Flow rate (m]/min)
Fig. 8 Effect of the gap of inlet and outlet expan-
sion pipe to now generated noise Fig. 11 Reduction of now generated noise by horn
type extension pipe
130-
120-
„ 110-
'3
•5 9°
u
C3
S 80
25n
25 Straiglit pipe
(Unit: m'/niin.)
10°
10'
Back pressure (mmHg)
Fig. 9 Flow generated noise of expansion type
mufflers
1 100
c
T3
2 80
u
c
c
20k
20 200 2k
Frequency (HI)
(1) Spectra of Expansion
chamber type
•L Ji L
10 15 2B 2
Flow rate (mVmin)
35
,'' " ij'' Fig. 12 Reduction of flow generated noise by horn
//~v'^-- type extension pipe of test muffler for
,'' ^"^ vehicles
20 200 2k 20
Frequency (H2j
(2) Spectra of Multi-holes
type with separator
Fig. 10 Change of flow generated noise spectra
by air flow rate
101
-------
(2) Resonance type
(a) When holes having a diameter of less than 10 mm are placed in two or three rows its noise
level rise is 2 to 3 dBA as compared with the straight pipe.
(b) When an opening diameter exceeds 20 mm, extreme whist'ing is produced.
(3) Perforated-pipe gas dispersion type
(a ) When the separator is installed to this type, it is the same as the expansion type,
(b) Without lepar.nor. whistling tends to be produced.
The characteristics of the fundamental elements are given in Table 2.
5.2.2 Reduction of Draft Noise
It is considered that draft noise is produced due to such factors as vortex, confliction, friction and
resonance when high-speed exhaust gas How passes through the muffler. Its spectrum is predominated by
a high-frequency components as shown in Fig. 10.
As a reduction means, it is important first to select a muffler having low draft noise level, especially a
hard-to-whistle element in the fundamental elements. As a very influential part of the internal
component, the edge is important. In order to prevent the edge from getting too close to the core of the
jet stream, the edge is mode horn-shaped (referred to as with R) which greatly reduces flow generated
noise.
Fig. 11 shows the effect of noise reduction in the expansion type, where the noise is reduced 10 to
20 dBA. When this is applied to the muffler for the vehicle, the effect shown in Fig. 12 is obtained.
5.2.3 Consideration of Back Pressure
Fig. 1 3 shows the back pressure-draft noise characteristics of the fundamental elements. Of the types,
especially the perforated-pipe gas dispersion type with the separator is in question, and it is approx. 1
folds as many as the straight pipe in pressure loss. An increase of pressure loss is a fatal defect for this
type. It is required to select a perforation rate of more than 1.5 as shown in Fig. 14. In practice, an
effect of 40% reduction in back pressure is achieved by selecting a perforation rate from 1.5 to 3.0, 2
folds as many as the original one in the prototype muffler for the vehicle.
Comparison of attenuation, draft noise level and back pressure based on the straight pipe is shown in
Table 2.
6. TESTS OF RADIATED NOISE FROM EXHAUST SYSTEM
6.1 Test Method
The schematic test system block diagram is shown in Fig. 15.
In this test, the exhaust system on the vehicle is vibrated on a base to investigate the vibration response
characteristics, and radiated noise from the pipe wall is typically measured on a close location mainly to
investigate the correlation between vibration and noise. For that reason, normal sine-wave vibration and
random vibration close to the condtions of running vehicle are selected.
6.2 Test Results
6.2.1 Shaker Test Result
Disturbance which the exhaust system suffers from the engine is 15 Gin maximum at the exhaust
manifold, and its predominant component ranges from 300 to 2,OOOIIz. When random vibration is
applied based on white noise of the exhaust system, a spectrum of each part obtained is almost similar to
a spectrum seen while the vehicle is running. A spectrum example under vibration is shown in Fig. 1 6.
102
-------
120
< 110
80
I 20 mj mm:
100
I II
[*- — : * L
jft Exhaust
fS\\
pipe _
'"Mu filer"
2_c Mik
I
Tail
e
pipe
"Shaker Fig. 1 5 Testing system of radiated noise from
exhaust system
537 975
'50 100 150
liack pressure (mnillg)
Fig. 13 Relation of flow generated noise and back
pressure of typical muffler elements
P. - p.
..=. 100
o
£
!
' \
v
\
; ' - ",F
/
s? * • »t
/- •• " ' — 1~5 — •
X* * • •
^
50 100 500 Ik
Frequency (HZ)
Fig. 16 Vibration response of exhaust system
(Random excitation)
+ 20
05 1 235 10
Perforation ratio '
Fig. 14 Static pressure coefficient vs. perforation
ratio
-£ -20
°10 20 50 100 200 500 Ik 2k
Frequency (HI)
Fig. '7 Vibration response of exhaust system
(Sinusoidal excitation)
Table 2. Rough characteristics of fundamental muffler elements
Expansion Multi-horn type
Straight pipe chamber Resonator /Separators / Separator noti
1 ' " \) I installed /
type
type
V installed ,
Attenuation (dBA) ff 4 to 6 .1 to 2 8 to 9 7 to 8
Draft noise Level (dBA) 0 15 to 20 2 to 5 10 to 20 15 to 25
Whistling None Small Middle None Large
Back pressure (%) 100 130 106 210 116
103
-------
The prominent peaks which appear in the spectrum depend upon resonant oscillation particular to
the system. The response acceleration ratio, obtained by the sine-wave vibration, more prominently
proves this fact. The comparison is shown in Fig. 17.
These peak frequencies often approximately correspond to calculated values of proper oscillation of
the model system (see Table 3).
Table 3 Measured and calculated resonant frequencies of exhaust system
(Unit: 11?)
1st
2nd
3rd
4th
5th
6th
7th
Measured values
Random excitation
-
50
106
218
356
537
975
Sinusoidial excitation
-
37
108
214
360
534
974
Calculated proper values
7.2
44.8
125.3
245.3
410.2
604.3
843.0
ft T Flexible piper
' J
\
'Engine '
Jl — IP 4 •._'''
^
/ \
\
Flexible pipe
not installed
\
~\J-
*" \ j
1 5 10 15 20 25 30
Measurement point
Fig. 18 Reduction of the radiated noise and the
vibration of exhaust system by insertion
of flexible pipe
80
0
Engine
(EJchaust
A'tj noise)
—^>v A
(Muffler W^i
,v/_ radiated P
noise) •=—'
pre- r(Temp.)
muffler ?Main muffler
".VI,
Engine 2500 rpm full load
100 -5 g
30
100
600
200 300 400 500
Exhaust gas temp.
Fig. 20 Influence of exhaust gas temperature on
exhaust system noise
104
-------
6.2.2 Effects of Anti-vibration Clements
There arc types of flexible pipe as anit-vibration elements which arc applicable to the exhaust system
But almost all the elements do not satisfy conditions such as heat proot'ness, u.is leakage, anti-vibration
performance aiul durability. In this test, an interlock type flexible pipe (known as bellows) is used.
'Ihe relationship between vibration response and radiated noise of the exhaust system in random
vibration is as shown in Fig. 18. In such an exhaust system model, there is almost'no sound pressure
attenuation in the exhaust pipe and its level tends to increase at the mid-section 01 (he mulller cavity
But radiated noise can be reduced approx. 10 dBA by using the anti-vibration element, and as example is
shown.
7. EXHAUST NOISE TEST ON VEHICLE
7.1 Test Method
To test exhaust noise and radiated noise from the exhaust system on the vehicle, an Fddy
dynamometer having 300 PS is mounted to control engine output. The system, shown in Fig. 19,
is used to measure only the noise from the exhaust system separating that Irom the engine noise.
For measurement, data are processed in online mode by the measuring vehicle which mounts a
miniature computer.
The measurement procedure is shown in Table 4.
Table 4 Outlines of Measurement Method
Type of test
Sound
Flow
gcnenoisc
Radiated
noise from
exhaust
system
Exhaust
noise in
vehicle
Radiated
noise from
exhaust
system of
vehicle
Noise
Mike position:
20 mm from
exhaust port
Mike position:
45°, 50 cm
from exhaust
port
Mike position:
50 mm from
pipe wall
Mike. position:
45°, 50 cm
from exhaust
port
Mike position:
100 mm from
pipe wall
Measurement method'
Back pressure
50 mm before
exhaust pipe
(1) 100 mm from
manifold out-
let
(2) 200 mm
before mani-
fold
As staled above
Temperature
1*0
Normal
20 to 40
Normal
(1) 100 mm from
manifold out-
let
(2) 200 mm
before mani-
fold
As stp*cd above
Accelera-
tion
g pick-up
g pick-up
Varied
press.
200 mm
before
manifold
As stated
above
Test condition
(1) Input sound
pressure: Constant
(2) Sound source:
(a) Sine wave
(b) Random
(c) Engine noise
Flow rate. 0 to 30
m'/min.
(1) Vibration input:
2 g rms
(2) Oscillator:
(a) Sine-wave
(b) Random
(1) Ensine speed:
1,0"00 to 2,500
rpm
(2) Load: 4/4
As stated above
105
-------
7.2 Test Results
As compared with the fundamental study of sound, draft noise and radiated noise mentioned above,
when studying the actual vehicle, one must also consider the effects of exhaust gas flow containing
exhaust pulsaiion and of temperature.
7.2.1 lixhaust Noise from Vehicle
During measuiemeiH of noise from the exhaust system on the vehicle, a factor to make the measure-
ment difficult is variation of exhaust gas temperature. Fig. 20 shows variation of exhaust system noise
with temperature. Spectra of exhaust noise and radiated noise under that condition are shown in Fig. 21.
1-xhaust noise level increases with temperature rise. 1'his is probably due to the increase of flow
generated noise caused by increased exhaust gas flow rate. In the meantime, radiated noise tends to
lower in level with temperature rise.
7.2.2 Reduction of Vehicle Exhaust Noise
Fig. 22 shows the relationship between exhaust noise reduction and back pressure in combination of
the prototype exhaust systems which have been fabricated for trial this'time.
(2) Premufflcr radiated noise
Low-temp
90
60
(3) Main muffler radiated noise l°°»n)
Low-temp.
202002k
Frequency (Hz)
CD
•o
Drum can
• ! —.. .-- Perforated-pipe Gas Dispersion
Premuffler Main I
-muffler .
-(2.2 folds as
much as
capacity 1)
50 100 150 200
Back pressure (mmHg)
250
Fig. 22 Relation of exhaust noise reduction and
back pressure of various exhaust system
arrangements
Fig. 21 Influence of exhaust gas temperature on
exhaust noise spectra
106
-------
Table 5 Performance of the typical exhaust models
Engine at 300 PS/2,500 rpm
Typical mode
I
n
in
N
3000
•
*
1
3000
- - -
)
000
DOO.^280
3^3
500^.90 ^n ,000.^80
t— 1
10
(=
, r
00.^2!
Goal
~
=rrr-).
1 1 ifUUU
0 1
2500
... —
)00, ,1280
=1-
values
Vdlidc ; IkMK'll
Exhaust
noise
0
(108.5)
-8.8
- 8
lU-k rn5S.(nimllc;),K:,JL!k\J nobe(JK-\» ^ __ c ,,,„ , ]-,, ,,-,„,„•
Mjnit'okl| licl'ore ' I're
nutlet ' muffler mul'ilei
0
(157)
+ 6
- 7
+ 16
0
(68)
— 9
-16
60 >
0
(97 6)
+ 2
+ 5.2
+ 3.6
Mut'ller (dm . Uii.M iJBA>
(104.3)
-5.5
°
+ 3.0
-4.8
r
16
20.3
20.3
20.3
i
0
(104.5)
-12.4
-12.4
Type C, shown in Fig. 5, is used as the main muffler here. It is delicately affected by the tail and
exhaust pipes, and thus it is important to select the most suitable length and elements when arranging
them in the exhaust system.
The typical models selected and required'layout for goal values for reduction are shown in Table 5.
Model I is a reference model having a muffler capacity of 33.2 lit. (2.23 folds as much as displacement
of the engine tested). Model II has a muffler capacity of 61.5 lit. which is about 2-folds as much as the
reference model' Model IV is 3.7 folds as much as the reference model in muffler capacity.
It is known that to clear a reduction target of -8 dBA, a muffler capacity which is 2.7 folds as much
as the reference model is required.
7.2.3 Radiated Noise from Vehicle Exhaust System
The vibration response characteristics of the exhaust system on the vehicle matches well with that of
bench test mentioned above. As far as radiated noise level on the vehicle is concerned, attenuation in
the muffler is poorer than the bench test as shown in Fig. 23. This is estimated that radiated noise from
the exhaust pipe close to the muffler, and also fromexhaust pulsation is amplified and transmitted. For
-10
^ o
-10
\
Tail pipe
Fig. 23 Example of radiated noise from exhaust
system of vehicle
107
-------
that reason, the flexible pipe which provided a reduction effect of more than —10 dBA for radiated
noise in the bench test provides only -2 to 4 dBA in this case.
As an effective measure for radiated noise, there is lagging. A lagging effect of 10 to 15 dBA is provided
by a heatproof anti-vibration material (t - 25 mm) and iron sheet (./ = 1.0 mm) but lagging is unavoidable
when a radiated noise measuie is essential.
8. CONCLUSION
The following guide lines were obtained for noise reduction of the exhaust system on the
diesel-engine vehicle through this study.
8.1 Acoustic Characteristics
A spectrum of exhaust noise without the muffler is almost close to that of white noise, and a com-
ponent in engine combustions over the low-frequency region varies 2 octaves or. so in the engine speed.
range. For that reason, it is ideal that the attenuation spectrum required for the mulfler is flat over
almost the entire frequency region and that attenuation level is high.
But it is impossible in practice to obtain the characteristics which are almost flat in the restricted size
range of the exhaust system. In this test, it is considered better that the perforated-pipe gas dispersion
type providing comparatively high attenuation in the high-frequency region and the expansion type to
permit high attenuation at lower than 1 kHz should be combined together, and that the region which
will need more attenuation even in this combination should be covered by the resonance type.
8.2 Draft Noise
If is desirable to avoid as much as possible the use of elements which tend to produce draft noise,
and shape the outlet insertion pipe to a horn when the expansion type is used. When using the
perforated-pipe gas dispersion type whether the separator is installed or not, consideration must be taken
to do so at the prestage of the muffler.
8.3 Radiated Noise
It is found that of types of noise from the exhaust system, noise radiated from its outer wall occupies
a large share, and that it is not negligible in noise measures. It is also qualitatively proven that exhaust
pulsation does greatly affect the level of radiated noise, and that in relation to this, mounting of the
premuffler is effective to reduce radiated noise.
Shut-off of transmission of engine vibration to the exhaust system and lagging effects are ascertained
as counter-measures tor radiated nose, but many problems still remain in'practical durability and
reliability.
The influence or rigidity of the exhaust system upon radiated noise and transmitting noise
characteristics were not covered by this investigation, and these will have to be solved through farther
research.
8.4 Back Pressure
As far as back pressure in the exhaust system is concerned, pressure losses in the exhaust pipe are
larger than in the muffler. It is important in design to increase the diameter of the exhaust pipe and
take a large radius of curvature at the bending sections when piping.
To reduce pressure losses in the muffler, it is required for the perforated-pipe gas dispersion type to
secure a perforation rate and for the expansion type, to design a horn-shaped outlet insertion pipe.
design.
Reference Literatures
(1) P.O.A.L., Davies, R. J. Alfrcdson, Design of Silencers for Internal Combustion Engine Exhaust System
(2) Fukuda and Okuda, Mechanical Society Magazine, 72-604, May 1969
(3) Tsutomu Kanai, NLR-12, June 1959
108
-------
METHOD AND APPARATUS
FOR
MEASURING MUFFLER PERFORMANCE
Peter Cheng
STEMCO MFG. CO.
Longview, Texas
109
-------
The measured quantity in our test facility is not the transmission
loss nor' insertion loss, but the pure exhaust noise under conditions
simulating those specified by state and federal truck noise laws.
The tool to evaluate the pure exhaust noise is a bench test conducted
at a test facility where total isolation of all other noise sources
is feasible. The cross section of exhaust noise test lab is shown
in Fig. 1.
The installation features an underground structure to mount test
engines and water brake dynamometers. This structure serves to
isolate the mechanical and air intake noises from the exhaust noise.
All the exhaust from the test engine is piped directly above the
ground to the muffler. The exhaust pipes are positioned in a manner
as close to that found on the vehicle as possible.
The site was chosen for it's compliance with SAE specification for
stationary and drive-by test. That is, it is an open space test
site with no nearby reflecting surfaces. Typical ambient sound
level is below 50 dB(A), well below the measured levels. The height
of microphone and separation between microphone and muffler is
specified as 4 ft. and 50 ft. respectively, so that the measured
exhaust noise level would be about the same as that from a moving
truck undergoing a drive-by test per SAE J-366b procedure.
Before the testing modes are introduced, let us review briefly
thru Fig. 2 the drive-by test per SAE 366b.
The vehicle under test approaches point A with 2/3 of the rated
engine rpm and begins -acceleration at point A under wide open trottle
so that the rated engine rpm can be reached somewhere within the end
zone.
To simulate the vehicle test conditions, three test modes are con-
ducted .
(A) Steady state mode
- rated engine speed and full load
(B) Varing speed full load mode
- engine speed slowly varied from rated speed to 2/3 of rated
speed at wide open throttle
(C) Acceleration mode - accelerate the engine from low idle to
governed speed until the engine speed stabilizes and return to
low idle by rapidly opening and closing the throttle under no
load conditions.
Modes (A) and (B) clearly have the drive-by-test in mind. Mode (C)
simulates the stationary vehicle noise test.
110
-------
(2)
The "sound level rating" in Stemco aftermarket catalog is the highest
recorded pure exhaust sound level measured in above mentioned test
modes.
The "sound level rating" defined above may be too conservative in
many cases. To illustrate this point, three hypothetical cases
listed below will be examined.
Sound Level (dBA)
Engine Speed(rpm) Muffler A Muffler B Muffler C
2100(rated) 71 71 71
1900 73 73 73
1400 75 70 77
In the case of Muffler A, the peak value of 75 dBA at 1400 rpm (2/3
of rated rpm) may not be a factor in the drive-by test. The distance
between the microphone and point A is 70.7 ft. instead of 50 ft, and
usually other noise sources do not peak until at higher rpm's.
Muffler A and B may yield identical total vehicle noise per drive-by
test. On the other hand, the peak level at 1400 rpm in Muffler C's
case may indeed affect the total vehicle noise in drive-by test. A
peak value at 1900 rpm or 2000 rpm may also be important because the
vehicle would be close to point B in Fig. 2 and be right in front of
the microphone.
It is therefore difficult to use one dBA level to correlate bench
test results and drive-by test results without being either too liberal
or too conservative. But to a large extent, muffler designers can
usually use .the bench test .results and judge how the muffler will
perform in a drive-by test.
Ill
-------
STEMCO EXHAUST NOISE
TEST FACILITY
50 fi
1fO MICROPHONE
4 ft
J.
MUFFLER
— GROUND LEVEL —
^A
DYNO-
AIR
INTAKE
Fig. 1
-------
A
B
C
I/ /
T7
1111111111111
END ZONE
50'
MICROPHONE
Fig. 2 Schematic Diagram of Drive-By Test Per SAE-366b
113
-------
"OPTIMUM DESIGN OF MUFFLERS
by
D. Baxat A. Baz**and A. Seireg**
University of Wisconsin
Madison, Wisconsin
* Assistant Professor, Department of Engineering.& Applied Science, UW-Ext
** Research Associate, Mechanical Engineering Department
*** Professor of Mechanical Engineering
115
-------
Optimum Design of Mufflers
by
D. Baxa, A. Baz and A. Seireg
University of Wisconsin
Madison, Wisconsin
Abstract
This paper describes a computer based-design procedure for selecting the
optimum configuration of automotive reactive mufflers and acoustic silencers.
The procedure utilizes a specially developed scheme that predicts the pressure
histories, and accordingly the accompanied attenuation or amplification of the
noise level, resulting from the simultaneous reflection and transmission of sound
waves propagating through variable impedance exhaust tubes.
The developed procedure is general in nature and can be used for synthesiz-
ing the optimal configuration of mufflers for any given operating parameters and
design objectives.
Several examples are given to illustrate the optimum muffler configurations
necessary to minimize the transmission of noise level at different working condi-
tions. The examples demonstrate the potential of the developed procedures.
The described computer aided design approach can be readily applied for dif-
ferent patterns of exhaust pressure waves, mufflers with excessive temperature
gradients and wall frictional losses as well as any other operating conditions
and design objectives.
116
-------
Introduction
The continuously increasing demand for high performance internal combustion
engines has forced the automotive engineers to raise considerably the cycle pres-
sures and the engine speed. Such modifications have contributed considerably to
the increase of the exhaust noise level to the extent that it became a major
environmental pollution problem. Consequently, efforts have been exerted to de-
velop several forms of exhaust silencing systems in order to meet the severe re-
quirements of the noise pollution statutory limits without reducing the engine
performance. Realizing the importance of developing better mufflers the automotive
industry in the USA is expected to spend $16. to $100. per car to meet the 1978
noise pollution standards [1] . Such figures will definitely be higher in years
to come to meet the growing need for cars with better handling, i.e. with low
center of gravity, and therefore with very limited space for the exhaust systems.
With the emission control components, the muffler designer will, thus, be under
pressures to develop even more efficient and compact silencing systems.
The development of automotive mufflers has generally relied on empirical
skills guided by past-experience and simple acoustic principles. Some design
guides can be also found for simple muffler configurations as given by Magrab [2].
Only in the recents years has the development of automotive exhaust systems taken
a more systematic and rational approach as can be seen in reference [zJ to [6].
These efforts have presented different simulation techniques that utilize the
wave propagation theory to predict the dynamic performance of reactive mufflers.
* Numbers between brackets refer to references at end of paper
117
-------
The validity of the developed muffler simulation models has generally been tested
either experimentally or against close-form theoretical formulas that are developed
for simple muffler configurations. Common also among these studies is the fact
th^t all have.been used only to analyze the performance of reactive mufflers at
different operating conditions rather than to devise means for selecting the op-
timum muffler that is best suited for a particular application. Few attempts
[7,8] have been made to optimize the performance of mufflers but they were based
on exhaustive-experimental 'search for the geometrical parameters or the properties
of the lining materials for a muffler of a particular configuration.
The purpose of this study is to develop a computer-based design procedure
to synthesize the optimal configuration of any reactive muffler for any given
operating conditions and design objectives. The analytical procedure is based
on a computerized one-dimerrsional wave propagation technique developed by Baxa
and Seireg [3], This technique is used to monitor continuously the reflection
and transmission of pressure waves as they propagate through variable impedance
exhaust tubes. Consequently, the pressure-time history at any location inside the
muffler can be determined together with the accompanied degree of attenuation
of the noise level.
This optimal design approach of mufflers will eliminate the exhaustive trial and
error search for the best muffler ft any given situation and therefore reduce
the cost of development of the car's exhaust silencing system.
118
-------
The optimization procedure used in this study is an adapted version of
that developed by Wallace and Seireg [9] to optimize the shape of prismatic
bars when subjected to longitudinal impact.
Computational Scheme for the Analysis of Have Propagation in Mufflers with
Step Changes in Impedance
The classical theory of one-dimensional wave propagation enables us to pre-
dict the pressure P at any location X and at time t by relating these para-
meters by the following eq :
ax2 c2 9t2
where C is the speed of propagation.
This theory assumes that there are small changes in the instantaneous density
and consequently the instantaneous value is approximately equal to the average
density p , that the wave propagation is frictionless, the medium is homogeneous,
and the sound levels are below 110 dB re 0.0002 microbar.
This equation has long been the basis for the analysis of one-dimensional
transmission of waves and their reflections where changes in impedance occur.
The evaluation of pressure variations in tubes can become more difficult as the
number of impedance changes increases. However, with appropriate schemes, such
as that developed by Baxa and Seireg [3], these problems can be conveniently
and economically analyzed.
119
-------
The following are some of the basic assumptions made in the developed
muffler analysis program:
(1) Pulse length is long compared with the tube diameter.
(2.) The source moves the entire cross-section with the same
particle velocity.
(3) Pressure fluctuation levels remain in the linear elastic region.
The first assumption implies that the wave would have a constant speed
of propagation, which is determined by:
c - PQ
where y = 1.4; PO = mean pressure; p~ = mean density. The second assumption
indicates that the waves move as plane waves through the tube. Finally, the
third assumption suggests that the waves and their reflected and transmitted
components can be combined by superposition.
The time necessary for a disturbance to propagate through a tube segment
of length L can be calculated from
tp = L/C (3)
In a complex tube comprised of many different segments (Figure 1), a
propagation time is determined for each segment length. By comparing propa-
gation times, a ratio of numbers K,, 1C,..., K is determined from the
following expression:
(t )1 (t )2 (t )n
t = — = —B— = ._ = —B (4.)
V V Y
K1 K2 Kn
120
-------
The significance of these integers is that it takes a wave K, units of time
(where one unit is t ) to travel the length of the first segment, K2 units to
travel the length of the second segment, etc. Because the propagation times
are multiples of the unit of time, t , the initial wave and all reflected and
transmitted waves will reach the interface at times which are some multiple of
V
Every section of the tube has an acoustical impedance which depends. upon
the mean density (PQ)> the velocity of propagation (c), and the cross-section
(S) of the pipe. The relationship is as follows:
Pnc
Z- -§- (5)
Pnc is often referred to as the characteristic impedance of the medium.
By considering the pressure and velocity equalities at the interface of
a wave going from tube 1 to tube 2, it can be shown [10] that the transmission
and reflection. of the velocities are as follows:
UR
Z2 + Zl
2Z, S2
u = —! -^ u,
'7+7 s
L2 L1 51
where UT, UD and UT are incident, reflected, and transmitted volume velocities,
1 K I
respectively; Z, and Z~ are the impedances of the two tubes. Since pressure
and volume velocity are related by:
P = U pQc/S (8)
121
-------
equations (6) and (7) become:
PR = L PI (9)
R Z2 + Z1 l ( '
2Z
PT = TTTpi do)
L2 Al
When the density and velocity are constant,
1 1
1C< = (7L-T^Pi = CRPI (ID
,S2 S1,PT _ ,S1 " S'
s2 s,
Where cDisthe reflection coefficient.
K
S? ' 2S
_2_)P _ { ] )P
T ~ CT T C\?}
'' T ~ i 4- i 1- «;+ ^ \'£/
1 ± ± Is
Q S I On
bg b1 ^
Where CT is the transmission coefficient.
Consequently, when the magnitude of the incident wave and the physical
properties of the gas in the tubes are known, the transmitted and reflected
portions of the wave can be determined from equations (11) and (12).
In order to analyze a general wave being emitted from the source, the
physical properties and initial conditions of the source and of every segment
of the tube must be known. These properties should include the impedance,
speed of wave propagation, area, and length. In the case of a homogeneous
gas the reflection and transmission coefficients can be reduced to a function
of area only. The ratio of the propagation times must also be known. The initial
122
-------
condition of the tube is considered to be that of no pressure waves inside.
Therefore, it can be seen that knowing the parameters of area (S.), length
(U), static pressure of the gas (PQ), static density of the gas (pQ), and
the ratio of the specific heat of the gas at constant pressure to that at con-
stant volume (y), one can determine the pressure history inside the tube. The
wave propagation speed can then be determined from the relationship c = \/ - — or
' P0
c = \ArT, where r is a constant dependent on the particular gas and T is the
temperature of the gas in degrees absolute. To determine the impedance of each
tube segment, the density (PQ)> the speed of wave propagation (c), and the area
of each segment (S.) are substituted in the equation Z = P0C. The propagation
1 S
times are determined from the segment lengths and the wave propagation speed
as t = L/C.
A'ratio of integers is found from this array of propagation times, either
by visual inspection or with the help of a computer program. Since it is assumed
that each tube segment contains the same gas at the same pressure and temperature,
the speed of wave propagation remains constant and the ratio of propagation times
will be the same as the ratio of segment lengths.
Once all the physical properties and initial conditions are known, the
pressure-time history can be determined as follows. After each unit of time,
each interface is checked and the reflected and transmitted portions of the
waves are calculated by using equations (11) and (12). All of the waves travel-
ling in the same direction from an interface are summed. By knowing the magni-
tude of all the waves arriving at and leaving a given interface, it is possible
123
-------
to construct the "pressure-time" history at every interface. This procedure is
repeated for each unit of time until a steady-state condition is .achieved.
The analysis scheme Utilizes this approach and can be used in one of two
modes. First, the response to a sinusoidal input can be determined and the
transmission loss can be calculated in decibels for the entire system. In the
second format, a general periodic pressure input can be read in and used to
calculate the pressure responses of the system. This second approach is particu-
larly useful in determining the effect of a tuned exhaust system on the pressure
history.
The computerized routine is developed to include as many segments as can
conveniently fit into the computer. Each segment corresponds to a particular
portion of the muffler. It is also possible to set the source and termination
impedance in order to investigate the effect of this variation on the system.
If the source or end is completely absorptive, the areas chosen would have the
same area as the connecting segment. If the source or end is completely reflec-
tive, the area chosen would be zero. A flow chart of the developed scheme is
shown in Fig. (2) to illustrate its different features.
Strategy for Designing Optimum Mufflers
The design of a stepped-configuration reactive muffler for attenuation of
exhaust noise levels is formulated as an optimal programming problem. The major
considerations in this formulation are the identification of the decision para-
meters, the description of the constraints imposed on the design, the explicit
statement of the objective and the development of a suitable search technique
for locating the optimum design parameters.
124
-------
Muffler Parameters
For the general case of a segmented muffler, as shown in Fig. (1), is sub-
jected to general periodic pressure waves of known amplitude, frequency and
temperature, then the system variables are: -
a. number of muffler segments .. n
b. Length 'L^'and area 'S_.'of each muffler segment where i = 1, .n
c. Source and termination impedances.
It can therefore be seen that for the n segment - muffler the total number of
system parameters is (2n + 2). Some of these parameters are specified beforehand.
The remaining variables represent the decision parameters and have to be selected
within the constraints imposed on them in such a way as to provide the highest
possible performance.
Explicit statement of Muffl-er design Objectives
An explicit statement of a merit criterion which accurately describes the
designer's objective constitutes a very important matter since this criterion guides
the search and determines the selection of the optimum values of the decision
par-- ^t-s.
Examples of the possible objective criterion for this class of problems are: -
(a) Maximization of the noise transmission losses at the engine
operating speed.
(b) Maximization of the noise transmission losses over a wide range
of engine speeds.
(c) Maximization of the negative pressures developed during the
suction stroke when using a tuned muffler.
125
-------
Other design objectives can be used to guide the selection of the muffler de-
sign parameters in order to meet the requirements for any particular situation.
Search Method
The steepest ascent method is utilized to serach for the optimum design
parameters of mufflers in order to achieve the maximum attenuation of the noise
level, or any other objective, associated with the incident pressure waves. The
optimization method guides the search for the optimum parameters along the di-
rection of maximum attenuation, or any other objective, by changing the value
of each design parameter X. independently by a small perturbation AX. and noting
the accompanied change in the noise level All. The new value of the design para-
meter X. is determined from the old value X. according to the following
Vl 'j
relationship:
Xij+1 = Xij + A (AU/AX.) i = 1.....M (13)
where M is the number of decision parameters. A is an optimally selected step
size that controls the changes between points j and j+1.
If no improvement occurs, the parameter is varied in the opposite direction.
If this also fails to produce an improvement in the merit value, this parameter
is kept constant for this step and the value of the other parameters is changed
in a similar way.
The details of the adopted optimization scheme are shown in the flow chart
of Fig. (3) to indicate the means included for selecting the maximum step size
without violating the constraints and for avoiding the termination of the search
at regions where the attenuation level vanishes. Such features make the use of
126
-------
the steepest ascent method very suitable for searching the complex design region
of the mufflers because it is extremely sensitve to parameter changes.
Therefore, for regions where no sharp ridges exist in the contours of the
objective criterion, this algorithm is equivalent to a gradient search. But
for situations where a ridge exists in the design space the algorithm is in effect
a univariate search.
Numerical Examples
The optimum design procedure is used to develop the optimum-muffler config-
uration necessary to maximize the attenuation of the noise level of a particular
pressure wave with a frequency of 1000 Hz and flowing through the mufflers at
a temperature of 70°F. The procedure is utilized to illustrate the effect of
changing the number of segments of the muffler on the degree of optimum attenua-
tion of the transmitted nofse. Mufflers having a fixed length of 3 feet but
with 3, 6, and 12 segments are considered to illustrate the potential of the
procedure in optimizing muffler configuration.
In all the considered examples the design problem is formulated as follows: -
Find the areas of the segments S. i = 2 —•> n-1
P.
To maximize the transmission loss .. TL = 20 log,n (pinPu ) db
IU output
such that S1 = Sinput
SIN= Soutput
min ^ i — max (14)
L. = l.f i = 1,... ,n
1 i
where each segment lengths L. is equal to a given value L,
1 i
127
-------
In the above formulation the muffler designer can select the desired limits on
the area and length of each muffler segment. Consequently S^ . , soutputj sm-jn
S and L,: are fixed values specified according to the designer requirements.
max T •
In the following examples these limits are taken as follows: -
S = S =1
input output
Smin/Sinput = 0.1
max input
Lf A - 3/n
where A is the wave length of the incident pressure waves
Example 1
Fig. (4) shows the results for a 3 segment muffler, as that shown in Fig.
(4-a). The optimization procedure with a initial configuration will produce the
configuration shown in Fig. (4-a). Such an optimal configuration results in a
noise transmission loss of 10.3 dB as compared to the 5.09 dB loss produced
by the configuration of Fig. (4-a). It is interesting to note that the area
of the middle segment in the optimal configuration, has increased to reach the
maximum allowable limit set by eq— (15). This agrees with the common practice
of single expansion chamber muffler discussed, for example, (2 and 3).
128
-------
Example 2
This example illustrates the effect of changing the number of segments of
the muffler on the noise attenuation while operating under the same conditions
as in the previous example.
Fig. (5-a) shows that starting with the 6 segment muffler illustrated in
Fig. (5-a-i) then the optimal configuration will be as shown in Fig. (5-a-ii),
and the noise transmission losses will be 9.6 dB which is a less efficient de-
sign than that produced by the 3 segment configuration of Fig. (4-b).
But if we start with the configuration of Fig. (5-b-i) then the optimum
configuration illustrated in Fig. (5-b-ii) shows a considerable improvement,
nearly 24.3%, over the optimum 3 segment muffler. If we consider, however, the
muffler configuration of Fig. (5-c-i) as the initial starting point for the
optimization routine, then "the obtained optimum configuration of Fig. (5-c-ii)
yields a considerable improvement of 61.4% over the optimum 3 segment muffler.
It can therefore be seen that increasing th*3 nurk^r ~ laments of a muffler
of a given total length, is expected to produce a considerable increase in noise
attenuation.
Also, it is interesting to note that starting with different initial con-
figurations does not produce the same optimum configuration. This is due to the
complexity of the design space and emphasizes the need for optimization tools
for designing mufflers and acoustic silencers.
129
-------
Example 3
This example shows the improvement in noise attenuation resulting from in-
creasing the number of segments of the muffler under consideration to 12 segments
Fig. (6-a) shows the initial and the optimized configurations which result
in a noise attenuation of 30.24 dB. This is almost three times as much as that
Of the optimum 3 segment configuration. This optimal 12 segment shape has been
obtained in a single iteration by the developed optimization routine.
Fig. (6-a-ii) shows another optimal configuration which is a symmetrical
arrangement of multi-connected expansion chambers.
If we consider the initial 12 segment configuration of Fig. (6-b-i) then the
resulting optimal muffler will attenuate the incident noise level by 32.34 dB
which is 6.94% better than that produced by the configuration of Fig. (6-a-ii).
Summary
The paper has described a computer-based design procedure for optimized
configurations of reactive mufflers with step changes in their acoustic impedance
when subjected to periodic pressure waves. The existence of multiple optimum
configurations is evident by the dependence of the final design on the selection
of the number of segments and the starting point of the search. The considered
examples illustrate the potential of the developed computerized optimization
approach as a powerful tool for synthesizing the optimal configurations of
reactive mufflers.
130
-------
Although the optimization in the considered examples is based on the
maximization of the noise transmission losses at one frequency, the technique
can be readily used to optimize the muffler design over a wide range of fre-
quencies as well as optimizing the exhaust pipes for improved engine performance
The procedure can also be applicable to situations where factors such as
mean flow, frictional losses, temperature gradients, variable source and
terminati-on impedances should be considered in the design scheme.
131
-------
References
1. Heath. R.A., "What's Ahead in the Automotive Muffler Field," Proc. of
National Noise and Vibration Control Conference, Chicago, IL, June 1974,
pp. 87.
2. Magrab. E.. "Environmental Noise Control," J. Wiley and Sons, New York,
1975, pp. 215.
3. Baxa. D. , and A. Seireg, "A Computer-based Procedure for the Analysis of
Reactive Mufflers and Tuned Exhaust Systems," Proc. of NOISEXPO, 1977.
4. Davies, P.O.A.L., "Mufflers for Internal Combustion Engines," Proc. of
Inter-Noise Conference, Denmark, 1973, pp. 143.
5. Karnopp, P.. J. Reed, D. Margolis, and H. Dwyer, "Design and Testing of
Reactive Mufflers," Proc. of Inter-Noise Conference, Washington, D.C.,
1974, pp. 325.
6. Okda, J., "Performance of Reactive Mufflers- and Calculation of Engine
Exhaust Noise," Proc. of Inter-Noise Conference, Sendai, 1975, pp. 655.
7. Lykkeberg. P., "Design of Reflection Type Silencers based on the Theory
of Reactive Acoustic Filters," Proc. of Inter-Noise Conference, Denmark,
1973, pp. 158.
8. Ver, I., "Design Optimi-zation of Gas Turbine Silencers," Proc. of Inter-
Noise Conference, Sendai, 1975, pp. 687.
9. Wallace, P., and A. Seireg, "Optimum Design of'Prismatic Bars subjected to
Longitudinal Impact," ASME Paper No. 70-DE-G, Trans, of ASME, J. of
Engineering for Industry.
10. Kinsler, I.E., and A.R. Frey, "Fundamentals of Acoustics," Second Edition,
J. Wiley and Sons, Inc., New York, 1962.
132
-------
SOURCE
^
w
SEGMENT S
1
Sl
5EGMEN
2
S2
T
S
JL
EGMENT SEGMENT
n- 1 n
Sn- 1
^^iii
Sn
FIG. I —A TUBE WITH n -SEGMENTS
133
-------
CRFAD THF NIIMRFR OF MUFFIFR SFGMPNTS )
1
A
(READ AREA, DENSITY AND LENGTH OF EACH SEGMENT )
I
(DETERMINE WAVE PROPAGATION RATIOS )
t
( CALCULATE
^L^ r\ L. r l_ L. L* I 1 w 1 1 r\/A 1
A
IMPEDANCES }
1
RANSMISSION AND A
OS AT EACH INTERFACE )
A
( PRINT MUFFLER PARAMETERS 3
SET INITIAL VALUES
REPRESENTING WAVES
1
OF VARIABLES
AND PRFSSIIRF
j
»
C CALCULATE INTERFACE "\
^ PRESSURE WAVES )
4
( PRINT INTERFACE
^
^x^THE STEA
<^ PRES
^SVV^ REA
( TIME INCREASED
PRESSURE )
S,
DY STATE^\^ YES
jUKLo ^x^ ^~~^^
CHED ^^
Jl NO
BY ONE TIME UNIT )
4
CALCULATE NEW PRESSURE WAVES ORIGINATING AT THE
INTERFACES (FROM THE SUPERPOSITION OF THE RE-
FLECTED & TRANSMITTED PORTIONS OF THE WAVES REACHING
THE INTERFACE)
Fig. 2 Block diagram of analysis program
134
-------
( READ NUMBER OF MUFFLER SEGMENTS
READ THE TYPE OF PRESSURE WAVE
INPUT TO MUFFLER
I
READ IF THE LAST SEGMENT OF THE
MUFFLER IS A VARIABLE
i
READ: PROPAGATION RATION AND AREAS
OF EACH MUFFLER SEGMENT
SET THE INITIAL CONDITIONS
CALCULATE SEGMENT IMPEDENCES
CALCULATE THE PRESSURE
HISTORIES AT THE INTERFACES
CALCULATE THE MERIT VALUE
WHAT IS THE VALUE OF LC?
PRINT VARIABLES
B
SET THE VARIABLE BACK
TO ITS ORIGINAL VALUE
DECREASE FIRST
VARIABLE BY 3%
ID THE
MERIT VALUE
INCREASE OVER THE
"NEW POINT'
VALUE
HAS
HE VARIABLE
BEEN INCREASED
NO
DERIVATIVE
= ' 0 '
CALCULATE THE
DERIVATIVE
INCREASE THE
VARIABLE BY 3%
AS TH
DERIVATIVE OF T
LAST VARIABLE BEEN
CALCULATED
Figure 3
135
-------
LJ
NO
= LJ =
PRINT THE VALUES
OF THE PARAMETERS
AT THE OPTIMUM POINT
RE ALL
THE DERIVATIVES
0 ?
SET ALL
DERIVATIVES
I
CALCULATE CHANGE
USING DERIVATIVES
NEW P01UT = OLD POINT +^
(NOTE: IF SIDE CONSTRAINT IS
VIOLATED THE VARIABLE IS SET EQUAL
TO THE LIMIT)
®
Figure 3 (Continued)
136
-------
Initial Muffler
2 3
TL= 5.09dB
.886 x .886 K .886
(i)
FIG.(4) Three Segment Reactive Muffler
224
uffler
1
3dB
2
3
i
q
it
k
1
> —
\
k
3.16
t
137
-------
Initial Muffler
23456
2.24
= 5.09db .443
.443
Optimum Muffler
23456
22
3.16
TL=9.6db
FIG.(5-a) Six Segment Reactive Muffler
•138
-------
443
443 443
Initial Muffler
1.73
TL= 6.86db
4 5
6
— ]
— i
1
r —
141
3.
^ —
Optimum Mufffer
TL= 12.81 db
1
2 3
4 5
6
— i
l
4
i
1
i
q
d
t
2.09
> — ^
i
2.10
r
»
» —
3.16
i —
FIG (5-b)Six Segment Reactive Muffler
139
-------
Initiol Muffler
1.73
5
6
. j
i
i
i
\
i
]
I
k
2
F
1
I
2.
443
-# * x X-
443 " 443
TL=7.38db
Optimum Muffler
r
i
1
2 3
i
1
4
4 5
i
6
j
3
\
1
r
'
— 1
2.94
'
~ '
2.96
3.11
i ,
^ ifi
f
TL 16.62 db
FIG.(5-C) Six Segment Reactive Muffler
140
-------
Initial Muffler
2 3 4 5 6 7 8 9 10 II 12
TL=5.09db
222A .22 2^ ,222 ^ ,222,
2.24
Optimum Muffler
1.38
.32
.36
2.23
2.26
TL 30.24db
FIG.(G-a) Twelve Segment Reactive Muffler
141
-------
Initial Muffler
2.83
1.73
1.41
-TL=3.39db
Optimum Muffler
3.12
2.84
1.91
1.71
ief"
TL= 32.34db
9
1
10 II
1
12
^
i
j
»
i
t
i
i
1.73
(
t
i
2
F
1
£45
1
i
3
t
3.
(l)
1.92
.96
2.90
3.02
3.16
FIG.(G-b) Twelve Segment Reactive Muffler
142
-------
BENCH TEST AND ANALOG SIMULATION TECHNIQUES FOR
ENGINE MUFFLER EVALUATION
by
Cecil R Sparks
Director, Engineering Physics
Applied Physics Division
Southwest Research Institute
San Antonio, Texas
Presented to:
Surface Transportation Exhaust System Noise Symposium
Sponsored by the U. S. Environmental Protection Agency
Chicago, Illinois
October 11, 12, 13, 1977
143
-------
ABSTRACT
The problems associated with laboratory evaluation of engine mufflers are
primarily those of (1) designing a facility which will provide a meaningful
measure of muffler noise reduction, and (?.} relating this physical (acoustic)
data to the action of the muffler when placed on a specific engine, exhaust sys-
tem. While a wide-band siren can be designed to provide a suitable noise
spectrum and source impedance, performance of any muffler must ultimately de-
pend on the exhaust pi'ping configuration into which it is placed. Experiment-
al work in the 1960's at SwRJ has shown that a bench test facility can provide
useful acoustic data if the candidate mufflers are being evaluated for a rela-
tively narrow range of engine applications, and a loudness evaluation technique
was evolved which could reliably relate data from the bench test facility to
performance (sone reduction) on an engine.
In addition, electronic simulation techniques have been evolved whereby
the entire exhaust system (muffler, manifold, and piping) can be quantitative-
ly evaluated on an electroacoustic analog. Although designed principly for
simulating pulsation filters, this anal.og has been extensively used for sim-
ulating the exhaust systems of reciprocating engines, and for the design of
mufflers specifically tailored for that engine, exhaust system, and range of
operating conditions.
144
-------
BENCH TEST AMD ANALOG SIMULATION TECHNIQUES FOR
ENGINE MUFFLER EVALUATION
BY
CECIL R. SPARKS
BACKGROUND
The problems associated with evolving a bench test procedure for eval-
uating the acoustic performance of mufflers lie chiefly in the fact that there's
no such thing as an inherently good muffler. Regardless of muffler design,
the MR afforded by any muffler is not a function of the muffler design alone, as
the muffler is merely one part of a complex acoustic piping system. The "best"
muffler for one engine may actually amplify noise from another.
Being -a passive acoustic network, a muffler's performance (amplification
or attenuation) depends not only upon its internal design but also upon, its
source and termination impedance (i.e., the attached piping), upon the spectral
distribution and amplitude of the engine noise spectrum, flow rate, pressure
drop and, of course, acoustic velocity (temperature and gas composition).
This is not to say that some muffler designs are not better than others for
a given range of conditions, or that an optimum muffler cannot be designed for
a specific set of conditions (and assuming a specific set of constraints on size,
etc.), but as soon as engine operating conditions change, or the muffler is appli-
ed to a different engine, its performance can suffer markedly- Normally, muffler
design is tailored to cover the range of engine operating conditions expected,
and is designed as an acoustic low pass filter with a minimum' of pass bands and
the lowest back pressure (flow resistance) possible. These are, in fact, the
major marks'of a "quality1' muffler.
The first step in seriously undertaking a program of bench testing, there-
fore, Ties in defining the application and operating conditions for which the
candidate muffler is to be evaluated. The more precise we can be in defining
these conditions and the more narrow the variations in application and operating
conditions are, the better job we can do both in designing a muffler and in
bench testing it.
He at SwRI did a study some 12 - 15 years ago for MERDEC (then ERDL) to
evaluate the feasibility of developing and utilizing a bench test facility as an
Army procurement aid for several classes of more or less similar stationary
engine applications. The most questionable part of the effort was simply to
define if the military standards engines used in these applications were suf-
ficiently similar in exhaust spectral content and the acoustic properties of
their exhaust system that any one set of bench facility tests would be of sig-
nificant value for extrapolating performance to all engines in the selected
class. Perhaps the results of this program will be of interest to this group
in defining just how .a bench facility might be utilized in testing muffler
"quality" and in defining some of its inherent limitations.
In this discussion, I regret that time will not permit a full discussion
and description of the exact design procedures used in evolving the bench test
facility (e.g., the siren), to analytically prove some of the assumptions made
145
-------
(linearization procedures in extrapolating acoustic system response] or in pro-
viding experimental documentation of the validity of scaling some of the com-
ponents. We could argue extensively about where to locate the microphone(s) at
the muffler exhaust. Nevertheless, the results of testing on the facility may
be worthy of note. I should also note that results of the bench test program
were published in SAE Paper 771A, dated October 1963, and entitled (appropri-
ately enough), "A Bench Test Facility for Engine Muffler Evaluation", by I. J.
Schumacher, 0. R. Sparks, and D. J. Skinner.
The first step in the program was to field test some half dozen different
engines, and 47 standard design mufflers from some 6 or 8 of the major sup-
pliers of mufflers for the MIL STD engines. This testing provided a data base
on the noise from the various standard engines with exhaust sizes ranging from 1
1/2 to 3 inches, data on the performance of various muffler designs (see Table
I), and data on the sensitivity of results to operating conditions.
From.this point work turned to the designing of a prototype facility, and to
developing techniques whereby facility data might be used to imply how a muf-
fler might perform on an engine, or at least show a means of differentiating
between obviously good and obviously bad mufflers for the application intended.
It was also recognized at this point that the facility had to be fool-proof in
the sense that "gimmicked" mufflers could not be designed which would show up
well on the facility but which would not work well on the engines (either be-
cause of noise or performance problems).
DESCRIPTION OF BENCH TEST COMPONENTS
A photograph of the first prototype of the bench test facility is shown in
Figure 1, and a schematic is shown in Figure 2. It may be seen that in addition
to its noise testing feature, the facility includes provisions for making both
static and dynamic backpressure measurements on the test mufflers at various
flow conditions. In order to optimize upon both the mechanical and operational
aspects of the facility and its component parts,, comprehensive studies were
made of these parameters in order to assure an optimum compromise between facil-
ity reliability and operational simplicity. Discussions of the major compon-
ents and the tests used to define their operational characteristics are pre-
sented below.
Siren Noise Source - The heart of the acoustic system is the siren exci-
tation source, shown at (1) in Figure 2. This siren produces wide band:, al-
most "white" noise and is a constant power source by virtue of the near crit-
ical pressure drop across it. This high impedance noise generator is used in-
stead of more conventional voice coil devices in order to simulate the impe-
dance characteristics of an engine noise source and 'thereby simulate loading
effects experienced when an exhaust system is attached to an engine noise
source. Discussions of performance testing of this device are given in the
following sections.
Manifold System - The second important component of the facility is an
acoustic conduit system which serves to couple test mufflers to the siren and
represents the manifolding system of an engine. For some types of testing,
this component is dispensable, and useful evaluation data can be taken without
146
-------
it. It serves chiefly to bring the absolute magnitude of the noise reduction
more in line with numerical data obtained in the field. For facility quali-
fication tests, this manifold is a specially designed piping component as shown
in Figure 2. For other tests involving the design of special purpose mufflers,
or for evaluating performance for a particular end-item application, excellent
correlation with field data can be obtained by using the actual engine exhaust
manifold.
Effect of Siren Pressure and Speed - A series of tests were conducted on
the wide band siren to evaluate the effect of operating pressure and speed.
These tests showed that the siren operates well at pressures from 2 psi to at
least 15 psi. The generated noise output varies directly with the source pres-
sure although the spectral distribution is essentially constant. The siren
operating speed has a decided effect on the spectral output of the siren. It
has been designed to produce wide band noise above 40. cps while operating at
approximately 240 rpm. At speeds above this level, the low frequency output
falls off markedly.
Microphone Position - Extensive tests were made on the piping configur-
ation for each size of muffler to evaluate the effects of microphone position.
A comparison of muffler performance characteristics measured at various micro-
phone positions show correlation is quite good so long as the microphone is lo-
cated in the acoustic far field. The exact position of the microphone is not as
important if one position is selected as a standard for each muffler size, and
so long as the microphone is not in the direct noise jet. Based on these tests
the microphone location was set at 45 deg. from the center line of the outlet.
Effects of Gas Temperature - The effects of gas temperature on muffler
performance are primarily i'n two areas:
1. Acoustic velocity varies directly with the square root of gas temper-
ature, and thus the cut-off and band-pass frequencies of a given muffler shift
in essentially the same proportions.
2. Gas viscosity increases with the temperature and thus dissipation
elements are generally more effective at elevated temperatures. .In general,
this means that the percent damping of each muffler will go up as temperature
increases (that is, the Q will decrease).
Test results showed that the measured octave band noise reduction character-
istics of the experimental mufflers differed slightly when measured with high
and low temperatures. As anticipated, the results showed that an increase in
cut-off frequency was experienced at high temperatures (450 F air temperature)
as well as a slight increase in the high frequency attenuation characteristics.
The use of high temperature air showed no particular advantage as far as dif-
ferentiating between high and low quality mufflers and as such did not warrant
the added complexity to the facility.
High F1ow Tests - A series of tests were conducted to evaluate the neces-
sity for and the effect of high flow through the muffler during acoustic tests.
tests. The most pertinent results from these facility tests conducted on all
three muffler sizes show that the quality mufflers can be conveniently differ-
entiated from the low quality or empty sirens without reproducing total muffler
147
-------
flow velocities experienced on the engine. Based on these tests no appreciable
improvement was realized from the acoustical tests conducted under high flow
conditions and as such, this requirement was excluded on.the facility design.
DESCRIPTION OF FACILITY MUFFLER EVALUATION TECHNIQUES
The output spectrum of the wide band siren is shown by curve A in Figure 3.
Shown by curve B on this p.l ot is facility unmuffled output with a typical engine
manifold attached to the siren. If now we superimpose on this plot curve C,
which shows output noise of the siren-manifold facility with a muffler attached,
the difference between curves B and C represents the noise reduction afforded by
the muffler. Since the siren is designed such that each octave interval shown
is rather completely filled with generated noise, specially tuned muffling de-
vices (as contrasted to high quality mufflers) may be shown to be relatively
ineffective in reducing total noise, and a numerical rating of noise attenua-
tion can be ascribed to each test muffler on the basis of the octave band noise
reduction measured.
In order to relate the octave .band noise reduction figures obtained from
the facility to muffler quality or loudness reduction, one must compensate for
the variation of ear sensitivity with frequency, and the dependency of this
frequency variation with absolute amplitude. In the program described, final
evaluation of muffler quality was based upon the reduction in sone loudness
afforded by a muffler when its decibel noise reduction properties are super-
imposed upon a typical engine noise spectrum. In order to illustrate both the
concept and the procedure involved, consider a muffler with facility-measured
decibel noise reduction properties as shown in Figure 4. If now we consider
that the unmuffled exhaust noise spectrum shown as curve A in Figure 5, is typ-
ical for engines which might use this muffler, we can attest quality of the test
muffler by computing the drop in loudness level (in sones) that the db noise
reduction of the muffler would produce when superimposed upon this spectrum. If
we graphically subtract the noise reduction figures from the engine noise spectrum,
we get the predicted muffled noise spectrum shown by curve B. When each of these
curves is converted to SAE sones, then the resulting tested quality of the muf-
fler is the difference in these sone levels. For convenience the sone loud-
ness scales are plotted directly on the octave ordinates of Figure 5, and it may
be seen from the nonlinearities of the scales that reduction in some of the
octaves is more important than in others insofar as loudness (sone) reduction is
concerned. In order to supply proper weighting to the reduction values obtain-
ed for each of the octaves, some typical engine noise spectrum must be -used.
In order to determine the final evaluation factor for each muffler subject-
ed to these tests, one needs merely to sum the sone reduction afforded in each
octave, or alternatively subtract the total calculated muffled sone loudness
from the sone loudness of the reference engine spectrum shown. The engine spec-
trum used is not critical, as variations in the band levels used as reference
have a second order effect on the octave band weighting factors used.
It may be seen that the process described above involves first of all, the
derivation of octave band noise reduction from the bench test facility, and
then the weighting of each of these noise reduction figures based upon noise
148
-------
conditions typical of those to which the muffler might be subjected in field
service. The entire process may be simplified considerably by graphical tech-
niques using the sone evaluation chart shown in Figure 6. This chart again has
the eight octave band ordinates. Measured muffler noise reduction values may be
plotted directly upon the ordinates, and corresponding values for sone reduc-
tion may be read directly. The typical engine spectrum weighting factors are
automatically included in the loudness reduction (db) figures on each ordinate.
To evolve the muffler quality factor (the sone reduction value) using this chart,
the process is as follows:
1. Obtain octave band NR figures for the test muffler from tests on the
bench test facility.
2. Plot these decibel values on the db ordinates in Figure 6.
3. Read the corresponding sone reduction figures from the right hand
scale of each ordinate.
4. Take the algebraic total of all inferred octave band sone reduction
values. This is the quality factor of the muffler.
After design and fabrication of the bench test facility shown in Figure 1,
an extensive series of tests were conducted on a series of mufflers with 1-1/2,
2, and 3 inch inlet sizes. It was shown that when a sophisticated simulation
of the exhaust system was utilized (for example, using the actual engine man-
ifold between the siren and muffler), facility tests ranked quality mufflers in
virtually the exact same relative order as engine tests. Such numerical cor-
relation is illustrated graphically in Figure 7, where loudness ratings from
field data on the 2 inch test mufflers are shown as the center ordinate, and
facil'ity rankings using two sone calculation techniques are shown on either
side. It may'be seen that both field and facility tests rate the mufflers in
virtually the same order, and that the facility easily differentiates the more
quality mufflers (B-12 through B-21) from the empty shell (6-11).
Similar tests, but using a different manifold were shown to rate the series
B-12 through B-21 in a different relative order, but they were still easily dif-
ferentiated from straight pipe sections or empty shells. Since the objective of
this development was a device to attest general muffler quality for use with a
variety of manifolds, the standardized manifold was adopted. The entire system
was thereby shown to be effective in differentiating between quality and non-
quality mufflers on a rather general basis.
MUFFLER BACK PRESSURE EVALUATION
The back pressure characteristics of the military standard mufflers is per-
haps the most important single evaluation criterion for most end-item applica-
tions. Since the military standard muffler design is not tailored to a specific
application, a compromise in the noise reduction characteristics was favored to
meet the maximum back pressure limits. An extensive series of tests were con-
ducted on the mufflers under a variety of both steady flow pulsating conditions.
Data were recorded using both a water manometer and a flush-mounted pressure
transducer, and were compared with field data obtained with a flush-mounted
149
-------
transducer installed in the engine exhaust system. The results showed that
under steady flow facility conditions (with siren off), excellent correlation
was obtained between field results and facility results using either a flush-
mounted transducer or a water manometer for facility measurements. The data
also indicated that full engine flow rates need not be simulated to perform
these tests and that the amount of flow required is dependent only upon the
resolution of the back pressure measuring system. Comparatively high flow rates
(240 scfm) are required for the large size mufflers in order to obtain necessary
reading accuracy when a water leg manometer is used. Alternately, lower flow
rates could be used with a more sensitive pressure transducer, but this system
would suffer from the complexity of calibration and data interpretation. The
correlation of steady flow back pressure measurements recorded on the facility
to engine back pressure data obtained during the field tests is presented in
Figure 8.
ANALOG SIMULATION TECHNIQUES
Another means for evaluating engine mufflers, at least in the difficult low
frequency portion of the spectrum, lies in electronic analog simulation of the
proposed muffler-manifolding configuration. The most sophisticated and well-
documented basis for this contention in the SGA Compressor Installation Analog,
developed and 'operated by Southwest Research Institute for the Southern Gas
'Association's Pipeline and Compressor Research Council (See Figure 10). While
the primary purpose of this analog is.to simulate pulsations in the piping sys-
tems of reciprocating compressors (to date some 3000 such studies have been
conducted), it is also useful and has been used as a tool for design and eval-
uation of engine muffler and exhaust systems. Using this analog, the total flow
characteristics (steady state and transient) of a piping system such as a muf-
fler and exhaust, system can be modeled using electronic delay line elements
which are simply coupled together to simulate the acoustic impedance network of
the exhaust system regardless of complexity. Lumping lengths can be chosen
arbitrarily short to accomodate whatever upper frequency limit is desired, but
pipe diameter does impose some upper frequency limitations. The simulation as-
sumes one-dimensional compressible flow, and is therefore limited in applicability
to frequencies whose wave lengths are large compared to pipe diameter. For a six
inch exhaust system, therefore, the upper frequency limit is on the order of 500 Hz.
It is readily noted, however, that it is precisely in the low frequency
ranges where muffler performance is difficult to predict analytically, and
where piping interaction effects are most important on muffler performance.
High frequency attenuation is relatively easy to achieve in a muffler, and once
low frequencies are controlled, the high frequencies normally take care of them-
selves. Standard acoustic theory ('viz. lined duct absorption effects) serves as
an adequate tool to design additional high frequency attenuation if it should be
desi rable.
The process of simulating an exhaust system on the analog is a relatively
straight-forward impedance simulation using a series of analogies where voltage
represents pressure (AC and DC), and current represents mass flow.
If we start with the equations of.motion, continuity and state for one-
dimensional , isothermal, compressible flow, and compare these to the electrical
150
-------
delay line equations, we find that a very convenient set of analogies occur
wherein
'Electrical Inductance a Acoustic Inertance
Electrical Inductance a Acoustic Compliance
Electrical Resistance cc Acoustic Damping.
Specifically, the electrical parameters of inductance (L), capacitance (C), and
resistance (R), per unit length of pipe are:
r — v •
<-* ~~ NO
C2
and
R = K3 M
where
p = flowing density
A = pipe fl ow area
c = acoustic vel ocity
M = mass flow rate
K = .constant
Using acoustic theory the same set of equations are derived, except that
the resistive term is assumed linear of the approximate form
R - 1.42 1/2 ^
irr
as contrasted to the fluid dynamic viscous resistance which is of the form
fc2
R = Ko -iS-* M
Considerable experimental work has been conducted to evaluate the relative mag-
nitude of the two resistive mechanisms, and results show that for all pipe
sizes of practical concern (i.e., larger than capilary tubing) and for all flow
rates on the order of several fps or greater, that the fluid dynamic term pre-
dominates. Thus for most systems, the non-flow acoustic resistance mechanisms
(e.g.. molecular relaxation) can be ignored with negligable effect.
It may be seen by inspection that of the three basic impedance terms defined
151
-------
(R, L and C) both L and C are quite linear with flow. Since these two parameters
determine electrical (and acoustic) propegation velocities, an excellent simula-
tion is achieved of muffler attenuation rates, cut-off frequencies, internal reson-
ances or pass-bands, and interaction frequencies caused by attached piping. The
only parameter undefined by R and C is the amplitude of the various resonance
peaks which are controlled by resistive damping. Since the R is non-linear with
flow, simulation can be achieved either by inserting nonlinear resistance circuits
into the delay lines, or by linearizing the R for the average mass flow rate M.
Experience with many simulations have proven either approach is adequate.
The question which usually comes up at this point is "What about perfor-
ations". Again, both analytical and experimental data shows that for non-flow
acoustics, perforation size must be quite small before the elements become re-
sistive rather than reactive. In Figure 11 perforation Q is plotted as a func-
tion of hole size for various frequencies. Mote that hole diameters must be
less than a quarter inch before the R predominates (i.e., before Q<1).
In the case of flow through perforations, analog data has been compared ex-
tensively with laboratory and field data, and again the results show that the
predominating effect in achieving pulsation damping is the same mechanism which
produces pressure drop. Specifically, the dynamic (acoustic or pulsation re-
sistance) is numerically equal to twice the steady state resistance, i.e.,
AC ~ DC ~
M steady flow
Using this approach, excellent correlation has been obtained between the an-
alog and field data for perforated element acoustic filters. An example is
given in Figure 12 which shows the pulsation spectrum from 0 - 100 Hz for a re-
ciprocating compressor. More specifically, the data shows the envelope of pul-
sation amplitudes as compressor speed varies over a range of ± 10%.
Again, the problem of using such a device for evaluating mufflers lies in
the question of what constitutes quality in a muffler. Although the analog will
accurately map filter attenuation as a function of frequency, including all pass-
bands and interaction effects of attached piping, the noise reduction data ob-
tained is for that particular exhaust system. If significant changes are made in the
manifold, tail pipe, etc., then data can be modified substantially. Figure 13 is
one example of analog data taken for a proposed muffler design for a large
stationary natural gas engine. Note that noise levels and spectra can be ob-
served anywhere in the system, but that as the piping configuration is changed,
output noise from the muffler will likewise change.
152
-------
FIG. 1. FIRST PROTOTYPE OF MUFFLER BENCH TEST FACILITY
153
-------
Table 1 - Field Results From Engine Tests on Experimental Muffler.
Engine Exhaust Size - 3 Ln.
Muffler No.
Noise Level, db
Loudneis, Sones
Engine
Noise
83
28.6
Open
Exhaust
105
102.2
A-l
104.5
93.8
A-2
101
68.4
A-3 A-4
A-5
103.5 100.5 102.
89.3 71.1 83.3
A-6
101
A-7
A-8 A-9
100.5 102 102.5
73.5 75.3 77.6 83.8
Engine Exhaust Size - 2 in.
Muffler No.
Noise Level, db
Loudness, Sones
Engine
Noise
76
19.8
Open
Exhaust
95.5
53.2
B-I1
92
38.3
B-12
94.5
45.9
B-13
94
42.7
B:14
94
46.7
B-15
93
39.6
B-16 B-17
86.5 89.5
28 .2 40.2
B-18
94.5
46.1
B-19
96
4S..7
B-20 B-21
95 93
45.1 36.<)
Engine Exhaust Size 1-1/2 in.
Muffler No.
Engine Open
Noise Exhaust C-23 C-24 C-25 C-26 C-27 C-28 C-29 C-30 C-31 C-32 C-33'C-34
Noise Level, db 74 94 88 89 85 88 90 91.5 91 88. 86 87 38.5 88
Loudness, Sones 17.4 37.6 28.1 31.2 26.7 29.7 29.4 33.1 31.6 39.0 24.8 28.5 30.1 24.5
SONICALLY CHO*ED
CONSTRICTIONS FOR
BACKPRESSURE TESTS
0.5'
1PHONE 25*
MICROPHONE
MANIFOLD
OCTAVE
BAND
ANALYZER
' —
SOUND
LEVEL
METEfl
—
PREAA1P
FIGURE 2
Schematic of Bench Test Facility
154
-------
f
0 73 ISO iOO SCO I2OO 24OO *8CO
5 ISO JOO. SCO 1200 24OO «00 ABOVE
OCTAVE BAND - tn
FIGURE 3
Octave Band Analysis of
Facility Noise Characteristics
_2S
20
75
75
ISO
150 JOO SCO I2OO 24OO 48CO
JCO SCO I2CO 2400 «00 lOKc
OCTAVE 3AWOS-CPS
FIGURE 4
Octave Band Analysis of Noise
Reduction Figures from. Typical
Experimental Muffler
90
30
OPEN STACK ENGINE. NOISE
NOISE REDUCTION
EQUIVALENT OUTPUTX« .
NOISE
- -4
s
ZO 75 I5O- 300 6OO I2OO 24CO 4&3O
7S ISO JOO 600 I2OO 2*OO 48OO IOKC
OCTAVE BANDS-CPS
FIGURE 5
Graphic Example of the Effect
of Muffling Action upon Exhaust
Noise Lo-udness
a
a
•
§20-
WO
So
23
O
UJ
Z 10-
OISE
FICATION(-)
..?. . ..1
Zj
a.
2 20-
i)
UJ
z
§OQ
a
.
-» M-
•
.
-ij _o_
.
-
10-
-4
-Z
-a
-10 ,0-
rw
I
J-.5O2O-
5-40
n
u
i -
n a
-" 30-
-so
- 20-
" :
-w 10-
-
:
-K3 "
-2S
-4'
"I
»
a «
is c
40-
.
30-
-
_Q
J
-4 ;
-4 "
•
-s
I"3 •
rw
p30jo-
n
ij
z
3 a
/i a
(4 40-
.
30-
12
20-
K)
«-
-3
r-10
IO-
-20
-3S
10
7)
u
6 a
9) O
J
.
30-
-%
20-
-4
- w-
-4
:,
L
f10 10-
1-20
J.
^-JO 20-
l/l
§ 03
l/l Q
~*
30-
~
"
-* 20^
-4
-t
1' ~
do
•(
fe-
Crt
Ld
O B
•VI Q
-J.S jo.
;
-1 20-
;
'. .
-1 "
-2
-4 O-
r«
Ml 20-
Ll4
n
j
2
3
/>
-2.S
-2
-I
-1
-2
r'4
I
FIGURE 6
Evolved Sone Evaluation Chart for
Direct Evaluation of Muffler Quality
from Test Bench Data
155
-------
120-
110-
a-ii
VI
m
^^
II
^
°u.
o° ;
t
|
t
H- FIELD OATA
FACILITY CATA — _.
t
f
t
ff
1
•
•24
•202
» UJ
I«
•l6^
Si
12 JS
J^
a ^
uz
z~
* z
u
A-l L-Z A-5 A-6 A-7 A-3
EQUIVALENT
FIELD LOUONESS
USING FACILITY
REDUCTION OATA
SAE SOMES'
FIELD LOUONESS
STEVENS SONES
8-16
EQUIVALENT
FIELD LOUONESS
USING FACILITY
REDUCTION OATA
STEVENS SONES
FIGURE 8
Comparison of Field and Facility
Muffler Backpressure Ratings
FIGURE 1
Comparison of Field and Facility Eval-
uations of Muffler Performance
FIGURE 9
Final Prototype Muffler Bench Test
Facility
156
-------
FIGURE 10
Electroacoustic Analog for Simulation of the
Acoustic Response of Piping Systems
-------
158
-------
Horiz.
Scale
Freq.
10 Hz
per div
Speed
Range
400-600
rpm
Field Data Analog Data
Comparison of Analog anu Field udta on the Performance of a
Perforated - Tube Pulsation Filter
FIGURE 12
159
-------
Muffler and Exhaust System Performance Characteristics
as Recorded on t'fie Electroacoustic Analog
FIGURE 13
160
-------
Comments on Evaluation Techniques
of Exhaust System Noise Control
Characteristics
D. W. Rowley
Donaldson Company, Inc.
Before discussing possible exhaust system bench evaluation techniques as
charged by Dr. Roper in his introductory comments yesterday, let me first
state my vantage point. In the area of surface transportation noise con-
trol, Donaldson is a manufacturer of both induction and exhaust system
products for medium and heavy duty trucks ... primarily intake air cleaner-
silencers and exhaust mufflers. Donaldson also provides products for
recreational vehicles, light aircraft, and for railroad locomotives.
This morning I would like to discuss with you those steps we find necessary
to insure ourselves and our customers that the muffler and exhaust system
for a given truck and engine indeed do the job for which they were intended.
Primarily I'll be speaking toward the heavy duty, diesel truck.
I'm going to review "how we do the job of developing hardware and then its
evaluation." To this pojlnt in the symposium, most of the speakers have
been heavily concerned with non-engine, bench test, acoustic theory. Well
now we're going to spend a few minutes concentrating on the real world of
engines, trucks, and their exhaust systems.
First, when a request is received for a given job, it's worthwhile to
determine if a suitable product is already in existence. For this a catalog
or recommendation sheet may be referred to, Fig. 1. The data shown is from
actual engine testing. Note that the performance of a particular product
depends on the engine and the exhaust system with which it is used.
If a muffler with the desired configuration and performance cannot be found
in the recommendation sheets, a computerized selection program may be used.
The program consists of two major listings. The first describes the flow
161
-------
and acoustic characteristics of approximately 135 engines, and the second
describes the flow loss and noise control properties of our standard line
of truck mufflers -- about '80 models are included.
By inputing the engine and truck type and the exhaust system to be used,
the computer will "match" the two lists, perform the required calculations,
and "select" those mufflers most applicable. Performance is predicted in a
form similar to the recommendation sheet. The accuracy of the prediction
is within 3 dBA of actual engine-dynamometer tests. It is also possible to
select a given muffler and predict that muffler's performance on all engines
for which it will "fit" backpressurewise.
These methods have been reviewed because either could conceivably be used in
a labeling scheme, -but please remember their accuracy, and again note, they
depend on engine-dynamometer testing as well as flow bench pressure drop
data for'a basis.
If a suitable product is not available, a development program must be
implemented. The design, and analytical stage involves utilization of math
model analysis techniques to provide an estimation of the muffler's trans-
mission and insertion loss. Next samples are obtained and evaluated. First,
the samples are tested on a flow bench to determine if flow pressure drop is
satisfactory. If OK, "non-engine" acoustic bench testing is then used to
evaluate the acoustic performance of the muffler and exhaust system. For
this loud speakers, sirens, shock tubes, air reinforced electrodynamic
speakers -- all have been employed. M.iny of these methods are worthwhile
development tools. They can, if properly utilized, rank mufflers by
performance quite effectively ... some methods better than others. The
closer to the actual exhaust system conditions, the more accurate the
ranking.
162
-------
To do a good job of evaluation on a "non-engine11 bench test, one must
somehow simulate actual engine exhaust system conditions of;
• Gas flow, temperature and temperature gradient down the exhaust system.
• The total exhaust system must be used: exhaust pipe and silencing
devices, connecting pipes and tailpipe, and probably most difficult,
something to simulate engine impedance.
• Generation of noise with a similar spectral content to the engine of
concern, and
• Of high enough amplitude (140 - 170 dBA) such that non-linear acoustic
conditions exist. Non-linearity cannot be ignored since it can
significantly affect acoustic velocity ... especially in a naturally
aspirated engine.
A large amount of complicated material to attempt to handle.1 Perhaps
someday it will be possible, but at the moment we can't do it with any-
whdre near the accuracy required.
Frankly, it's easier to "obtain an engine, provide adequate control measures,
and perform the tests on the actual engine and exhaust system. This in itself
is quite demanding. The engine must be right. It must have proper fuel and
intake air flow, with rated power out-put and normal exhaust gas temperatures.
A top-notch technician to perform the test is a must, along with equally top-
notch instrumentation.
We're almost ready to talk about engine test data, but first let's define
exhaust noise. Fig. 2 is an illustration of exhaust noise ... being made
up of tailpipe discharge noise, muffler shell noise, exhaust pipe surface
radiated noise, and also the noise transmitted through any leaks in the
exhaust system. At the bottom of the figure is a typical example of the
levels of these subsources required for 1978 trucks.
163
-------
Let me explain. Although the manufacturers are faced with meeting an
83 dBA overall truck level, their prototype truck design goal, because
of regulated test methods and manufacturing variations, is from 80 to
81 dBA in order to be safely under the 83. And since it is oftentimes
desirable to reduce exhaust noise so that it is essentially a non-
contributor, the goal for exhaust noise becomes 10 dBA less ... or the
low 70's. This in turn then requires the very low values shown for the
subsources.
Now as we look ahead to the 80 dBA 1982 truck, the subsources will become
that much more difficult to control to the very low levels required,
Fig. 2.
The subsources can in turn be broken down ... sub-subsources, as presented
in Fig. 3., The tailpipe discharge noise is made up of the exhaust noise
created by the engine that escapes through the muffler and is radiated out
the tailpipe. It also includes muffler generated noise caused by gas flow
through the muffler, and "jet" noise created by high velocity exhaust gases
escaping into the atmospTierel
Exhaust pipe surface noise is caused by the high internal dynamic pressure
within the exhaust piping.
Muffler shell noise isn't as straight forward as it might appear. It's
mainly caused by the internal pressures within the muffler, but it also
radiates engine and chassis vibrations that are transmitted to it via the
exhaust system. The muffler surface can also radiate exhaust pipe vibra-
tions as set up bv the internal dynamic pressures.
Now with that background, let's get into engine testing. Fig. 4 presents
50 ft. exhaust noise from a fully loaded engine. The information was
gathered by isolating engine mechanical noise by using a full enclosure
164
-------
and a heavy isolation wall. The wall is acoustically treated on the
outside, creating a free field above 150 hz. The data in the figure
is within 1 dBA of a completely free field over a reflecting plane.
This particular engine is rated at 2100 rpm. The engine is warmed up
and set to full load at 2100 rpm. The exhaust system is allowed to
stabilize at operating temperatures. Under these conditions much of
the analysis work is done ... spectrum, octave band, wave shape, and
the muffler internal elements are evaluated. In this particular case,
a 72 dBA would be reported at full load and rated rpm. Then the
"lug-down" mode is run. For this, load is taken off the engine until
it speeds up against the governor. In this case the governor is controlling
the engine rpm to 2400. Then load is slowly added, such that the engine is
"lugged" down through its operating range to approximately 2/3 rated rpm.
The 2/3 is important because of the agreement with the SAE 366b drive-by
test. Only one serious peak was found ... 75 dBA at 1500 rpm which would
be reported accordingly.
One other test mode is considered, Fig. 5. This is the sudden acceleration,
run up, goose, idle-max-idle.(IMI), or whatever. Notice the differences
from the lug mode.. Values of 73 dBA at 1700 rpm and 73.5 at 2250. Both
would be reported.
There is yet another test mode required ... one that will show the effect
of temperature on system performance. Surface radiated noise becomes of
more importance as muffler attenuation increases. Surface noise is a
function of the tempera.ture of the exhaust system parts. If the surface
is cold, it is more "live" (high Q) with a resulting greater surface
radiated noise. This is demonstrated in Fig. 6, Muffler, and again in
Fig. 7, Exhaust Pipe. These are copies of the actual work sheets. Note
the difference between stabilized conditions in the exhaust system and
cool conditions ... approximately a 5 dBA difference for the muffler, and
about 7 for the pipe ... quite considerable.
165
-------
In essence, five or six pieces of peak data are recorded. Obviously we're
looking for the worst condition. That's the condition very probably that
the truck manufacturer would run into, or possibly could run into, as he
evaluates his truck.
Pipe surface noise was further investigaced as a function of time, Fig. 8.
A 55 dBA can be seen for pipe surface radiated noise at idle, 500-600 rpm.
Then the throttle was punched wide open creating an exhaust noise peak of
78 dBA. As the momentum of the engine is overcome, the level drops down
to 65. At that point, load was put on the engine. Immediately, the pipe
surface noise went up to 75 dBA and then as.the system absorbed heat and
the temperature of the material increased to a stabilized condition, the
pipe noise likewise decreased.
The purpose of presenting the last series of figures was to provide some
indication of the difficulty of rating system performance even while
testing with the actual engine and system.
Now let's look at-other problems of evaluating systems ... in this case
distributed systems, Fig. 9, which are becoming more popular in the
industry. Distributed systems contain more than one silencing device.
These additional components are acoustically interrelated with the primary
muffler and one another. That is, the performance of the primary muffler
is affected by other devices in the system, and vice versa. Fig. 10 is
further evidence of this. Consequently, it's very difficult to say this
particular muffler or silencing device has such and such acoustic
characteristics without referring to the performance in an actual system.
The "whole" system must be evaluated.
With the complete data from an engine-dynamometer test, we have a pretty
good handle on the performance of the exhaust system on a given engine;
166
-------
but, we're still not completely convinced. So the next step obviously is
going to a truck, which is the real "proof of the pudding" (includes truck
noise source identification). The type of data gathered from a truck test
is shown in Figs. 11 and 12.
By utilizing the type of testing just reviewed, we try to meet our
committment to the truck manufacturers and the trucking industry ... striving
to make certain that the exhaust system controls the noise as intended and
without compromising engine performance. It is also required via testing
to provide proof of conformance to manufacturers' specifications.
In conclusion, any evaluation method selected must meet certain degrees of
accuracy. The lower the overall truck noise levels established, the more
sophisticated the mufflers and other silencing components will become; and
it follows, the more critical the accuracy of evaluation also becomes. As
of this point in time, this can best be done with an engine-dynamometer
type of test.
Presented at: EPA Surface Transportation Noise Symposium
Chicago, Illinois
October 12, 1977
Reference: SAE Paper No. 770893. "Exhaust System Considerations for
1982 Heavy Duty Trucks."
DWR/dp
167
-------
Figure 1
168
-------
Discharge Noise
EXHAUST NOISE SOURCES
* TYPICAL TC ENGINE •
Muffler Shell Noise
Leak -Noise
fllfflfllfl
"Exhaust \
' \ -••"• •••.;;:,,••;-. ;y". '|«Noise ^ . . ,. .
•"-. •• . ,...;••• •'.. ••'• .-•-. .•, .-v- •;; ,. "<-\ * • _ .
••"."' ' ° ...''' O •
Discharge+MuffleM-Leak+Pipe noises =Exhaust Noise
Example. '^TSTtuck: ^0*66^^62+62 ^"72J5 dBA
lExample. *82Truck: ;61*59+55*59 -65d8A
Figure 3
-------
'
Tail 'Pipe nisriharnft'
, "^ .. • ''^ • ,
- :Exhaust noise ^
Mi if f Ipr gpnerateH
Exhaust Pipe ^•irfanftlNQise
Muffler Shell Noise
ilntemal ifnuf f ler pressures (SRLim^^ _^_3
fEngine /^hasis^braW
"'"Vx - .. .„ -- " '-*%^ • . - --.^^ -w - - -: - . -r^^ --w • ^ ,
1$ lExhaustp
Astern p^ak INt>ise
KiP::;^v-fc-^^i:v^^ ^^:--r^:-^
Figure 3
-------
III
VS ;RPM
weak 'Torqtie
JRatedJRPM
and H.P.
*2/3 tRated RPM
Lug'RPM (x 100)
Figure 4
-------
Figure 5
-------
MUFFLER EVALUATION
SPL vs RPM
DD8V-71
DHV
3-18-77
dBA
at 50'
Cold
Cold
Warm
Hot
Hot
24 27 12
Figure 6
-------
Figure 7
-------
PIPE SURFACE NOISE
SPL vs Time
(Cold start -110°)
'Run-up
2100
EL —
Idle
Idle
Figure 8
-------
-------
Figure 10
-------
Figure 11
-------
Figure 12
-------
A BENCH TEST FOR
RAPID EVALUATION OF MUFFLER PERFORMANCE
by
A. F. Seybert
Department of Mechanical Engineering
University of Kentucky
Lexington, Kentucky 40506
181
-------
INTRODUCTION
The United States Environmental Protection Agency has published
general provisions for nois.e labeling standards [1]. Among other
things, these provisions indicate the need for test methodologies
for the evaluation of the acoustic characteristics of products to
be labeled. This paper discusses some of the problems associated
with the prediction of exhaust system performance and presents a
novel technique for the measurement of muffler characteristics.
It is shown that exhaust system performance can be predicted using
measured muffler characteristics in conjunction with other known
information such as engine impedance and pipe lengths.
BACKGROUND: FACTORS INFLUENCING EXHAUST SYSTEM PERFORMANCE
Figure 1 shows some of the factors influencing overall exhaust
system performance, where "performance" can be measured by some
acoustic descriptor such as the sound power radiated by the tail
pipe outlet or the sound pressure at some point in space at a fixed
distance from the tail pipe outlet. There seems to be mild confusion
and some misunderstanding within the automotive industry on how the
factors in Figure 1 interrelate in determining overall exhaust
system performance. Yet, it is essential that we understand these
effects if we are to develop a rational, workable test methodology
suitable for muffler labeling. For example, if we know quantita-
tively how engine source impedance and source strength affect exhaust
182
-------
system performance, we may possibly develop a bench-test methodology
in which the engine is replaced with an electronic noise source
such as an acoustic driver or loudspeaker. The data obtained from
the bench test would be used to predict the overall exhaust system
performance for any engine for which source impedance and source
strength information are available. In a similar way we would like
to account for variations in exhaust and tail-pipe lengths in order
that a standard pair of pipes can be used for the bench test. Thus,
by increasing our understanding of exhaust system behavior, we can
develop a simplified test methodology suitable for muffler labeling.
We can divide the factors listed in Figure 1 into two categories
factors that can be accounted for using proven acoustical theory,
and factors that must be accounted for with empirical data. Source
impedance and source strength are examples of the latter category.
On the other hand, pipes are classical acoustical systems, and the
effect of pipe length and diameter on sound propagation and radia-
tion is well known.
In general, muffler characteristics must be determined imper-
ically, except for very simple geometries, in which case analytical
results are reasonably accurate.
EXHAUST SYSTEM MODELING
Exhaust system modeling has evolved over a period of about 50
years since Stewart [2] analyzed muffler systems using lumped
parameter approximations.* Davis et al. [4] made significant
advances in exhaust system modeling by applying traveling-wave
techniques to evaluate expansion chamber and side-branch
*Crocker [3] has recently reviewed exhaust system modeling.
183
-------
configurations. Following this v/ork, Igarashi [5] applied electrical
four-pole techniques to exhaust system modeling. Recent developments
in exhaust system modeling are reviewed by Sullivan [6].
The four-pole theory used by Igarashi is .very powerful and easy
to apply, and seems to be an ideal method for exhaust system design.
Four-pole theory is based on the concept that in any linear, invariant
system the input and output quantities can be related by four
"system" parameters, called the "four-pole parameters." As an
example, consider a straight section of pipe of length L and cross-
sectional area S, Figure 2. The input and output quantities are
the acoustic pressure and volume velocity at each end of the pipe.
The expressions relating these quantities are:
VailP2
Vl=a21P2 + a22V2
where P.. and V are the acoustic pressure and volume velocity at
the pipe entrance, and P and V are the acoustic pressure and
volume velocity at the pipe exit. The four-pole parameters for the
pipe--a , a'-<2' a?i' anc^ a» --are functions of frequency, pipe
diameter, and pipe length:
a =cos kL a =(pc/S)jsin kL
a2.1= ^cs)1 Gin kL a =coskL
184
-------
where k=2irf/c, c is the speed of sound, n is the density of air, and
j denotes -imaginary quantity.
For complex acoustical systems (e.g. a muffler) the four-poj_ed
parameters can be computed from measured impedances. It can be
shown [7] that the four-pole parameters are related to the driving
point and transfer impedances:
Z11/Z'12 a!2~(ZllZ22 Z12)/Z12
a,,- — 1 / Z, „ a — Z „ „ / Z „
22 12 2.2. \_2.
where Z •, and Z are the driving point acoustical impedances looking
into the acoustical system at the entrance and exit respectively,
and Z „ is the transfer impedance (defined as the ratio of the
acoustic pressure P at the entrance to the acoustic volume velocity
V at the exit). If we can measure the impedances of a complex
system, then we will have the four-pole parameters for the system.
The four-pole theory is useful in combining 'acoustical sub-
systems, such as mufflers and pipes, to obtain overall system
performance. This can be illustrated by representing an exhaust
system in terms of four-pole parameters as shown in Figure 3. In
Figure 3, Z is the engine source impedance and V is the engine
source strength (the acoustic volume velocity of the engine). The
various subsystems are represented by cascaded four-pole, parameters,
and Z is the radiation impedance of the tail pipe. For the four-
pole model shown .in Figure 3, V , Z , and the muffler four-pole
parameters must be obtained empirically; but the four-pole parameters
185
-------
for the exhaust and tail pipes ace given in Equation 3. The radiation
impedance Z is knov/n from theory [8] .
Equations 1 and 2 can be written in matrix form:
pl
KJ
=
all a!2
.S21 322.
'P2
V2\
=
A
'P2
V2.
(5;
Likewise, the relationship between acoustic pressure and volume
velocity at the entrance and exit of the muffler can be expressed as
"P2"
Lv2J
=
bll b!2
Lb21 b22-
TPJ
kr
B
P3~
W
(6!
Equations 5 and 6 can be combined:
LV1J
A
B
V.
This process can be continued to yield
1 = D
V.
V
186
-------
where
D
11
12
'22]
A
B
Because the .four-pole parameters for A, B, and C are known (either
from theory or experiment), the overall four-pole elements of the
matrix D are also known. We can rewrite Equation 8 as:
d!2V4
d22V4
We.know also that P./V.-Z and V =V -p,/Z' . Combining these
4 4 r 1 e ^1 e ^
equations with Equations 10 and 11 to eliminate PI, V , and V yields
(12)
The insertion loss (IL) is a useful parameter for evaluating
the acoustic performance of exhaust systems. One way to express
insertion loss is to compare the acoustic pressure at the exhaust
system exit (e.g. Equation 12) with the acoustic pressure P at the
exit of the exhaust manifold when no exhaust system is present.
That is:
IL=10 Log
(13)
187
-------
The analogous circuit for the en i i. nc with no exhaust system is
shown in Figure 4, where Z is the radiation impedance of the
exhaust manifold. From Fiqurc 4:
r e
The insertion loss is found by combining Equations 12 and 14 with
Equation 13 .
IL=20 Log
Z +Z
e r
'This equation shows-clearly the relationship between the exhaust
system variables and hov; each affects exhaust system performance.
MEASUREMENT OF ENGINE AND MUFFLER PARAMETERS
-Equation 15 shows that we can predict exhaust system performance
for a given combination of engine, muffler, and exhaust and tail
pipes, providing we have the appropriate information. As mentioned
previously, the four-pole parameters for the exhaust and tail pipes
are known from theory, as is the radiation impedance Z , but the
engine source impedance and the muffler impedances must usually be
measured. This section will describe a novel method of impedance
measurement. This method, referred to as the "two-microphone,
random-excitation" technique was developed about two years ago by
D. F. Ross and the author at the Ray W. Ilerrick Laboratories, Purdue
University- The theoretical basis for the technique, as well as a
188
-------
literature survey of other techniques used to measure acoustical
properties, is the subject of a recent paper [9]; only the practical
aspects related to the measurement of exhaust system properties will
be presented here.
The experimental setup used for the measurement of muffler
properties is shown in Figure 5. With this arrangement, one can
determine the muffler impedances from which the four-pole parameters
for the muffler, b ., b „, b , and b _, can be obtained (using
equations like Equation 4). At the same time one can also determine
other muffler parameters such as the transmission loss, the reflec-
tion coefficient, and the absorption coefficient. It should be
emphasized, however, that these properties are not suitable for the
prediction of overall exhaust system performance.
Referring to Figure 5, random noise is introduced into a pipe
on one side of the muffler to be tested. Air flow may be introduced
to simulate actual operating conditions, if necessary. Two micro-
phones, located on the source side of the muffler and mounted flush
with the inside of the pipe, sample the sound pressure. The micro-
phones are separated a distance of approximately 50mm and located
as close to the muffler as is physically possible (to minimize
attenuation effects in the pipe). The microphone signals are
digitized and stored in a Fourier Analyzer or Fast Fourier Transform
(FFT) processor. A spectral processing technique [9] is used to
decompose the sound field in the pipe into incident- and reflected-
wave spectra. The muffler'impedance and other muffler parameters
can be determined from these spectra. To test the accuracy of the
technique, the input impedance of a straight tail pipe was measured
and compared with theory. Figure 6 shows the experimental and
theoretical data of the real (resistive) and imaginary (reactive)
189
-------
components of the tail pipe impedance. The excellent agreement
between theory and experiment verifies the experimental technique
and, at the same time, shows the- accuracy of the theory [10] .
This data supports earlier statements which noted that exhaust
and tail pipe properties could be accounted for by using theoretical
models.
In a second test the transmission loss of a prototype muffler
was measured and compared to data obtained using the conventional
standing wave ratio method. This data is presented in Figure 7;
again, excellent agreement is noted.
Figure 8 shows how the two-microphone, random-excitation
technique might be used to measure engine source impedance. The
measurement of engine source impedance has not yet been demonstrated,
but this and other work is underway at the University of Kentucky,
Figure 9.
The two-microphone, random-excitation technique has several
advantages over conventional methods of measuring acoustic properties.
Conventional techniques such as the standing wave method [11] use
traversing probe-tube microphones that are of complex design. In
addition, flow-generated noise may influence microphone measurements
made within exhaust pipes. The stationary, wall-mounted microphones
used in the two-microphone, random-excitation technique avoid
these problems. A second advantage is increased resolution. Because
random excitation is used, the computed acoustical properties are
essentially continuous in the frequency domain. With conventional
methods using discrete frequency (sinusoidal) testing, data is also
discrete, and important aspects of the acoustical properties (i.e.
190.
-------
occurring between test f rcquenci< ••;) can be overlooked. A third
advantage is increased speed. Because random excitation is used,
and because the data is acquired and processed automatically,
impedance measurements are conducted rapidly. Only about 7 seconds
of actual measurement time was needed to obtain the' data in Figures
6 and 7.
The two-microphone, random-excitation technique is simple in
design, and because the test is essentially a "hands off" test, the
technique should yield highly consistent results. This is an impor-
tant aspect of any testing technique that is to be used by a largo
number of individuals or groups in different regions of the country.
SUMMARY - A TEST METHODOLOGY FOR MUFFLER LABELING
The above discussion indicates that the insertion loss is a
suitable parameter for predicting exhaust system performance. It
is not practical to measure insertion loss for every engine and
exhaust system configuration, but insertion loss can be predicted
(e.g. Equation 15) using proven theory in conjunction with empirical
data for engine and muffler impedances.
Much research remains before a test methodology suitable for
muffler labeling can be implemented. For examp]e, our knowledge
of engine source impedance is quite incomplete. In predicting the
insertion loss using Equation 15, how accurate must we know engine
source impedance? Does engine source impedance depend on engine
type? Load? Speed? The derivation of Equation 15 neglected the
effects of flow and temperature gradients. How important are these
effects in predicting insertion loss? Tan these effects be included
191
-------
using some type of "correction factors" or is a rigorous analysis
called for here?
In conclusion, it appears that additional research is needed
to answer some of these questions and to test the feasibility of
using a semi-empirical test methodology, such as described in this
paper, as a basis for muffler labeling.
192
-------
REFERENCES
1. Federal Register, June 22, 1977, pp. 31722-31728.
2. Stewart, G. I-,7., "Acoustic Wave Filters," Phys. Rev lev;, 20,
192-2, p. 528.
3. Crocker, M. J., "Internal Combustion Engine Exhaust Muffling,"
Proceedings, Noise-Con '77, Washington, D.C., Oct. 17-19, 1977.
4. Davis, D. D. Jr., Stokes, G. M., Moore, D., Stevens, G. L.,
"Theoretical and Experimental Investigation of Mufflers with
Comments on Engine Exhaust Muffler Design," NACA 1192, 1954.
5. Igarashi, J. , Toyama, M., Fundamentals of Acoustical Silencers
(I), Report No. 339, Aeronautical Research Inst., University of
Tokyo, Dec. 1958, pp. 223-241.
6. Sullivan, J'. W. , "Modeling of Engine Exhaust System Noise,"
Noise and Fluids Engineering, R. Hickling, Ed., pp. 161-169.
7. Rschevkin, S. N., A Course of Lectures on the Theory of -Sound,
MacMillan Co., New York, l~9~6T.
8. Levine-, H. , Schwinger, J. , "On the Radiation of Sound from an
Unflanged Circular Pipe," Phys. Review, 73 (4), 1948, pp. 383-
406.
9. Seybert, A. F., Ross, D. F., "Experimental Determination of
Acoustic Properties Using a Two-Microphone, Random-Excitation
Technique," J. Acoust. Soc. Am., 61 (5), May 1977, pp. 1362-
1370.
10.. Morse, P. M. , Ingard., K. U., Theoretical Acoustics, McGraw-
Hill, 1968, Section 9.1..
11. ASTM C38.4-58, 1972, "Standard Method of Test for Impedance and
Absorption of Acoustical Materials by the Tube Method."
193
-------
VO
ENGINE
CHARACTERISTICS
EXHAUST PIPE
CHARACTERISTICS
MUFFLER
CHARACTERISTICS
OTHER FACTORS:
TAIL PIPE
CHARACTERISTICS
SOURCE IMPEDANCE
SOURCE STRENGTH
LENGTH
DIAMETER
TRANSMISSION
LOSS
or
TRANSMISSION
IMPEDANCES
LENGTH
DIAMETER
RADIATION
OUTLET
1. Gas Flow
2. Temperature Gradients
3. Bends
4. Shell Radiation
Figure 1. Some factors influencing exhaust system performance.
-------
V.
L
Figure 2. Straight pipe of length L with acoustical
variables P1 and V., at the pipe entrance, and
P and V2
at the pipe exit.
195
-------
V
ENGINE EXHAUST PIPE
) |
f \ t
Pl
}
Z
e
•r—
Sll a!2
321 a22
\
P2
V2
MUFFLER
- 1
bll b!2
b21 b22
P3
V3
TAIL PIPE
TAIL PIPE OUTLET
i i
- L — • I
C21 °22
/ \
P4
4
Z
r
1
Figure 3. Representation of engine and exhaust system using four-pole theory.
-------
V V
Figure 4. Model for engine and exhaust manifold without
exhaust system.
197
-------
FOURIER
ANALYZER
10'-12'
HHITE NOISE
SOUND SOURCE
MIFFLER
MICROPHONE
AMPLIFIERS
u u
AIR FLOW
MICROPHONES
Figure 5. Measurement of muffler characteristics.
-------
4.000-1
3.000-
UJ
u
cr
a
UJ
8.000-
UJ
CO
t-~(
co
UJ
ct:
•1.000 -
0.000'
$
•F
600.
- 1 i —
1200. 1800. 2400.
FREQUENCY (HZ)
i 1
3000. 3COO.
lO.OO-i
6.00-
UJ
LJ
2
cr
a
iij
0_ 0.00-
51
UJ
-5.00-
UJ
ct:
-10.00-
i r
600. 1200. 1800. 2400.
FREQUENCY (HZ)
3000.
3UOO.
Figure 6. Resistive and reactive tail pipe impedances;
solid line=theory; symbols=experiment.
199
-------
3D .UU -
a eo.oo-
,__,
in
CO
i
g 10.00-
co
(.O
:c
CO
£n o.uG-
t—
-10.00-
+
4-
rf ^ ^+ ? +«
++ H^ + * "*" a
! V s t •J»^-®*/
"*" ^ 4 "*" P + +
jW -Ah ^J
* ^* V * \if^
^ 0*
e
i i i i
o.
CQQ.
A
-
1200. 1000. esoo.
FREQUENCY (HZ)
—r__
:IOOQ.
:'160Q.
Figure 7. Transmission loss of prototype muffler; +=two
microphone, random-excitation method; ^
wave method.
200
-------
FOURIER
ANALYZER
K'HITE NOISE
SOUND SOURCE
MICROPHONE
AMPLIFIERS
1ICRQPHONES
Figure 8. Measurement of engine source impedance.
-------
Figure 9. Internal combustion engine noise research at the
University of Kentucky.
202
-------
Analytical and Experimental Testing
Procedures for Quieting Two-Stroke Engines
by
Donald L. Margolis
Assistant Professor
Dean C. Karnopp
Professor
and
Harry A. Dwyer
Professor
Department of Mechanical Engineering
University of California
Davis, California 95616
203
-------
Abstract
The results of a research effort sponsored by Yamaha Motor Co. of
Japan are presented. The main objective of the project was to quiet
the exhaust from 2-stroke engines without sacrificing (too much)
performance.
Analytical and experimental programs were undertaken to acquire
a fundamental understanding of 2-stroke engine dynamics, to measure and
predict noise levels associated with various exhaust systems, and to
design innovative muffling systems. The results show that predicting
absolute noise levels is difficult; however, comparative studies are
well suited to analytical techniques.
Primary emphasis is placed on experimental procedures which allow
testing of mufflers in an anechoic chamber and in the absence of an
operating engine. One of these is a positive displacement acoustic
level source to which mufflers can be attached and sound power levels
determined. This procedure was used to corroborate acoustic theory
and to determine the extent to which acoustic theory could be used in
the design of engine mounted mufflers.
Another procedure involves the use of a rotary valve and compressed
air to generate very realistic (motorcycle-like) large amplitude pulses
with the proper through-flow and frequency content. This very clean
experiment has proven to be a very excellent method for duplicating
actual engine tests. It is anticipated that further development will
result in a variable displacement, variable through-flow rotary valve air
motor that can be used to accurately assess real muffler performance.
204
-------
Introduction
Under sponsorship of Yamaha Motor Company of Japan, a research effort
was initiated at the University of California, Davis to study exhaust
silencing of two-stroke engines. The three authors were coinvestigators
*
on the project. The project resulted in. several publications (refs. [1]
through [7]), two patents for Yamaha, and supported several graduate
research assistants.
The principal objective of the effort was to quiet two-stroke engine
exhausts without sacrificing performance. To accomplish this goal, the
research was channeled into several-parallel' paths. One of these involved
a major analytical and experimental study of the gas dynamics and mechanical
dynamics of the two-stroke engine in order to gain a fundamental understanding
of its operation and why it produces (so much) noise in the first place. This
study is representative of refs. [1], [4], [5], [6], [7]. Another major
research channel involved analytical modeling and experimental testing of
mufflers in the University of California, Davis anechoic chamber. This aspect
is described in refs. [2] and [3].
In the following section the operation of a two-stroke engine will be
briefly described in order to gain a qualitative understanding of the noise
generation problems involved. Following this, the analytical engine and
exhaust modeling are described in some detail along with noise prediction
models. Finally, the analytical and experimental anechoic chamb'er tests are
presented and the entire project summarized with emphasis on regulatory tests
for EPA monitoring and control of motorcycle noise.
Numbers in brackets [ ] refer to references
205
-------
Two-Stroke' Engine Operation
The two-stroke engine is shown schematically in figure 1 for two
different crank positions. The associated conventional expansion chamber
is -shown in figure 2. Assuming a fresh charge of air/fuel mixture has just
been ignited, the piston is driven downward on its. power stroke. It first
uncovers the exhaust port (EP) and most of the exhaust gasses are forced
into the exhaust pipe due to the still relatively high pressure inside the
cylinder. Also, as the piston moves down, it compresses the fresh charge
of fuel already resident in the crankcase. As the transfer port (TP) is
uncovered this fresh mixture is forced through the transfer passages and into
the cylinder above the piston. As the piston moves upward from bottom dead
center (BDC) it first uncovers the inlet port (IP) and fresh mixture flows
into the crankcase as a result of the increasing crankcase volume. The
piston then covers the TP and finally the EP and compresses the remaining
fresh charge in readiness for the next spark ignition.
Some of the factors influencing the overall engine performance are the
amount of fresh charge inducted through the IP, the amount of fresh charge
pushed through the TP, and the amount of fresh charge that leaks out through
the EP prior to EP closure. These considerations are what make the two-stroke
engine a most interesting dynamic system. Qualitatively, it is the "inertia" of
the gasses in the intake passage and transfer passage that insure proper charging
of the combustion chamber, and it is the expansion chamber that controls the loss
of fresh charge into the exhaust system.
When the exhaust gasses are forced through the EP, a large amplitude pressure
wave begins propagating down the exhaust system (see fig. 2). As this wave passes
through the "diverging cone", a negative (or rarefaction) wave propagates back up-
stream and helps empty the cylinder of exhaust gasses. This process is called
scavenging. When the pressure wave reaches the "stinger", most of
206
-------
the energy is reflected and this returning pressure wave either pushes
fresh charge back into the cylinder or prevents too much from leaking away.
This "stuffing" phenomenon of course depends on engine RPM, exhaust system
length and various other system parameters. From the point of view of
performance' this type of expansion chamber can provide significant super-
charging of the combustion chamber. From the point of view of noise, the
straight through-flow expansion chamber is perhaps the worst possible
design.
In the following section the analytical modeling of two-stroke engines
and their exhaust systems is described along with noise prediction.
207
-------
Analytical Models for Performance and Noise Prediction
The model used for performance prediction is described in ref. [5].
Since performance is not the main consideration here, this model will not
be described in great detail. It consists basically of a bond graph [8]
model of the complete engine coupled with an approximate model of the
exhaust system. Dynamic considerations include the intake, exhaust, and
transfer passages as well as crankcase compression and combustion. The
model is ideal for performing extensive parametric studies of port timing,
port geometry, crankcase volume, exhaust system dimensions, etc. The
operation and capability of the model are discussed completely in ref. [5].
Of more importance with respect to noise prediction is the gas dynamic
modeling of the exhaust system. The gas flow was assumed to be one-dimensional
and time dependent. The equations of motion describing this flow are
|t(pA) = |x(puA) (Continuity)
ft(puA) = - |x(puEA + pA) + p ^ _ pAF
(Momentum)
^^ = - ~ (u{ Es + PA) ) - Work
(Energy)
Es - PA(CvT + U2/2)
p - pRT
where p, p and T are the thermodynamic properties pressure, density and
temperature; u - the fluid velocity; A - channel area; t - time; x - posi-
tion; Cv - specific .heat at constant volume; and R the gas constant. The
fractional losses have been included in the term pAF where F is given by the
208
-------
following expression
r _ 4f U
F - D 2
(f and D are the friction factor and diameter respectively). The procedure
for solving the above equations is given in ref. [1] where all unusual
circumstances such as boundary conditions and internal choking are discussed.
For an average case, 150 spatial node points, similar to figure 2 were used
throughout the engine and exhaust system and 800 time steps were needed to
complete one engine cycle. As can be surmised from the above comments and
equations the numerical simulation is very complete and general, and capable
of good spatial and time resolution. The spatial and time resolution is
extremely important for making noise predictions since high frequency waves and
large sound speeds are common in two-stroke engines.
The model is capable of predicting pressure, flows, temperature, etc.
throughout the entire exhaust system; however, for the purpose of this paper
only results associated with the "stinger" will be presented (see figure 2).
Also, all results are for a Yamaha 360 MX engine.
Figure 3 shows the predicted volume flow rate from the "stinger" into the
atmosphere for the engine operating at full throttle, under load, at 7000 RPM.
This is approximately the maximum power RPM for the 360 cc engine. The steep
fronted wave in the center of the figure is the dominant cause of the very
loud, high frequency snap associated with two-stroke engine's. This is also
apparent from figure 4 where pressure and velocity inside the stinger section
are shown. Pressure in excess of two atmospheres is predicted with velocity
surges in excess of 450 m/s. If we assume that any realistic muffling device
will not change the engine performance too much, then we see that extremely
large amplitude, high frequency waves will exist at the muffler entrance.
209
-------
This suggests that the type of nonlinear modeling presented here is essential
for accurate prediction of muffler performance for small, high performance
power plants.
To predict exhaust noise levels for this engine, the volume velocity of
figure 3 was assumed to be that of a simple source radiating into an anechoic
far field. The pressure predicted at 50 feet from the source was digitally
transformed into a frequency spectrum and is shown in figure 5. An A-weighted
sound scale was assumed. A significant characteristic of the spectrum is that
it is relatively flat and contains a broad band of frequencies. Also, there
is very substantial contribution from frequencies over 1000 cycles per second.
The total SPL, weighted for'the A scale, that is-associated with the spectrum
is 102.85 db for 50 feet from the simple source. This number is in good
agreement with SPL measurements on unmuffled expansion chambers.
The next results to be presented are concerned with the addition of
mufflers to the exhaust system. In figure 6 is shown the geometry of two
mufflers analyzed. The nonlinear muffler shown in the top of figure 6 was
analyzed with the new methods mentioned previously, while the lumped parameter
muffler was analyzed with classical acoustical type approximations. In figure
7 the volume flow rate out of the nonlinear muffler is shown. It can easily be
seen by comparing with figure 3 for the unmuffled case that considerable
smoothing has occurred due to the muffler. However, there is a very distinct
and regular high frequency variation in the flow. This regular variation is
due to the reflection and formation of waves in the muffler itself, and the
frequency is characteristic of the muffler dimensions and gas sound speed. This
frequency and its harmonics are very evident in the sound square spectrum shown
in figure 8. It is also apparent from the spectrum that frequencies below
1000 cycles/sec and very high frequencies have been substantially attenuated.
210
-------
The overall SPL for the nonlinear muffler is 95.8, which is less than the
unmuffled case, but still not very liveable.
One of the primary reasons for solving the lumped parameter muffler
was to compare with the nonlinear case and to make an asessment of the
quantitative value of standard acoustical approximations. In the modeling
of the lumped parameter muffler the system is represented by two volumes, two
.nonlinear resistances and two inertias and this system is solved simultaneously
with the flow in the engine and expansion chamber. The volume flow rate from
'the lumped parameter muffler is shown in figure 9 and it is seen to be extremely
smooth. The spectrum shown in figure 10 illustrates that all frequencies have
been suppressed by the lumped parameter muffler and the SPL was 58.9db. Since
the dimensions of the nonlinear and lumped parameter muffler are very similar
it must be concluded that the use of the lumped parameter analysis for the
large amplitudes waves in two stroke engines is questionable. The one region
of the spectrum where there is qualitative agreement between the two mufflers
is in the low frequency part of the spectrum.
Another important interaction between the muffler and exhaust system that
should be mentioned is the influence of back pressure caused by frictional
losses on the transfer ofgasseslnto and out of the engine cylinder. For both
the mufflers analyzed there was enough back pressure to cause a significant
amount of exhaust gasses to be left behind in the engine cylinder.
The mufflers analyzed here are quite primitive; fowever, the new technique
employed reveals some interesting physical processes which are not included in
classical approaches to the subject. The simple source assumption used to convert
exit volume flow rate into a SPL prediction proved to be quite accurate when
compared to actual drive-by tests (see ref. [4]). Further development of the
nonlinear analysis discussed here seems to offer the hope of gaining considerably
greater insight into the nonlinear physical processes in mufflers and two-stroke
engine expansion chambers.
211
-------
The Experimental Program
Coupled closely to the analytical effort, the experimental program was
designed to first corroborate, in so far as possible, the computer models
developed for performance and noise prediction. This aspect of the program
is discussed thoroughly in refs. [3], [4], [5], [6], and [7]. At this time
this corroborative experimentation is not directly applicable to muffler
evaluation and will not be discussed further.
Another aspect of the experimental program was the design of procedures
and devices for evaluating mufflers in the University of California, Davis
anechoic facility. The main purpose of these experiments was to test muffler
models designed from acoustic considerations and to compare muffler devices
subject to realistic large amplitude inputs. Two experimental apparatus
were developed. These are described next.
212
-------
Acoustic Filter Apparatus
The acoustic filter apparatus was designed to test mufflers subject to
small amplitude volume flow inputs. This device is shown schematically in
figure 11 and pictorially in figure 12. Basically it consists of a high
impedance electromagnetic shaker driving a piston and this producing a known
frequency dependent flow source. As shown in figures 11 and 12, the shaker
and piston are enclosed in a thick wall pipe to prevent acoustic leakage.
The device could be modified to include mean flow but at this time no mean
flow is available. This device is perfect for measuring insertion loss of
muffling schemes; however, it is restricted to small amplitude input and
correlation with actual muffler performance is questionable.
213
-------
Large Amplitude Simulator Apparatus
In order to use the anechoic facility to test mufflers subject to
realistic input, the apparatus of figures 13 and 14 was developed. It
consists of a high pressure supply to a olenum chamber which feeds one
side of a rotating cylinder driven by a 1/15 horsepower electric motor
The inside cylinder has a port which allows charging with high pressure
air as the port rotates past the plenum opening and then subsequent dis-
charging as the port uncovers the exhaust opening. This simple device,
when connected to a stock Yamaha 360 MX expansion chamber produces pres-
sure spectra which are virtually identical to that shown in figure 5.
Thus far, the rotary valve has been used for qualitative comparison
studies of various muffling schemes and has proven 100% effective with
respect to comparison noise studies of actual motorcycle tests. It was
not attempted to duplicate quantitative results as this was not essential
for the Yamaha project. However, there is no fundamental reason why the
rotary valve could not be used to produce quantitative comparisons of
anechoic chamber versus actual motorcycle tests. It appears that attention
need only be given to exhaust pulse amplitude, volume through-put, and
gas temperature in order to obtain quantitative comparisons.
The question of performance degradation associated with various muffling
devices is not as easy to infer from the bench tests as was the noise com-
parisons. Again, however, it appears that if some attention is given to
this specific problem, there is no fundamental reason why correlation cannot
be obtained.
214
-------
Does a bench test procedure exist for certifying motorcycle exhaust system
performance with respect to noise and performance constraints?
At the present time, such a procedure does not exist. However, it is
felt that rotary valve is a candidate for development into a dependable,
inexpensive, and fast procedure for evaluating, at the very least, two-
stroke engines for motorcycles and snowmobiles. It is also anticipated
that small, four-stroke power plants can be tested in a similar fashion.
What is required is a research effort directed specifically at the cer-
tification issue and relying heavily on the research results already
developed.
215
-------
List of Figures
Fig. 1 Schematic of a two-stroke engine.
Fig. 2 Conventional expansion chamber.
Fig. 3 Predicted volume flow rate from "stinger"
of unmuffled 360 MX at 7000 RPM.
Fig. 4. Predicted pressure and velocity inside
the "stinger" for unmuffled 360 MS at
7000 RPM.
Fig. 5. Pressure squared frequency spectrum
resulting from Fig. 3.
Fig. 6. Muffler geometries.
Fig. 7. Exit volume flow rate for the nonlinear
muffler.
Fig. 8. Pressure squared spectrum resulting
from the flow of Fig. 7.
Fig. 9. Exit volume flow rate for the lumped
parameter muffler.
Fig. 10. Pressure squared spectrum for volume
flow of Fig. 9.
Fig. 11. Schematic of the acoustic filter apparatus
Fig. 12. The acoustic filter apparatus set-up in
the anechoic chamber.
Fig. 13. Schematic of rotary valve.
Fig. 14. Rotary valve set-up in the anechoic
facility.
216-
-------
References
[1] Dwyer, H., Allen, R., Ward, M. , Karnopp, D., and
Margolis, D., "Shock Capturing Finite Difference
Methods for Unsteady Gas Transfer", AIAA Paper No.
74-521, AIAA 7th Fluid and Plasma Dynamics Conference,
Palo Alto, CA., June, 1974.
[2] Karnopp,-D., Reed, J., and Margolis, D., "Design
and Testing of Reactive Mufflers", Proc. of the
1974 International Conference on Noise Control
Engineering, Institute of Noise Control Eng.,
pp. 325-330.
[3] Karnopp, D., Reed, J., Margolis, D., and Dwyer, H.,
"Computer-Aided Design of Acoustic Filters sing
Bond Graphs", Noise Control Engineering, Vol. 4,
No. 3, May, 1975.
[4] Karnopp, D.C., Dwyer, H., and Margolis, D.L.,
"Computer Prediction of Power and Noise for Two-
Stroke Engines with Power-Tuned, Silence Exhausts",
SAE Paper 750708, August, 1975.
[5] Margolis, D.L., "Modeling of Two-Stroke Internal
Combustion Engine Dynamics Using the Bond Graph
Technique", SAE Transaction, September, 1975,
pp. 2263-2275.
[6] Kelsay, R.E.-, and Margolis, D.L., "An Experimental
Investigation of Two-Stroke Internal Combustion Engine
Performance", SAE Transaction, 1975, pp. 2251-2262.
[7] Ospring, M., Karnopp, D., and Margolis, D.L., "Comparison
of Computer Predictions and Experimental Tests for Two-
Stroke Engine Exhaust Systems", SAE Paper 760172',
February, .1976.
[8] Karnopp, D.C., and Rosenberg, R.C., System Dynamics:
A Unified Approach_, John Wiley and Sons, New York, 1975.
217
-------
TRANSFER
PORT
j KXHAUST
PORT
TRANSFER
SECTION
CRANKCASE
INLET
COMBUSTION
CHAMBER
•SCHEMATIC DIAGRAM OF TWO STROKE ENGINE OPERATION
Figure 1
218
-------
CRANKCASE
COMBUSTION
CHAMBER
EXHAUST
PORT
EXPANSION
CHAMBER
FIGURE 2
-------
N3
N>
O
CO
o
o
LU u
CO _
O
UJ
CE
CE
o
" d
cc
o
i
o
i
! h
-H 1 1 h
40
80
120 160 200 240
CRRNK RNGLE (DEGREES)
280
320
353
360 MX- 7,000 RPM
FIGURE 3
-------
CD -r
40
80 120 160 200 240
CRRNK RNGLE (DEGREES)
280
320
360
- 7,000 RPM
FIGURE 4
350
-------
1.0
- "'QQO
5CRLE
-V
.01
o
UJ
or
a:
13
O
CO
UJ.Q91-
cc
07
UJ
.000011-
10
ioo
IQOO
IOCJOO
FIGURE 5
-------
,121 —t
JLQJj LINEAR .MUFFL.FR
1
.1521
ULn_£UI_PAJR.A M E T E R M II F F i F R
FIGURE 6
-------
10
M
-P-
CO
o
O
UJ CD
ID
O
LU
I—
or
o
__]
u_
o
oz
Lu
o
i
o
I
H 1 1 1 ! 1 ! 1 1 1 i 1 1 1 1 1 1 H 1 1 1 1 ( 1 1 1 =) 1 1 1 1 1 1 H
40
80 120 160 200 240
CRRNK flNGLE (DEGREES)
360 MX- 7,000 RPM
280
320 350
FIGURE 7
-------
.'M
Ul
10
-------
00
0
O
UJ co
o ^
o
LU
I—
cr
cc ™
o
° d |
1—• I
GZ
LJ
>~
-------
KJ
l.-T
1 +
(X)
en
Q_
.oi4-
CZ5
LU
CT.
X-t—
ID
CH
ID
CTl
00
> . i
r ;~
in
RPM- 7COD.
fl SCPLE
FREQUENCT
FIGURE 10
-------
mu
ff tar
I
I
///S/
FIGURE 11
-------
y*rr.
*L*
FIGURE 12
229.
-------
FIGURE 13
230
-------
231
-------
POWER OR PRESSURE - A DISCUSSION OF CURRENT ALTERNATIVES IN
EXHAUST SYSTEM ACOUSTIC EVALUATION
by
Larry J. Eriksson
Nelson Industries, Inc.
Stoughton, WI 53589
Presented at the
United States Environmental Protection Agency
Surface Transportation Exhaust Noise Symposium
Chicago, 111ino is
October 11-13, 1977
233
-------
ABSTRACT
Various procedures for the evaluation of exhaust system performance are
presented and discussed. Analytical as well as experimental techniques are
considered. Comparisons are made with measurements on actual engine exhaust
noise. The major approaches are ranked with respect to accuracy and cost.
INTRODUCTION
In order to select an appropriate technique for the evaluation of exhaust
system performance, the specific goals of the evaluation must be determined.
The needs of the development engineer are quite different than those of the non-
technical consumer. This paper will attempt to present the various considerations
present in making such a selection and to illustrate a wide variety of available
techniques.
There are essentially no "good" or "bad" mufflers. A given muffler may
produce good noise control results on a given system or application while producing
poor results for another. In'addition, many secondary parameters must be in-
cluded in order to fully characterize the performance of a given muffler. A
summary of some basic design considerations is given in Fig. 1. Thus, to obtain
an accurate statement of the muffler's performance, it is necessary to specify
the precise exhaust system configuration and engine application including
operating conditions such as speed and load.
TWD of the primary acoustic considerations are whether to measure sound
pressure or sound power and whether to use the actual level produced or the
difference between the silenced and unsilenced levels. A "difference approach"
has the advantage of relating more directly to the muffler performance independent
of the noise source involved, while a "level approach" has the advantage of re-
lating more directly to the sound perceived by the listener and associated
loudness.
234
-------
The choice between sound pressure and sound power is essentially a choice
between a "point measurement" versus an "area measurement". Each approach has
certain advantages. Sound pressure level must be given for a specified location
and is most appropriate when such a location may be clearly determined. Sound
power level is determined from a measurement of the average sound pressure level
over area and, thus, may be more appropriate when the location of persons near
the exhaust system is not clearly determined. Some of the practical considerations
in making these measurements will be presented later.
I. EVALUATION TECHNIQUES
A flow chart of some of the major evaluation techniques that are available
is shown in Fig. 2. Analytical and experimental approaches are listed and will
be discussed in more detail in the following sections. The complexity of an actual
engine exhaust svstem makes the selection of a single technique difficult. Severe
temperature gradients, rapidly varying turbulent flow, high amplitude pressure
variations and non-linear effects are among the primary factors contributing to
this complexity. For this reason, most actual exhaust system engineering uses a
combination of techniques to assist the exhaust system designer in obtaining
optimum performance.
A wide variety of parameters are available for use by the designer in
specifying the exhaust system performance (1-3). Some of these are listed in
Fig. 3. In general, transmission loss is preferred for theoretical calculations
because it does not depend on the engine source impedance. The determination
of engine source impedance is a difficult problem that has received only limited
study. For experimehtal work, insertion loss and noise reduction have come to be
preferred because of their relative ease of determination.
235
-------
The method of excitation used varies from the actual engine to a white
noise source. While white noise has been recormiended in the past as a solution
to the problem of measuring the performance of highly tuned mufflers (4), in fact,
this can be inadequate. A white noise source can produce conservative
or optimistic predictions of a muffler's performance depending on the specific
source, exhaust" system, and measurement procedure used. Shock wave excitation
has received considerable past study and has specific advantages in evaluating
exhaustsystems used on high-performance engines.(5,6)
II. ANALYTICAL TECHNIQUES
Analytical techniques offer the advantage of not requiring the time or
ost of experimental procedures. They can range from simple parametric analysis
chniques such as the use of muffler volume, as shown in Figs. 4 and 5, to
tplex acoustic models. (7-9) In general, the parameter technique is quite
ude in comparison to acoustic modelling although very simple to apply.
The acoustic model developed and used at Nelson includes the effects of
devated temperatures, temperature gradients, mean flow, termination impedance,
source impedance, higher order modes, and a wide variety of silencing configurations
or elements. Derived from work by Alfredson and Davies (10-13), this model has
been considerably improved and extended at Nelson to be applicable in a wider
variety of cases. Although useful from a design standpoint, quantitative
agreement with actual engine measurement is undergoing continued study in order
to obtain improved correlation. "Typical results are shown in Fig. 6. The
predicted transmission loss plot shows major minima at about 425 Hz, 850 Hz and
so on corresponding to the length of the expansion chamber equalling a multiple
of a half wavelength. Additional secondary minima are present at about 150 Hz,
300 Hz and so on corresponding to the length of the tailpipe equalling a multiple
of a half wavelength. The predicted insertion loss plot illustrates somewhat
increased complexity, partially due to the effect of the exhaust pipe. Neither
r is in good quantitative agreement with the engine measured insertion loss,
236
-------
although the frequency characteristics show some qualitative agreement and the
amplitudes reflect some general trends. Even with these limitations, this
computer model has been successfully utilized in a number of corrmercial design
activities.
III. EXPERIMENTAL TECHNIQUES
Experimental techniques fall into the two general categories of closed and
open system techniques. A variety of these techniques are illustrated in Fig. 7.
Closed system measurements do not include the radiation from the walls of the
muffler shell or exhaust system piping. The most conrron example of such a system
is the impedance tube. (14-16) Typically used to measure transmission loss
using a pure tone source and anechoic termination, this device can also be used
with a white noise source. Very similar results are obtained in considerably
less time. Results from such measurements are illustrated in Fig. 8 along with
results from the Nelson analytical model. The agreement between the top two
curves is very good and typical of the results obtained using this technique
with the pure tone or white noise source. In this example, the solid extended
inlet and outlet of the pass muffler are approximately equal to half the length
of the muffler resulting in the peaks at about 300 Hz, 900 Hz and so on.
Measurements may also be made using taped engine noise and other terminations as
will be shown later.
The closed impedance tube may also be used in the time domain as a "pulse
tube". This technique,'which has received considerable development at Nelson,
offers the advantage of presenting the pressure waveform as perceived by the
listener and as associated with the engine in the time domain. Results will be
shown below.
Open system measurements include the noise radiated from muffler and tailpipe
walls by terminating the impedance tube in an open space such as a semi-anechoic
or reverberant chamber. The excitation may be typically an electronic noise
237
-------
source, blower, standardized engine, or actual engine. At Nelson, two semi-anechoic
chambers and a reverberant chamber are available for use in such measurements
as shown in Figs. 9 and 10. (17) The semi-anechoic chamber is the most widely
utilized sound chamber for muffler evaluation. Its primary advantage is its
correlation without the associated weather problems with measurements
made outdoors on actual equipment. The reverberant chamber allows measurement
of the spatially averaged sound pressure level from which the sound power level
may be readily calculated. For applications in which the desired point of
measurement is not readily apparent, the reverberant room measurement provides a
potential advantage in that the average value is obtained. However, if the measure-
ment in the semi-anechoic chamber is simply made at the angle of maximum sound
pressure level, this advantage is minimized since the spatial average will be
strongly dominated by this maximum value. Thus, for muffler work, the main
advantages of the reverberant chamber become its lack of anechoic wedges allowing
greater flexibility in exhaust system piping and a decrease in installation and
maintenance expense.
IV. COMPARISON OF TECHNIQUES
A. BASIC SILENCING ELEMENT
The performance of a basic expansion chamber silencing element was evaluated
using a variety of the above techniques. In Fig. 11, results using the analytical
model with an anechoic termination and free-field termination are compared to
results measured on the impedance tube developed at Nelson. The expansion chamber
and tailpipe effects as well as the higher order mode effects (at about 2800 Hz) are
predicted with fair accuracy, especially for the anechoic termination case, by
the analytical model.
In Fig. 12, results for the same unit using the analytical model with an
anechoic termination, tailpipe, and tailpipe/exhaust pipe combination including
source impedance effects to obtain insertion loss are compared to results measured-
238
-------
on an actual engine. The qualitative agreement is fair, but the amplitude and
details of the frequency dependance again show considerable lack of quantitative
correlation. Many of the same features mentioned in Fig. 11 are again evident.
In Fig. 13, results for the same unit using various arrangements of the
impedance tube are compared to results measured on an actual engine. Agreement
of the simulated tests with the analytical results in Fig. 12 is fairly good,
but agreement with the engine results is again less than desired even with proper
correction for the higher exhaust gas temperatures.
In addition to the transmission loss and insertion loss measurements il-
lustrated above, transfer function measurements may also be made as shown in Fig.
14 along with the associated coherence. (18) The inversion of the transfer
function plot produces a curve proportional to the transmission loss plots
presented earlier. The minima -and maxima agree quite well with the values expected
from analytical considerations for this pass muffler.
While frequency domain analysis is most commonly used in muffler analysis,
time domain analysis using the pulse tube approach described above can provide
a useful alternative. At Nelson a pulse tube has been developed for this purpose.
Results of such a measurement are shown in Fig. 15 for a variety of expansion
chambers. The transmitted pressure pulses show good agreement with the analytically
expected values of amplitude and timing. Specifically, the time between output
pulses may be calculated to be about 2 msec corresponding to a rourid trip
distance of about 2 feet or twice the chamber length,
B. INDUSTRIAL MUFFLER
The performance of a typical industrial muffler was evaluated using white
noise excitation with the impedance tube and the intake and exhaust noise from
an actual engine as shown in Fig. 16. (19) The lack of agreement of the insertion
loss measured on the intake to the impedance tube results is increased by flow
generated noise in the intake system. The lack of agreement of the insertion
loss measured on the exhaust to the impedance tube results is increased by
239
-------
interference effects due to floor reflections. The overall A-weighted sound
levels were reduced from 117 dBA to 99 dBA for the white noise source, from 100 dBA
to 88 dBA for the intake noise and from 119 dBA to 94 dBA for the exhaust noise.
C. TRUCK MUFFLER
The performance of a typical truck muffler was evaluated using white
noise excitation with the inpedance tube and the exhaust noise from an actual
engine as shown in Fig. 17. The lack of detailed correlation is again readily
noted. The overall A-weighted sound levels were reduced from 115 dBA to 78 dBA
for the white noise source and from 111 dBA to 72 dBA for the engine noise.
Other detailed studies at Nelson have demonstrated the dependence of
exhaust noise on exhaust system configuration as shown in Fig. 18. (20) The
overall A-weighted sound level can be seen to vary as much as 7 dB for the same
muffler. This again emphasizes the importance of specifying the application for
a given muffler. In addition, the directivity pattern from an exhaust outlet
can be an important variable as shown in Fig. 19. The shape of the spectra
varies considerably as a function of angle from the outlet. As discussed
previously, in a semi-anechoic chamber, the measurement location must be carefully
selected, usually on the basis of maximum sound pressure level. In a reverberant
chamber, this problem is avoided by obtaining a spatial average of the sound
pressure level. Of course, directivity information is lost in such a sound power
measurement.
V. SUMMARY
The selection of an evaluation technique must be based on the specific
goals of the evaluation procedure. In Fig. 20, the major techniques described
above have been ranked according to the primary characteristics of accuracy and
cost. It is clear that many tradeoffs must be considered before a given technique
can te selected. Although various approaches can be useful mainly for design
240
-------
purposes, final muffler evaluation usually demands an actual engine test.
Only in this way can the required accuracy be achieved (21). Errors of 5-10 dB
in muffler performance prediction, often encountered in other techniques, are
not acceptable for today's application problems.
The assistance of Dr. Ivan Morse of the University of Cincinnati in providing
the transfer function measurements and D. Olson, D. Flanders, R. Hoops, and
G. Goplen of Nelson in providing supplementary data is gratefully acknowledged.
241
-------
REFERENCES
1) Don D. Davis, Jr., "Handbook of Noise Control", C. M. Harris,
Jr. (eel). New York: McGraw-Hill, 1957, Chapter 21, p. 21-5.
2) T. F. W. Embleton, ''Noise and Vibration Control", Leo L. Beranek, (ed.).
New York:' McGraw-Hill, 1971, Chapter 12, pp. 363-364.
3) Norman Doelling, "Noise Reduction", Leo L. .Beranek, (ed.). New York:
McGraw-Hill, 1960, Chapter 17, pp. 435-437.
4) Irwin J. Schumacher, Cecil R-. Sparks, and Douglas .J. Skinner, "A Bench
Test Facility for Engine Muffler Evaluation.". Paper 771A Presented
at SAE National Powerplant Meeting, Chicago, October, 1963.
5) B. Sturtevant, "Investigation of Finite Amplitude Sound Waves", Presented
at Interagency Symposium on University Research in Transportation Noise,
March 28-30, 1973, Stanford, California.
6) B. Sturtevant, "Relationship Between Exhaust Noise and Power Output of
Small High-Performance Internal Combustion Engines", Final Report to Nelson
Industries, January, 1975.
7) L. J. Eriksson, "Exhaust Systems for High-Performance, Four-Stroke Engines."
Presented at Noise-Con 73, Washington, October, 1973.
8) "Muffler Design Guide." Bulletin No. 74500, Nelson Muffler, Stoughton,
Wisconsin.
9) Erich K. Bender and Anthony J- Branmer, "Internal Combustion Engine Intake
and Exhaust System Noise", J. Acoust. Soc. Am. 58^ (1) ' 22-30 (1975).
10) Robin J. Alfredson, "The Design and Optimization of Exhaust Silencers."
Ph. D. Thesis, Inst. Sound and Vib. Res., Univ. of Southampton, July, 1970.
11) R. J. Alfredson and P. 0. A. L. Davies, "Performance of Exhaust Silencer
Components", J. Sound Vib. 15_ (2), 175-196 (1971).
12) Tony L. Parrott, "An Improved Method for Design of Expansion-Chamber
Mufflers with Application to an Operational Helicopter", NASA Technical
Note TN D-7309, Washington, October, 1973..
13) John E. Sneckenberger, "Recent Results Toward Experimental and Analytical
Predictions of Basic Engine Exhasut System Performance - Part II - Some
Progress in Computer-Aided Design for Analysis and Optimization of Basic
Exhaust Systems." Presented at the Eighth Annual Noise Control in Internal
Combustion Engines Seminar, University of Wisconsin, Madison, January, 1976.
242
-------
(Cont.)
14) Don D. Davis, Jr., et. al., "Theoretical and Experimental Investigation of
Mufflers with Garments on Engine Exhaust Muffler Design." NACA Report 1192,
(1954).
15) Don D. Davis, Jr., "Handbook of Noise Control", C. M. Harris, Jr. (ed.).
New York: McGraw-Hill, 1957, Chapter 21, p. 21-43.
16) D. C. Flanders, "Recent Results Toward Experimental and Analytical Predictions
of Basic Engine Exhaust System Performance - Part I - Impedance Tube: A
Tool for the Research and Development of Basic Muffler Elements." Presented
at the Eighth Annual Noise Control in Internal Combustion Engines Seminar,
University of Wisconsin, Madison, January,,1976.
17) L. J. Eriksson, "Innovative Approaches to Engine Noise Measurement." Presented
at the Eighth Annual Noise Control in Internal Combustion Engines Seminar,
University of Wisconsin, Madison, January, 1976.
18) Data provided by Dr. Ivan Morse, University of Cincinnati, Cincinnati, Ohio.
19) D. A. Olson, D. C. Flanders, and L. J. Eriksson, "An Integrated Approach to
Exhaust and Intake Noise." Paper 760602 presented at SAE V/est Coast Meeting,
San Francisco, August, 1976 and SAE Off-Highway Meeting, Milwaukee, September,
1976.
20) D. A. Olson, K. D. Nordlie, and E. J. Seils, "Techniques and Problems of
Truck Exhaust System Noise Measurement." Paper 770895 presented at SAE
Truck Meeting, Cleveland, October, 1977.
21) L. J. Eriksson, "Discussion of Proposed SAE Recommended Practice XJ1207,
Measurement Procedure for Determination of Silencer Effectiveness in Reducing
Engine Intake on Exhaust Sound Level", Presented at U.S.E.P.A. Surface
Transportation Exhaust Noise Symposium, Chicago,-October, 1977.
243
-------
NOTATION FOR FIGURES
A Incident pressure amplitude
n
B Reflected pressure amplitude
n
Z Impedance
Q Directivity factor
A Room constant
R Measurement distance
M Muffler volume
D Engine displacement
IL Insertion loss (LTT)
TL Transmission loss (!+_ )
5.6X24 5.6 inch diameter, 24 inch long muffler
65 tailpipe 65 inch long tailpipe
18 exhaust pipe 18 inch long exhaust pipe
F/S feet per second
70F 70 degree Fahrenheit average exhaust gas temperature
DB Unit for sound pressure level in decibels
DBA Unit for A-weighted sound level in decibels
3600 RPM 3600 RPM engine speed
244
-------
FIGURE CAPTIONS
Figure 1 - Surrmary of Basic Desipn Considerations
Fipure 2 - Flow Chart of Manor Evaluation Techniques
Figure 3 - Exhaust System Schematic and Evaluation Parameters
Fifrure 4 - Insertion Loss Versus Muffler Volume to Engine Disnlacement
Ratio for a Wide Variety of Applications
Fifrure 5 - Desifm Guide Derived from Data Such as That Shown in Fig. 4
Finure 6 - Transmission Loss and Insertion LOSS from Nelson .Analytical Model
Compared to Insertion Loss Measured on a Single Cylinder,
Four Stroke Engine Under Full Load at 3600 RPM
Figure 7 - Summary, of- Experimental Techniques
Figure 8 - Typical Results from Impedance Tube Insertion Loss Measurements
Using White Noise Excitation and Transmission Loss Measurements
Using Sine Wave Excitation Compared to Analytical Results
Figure 9 - Nelson Large Reverberant Chamber and Semi-Anechoic Chamber
Figure 10- Cutaway View of Nelson Large Engine Test Facilities
Figure 11- Comparison of Analytical to Experimental Results Using Impedance
Tube and Floor .Mounted Microphone
Figure 12- Comparison of Analytical to Experimental Results Using Single
Cylinder, Four Stroke Engine Under Full Load at 3600 RPM With
Floor Mounted Microphone
Figure 13- Comparison of Impedance Tube to Engine Run Results Using Single
Cylinder, Four Stroke Engine Under Full Load at 3600 PPM With
Floor Mounted Microphone
Figure 14- Transfer Function and Coherence Measurements for Simple Pass Muffler
With 4.5 Inch Solid Extended Inlet and Outlet Tubes Prior to
Perforations
Figure 15- Time Domain Evaluations of Expansion Chambers Using Single Pulse
Excitation
Figure 16- Comparison of Insertion Loss on Typical Industrial Muffler Using
Three 'Different Sources (Microphone at 30 Inch Height for Intake
Measurements and 24 Inch Height for Exhaust Measurements)
Figure 17- Comparison of Impedance Tube to Engine Results With Microphone
50 Feet From Outlet and Four Feet High
245
-------
FIGURE CAPTIONS (Cont. )
Figure 18 - Effect of Varying Tailpipe and Exhaust Pipe Length on Large Engine
Exhaust Noise
Figure 19 - Effect of Measurement Position on Exhaust Noise From Single
Cylinder, Four Stroke Engine Under Full Load at 3600 PPM
Figure 20 - Major Techniques Ranked According to Accuracy and Cost
246
-------
MINIMUM NOISE LEVEL
MAXIMUM ENGINE PERFORMANCE
MINIMUM WEIGHT
MINIMUM SIZE
MINIMUM COST
LONG LIFE
GOOD TONAL QUALITY
EASY TO MANUFACTURE
CONVENIENT SHAPE
MINIMUM TEMPERATURE
ATTRACTIVE APPEARANCE
247
Figure 1
-------
MUFFLER :
EVALUATION
ANALYTICAL
r
— — — EXPERIMENTAL
SECONDARY
PRIMARY
CLOSED
SYSTEMS
BASIC
PHYSICAL
MODEL
PHYSICAL
MODEL
W/EXP.INPUT
CRITICAL
PARAMETER
METHOD
OPEN
SYSTEMS
ANECHOIC
TERMINATION
NON-ANECHOIC
TERMINATION
FREE
FIELD
31 C
1
__ —
REVERBERANT
FIELD
I
-------
ENGINE
EXHAUST PIPE
MUFFLER
TAILPIPE OPENING
A.
Z SOURCE
*••
_« —
. f*Tf* rw^
-j^nn- -nm-j-
T T
— ^«
^
»-vvv orv-k
-jrrrv- -nnr^
T T
>-
4
|-T-V» /^»^^
of^ "^r
T T
^
<^OT-
z^" -L
RAD ^
B4 —Tp
10 LOG
_O
rr R
2 +
(R2,A IN FEET27LW RE 10"12W.)
L TL= 10 LOG
LIL " Lp (WITHOUT MUFFLER) - Lp (WITH MUFFLER)
NR = Lp (INPUT) - Lp (OUTPUT)
0)
CO
-------
35
30
25
20
IL-DBA
15.
10
x
X
M/D
16"
250
Figure 4
-------
N3
Ln
R 6
Engine Rating Numbers _ 5
ENGINE RATING CHART
Hi-Perform.
Nal. Asp.
4-Stroke
or2Cyl.
Spark Ign.
Total Possible = 5
Muffler Volume Required Equals
R xTotal Engine Displacement
Muffler Volume
Engine Displacement
20 30
INSERTION LOSS IN dB(A)
CD
Ul
-------
40 J
20 "I
IL-DB
5.6X24 EXPANSION CHAMBER
65 TAILPIPE - 1700 F/S
ANALYTICAL MODEL
5.6X24 EXPANSION CHAMBER
65 TAILPIPE-18 i EXHAUST PIPE-1700 F/S
ANALYTICAL ft MODEL
5.6X24 EXPANSION CHAMBER
65 TAILPIPE - 18 EXHAUST PIPE
ENGINE DATA - 1700 F/S
IK
FREQUENCY-HZ
Figure 6
252
-------
AMPLIFIER
Ul
OJ
SINE WAVE
OSCILLATOR
WHITE
NOISE
-GENERATOR
TAPE
RECORDER
PULSE
GENERATOR
MIC
SLIP TUBE
DRIVER
FFT ANALYZER
MIC
1 RTA ANALYZER
1
> SEMI-
ANECHOIC
—CHAMBER
AAAAAAAA/V
REVERBERANT
CHAMBER
ANECHOIC TERMINATION
REFLECTING TERMINATION
OPEN PIPE TERMINATION
-------
40
IL-DB
20
5.6X24 PASS MUFFLER ON IMPEDANCE TUBE
WHITE NOISE EXCITATION - 70 F
AN ECHOIC TERMINATION
IK
2K
40
TL-DB
20
5.6X24 PASS MUFFLER ON IMPEDANCE TUBE
SINE WAVE EXCITATION - 70 F
ANECHOIC TERMINATION
IK
2K
40
TL-DB
20
5.6X24 PASS MUFFLER
ANALYTICAL MODEL - 70 F
ANECHOIC TERMINATION
2K
FREOUENCY-HZ
254
Figure 8
-------
s
a
1'
j
LO
CN
-------
cui
of Technical CenUr
Engine Semi-Anachoic Chamber
Unapfground Large Engine Test Cells
Ervg.ne Enhausl Col'eCfor
Large Engine ConUol Room
Vehtcle Enltance
Sour,a insliumenialion Room
Small Engine S«meAnechoic Chamber
-------
'40
TL-DB
20 H
6X12 EXPANSION CHAMBER-ANALYTICAL MODEL - 70 F
ANECHOIC TERMINATION
5K
IOK
6X12 EXPANSION CHAMBER ON IMPEDANCE TUBE
WHITE NOISE EXCITATION - 70 F
ANECHOIC TERMINATION
5K
40 -
TL-DB
20 1
0
-10
6XI2 EXPANSION CHAMBER-ANALYTICAL MODEL - 70 F
66 TAILPIPE-OPEN PIPE/FREE-FIELD TERMINATION
• II
0
1 1 1 1 1
5K
IOK
30 -
20
IL-DB
10
0
6X12 EXPANSION CHAMBER ON IMPEDANCE TUBE
66 TAILPIPE INTO SEMI-ANECHOIC CHAMBER
WHITE NOISE EXCITATION - 70 F
5K
FREQUENCY-HZ
257
IOK
Figure 11
-------
40 -
6X12 EXPANSION CHAMBER-ANALYTICAL MODEL - 890 F
ANECHOIC TERMINATION
TL-DB'
0
5K
IOK
40 •
6X12 EXPANSION CHAMBER-ANALYTICAL MODEL - 890 F
66 TAILPIPE -OPEN PIPE/FREE-FIELD TERMINATION
TL-DB-
IOK
40-
6X12 EXPANSION CHAMBER-ANALYTICAL MODEL - 890 F
66 TAILPIPE - 22 EXHAUST PIPE -OPEN PIPE/FREE-FIELD
TERMINATION
5K
IOK
6X12 EXPANSION CHAMBER ON ENGINE
66 TAILPIPE - 22 EXHAUST PIPE
3600 RPM - 890 F
5K
FREQUENCY-HZ
258
Figure 12
-------
30
6X12 EXPANSION CHAMBER ON IMPEDANCE TUBE
ANECHOIC TERMINATION
WHITE NOISE EXCITATION - 70 F
5K
30 -
20 •
IL-DB
10
0
30
6X12 EXPANSION CHAMBER ON IMPEDANCE TUBE
66 TAILPIPE INTO SEMI-ANECHOIC CHAMBER
WHITE NOISE EXCITATION - 70 F
5K
IOK
6X12 EXPANSION CHAMBER ON IMPEDANCE TUBE
66 TAILPIPE INTO SEMI-ANECHOIC CHAMBER
TAPED ENGINE NOISE - 70 F
0
6X12 EXPANSION CHAMBER ON ENGINE
66 TAILPIPE - 22 EXHAUST PIPE
3600 RPM-. 890 F
5K
FREQUENCY-HZ
259
IOK
Figure 13
-------
-10 -
-20 -
DB
-30
-40 -
-50 .
500
5X15 PASS MUFFLER
TRANSFER FUNCTION
WHITE NOISE EXCITATION
70 F
1.2 -
5X15 PASS MUFFLER
COHERENCE FUNCTION
WHITE NOISE EXCITATION - 70 F
500
FREQUENCY-HZ
Figure 14
260
-------
0.41V
3.18X12 EXPANSION CHAMBER
PULSE EXCITATION - 70 F
ANECHOIC TERMINATION
4.18X12 EXPANSION CHAMBER
4.9X12 EXPANSION CHAMBER
0.16V
6.0X12 EXPANSION CHAMBER
0.5MSEC/DIV
261
Figure 15
-------
40 .
30
20 -
IL-DB
10 '
0
INDUSTRIAL MUFFLER ON IMPEDANCE TUBE
WHITE NOISE EXCITATION - 70 F - ANECHOIC TERM.
IK
2K
40 '
30
20
IL-DB
10 -
20
IL-DB
10
INDUSTRIAL MUFFLER ON ENGINE INTAKE
2900 RPM - 70 F
2K
INDUSTRIAL MUFFLER ON ENGINE EXHAUST
2900 RPM - 1250 F
IK '
FREQUENCY-HZ
262
2K
Figure 16
-------
60 '
50
40
IL-DB
30 '
20 -\
10
TRUCK MUFFLER ON IMPEDANCE TUBE
WHITE NOISE EXCITATION - 70 F - ANECHOIC TERMINATION
IK
FREQUENCY-HZ
2K
50
40 1
IL-DB
30 •
20 1
10
TRUCK MUFFLER ON 8V-7IN
2IOO RPM - 970 F
IK
FREQUENCY-HZ
263
Figure 17
-------
SHELL NOISE
INCLUDED EXCLUDED
50'
@
FLOOR
MICROPHONE
POSITION
-P-
O
75
90
105
. 120
x
X
135
150
50'
(a)
4' VERTICAL
0>
I—1
oo
15
M-VARIABLE MUFFLER POSITION
E-VARIABLE EXHAUST PIPE LENGTH
T-VARIABLE TAILPIPE LENGTH
S-VARJABLE TAILPIPE LENGTH
AND SOURCE HEIGHT
A-VARIABLE EXHAUST PIPE LENGTH
@ 100" SOURCE HEIGHT
CONSTANT: 144" STANDARD
SOURCE HEIGHT
SI
Ml
Tl
S2
M2
T2
S3 A6
A5
A4
A3
M3
A2
Al
T3
S4
M4
T4
S5
M5
T5
S6
M6
E6
E5
E4
E3
E2
T6 El
30 45 60
TAILPIPE LENGTH
(INCHES)
75
90
1
2
3
4
5
6
M
76.5
71.0
71.5
73.5
72.5
E
74.5
72.0
71.5
71.5
73.0
72.5
T
76.5
74.0
76.0
77.0
78.0
74.5
S
76.5
74.5
73.5
73.5
74.0
72.5
A
75.0
72.5
71.5
72.0
72.5
73.5
50' @4' VERTICAL
SHELL NOISE INCLUDED
1
2
3
4
5
6
M
72.5
68.5
67.5
68.5
69.5
E
70.5
68.5
67.5
67.0
69.0
69.5
T
72.5
71.5
70.5
72.5
73.5
70.5
S
75.0 '
70.0
70.5
71.0
71.5
69.5
A
70.5
68.5
67.5
68.5
69.0
70.5
50' (Si 4' VERTICAL
SHELL NOISE EXCLUDED
dBA
dBA
-------
100
80
70
9
-AAAA
AAA/V
OPEN PIPE-ENGINE
30 DEGREES
5K
OPEN PIPE-ENGINE
45 DEGREES
IK"
,-U
IOK
IOK
70
OPEN PIPE-ENGINE
60 DEGREES
5K
IOK
90 -
SPL-DBA
80 J
70
OPEN PIPE-ENGINE
75 DEGREES
5K
IOK
OPEN PIPE-ENGINE
90 DEGREES
60
5K
FREQUENCY-HZ
265
Figure 19
-------
COMPARISON OF EVALUATION METHODS
MOST ACCURATE?
I) ACTUAL ENGINE
2) STANDARD ENGINE
3) SIMULATED SOURCE
4) ANALYTICAL MODEL
5) PARAMETER EVALUATION
LOWEST COST?
I) PARAMETER EVALUATION
2) SIMULATED SOURCE
3) ANALYTICAL MODEL
4) STANDARD ENGINE
5) ACTUAL ENGINE
Figure 20
266
-------
A COMPUTER-AIDED APPROACH TOWARD PERFORMANCE PREDICTIONS
FOR ENGINE EXHAUST MUFFLER
John E. Sneckenberger
Associate Professor
West Virginia University
College of Engineering
Morgantown, WV 26506
presented at
U. S. Environmental Protection Agency
Surface Transportation' Exhaust Noise Symposium
Chicago, IL
October 1977
organized by
McDonnell-Douglas Co.
267
-------
INTRODUCTION
Engineering acoustics has been an area of study in the Mechanical Engi-
neering and Mechanics Depar :ment at West Virginia University since 1971, with
student involvement from freshman projects to graduate research. Somewhat of
interest to some might be the fact that muffler design, development and testing
is taught to freshman engineering students, and in only three weeks, during
only one day each week, and only for three hours in the afternoons of these
three days. Thus, because of student project grading requirements, I have
been evaluating and "labeling" mufflers - with a letter grade - for years.
My "regulatory policy1 for muffler labeling must be a good one and maybe, quite
humorously of course, should be considered bv the Environmental Protection
Agency because I have yet to be taken to court concerning my regulatory policy.
During the summer of 1975, I participated as one of two summer faculty
research participants at Nelson Industries, Inc. of Stoughton, WI under a
National Science Foundation grant to the Nelson Research Department. As Larry
Eriksson, vice-president of research, and I formulated a work plan for the ten-
week period that summer, it was decided to attempt to expand the existing com-
puter-aided design capabilities at Nelson Industries. At that time, improved
computer-aided design was visualized as being an important compliment to an
on-going impedance tube muffler development study. Now, today at this symposium,
after considerable success as an analytical development, design, evaluation and
(potentially) optimization tool for the manufacture of mufflers, this "Computer-
Aided Approach Toward Performance Prediction for Engine Exhaust Mufflers" is
268
-------
being presented to exhibit the increased extent, possible merit, etc of this
computer-aided methodology to predict and to communicate noise reduction
charac t'er is t ics of vehicle exhaust systems. My presentation here will be an
extension of a paper (1) presented in January 1976 at the Eighth Annual Noise
in Internal Combustion Engines Seminar base on the initial work completed at
Nelson Industries the previous summer. Presentation of information contained
in that paper entilted "Some Progress in Computer-Aided Design for Analysis
and Optimization of Basic Exhaust Systems" will be followed by some Comments
on the state of the computer program as it exists today as well as on the
judged applicability of the computer program to function as an analytical
simulation technique toward usefulness as a "bench-type1 methodology in regu-
latory muffler labeling.
This 1976 seminar paper just mentioned began with a brief description of
three of the more recent approaches which seemingly offered potential for con-
tinuing future progress toward effective computer-aided design of exhaust
systems. Secondly, the paper then discussed exten.sion features which were in-
corporated into a recent National Aeronautics and Space Administration prepared
computer-aided muffler design program to provide improved capabilities for
Nelson Industries to complement its on-going muffler development work utiliz-
ingimpedance tube experimentation. Thirdly, the paper then provided an example
of how this extended NASA computer program permitted a parametric study for an
extended inlet-extended outlet muffler to produce generalized computer-aided
muffler design curves. Finally, several potential additions co expand the design
analysis and optimization capabilities of the extended computer program were
identified. This material will be presented in the next four section of this
paper.
269
-------
RECENT COMPUTER-AIDED MUFFLER DESIGN METHODS
Munjal (2) had recently proposed a revised transfer matrix method,
utilizing a modification to a previously defined velocity ratio function, for
the computer evaluation of insertion loss for exhaust mufflers with mean flow.
Acoustic pressure and mass velocity were redefined considering the convective
coupling between acoustic phenomena and incompressible mean flow. Transfer
matrices for various basic muffler elements were derived. Unlike the case
for zero mean flow where each of the transfer matrices corresponded to one of
the three types of impedances, such a correspondence did not appear to be the
case for non-zero mean flow. See Figure 1.
Work by Karnopp, et al. , (3) on modeling engine exhaust mufflers in bond
graph terms had been recently reported in connection with the computer pre-
diction of power and noise for two-stroke engines with power tuned, silenced
exhausts. From the equivalent bond graph model of a lumped muffler (See
Figure 2), recursion formulas relating acoustic pressure and volume flow rate
in terms of the volume of fluid stored by the compliance element and the
momentum of the fluid of an inertial element were formulated. The associated
finite element computer program was developed to handle the one-dimensional
effects of nonlinear wave steeping, flow resistance and high mean flow. The
conclusion, however, seemed to be that such a one-dimensional computer program
could not accurately describe complicated muffler configurations in which
three dimensional effects are important.
In a then recent paper, Young and Crocker (4) used variational methods to
formulate a mathematical description of the acoustic field existing in a
muffler. See Figure 3. Solution of this variational method formulation for
the acoustic field was obtained by finite element methods. For this approx-
270
-------
imate solution numerical method approach, the muffler is divided into a number
of subregions of nodal elements. Nodal parameters descriptive of the varia-
tion of acoustic pressure at each node were then defined. The prediction of
the desired muffler transmission loss was then made by forming the equivalent
acoustic four-terminal transmission network for which the nodal parameters
are used to determine the four-terminal constants. Future papers were then
planned to show that when applied to mufflers with complicated shaped chambers
for which plane wave theory predictions are not available, transmission loss
predictions using this method are in good agreement with experiments.
EXTENSIONS TO NASA MUFFLER DESIGN COMPUTER PROGRAM
The above three relatively new methods of computer-aided muffler design,
as well as other possible methods which were aot mentioned, indeed projected
prospects for more progress in the analysis and optimization of exhaust mufflers
in the near future. However, for immediate short term (ten weeks) applicability
with some potential for later extension, it seemed most appropriate at that
time to develop computer-aided design capabilities using the most complete
computer program available based on muffler modeling which used essentially
linear wave equation theory. Figure 4 illustrates how the planned computer-
aided design capability would be incorporated into the overall scheme of manu-
facturing mufflers from specifications.
Such a rather well developed computer-aided muffler design program as
suggested above for reactive extended inlet-extended outlet expansion chamber
mufflers had been made available by the NASA through Technical Note TN D-7309.
271
-------
This computer program is largely based on the work of Alfredson and Davies (5).
The key features of the NASA computer program are listed in Figure 5.
In order to appreciate the complexity of a typical commerical muffler
relative to the existing capability of the NASA computer program, Figure 6 (a)
shows a drawing of a two tube-three pass muffler taken from page 31-17 of the
Handbook of Noise Control, Harris, ed.. As the projected and unfolded ver-
sion of this muffler shown in Figure 6(b) illustrates, several features, such
as multiported chambers and perforated tubes, are not readily handled by the
existing NASA computer program.
As an initial effort to extend the NASA computer program, the program
was converted from "complete chamber" analysis to 'individual section1 analysis.
Further, efforts were directed at providing sectional models for plug and two-
pass muffler sections which are quite common in Nelson mufflers. For all sections,
variable diameter pipes and chambers were now permitted. A pictorial description
of these initial extensions to the NASA computer program is shown in Figure 7.
Figure 8 shows a more detailed definition of how various example mufflers would
be sectioned for inputing to the extended NASA computer program.
Using the sectional approach to the prediction of transmission loss for a
particular muffler required internal modification to the flow logic of the NASA
computer program. A flow diagram depicting how the transmission loss is deter-
mined by stepping individually through the sectional subroutines, compiling and
storing the results until the complete muffler performance is printed out in
either tabular and/or plotted form is shown in Figure 9. Sectioning of the
exhaust system is performed by first defining the type of tailpipe radiation
environment and proceeding up to and including the type of engine source impedance.
272
-------
EXAMPLE OF COMPUTER-AIDED STUDY OF MUFFLER DESIGN
The extended computer program served a primary function of confirming,
evaluating, predicting, etc, the theoretical transmission loss for experimental
basic muffler models as they were evaluated using the impedance tube technique.
Another function of the extended NASA computer program was its capability to
perform analysis of muffler transmission loss behavior as a function of parti-
cular muffler design parameters. For example, consider the extended inlet-
extended outlet expansion chamber muffler with both extensions initially one-
fourth the length of the chamber. Keeping the distance between the internal
ends of the extended inlet and extended outlet pipes constant, this fixed distance
was then offset by the varying amount S . See Figure 10. In Figure 10 below
the sketch of the muffler being considered is a tabular example showing the
changes in value of the quarter-wave length resonances with amount of offset ^ .
Figure 11 provides an appreciation of the resultant influence on transmission
loss for several values of offset 3 . Generalized curves representing the
behavior of the resonant frequencies are shown in Figure 12. Observe that as
the .centered fixed distance representing a double resonant frequency at say
1000 hz is offset to the maximum value, the one resonant frequency for the
lengthing inlet (or lengthing outlet) approaches one half its initial value or
500 hz, while the other resonant frequency for the shorting outlet («r shorting
inlet) rapidly increases toward infinity. Of additional note is the decreasing
resonant frequency from 3000 hz to 1500 hz with offset distance o which could
contribute to certain advantageous transmission loss features in specific situ-
ations. Many such parametric studies of muffler geometry, etc can be conceived
and readily performed using the extended computer program.
273
-------
POTENTIAL ADDITIONS TO FURTHER EXPAND COMPUTER DESIGN,CAPABILITIES
With the computer program operational and functioning both in its initial
intended role as a compliment to the impedance tube study and in its inherient
capacity to perform parameter variation studies of muffler performance, pro-
jections were made at the end of the ten weeks of possible additional extensions
that could contribute to the further development of the extended NASA computer
program. These extensions included a) sectional model for a flow reversing
chamber muffler (6) and b) sectional model for a parallel duct muffler (7).
Theoretical development and experimental verification both offer attractive
encouragement to their possible inclusion in muffler systems. Descriptional
and performance features from the literature for the flow reversing chamber
muffler is shown in Figure 13. This type of chamber is quite common in commeri-
cal mufflers. A parallel duct muffler is described and experimental perform-
ance results shown in Figure 14. The experimental curve 6n the left shows
quite good wideband transmission loss.
Addition of muffler sections such as these two mentioned offered increased
improvement to the extended NASA computer program as it had been developed at
that time about two years ago.
COMMENTS ON ADDITIONALLY EXPANDED CAPABILITIES OF COMPUTER PROGRAM
Growth of the computer-aided design capabilities for exhaust muffler
analysis since the initial summer development work by the author has been quite
substantial. Efforts by Nelson research personnel have made advances toward
the addition of temperature gradient effects, reversing chambers, perforated
274
-------
tubes, and higher-order modes within the exhaust system as well as the incor-
poration of engine source impedance description for permitting insertion loss
prediction. Experimental work is currently being undertaken at West Virginia
University to better define engine source impedance for use in the computer
program.
The principal uses made of this continuously expanding computer-aided
approach for muffler design by Nelson Industries have been 1) as the theoret-
ical predictor of transmission loss for conceptual muf.flers and large industrial
silencers proposed by persons within and outside the Research Department, and
2) as the analytical compliment to assist the direction of experimental bench
and/or laboratory engine muffler development research projects, such as the
initially intended impedance tube muffler development study (8). Evidence of
the computer program's successful application as a compliment to experimental
engine-exhaust system studies in terms of providing analytical comparison pre-
diction plots is provided by Figures 6, 11, 12 and 13 of the paper by Larry J.
Eriksson entitled 'Power or Pressure - a Discussion of Current Alternatives in
Exhaust System Acoustic Evaluation1 presented at this Symposium. (Reference 9)
Additional expression of the computer program availability for incorporation
into experimental studies conducted at Nelson Industries can be found in Refer-
ence 10. Figure 15 and Figure 16 of this paper provide comparative analysis of
the ability of the computer program to predict the measured acoustic performance
of typical pass and plug exhaust mufflers respectively at engine operating
conditions.
The optimization capability of the computer program has served a limited
purpose and use to this time, mainly because its cost-effectiveness operation
has not been totally explored.
275
-------
APPLICABILITY OF COMPUTER PROGRAM IN REGULATORY MUFFLER LABELING
In regards to the possible applicability of this analytical simulation
technique toward usefulness as a 'bench test1 methodology in regulatory
muffler labeling, the following four statements seem appropriate: 1) this
'methodology' potentially can "measure" (by theoretical calculation) the noise
reduction characteristics (transmission loss, insertion loss, etc.) of engine-
exhaust systems, assuming continued successful efforts toward definition of
muffler sectional configurations, engine source impedance, etc.; 2) this
"methodology" can communicate the noise reduction characteristics by means'of
single number (overall) and frequency band (third octave, etc) evaluation and
could compare these evaluations with any applicable standards. Also, through
a design optimization procedure, suggestions for exhaust system improvement
might be made; 3) this "'methodology' cannot provide "total vehicle" evaluation
toward labeling ofsurfaee transportation vehicles with respect t.o all possible
vehicle noise sources; 4) this ' methodology' might provide information which
would be compatiable with regulatory policy once the regualtory policy itself
is eventually formulated. Currently, this 'methodology' is quite useful for
muffler design purposes which was its initial intent.
276
-------
SELECTED REFERENCES
1. Sneckenberger, J. E., "Some Progress in Computer-Aided Design for
Analysis and Optimization of Basic Exhaust Systems", Eighth Annual
Noise Control in Internal Combustion Engine Seminar, 23 pp., January
1976.
2. Munjal, M. L., "Velocity Ratio-Cum-Transfer Method for the Evaluation
of a Muffler with Mean Flow", J. Sound and Vibrations, Vol. 39, No. 1,
pp. 105-19, March 1975.
3. Karnopp, D. C. , Dwyer, H. A. and Margolis, D. L., "Computer Prediction
of Power and Noise for Two-Stroke Engines with Power Tuned, Silenced
Exhausts", SAE Paper 750708, 14 pp., August 1975.
4. Young, C. J. and Crocker, M. J., "Prediction of Transmission Loss in
Mufflers by the Finite Element Method", ^J. Acoustical Society of America,
Vol. 57, No. .1, pp. 144-8, January 1975.
5. Alfredson, R. J. and Davies, P.O.A.L., "Performance of Exhaust Silencer
Components", _J. Sound and Vibrations, Vol. 15, No. 2, pp. 175-96,
April 1971.
6. Tengtrairat, P. and Diboll, W. B., "Adjoint Inlet-Outlet Quarter-Wave
Length Muffler", Proceedings: 1973 Noise-Con, pp.212-7, October 1973.
7- Patrick, W. P., "Systematic Method to Determine the Acoustical Character-
istics of Series-Parallel Duct Configurations Using Transmission Matrices",
Proceedings: Third Interagency Symposium on University Research in
Transportation Noise", pp. 533-45, November 1975.
8. Flanders, D. C., "Impedance Tube: A Tool for the Research and Development
of Basic Muffler Elements", Eighth Annual Noise Control in Internal
Combustion Engine Seminar, 32 pp., January 1976.
9. Eriksson, L. J., "Power or Pressure - A Discussion of Current Alternatives
in Exhaust System Acoustic Evaluation", EPA Surface Transportation
Exhaust Noise Symposium, 33 pp., October 1977.
10. Olson, D. A., Nordlie, K. D. and Seils, E. J., "Techniques and Problems
of Truck Exhaust System Noise Measurement", SAE Paper 770895, 14 pp.,
October 1977.
277
-------
12* II 10 9 8 7
5 4 3
(o)
-&._..v_-..i.
*T«T I *l'j
n*
(b)
r.O
(C)
(a) A TYPICAL STRAIGHT-THROUGH EXHAUST MUFFLER
(b) ANALOGOUS CIRCUIT FOR THE EVALUATION OF VRn+1
(c) ANALOGOUS CIRCUIT FOR THE EVALUATION OF VRC
Figure 1. FORMULATION FOR VELOCITY RATIO-CUM-TRANSFER MATRIX METHOD.
278
-------
FLOW.
(a)
Jiff J-BJ' fc f f if
-»6- -6-j—o.i-A»*i..-.S-.J -3-1-.:
t 1 I
R R R;
(b)
(a) MODEL OF EXPANSION CHAMBER MUFFLER
(b) EQUIVALENT BOND GRAPH FOR MUFFLER
Figure 2. FORMULATION FOR BOND GRAPH METHOD.
279
-------
lut
i.t
(a)
P - 0 or - 0
(b)
(a) GENERALIZED ACOUSTICAL SYSTEM
(b) FORMULATION APPLIED TO EXPANSION CHAMBER MUFFLER
Figure 3, FORMULATION FOR VARIATIONAL METHOD.
280
-------
YESTERDAY
MANUFACTURING
EXPERIENCE
GUESSING
AND TESTING
k
!
v
TODAY
GUESSING
AND TESTING
MANUFACTURING
EXPERIENCE
SPECIFICATIONS
\ \ \ \
IMPEDANCE TUBE
STUDIES
COMPUTER
AIDED
DESIG
ACOUSTIC
THEORY
MUFFLER
Figure 4. COMBINED IMPEDANCE TUBE-COMPUTER PROGRAM APPROACH.
281
-------
NASA IN D-7309
AN IMPROVED METHOD FOR DESIGN OF EXPANSION-CHAMBER
MUFFLERS WITH APPLICATION TO
AN OPERATIONAL HELICOPTER
KEY FEATURES OF COMPUTER PROGRAM
{2)
(1) 4
INLET
.M,
4 , L6
L5
L4 .
TVvil PlPP
• CALCULATES TRANSMISSION LOSS
• HANDLES UP TO FIVE EXPANSION CHAMBERS
• INCLUDES MEAN FLOW EFFECTS
• VARIES COMPONENT LENGTHS WITHIN SPECIFIED
LIMITS TO OPTIMIZE PERFORMANCE
Figure 5. FEATURES OF NASA MUFFLER DESIGN COMPUTER PROGRAM.
282
-------
/*.
T
K
A
k
] L
T
(a)
(b)
(a) as constructed
(b) as unfolded
Figure 6. UNFOLDED VERSION OF A TWO TUBE-THREE PASS MUFFLER.
283
-------
PREVIOUS
is.
EXTENDED '
INLET
STRAIGHT PIPE
'EXTENDED OUTLET
FR06R6SS
(up to 5)
V
..' 54.
I
f~
M • • • • ^MWMkMMV* • • •• mm
Sj.
±3j J
i — j i n
s,
ABOVE SECTIONS
PLUS PLUG AND TWO-f&SS SECTIONS
ONE SECTION
(up to 20)
Figure 7. EXTENSIONS TO NASA MUFFLER DESIGN COMPUTER PROGRAM.
284
-------
1
'OUTLET
(typ.)
Section 3D Numbers
INLET
(typ.)
lo
ft t
^ 10 ^:
Ex: AM RLE
3
i a : i
10 j a '
*-~h-i
Figure 8. EXAMPLES OF SECTIONING OF EXHAUST MUFFLERS.
285
-------
c
IKJPUT
TYPE JN DATA
Figure 9. FLOW DIAGRAM FOR SECTIONALIZED MUFFLER DESIGN COMPUTER PROGRAM.
286
-------
EXTENDED INLET AND OUTLET MUFFLER WITH OFFSET OPENING
TABULAR EXAMPLE
Given L = 14.4 inches; for ,£ = 7.2 inches
V'"i
0
.5
1.0
1.5
2.0
2.5
3.0
3.6
fl '
1000
877
780
710
645
600
540
500
f2
1000
1163
1380
1710
1945
2921
4732
oO
f3
3000
2631
2340
2130
1935
1800
1620
1500
f4
3000
.
•
•
I
•
'
*
Figure 10. EXAMPLE OF STUDY USING MUFFLER DESIGN COMPUTER PROGRAM.
287~
-------
1-0
CD
CO
*>'
500
1000
FREQUENCY, hz
1500
2000
Figure 11. FAMILY OF TRANSMISSION LOSS CURVES FOR SEVERAL AMOUNTS OF OFFSET PIPE OPENING.
-------
3 -
I
o
S 2
4J
C
C
o
"O
(1)
C
a)
o
o
C
3
cr i
a) *•
C
n)
o
O
M
H
0 .2 .4 .6 .8 1.0
RATIO: Offset Displacement/ Maximum Offset Displacement
Figure 12. GENERALIZED CURVES FOR RESONANT FREQUENCY VS OFFSET DISPLACEMENT.
289
-------
t A «"1Ka
A A ' Al e
° * ^ B "*
V Bl 1 V
T
83 A3
-i-.
A2
B2
S
2
x«0
(a)
Inlet
.12 ft. ID
Outlet
.12 ft. ID
ft-
.36 ft.
I.D.
40 •
C
O
•i-l
it
jj
<
20 •
Tailpipe - 1 ft
(a)
(b)
0 100 200 300 400
Frequency, hz
(b)
theoretical model for Flow Reversing Chamber Muffler.
-Measured Transmission Loss for Flow Reversing Chamber Muffler.
700
Figure 13. DESCRIPTION AND PERFORMANCE OF FLOW REVERSING CHAMBER MUFFLER.
290
-------
P1,u1,s1
.75"
_L
l r
b
16
0.3O la
Duct A
S.
1B
1B
Duct B
(a)
TL - 20 log(P1/P2)
TL
2B
J2B
...•.«•••••..•.• • • ••.•.•.••.*•••••••.
..•••.••...•.• •« • . . .'.••.•••.
..,«..... . • •. •..'.•.•••••
P2,urst
joo joo u
Frequency
300 . 500 IK
Frequency
(b)
(a) Theoretical Model for Parallel Absorptive Duct Muffler.
(b) Measured Transmission Loss for Parallel Absorptive Duct Muffler.
Figure U. DESCRIPTION AND PERFORMANCE OF PARALLEL DUCT MUFFLER.
291
-------
40
IL-DB
20
5.6X24 PASS MUFFLER ON ENGINE
66TAILP1PE-I8 EXHAUST PIPE
3600 RPM - 890 F
, P-
IK
FREQUENCY, HZ
—r
2K
40 -
TL-DB
20
5.6X24 PASS MUFFLER-ANALYTICAL MODEL
66 TAILPIPE - 890 F
IK
2K
40'
TL-DB
20-
5.6X24 PASS MUFFLER-ANALYTICAL MODEL
ANECHOIC TERMINATION - 890 F
2K
Figure 15. Analytical and Experimental Results for a Pass Muffler.
292
-------
40.
IL-DB
20.
5.6X9.8 PLUG MUFFLER ON ENGINE
76TAILPIPE-20 EXHAUST PIPE
3600 RPM - 890 F
5K .
FREQUENCY-HZ
IOK
40 H
TL-DB
20 H
5.6X9.8 PLUG MUFFLER - ANALYTICAL MODEL
ANECHOIC TERMINATION - 890 F
5K
FREQUENCY-HZ
Figure 16. Analytical and Experimental Results for a Plug Muffler.
293
-------
REVIEW OF INTERNAL COMBUSTION
ENGINE EXHAUST MUFFLING
by
Malcolm J. Crocker
Ray W. Herrick Laboratories
School of Mechanical Engineering
Purdue University
Uest Lafayette, Indiana, USA
SUMMARY
This paper will describe types of mufflers in existence,
discuss definitions of muffler performance, briefly review
historically some of the theory developed to predict muffler
acoustic performance, describe some of the work done at the
Herrick Laboratories'on predicting muffler attenuation, and
lastly comment on the possibility of designing a practical
bench test for a muffler which does not involve an engine
as a source.
INTRODUCTION
Exhaust noise is the predominant noise source with
most internal combustion engines and thus mufflers and
silencers have been designed to reduce this noise.
Unfortunately, although the acoustic performance of
a muffler can sometines be successfully predicted in
the laboratory with artificial (loudspeaker type) sources,
295
-------
until recently most attempts 'co predict the perforii.^.iCt
of a muffler on an engine havt been disappointing. How
ever, in the last few years progress has ^oen made .and
now prediction of the acoustic performance cf real muff let.-.
on engines can be made with more accuracy, although un-
known effects still remain.
Most muffler designs manufactured still rely heavily
on a great deal of empiricism, experience and experiment.
Recent U.S. legislation to improve fuel efficiency of
automobiles has produced increased pressure to save
weight in mufflers and optimize acoustic performance.
It is to be expected that this pressure will increase
efforts to improve theoretical models -of the
acoustic performance of mufflers still further in the
near future.
MUFFLER CLASSIFICATION
Mufflers.can be classified into.two main types,
reactive and dissipative. Reactive mufflers are composed
of chambers of different volume and shape and work by
reflecting most of the incident acoustic energy bacK towards
the source (the engine). Dissipative mufflers on the othei
hand are lined with acoustip material which absorbs the
sound energy and converts it into heat [1,2,3]. Mufflers
can be designed to be partly reactive and partly dissipa-
tive and in fact some internal combustion engine mufflers
do sometimes incorporate absorbing materials. However,
296
-------
this material usually deteriorates because of the severe
temperature conditions and becomes clogged, melts or
fatigues. Thus most automobile mufflers manufactured
today are of the reactive type and do not incorporate
absorbing materials. Nevertheless some dissipation
can still occur in a reactive muffler due to viscous
dissipation.
Reactive mufflers can be further subdivided into
straight-through and reverse-flow types [4,5]. Figure 1
shows some typical straight-through types. These
mufflers are usually comprised mainly of expansion
chambers (chambers in which the area is suddenly increased
then decreased) and concentric tube resonators (side
branch Helmholtz resonators) . Reverse-flow types car.
be built in many different configurations. A typical
reverse-flow muffler is shown in Figure 2. Figure 3
shows a photograph of another similar reverse-flow
muffler. As shown such mufflers consist of several
chambers connected by straight pipes. There are usually
two end chambers in which the flow is reversed and one
or more large low-frequency Helmholtz resonators. Some-
times louver patches are used to produce side branch
Helmholtz resonators (which reflect high frequency
noise). In addition cross flow is often allowed to occur
and attenuation is then created by interference o£ Bound
traveling over different path lengths. Most automobile
mufflers are of the reverse-flow type, although trucks
297
-------
can use either reverse-flow or straight through mufflers.
DEFINITIONS
The definitions of muffler performance in most common
use will be given here [5,6,7,8]. It should be noted,
however, that some authors use different nomenclature
and confusion can sometimes arise.
A. Insertion Loss (IL). This is the difference in the
sound pressure level measured at one point in space with
and without the muffler inserted between that point and
the source [7,8]. Insertion loss is a convenient quantity
to measure and its use is favored by manufacturers.
B. Transmission Loss (TL) . This is defined as 10 log,_
of the ratio of the sound power incident on the muffler to
the sound power transmitted. This is the quantity which
is most easily predicted theoretically and its use is
favored by those engaged in research.
C. Noise Reduction (NR). This is the difference in sound
pressure levels measured upstream and downstream of the
muffler.
D. Attenuation. This is the decrease in propagating sound
power between two points in an acoustical system. This
quantity is often used in describing absorption in lined
ducts where the decrease in sound pressure level per unit
length is measured [7,8].
298
-------
The first three definitions are used frequently in
work- on mufflers for automobile engines and they are
illustrated in Figure 4. it is. of interest to note
that these definitions are also used with similar
meanings to describev sound transmission through walls
or enclosures.
In general, the insertion loss, the transmission
loss and the noise reduction are not simply related,
since, except for the transmission loss, they depend
on the internal impedance of the source (engine) and
the termination impedance (radiation impedance of the
tail pipe). However, if the source and termination
impedances are equal to pc/S (i.e., the source and
the termination are non-reflecting), then
IL = TL < NR,
and usually,
NR - TL - 3dB.
DEVELOPMENT OF MUFFLER THEORIES
Although Quincke in the last century discussed the
interference of sound propagation through different length
pipes, theory of real use in muffler design was not
developed until the 1920's. This was probably partly-
because prior to this time it was difficult (if not
impossible to measure sound pressure quantitatively)
due to the lack of suitable microphones and partly due to
less need, because of the lower noise produced by engines.
299
-------
In 1922 Stewart, in the USA began developing acoustic
filter theory using a lumped parameter approach [9]. In
1927 Mason developed this theory further [10]. In Britain
and Germany in the 1930's work was conducted on designing
mufflers for aircraft [11] and single cylinder engines [12]
However it was not until the 1950's when another signi-
ficant improvement in muffler theory occured. Davis and
his co-workers [13,14] then developed theory for plane
wave propagation in multiple expansion chambers and side
branch resonators. They made many experiments and found
that in general their predictions of transmission loss
were good provided the cut-off frequency in the pipes
and chambers was not exceeded in practice. Above this
frequency, cross modes in addition to plane waves can
exist and one of their theoretical assumptions was
violated.
When Davis et al tried to use their theory to design
a helicopter muffler, their prediction was very disappoint-
ing, since they only measured about 10 dB insertion loss,
compared with the 20 dB they had expected from their
transmission loss theory. Davis et al tried to explain
this by saying that finite amplitude wave effects must
be important. However a more likely reason is their
neglect of mean flow which can be of particular importance
in insertion loss predictions. For a more complete
discussion of the assumptions made by Davis et al in their
theory see [5] .
300
-------
In the late 1950's Igarashi et al began to calculate
4
the transmission properties of mufflers using equivalent
electric circuits [15,16,17]. This approach is very con-
venient. The total acoustic pressure and total acoustic
volume velocity are related before and after the muffler
by using the product of four-terminal transmission
matrices for each muffler element [5]. The equivalent
electrical analog for a muffler is quite convenient since
electrical theory and insight may be brought to bear.
The four-terminal transmission matrices are also useful
since it is only necessary to know the four parameters
A, B, C, D which characterize the system. The parameter
values are not affected by connections to elements up-
stream or downstream as long as the system elements can
be assumed to be linear and passive.
Several transmission matrices have been evaluated
for various muffler elements by Igarashi et al [15,16,17]
and Fukuda et al [21,22,23]. Parrott [18] also gives
results for transmission matrices, some of which include
the effects of a mean flow. However, note that the
matrix given for a straight pipe carrying a mean flow
of Mach number M (equation 28 in [18]) is in error.
Sullivan has given the corrected result in [24].
In the middle and late 1960's and early 1970's
several workers including first Davies [25,26] and then
Blair, Goulbourn, Benson, Baites and Coates [27-32]
developed an alternative method of predicting muffler
301.
-------
performance based on shock wave theory. Perhaps this
work -was inspired by Davis's belief [13] that the failure
of his helicopter muffler design was caused by the fact
exhaust pressures are much greater than normally assumed
in acoustic theory so that finite amplitude affects
become important. This alternative method involves the
use of the method of characteristics and can successfully
predict the pressure-time history in the exhaust system.
Also, one-third octave spectra of the acoustic noise
have been predicted [32]. However, the method is time
consuming and expensive and has difficulties in dealing
with complex geometries and some boundary conditions.
Although such an approach is probably necessary and
useful with the design of mufflers for single cylinder
engines, so far this method has found little favor with
manufacturers of mufflers for multicylinder engines.
It appears furthermore that Davis's belief [13] may
have been incorrect. There are several other possible
reasons why Davis failed to obtain better agreement
between theory and experiment, each of which can be
important. These include [33]: neglect of mean gas
flow (and its effect on net energy transport), incorrect
boundary conditions for exhaust ports and tail pipe,
neglect of interaction between mean gas flow and sound
in region-s of disturbed flow, and, neglect of mean
temperature gradients in the exhaust system.
302
-------
In 1970 Alfredson and Davies published work which
shed new light on the acoustic performance of mufflers
[33,34,35,36,37], Alfredson working at Southampton
University mainly considered the design of long expan-
sion chamber type mufflers commonly used on diesel
engines. Alfredson's work has been important since
he has shown that (at least with the mufflers and engine
he studied) that acoustic theory could be used to predict
the radiated exhaust sound and the transmission loss of
a muffler and that finite amplitude effects could be
neglected, provided that mean gas flow effects were
included in the theory. Alfredson concluded that as
the mean flow Mach number approached M = 0.1 or 0.2
in the tail pipe, the zero flow theory overpredicted
the muffler effectiveness by 5 to 10 dB or more. The
most serious discrepancy occurred for values of reflection
coefficient R -> 1, This would occur for low frequency
(large wavelength). Alfredson computed this error to be
Error = 10 Iog1(){ [ (1 + M) 2 - (1 - M)2R2]/[1 - R2] } (1)
and the result is plotted in Figure 5.
As a check on his acoustic theory and on Equation (1),
Alfredson later measured the attenuation of an expansion
chamber and compared it with theory [35]. The result is
shown in Figure 6. The good agreement between theory
(with flow included) and experiment and poor agreement
with theory when flow was neglected seem to confirm
303
-------
that acoustic theory is probably adequate in many instances
in muffler design provided the effects of mean flow are
included in the model where necessary. These conclusions
are very important.
Another new development occured in 1970 when Young
and Crocker began the use of finite elements to analyze
the transmission loss of muffler elements [38]. The
reason for the use of finite elements is that some
chambers in reverse-flow mufflers (e.g., flow-reversing
end chambers and end-chamber/Helmholtz-resonators combinations)
are not a'xi-symmetric and thus difficult, if not impossible,
to analyze using classical assumptions of continuity of
pressure and volume velocity at discontinuities, even
in the plane wave region. The use.of a numerical technique
such as finite element analysis makes the acoustic per-
formance of complicated-shaped chambers possible to predict
even in the higher frequency cross-mode region. The work
of Young and Crocker [38,39,40,41,42] will be described in
some detail later in this paper.
Other investigators have since used finite elements
in muffler design. Kagawe and Omote [43] have used two-
dimensional triangular ring elements. Craggs [44] has
used isoparametric three-dimensional elements, while
Ling [45], using a Galerkin approach, included mean
flow in his acoustic finite element model. However,
Ling's work was mainly concentrated on propagation in
ducts rather than muffler design.
304
-------
Side branch resonators (known by manufacturers as
beaa cans or spit chambers), see Figures 2 and 3, have
recently been studied by Sullivan and Crocker [46,47]
in practical situations, axial standing waves can exist
in the outer concentric cavity of the resonator. Previous
theories have been unable to account for this phenomenon
(assuming the cavity acts like a lumped parameter stiffness)
Sullivan's work will be described in more detail later
in the paper.
Other developments in muffler design have included
the Bond Graph approach by Karnopp [48,49]. It is claimed
that this approach can extend the frequency range of
lumped parameter filter elements.
Another important topic little touched on so far is
the effect of flpw in mufflers. Various phenomena can
occur. Noise can be generated by the flow process.
Interactions can occur between the flow and sound waves.
Fricke and Crocker found that the transmission loss of
short expansion chambers could be considerably reduced
[50]. The effect appeared to be amplitude dependent
and a feedback mechanism was postulated. Kirata and
Itow [51] have studied the influence of air flow on side
branch resonators and concluded that the peak attenuation
is considerably reduced by flow. Anderson [52] has con-
cluded that a mean air flow causes an increase in the
fundamental resonance frequency of a simple single side-
branch Helmholtz resonator connected to a duct.
305
-------
Perhaps the most important development recently is
the two microphone method for determining acoustic pro-
perties described by Seybert and Ross [53] in work con-
ducted at the Herrick Laboratories. White noise is used
as a source. Two flush-mounted wall microphones are
used and measurements of the auto and .cross spectra
enable incident and reflected wave spectra and the
phase angj.e between the incident and reflected waves
to be determined. The method can be used to measure
impedance and transmission loss. Agreement between this
two microphone random noise method and the traditional
standing wave tube method is very good and the method
is very much more rapid (only 7 seconds of data were
used to obtain the plots given in Figures 7 and 8).
Figure 7 shows a.comparison between theory and experiment
2
for the power reflection coefficient R for an open end
tube and the phase angle. Figure 8 shows the transmission
loss, TL, of a prototype automobile muffler with a com-
parison between this method and the classical standing
wave ratio (probe tube) method (SWR). For TL measurements,
a third microphone was used downstream of the muffler.
CLASSICAL MUFFLER THEORY
A. Transmission Line Theory
We will first make some simplifying assumptions:
a) sound pressures are small compared with the mean pressure,
b) there are no mean temperature gradients or mean flow and
306
-------
c) viscosity can be neglected. If plane waves are assumed
to exist in a muffler element (see Figure 9) then the
acoustic pressure p anywhere in the muffler element can
be represented as the sum of left and right traveling
waves p+ and p~ respectively
p = p+ + p~, (2a)
p = P+ e~ikx + P~ eikx, (2b)
V = V+ + V", (3a)
V = (S/pc)(P+ e~lkx - P~ elkx), (3b)
V = (S/pc)(p+ - p~). (3c)
Note that p and V represent the magnitude (and phase) of
the total acoustig pressure and volume velocity. The time
dependence (constant multiplying factor e u ) has been
omitted for brevity. The right and left traveling acoustic
waves are represented by the + and ~ superscripts, respectively,
while P represents the pressure amplitude, S the cross
sectional area, pc/S the characteristic acoustic impedance
(traveling wave pressure divided by traveling wave volume
velocity), k = w/c, the acoustic wave number, w the angular
frequency, c the speed of sound, and p the fluid density.
Davis et al used theory such as this to predict the
transmission loss of various expansion chamber type
mufflers [13,14] by assuming 1) continuity of pressure
and 2) continuity of volume velocity at discontinuities.
307
-------
For example if there is a sudden increase in area at
station 1 and a sudden decrease in area at station 2,
then the chamber is known as an expansion chamber and
its transmission loss is given by:
TL = 10 log(|Pi/Pt|)2,
TL = 10 log1Q[l + i(m - 1/iri) 2sin2kL] . (4)
Equation (4) is easily derived from equations (2) and (3)
above by assuming the sudden area changes occur at
x = 0 and x - L and by assuming the continuity of pressure
and volume velocity at the area discontinuities. In
Equation (4), P. and P. are the pressure amplitudes of
the right traveling waves incident and transmitted by
the expansion chamber. Figure 10 gives a comparison between
theory (Equation (4)) and experiment from Davis et al
[13,14] .
B. Transfer Matrix Theory
An alternative approach is to assume that the pressure
p and volume velocity V at stations 1 and 2 in Figure 9 can
be related by:
P, = Ap + BV9, (5)
j_ t t.
and
V1 = Cp2 + DV2. (6)
An electrical circuit analogy can be used where the
pressure p is analogous to voltage and volume velocity V
308
-------
to current. This is known as the impedance analogy.
Note that an alternative mobility analogy is sometimes
used [5]. The circuit element can be represented by
the four pole element shown in Figure 11. If the muffler
section is simply a rigid straight pipe of constant cross-
section, then from Equations (2b) and (3b), the pressure
and volume velocity at stations 1 and 2 are:
PI = P+ + P~ , (7)
p = P+ e~ikL + P- eikL' (8)
'2
V1 = (S/pc) (P+ - P-), (9)
and V2 = (S/pc)(P+ e~lkL - p" elkL). (10)
The parameters A, B, C and D may be evaluated using
a "black box" system identification technique. To evaluate
A and C, assume that the matrix output terminals are open
i2kL
circuit, or V2 = 0. Then Equation (10) gives P+/P- = e
and Equations (5) and (6) give: A = P-i/Po and c = Vi/p2'
Using this result for P+/P", and Equations (7), (8) and (9),
after some manipulation, it is found that A = cos kL and
C = (S/pc) i sin kL. Similarly, to evaluate B -and D assume
that the matrix output terminals are short-circuited and
P2 = 0. Then Equation (8) gives P+/P~ = -e1 L and Equa-
tions (5) and (6) give B = P1A2 and D = V^Vj. Using this
result for P+/P~ and Equations (7), (9) and (10) , it is
found that B = (pc/S) i sin kL and D = cos kL.
309
-------
Substituting these results for A, B, C and D into
Equations (5) and (6) and writing them in matrix form
gives:
"PI"
_vl.
=
"A B"
C D
V
.V2.
:n
where the four pole constants (for a straight pipe of
length L) are:
A B
C D
cos kL
i(S/pc)sin kL
i(pc/S)sin kL
cos kL
(12;
Note that AD - BC = 1. This is a useful check on the derived
values of the four-pole parameters and is a consequence of
the fact that the system obeys the reciprocity principle [5].
The matrix in Equation (12) relates the total acoustic
pressure and volume velocity at two stations in a straight
pipe.
If several component systems are connected together in
series, as in Figure 12 then the transmission matrix of the
complete system is given by the product of the individual
system matrices:
B.
V
C2 D2
310
-------
This matrix formulation is very con-venient particularly
where a digital computer is used. The four pole constants
A, B, C and D can be found easily for simple muffler
elements such as expansion chambers and straight pipes
as has just been shown (see Equation (12)). They can
also be found in a similar manner for more complex
muffler shapes (reversing end-chambers and reversing
end-chamber/Helmholtz resonator combinations) by the
finite element method using the same black box identification
technique mentioned above (with alternatively P~ = 0 and
V2 = 0).
EXHAUST SYSTEM MODELING
It will now be shown that for any linear passive muffler
element that the transmission loss is a property only of the
muffler geometry (i.e., four-terminal constants A, B, C
and D) and unaffected by connection of subsequent muffler
elements or source or load impedances. On the other hand, it
will be shown that the insertion loss is affected by the
source and load impedances. Finally if it is desired to
predict the sound pressure level,outside of the tail pipe
it is necessary to have a knowledge not only of the source
(engine) impedance and load impedance but also of the
source (engine) strength - either pressure or volume velocity.
The transmission loss of a muffler is the quantity most
easily predicted theoretically and is certainly of guidance
in muffler design. However insertion loss or a prediction
311
-------
of the sound pressure radiated from the tail pipe are
much more useful to the muffler designer and these are
now discussed.
A. Transmission Loss
The engine-muffler-termination system may be modeled
as an equivalent electric circuit [19,20,24,54]. The
velocity .source model in Figure 13b will be used in the
derivations of TL (although the pressure source model gives
the same result). For simplicity, the mean-flow Mach
number M = 0, the cross-sectional areas of the muffler
inlet pipes S 'are assumed equal and there is no mean
temperature gradient in the muffler system. To determine
the transmission loss, the incident and transmitted pressure
amplitudes |p, ] and |pi| are needed. The transmitted pres-
sure Ip^l is most easily determined by making the tail
pipe non-reflecting (Z = pc/S ). Thus p~ = 0.
From Figure 13b (see Equations (2a) and (3c)):
Pl = pl + pl' (14)
V1 = (SQ/pc)(p+ - p~), (15)
V2 = (So/pc)p+, (16)
and from Equation (11):
P! + P! = A p+ + B p+(So/pc), (17)
(SQ/pc)(p+ - p~) = C p+ + D p+(So/pc), (18)
From the definition in Figure 4b:
312
-------
+
2 + 2
TL = 10 log 2 = 20 log^ | p / | p+ | . (19)
IPJI /PC
Then eliminating p~ in Equations (17) and (18) and
substituting into Equation (19) gives:
TL = 20 logl0{|A + B(SQ/pc) + C/(SQ/pc) + DJ/2}. (20)
Equation (20) is a similar result to that obtained by
Young and Crocker [40] . Except note that in [40] particle
velocity was used instead of volume velocity and so A, B,
C and D have slighly different definitions. Sullivan [24]
has also derived a result similar to Equation (20) in which
the mean temperature, cross-sectional area and mean flow in
pipes 1 and 2 are different.
The transmission loss TL is convenient to predict but
inconvenient to measure experimentally. With some care it
is possible to construct an anechoic termination from an
absorbently lined horn or absorbent packing [15,41] enabling
| pi | to be measured* directly. The quantity | p+| can also
be determined when the source (in Figure 13) is a loudspeaker,
by measuring the standing wave in the exhaust pipe, using a
microphone probe tube (although it is a laborious process) .
However if the transmission loss is determined in the
"real-life" situation with an automobile engine as a
source, the microphone probe tube is placed under severe
environmental conditions of high temperature and moisture condensatioi
Alternatively the transmission loss can be measured using
313
-------
two microphones instead of a probe tube as suggested by
Seybert and Ross [53] . However if a tail pipe anechoic
termination is used it , must be of special design to with-
stand the high temperature. Of much more practical
interest and much easier to measure with an engine as
a source is the insertion loss which is discussed next.
B. Insertion Loss. Using Figure 13b again gives:
Vl = Ve " Pl/Ze' (21)
V2 = P2/Zr, (22)
where Z and Z are the engine internal impedance and tail
pipe radiation impedance, respectively. Then from Equation
(11) :
.PI ^ Ap2 + Bp2/Zr, (23)
Vl = Cp2 + Dp2/Zr' (24)
Substituting for V, from Equation (21) into Equation (24)
a-nd combining Equations (23) and (24) to eliminate p,
gives :
P2 = ZeZrVe/(AZr + B + CZeZr + DZe) ' (25)
If a different muffler with four-terminal parameters A' ,
B1, C" and D1 is now connected to the engine, a new
pressure p results:
P2 = zeZrVe/(A'Zr + B' + c'zeZr + D'Ze)' (26)
314
-------
Thus
pi AZ + B + CZ Z + DZ
11 = * §_£ § (27)
P2 A'Zr + B1 + c'zezr + D'Ze'
This result is similar to that obtained by Sullivan [24].
If pi is measured with no muffler in place and only a short
(in wavelengths) exhaust pipe7 then A1 = D1 = 1, and B1 = C1 = 0
Then
pi AZ + B + CZ Z + DZ
P2 z
e
This result is similar to that obtained in [20] . Since
IL = 20 log Jp'/p2| it is seen from either Equation (27)
or (28) that unlike the TL, IL depends on both the internal
impedance of the engine and the tail pipe radiation impedance,
.besides the transmission characteristics of the muffler
itself. Several workers have predicted the insertion loss (IL)
of mufflers installed on engines, e.g., Young [40] and
Davies [55]. However they have normally had to rely on
assumed values of engine impedance (e.g., Z = 0,
pc/S or °°) , since measured values have not become
available until recen-tly.- Young's results for IL, [40],
will be discussed later.
In prediction of insertion loss, Zr must also be known.
Discussion on the problems of estimating Z and Zr follows
in a later section.
If the engine and radiation impedances are assumed to
be Z = Z = pc/S , then Equation (28) becomes:
315
-------
pi A(pc/S ) + B + C(pc/S )2 + D(pc/S )
^ OOU
and
IL = 20 Iog10|p^/p2| ,
IL =-20 log1(){|A + B(SQ/pc) + C/(SQ/pc) + D|/2}; (30)
a result identical to Equation (20). This demonstrates
the general case that the muffler transmission loss is
not equal to the insertion loss except when the insertion
loss is "measured with source and termination impedances
equal to the characteristic acoustic impedance pc/S . The
same conclusion can be reached intuitively or theoretically
(although it is more difficult than with transmission
matrix theory) by^studying the travelling wave solutions
(transmission line theory) in mufflers and the exhaust
and tail pipes.
C. Sound Pressure Radiated From Tail Pipe
A prediction of this quantity is of probably more impor-
tance to muffler designers than a knowledge of either trans-
mission loss or insertion loss. After all, the -radiated
sound pressure level is the quantity which finally deter-
mines the acceptability of a muffler. Examining Equation
(25), shows that if the engine volume velocity source
strength VQ, engine impedance Zg, radiation resistance Z
and muffler four-terminal (fourpole) parameters A, B, C and D
are known, then the total pressure amplitude (and phase)
316
-------
at the end of the tail pipe p2 can be calculated. It is a
fairly simple matter to calculate the radiated pressure
amplitude |p | at- distance r from the tail pipe outlet
[33,34,36], The method used is to assume monopole radia-
tion from the tail pipe so that the net acoustic intensity
transmitted out of the tail pipe is equal to the intensity
in the diverging spherical wave at radius r. This gives:
2TT a2(|p+|2/2p2c2){(l + M)2 - (1 - M)2R2(M)}
= 4u r2|pr|2/2poco (31)
where a is the tail pipe radiusf and R(M) the tail pipe re-
flection coefficient (dependent on Mach number) of the
mean flow. Subscript 2 refers to conditions just inside
the tail pipe. From Equations (2a) and (3c), at any
station in the muffler:
2p+ = p + (pc/So)V, (32)
and at the tail pipe exit:
P2 = V2Zr. (33)
Thus, at the tail pipe exit, from Equations (32) and (33):
p2 = 2p+/[l + (pc/So)/Zr] (34)
and substituting Equation (34) into (25) gives:
P2 = VeZe(Zr + PC/SQ)/2[AZr + B + CZ^^. + DZQ] . (35)
317
-------
Taking the modulus of Equation (35) and substituting it
into Equation (31) eliminates p+ and gives the pressure
|p | in terms of the source volume velocity, V , the
engine and tail pipe radiation impedances, Z and Z ,
the muffler fourpole parameters, the tail pipe reflection
coefficient R(M) and the mean-flow Mach number in the
tail pipe, M.
TAIL PIPE RADIATION IMPEDANCE, ENGINE IMPEDANCE AND SOURCE
STRENGTH
A. Tail Pipe Radiation
Early work on mufflers was hampered by a lack of know-
ledge of the reflection of waves at the end of the tail pipe.
As Alfredson discusses [33], various assumptions have been
made in the past about the magnitude and phase of the
reflection (some workers assuming the reflection coefficient
R was zero and some, one). In 1948, Levine and Schwinger [56]
published a rigorous, lengthy theoretical derivation of the
reflected wave from an unflanged circular pipe. The
solution assumes plane wave propagation in the pipe and no
mean flow. In 1970, Alfredson measured the reflection
coefficient R and phase angle 6 of waves in an engine tail
pipe using the engine exhaust as the source signal. The
motivation was to determine if a mean flow and an elevated
temperature had a significant effect on the zero flow reflection
coefficient and phase calculated by Levine and Schwinger.
Both the theoretical results of Levine'and Schwinger and
Alfredson's experimental results are given in Figure 14.
318
-------
Alredson's experimental results show only a 3 to 5 per-
centage increase in the reflection coefficient and virtually
no change in the phase angle , as the flow and temperature
increase to 'those conditions found in a typical engine tail pipe
Either Alfredson's or Levine and Schwinger's results for
R and 8 can be used to determine the tail pipe radiation-
impedance Z used in insertion loss or sound pressure
predictions [Equations (27) and (28) or (25) and (35)].
The ratio of the pressure and volume velocity at
the tail pipe exit yields the radiation impedance Z :
p2 = p+ + p- = p+(l + Re16),
V2 = (SQ/P2c2)(P+ - p-) = P^(So/p2c2)(l - Re16),
i A i ft
" Zr = P2/V2 = (P2C2/SJ (1 + Re )/(1 ~ Re >• (36)
B. Engine Impedance and Source Strength
Until recently, values of engine impedance have been
completely speculative. Values of Z of 0, pc/S and °°
have been assumed by various workers in making insertion
loss calculations. Other experimenters have tried to
simulate these different values in their idealized experi-
mental arrangements. Values of Z = °° and 0, correspond
to constant volume velocity (current) and constant pressure
(voltage) sources, respectively. Suppose the muffler and
termination impedances shown in Figure 13 are lumped
together as a load impedance, then Figures 13b and 13c
reduce to Figures 15a and 15b respectively.
319
-------
For the volume velocity source, V1 = V Z /(Z + Z.)
and if the internal impedance Z -»• °°, V, -> V . A constant
volume velocity is supplied to the load, independent of its
impedance value, (provided it remains finite). When Z -* °°,
this source is known as a constant volume velocity source.
For the pressure source, p, = p Z /(Z + Z ) and if the
_L 6 X/ Q X-
internal impedance Z -> 0, p. ->- p . A constant acoustic
ti .L c ^™—«P»«^—^
pressure is supplied to the load terminals independent of
of the impedance value (provided it remains finite also).
When Z -> 0 this source is known as a constant pressure
source. Note that if Z pc/S in either model, that
constant sources are not obtained in either model. These
constant volume velocity and constant pressure sources are
equivalent to constant current and voltage sources which
are well known in .electrical circuits (see, e.g., [57]).
It is of course unlikely that engine impedance approxi-
mates either 0, pc/S or °°. However, it could approach one
of these values in certain frequency ranges. Some have
even questioned the meaning of engine impedance since it
must vary with time as exhaust ports close and open.
There are at least three approaches to model the engine
source characteristics. Without directly using the con-
cept of engine impedance as such, Mutyala and Soedel
[58,59], working at the Herrick Laboratories, have used
a mathematical model of a single-cylinder two-stroke
engine connected to a simple expansion chamber muffler.
The passages and volumes are treated as lumped parameters
320
-------
and kinematic, thermodynamic and mass balance equations are
used. Good agreement between theory and experiment was
obtained for the radiated exhaust noise.
Galaitsis and Bender ['60] have used an empirical approach
to measure engine impedance directly. Using an electro-
magnetic pure tone source and by measuring standing waves
in an impedance tube connected to a running engine they
were able to determine the engine internal impedance. At
low RPM the impedance fluctuated. However, at high RPM
the impedance approached pc/S at higher frequency. Ross
[61] ,has also, used a similar technique.
A third approach to the determination of engine impedance
(and source strength) is the two load method. This method
is well known in electricity but has been little tried in
acoustics. Kathuriya and Munjal [54] have recently discussed
this method theoretically but apparently have yet to try it
in practice.
Using the. pressure source representation [54] (see Figure
15b) and-two different known loads Z^ and Z^, two simultaneous
equations are obtained:
P! - PeV(Ze + V' (37)
p'=pZ'/(Z+Z'). (38)
Jr \ Lr Q v fv P
Eliminating p in Equations (37) and (38) gives:
2 = (P-i - P-I)/ (P-J/Zn - P-i/Zj). (39)
6 J. -L J. J6 _L X
321
-------
Substitution of Z in Equation (37) or (38) now gives the
source strength p . Kathuriya and Munjal suggest using
two different length pipes so that there is little change
in back pressure and so that (presumably) the load impedances, z
and Z^ (comprised of straight pipe and radiation impedance) are
well known. In order to remove the necessity to measure
p, inside the tail pipe (where the exhaust gas is hot) it
should be possible to measure the sound pressure radiated
from the tail pipe p since this can be related to the
pressure p, in the straight pipe by equations such as
(31) and (34) .
Egolf [62] has used this two load method in the design
of a hearing aid. Sullivan [24] discusses the limitations
of the method.
RESEARCH WORK ON MUFFLER DESIGN AT HERRICK LABORATORIES
A program of research on the acoustic performance of
automobile mufflers has been conducted at Herrick Laboratories
since 1970.
Finite Element Analysis
Young and Crocker [38,39,40,41,42] were the first to
use finite element analysis in muffler design. So far in
this paper it has been assumed that acoustic filter theory
[13,14] provides a sufficient theoretical explanation for
the behavior of muffler elements. This filter theory is
normally based on the plane wave assumption. However when
a certain frequency limit is reached (known as the cut-off
322'
-------
frequency), the filter ceases to behave according to plane
wave theory. (This cut-off frequency is usually proportional
to the pipe or chamber diameter.) In addition, if the muffler
element shape is complicated, the simple plane wave assumptions
and the boundary conditions are difficult to apply.
In Young and Crocker's work a numerical method was
produced to predict the transmission loss of complicated
shaped muffler elements. In this approach,variational
methods were used to formulate the problem instead of the
wave equation. The theoretical approach is described in
detail in [38-42] and will not be given in detail here.
It is assumed that the muffler element is composed of a
volume V of perfect gas with a surface area S. The surface S
is composed of two parts: one area over which the normal
acoustic displacement, is prescribed and the other .area
over which the pressure is prescribed. The pressure field
in the muffler element is solved by making the Langrangian
function stationary [38]. Thus this approach is essentially
an approximate energy approach. The muffler element is
divided into, a number of subregions (finite elements).
At the corners of the elements the acoustic pressure and
volume velocity are determined. The four pole parameters
A, B, C and D relating the pressure and volume velocity
before and after the muffler element are obtained in a
similar manner to that described above assuming that
the matrix output terminals are alternately open-circuited
or short-circuited [38].
323
-------
At the corners of the elements the acoustic pressure
and volume velocity are determined. The four pole para-
meters A, B, C, D relating the pressure and volume velocity
before and after the muffler element are obtained in a
similar manner to that described above assuming that the matrix
output terminals are alternately open-circuited or short-
circuited [38] .
In order, to check the finite element approach and
computer program, it was first applied to the classical
expansion chamber case [40]. The dimensions of the simple
expansion chamber used are given in Figure 16a. The
chamber was 8 inches (0.20 m) long and 10 inches (0.25 m)
in diameter. Since the chamber was symmetrical, only
half the chamber was represented with finite elements.
Three finite element -models were studied. The first had
8 elements with 16 nodaj. points, the second had 16 elements
with 28 nodal points (see Figure 16b). The third had
24 elements with 38 nodal points.
Figure 17 shows the transmission loss predicted by
the three finite element models and by the classical
theory for an expansion chamber (see Equation (4)). Figure
17 shows the rapid convergence of the finite element
approximation. Eight elements are insufficient to predict
the transmission loss (TL), although the TL predicted
by 16 or 24 elements is about the same. Note, however,
that above about 1100 Hz, the classical theory and the
324
-------
finite element TL predictions diverge. Above this
frequency the chamber-diameter-to-wavelength-ratio
becomes less than 0.8 and higher modes, in addition
to plane waves, can exist in the expansion chamber.
However, the classical theory (Equation (4)) only
predicts the plane wave performance.
Having shown that the finite element program could
be used to' predict transmission loss successfully on known
chambers., it was now used to examine chambers such as
reversing flow end chambers (see Figure 3), end chamber
Helmholtz resonator combinations and finally mufflers
comprised of combinations of straight pipes, end
chambers and up to two Helmholtz resonators.
A typical end chamber examined is shown in Figure 18.
The measurement of transmission loss was based on the
standing wave method, see Figure 19. An acoustic
driver (H) was used to supply a pure tone signal and
the standing wave in the test section (J) was measured
with the microphone probe tube (I). Using standing wave
theory the amplitude of the incident wave was determined
by measuring the maxima and minima of the standing wave
at different frequencies. The transmitted wave was deter-
mined by a single microphone (M) since the reflections
were minimized by the anechoic termination (L). A steady
mean air flow could be supplied to the plenum chamber
(G) and was used to investigate flow effects on transmission
loss in some experiments.
325
-------
Figures 20 and 21 show the predicted and measured
transmission loss of two different shape reversing end
chambers, with and without a mean air flow of 110 ft/sec
(33.5 m/s). Neither end chamber examined had a pass tube.
The first chamber has side-in side-out (SI-SO) tubes and
the second side-in center-out (SI-CO) tubes. It is
observed that experimental agreement with theory is
good and that flow effects appear small at the mean flow
velocity (Mach number) used. Part of the volume appeared
to act as a side-branch with the SI-CO chamber (Figure 21).
The theory developed was then used to conduct a theoretical
parametric study on reversing end chambers as dimensions,
and locations of inlet, outlet and pass tubes were changed.
The results are given in [41].
Figures 22 a'nd 23 show the predicted and measured
transmission loss of similar SI-SO and SI-CO end chambers
both of which have pass tubes. Both the cases when the
end chambers have Helmholtz resonators attached (solid
line) and when there are no resonators (broken line)
are shown. The no-resonator cases are similar to Figures
20 and 21, except that here pass tubes are present.
It should be noted that the experimental points were
measured without flow but with resonators attached.
The predictions were made by dividing both the end
chamber and the resonator into finite elements [41].
Although only two-dimensional finite elements were
used, the third dimension and the elliptical cross-
326
-------
sectional shape were allowed for by varying the mass of
the elements corresponding to their thickness [38-42].
It is noted in Figures 22 and 23 that the addition of
the Helmholtz resonators produces sharp attenuation-
peaks in the transmission loss curves. The first
resonance frequency peak at 350 Hz agrees well with
the value of 356 Hz calculated for the resonance fre-
quency of a Helmholtz resonator using lumped parameter
(mass-spring) theory [42]. The higher frequency peak
must be produced by a higher mode resonance caused by
interactions between the Helmholtz resonators and the
end chambers.
Figure 24 shows that the positioning of the resonator
neck is theoretically an important factor in determining
the transmission loss curve [42].
Figures 25, 26 and 27 show the predicted and measured
transmission loss for three different muffler combinations,
The predictions were made by combining the predicted four
pole parameters of the end chamber systems with those
of the straight pipes using the matrix multiplication
method discussed earlier (see Equation (13)). The
muffler combinations shown, in Figures 25, 26 and 27
are typical of automobile reverse flow mufflers used
in the USA except that cross flow elements and side
branch concentric resonators are absent. It was shown
that at least at the low Mach number used (flow velocity
of 32 m/s) that there was very little difference in the
327
-------
transmission loss measured with or without flow. Flow
effects may be more important at higher flow rates (correspond-
ing to higher engine loads). Also flow is expected to have
a greater effect on the radiated sound (see Equation (1)
and Figure 5).
PREDICTION OF CONCENTRIC TUBE SIDE BRANCH RESONATORS
Sullivan and Crocker [46,47] have examined the trans-
mission loss of concentric tube resonators (sometimes
known as "spit chambers" or "bean cans", (See Figure 3).
These resonators which are often used to provide higher
frequency attenuation are constructed by placing a
rigid cylindrical shell around a length of perforated
tube, thus forming an unpartioned cavity. Sullivan and
Crocke'r used a one-dimensional control volume approach
to derive a theoretical model which accounted for the
longitudinal wave motion in the cavity and the coupling
between the cavity and the tube via the impedance of
the perforate.
Figures 28 and 29 show the transmission loss for
both short and long resonators [46,47]. In short resonators
the primary resonance frequency f is less than the^first
axial modal frequency f-^ of the cavity, (f^ = c/2£) where
c is the speed of sound and £ the length. If fr > f^,
then the cavity is said to be long. The transmission
loss of short resonators (Figure 28) is characterized by
two peaks. The first resonance peak results from the
328
-------
coupling of the center tube with the concentric cavity
and its frequency f can be calculated approximately
from the branch Helmholtz equation [46,47]. However
in Figure 28, the Holmholtz frequency f is less than
the fundamental frequency f by 27%. The frequency
of the second peak in Figure 28 is related but not
equal to the first axial cavity modal frequency f, = c/2£
The performance of concentric tube resonators is
dependent on the parameter k £ where k = 2ir f /c =
Here k is the wave number of the Helmholtz resonance
frequency f , c is the speed of sound, and C, V and
£ are the conductivity, volume and axial length of
the resonator respectively -
In Figure 29 the transmis: 'on loss of a long resonator
is shown. Here the primary res ance frequency f occurs
above the first and several other cavity longitudinal
standing wave modal frequencies. F' ure 30 shovs the
theoretical effect of changing the porosity of a resonator
of constant length 66.7 mm so that as the porosity is
increased from 0.5% to 5.0%, the primary resonance fre-
quency f and the first axial modal frequency f.. are
gradually merged to provide a wide band of high trans-
mission loss [46,47].
INSERTION LOSS
The effect of source impedance on insertion loss
was investigated theoretically by Young [39]. Some results
329
-------
are shown in Figures 31 and 32. In Figure 31 it is
seen that there is a large difference between insertion
loss curves for a muffler for the three different
source impedances investigated: Z^ = 0, pc/S, and °°,
when the prediction is made for discrete frequencies.
However Figure 32 shows that'if the insertion loss is
averaged on an energy basis (with a theoretical 25 Hz
filter) that the differences in insertion loss predictions
are much less. Note that the vertical scales in Figures
31 and 32 are different and that a different engine firing
frequency is chosen. Also of considerable interest is
the fact that in both figures the transmission loss
curve passes through the middle of the insertion loss.
curves. In Figure 32, the hills and valleys in the
insertion loss curves are thought to be caused by
standing waves in the lengths of straight (exhaust and
tail) pipes in the muffler systems.
CONCLUSIONS
This paper has reviewed briefly the historical develop-
ment of theory to predict the acoustic performance of
mufflers (silencers) used on internal combustion engines.
Research conducted at Herrick Laboratories has been
reviewed in a little more detail.
It seems that theory has now been developed which
can predict fairly accurately the transmission loss (TL)
of mufflers particularly when loudspeaker (or acoustic
330
-------
driver) type source-s are used. It is more difficult
to predict the transmission loss of a muffler when it
is installed on an engine and high mean flow rates
and severe temperature gradients exist in the muffler.
It was shown theoretically that if it is desired
to predict the insertion loss of a muffler, then it
is necessary to know the source (engine) and radiation
impedance. Although the radiation impedance of a
tail pipe has been known theoretically for some time
[56] , the impedance of engines has only recently
been measured [60,61]. However Young has shown
theoretically [39] that source (engine) impedance be-
comes less important, provided narrow band predictions
of insertion loss, IL, are not required and some fre-
quency averaging can .be tolerated.
It would seem that for the purposes of a quick
bench test to compare the transmission loss and/or
insertion loss of different mufflers, an acoustic
driver source could be used. However, in this case,
flow effects and temperature gradient effects would be
lost. These, however, may be less important in trans-
mission loss predictions then in insertion loss pre-
dictions. Flow effects could be included by supplying
a mean flow through the muffler from a fan or blower
source. Insertion loss could be measured with such an
experimental set-up provided narrow band results are
not required.
331
-------
Because flow, temperature gradient (and engine
impedance) effects are known to be important in muffler
acoustic performance, the only real way to test a muffler
is on a real engine. Thus a "standard" engine could be
used and insertion loss of different mufflers measured
with it and compared with each other. The comparisons
between mufflers should be applicable to other engines
provided the mean flow is not vastly different and
provided some frequency averaging is used. In any
case it may be almost as easy to use an engine as a
source,than to try to make an artificial source from
an acoustic driver and fan or blower combination.
ACKNOWLEDGMENTS
This paper is based in part on a paper which appeared
in the NOISE-CON 77 Proceedings and on results presented
previously in some other papers. However, some sections
of the paper are new. Much of the research work conducted
at Herrick Laboratories which was described in this paper
was funded by' contracts from Arvin Industries, Columbus,
Indiana.
332
-------
REFERENCES
1. M.J. Crocker, Mufflers, in Tutorial Papers on Noise
Control, Edited by M.J. Crocker, INTER-NOISE 72,
pp. 40-44. Also published in Reduction of Machinery
Noise (Rev. Ed.), edited by M.J. Crocker, Purdue
University, 1975, pp. 112-116.
2. I. Dyer, "Noise Attenuation of Dissipative Mufflers,"
Noise Control, 2_, 3, 1956, pp. 50-57.
3. G.J. Sanders, "Silencers: Their Design and Application,"
Sound and Vibration, February 1968, pp. 6-13.
4. C.E. Nelson, "Truck Muffler Design," Noise Control, 2_,
3, 1956, pp. 24-27 and 77.
5. M.J. Crocker, "Internal Combustion Engine Exhaust
Muffling," NOISE-CON 77 Proceedings, October 17-19,
1977, pp. 331-358.
6. P.A. Franken, "Reactive Mufflers," Chapter 16 in Noise,
Reduction, L.L. Beranek (Editor), McGraw-Hill Book Co.,
New York, 1960.
7. N. Doelling, "Dissipative Mufflers," Chapter 17 in
Noise Reduction, L.L. Beranek (Editor), McGraw-Hill
Book Co., New York, 1960.
8. T.F.W. Embleton, "Mufflers," Chapter 12 in Noise and
Vibration Control, L.L. Beranek (Editor), McGraw-Hill
Book Co., New York, 1971.
9. G.W. Stewart, "Acoustic Wave Filters," Phys. Review,
20, 1922, p. 528. See also G.W. Stewart, Phys. Review,
23, 1924, pp. 520, and -G.W. Stewart, Phys. Review. 25,
1925, pp. 9.0.
10. W.P- Mason, Bell Sys. Tech. Journal, 6_, 1927, pp. 258.
11. A.W. Morley, Progress in Experiments in Aero-Engine Ex-
haust Silencing, R&M No. 1760, British A.R.C., 1937.
333
-------
12. H. Martin, V. Schmidt and W. Willins, The Present Stage
of Development of Exhaust Silencers, RTF TIB Translation
No. 2596,British Ministry of Aircraft Production (from
MTZ, No. 12, 1940).
13. D.D. Davis, Jr., G.M. Stokes, D. Moore and G.L. Stevens, Jr.,
"Theoretical and Experimental Investigation of Mufflers
with Comments on Engine Exhaust Muffler Design," NACA
1192, 1954.
14. D.D. Davis, Jr., "Acoustical Filters and Mufflers,"
Chapter 21 in Handbook of Noise Control, C.M. Harris
(Editor), McGraw-Hill Book Co., New York, 1957.
15. J. Igarashi and M. Toyama, Fundamental of Acoustical
Silencers (I), Report No. 339, Aeronautical Research
Institute, University of Tokyo, December 1958, pp. 223-
241.
16. T. Miwa and J. Igarashi, Fundamentals of Acoustical
Silencers (II) , Report No. 344, Aeronautical Research
Institute, University of Tokyo, May 1959, pp. 67-85.
17, J. Igarashi and M. Arai, Fundamentals of Aco'ustical
Silencers, (III) , Report No. 351, Aeronautical Research
Institute, University of Tokyo, February 1960, pp. 17-31.
18. T.L. Parrott, "An Improved Method for Design of Expansion-
Chamber Mufflers with Application to an Operational
Helicopter," NASA TN D-7309.
19. K.U. Ingard (unpublished information) as reported in
Chapter 21, Handbook of Noise Control, C.M. Harris
(Editor), McGraw-Hill Book Co., New York, 1957 (see
Figure 21.47).
20. E.K. Bender and A.J. Brammer, "Internal Combustion
Engine Intake and Exhaust System Noise," J. Acoust.
Soc. Am. , 58 , 1, 1975. ~'
21. M. Fukuda, A Study on the Exhaust Muffler of Internal
Combustion Engines, Bulletin of JSME, 6_, 22, 1963,
pp. 255-269. ~
22. M. Fukuda, A Study on Characteristics' of Cavity-Type
Mufflers, (1st Report), Bulletin of JSME, 12, 50, 1969,
pp. 333-349.
23. M. Fukuda and J. Okuda, A Study on Characteristics of
Cavity-Type Mufflers (2nd Report) Bulletins of JSME, 13,
55, 1970, pp. 96-104.
334
-------
24. J.W. Sullivan, Modelling of Engine Exhaust System Noise,
paper to be presented at ASME Winter Conference, Atlanta,
Georgia, November 1977.
25. P.O.A.L. Davies, "The Design of Silencers for Internal
Combustion Engine," Jour. Sound and Vib. .!_, 2, 1964,
pp. 185-201.
26. P.O.A.L. Davies and M.J. Dwyer, A Simple Theory for
Pressure 'Pulses in Pipes, Proc. Inst. Mech. Eng.,
London, 1963.
27. G.P. Blair and J.R. Goulbourn, Pressure-Time History in
the Exhaust System of a High Speed Reciprocating Internal
Combustion Engine, SAE Transactions, 76, paper 670477, 1967.
28. R.S. Benson and J.S. Foxcroft, Nonsteady Flow in Internal
Combustion Engine Inlet and Exhaust Systems, Inst. Mech.
Eng. (London), paper no. 3, 1970, .pp. 2-26.
29. G.P- Blair and J.A. Spechko, Sound Pressure Levels
Generated by Internal Combustion Engine Exhaust Systems,
SAE Paper 720155, 1972.
30. G.P. Blair and S.W. Coates, Noise Procuded by Unsteady
Exhaust Efflux from an Internal Combustion Engine, SAE
Transactions, 82, paper 730160, 1973.
31. S.W. Coates,,The. Prediction of Exhaust Noise Characteristics
of Internal Combustion Engines, Ph.D. Thesis, Queen's
University of Belfast, Northern Ireland, April 1974.
32. S.W. Coates and G.P. Blair, Further Studies of Noise
Characteristics of Internal Combustion Engine Exhaust
Systems, SAE Paper 740713, 1974.
33. R.J. Alfredson and P.O.A.L. Davies, "The Radiation of
Sound from an Engine Exhaust," Jour. Sound and Vib.,
I3_, 4, 1970, pp. 389-408.
34. R.J. Alfredson, The Design and Optimization of Exhaust
Silencers, Ph.D. Thesis, University of Southampton, 1970.
35. R.J. Alfredson and P.O.A.L. Davies, "Performance of
Exhaust Silencer Components," Jour., Sound and Vib.,
15_, 2, 1971, pp. 175-196.
36. R.J. Alfredson, The Design of Exhaust Mufflers Using
Linearized Theoretical Models, SAE Paper 719139,
pp. 1048-1052.
335
-------
37. P.O.A.L. Davies and R.J. Alfredson, Design of Silencers
for Internal Combustion Engine Exhaust Systems; Proc. J.
Mech. E.,'Conference Vibration and Noise in Motor
Vehicles, pp. 17-23.
38". C.-I. J. Young and M.J. Crocker, Muffler Analysis by
Finite Element Method, Ray W. Herrick Laboratories,
Purdue University Report No. HL 71-33, December 1971.
39. C.-I. J. Young, Acoustic Analysis of Mufflers for Engine
Exhaust Systems, Ph.D. Thesis, Purdue University,
August 1973.
40. C.-I. J. Young and M.J. Crocker, "Prediction,of Trans-
mission Loss in Mufflers by the Finite-Element Method,"
Acoust. Soc. Am., 57, 1, 1975, pp. ,144-148.
41. C.-I. J. Young and M.J. Crocker, "Acoustical Analysis,
Testing, and Design of Flow-Reversing Muffler Chambers,"
Acoust. Soc.'Am., 60, 5, 1976, pp. 1*111-1118.
42. C.-I. J. Young and M.J. Crocker, Finite-Element Acoustical
Analysis of Complex Muffler Systems With and Without Wall
Vibrations, NOISE CONTROL ENGINEERING, 1977, 9_, 2 , pp. 86-93
43. Y. Kagawa and T. Omote, "Finite-Element Simulation of
Acoustic Filters of Arbitrary Profile with Circular
Cross Section," Jour. Acoust. Soc. Am., 1976, 60, 5,
pp. 1003-1013.
44. A. Craggs, "A Finite Element Method for Damped Acoustic
Systems: An Application to Evaluate the Performance of
Reactive Mufflers," Journ. Sound and Vib., 48, 3, 1976,
pp. 377-392. —
45. S.F. Ling, A Finite Element Method for Duct Acoustics
Problems, Ph.D. Thesis, Purdue University, August 1976.
46. J.W. Sullivan, Theory and Methods for Modelling Acoustically-
Long, Unpartitioned Cavity Resonators for^Engine Exhaust
Systems, Ph.D. Thesis, Purdue University, Decembe-r .1974.
47. J.W. Sullivan and M.J. Crocker, "A Mathematical Model
for Concentric Tube Resonators," submitted to J. Acoust.
Soc, Am. for publication, May 1977.
48. D. Karnopp, "Lumped Parameter Models of Acoustic Filters
Using Normal Modes and Bond Graphs," J. Sound Vib., 42,
4, 1975, pp. 437-446. —
336
-------
49, D. Karnopp et al, "Computer-Aided Design of Acoustic
Filters Using Bond Graphs," NOISE CONTROL ENGINEERING, 1975
4_, 3, pp. 114-118.
50. F.R. Fricke and M.J. Crocker, Sound Amplification in
Expansion Chambers, Inter-Noise 75.
51. Y. Kirata and T. Itow, "Influence of Air Flow on the
Attenuation Characteristics of Resonator Type Mufflers,"
Acustica, _28_, 1973, pp. 115-120.
52. J.S. Anderson, "The Effect of an Air Flow on a Single
Side Branch Helmholtz Resonator in a Circular Duct,"
J. Sound Vib., 5,2, 3, 1977, pp. 423-431.
53. A.F. Seybert and D.F. Ross, "Experimental Determination
of Acoustic Properties Using a Two-Microphone Random-
Excitation Technique," Jour. Acoust. Soc. Am., 61, 5,
1977, pp. 1362-1370.
54. M.L. Kathuriya and M.L. Munjal, A Method for the Experi-
mental Evaluation of the Acoustic Characteristics of an
Engine Exhaust System in the Presence of Mean Flow. J.
Acoust. Soc. Am., 60, 3, 1976, pp. 745-751.
55. P.O.A.L. Davies, Exhaust System Silencing the Institution
of Marine Engineers, 1972, pp. 46-51.
56. M. Levine and J.. Schwinger, "On the Radiation of Sound
from an Unflanged Circular Pipe," Physical Review, 73,
4, 1948, pp. 383-406.
57. H.H. Skilling, Electrical Engineering Circuits, John
Wiley and Sons, Inc., New York, 1957, pp. 23-24.
58. B.R.C. Mutyala, A Mathematical Model of Helmholtz
Resonator Type Gas Oscillation Discharges ofrTwo-
Cycle Engines," Ph.D. Thesis, Purdue University,
December 1975.
59. B.R.C. Mutyala and W. Soedel, "A Mathematical Model
of Holmholtz Resonator Type Gas Oscillation Discharges
of Two-Stroke Cycle Engines," Journ. Sound and Vib.,
4^, 4, 1976, pp. 479-491.
60. A.G. Galaitsis and E.K. Bender, "Measurement of the
Acoustic Impedance of an Internal Combustion Engine,"
Jour. Acoust. Soc. Am., 58^ (Supplement No. 1), Fall
1975.
337
-------
61. D. Ross, Experimental Determination of the Normal
Specific Acoustic Impedance of an Internal Combustion
Engine, Ph.D. Thesis, Purdue University, 1976
(unpublished).
62. D.P. Egolf, A Mathematical Scheme for Predicting the
Electro-Acoustic Frequency Response of Hearing Aid
Receivers - Earmold - Ear Systems, Ph.D. Thesis,
Purdue University, August 1976.
338
-------
Figure 1a. Single Expansion Chamber
Figure 1b. Double Expansion Chamber With Internal
Connecting Tubes
Figure 1c. Single Chamber Resonator
Figure 1d. Double Chamber Resonator
END CHAMBER (I)
SIDE BRANCH
RESONATOR
FLOW IN
FLOW OUT
CROSS FLOW
END CHAMBER (21
HELMHOLTZ
RESONATOR
Figure 2. Typical Reverse - Flow Automobile Muffler
339
-------
CONCENTRIC TUBE
RESONATORS
.OUVERS
HELMHDLTZ RESONATORS
FLCH-REVERSING CHAMBER
WITH TWO HELMHOLTZ
RESONATORS CONNECTED
SIMPLE
FLOW-REVERSH
CHAMBER
INLET
OUTLET
END PLATE
CROSS-FLOW
CHAMBER
\|
THROATS OF THE
HELMHOLTZ RESONATORS
Figure 3 - Photograph showing cross section of connon US reverse - flow
automobile muffler with different parts indicated
340
-------
AREAS.
INTENSITY
a) INSERTION LOSS
IL =L • L
AREA S
b) TRANSMISSION LOSS
TL = 10 log 10S|l'/St't
c) NOISE REDUCTION
NR = L -L
P1 P2
Figure 4. Definitions of Muffler Performance
10
LU
>
§
cc
cc
1.0
OUTLET
MACHNO.M
MACH NUMBERS
0.9 0.8 '0.7
REFLECTION COEFFICIENT,R
0.6
Figure 5. Radiated Sound Pressure Level Error Due to Neglect of Mean Flow, Pr = Ft.Pe Radiated Sound Pressure
Level Calculated Neglecting Mean Flow is Low by the Amount Given Here as Error
341
-------
CD --
3 FT
RADIUS
u.
O
Q.
Q
HI
1-
5
cc
ro
Z
0
1-
u
ui
in
J
a.
in
-20
-40
-60
-80
-100
I I
1
\l
1
3 4
WAVELENGTH (FT)
10
Figure 6. Influence of Mean Gas Flow on Effectiveness of Silencer. o.Measured values:
, Calculated, M = 0-0.
-Calculated, M = 0-15;
'342
-------
1.000
0.000
"00. 1200. 1800. 2400. 3000. 3600.
FREQUENCY (Hz)
180.0 n
g 90.0
LU
_l
u
I o.o -I
I
^ -90.0
-180.0
0 600. 1200. 1800. 2400. 3000. 3600.
FREQUENCY (Hz)
Figure 7. Power Reflection Coefficient and Phase Angle for Open End Tube. Solid Line: Theory; Open Square,
Open Triangle: experiment.
30.00
W20.00
8
z
10.00^
0.00
-10.00
•%
*
0 600. 1200. 1800. 2400. 3000. 3600.
FREQUENCY (Hz)
Figure 8. Measured Values of Transmission Loss for Prototype Automotive Muffler. Filled Circle: SWR Method;
Operi Square, Plus Sign: Two-Microphone Random-Excitation Method
343
-------
Figure 9. Muffler Element
1111\
p
p"
P2,v2
0 12 24
MUFFLER SCALE, IN
1
m.4
THEORETICAL
• MEASURED
MUFFLER
5
0 12 24
SCALE, IN.
_TL
T__r
m.16
m.16
m.16
m.16
200 400 600
FREQUENCY ,f,cps
(al EFFECT OF EXPANSION RATIO m.
(b)
50
40
30
20
10
0
50
40
30
m 20
z 10
50
40
30
20
10
0
50
40
30
20
10
THEORETICAL
• MEASURED
0 200 400 600
FREQUENCY.f.eps
(b) EFFECT OF LENGTH
Figure 10. Comparison of Theoretical and Experimental Attenuation Characteristics Single-Expansion-Chamber
Mufflers; M is Tube Area Expansion Ratio
344
-------
Figure 11. Four Pole Representation of Muffler Element
Al Bl
cl Dl
'2
\
A2 B2
C2 °2
1 3
^
A3 B3
C3 °3
*k
\
Figure 12. Series Connection of Transmission Matrices
Engine
Figure 13a. Real Engine-Muffler-Exhaust
System
Exhaust Muffler |T n pipe
-J—
Figure 13b. Volume Velocity Analog
Figure 13c. Pressure Source Analog
^v_
n ^
-7C_,
Manifold
<£
IH
\
Ze
I
Z
e
l /- - - r
\ I-"- V
n V
\
\
VI
i
) /
A B
C D
A B
C D
V >
; ^2
i
_^ i
V2 Z
r
i
l
I
v— »• 1
2 'r
1
i
1 ^^ n fc_ — -*
Source Muffler Trans- Termination
mission Matrix
345
-------
on (tow
(o)
2
5
<
o:
=> 1C
LiJ
_l
z
HI
1/5
e© Z^
1
346
-------
Figure 16. Simple Expansion Chamber Showing Division Into 16 Finite Elements and 28 Node Points
347
-------
60 1
CO £0.
-o
40. -
O 30. i
a:
O.
-/o.
4- PLANE WAVE ACOUSTIC FILTER
8 ELEMENTS
A 16
0
0, 30O. 600. 900. /£00. /500. /BOO.
FREQUElsiCY (Hz)
Figure 17. Transmission Loss of Simple Expansion Chamber, fi = 8.0 in., m = 5 in.
348
-------
Pass Tube
End -
Figure 18. Flow-Reversing Chamber With Pass Tube and
End Plate. C — Distance Between Centers of
Inlet and Outlet Tubes; H — Height of
Chamber; L - Length of Chamber; W -Width
of Chamber; and d — pipe Diameter
Figure 19. Experimental System for Measuring the
Transmission Loss. A — Frequency Counter;
B —Amplifier; C — Frequency Oscillator;
D — Oscilloscope; E — Level Recorder;
F — Spectrometer; G — Plenum Chamber;
H — Acoustic Driver; I — Microphone Probe;
J — Standing Wave Tube; K — Flow-Re_versing
Chamber; L — Anechoic Termination; and
M — Microphone Port.
/SIDE BRANCH
111
CD
X3
a :o.
I
" ID
-10
300 GOO iOO 1200 1500 1100
f requency, Hz
40
. 3°
x>
•&
3 20
2 10
E
-10
0 300 600 900 1200 1500 1800
Frequency, Hz
Figure 20. Transmission Loss for SI-SO Flow-Reversing
Chamber (L = 2.0 in., H = 9.0 in. W = 4.75 in.
Open- Square—Predicted by Theory for No
Flow Condition. Open Triangle—Measured
Without Flow. Circle- Measured With Flow at
110 ft/sec.
Figure 21. Transmission Loss for SI-CO Flow-Reversing
Chamber (L = 2.0 in., H = 9.0 in., W = 4.75 in.).
Open Square—Predicted by Theory.
Plus-Measured Without Flow; Open Triangle-
Measured Without Flow (End Plate Vibration
Eliminated); Circle-Measured With Flow at
110 ft/sec.
349
-------
50
40
30
.2 20
Figure 22. Transmission loss characteristics for SI-SO ;
muffler chambers:- flow-reversing \
chamber only- predicted;—flow-
reversing chamber and resonator- predicted;
and + is the measured data for the flow-
reversing chamber and resonator
10
-10
300 600 900
Frequency, Hz
1200
1500
1800
Figure 23. Transmission loss characteristics for SI-CO
muffler chambers: flow-reversing
chamber only- predicted; flow-
reversing chamber and resonator — predicted;
and + is the measured data for the flow-
reversing chamber and resonator
50
40
30 -
20 -
£ 10 -
0 '
-10
600 900 1200
Frequency, Hz
1500 1800
350
-------
50
40
30
20
10
-10
300 600 900
Frequency, Hz
1200
1500 1800
Figure 24. Predicted Transmission Losses for Combination of SI-CO Flow-Reversing Chamber and Helmholtz
Resonator, With Different Throat Locations: Side-Located Throat and Centrally
Located Throat
351
-------
|~0.406m-1
Figure 25. Transmission loss characteris-
tics for combination of SI-CO and CI-SO
flow reversing chambers; is predicted;
o is measured without flow and • is mea-
sured with a flow speed of 32 m/s
Figure 26. Transmission loss characteris-
tics for combination of SI-CO and CI-SO
flow reversing chambers; is predicted;
o is measured without flow and • is mea-
sured with a flow speed of 32 m/s
Figure 27. Transmission loss characteris-
tics for combination of CI-SO and SI-SO
flow reversing chambers with two reson-
ators; is predicted; o measured with-
out flow; and • is measured with a flow
speed of 32 m/s
200 400 60O 8OO KKX> I2OO WOO I6OO I8OO
FREQUENCY, Hz
f-0.406m-|
(fi 30
O
200 400 6OO 8OO 1000 I2OO 1400 I6OO BOO
FREQUENCY, Hz
0.406m-I
u> 30
O
200 400 eoo aoo looo 1200 1400 ieoo isoo
FREQUENCY, Hz
352
-------
30.00
^
00
2.
20.00
CO
CO
3
z
o
g 1000
2
to
z
cr
000
CAVITY' LGTH 66.7 mm
CAVITY OD 76.2 mm
CAVITY ID 50.8 mm
- POROSITY 3.7 %
o o
^0
O \
f \ °
/ Xon |
/ \° 0
y^° Voo7
}
o
Xo \O \
o^0'0 ° V
_o5^cf°'0' \0__
1 I 1 I 1 1 1 1 1 1 1 1
6OO
I20O I8OO 24OO
FREQUENCY (Hz)
3000
3600
Figure 28. Transmission Loss for a Short Resonator, (Predicted , Measured o)
353
-------
30.00
£?20.00
CO
CO
Z 10.00
o
CO
o.oo
CAVITY LGTH 257.2 mm
CAVITY 00 76.2 mm
CAVITY ID 5O.8 mm
POROSITY 3.8 %
600
1200 1800 2400
FREQUENCY (H2)
3000
3600
Figure 29. Transmission Loss for a Long.Resonator, (Predicted —; Measured o)
354
-------
2 20! U = 0-72 TT
CAVITY LGTH 66.7 mm
CAVITY OD 76.2 mm
CAVITY ID 50.8 mm
500 1000 1600 £000 3600 3000 3500
H 1 » 1 1 1 1 1 1 1 1 1 1
I I 1 1 1 1 1 1 1 1
Porosity
5.0 %
4.5 %
4 %
i—•—i—i—i 1—i—i—»—i—i
k0l » 0.46TT
2 %
H 1 1 1 1 1 »-
H 1 1 »-
K0I = 0.2371
ID--
H 1 1 1 * t I »-
0.5 %
600 1000 1500 2000 2600 3000 3600
FREQUENCY (Hz)
Figure 30. Effect of Porosity on Transmission Loss for a Short Resonator, Predicted From Mathematical Model
355
-------
8
o"
CD
O
O
D
r-
o
o
o
in
o
^—t
I—
CC
a:
o
o
o
o
o"
CO
I
o
o
O I.L.with zero source impedance
A I.L.with anechoic source termination
-j- I.L. with infinite source impedance
y^ Transmission loss
( without filter )
o
in
220. 4SO. 660. 880.
FREQUENCY (Hz )
1100.
1320.
Figure 31. Theoretical Insertion Losses and Transmission Loss for Engine Exhaust Muffler System with Actual
Exhaust Temperature Profile at Firing Frequency 100 Hz
356
-------
o
o
o
co
o
in
o
o
TJC3
(X
o
O
O
a
o
a
O
/A
-{-
I.L. with zero source impedance
I.L. with anechoic source termination
I.L. with infinite source impedance
Transmission loss
( with 25 H filter )
Q.
220.
440. G60. 830.
FREQUENCY (Hz)
1100.
1320.
Figure 32. Theoretical Insertion Losses and Transmission Loss for Engine Exhaust Muffler System at Elevated
Temperature at Firing Frequency 70 Hz
357
-------
SHOCK-TUBE METHODS FOR SIMULATING EXHAUST PRESSURE
PULSES OF SMALL HIGH-PERFORMANCE ENGINES
B. Sturtevant and J. E. Craig
California Institute of Technology
Pasadena, California
ABSTRACT
The unique aspects of steep-fronted, large-amplitude pressure
pulses that occur in the exhaust systems of small high-performance
internal-combustion engines are reviewed. Some special analytical
and experimental techniques that are useful for testing, simulating and
analyzing such exhaust systems are described. Two examples are given
of wave-diffraction effects which are particularly important when the
incident waves are steep-fronted and which significantly affect the per-
formance of simple muffler elements in these circumstances. The
radiated noise due to these diffracted waves after their passage through
the exhaust system can be strongly affected by gas dynamic nonlinearity,
It is concluded that any procedure for qualifying mufflers of high-
performance engines must accurately simulate the unique features of
the exhaust dynamics of these systems.
359
-------
1. Introduction
In this paper we review the unique aspects of the exhaust
dynamic's of small, high-performance internal-combustion engines
and the special techniques that should be used in testing, simulating
and analyzing their exhaust systems. In this regard, the most important
feature of small engines operating at high rpm is the fact that the pulses
generated by the opening of the exhaust valve or port tend to be steep-
fronted and of large amplitude. Risetimes of pressures measured near
the exhaust port of both 2- and 4-stroke engines commonly range from
0. 1 to 1 msec (Refs. 1-4), so the thickness of the first pulse as it exits
the exhaust port is in the range 2-20 cm. Furthermore, a large-
amplitude pulse tends to get thinner as it propagates, by nonlinear
steepening. A pulse with amplitude 0.5 bar will steepen to a discontinuity
after propagating a distance only 3 times its initial thickness. Therefore,
for example, a pulse with an initial risetime of 3/4 msec will steepen
to a discontinuity after propagating 0, 8 m.
When steep-fronted pulses occur in an acoustics problem it is
more natural to treat the_ problem in the context of the theory of geometrical
acoustics (Ref. 5), than by spectral decomposition and harmonic analysis.
In geometrical acoustics the analysis is carried out in the time domain,
so the physical processes are more transparent and the results more
intuitively obvious. The theory of geometrical acoustics has been
extensively developed, including the treatment of diffraction effects (Ref. 6).
Application of nonlinear boundary conditions is straightforward. Furthermore
pulse theory can be directly extended to account for effects of gasdynamic
nonlinearity (Ref. 7), while consideration of nonlinear effects in the
frequency domain is cumbersome and unproductive.
Therefore, when the thickness of the compressive portions of
the pressure pulses in the exhaust systems of small high-performance
engines is of the order of or smaller than typical transverse dimensions
(i.e. , the largest diameter), it is useful for determining acoustic
performance to trace the propagation of the pulses through the system
and to study their interactions. This is especially true if one is interested
in the emitted noise because noise in the far field is generated by the
360
-------
rate of change of volume flux at the source. Therefore, most of the noise
originates at the steep fronts of the waves. At Caltech we have conducted
some experiments in shock-tube facilities*, in which the pulses incident
on exhaust systems are discontinuous fronts (weak shock waves). This
simplification has permitted the observation of two previously unexpected
diffraction effects which may be important sources of noise (self noise)
in applications with steep-fronted pulses. The spiked waveforms typical
of diffracted waves are sensitive to the effects of gasdynamic nonlinearity,
so propagation in straight sections of pipe (e.g. , the tailpipe) can have
important effects on the emitted noise.
It is concluded that any procedure for testing mufflers for
small high-performance engines must include provision for measuring
the effects of fast pulse risetimes and finite amplitudes. Though the
apparatus used at Caltech has not been developed for use in a standardized
procedure, it is possible that shock-tube facilities can be used to simulate
these features of exhaust pulses of high-performance engines. Of course,
shock tubes do not duplicate all the characteristics of engine noise sources,
so they should be used only to supplement the information obtained in
other, perhaps more conventional, tests.
In this paper we first describe the test apparatus and then cite,
as proof that finite-amplitude effects must be accounted for, two examples
of two-dimensional diffraction effects which are influenced by gasdynamic
nonlinearity.
2. Experimental Apparatus
In systems with large - amplitude unsteady motion, the max-
imum instantaneous flow velocity may be substantially larger than
the mean velocity. Therefore, there may be substantial inflow from
the atmosphere into the exhaust system during certain portions of
the cycle. Because of viscous effects and separation, flow out of
an area expansion (jet flow), is fundamentally different from flow into a con-
verging section of tube (sink flow), so the occurence of flow reversal
Jf
'i-
Complete details of the experimental apparatus, the research program
and some findings of the fundamental behavior of finite-amplitude waves
in exhaust systems may be found in Ref. 1.
361
-------
during a portion of the cycle can be an important source of departure
from ideal acoustic behavior. For example, our work has shown that
the performance of perforated tubes in mufflers can be greatly affected
by the existence of inflow into the muffler from the atmosphere before
arrival of the main pulse. In the present experiments we use two different
facilities, a periodic source and a single-shot source, to bracket the
effects of inflow. The two devices are represented schematically in
Figure 1.
Resonance Tube The resonance tube (Figure 2) is a long
gas-filled tube which is excited at one end by a reciprocating piston and
terminated at the other end with the exhaust system to be studied. The
piston is driven at the fundamental acoustic resonance frequency of the
tube, and its amplitude is large enough that at resonance the compressive
portions of the waveform steepen to form a shock wave travelling back
and forth in the tube. Thus, the resonance tube is used as a wave
generator to supply large-amplitude steep-fronted periodic waves for
exciting the exhaust system. A comparison between the resonance-tube
waveform and a typical pressure history measured at the exhaust port
of a 250 cc single-cylinder two-stroke engine, when both sources are
connected to a high-performance expansion chamber exhaust system, is
given in Figure 3.
Provisipn is made for measuring internal pressures at several
locations in the exhaust system and for measuring free-field radiated
noise. Data are acquired by a computer-controlled data acquisition
system, and all data are processed in real time and the results are
output in plotted format shortly after completion of a run. The data
acquisition is synchronized with the piston crank mechanism through
a 256-tooth gear mounted on the crank shaft and a magnetic pickup.
This has the important consequence that spectra calculated by a
Fast Fourier Transform (FFT) algorithm are actually exact Fourier
analyses of the periodic signal, and it is not necessary to apply window
functions, etc. , to the sampled data to insure adequate accuracy of the
results.
362
-------
Shock Tube. The shock tube (Figure 4) is a conventional
pressure-driven shock tube to which is attached the exhaust system to
be studied. In order to maximize the uniformity of the input shock wave
a "cookie-cutter" configuration, in which the exhaust pipe is extended
inside the shock tube, is used. Provision is made for measuring internal
pressures and free-field radiated noise. The same data-acquisition
system as was used with the periodic system described above is also
used with the single-shot shock tube. Further details of the experimental
technique are given in Ref. 1.
Only very simple muffler configurations have been studied
in this work, for the purpose of examining the fundamentals of wave-
propagation in exhaust systems. However, the results are sufficient
to demonstrate the utility of the experimental method. The repeatability
of the results and the accuracy of the measurements are such that many
effects related to noise suppression are easily visible on the pressure
traces. Therefore, the method is also useful for diagnostic analysis
and for muffler-design optimization.
3. Perforated Tubes in High-Performance Mufflers
Experiments have been carried out to determine the mechanism
by which perforated tubes in mufflers attenuate acoustic pulses.
Figure 5 shows the simple straight-through configurations tested
(enclosures A, B and C are defined in Figure 9) and identifies the
notation for the transducer locations U, Dl and D2 used in subsequent
figures. The perforations are 6.35 mm dia drilled holes and are
arranged so that the open area per unit wall area is approximately 1/6.
The total area A of the perforations in a given test is set by the number
lij
of holes in the tube and is characterized by the ratio A /A, where A is
Hj
the tube cross-sectional area.
Oscilloscope traces of internal pressures measured at three
different locations in a single-pulse excited system, with three different
values of A for an "infinite" enclosure (perforations open to the room)
hj
are shown in Figure 6. They generally confirm results obtained in
previous studies of perforated tubes (Refs. 8 and 9). The upstream
traces show the incident shock followed by an expansion wave reflected
363
-------
from the perforations. The downstream traces show the detailed
structure of the transmitted wave. The final steady-state pressure
behind the transmitted wave is well accounted for by a simple analytical
model of the sink effect of the flow through the perturbations (with a
reduced orifice discharge coefficient due to axial momentum in the jets),
but the spike and pressure minimum observed especially for large A /A
are unpredicted 2 - dimensional effects and are obviously important
with regard to noise emission. When the perforation area is large,
evidently the shock is not immediately attenuated to its theoretical
value. Particularly in the case A /A = 0. 89 in the figure, the effect
of propagating between Dl and D2 in the tailpipe is evident; the shock
discontinuity and the very rapid expansion wave, 'which is probably made
up of (2 dimensional) diffracted waves from the numerous orifices,
interact, resulting in an attenuation (and slow disappearance) of the pressure
spike. This attenuation is due entirely to gasdynamic nonlinearity; if
there were no nonlinear effects the spike would be much larger, a fact
which is born out by the fact that it shows up much more strongly for
the weaker waves in our experiments (Figure 6) than for stronger
waves, where nonlinear effects are larger. The fact that important
attenuation can occur during propagation down the straight tailpipe
emphasizes the importance of testing complete muffler systems in
obtaining noise suppression data for high-performance engines.
Figure 7 summarizes the overall effect of perforations on
radiated noise. Though a small spike persists at D2, the main effect
has been to slow the rise of the compression in the pipe to a very much
larger value than that of the input discontinuity- vastly reducing-the
far-field (location F) noise level (a shock of the same amplitude would
yield about 1 mBar amplitude, vs. the 0. 12 observed). However, the small
surviving pressure spike remains the major noise source!
Figure 8 shows the effect of finite enclosures surrbunding
the perforations. The effects of waves excited by the passage of the
incident .shock reflecting back and forth in the enclosures are evident,
particularly in the radiated noise, where secondary- spikes now
occur. With the experimental technique used in this work it is even
possible to see that the odd - numbered secondary peaks at Dl are
364
-------
smoother than the even-numbered, due to the nature of wave .propagation
in the muffler, with the consequence that the corresponding spikes in
the far field are much -weaker!
4. Expansion Chambers
Figure 9 shows the simple expansion chamber configurations
tested in the present work. It is well known that when the acoustics
of expansion chambers is considered from the pulse point of view one
can trace the waves as they reflect back and forth in the expansion chamber
interacting with the discontinuous area, changes, as shown schematically
in Figure 10. Indeed, each and all of the infinite number of infinite
series of waves can be summed in closed form to give the overall
transmitted wave field, but this always gives too large a value for the
radiated noise because viscous dissipation during the wave interactions
has been neglected. However, within the pulse point of view it is a very
direct and effective artifice to. simply truncate the series at some finite
number of terms to provide a first-order correction for the effects of
dissipation. In any case, if the spectrum of the transmitted waveform
is calculated it is seen that the multiple reflections of the discrete
fronts have the same effect as the familiar superposition of incident
and reflected waves in a spectrum of harmonic excitations, both points
of view showing the effects of destructive interference.
The geometrical point of view shows immediately that the
manner in which an expansion chamber serves to-attenuate an acoustic
pulse is to break up the single incident pulse into a series of weaker
waves. In a sense, the transmitted wave is stretched out into a more
gradual compression, so the net effect is the same as with the perforated
tube discussed above. Indeed, after a comparative study of both devices,
one would conclude that the optimum combination of elements in systems
where wave amplitudes are large would be a series arrangement with
the expansion chamber first, followed by the perforated tube (cf. Ref. 1).
However, one phenomenon that one-dimensional theory can not
predict is the diffraction of wave fronts at discontinuous area changes.
Figure 11 depicts schematically the geometry of the actual wave fronts
365
-------
generated when a wave diffracts from the end of an extended inlet and,
in the bottom sketch, the representation of the process by one-dimensional
theory. To the extent that the multitude of diffracted fronts persist
as they propagate in straight sections of tube, the noise emitted by the
system may be seriously underestimated by one-dimensional considerations.
Figures 12 and 13 show two examples of interior and free-
field wave forms observed in experiments with two different expansion
chambers. The multiple reflections of the incident front in the ex-
pansion chamber are evident in the reflected and tiansmitted waves,
but superimposed on these waves are very high frequency fluctuations
due to diffracted waves. In this case, contrary to the behavior in
perforated tubes^ gasdynamic nonlinearity aggravates the situation,
because, as is well known, the wavelength of a nonlinear sawtooth
wavetrain tends to saturate at a constant value, while linear diffracted
waves tend to "merge" simply by geometrical spreading from their point
of origin. In Figure 13 the diffracted waves at location D3 have formed
a sawtooth waveform containing shocks and have the same spacing as
at Dl, indicating nonlinear saturation. Their large contribution to the
radiated noise at location-F is obvious. At D3 the amplitude of several
of the diffracted waves is more than 10$ of the amplitude of the single
incident shock. The relative strength of the diffracted waves increases
as the expansion chamber diameter increases, so in fact the noise
attenuation of an expansion chamber peaks out at a particular area ratio
and fails to increase beyond that value.
5. Conclusions
It has been shown that some unique features of the steep-
fronted large-amplitude pressure pulses in the exhaust systems of high-
performance internal-combustion engines require accurate simulation in
procedures for testing and qualifying mufflers. An experimental technique
which simulates the actual pulses with discontinuous press.ure rises,
(weak shocks) is described. The technique has the advantage that is also
useful to the designer for diagnostics and design modification. Two
examples have been given of two-dimensional phenomena that are not
accounted for in one-dimensional analyses but which are particularly
important when the pulses are steep-fronted.
366
-------
References
1. J.'E. Craig, "Weak Shocks in Open-Ended Ducts with Complex
Geometry", Ph.D. Thesis, California Institute of Technology,
Pasadena, CA. (1977), Figure 7.
2. G. P. Blair and J.A. Spechko, "Sound Pressure Levels Generated
by Internal Combustion Engine Exhaust Systems", SAE Trans. 81,
563 (1972), Figure 4.
3. W.A. Huelsse, "Investigation and Tuning of the Exhaust System
of Small Two-Stroke Cycle Engines", SAE Trans. 77, 563
(1968), Figure 17.
4. M. Leiber, "The Exhaust System of the Two-Stroke Cycle
Engine", SAE Trans. _77, 1846 (1968), Figures 22, 24.
5. J.B. Keller, "Geometrical Acoustics. I. The Theory of Weak
Shockwaves", Jour. App. Phys . 25, 938(1954).
6. F.G. Friedlander, Sound Pulses, Cambridge University Press
(1958).
7. G.B. Whitham, Linear and Nonlinear Waves, John Wiley and
Sons (1974), Ch.~
8. J.H.T. Wu and P. P. Ostrowski, "Shock Attenuation in a Perforated
Duct", in Shock Tube Research (ed. J.L. Stollery, A. G. Gaydon
and P.R. Owen), Chapman and Hall, London (1971).
9. A. P. Szumowski, "Motion of a Shock Wave Along a Perforated
Duct", Prace Nauk. Mech. , Politech. Warszawska, Nr. 18 (1972).
367
-------
MUFFLER
•TEST SECTION
GALCIT
SHOCK TUBE
RESONANCE
TUBE
DIAPHRAGM-
FIGURE I THE GEOMETRY OF EXPERIMENTAL
FACILITIES
368
-------
50 cm
670cm
Ground Plane
210cm
Resonance Tube
76 rnm *-
J.A. Prestwick
100mm- Stroke
80mm -Bore —
Gear
ADC Controller
Phase Locked Loop
Beta
Clock
TDC
i—Magnetic Pickup
,
15 H.P
D.C. Motor
( Location F )
H. P. 2100
A-D
Converter
CPU
Magnetic
Tape
Disc
FIGURE 2
-------
OJ
•^i
o
(a)
front
(arbi-trary scale)
17.8
(b)
Vertical scale 138 —
cm
Horizontal scale 1
2.5
mS
cm
t
IOG
0 . 33.0, 63.5"
FIGURE 3 COMPARISON OF RESONANCE TUBE, A, AND MOTORCYCLE
ENGINE, B, PRESSURE WAVE FORMS
- C/rjj
-------
OJ
(Location F )
Wall
Diaphragm
Driver
Section
ll
Origin
Main
Section
Cookie
Cutter
Test
Section
3.SI
C/r>
II-Om-
3.2-l/n-
FIGURE 4 GALCIT SIX INCH SHOCK TUBE
-------
LO
•^1
N3
Resonance Tube
74.6'
Shock Tube
1370
-o-
4.6
-76.2
UI U
•Enclosure A,B, Or C
30.5
Transducer
Locations —
39.7
Dl
EXIT
25.5
-•—42.5-
D2
All Dimensions Cm.
FIGURE 5 PERFORATION SYSTEMS
-------
AF/A 0.44
0.89
11 •
.78
THE EFFECT OF PERFORATED AREA RATIO
ON REFLECTED AND TRANSMITTED WAVES
C MACH NO. = 1.13) -
FIGURE 6
-------
OJ
^J
-P-
o
o
8
o
CD
E
Location U
Tms
12.0
Location F
O
6
ro
o
co
E
Q_
Q
o
o
C)
o
Tms
10.0
E
CL
q
d
?0
Location D
Tms
12.0
Location D2
Tms
12.0
FIGURE 7
PRESSURE HISTORIES OF SHOCK PROPAGATION PAST A PERFORATED
TUBE AE/A = 4.00, SHOCK TUBE
-------
Enclosure A
B
c
o
o
o
o
_J
AE/A = 1.78 Scales:Top And Middle Rows,lOOmBar-I m Sec
Bottom Row, 10p-Bor-1 m Sec
FIGURE 8 PRESSURE HISTORIES OF SHOCK PROPAGATION PAST PERFORATED TUBES,
RESONANCE TUBE
-------
Resonance Tube
7.6
3.8
Shock Tube
15.2
Removable Extensions
Transducer
Locations
FIGURE 9 EXPANSION CHAMBER SYSTEMS
EXIT
• J e
f 4.6
—rs* f*
74.6
)
^ J -
-" 30.5 ^
— v.
^4.6
68.0 H
In Cm.
Chamber
A
B
C
D2
6.35
8.25
11.4
A2/A|
2.77
4.69
9.00
-------
FIGURE 10 EXTENDED INLET SYSTEM
377
-------
FIG. II SHOCK INTERACTION WITH AN EXTENDED INLET
378
-------
o
C)
o
m
o
op
E
Q.
iT
''
0
Location Ul
Tms
CD
E
a_
ro
Location Dl ,
_^_J -^^JL J
12.0 £0
Or
Location F c\J j Location D3
ro
o
CD
£
Q_
q
o
Tms
Tms
10.0 0
Tms
12.0
T,
12.0
FIGURE 12 PRESSURE HISTORIES OF SHOCK PROPAGATION THROUGH EXPANSION
CHAMBER,B. Ms = 1.17, SHOCK TUBE
-------
00
o
12.0
Tms
9.0
12.0
Tms
12.0
FIGURE 13 PRESSURE HISTORIES OF SHOCK PROPAGATION THROUGH EXPANSION
CHAMBER, C. Ms = I. 07, SHOCK TUBE
-------
NOISE SYMPOSIUM IN CHICAGO - OCTOBER 11-13, 1977
CORRELATION OR NOT BETWEEN BENCH TESTS AND OUTSIDE MEASUREMENTS
FOR SNOWMOBILES.
As you probably know, our company, BOMBARDIER LIMITED, is
involved in recreational vehicles and more particularly
in SKI-DOO snowmobiles.
With snowmobiles we are faced to three certification standards:
See sli de no. 1
SSCC-55 which is a 15 MPH pass-by test;
SAE J-192a which is a full acceleration test;
ISO R-362 which is the European procedure.
During this symposium, up long, we have heard a lot in
theoritical predictions versus practical measurements on
bench tests. In this presentation I do want to go away from
this interesting aspect for having a good exhaust labelling.
I will try to compare practical bench test measurements to
actual measurements on the snowmobile itself.
WHY?
Becau.se I am interested in the consumer point of view.
Fora "future buyer of any transportation vehicle, it is
important to give him the truth.
381
-------
So we try to take the problem by the end.. Let us suppose
we have the right method to obtain practical measurements
on bench test and let us try to see what is going to happen
on the actual field test.
And, from now we are going to notice all the parameters
which are involved in the sound of the exhaust. And, I am
sure, that any of you can.find even more than what we are
g.oing to speak of.
In order to eliminate partially the discussion of the
influence of the other sources (air intake, track etc...)
we use a vehicle in which muffler noise>was supposed to be
the greater source at least by 3 dB at fifty feet. You will
ask why not more than 10 dB? Because this is never an actual
situation and we were interested in seeing how changing
muffler is combining_ in the spectrum with the other components,
At this point, concerning a possible method to measure exhaust
noise at bench, please refer to next speaker, Jim Moore who
is going to show you how bench test and outside measurement
correlate in some particular conditions.
382
-------
I OUTSIDE EFFECTS
First of all we have physical- parameters which are generally:
i) WIND, which should not be more than 12 MPH.
But from "0" to 12 MPH you can easily imagine the consequences
on performance (with free air engine),-temperature of
exhaust and angle of incidence which can help you a lot
or not. Differences: up to 1.8 dB(A)
ii) AIR PRESSURE, we know that it affects sound transmissibility
and performance. Not a lot for sure but enough to be
considered. Differences: up to .8 dB(A)
iii) AIR TEMPERATURE, this of course is quite an important factor
o o
especially on snowmobiles which will run in a -40 C to 0 C
range, and it is not easy to mix cold chamber and a
semi-anechoic chamber!
And of course, temperature will affect the muffler itself
but also the spectrum and the total value of each other
sources. So it is quite a job to' separate those effects
and to obtain a significant comparison or typical values
between different mufflers.
383
-------
I OUTSIDE EFFECTS
iii) cont'd
Remember that a two-stroke engine with free air or
fan cooled version, it is much more affected by the
exhaust temperature than any liquid cooled engine.
Differences: up to 2.0 dB(A).
iv) RELATIVE HUMIDITY, every one of us know that it could
affect performance quite a lot. It affects also sound
reflexion and transmissibility. So are we going to take
care of the humidity? You can control it on bench test.
Yes, but for certifying a muffler, are you going to make
this humidity vary from step to step to see where is the
maximum? Certainly not. For development purposes, yes,
but not for obtaining a rating level of the exhaust
noi se .
Differences: up to .8 dB(A).
384
-------
II GROUND EFFECTS
Now speak of the most important point: ground effects.
This is quite particular to snowmobiles.
See slide no. 2 .
In the procedure they tell us that you can use:
firstly: packed snow with not more than 3 inches of ordinary
snow.
secondly: dry grass, 3 inches.
The problems are:
What is exactly packed snow? It could be ice, it could be
just packed by passin'g on with a snowmobile.
What sort of grass and underground? We could get more than
1.5 dB(A) difference with the same grass type but with soft
or hard ground underneath.
And also we have to speak of the "fact that some models are
unaffected when compared between grass and snow. Others
could get differences up to 2.5, even 3 dB(A).
We know that snow is much better than grass and of course
asphalt, .to absorb low frequencies.
See spectrum no.. 1.
385
-------
_l _ ._.
Potentiometer Range'
I ( (
mm/sec. Paper Speed: mm/sec
OP 1134
Multiply Frequency Scale by:
500- 1000 2000
Zero Level:
(1612/2112)
ABC Lin.
-------
Let us go now with practical experience in the snow:
3" snow
muffler output
As we can see, distance from ground, reflexion incidence
regarding the exhaust are not always the same. So?
And, remember in the snowmobiles trails it is much more often
like that:
rather than in a straight line.
And a snowmobile is normally running on snow, so according
to me you have to watch this situation very carefully.
387
-------
Now speak of orientation of the output.
If you look at all sorts of mufflers on the market, you can
have an output like:
frame
ground
footrest
frame
r ame
til
'footrest
''S •/' ' // ; ' s s 's.,--'
It is anothe'r factor that you have to consider.
For this we have made isosonic curves by having maximum HP/RPM
on a static vehicle.
See slide no. 2
388
-------
For finishing: sound direction related to spe e d.
I t>r
-------
So, facing all these factors, we have tried to find an
empirical formula which could be used of the major puts
of what we have explained. We were interested in predicting
the influence of any exhaust if set-up on any kind of
vehicle in any kind of conditions.
For doing this we put on a vehicle sensors in order to
get temperature of exhaust (near the end of the muffler),
temperature and pressure at the spark plug, temperature
of the air intake, RPM (measured at the drive pulley),
real vehicle speed (measured at the driven pulley with
appropriate correction for gearing), and of course we
measured "external temperature, humidity, pressure, wind
and di recti on.
We also put coefficients for sort of packed snow, for
thickness of packed snow, for sort of above snow, for
thickness of above snow, for dry grass, for wet grass,
for hard ground, for soft ground, for asphalt,and also
using isosonic curves for orientation effect.
Mixed track: asphalt and grass (or asphalt and snow).
See slide no. 4.
390
-------
10/.. .
A statistical analysis has been done in order to find the
influence of each parameters. We want to have something
absolutely general with no particular test site conditions
or particular muffler with a particular engine. This is
going on right now. The first tries are not very good
(± 5 dB(A)). We have to make some changes in factors to be
considered and in the program itself.
Conclusion
This 'statistical approach has the advantage of being not
very complicated and not very heavy in terms of dollars.
It has the quality of being very near the field result
that is to say, very near from what consumer people will
really obtain. The. results that we have obtained seem to
confirm that it is quite difficult to p-redict field result
with good correlation fdr snowmobiles.
391
-------
ll/...
INSTRUMENTATION USED:
Sound level meter BRUEL & KJAER #2204
FM recorder BRUFL & KJAER #7003
Low pass filter HP #5489a
Power supply HP #73a
WESTON voltmeter #4442
Electronic conditioner HP #5216a
Spectrum display HP #3720a
Correlator HP #3721a
Digital recorder HP #5055a
Statistical description analyser BRUEL & KJAER #4420
Plotter X, Y HP #44a
392
-------
Figure 1
-------
Fiaure 2
-------
U).
^o
Ln
Figure
-------
Figure 4
-------
JOHN DEERE HORICON WORKS
220 EAST uAKt HORiCON WISCONSIN 5J032
31 October 1977
JAMES W. MOORE
MEASUREMENT OF ENGINE EXHAUST
NOISE IN DYNAMOMETER ROOMS
A method of measuring engine exhaust noise has been developed
as a substitute for the more complicated anechoic room or field
tests. It is simple and easy to use and does not require
expensive test facilities and equipment or modifications to
the exhaust system. The sound readings and insertion loss can
be determined simultaneously with dynamometer power measurements.
The results have shown good repeatability'and are not subject to
the variations in weather conditions encountered during field tests
The test procedure was developed by Richard Kostecki of
ACS Engineering in Toronto, Canada and has been used success-
fully by ACS for exhaust system development for several years.
A similar test method is also used by two other snowmobile
manufacturers. John Deere has used it extensively in the
development, comparison, and selection of snowr>cbile and small
four cycle engine exhaust systems.
Figure 1 is a schematic diagram of the test system. The exhaust
gas discharges from the muffler (1) into a 4-foot long, 2-inch
diameter, flexible exhaust pipe (2) which is anchored at the loose
end to a 60-pound steel block(5). The exhaust gasses can be
evacuated from the test cell by the collector (6). The sound
pressure is measured through a hole in the end of the pipe by a
microphone (3) in a special water-cooled, mounting (4) .
The length and diameter of the flexible pipe were selected after
extensive experimentation and are designed to isolate the
microphone from the engine vibration and noise, and to provide
adaptability to various exhaust system geometries. Engine
performance and exhaust noise generation are not affected by the
measurement .system.
The sound level is read on a sound meter (7). Octave band
measurements can also be taken <8). Correction factors are
applied to each octave band to compensate for noniinearities
in the measurement system and for comparisons to field tests.
This correction process is simplified by a spectrum equalizer (9).
397.
-------
JOHN DEERE HORICON WORKS
PAGE 2
A METHOD OF ENGINE EXHAUST NOISE MEASUREMENT IN DYNAMOMETER ROOMS
The upper curve in Figure 2 shows a typical exhaust noise spectrum
of a snowmobile muffler measured on the test fixture. A correction
factor is subtracted from each of the seven octave readings to
extrapolate to the exhaust noise -spectrum in the lower curve that
would result from a snowmobile driveby sound test at 50 feet.
The sum of the corrected octave bands produces the overall
A-weighted level.
Figure 3 shows the spectrum of correction factors that are
subtracted from each octave of exhaust noise measured on the
test fixture. The upper curve is the difference between exhaust
noise measurements made in an anechoic chamber and with the test
fixture. It corrects the noise measured with the fixture to an
A-weighted, "free field" sound level at a distance of 1 foot.
(Narrow band measurements have shown that the frequency linearity
of the measurement system is excellent within each octave band.
A correction in the wider octave bands is all that is necessary
to compensate for the nonlinear effect of the 4-foot long flexible
pipe.) The middle curve converts the 1-foot measurement to 50 feet.
The total correction is shown in the lower -curve.
Figure 4 demonstrates how the 50-foot correction factor was
developed. Octave bands of white, random noise produced by an
acoustic driver were measured over a grass test site at a distance
of 50 feet. The microphone was located 4 feet from the ground
surface, and the sound source was placed at 1/8, 1/2, 1 and 2 feet
above the ground. (The test site confirmed to the requirements
of "SAE J192 , Sound Level Measurement Procedure for Snow Vehicles".)
The variations in sound level with source height are caused by
ground reflections (see SAE Publication 740211, "Effect of Ground
On Near Horizontal Sound Propagation" by Pie.tcy and Embleton) .
The 1/2-foot level, which is about the height of a snowmobile
exhaust, provides the 50-foot correction factors shown in Figure 3.
Tests have shown that this exhaust noise measuring system gives
sound levels within 2 dB Df measurements rrade in an anechoic
chamber. Correlation with the exhaust noise predicted in snow-
mobile passby tests is also excellent. The sound level difference
between similar exhaust systems on the same engine or in the same
vehicle can be compared, within 1 dB. The convenience, repeatability,
•and simplicity of th' s nethod of exhaust noise measurement makes
it very useful in small -ngine muffler development, selection and
rating.
Noise measurements havt not been attempted on exhaust systems
^t.her than those on SHIP I1 , two cycle and four cycle engines.
398
-------
SCHEMATIC DIAGRAM
Acoustic (Mechanical) Components
Electronic Components
FIGURE 1
dB
MEASURED
EXHAUST SOUND LEVEL
150
130
120
110
70
dBA 60
CORRECTED-
(50' DRIVE BY)
\
\
16 32 63 125 250 500 1000 2000 4000 8000
OCTAVE BAND CENTER FREQUENCY-Hz
399
LIN
-------
SOUND LEVEL
CORRECTION
FACTORS dB
-10
-20
-30
-40
-50
-60
-80
For test fixture
(O to 1' distance)
For ground effect
(V to 50'distance)
Total
•
X—-X
16 32 63 125 250 500 1000 2000 4000 8000
OCTAVE BAND CENTER FREQUENCY-Hz
FIGURE 3
—o
—10
-20
GROUND
EFFECT
ATTENUATION-
FROM D = V
TO D=50'
dB
—40
— 50
-60
-70
DISTANCE
RULE
SOURCE
HEIGHT-FT.
16 32 63 125 250 500 1000 2000 4000 8000 16000
OCTAVE BAND CENTER FREQUENCY Hz
FIGURE 4
400
-------
THE APPLICATION OF THE FINITE ELEMENT METHOD
TO STUDYING THE PERFORMANCE OF REACTIVE &
DISSIPATIVE MUFFLERS UITH ZERO MEAN FLOW.
by
A. Craggs
Dept. of Mechanical Engineering
University of Alberta, Edmonton
Alberta, Canada.
INTRODUCTION:
This paper gives a brief review of some of work carried out by
the author on the application of acoustic finite elements to studying muf-
fler performance. It is shown that the method can give plausible results
for a models having a simple geometry because the results compare very favour-
ably with those obtained by other methods. Because the elements used in the
work have a variable shape they can be used to simulate systems which might
have a difficult geometry and still give meaningful information. This is
one of the prime virtues of the method.
In two recent papers (1) and (2) it was shown that for transmis-
sion loss calculations the'muffler has to be treated as one which has damp-
ing even when the muffler is a reactive one. This is because reactive muf-
flers lose energy through radiation at the inlet and exhaust parts. As
such the equations which govern the motion of the system are expressed in
terms of complex quantities. The general form of the equations are the same
for both transmission loss and insertion loss calculations.
As the theory is available elsewhere (1) and (2) it is kept to a
minimum in this presentation. However, the concept of an absorption element
has not been used before and it is introduced here. These elements are par-
ticularly useful when dealing with absorptive boundaries having an extended
reaction. A brief application of these elements is discussed at the end of
the paper.
2.0 GENERAL THEORY
The application of the finite element method results in a set of
linear equations. Because all of the situations are essentially for damped
systems the problem has to be formulated in terms of complex quantities.
However, using the method given in reference (1), the real and imaginary
401
-------
2.
parts can be separated and the- system equations can be expressed entirely in
terms of real quantities. When this is done, the equation for reactive and
dissipative mufflers all have the general form shown below:
[A] - k2[B] - k[Cj] I - k[CRJ
+ k[CRJ j [A] - k2[B] - k[Cz]
Kl
k
=
V
^
(1)
Here PR is the real part of the acoustic pressure; PI is the imaginary part;
QR is the real source vector; Oj the imaginary source vector; [A] and [B]
are the kinetic energy and strain energy matrices respectively. The matrix
[C] is a dissipation matrix which only has non-zero elements at points cor-
responding to the boundary nodes where the energy is lost either through
absorption as with mufflers having a dissipative lining or through radiation
at the input and output parts as in a reactive muffler. In the general
problem the matrix [C] has the real and imaginary components [CR] and [Cj].
Thus if we have a given sound source {Q} then the acoustic pres-
sure at any point within the system may be found through matrix inversion,
using standard computer subroutines.
2.1 TRANSMISSION LOSS CALCULATIONS:
The transmission loss refers to the performance of a muffler
when it is inserted into an infinite transmission line. See Figure 1.
The source is due to an incident progressive wave, of magnitude p'1", which
strikes the entrance of the muffler. The response then contains the reflect-
ed wave, P- and the transmitted wave pj, and pressures at numerous points
inside. The transmission loss is calculated from the formula, (see references
(1) and (2) :
M
T.L = 20 log
Because of the infinite line there are no reflected waves either at the input
of the output stations, and the impedance at these stations is accordingly
entirely real; being equal to pc, where p is the mass density of air and c
is the speed of sound.
Transmission loss calculations are usually the first step carried
out in the design of a muffler. However, because of the highly idealised
situation which is applied some caution is needed when interpreting the
402
-------
3.
the results for a practical situation where reflected waves are present both
on the input and output lines. A much more meaningful calculation is for the
Insertion Loss.
2.2 INSERTION LOSS CALCULATIONS
The insertion loss refers to the difference in the sound intensity
levels at a point before and after the insertion of the muffler. In general,
.then, two sets of calculations are required; one for calculating the response
in the original situation and another for the situation including the muffler.
The results will depend upon the nature of the source and the output radia-
tion impedance. There is not a unique value for insertion loss and the result
will clearly depend upon the individual case. Two different models are
shown in Figure 1; one case Figure 1 (b) having a constant velocity piston
source with the muffler terminated in an infinite transmission line and the
other, Figure 1 (c), having a similar source, but being terminated into a
half space through an infinite baffle. The finite element results for the
transmission loss and insertion loss problems shown in figure 1 are discussed
in a later section.
3.0 THE ACOUSTIC FINITE ELEMENT MODEL
The acoustic finite element used to obtain the results for this
paper is shown in figure 2. It is a hexahedral element having 8 nodes and
allows for a linear variation of pressure between the node points. Because
the element is an isoparametric element it can be distorted to any reasonable
shape. Therefore the use of this element enables problems having a difficult
geometry to be treated. For the results given here only axi-symmetric cases
were studied. With axi-symmetry, the three dimensional problem can be
treated as a two dimensional one with a substantial redu :tion in the size of
the problem. In this case the reduction in size was achieved by forming the
hexahedran into a segment of a thick cylinder, then equating the pressures
having equal radii and length coordinates. The element thus used has effec-
tively 4 nodes instead of 8. (see reference 1).
A typical grid used for a simple expansion chamber model is shown
in figure 3. Although this is quite crude compared with those required by
many other finite element solutions the results obtained were quite accurate.
403
-------
4.
4.0 RESULTS
Most of the results given below are to validate the method. Many
of these can be obtained from simple models of the system and they form a
useful check on the procedure. This is particularly true for reactive
mufflers when it can be assumed that acoustics within the expansion chamber
is strictly plane-wave and thus one dimensional. However, the plane wave
solution breaks down when the wavelength approaches the chamber diameter.
It is then that the finite element model shows a distinct advantage.
Results are discussed in turn for reactive mufflers, dissipative
mufflers with a locally reacting boundary and finally for lined mufflers with
extended reaction at the boundaries. The extended reaction is modelled by
extending the finite element approach to an absorptive material and then form-
ing-an acoustic-absorption model.
4.1 REACTIVE MUFFLERS: TRANSMISSION LOSS
The transmission loss of a simple expansion chamber in terms of
the area expansion ratio, m, length 1 and wave number k is given by a formula
due to Davis (3) :
T.L. = 10 log1Q (1 + l/4(m - 1/m)2 sin2 kl)
The finite element results are compared with those obtained form this form-
ula in figure 4. There is excellent agreement. Further results correspond-
ing to higher frequencies are given in reference (1), they show that when
diametral modes are excited they can either act as passing filters and thus
reduce the transmission loss or as blocking modes.
Figure (5) show the effects of extended inlet and outlet
pipes within the chamber. These act as quarter-wavelength filters which give
high transmission-loss values whenever the length of the extended pipe, le,
is given by Kle = nir/4, when n is any odd integer. The finite element results
show this to be the case.
4.1 INSERTION LOSS
Figure (6) compares the transmission loss results with the insertion
losses calculated for the two situations shown in figure 1. There is an
enormous difference and in one case, where the muffler is terminated into
a semi-infinite space the insertion loss shows negative values, thus the muf-
fler is enhancing the sound, where transmission loss calculations would indi-
404
-------
5.
cate a substantial reduction.
4.2 DISSIPATIVE MUFFLERS : LOCALLY REACTING BOUNDARIES
The calculation of the transmission loss for an expansion chamber
having a cylindrical absorptive lining is not a simple matter, although design
procedures do exist. See Beranek (4). It can be handled with a finite
element model by solving the general equations given in equation.]. When an
absorptive lining exists the terms in [CjJ and [CRJ are non-zero at points
corresponding to the boundary nodes where the liner is attached. The terms
in [CiJ and [C^] depend upon the form of the liner impedance. In this model,
the liner was assumed to be locally reacting with the impedances given by the
empirical equations developed by Delany and Bazley (5). See also reference (2)
These equations allowed for a semi-rigid porous material in which the charac-
teristic impedance was a function of the materials resistivity. The imped-
ance for any thickness was then calculated by assuming that the outer end of
the layer was attached to a rigid layer.
Results for the transmission loss are shown in figure 6, these
show the changes which occur when the thickness of the liner is increased.
With a thin liner, there i's little change from the unlined reactive case.
As the thickness increases, the multiple hump transmission loss character-
istic of the reactive muffler is replaced by a single hump which has a max-
imum when the thickness of the liner is 'approximately equal' to a quarter
wavelength. Thus the maximum value occurs at lower frequencies as the thick-
ness is increased.
However, there comes a point when the thickness is too great and
the magnitude of the reflected wave from the hard boundary is small, in
which case the boundary impedance of the liner approaches the characteristic
impedance of the liner material and no further changes in the transmission
Icrs occur.
DISSIPATIVE MUFFLERS WITH EXTENDED REACTION
An improved model of the acoustic lining is 'obtained if the
assumption that the boundary is locally reacting is removed. In order to
achieve this an acoustic absorption element has been developed based on a
Rayleigh model for a rigid-porous material. This element is again hexahedral
in form and is entirely compatible with the previously mentioned acoustic"
405
-------
6.
element. The general form of the response within the medium is again govern-
ed by an equation similar to (1), the differences with the acoustic equation
being found in the matrix[C]. For the absorption equations this matrix is
now fully populated and the magnitude of the terms are proportional to the
resistivity of the material. Further, details of this element are to be
published in reference (6).
The absorption elements can be joined to acoustic elements by equat-
ing the pressures at the common node points. A typical axi-symmetric model
is shown in Figure 7; this represents a cylindrical expansion chamber with
a thick lining. Results for such a chamber are also shown. When the resis-
tivity R = 0, the model is then of a simple reactive chamber and the trans-
mission loss has the typical "squared sine wave" appearance. The lining great-
ly increases the transmission loss when the resistivity R = 10,000 Rayls/
metre . Although experiments need to be carried out to verify the results,
the general form of the curve is in agreement with those obtained from lined
duct silencers used in ventilating systems.
COMMENTS.
The use of acoustic finite elements for modelling silencer systems
has been described. The method at this stage is particularly valuable when
difficult geometries are to be simulated and for predicting the performance
at high frequencies when the wavelength approaches the diameter of the expan-
sion chamber and one dimensional theories no longer apply. It is also use-
ful for modelling dissipative liners either with, locally reacting model in
which there is no substantial increase in the size of the matrices compared
with the reactive case or with absorption elements. The method can easily
be applied to Transmission Loss or Insertion Loss calculations.
The contents of this paper are mainly concerned with the work of
the author. However, the method has been applied to mufflers by other authors
with some success. Young and Crocker (8) calculated the transmission loss
of an expansion chamber using rectangular elements. Kagawa and Omote (9)
considered reactive mufflers using axi-symmetric ring elements and later
Kagawa, Yamabuchi and Mori (10) considered the transmission loss of a muffler
with a sound absorbing wall.
406
-------
REFERENCES
(1) A. Craggs 1976 "A Finite element method for damped acoustic systems;
an application to evaluate the performance of reactive
mufflers." Journal of Sound & Vibration (48) (3) p. 377-392
(2) A. Craggs A 1977 "A Finite element method for modelling dissipative
mufflers with a locally reactive lining." Journal of
Sound & Vibration 54 (2) p. 285-296
(3) D.D. Davis Jr. 1975 in Handbook of Noise Control, C.M. Harris (editor)
New York : McGraw-Hill Book Company Inc. Acoustical Filters
and Mufflers. Chapter 21.
(4) L. Beranek 1971 Noise and Vibration Control New York, McGraw-Hill Book Co.
(5) M.E. Delany and E.N. Bazley 1970 Applied Acoustics 3 105-116. Acoustic
properites of fibrous absorbent materials.
(6) A. Craggs A Finite Element Model for Rigid Porous Absorbing Materials
to be published.
(7) C.I.J. Young and M.J. Crocker 1975. Journal of the Acoustical Society
of America 57 p 144-148. Prediction of the Transmis-
sion Loss in Mufflers using the finite element method
(8) Y. Kagawa and T. Omote 1976. "Finite Element Simulation of Acoustic
Filters of Arbitrary Profile with Circular Cross Section."
Journal of Acoustic Society Am. 60 (5) p. 1003-1013
(9) Y. Kagawa, T. Yamabuchi and A. Mori. 197 "Finite Element Simulation of
Axi-Symmetric Acoustic Transmission System with a Sound
Absorbing Wall. Journal of Sound & Vibration
407
-------
Figure Captions
Figure 1 Models for Transmission Loss and Insertion Loss calculations.
(a) Transmission Loss (b) Insertion Loss : Constant Velocity
source terminated in an infinite line (c) Insertion Loss Constant
velocity source terminated infinite baffle.
Figure 2 The eight node isoparametric hexahedral element.
Figure 3 Two-dimension grid for an axi-symmetric expansion chamber model.
Figure 4 Transmission Loss, comparison of finite element results with
exact one dimensional solution at different expansion ratios m.
Figure 5 Finite Element results for the effect of extended inlet and out-
let pipes.
Figure 6 Comparison of Transmission Loss with Insertion Loss. Finite
element results. See Figure 1.
Figure 7 Transmission Loss for Expansion chamber with a cylindrical
absorbent lining. Impedance calculated using Delany & Bazley
equations. Figure shows effect of lining thickness. (m=a)
Figure 8 (a) The Axi-symmetric Acoustic-Absorbent finite element grid.
(b) Transmission-Loss with and without any absorption.
408
-------
J(u>t-kz)
p-e j(ut+kz)
(a)
,(wt-kz)
MUFFLER ELEMENT
(b)
Ue
jcot
oO
(c)
Ue
INFINITE BAFFLE
Figure
409
-------
3 NODE ISOPARAMETRIC
ACOUSTIC ELEMENT
Figure 2
410
-------
-*-
"T
d D
I
Figure 3
All
-------
m = 100
TL
(dB)
rr (kl)
Figure 4
412
-------
B
Figure 5
413
-------
Transmission Loss
Insertion Loss - Infinite
Insertion Loss -Open end
teminatjcn
Figure 6
414
-------
50
40
30
10
0
O05
0.20
3TT
4-ar
Figure 7
415
-------
A
1
Abscrt>tvov\
s
.A
s /
4-
s" y
41-
TL
Rnif*.
Figure 8
-------
A COMPARISON OF STATIC VS. DYNAMIC
TESTING PROCEDURES FOR MUFFLER EVALUATION
W. L. Ronci
10/21/77
417
-------
INTRODUCTION:
For the past ten years, Original Equipment exhaust systems have been
designed to meet the requirements of SAE Test Procedure J986a-. J986a
was the first noise test standard for light vehicles in this country -
The original development work on the procedure was done in early 1966.
The standard was first applied to new vehicles in 1967 and was
revised to its current version in 1968.
SAE Test Procedure J986a formed the basis for the so-called California
Passby Test. The California Passby Test is required under California
Vehicle Code 27160, for new motor vehicles under 6,000# gross vehicle
weight. The code first became effective in 1968. It has been revised
twice since, first in 1972 and again in 1973, when the current version
became effective. The California Passby Test procedure is defined
under Title 13, of the California Administrative Code.
A detailed comparison of the California Passby Test and the J986a
Passby Test will disclose that there are differences between the two
procedures. In actual practice the differences are minor. Test
results obtained by the two procedures correlate extremely well.
Walker uses the SAE procedure as specified by their Original Equip-
ment customers.
J986a TEST PROCEDURE
To conduct the test, a sound level meter microphone is placed 50-feet
off to the side from the center line of vehicle travel as shown in
Figure 1. The microphone is located four feet above the test surface.
The procedure calls for a flat open area, free from obstructions for
a distance of 100 feet in all directions.
Under the procedure, the test vehicle approaches the test section
at a steady state speed of 30 MPH. When the vehicle reaches 25 feet
from the test point, it is accelerated at wide open throttle. The
lowest gear ratio is used which will permit at least 50 feet of
accelerating distance without over speeding the engine. Passbys are
made under these conditions in both directions and the maximum ob-
served total sound pressure level for each passby is recorded. The
average of the two highest observations within two dB of each other
is reported as the test value for the vehicle. The test results are
reported for the noisier side of the vehicle.
It should be emphasized that the California Passby Test regulated
only new vehicles sold in that state. It did not regulate existing
vehicles. Nor did it regulate the replacement of noise-producing
or noise-silencing components, nor of vehicle modifications which
increase the total, vehicle noise.
418
-------
20" STATIC TEST PROCEDURE
Accordingly, the 1971 session of the California Legislature enacted
Vehicle Code 23130 which regulates aftermarket replacement exhaust
systems. The Commissioner of the California Highway Patrol was
directed to conduct a study to define procedures and standards by
which exhaust systems could be certified as meeting the established
allowable total vehicle noise levels. The California Highway Patrol
commissioned the McDonnell Douglas Company to develop a certification
program, stationary test methodology and related law enforcement
techniques. The study formed the basis for the regulations promul-
gated in November of "75 under Title 13 of the California Administrative
Code.
The test procedure adopted in the code was the so-called California
20" static test. The choice of a static test procedure was based
in large measure on the ineffectiveness of the driveby test proce-
dure in urban areas. The coverage attainable using the driveby test
in urban areas was limited because of the lack of suitable enforce-
ment sites with sufficient open area and low ambient noise levels.
The passby test was more appropriate to rural highways or freeways.
Moreover, being a total vehicle noise test, it was unsuitable for
regulating replacement mufflers. There was no simple enforcement-
means to.ensure that a cited vehicle was subsequently made legal.
The 20" Static Test Procedure specifies that the test be conducted
on an outdoor pavement or on a shop floor. A clear open area around
the test site of only ten feet is required. The microphone location
is dependent upon the-tailpipe routing as shown in Figure 2. Typi-
cally it is located 20" from the end of the tailpipe, 45° off-axis,
at the height of the tailpipe exit. The procedure calls for opera-
tion of the vehicle, after a suitable warmup, at 3/4 of rated RPM,
with the transmission in neutral. The value reported for the exhaust
system is the highest reading obtained, disregarding extraneous peaks.
CORRELATION STUDY
With the addition of a 20" Static Test Procedure which was to be-
come effective January 1, 1977, it was evident that the potential
existed for a dual design standard for exhaust system development.
Accordingly, Walker set about to determine whether there was suffi-
cient correlation between the two test methods to permit the pre-
diction of static test performance based on driveby tests, which
were currently being conducted for Original Equipment product.
The prime motivation for this was to reduce the total engineering
test load and to establish a single acoustic design and acceptance
test criteria. Data was taken on a variety of new vehicles. A
representative mixture of four, six, and eight cylinder passenger
cars v;<;re used in the tests.
419
-------
A number of different types of mufflers were tested. These included
the Original Equipment systems, with which the new vehicles came
equipped. The Original Equipment system sometimes incorporates a
smaller muffler or resonator. The system is usually made up on one
or more assemblies, with the pipe welded to the muffler. Figure 3
shows a typical example of an OE system assembly. Welded assemblies
are used to minimize the installation labor in the car factories.
The O.E. system is designed to meet both the objective requirements
of J986a and the particular car company's subjective sound quality
as it relates to the image of the vehicle in question.
Walker's regular aftermarket mufflers and resonators were also tested.
Regular mufflers and resonators are sold as separate units with the
system held together by clamps. Figure 4 shows a cut-away view of
a typical regular aftermarket muffler. Walker follows the practice,
which is common in the replacement exhaust system industry, of con-
solidating a number of Original Equipment designs into one after-
market design in order to achieve some economies of scale in produc-
tion and to minimize the stocking and inventory problems that would
otherwise exist. Walker's, indeed the industry's, ability to pro-
vide the consumer with an economically priced replacement part, on
a moment's notice, is heavily dependent upon its ability to consoli-
date O.E. Designs.
The construction techniques and acoustic design techniques of
Walker's regular muffler line, is quite similar to the Original
Equipment. Figure 5 shows a cut-away view of an OE design for
comparison. The subjective sound quality of the regular line con-
forms to Walker's own-corporate standards for preserving the Orig-
inal Equipment image of the vehicle. A Cadillac owner expects his
vehicle to sound like a Cadillac; a Corvette, like a Corvette.
Also included in the tests were Walker's WACO mufflers. These are
a highly consolidated line for certain customers such as K-Mart arid
Montgomery-Ward. The line is built to the same high quality and
construction standards as the regular line. Bushing adapters are
used to accommodate a wider variety of applications. On average
they are slightly smaller in size than the regular aftermarket muf-
fler or the Original Equipment design which they replace. Figure 6
shows a cut-away view of a typical WACO unit.
Walker's Unitized line was tested as well. The Unitized muffler
is a 4" round tubular design with swaged ends. This line has a
reasonably high degree of consolidation. Generally it uses a "Tri-
flow" acoustic design (See Figure 7) and is not as efficient at
the low frequencies because of the smaller physical volume. Single
and double tuned resonators are not used. The Unitized line was
introduced to satisfy the needs of car owners with older vehicles,
who are interested in economy.
420
-------
The fifth type of muffler included in the tests were glass packs.
Walker's glass packs also employ a 4" round construction with
swaged ends. In external appearance they look very much like a
Unitized muffler. Acoustically they are quite different. They
employ a straight thru design with a concentric perforated tube
surrounded by fiberglass, as shown in Figure 8. The design is
effective at absorbing high frequencies and is characterized by
a throaty, straight-thru sound quality. Generally it is both
objectively and subjectively louder than the other lines.
In total 305 systems were tested using both the 20" static test
procedure and the J986a passby method. Fifty-nine Original Equip-
ment systems were evaluated along with 110 regular mufflers, 50
WACO units and a combined total of 86 Unitized and glass pack
versions.
ANALYSIS OF RESULTS
The test data was analyzed using standard computer statistical
'techniques. The data was examined in a variety of ways. Simple
statistics were determined for each test method and each class
of muffler system; that is, the mean, the range and the standard
•deviation.. The simple statistics, while not very informative,
are presented in Tables I and 2.
Each class of muffler and the total population were subjected to
a correlation analysis from which the correlation coefficient was
determined. A correlation coefficient of one means a one-to-one
correspondence between the two test methods. A correlation coeffi-
cient of 0 indicates a totally random relationship between the
two tests. The results of the correlation analysis are shown in
Table 3. It is evident that there is no significant correlation
between the two. The data was also subjected to a regression
analysis. From this, a best, least-squares relationship between
the two test methods was eatablished. The lack of correlation is
very evident from the scatter diagrams shown in Figures 9 thru 13.
It can be seen that the predictive accuracy of the J986a test is
about t 20 to 30 dbA.
From the analysis it is apparent that there are different: accept-
ance criteria required for O.E. and aftermarket product. It is
eveident one cannot eliminate the need for running both tests.
It was also evident that potentially different design approaches
would be required for aftermarket and O.E. product.
It appeared that the internal construction of the muffler affects
the relationship between the test results obtained by the two
methods. This is apparent from the different correlation coeffi-
cients for the regular, WACO and Unitized mufflers configurations.
The increased correlation shown by the Unitized and glass pack
mufflers was probably attributable to the'lack of some low frequency
421
-------
tuning elements in these designs, and to the presence of a larger
component of exhaust noise in the passby test.
The test results lend jredence to another set of conclusions that
can be reached about the process by which the two California laws
were developed. A new vehicle law was passed first, which regu-
lated total vehicle noise without defining what the exhaust system
contribution to it would be, and without adequate provisions as to
how exhaust noise would be regulated on older vehicles. Next an
aftermarket law was passed to regulate exhaust systems. The end
result is two standards of acceptance of exhaust systems which
bear little relationship to each other. Perhaps this could have
been avoided had both O.E. and aftermarket been considered together
from the start.
These light vehicle standards have now been adopted almost without
change by the state of Florida and are being followed with interest
by the state of Oregon. The ultimate impact of these tests on the
industry's ability to continue the important practice of consolida-
tion is not yet fully known.
The federal government is presently developing a new set of accept-
ance criteria for passenger cars. This one will probably be based
on a totally different passby test. We have been meeting here the
last few days to discuss yet another criteria, this one a bench
test suitable for labeling exhaust system replacement parts. The
question of correlation between these two federal test methodologies
should be considered from the onset in their development.
The importance of considering the impact of these new regulations
on the industry's ability to consolidate Original Equipment designs
cannot be overemphasized. Should the industry lose this ability
and the number of replacement parts proliferate, the result would
be increased engineering costs, shorter production runs, increased
warehousing space and higher inventory costs. The end result of all
that will certainly be higher prices to the consumer and potentially,
delays on the part of the installer in finding a replacement part
for his customer's vehicle.
422
-------
itv
Mi
Approach
at 30 MPH
Basic Site Layout
J-986a Passby Test
r
Open throttle fully
25'
50'
Microphone-
Figure 1
-------
Microphone Placement
>8 In.
Microphone Height
20" Static Test Layout - Figure 2
-------
Muffler Type
O.E.
Regular
WACO
Unitized & Glass Pack
Total Composite
Muffler. Type
O.E.
Regular
WACO
Unitized & Glass Pack
Total Composite
Mean
Value
76.6
77.8
80.8
80.7
78.8
J-986a Test Resi
Table 1
Mean
Value
84.4
85.4
86.6
92.2
87.3
20" Static Test
Table 2
Standard
Deviation
3.2
3.5
3.2
4.1
4.0
alts
Standard
Deviation
3.6
4.5
4.3
5.3
5.5
Results
Lo
71.7
71.2
74.4
73.3
71.2
R
Lo
78.1
78.2
79.0
82.6
78.1
Range
Hi
87.8
88.2
88.2
94.0
94 .0
ange
Hi
92.1
95.8
96.9
105.4
105.4
Muffler Type Correlation No. of
Coefficient Observations
O.E. .245 59
Regular .333 110
WACO .282 50
Unitized & .Glass Pack .451 86
Total Composite .462 305
Correlation Analysis Results
Table 3
425
-------
Typical Original Equipment Exhaust System
Figure 3
7^rSSF55?rS
• -'
C u t a \v a y View -
Regular Aftermarket Muffler
Fiaurf 4
426
-------
Cut-away View - Typical OE Muffler
Figure 5
Cut-away View - WACO Muffler
Figure 6
427
-------
Cut-away View - Unitized Muffler
Figure 7
Cutaway View - Glass Pack Muffler
Figure 8
428
-------
OE Mufflers
90
m
-p
in
0)
EH
U
•H
-M
-------
100 t
90
m
(fl
dj
EH
U
•H
-P
(fl
4J
O
(N
80
Walker Regular Mufflers
O •
0 «
«
a o »3 a «
to • 99 9
• • •
Regression
line
• - Single Point
O - Two Points
70
80
J-986a Passby Test - dBA
Figure 10
430
90
-------
100-r
Walker WACO Mufflers
w
•a
i
-p
Q)
EH
U
•H
-P
90-
Regression
line
o
(N
80- -
70
80
•T-986a Passby Test - dBA
Figure 11
90
431
-------
110 -r
Walker Unitized & Glass Pack Mufflers
O
100
m
4-1
W
0)
EH
O
-H
4-1
4J
O
CN
90
• •
Regression
line
• - Single Point
O - Two points
80
70
80
J-986a Passby Test - dBA
Figure 12
432'
90
-------
110 -r
All Mufflers
100
PQ
-p
in
OJ
90
10
O
80 —
Regression
Line
•oo
• o •
o ••• •• • •
O •• 09 •
• • • • 0 • ••
o •• ./ • «• •
Oca • • •
o • •
o* • • • • •
o* • • • a* •
'•o • • •• • o •
• • • • • • •
••O OO»»OOOO O • « •••
• • oo»» ••• • *
X
X
X
• • • • •
o ••• •
•• • • •
• •• • ••
• •
• • •
x
X 95% Limit
X
X
X
X
X
• - Single Point
O - Two Points
+ - Three Points
70
90
J-986a Passby Test - dBA
Figure 13
433
-------
DISCUSSION OF PROPOSED SAE RECOMMENDED PRACTICE
XJ1207, MEASUREMENT PROCEDURE FOR DETERMINATION OF
SILENCER EFFECTIVENESS IN REDUCING ENGINE INTAKE OR EXHAUST SOUND LEVEL
by
Larry J. Eriksson
Nelson Industries, Inc.
Stoughton, WI 53589
Presented at the
United States Environmental Protection Agency
Surface Transportation Exhaust Noise Symposium
Chicago, Illinois
October 11-13, 1977
435
-------
Discussion of Proposed SAE Recommended Practice
XJ1207, Measurement Procedure for Determination of
Silencer Effectiveness in Reducing Engine Intake or Exhaust Sound Level
by
Larry J. Eriksson
Nelson Industries, Inc.
Stoughton, Wisconsin
ABSTRACT
The development of Proposed SAE Recommended Practice XJ1207, Measurement
Procedure for Determination of Silencer Effectiveness in Reducing Engine
Intake or Exhaust Sound Level is reviewed. This Recommended Practice describes
a procedure for a measurement of the actual sound level produced. Successive
measurement may be performed to obtain relative performance values or insertion
loss. Various considerations in the writing of the procedure are discussed
and limitations reviewed.
IN RESPONSE to a need for a standardized test procedure for exhaust and
intake silencers, the SAE Vehicle Sound Level Committee (VSLC) formed the
Exhaust and Induction Silencer Subcommittee in December of 1974. The objective
of this subcommittee was to develop "insertion loss measurement methods in
order to provide a rating for the respective devices." Since it was felt
that a single procedure was feasible for exhaust and intake silencers, the
standard was to be developed in close liason with the Air Cleaner Test Code
Subcommittee of the SAE Engine Committee.
BACKGROUND
Membership was sought for the Exhaust and Induction Silencer Subcommittee
(EISSC) from a broad spectrum of technical personnel including those involved
with exhaust silencers, intake silencers, engines, and vehicle applications.
An organizational meeting was held in March of 1975 to review possible directions
for the Subcommittee's work. Numerous existing test procedures were reviewed
at this meeting as well as subsequent meetings. These included SAE Recommended
Practice J1074, Engine Sound Level Measurement Procedure, and the SAE Recommended
Practice J1096, Measurement of Exterior Sound Levels for Heavy Trucks Under
Stationary Conditions, as well as procedures developed by such organizations as
43F
-------
the Industrial Silencer Manufacturers Association (ISMA) and Department of
Transportation (DOT). Although useful ideas were .obtained from many of these
sources, no procedure was found to meet the requirements for a standard test
procedure, for exhaust and intake silencers as specified by the VSLC charge to
the Subcommittee.
MAJOR CONSIDERATIONS
Two major areas of concern were discussed in detail. The first was the
type of noise source to be used in the evaluation of the silencer. Among
those considered were.a speaker, a blower, and a standardized engine. Finally,
it was concluded that in order to obtain sufficient accuracy, compatible with
other SAE Recommended Practices, it would be necessary to use the actual
engine and silencer system for which the silencer was to be applied. This
approach was thought to have the potential of providing the most accurate
engineering data for these types of units.
The second major area discussed was the type of measurement that should
be made on the silencers. Again, a broad range of possibilities were considered.
These included insertion loss, transmission loss, transfer function, and actual
sound level. It was concluded in this case that in order to meet the dual goals
of a test procedure that could be widely used as well as provide usable data
that could be related to other measurements, the actual sound level produced
with the silencer system installed on a given engine should be the measured
quantity. It was further noted that the option remained for the test procedure
in this form, to be applied successively to different silencers to obtain relative
performance values or to silenced and unsilenced cases to obtain insertion loss
(IL).
437
-------
ADDITIONAL CONSIDERATIONS
Other areas discussed included the wide range of sizes of silencers,
engines, and test facilities that would be involved in using the desired test
procedure. While the subcommittee felt that measurements at 15 metres (50 feet)
from the silencer were most desirable to be consistent with other test methods,
it was thought that other distances should be allowed in order to make the
procedure practical for use with small engines and light duty applications where
the available measurement distances are often considerably less than 15 metres
(50 feet).
It was also concluded that the procedure should allow for measurements
in a free field above a reflecting plane. This may be obtained either in a
flat open space or semi-anechoic chamber. The former offers the advantage
of a potentially better free-field condition, but also the disadvantage of
potentially more problems with ambient noise, wind, temperature gradients,
and other weather variables. The latter approach, the semi-anechoic chamber
requires extensive wall treatment to obtain adequate free-field behavior,
but offers better control over weather conditions and ambient noise. In view
of these tradeoffs, the subcommittee decided to include both approaches with
specific requirements for both, This decision also resulted in data that were
more widely obtainable as well as comparable to those obtained using other test
procedures usually performed outdoors,
INITIAL DRAFT
Following these discussions, the first draft of the test procedure was
completed in September of 1975. It included the above factors and required
an isolated test cell containing the specific engine to be used.in the measure-
ment with an adjacent free field above a reflecting plane. The exhaust or
intake system was to be piped to this open space' and placed in an orientation to
the ground as similar as possible to the actual end application, The piping
438
-------
from the engine to the silencer was to be acoustically treated to eliminate
all contributions to the measured level from this pipe. This was done since
some pipe had to be excluded in order to connect to the isolated engine and
thus, excluding all of this noise was the only practical method to standardize
various test facilities that might be used. However, all noise from the surface
of the silencer as well as the tailpipe must be included in the measurement
along with noise from the acoustical outlet.
This first draft was subsequently extensively modified'until finally
reaching its final form as approved by the VSLC in June of 1977 and balloted
to the SAF Motor Vehicle Council (MVC) in August of 1977.
COMMENTS ON FINAL DRAFT
Among the areas receiving considerable attention during the various
revisions was instrumentation. The primary concern was to obtain sufficient
information to determine that the engine was functioning properly. The mode.s
of engine operation were also reviewed in detail. It was determined by the
subcommittee that the peak sound level could occur under a fairly wide variety
of conditions depending upon the specific silencer-engine combination being
tested. Thus, a steady state and varying speed mode are required along with
an acceleration test for governed engines. Fast dynamic response of the
sound level meter was selected for all modes as providing adequate results
with minimum potential for error.
The final version of the test procedure does not include any measurement
of the restriction of the silencer system. While this is acknowledged to often
be an important parameter along with many other specifications, it was not felt
to be directly related to the sound level measurement and as such was excluded.
439
-------
Because of the wide variety of test set-ups this procedure applies to,
it is recommended that a photo or diagram of the test set-up be included with
the test results.
LIMITATIONS
Among the limitations of this test procedure are the lack of a direct
correlation to other overall vehicle pass-by tests as well as the lack of
specification of the subjective quality of the exhaust or intake noise. This
aspect can be quite important for many applications in which the overall
A-weighted sound level is not an adequate description of the acoustic acceptability
of a silencer.
440
-------
APPENDIX A
Members of Subcommittee During Development
Name
*J. Cahill (Secretary)
*P. Cheng
W. Dreyer
*J. Dreznes
*F. Egbert
*L. Eriksson (Chairman^
R. Heath
R. Hunt
,S. Koehler
*K. Li got
*K. .Nowak
*W. O'Neill
*R. Palmer
C. Reinhart
*D. Rowley
G. Shaltz
*D. Thomas
Affiliation
Stemco Manufacturing Co.
Stemco Manufacturing Co.
Walker Manufacturing Co.
United Air Cleaner
International Harvester Co
Nelson Industries, Inc.
Walker Manufacturing Co.
Stemco Manufacturing Co.
Donaldson Co.
Walker Manufacturing Co.
Cosmocon, Ltd.
Fram Corporation
AP Parts Co.
Donaldson Co.
Donaldson Co.
United Air Cleaner
. . . with contributions from many others
* Current Members
441
-------
APPENDIX B
MEASUREMENT PROCEDURE FOR DETERMINATION OF SILENCER
EFFECTIVENESS IN REDUCING ENGINE INTAKE OR EXHAUST SOUND LEVEL
XJ1207
1.0 Scope - This SAE Recommended Practice sets forth the instrumentation,
environment, and test procedures to be used in measuring the silencer system
effectiveness in reducing intake or exhaust sound level of internal combustion
engines. The system shall include the intake or exhaust silencer, related
piping and components. This procedure is intended for engine-dynamometer
testing and is not necessarily applicable to vehicle testing (see Appendix
A). The effect of the exhaust or intake system on the sound level of the
overall machine must be determined using other procedures. This procedure
may be successively applied to various silencer configurations to determine
relative effectiveness. Insertion loss for individual silencers may be
calculated through measurement of the silenced and unsilenced system.
2.0 Instrumentation - The following instrumentation shall be used for the
measurement required:
2.1 A sound level meter which meets the Type 1 or S1A requirements of
American National Standard Specification for Sound Level Meters, SI.4-1971
(R1976).
2.2 As an alternative to making direct measurements using a sound level meter,
a microphone or sound level meter may be used with a magnetic tape re-
corder and/or a graphic level recorder or indicating instrument, providing
the system meets the requirements of SAE Recommended Practice, Qualifying
A Sound Data Acquisition System - J184.
2.3 A sound level calibrator having an accuracy within +_ 0.5 dB. (See
paragraph 6.2.4.)
2.4 A windscreen may be used. The windscreen must not affect the microphone
response more than +_ 1 dB for frequencies of 20 - 4,000 Hz or +_ 1.5 dB
for frequencies of 4,000 - 10,000 Hz. (See paragraph 6.3.) ~
442
-------
2.5 If outside tests are being performed, an anemometer or other means for
determination of ambient wind speed having an accuracy within + 10% at
19 km/h (12 mph).
2.6 A thermometer or other means for determination of ambient and engine
intake air temperature, having an accuracy within j^l C (+_ 2 F).
2.7 A thermometer or other means for determination of fuel temperature at
the fuel pump inlet having an accuracy within +_ 1°C (+_ 2 F).
2.8 A barometer or other means for determination of ambient and engine
intake air barometric pressure, having an accuracy within j^ 0.5% of
the actual value.
2.9 A psychrometer or other means for determination of ambient and engine
intake air relative humidity, having an accuracy within +_ 5% of the
actual value.
2.10 An engine dynamometer with engine speed and torque (or power) indicators
having an accuracy within +_ 2% of the rated engine speed and torque
(or power).
2.11 A flowmeter or other means for determination of engine fuel rate having
an accuracy within +_ 1% of the rated fuel flow.
3.0 Environment - The silencer shall be measured in an environment such that
results are equivalent to those obtained in a free field above a reflecting
plane. Measurements may'be made at a flat open space or- in an acoustically
equivalent test site as described in Appendix B.
3.1 The flat open space or requivalent test site shall be free from the
effect of a large reflecting surface, such as a building or hillside
located within 30 m (100 ft) of either the silencer opening or micro-
phone. The area directly between the silencer opening and the micro-
phone shall be concrete or sealed asphalt with a total deviation of
+ 0.05m(j^2 in.) from a plane extending at least 3.0 m (10 ft.) in all
"directions from all points on the line segment between the silencer
outlet and the microphone'.
3.2 The ambient A-weighted sound level (including wind effects and other
noise sources such as the engine) shall be at least 10 dB lower than
the level being measured.
3.3 Not more than one person other than the observer reading the meter shall
be within 15 m '(50 ft) of the silencer opening or microphone, and that
person shall be directly behind the observer who is reading the meter,
on a line through the microphone and the observer, or behind the silencer
under test.
443
-------
4.0 Procedure
4.1 The silencer shall be tested on the engine and silencer system for which
data will be reported.
4.2 The specified silencer system configuration shall provide for measurement
of the acoustical radiation from the surface of the silencer or silencers,
connecting pipes, and the acoustical outlet of the system. This does not
include piping from the engine to the silencer. The silencer system
should be oriented in the same relative position to the- ground as for
the actual application. Any deviation must be reported with the test data.
All system connections are to be free from leaks. For determining the
insertion loss, the unsilenced system shall include a pipe of physical
length equal to the silencer.
4.3 The engine and fuel rate shall be measured at full load from 2/3 of rated
speed to governed speed, or to rated speed on ungoverned engines, to
determine whether the engine is within the engine manufacturer's performance
specifications prior to proceeding with this test procedure,
4.4 The engine shall be operated in the following modes after reaching normal
operating conditions:
(a) Steady state mode - rated engine speed and full load.
(b) Varying speed full load mode - engine speed to be slowly varied
from rated speed to 2/3 of rated speed at wide open throttle.
For governed engines only:
(c) Acceleration mode - accelerate the engine from idle to governed
speed until the engine speed stabilizes. and return to idle by
rapidly opening and closing the throttle under no load conditions.
5.0 Measurements
5.1 The microphone shall be located at a height of 1.2 m (4 ft) above the
ground plane and at a horizontal distance of 15 m (50 ft) from the
center! ine of the silencer system. Other optional distances such as 7.5 m
(25 ft) may be used and must be reported. The angular location of the
microphone relative to the silencer system opening shall be recorded.
5.2 The sound level meter shall be set for fast dynamic response and for the
A-weighted network,
5.3 For the procedure specified in Paragraphs 4,3 and 4,4, report:
444
-------
(a) Engine power and fuel rate as determined in Paragraph 4.3,
(b) Ambient wind speed, ambient temperature, ambient barometric pressure,
ambient relative humidity, and ambient A-weighted sound levels for
the test site.
(c) Maximum A-weighted sound level measured for each test mode in
Paragraph 4.4.
(d) Torque (or power), engine speed, engine intake air temperature,
barometric pressure, and relative humidity at which the maximum
sound level was obtained.
(e) Any deviations from recommended test procedure as described in
Section 4.2.
(f) The angular location and distance of the microphone relative to
the silencer opening.
(g) Description of the test configuration, including .all pertinent
lengths.
i
6.0 General Comments
6.1 It is essential that persons technically trained and experienced in the
current techniques of sound measurement select the equipment and1conduct
the tests.
6.2 Proper use of all test instrumentation is essential to obtain valid
measurements. Operating manuals or other literature furnished by the
instrument and manufacturer should be referred to for both recommended
operation of the instrument and precautions -to be' observed. Specific
items to be considered are:
6.2.1 The type of microphone, its directional response characteristics,
and its orientation relative to the ground plane and source of
noise.
6.2.2 The effects of ambient weather conditions on the performance of
all instruments (for example, temperature, humidity, and barometric
pressure). Instrumentation can be influenced by low temperature
and caution should be exercised.
6.2.3 Proper signal levels, terminating impedances, and cable lengths
on multi-instrument measurement systems.
6.2,4 Proper acoustical calibration procedure, to include the influence
of extension cables, etc. Field calibration shall be made immediately
before and after each test sequence. Internal calibration means
is acceptable for field use, provided that external calibration
is accomplished immediately before and after field use.
445
-------
6.3 It is recommended that measurements be made only when wind speed is below
19 km/h (12 mph).
6.4 It is recommended that a drawing or photograph of the test configuration
be included in the reported results.
7.0 References - Documents referenced in this Recommended Practice are:
7.1 ANSI SI.4-1971 (R1976), Specification for Sound Level Meters.
7.2 SAE J184, Qualifying a Sound Data Acquisition System.
7.3 ANSI SI.13-1971 (R1976), Methods for Measurement of Sound Pressure Levels.
ANSI documents available from American National Stds. I'nst., 1430 Broadway,
New York, NY 10018.
446"
-------
APPENDIX A
A typical test layout may include an engine-dynamometer located in an acoustically
isolated test cell adjacent to the test site. The piping from the engine to
the silencer should extend from the isolated test cell to the test site. The
silencer system should be oriented in the same relative position to the ground
as for the actual application. All piping.between the engine and silencer
should be acoustically treated to meet the requirements of Paragraph 3.2
The sound level measured during the test should include outlet sound as well
as shell sound from the silencer and connecting pipes, but not including the
piping from the engine to the silencer. The test site may consist of a flat
open space or acoustically equivalent indoor or outdoor test site.
APPENDIX B
If a facility other than a flat open space (Paragraph 3.1) is used, the
A-weighted sound level from a broad band sound source must not deviate over
the test distance from the response in a free field above a reflecting plane
more than +_ 1 dB. Measurement considerations in American National Standard
Methods for Measurement of Sound Pressure Levels, ANSI SI.13 - 1971 (R1976),
shall be used.
.447
-------
A Theoretical Examination of the Relevant Parameters
for Dynamometer Testing of 2-Cycle Engine Mufflers
by
Professor G. P. Blair
Department of Mechanical and Industrial Engineering,
The Queen's University of Belfast
Abstract
A powerful design tool has been developed for the prediction of
noise and performance characteristics for two-stroke cycle engines of
the type used for motorcycles, chainsaws, outboard marine units, or
snowmobiles. Here it is used to assess the various parameters involved
in dynamometer testing of an engine when fitted with an exhaust muffler
by comparison with the normal utilization of the product. A motorcycle
example is used to illustrate the several problems inherent in such a
technique and the effectiveness of the computer program in providing
solutions to them. The precise usage of the computer program is presented
in an appendix.
449
-------
1.1 Introduction
The history of the internal combustion engine is peppered with
theoreticians whose dream it is to predict the performance of some
particular unit, or type. The history of i.e. engine silencers, or
mufflers as they are referred to in the United States, is equally laced
with theoreticians with absolute design pretensions. It has always
amazed this author that the former group rarely include the detailed
geometry of an exhaust (or intake) silencer as part and parcel of their
design for engine power or efficiency and that the latter section will
cheerfully design a muffler in acoustic, pseudo-acoustic, or in
electrically analagous terms as if the engine barely existed. Yet
the interrelation of these components is all too obvious.
The blunt truth is that designers of either type have, with some
notable exceptions, failed to attempt their theoretical design procedures
based on reality, namely the mathematical tracing of the thermodynamic
state, position and velocity for every particle of gas from the time it
enters the "system" until it leaves it. The "system" is of course the
engine and its intake and exhaust silencers. Should such a calculation
be carried out then in engine terms its performance characteristics can
be deduced as power, torque, fuel and air consumption and thermal efficiency
at some particular rotational speed and in noise terms the separated
intake and exhaust noise spectra and levels can be determined at any
desired location in space from their sources at the "system". That i_s_
a design procedure, for then the effect of changing the most detailed
of geometry on both noise and performance can be evaluated.
It will be noted in the foregoing that no mention has been made of
two-stroke or four-stroke cycle, Diesel or spark-ignition, rotary or
reciprocating piston, super/turbo-charged or naturally aspirated engine;
nor is there need to for the theories of unsteady gas dynamics are as
450
-------
catholic in application as the particles of air are non-sectarian on
the topic of into which engine type they should be ingested.
2.1 Theory
Computer programs can, and have, been assembled for the derivation
of performance characteristics for most of the engine types listed in 1.1,
but not many of these solutions have been extended to deriving the intake
and exhaust noise spectra created. In the appendix to this paper there
is a report issued from the Queen's University of Belfast, report No.1096,
describing the input and output data from such a calculation for a
single-cylinder, naturally aspirated, spark-ignition, gasoline burning,
crankcase compression, two-stroke cycle engine; several species of intake
valving can be catered for as can the most complex geometry for the "system"
for this common type of i.e. engine. The references in that appendix
describe the background experimental and theoretical work over the last
thirteen years and the level of correlation between measurement and
calculation which now justifies the computational method as a working
design tool. Further discussion here would be verbiage.
One of the computer programs, type GPB2, will be used here to illustrate
the various problems associated with testing mufflers on a dynamometer
as a means of evaluating their performance in their natural environment.
As can be seen in the appendix, program GPB2 describes a typical single-
cylinder engine with piston controlled inlet porting and having a
performance tuned exhaust system but with exhaust silencer consisting of
four expansion boxes in series and with a single expansion box type of
induction silencer. The actual data used is for an existing 250 cm^
machine sold in the United States for 'enduro' or 'desert' racing. A
listing of the 'standard' data is shown in Fig.1 with certain of the
values covered, for the data and the engine form part of a design developed
451
-------
at QUB for a particular manufacturer and are consequently of some
confidentiality. Also shown on Fig.1 is the output for the peak
horsepower speed of 8000 rev/rain and the description of the symbols
and the data nomenclature is given in the appendix.
2.2 Theoretical solutions to some problem areas
2.2.1 When a motorcycle engine is being tested on a dynamometer, either without
or within its production chassis, and a microphone is placed in the
dynamometer- test area,unless some acoustic cover is provided for it
then it will record the summations of the various noise sources, namely
intake, exhaust and mechanical noise. In the nomenclature for program
GPB2 the microphone is positioned at distance RPATHI and RPATHE from the
intake and exhaust noise sources. The program provides no information
as to mechanical noise levels.
The possible experimental solution to the dynamometer assessment
of the effectiveness or otherwise of an exhaust muffler would be to
acoustically shield the entire test area but have the exhaust orifice
appear outside that shield and the positioning of the microphone at
RPATHE from that orifice becomes a less critical factor.
A theoretical examination of these possibilities appears in section
3.2.1 by comparison with the noise made jointly by intake and exhaust
noise sources under the test conditions imposed by typical acceleration
test procedures at 7.5 or 15.0 m employed by several legislative
authorities.
2.2.2 One of the simplest methods of silencing any engine device is to throttle
the intake or exhaust systems; this has the distinct commercial ana
ecological disadvantage in that, almost certainly, engine performance
and efficiency deteriorate respectively. An examination of the effectiveness
or otherwise of this approach is discussed in section 3.2.2.
452
-------
2.2.3 Under acceleration test conditions on a track the vehicle passes through
a torque and power speed range as well as a noise-speed related spectrum.
The theoretical program allows one to examine in detail the performance
and noise-speed spectrum in detail and permits the redesign of the silencer
so as to eliminate the worst noise case at a particular speed point without
reducing the overall engine performance; for it is that 'worst1 noise
point which will register on an acceleration test. Some riders of
motorcycles have demonstrated their ability to record lower (by 1 or 2 dB)
noise values under acceleration test conditions and this is managed by
their instinctive ability to hold that 'worst' noise-speed point to be
either well before or well after the minimum microphone to machine distance
point. Further discussion of this is contained in section 3.2.1 where
actual values are quoted.
2.2.4 One of the difficult assessment problems as to the effectiveness or
• otherwise of an exhaust muffler, and it applies equally to dynamometer
and acceleration truck testing, is when an exhaust muffler is being
employed in the presence of an intake noise level which is either equal
to, or is in excess of, that emanating from the exhaust source. The same
comments apply tequally to mechanical noise but that is outside the scope
of the theoretical examination here. Discussion of this problem with
predictions from program GPB2 to assist in its illumination are presented
in section 3.2.3.
453
-------
Discussion
3.1 The information presented here is but a minor fraction of the total
available from the several computer runs involved in numerically
highlighting the general nature of potential problems in sections
2.2.1 to 2.2.4.
3.2.1 A summary of the main performance characteristics of the engine are
shown in Fig.2 over the speed range between 5000 - 8000 rev/min which
would be that employed for a typical acceleration test, irrespective of
microphone positioning and test conditions. Presented on Fig.2 are
both experimental and theoretical values at each speed point for
power (bhp), delivery ratio and brake specific fuel consumption (Ib/hp.hr)
The theoretical values are predicted by the program GPB2 for the listed
data in Fig.1 and the experimental or measured values were provided by
the engine manufacturer; thus not all theoretical values predicted here
have a measured equivalent. The engine is running at full throttle
both theoretically and on the measured dyho test data, and as it would
be for an acceleration noise test. The theoretical/experimental
correlation is quite good.
"Acceleration Test"
The contribution of the intake and exhaust noise sources to the
overall noise levels at each speed point on the 'acceleration' test
are shown in Fig.3, as predicted theoretically for microphone positions
of 15.0 m for both sources. The noise levels on Fig.3 are computed
as dBa while the equivalent data for the same situation but with total
noise levels calculated are plotted as dBLIN on Fig.4. It can be seen
that the intake noise is lower than the exhaust noise in general, but
has two quite distinct peaks at 5500 and 7000 rev/min. It will be
noted that the peak exhaust noise occurs at 6500 rev/min. The
454
-------
7000 rev/min, irrespective of -hether the noise recording occurs by
dBa or DBLIN criteria. The overall noise/speed spectrum is quite
flat, produced mainly by a noisier and "flat" exhaust noise/speed
characteristic. Should the intake noise have been at a higher level
a totally different situation would have occurred.
A Test Muffler Problem
The kernel of a potential problem for muffler assessment appears
here; let us assume for a moment that the above defined system passed
the "test", just. Let us suppose that a new exhaust muffler is to be
assessed and it is found that this alternative device has a noise/speed
characteristic no higher in peak value than the standard unit, at
75.9 dBa, but the peak occurs at 7000 rev/min and not at the 6500 rev/min
for the initial silencer. The nett effect would be that the peak intake
and exhaust noise/speed points would coincide and produce a peak noise at
6500 rev/min perhaps 2dB higher than the current highest value. Does
this silencer then fail the "acceleration" test; almost certainly for
the peaks tend to get recorded!
Typical Noise Spectra
The program predicts the intake, exhaust and overall noise spectra
at whatever independent microphone position is selected. Present in
Fig.5 is the noise spectra from the 7000 rev/min positions in the
calcuations discussed above. It can be seen that the principal source
of noise is the peak in the exhaust noise spectrum between 450 and 700 Hz,
whereas the intake noise spectrum has a dip at that position, otherwise
the overall noise peak would have been even higher. It can be seen that
the exhaust noise spectrum falls off rapidly after 1000 Hz whereas the
intake spectrum stays very flat until 2000 Hz. The combination of these
two characteristics results in a sustained noise source with a relatively
flat overall residual spectrum, influences the overall sound level and
should be the frequency to be tackled by (say) a suitable side resonator
element in any redesign of the unit
455'
-------
Microphone Positioning
In a dynamometer test situation where the intake (and mechanical)
noise is not shielded from the microphone which is being used to
record (or attempt to record) the exhaust noise then the microphone
positioning becomes critical. The relatively obvious conclusion is to
place it as close to the exhaust noise outlet as is practical. An attempt
to illustrate this point is made in Figs. 6 and 7 in the form of tabular
data and in Fig.8 as a graphical representation.
In Fig.6 is shown the intake, exhaust and overall sound pressure
levels (dBa) for several combinations of microphone positioning relative
to the intake source point (RPATHI) and the exhaust outlet (RPATHE),
with the relative positioning being mostly 0.5 m nearer to the inlet
in most cases for dyno work and 7.5/7.5 or 15.0/15.0 m to represent
the acceleration equivalent. The reverse situation is shown in Fig.6
where the microphone is more logically placed closer to the exhaust
outlet.
At equal/equal microphone positioning it will be remembered that
the exhaust noise is some 2dB greater overall than the intake level.
A close examination of the figures reveals the relatively obvious,
namely, the closer one approaches the exhaust outlet with the microphone
the more nearly does the exhuast noise level and the overall noise
level coincide. Thus any careless positioning of the microphone, such
as positioning (b) or (c) in Fig.6, would mitigate against any clear
assessment of a 1 or 2dB difference in the performance of any particular
exhaust muffler. The curves of noise levels for intake and exhaust
noise at various independent microphone positions are shown together
on Fig.8. While equal/equal microphone positioning produces an
approximately constant 2dB differential, the differential microphone
positioning for equal noise levels from both sources inci. a.ses with
456
-------
distance. In other words at lOOdB noise level from both sources
the differential microphone positioning is 0.2 m at about 0.75 m median
value but for 76dB equality the differential spacing is 3.2 m on a 12.3 m
median point.
Close positioning of the microphone to the exhaust outlet would not
necessarily require the acoustic shielding of other noise sources for
dynamometer test purposes.
3.2.2 Throttling the Exhaust Outlet
The four-box silencer used in the relatively simple silencer
design discussed in the previous sections has basically four elements of
different volumes connected by 24 mm diameter tubes. The calculation
at 7000 rev/min was repeated for a microphone positioning of 7.5/7.5 m
equality of distance from intake and exhaust inlet/outlets respectively.
It will be remembered that 7000 rev/min was the highest noise point on
the noise/speed characteristic. In each of five calculations the diameters
DD1, DD1R, DD2 and DD2R were changed successively from 16.0 to 18.0 to
20.0 to 22.0 and to 24.0 mm; the latter value being the original standard
calculation. In other words the final outlet tube diameter was changed
from 16.0 mm to the standard 24.0 mm value in several steps. The results
for power, delivery ratio, brake specific fuel consumption and exhaust,
intake and overall noise are shown on Fig.9.
There is no doubt that throttling the exhaust outlet down to 16.0 mm
from 24.0 mm diameter certainly reduces the overall noise by some 4dB,
but more significantly to below the levels for the intake noise which now
becomes the predominant source. The equality of noise level occurs at
an outlet diameter of 22 mm; here the overall noise level is reduced by
just 1.5 dB for 0.5 hp penalty in power and none in fuel consumption.
Significantly, although the air flow was reduced by some 2% the intake
noise slightly increas_ed.
457
-------
Further throttling to 16.0 mm produces a considerable drop in
power (6 hp), a deterioration in engine efficiency (the bsfc increased
by some 10%); while the air flow rate decreased by some 15% the intake
noise barely altered, indeed it actually increased by IdB at ,the point
where the outlet diameter was 20 mm.
It can be seen that in any muffler assessment program, a device
which is overly restrictive on the entire system reduces both engine
power and efficiency, and must be recognised and categorized as such a
device. The test methods should be capable of differentiating between
the silencer which is allowing the engine to produce its rated power
and efficiency within the noise limits and the badly designed or produced
device which derates the power unit so as to fit within the legislative
framework. In these ecologically-conscious days retention of high engine
thermal efficiency is as important as excessive noise.
3.2.3 In section 3.2.1 the importance of the design of the intake silencer
was pointed out; particularly emphasized was the necessity to ensure
that the noise peak in the intake spectrum did not coincide with that
from the exhaust system.
On the "standard" engine the intake box, Box 1, had a volume of
7200 cm3 with a 40 mm outlet tube diameter (all diameters DS1 - DS2R).
This was replaced by a smaller box, Box 2, of 2500 cm3 volume and a
tube of 44 mm diameter of the same length. This was so arranged as to
produce the same total air flow at 7000 rev/min and therefore the same
power from the engine with a common "standard" exhaust system for each
"paper-engine computer-dynamometer test" situation. The exhaust noise
is unaltered in consequence.
The overall noise (intake) levels and their frequency spectrum
are shown in Fig.10 and the first point to be observed is greatly
increased overall sound pressure level peak (dBLIN) at the first
458
-------
harmonic (116.7 Hz). It is at this point that one must observe that
one has grave doubts about the legitimacy of the A-weighting factor at
this frequency; for be assured that should one ride a motorcycle with
such a replacement (Box 2) intake silencer box then this low frequency
noise peak would be obtrusive and unpleasant. As the facts stand the
application of the A-weighting characteristic produces an overall sound
level for Box 2 only 0.6dB higher than the original design. Perhaps
it is time to reconsider the application of a total sound pressure
level (dBLIN) criteria for legislative purposes.
Conclusions
The theoretical procedures illustrated here show the usefulness of a
design tool which is that in a true sense; it has the capability to reveal
the separate intake and exhaust noise production at independent distance
assessment points as well as the interaction of the intake and exhaust mufflers
on the engine and its performance parameters.
The program here is oriented towards the two-cycle motorcycle, outboard,
snowmobile, chainsaw, or industrial engine type; there is no theoretical
barrier to its application to any internal combustion engine which inhales or
exhales in the commonly unsteady manner.
Acknowledgements
The author would like to acknowledge the efforts of past research students
at The Queen's University of Belfast who laid the groundbase for this computer
program series and who carried out the experimental work to verify the accuracy
of their theoretical premises; Dr. J. A. Spechko (at Warner Electric),
Dr. S. W. Coates (at Mercury Marine) on the noise programs; Dr. W. L. Cahoon
(at Mercury Marine) and Dr. M. C. Ashe (at Kohler) on the engine gas flow studies.
459
-------
iA«t A'nV IVI'F
~\\
U
5 UA 1 A T"
1.
F»ir i IH.N ^ AC rims
i"i
HP-
8.'
SK
it"
101
''
Kt r.ifl |U"b
(.'i- ''US 1 lu1^ P
K'l>-tH
n H p K .1 L ^
•'! 29|b 22!,! i'!
Trt > fcL" "i
1 JJ .1
•J . ' 1 f
bj.t J
hh» 7
« , ' '
UJ j J
1 >G^ 7
1 ? 1 1 '* ')
1JJJ J
MOD 7
I7JJ 1
'1'!" ;
AL i"iA*t iiiii,E
LI." 4-1
- Tn
LMP
A F 1
1 1
H b
L!L*b
LA
3. , vl .'
L"
(,.'. .1
Lt.
9 'i.'.
LT/
Ol'.iLI
A.iSPi
A . .
Hi F
I'I.HL 1
.'ii.r>
ii
Ib
ij TIM ,5Pm»iu
i 74,i;.j
IB 40
J.b"
T H T K 4 u
3. /in
'lP.(T..lll FfilKAl) FNhHAL)
L"
H b , .J ' >
Ht:
122,1'.'
I1 ll 1 .1
,.i.i 4b. 'n
JA 1 JA1K
2 4., IIP <;4.Hil
UrJl 1>H1H
24. ,1.1 24,^.1
»C1 UC1«
24.'l,i 24,^..
U')l DU1><
24. .1,1 24, /I.
KIM TLF _ Lb t>!>1
tip
FFL
.I1./I4
I''
15.
A W ft M t
OSI-C
1* Ll
74 f.
/7 l'.
1 •) T A K
L I N
o9 1
02.9
DH , f
U2.B
J J ^
3-1,7
D ^ 4
3 1 y
bjl?
31.3
b3.2
b3. 1
S3. 1
33.4
UH
27J.
Ft- A
ll . ' • <•
J4M.
•'•>''
ILRb
4b 9
E Ho
A«T
5-O
b).2
49 2
b4.J
54^
b2 ^
b2i2
bJ.7
bJ.°
b4^ 1
54,3
TuTAL
ii
.Hi 4().u^
FFI FFT
0 w.l"i L6M L7H
S2?.Hki >» -J- 1222. «tf IJ22..HU
o3^ U4rf -^*, 060 Li/fc1 U7H
75.1H) V^.llb f*c 'J4.5D 24. SH 24. Ski
I/ 4 2 n v'« H
24. IM Ik2.«k)
UH2M VHH
24. HIT 2"?,^.' /.
UC2H VCU XC«
24.DH 218.21 8H.H>
U02H VOB XDB
24. UH 331.3k- 122. «H
Db2 US2H Vb'U XbB
4j.k)ki 4^,^11^ /2k1kl.lilk!l J^kl.Mf*
AIUF
11. Ho
MtP FMLP UN CL IE SE P1RAP PHFL PfAX T««L 3CAV
KPA PSI KPA
38.3 8.4 57. 0 kl.7? H.56 11,78 k),87 2,2 6,4 55,2 565, 43,
38,3 .B,4 57.6 ».8t) H.58 M.72 M.9B 2.2 6.2 57.4 565, 47,
TOTAL NUISE SPECTKUM UB
LlN
71.1
71.2
74^4
ort.b
55.6
b3l8
b4.3
b3,3
b.5.4
*) J , J
bJ.1
b3.2
b3.o
EXHAUjT NOlSt I)H (IVFHALL
1 1-17
Ll'-
1 a •}
AW 1
b4.5
62.4
M.3
74 b
06,3
54,5
Sb.8
53,8
54,6
b3.B
b 4 . i
54.2
'J4.4
b4.«
bb.n
•-UlSt LEVtL OB
4*1
t -\ ^
I I .1
O".?
7~> . I
/?,7
78,2
IE St P1KAP PHEL PHAX 1»AL SCAV
HPn I'n^EH uSFC BrltP I"LP PU"P«[P F«tP UH Ct
ll.IP «.. LH «t PSI hPA PSI KPA PSI KPA PSI KPA
7HHM. 'Jt \ lo.l Ki./9 '/.4b 92,3 030. J l«b,4 726.8 b'.B 4U.1 7,3,5Bj4 «,7H f.blj 0.75 U.91 2,1 B.I 55.1 54B. 4B,
•- ' 7.1.2 |!!77 .'!47 11'2,8 7k'8.b lib.7 798.1 5.6 38.9 7.3 5J.4 K.85 f.59 U.69 H.93 2,j b,9 5IC.S* »4B, 44.
T« flf*
1 1 h ,
?JJ,
J '•> •' ,
*1M J '
7. •'!
h 1 n
•JJJ.
1 1 t"i|
14/1*
I Mf/l
1 " JJ,
1 '?•' ,
1 00*1 ,
»L l'ilA«
Ll •
7 b . /
r)Mp
1 . P.I . fl
HZ
7
J
7
1
7
J
7
J
7
J
7
J
t ..Jlbt
ft-!
7J. 1
Kh
* "• L'
1 h . 4 .1 .
1 M
Ll
03
39
0.1
0 J
03
ul
3?
ol
o .'
D .
3 J
39
39
l-t
•Jb>
rt«S
AKE n
^ 4 w
6 52
7 b3
9 53
? oJ
4 0.'
9 02
1 0?
9 01
4 n?
7 ii?
H 0 1
7 Ol
o.)
H Ii.1
J '!•'
,' 0''
6 - 01
TnT
1.
/
c
Kll
)
4
7
t
7
4
1
'1
2
7
4
3
2
L Fx
!••
M.f,
h'lt
•"<1
F X'l
"LI
01
07
05
07
77
ob
bl
b2
b ft
b3
4 S
d r1
Jt.
d >l
22
Jl
MAUST
A,
7b
p
hP4
AUST OH
N A«T
.8 44.fr
,1 b7.2
.9 b9,4
.6 63.?
. 1 74,1
,1 oJ.J
.2 bd. 1
.0 b2.2
,4 50.4
.2 5J.4
.1 •l-'.S
.7 41.2
> 37.6
fi J h , S
.7 41, B
,7 23.9
.1 J2.4
nulbt Oil
1
, 1
I«tP
PSI KPA
i. 1U.2 711
TOTAL
LIN
71,
68 ,
66.
7-1.
77.
67.
63.
62.
r>3.
r.2.
6 1 ,
61 ,
61 .
'b9.
b9.
59,
nvEi'AL
LI«
01,4
Pu
PSI
.8 T.
H
'£
7
g
9
3
0
4
4
3
3
^
M
^
D
3
M
6
L
M
7
USE SHFCTnUH Oh
A«l
53. «•
5B , 8
6.1.3
66. b
74,3
6b,e
62.3
62,1
03', 4
' 62. b
61.3
61.3
0" , 7
6t , 5
6^,4
61,2
61,1
'JUlSF LEVEL f>a
A»l
7/.2
r'"EP M'tP DM
KPA PSI KPA
39.6 6.3 43.2 n.82
Ct
St
JIRAP PKEL P"AI 1>.A
1.5 4.6 49.ll 4b1. 36.
460
-------
30rO-9Q
Q.
XI
20
1
Js.
210
0-90
or
Q
065
0-70
0
CQ
C
O
"o.
e
C/l
c
o
- o
.0-90
0-800-85
0750-80
.0-75
^000
5000 6000
engine speed - rev. /mm.
7000
8000
Fig_Z
-------
overall noise
75
o
CD
T)
70
Q;
a;
in
o
c
65
60
mike at 150 m
5000
6000 7000
engine speed rev/rr.in.
8000
EigJL
-------
80
D\pverall noise
75
m
70
CD
I/)
O
c
65
60
mike at 15'0 m
5000
6000 7000
engine speed - rev/min
8000
463
-------
70
65
60[
55|
50!
T3
cu
•5
c
35|
engine speed ; 7000 rev./min
mike at 15-0 m .
30
25l
ro p c^ o
CO
o
n <±>
D m
n o
CD *-
m
CO
CM
p
o
O
rp
tb
T—
LT)
to
o
cb
in
to
U3
CO
10
15
17
Harmonics
454
-------
MICROPHONE POSITIONS
RPATHI RPATHE INTAKE dBA EXHAUST dBA OVERALL NOISE dBA
(a)
(b)
(c)
(d)
(e)
(f)
0.25
0.5
1.0
2.0
7.5
15.0
0.75
1.0
1.5
2.5
7.5
15.0
108.6
102.6
96.6
90.6
79.1
73.1
101.1
98.6
95.1
90.6
81.1
75.1
109.3
104.1
98.9
93.6
83.2
77.2
FIG. 6 - MICROPHONE PLACED NEARER TO INTAKE SOURCE
MICROPHONE POSITIONS
RPATHI RPATHE INTAKE dBA EXHAUST dBA OVERALL NOISE dBA
(a)
(b)
(c)
(d)
(e)
(f)
0.75
1.0
1.5
2.5
7.5
15.0
0.25
0.5
1.0
2.0
7.5
15.0
99.1
96.6
93.1
88.6
79.1
73.1
110.6
104.6
98.6
92.6
81.1
75.1
110.9
105.3
99.7
94.1
83.2
77.2
FIG. 7 - MICROPHONE PLACED NEARER TO EXHAUST SOURCE
465
-------
100
100
o
00
T)
90
o
c
80
intake
70
engine speed '. 7000 rev./min,
J
0-0
5-0 10-0
microphone distance
- m
466
15-0
EigJL
-------
85 r
cc
Q
I
2- I O
5 -2 CD
TJ
I
80
'0-85r ~a>
>
28r
26
22
20
0-80-
075
Q70
-O65-
75
70
overall noise.
mike at 7-5 / 7-5m
engine speed : 7000 rev/min.
bsfc
0-85
g
"o.
£
-
0-80
o
Q_
LO
CD
0-75
Fig. 9.
16
18 - 20 22
silencer outlet tube diameters DDI - DD2RJmm
-------
80
CD
TD
70
Qj
3!
o
60
50
intake box 2 - LIN
Intake noise levels - dLIN dBa.
!"standard)Box 1
Box 2
81 •?
89-4
791
79-7
intake Box 1 - AWT
intake\\
Box 2
AWT \
mike at 7-5 m
engine speed 7000 rev./mm.
Frequency- Hz
8
in
p
o
to
o
fi
CO
rn
oo
CN
p
o
o
(T)
LO
B
5 10
Harmonics
15
17
468
Fin
-------
APPENDIX
Report No.1096 of The Queen's University of Belfast
on a Computer Program for the Prediction of Noise and
Performance Characteristics of a Two-Cycle Engine
by
Professor G. P. Blair
469
-------
r
A Computer Program for the
Prediction of Noise and Performance
Characteristics of a Two-Cycle Engine
by
Professor G. P. Blair
Report No. 1096
L J
Summary
This report contains a description of the data sheets for the
use of a computer program called "THROUGHFLOW" which predicts
the performance characteristics of power, torque, fuel
consumption, air flow, etc., as well as the separate intake
and exhaust noise spectra and their overall separate and
combined noise levels. A brief description of the input and output
data is included, as is reference material for further study and
as background material and as experimental proof of the accuracy
of the prediction method.
Ashby Institute, Stranmillis Road, Belfast BT9 5AH Telephone /15133 Telex 74487
661111
47U
-------
'THROUGHFLOW - a computer program to predict the performance
and noise characteristics of a crankcase compression two-stroke
cycle engine
Research work at The Queen's University of Belfast over the period
1964 to the present day has been aimed at understanding the unsteady gas
flow behaviour of all types of engines, two and four-stroke cycle, Diesel
or spark ignition, supercharged or naturally aspirated, with reciprocating
or rotary piston mechanisms.
Recent work published by Blair and Cahoon (1), Blair and Ashe (2) and
Blair (3) shows how this research work has moved with a natural progression
from prediction of gas flow through the engine to direct evaluation of the
engine's performance characteristics of power, torque and specific fuel
consumption. Related work by Blair and Coates (<4) and (5) described the
method of evaluating gas-borne noise created by pulsating pipe systems and
this has now been incorporated with the above-mentioned prediction computer
program to give noise characteristics for the intake and exhaust systems
or their combined effect.
The data sheets which follow this section detail the geometrical
details of the naturally aspirated, gasoline burning, crankcase compression,
spark ignition two-stroke cycle engines which can be analysed with this
program. There are several variations of intake and exhaust systems which
can be handled, and for the several types of induction system such as piston,
reed and disc valve control.
The main types of engine handled are
(a) exhaust tuned units (motorcycles and snowmobiles)
(b) non-exhaust tuned engines (industrials, chainsaws, lawnmowers)
(c) the 'in-between1 units or part exhaust tuned (outboards)
The signature of the programs applying mainly to units typified in
(a) are:-
GPB2 and GPB6.
471
-------
The signature of the programs applying mainly to units typified in (b) and
(c) are:-
GPB1, GPB3 and GPB5
The middle initial P refers to the program indexing a "piston-ported"
induction process, with the data oriented in sequence to suit that program.
Middle initials R and D refer to "reed-valve" and "disc-valve" induction
characteristics. In other words program GPB1 refers to a piston-ported
industrial engine with a single exhaust and a single intake box silencer (see
data sheet later) and programsGRB1 and GOBI would calculate the alternate
noise and performance characteristics for the same systems but for 'reed'
and 'disc1 valved units.
The numeric symbol 1-6 defines the type of exhaust system attached to
the engine, all units having a single "box and tube" intake silencer. To
illustrate this, apart from examining the sketches in the data sheets which
follow -
Program GPBJ^ has a single box/tube exhaust silencer, without a tuned exhaust
system.
Program GPB2_ has a set of four box/tube exhaust silencers, with a tuned
system.
Program GPBJ3^ has two box/tube silencers, without a tuned exhaust system.
Program GPB_5^ has two box/tube silencers with one tube perforated, and without
a tuned exhaust system.
Program GPB6_ has a single perforated tube silencer and a tuned exhaust pipe
system.
The following page, Fig.A, is a reproduction of an actual computer output
for program GPB1 - a piston-ported induction unit, actually of the chainsaw type.
The first half of the 'output' from the program is the "input" data as
specified in the data sheets which follow and in the exact order of the data
listed in that section. In other words from BORE to ATOF (cylinder bore, mm
to air to fuel ratio) is -the data listing for the engine. The units are metric
472
-------
(SI) and linear dimensions are mm, with exhaust temperature (TWAL) listed
as C.
The second half of t'he output is the result of the calculations for the
first six cycles of the engine running on the computer as a 'paper engine',
with the fifth and sixth cycle calculations printed out for power BHP, brake
specific fuel consumption BSFC, etc., at the input value of engine speed, RPM.
The noise calculations, spectrum or overall values are for the last (sixth)
cycle only.
The pressure-crankshaft angle pictures are also drawn by the compute!
graph plotter for the last (sixth) cycle calculation, see Fig.B, and an
explanation of the relevance of the particular graphs is written on that
figure.
The output contains symbols defined below:
RPM: engine speed rev/min (also an input data value)
POWER: engine power as
BHP - based on brake horsepower (7A6W)
or KW kilowatts, kW
BSFC: brake specific fuel consumption as
LB - Ib/hp hr
or KG - kg/kW h
BMEP: brake mean effective pressure as
PSI - lb/in2
or KPA - kPa
IMEP: indicated mean effective pressure as
PSI - lb/in2
or KPA - kPa
PUMPMEP: crankcase pumping mean effective pressure as
PSI - lb/in2
or KPA - kPa
473
-------
FMEP: friction mean effective pressure as
PSI - lb/in2
or KPA - kPa
DR: delivery ratio defined as
mass air flow induced per cycle
mass of engine's swept volume at STP
' where STP is "standard temperature (20 C) and pressure
(760 mm Hg or 101.326 kPa)"
CE charging efficiency defined as
mass of air trapped per cycle
mass of engine s swept volume at STP
TE: trapping efficiency defined as
mass of air trapped per cycle
mass of air induced per cycle
SE: scavenging efficiency defined as
mass of air trapped per cycle
total mass trapped per cycle
(also can be seen as 'trapped charge purity')
PTRAP: trapping pressure, or pressure at exhaust port closure in
units of atm.
PREL: release pressure, or pres'sure at exhaust port opening in
units of atm.
PMAX: maximum cylinder pressure during combustion in units of atm.
TWAL: also an input value, exhaust temperature, C.
SCAV: SCAVDEG, the number of degrees of "perfect1 scavenging after
transfer port opening. For'a fuller explanation see reference (2).
The next section of output deals with the noise output analysed over
the last (sixth) cycle of calculation. The first part shows the noise spectrum
for the first to the nth harmonic up to a maximum of frequency of 2000 Hz
applied to the intake system and the exhaust system at their respective
distance (RPATHI and RPATHE) from the 'microphone'. Also shown is the total
or overall noise spectra, the combined noise spectra of the intakevand exhaust
474
-------
system. The values are in dB and are analysed as LIN (overall sound pressure
level in dB) or as AWT (weighted according to the A-weighting scale factors
in dBA).
The last line of the output shows the summation of all of these spectra
to give the total intake noise (LIN and AWT), the total exhaust noise (LIN
and AWT), and the combined noise for both noise sources (LIN and AWT).
The graphical output in Fig.B shows the pressure-time histories in two
sets, for reasons of clarity.
Set I: at the top of the picture are the crankcase and inlet port pressures
(in atm.) with the horizontal line being atmospheric pressure
(1.0 atm.).
Set II: at the bottom of the picture are the cylinder, exhaust port and the
(middle of) transfer duct pressures (in atm.) with the horizontal
line being atmospheric pressure (1.0 atm.).
The x-axis of the pictures run from TDC to TDC (on the sixth cycle) or
'360° crankshaft where BDC at 180° is the centre of the picture. TDC and BDC
refer to top-dead-centre and bottom-dead-centre piston positions respectively
The vertical lines drawn on the diagram; apart from TDC, BDC and TDC are 10
and 1C (inlet port opening and closing), TO and TC (transfer port opening and
closing), and EO and EC (exhaust port opening and closing).
475
-------
ENGINE MAKt AND TYPE STjTVty_ C?°)O
TWO-STROKE PP INDUSTRIAL ENGINE HITH ONE EXP. BOX INTAKE AND ONE EXP, BOX EXHAUST SlLEnCEh
P P UG K A M GPbl
BORE
6 6 , U U
STROKE
4 H . Vi a
CHL
73.dll
RPM EXHSOPfcN TRANSilPfcN
8H0H.Hk! 1^6,111] Il9.«)tt
ENuPEN
TRAPCR
EXHAUST PORT DATA EXPNU EXHSPRTWID
TRANSFER PORT DATA TRANSPNO TRANSPRT«ID
2 , kH! 4 4 , kl k,'
EXTPAO EXBRAD EXHSPRTHTMAX
b.urt b, n H u. w iJ
TRTRAD TR.BHAD
INLET PORT DATA ENPMO ENPRTWID
l.iHI 42. Hi)
ENTHAO
E N B R A D ENPRTHTMAX
5.Hw 13,^'U
TRANSFER DUCT DATA
bbu.au
EXHAUST PIPE LENGTHS LI
L8
35.UH
CRANHCR
I,b4
L6
35. P*
L/
1 H (* . '<) i
DU
26.
DAI
EXHAUST PIPE DIAMETERS
EXH. BOX A DATA _ LA
INT BOX S DATA THROTTLE US
FRICTION FACTORS .FFE FFA
MIKE POSITIONS
COMBUSTION PARAMETERS SPARK BURNHEAT
pi
DA1R
DS1 f
35. b«) :
DIR
26. bn
OA2
26. bH
1S\K
DA2R
26.b;i
DS2
35.bk)
VA8
DS2H
XAlJ
7b.kU<
22.UH
XSb
5 k' . U U
FFI
INTAKE EXHAUST
0,762 «.7b2
FFT
FFS
SURNDEG
AT Of
RPM
POHER bSFC bMEP IMEP PUMPMEP
BHP Kk. LH KG PSI KPA PSI KPA PSI KpA
1H.2 7.6 kl.bl 0,5.) 60.b 417.1 7H.H 4RB.2 4.7 32.7
1H.3 7.7 a.79 H,4« 61.2 421.7 71.b 492.7 4.7 32.0
SE PIRAP PnEL PMAX
FM£P DR CE TE
Pbl KPA
5.0 38.4 tf.61 H.3B 10.62 M.69 1.3 3.2 32.b
5.6 38.4 Id.6^ 1-1.37 t>. b 2 *>. ft y 1.3 3.1 3 i . 7
SCAV
i ENTR FSEQ
133.
26b.
4 /, H.
533.
666,
8 k^ H
933,
1H66.
1333|
1 6 'ikl
1733^
1666,
2f1UH.
HZ
3
7
0
3
7
v\
3
7
M
3
7
it
3
7
H
TOTAL INTAKE NOI
i J *i
119.2
HPH PQ*t
BHP
75^'. \»'.\
A*T
lib. 4
INTAKE ou
LIN Ak«T
1V13.
10B.
114.
111.
114.
99.
yb.
iwl .
94.
8«.
as!
84^
8 4
73.
SE l)b
g
R
7
3
3
^
2
8
5
b
^
^
9
3
1
B8
IvHI
1-J9
11?
97
94
H'l
9 4
69
89
91
85
85
74
.2
v)
.2
.7
. 1
.8
.8
,4
,M
p
',4
|9
!&
TOTAL
1
ER BSFC
Krt
7.b
/.b
LB
Kt
}
LIN
15.
b
PSI
63
EXHAUST OB
LIN AWT
102.
1U">.
1
1
1
1
98,
«J9.
i!6.
12.
tft .
97.
91.
9H|
98.
97.
'98.
96.
EXHAUST
7 1
M£p.
AWT
13.
1
4
9
2
1
'.•1
8
1
2
4
3
I
3
b
86
91
93
lt>5
lw3
1 1 «^
• im
97
91
8H
99
- 99
98
99
97
•
•
•
•
•
•
•
f
•
•
NOISE
9
KPA
.b 43d. 1
!'IB3 '.';-i.- ^3.7 439.1
IM
PSI
73.
74.
E
b
a
5
6
3
5
9
9
9
2
5
5
1
2
1
5
9
DB
P
K
b
b
TOTAL NOISE SPECTRUM DH
LIN A H T
' '.5
6.1
109.4
1 14.9 1U9.3
113.4 HJ9.H
114.9 1 12.7
112.3 111.1
1W2.7 1U2.3
1^3.1 lk)3,2
96.b
89.6
99.5
99.8
98.4
99.0
96.2
89. I
98.8
99.tf
97.3
98.4
96.6
9H.t
OVERALL NOISE LEVEL
LIN AM
12H.8 117.7
PSI
5.1
5.1
KPA
Pbl KPA
b,2 3o. u
b.? Jo.,-
D«
CE
.6b H.39
..hfS --..vi
IE.
SE PTHAP PM(• L
'.7i
1.3
1 . s
3.2 33.J
4SU. 24.
SCAV
"24.
-------
477
-------
REFERENCES
1. G. P. Blair and W. L. Gaboon, "A More Complete Analysis of
Unsteady Gas Flow Through a High-Specific-Output Two-Cycle
Engine", SAE Transactions Vol.81, 1972, SAE 720156.
2. G. P. Blair and M. C. Ashe, "The Unsteady Gas Exchange
Characteristics of a Two-Cycle Engine", SAE Off-Highway
Vehicle Meeting, Milwaukee, Sept. 1976, SAE 760644.
3. G. P. Blair, "Prediction of Two-Cycle Engine Performance
Characteristics", SAE Off-Highway Vehicle Meeting, Milwaukee,
Sept. 1976, SAE 76064S
4. G. P. Blair and S. W. Coates, "Noise Produced by Unsteady
Exhaust Efflux from an Internal Combustion Engine", SAE
Transactions Vol.82, 1973, SAE 730160.
5. S. W. Coates and G. P. Blair, "Further Studies of Noise
Characteristics of Internal Combustion Engine Exhaust Systems",
SAE FCIM Meeting, Milwaukee, Sept. 1974, SAE 740713.
478
-------
Data Sheet for "Throughflow" - a complete scavenging,
induction and exhaust analysis of a crankcase
compression two-stroke cycle engine.
Professor G. P. Blair
ENGINE NAME
ENGINE TYPE
NO. OF CYLINDERS
INDUCTION SYSTEM: (a) Piston ported
(b) Disc Valve
(c) Reed Valve
Dimension
1. Cylinder Bore, diameter
2. Cylinder Stroke, length
3. Connecting rod centres, length
4. Crankshaft speed
5. Exhaust port timing, at opening,
degrees ATDC
6. Transfer port timing at opening,
degrees ATDC
7. Inlet port opening, degrees BTDC
8. Cylinder trapped compression ratio
9. Crankcase geometric compression
ratio, including transfer duct
volume
OR Crankcase clearance volume
including volume of all
transfer ducts
Symbol
BORE
STROKE
CRL
RPM
EXHSOPEN
TRANSOPEN
ENOPEN
TRAPCR
CRANKCR
CRANKCVOL
Units
mm
mm
mm
Rev/min
degrees
degrees
degrees
Cm3
Data Value
TABLE I
See Figs.l and 2, for further details
479
-------
..volume trapped
Fig.l Crankshaft position shown at exhaust closing position, the trapping
position, usually EXHSOPEN deg BTDC.
TRAPCR =
VOLUME TRAPPED
CLEARANCE VOLUME of COMBUSTION
CHAMBER WITH PISTON at TDC
i
o
-J
-CRANKCVOL, cm3
Fig.2 Crankshaft position shown at bottom dead centre, B.D.C. - note all
transfer ducts are open, and the volume under the piston is then
the crankcase clearance volume, measured in Cm3. If SV is the swept
volume per cylinder, Cm^ then -
CRANKCR =
SV + CRANKCVOL
CRANKCVOL
480
-------
TABLE
Dimension
Symbol
Units
Data
Values
10. number of exhaust ports
11. maximum effective width of each
exhaust port
12. corner radius on top edge of
each exhaust port
13. corner radius on bottom edge
of each exhaust port
14. maximum height of exhaust port
i.e. not extended to piston
BDC position
EXPNO
EX11SPRTWID
EXTRAD
EXBRAD
EXHSPRTHTMAX
mm
mm
mm
mm
Note: A data value for EXHSPRTHTMAX of 0.0 in the program indicates that
the exhaust port height extends to BDC.
Fig.3 Plan section on exhaust ports
Fig.4 Elevation on an exhaust port
481
-------
Dimension
15. Number of transfer ports
16. Total effective transfer port
width (usually 2(a + b + c) )
OR WIDTH (a)
WIDTH (b)
WIDTH (c)
16a. Port elevation angles
17. Corner radius on upper edge
of transfer port
18. Corner radius on lower edge
of transfer port
Symbol
TRANSPNO
TRANSPRTWID
a
b
c
6A
9B
9c
TRTRAD
TRBRAD
Units
mm
mm
nun
mm
degrees
degrees
degrees
mm
mm
Data Values
— — —
TRANSFER PORT
WIDTHS
Fig.6 Plan Section through transfer ports
PORT ELEVATION ANGLES
L8
A , port type A
'„, port type B
, port type C
deg
deg
deg
Fig.6 section, elevation, through port A, B, or C-
482
-------
TABLE
Dimension
19. number of inlet ports
20. maximum effective width of
each inlet port
21. corner radius on top edge of
each inlet port
22. corner radius on bottom edge
of end inlet port
23. maximum possible inlet port
height
24. carburettor flow diameter
25. inlet port down draught angle
wrt cylinder centre-line
26. length from piston face to the
position where tract area
equals carburettor flow area
27. length inlet tract where trace
area essentially equals
carburettor flow area
Symbol
ENPNO
ENPRTWID
ENTRAD
ENBRAD
ENPRTHTMAX
DIP
DOWN DRAFT
L6
L7
Units
mm
mm
mm
mm
mm
degrees
mm
mm
Data
Values
DOWNDRAFT
Fig.7 Section through inlet tract for piston-port
induction system.
483
-------
FOR PROGRAMS GD _, INDICATING THAT THE PROGRAM REFERS TO A TWO-
STROKE DISC VALVE (D) ENGINE.
Disc valve, or Rotary Valve induction
data values indicating the following: ENPRTHMAX, ENTRAD, ENBRAD, ENPRTWID
on R MEAN should be entered on Table 4 as the equivalent named data values
numbered 23, 21, 22, 20 and also data number 28 below.
ENPRTHTMAX
.ENTRAD andENBRAD
mean radius
RMEAN
Fig.8 elevation on face covered by rotary disc
disc
DIP
air
Fig.9 induction tract length/diameter characteristics
for disc valve engines.
Dimension
28
Mean radius of inlet port
for disc valve induction
Symbol
R MEAN
Units
mm
484
-------
Transfer Duct, length and entry areas
TABLE 5
Dimension
28. effective area to each transfer
duct at entry frcm crankcase
(see Fig. 5)
individual area arean area
A, D, L,
and (usually 2A + 2B •+• 2C) total area
29. centre line length of transfer
duct from crankcase entry to
cylinder exit (see Fig. 6)
Symbol
FTRDUCT
L8
Units
mm2
9
mm
mm
Data Values
Often individual transfer port and duct designs do not conform to the form or
type indicated here. Please sketch below if this is not the case:
EXHAUST GAS TEMPERATURE, TWAL °C
SHOULD INFORMATION BE AVAILABLE AS TO THE EXHAUST GAS TEMPERATURE,
°C, (OR '°F) TAKEN PREFERABLY IN BOX A FOR PROGRAMS GPB1, GPB3
AND GPB5, OR TAKEN BETWEEN D50 AND D60 FOR PROGRAMS GPB2 AND
GPB6 THEN IT WOULD BE HELPFUL TO THE PROGRAMMER TO LIST THEM
FOR EACH POTENTIAL CALCULATION SPEED (RPM) OR OTHER OPERATING
VARIABLE.
485
-------
Microphone Position
DS2
THROTTLE
(area ratio)
Program: G.P.B.1 EXHAUST AND INTAKE SILENCING BOX PARAMETERS REQUIRED FOR PROGRAM
-------
EXHAUST AND INTAKE SILENCER BOX DATA FOR PROGRAM CPB1
EXHAUST PIPE:
LENGTHS
LI
DIAMETERS
DO
Dl
D1R
mm
mm
mm
BOX A DATA:
LA DAI PAIR
mm mm
IKTAKE BOX S DATA:
Throttle LS
mm
(area ratio)
DA2 DA2R
mm mm mm
VAB XAB
n3
cm
MICROPHONE POSITIONS:
RPATHI
m
DS1 DS1R
DS2 DS2R VSB
XSB
CHI-
RP ATHE
m
(SPARK) IGNITION TIMING:
(ATOF) AIR TO FUEL RATIO:
BTDC
REENTRANT TUBE LENGTHS
'Lll
LAA
LSS
487
-------
•CO
00
Microphone Position
^
"^
\
VB3 VCB
XBB / • XC3 /XDB
VDB
DB1R p1r
-i~( T~
/•
LCD
1?
fit
f-1
DA!R t-LAA-
DAI
f
A '-1-
i~r
1 1
DA2'3A2R
It
-Li r
_,_.
or ;c I p-i
L/ »- 1 1 . O k^ 1
• ' f< J
/ — — *1 1
I -T-
n/^o -(^7"
t-J t* *. L* ^ Z n
IT
IDD2R
DD2
D60
BOX A BOX 6 BOX C BOX D
EXHAUST AND INTAKE SILENCING BOX PARAMETERS REQUIRED FOR PROGRAM GP32.
-------
EXHAUST AND INTAKE SILENCER BOX DATA FOR PROGRAM CPB2
EXHAUST PIPE:
LENGTHS
DIAMETERS
L10
L20
L30
L40
L50
L60
L70
mm
mm
mm
nra
mm
DO
DIO
D20
D30
D40
D50
D60
D70
D7R
mm
mm
mm
mm
mm
mm
mm
mm
nira
BOX A DATA:
LA DAI PAIR DA2 DA2R VAB
mm tran mm mm mm
XAB
BOX B DATA:
LB
DB1
DB1R
DB2
DB2R
VBB
XBB
nun
mm
mm
mm
mm
BOX C DATA:
LC DC1 DCIR DC2 DC2R VCB
XCR
mm
mm
mm
mm
mm
BOX D DATA:
LD
DD1
DD1R
DD2
DD2R
VDB
XDB
mm
INTAKE BOX S DATA:
Throttle LS
DS1
DS1R
DS2
DS2R
VSD
XSB
(.ire.i ratio)
MICROPHONE POSITIONS:
RPATHI
(SPARK) IGNITION TIMING:
(ATOF) AIR TO FUEL RATIO:
REENTRANT TUBE LENGTHS
LU LAA LBB
mm n™
Rl'ATHE
o
BTDC
LCC
LDD
LSS
mm
mm
mm
489
-------
Mcrophcre Position
DS2
THROTTLE
(area ratio)
Program: G.P.B. 3 EXHAUST AND INTAKE SILENCING BOX PARAMETERS REQUIRED TOR PROGRAM
-------
EXHAUST AND INTAKE SILENCER BOX DATA FOR PROGRAM GPR3
EXHAUST PIPE:
LENGTHS
LI
rnin
BOX A DATA:
LA DAI
DA1R
DIAMETERS
DO
Dl
D1R
mm
DA2
DA2R
VAB
XAB
mm
tnm
BOX B DATA:
LB DB1 DB1R DB2 DB2R VBB
mm
mm mm
INTAKE BOX S DATA:
Throttle LS
mm
DS1
(area ratio)
MICROPHONE POSITIONS:
RPATHI
m
DS1R DS2
mm
RPATHE
m
XBB
DS2R
cnr
VSB
mm
XSB
cnr
(SPARK) IGNITION TIMING:
(ATOF) AIR TO FUEL RATIO:
REENTRANT TUBE LENGTHS
Lll
'mm
LAA
mm
BTDC
LBB
LSS
mm
491
-------
Microphone Position
DA2
DA2R[^ 0-;. vA,<.°'°.'Vg'T';
VPB
EXHAUST AND INTAKE SILENCING BOX PARAMETERS REQUIRED FOR PROGRAM GPB 5-
-------
EXHAUST AND INTAKE SILENCER BOX DATA FOR PROGRAM GPB5
EXHAUST PIPE;
LENGTHS DIAMETERS
DO mm
LI mm Dl mm
DIR mm
BOX A DATA:
LA DAI PAIR DA2 DA2R VAB XAB
,3
mm mm mm mm mm cirr
BOX P DATA:
LP N holes 4)P VPB XPB
mm cm
INTAKE BOX S DATA:
Throttle LS DS1 DS1R DS2 DS2R VSB XSB
mm mm mm mm mm cm
(area ratio)
MICROPHONE POSITIONS
RPATHI RPATHE
(SPARK) IGNITION TIMING: °BTDC
(ATOF) AIR TO FUEL RATIO:
REENTRANT TUBE LENGTHS
Lll LAA LBB LSS
mm
493
-------
Microphone Position
THROTTLE
(area ratio)
DTP
030
DiO
N holes of
diameter
D50
EXHAUST AND INTAKE SILENCING BOX PARAMETERS REQUIRED FOR PROGRAM FILE GPB 6
(moto -cross motorcycle). FOR PROGRAM FILE GPB 7 (for road racing no intake silencer S)
-------
EXHAUST AND INTAKE SILENCER DATA FOR PROGRAMS
GPB6 AND GPB7
EXHAUST PIPE:
LENGTHS
DIAMETERS
L10
L20
L30
L40
L50
L60
L70
mm
nun
mm
mm
mm
DO
D10
D20
D30
D40
D50
D60
D70
mm
mm
mm
mm
mm
PERFORATED PIPE AND BOX DATA:
LP
VPB
XPB
No. Holes
PHI
mm
INTAKE BOX S DATA:
Throttle
LS
DS1
mm
TAIL PIPE DATA:
LTP
MICROPHONE POSITIONS
RPATHI
mm
mm
DS1R
PHI
VPB
XPB
mm
DTP
mm
RPATHE
(SPARK) IGNITION TIMING
(ATOF) AIR TO FUEL RATIO
BTDC
495
-------
PANEL DISCUSSION
Thursday Afternoon - October 13, 1977
The panel discussion was conducted in two parts as follows:
Part I: Panel members were asked to discuss specific
issues presented to them.
Panel Members
Dr. R. J. Alfredson - Monash Univ., Australia
Dwight Blaser - General Motors Tech. Center
Dr. A. Bramer - Nat'l Research Council, Canada
Peter Cheng - Stemco Mfg. Co.
Prof. P.O.A.L. Uavies - Univ. of Southampton
Larry Erikkson - Nelson Industries Inc.
Doug Rowley - Donaldson Co.
Dr. Andy Seybert - Univ. of Kentucky
Cecil Sparks - Southwest Research Instit.
Part II: EPA representatives from the office of noise
control and abatement and from enforcement,
ansv/ered questions from the floor
EPA Personnel
Dr. William Roper - Branch Chief, ONAC
Scott Edwards - ONAC
Charles Ma Hoy - ONAC
John Thomas - ONAC
Jim Kerr - Enforcement
Vic Petrolotti - Enforcement
497
-------
PART I - PANEL DISCUSSION
Ernie Oddo
For the past two and half days we've listened to experts in industry
and universities tell us about their work on various methodologies
being studied, developed and employed to predict the performance
of mufflers on various surface transportation vehicles. I believe
it's fair to say that most of this work has been done to aid in the
design of effective mufflers. All of us present, at this symposium
I am sure, have a real appreciation for the complexities involved
in dealing with the significant parameters which must be considered
in any muffler performance prediction technique. Bearing in mind
these complexities then., we would like to address the objectives
of the EPA muffler labeling contract and the specific areas in which
we need assistance from panel members and members of the audience.
To open .these discussions I'd like to call upon Dr. Bill Roper from
the EPA Office of Noise Abatement and Control who will elaborate
on these objectives.
Dr. Bill Roper
I would .like to go back and read over the four objectives that I
mentioned in my opening statement to this meeting which outlines the
specific objectives of the EPA general labeling program, which I think
is very applicable here this afternoon and applicable to this entire
symposium. The first objective is the provision for accurate and
understandable information to be provided to product purchasers and
users regarding the acoustical performance of designated products
so that a meaningful comparison could be made concerning the acoustical
performance of the product as part of the purchaser's use decision.
This objective I think, is a particularly important one with regard
to the subject of this symposium. The second objective is to provide
498
-------
accurate and understandable information on product noise emission
performance to consumers with minimal federal involvement. The
third objective is the promotion of public awareness and understanding
of environmental noise and associated terms and concept. And the
fourth objective is the encouragement of effective voluntary noise
reduction and noise labeling efforts on the part of product manu-
facturers and suppliers. With that quick review of the principal
objectives of the EPA general labeling program I would like to go
back and focus on objective one which dealt with providing the consumer
with information at the point of purchase-decision relative to the
acoustical performance of a product. Now, that doesn't necessarily
mean that a product would have to be quote "physically labeled".
Information could be provided to the consumer in a number of different
ways. It is essential to provide him with information on the acoustical
performance of a product at the time he makes the purchase decision.
We think this is an important concept. As consumers utilized the
acoustical information in their purchase decision it is felt that
such selective decisions will have an impact on the noiseiness of products
used in this country. It's a way of potentially getting noise reduction
resolved without any required federal regulatory standards on the
manufacturer of new products or aftermarket part replacement manu-
facturers. In looking at the problem from the aspect of a voluntary
standard, consider that the consumer, given the right information,
can make a voluntary decision on whether they want to buy a noisy
product or a quieter product. Without the acoustical performance
information however, he really can't make that decision. In a
general sense, that's one of the principal reasons that EPA is interested
in labeling vehicle exhaust systems and is collecting information at
this time for use as background data to eventually put into a format
for decision making within the agency.
I'd like to look back at what I consider two separate parts of the
labeling background study that would have to be developed in order
to have the necessary information to implement such a program. One
deals with the technical performance data relative to, in this case,
499
-------
MATRIX
Categories of engines vs. current best muffler assessment approaches
Huffier Assessment Lt. Heavy Auto- Motor- Snov/-
Approach Truck Truck mobile Buses cycles mobiles
Parametric Analysis
Acoustic Modeling
Engine Simulation
Standard Engine
Actual Engine
Other
500
-------
INSTRUCTIONS TO THE PANEL
Mease consider the following two questions for application across
all surface transportation vehicles or to a logical grouping of these
vehicles. (Light and heavy trucks, autos, busses, motorcycles,
snowmobiles, motorboats)
Also consider approaches that:
1. Do not use the engine, such as
A. 'Parametric approaches
B. Analytical techniques
C. Engine simulation
2. Use an engine, either
A. Standard engine
.B. Actual engine
QUESTIONS
1. Is there an existing bench test methodology that could be used
to test mufflers, which would give values that:
A. Could be added to the noise contribution of other
predominant sources on a vehicle, to accurately
predict the total vehicle noise level, or
B. Would characterize the performance of a replacement
muffler, compared to a vehicle's OEM muffler.
2. If not, can the panel make recommendations on the most promising
bench test candidates that would meet the objectives of question
one, and the stage of development of these tests.
501
-------
the exhaust system; the other deals with communication of that
information to the lay purchaser or user. I'd like to separate the
latter one from the discussion today and concentrate on the technical
performance aspect. A major part of that consideration is of course
the measurement methodology procedure through which you can collect
the required data. Selection of a methodology must be based on a
whole series of considerations. There are many trade-offs. To name
a few there's the accuracy of the procedures, the repeatibility, the
simplicity, the cost involved in both the operation, and the equip-
ment instrumentation. These trade-offs will directly impact whoever
is using the measurement procedures, as part of his design or production
process. For the past 3 days this symposium has focused on one type
of measurement methodology the bench test, to determine what was available,
problems that might be involved in utilizing what is available or more
basically, if such a measurement methodology was even available.
This methodology referred to is the use of bench testing for determining
exhaust system performance. I think from a labeling standpoint we
would be'looking at the muffler particularly, although I recognize
that many, or perhaps all of the people that have participated in the
symposium have stressed the importance of looking at the total system.
I think from a labeling standpoint the most important part of the
exhaust system is the muffler, although you'd have to consider the
total system in developing the information base to properly identify
or characterize the muffler. I think another element here is the fact
that in carrying out this program, conducting this symposium and
investigating what procedures are available for measuring exhaust system
noise we at EPA recognize that the industry and the people such as
yourselves, who have done research in this area over the years are
the experts in the field. You are the ones that know what can and
can't be done both from a theoretical and a practical standpoint and
we would like to benefit from the knowledge that you have and receive
recommendations from you based on the best information, that's available
on what you would recommend to EPA as far as any measurement methodology
for vehicle exhaust systems is concerned, flow, we get down to the real
502
-------
practical aspect of the task we at EPA have.to accomplish, which is
to look at specific exhaust systems and to determine the most practical,
available test procedures to use, to obtain representative acoustical
data. I have broken out here on the viewgraph the 7 major categories
of products that we are looking at at this time. I would like to focus
the attention of all those on the panel and the audience on these
7 categories and based on the information and reports that we've had
in the last three days I would like to challenge you to come up with
your best recommendation on how we might measure and characterize
muffler performance on these 7 categories of vehicles. Now, I recognize
that none of the presentations have specifically broken the products
out this way although I think there are possibilities here for combining
certain categories. I would be very interested in the comments that
might come forth on these particular applications. Now, I have gone
ahead and taken the liberty of using some of Larry Erikkson's breakout
of a general approach to muffler assessment and listed some of those
down the vertical axis here and I guess the question conies down to how
much of that matrix can we fill out? What's available today? And
perhaps if there are two or three procedures available for testing in
one category here, maybe v/e should talk about a ranking of which of
those three are best for use in that particular application. I think
as we move into that discussion, since you are the experts in the field,
you can also interject your concerns for the other elements of measure-
ment methodology that have to be considered at some point such as
simplicity, cost, accuracy and similar things. The EPA program, from
a time standpoint calls for our contractor McDonnell Douglas Astronautics
to pull together and present to us in approximately one month, the
recommendations from this symposium, along with their own views on
this question. These recommendations will be used by the EPA to
make decisions on a procedure or procedures to be used in our testing
program for measuring exhaust system muffler performance; a procedure
other than for measuring total vehicle sound level. Our contractor
will be conducting tests using both total vehicle and whatever other
bench test procedure we have selected, starting the first part of
next year. I have briefly summarized the program schedule that we're
working under and the purpose and objectives of this symposium.
503
-------
Our primary objective in this symposium is to come up with the best
available test procedures to be used in assessing a muffler exhaust
system performance, other than by using total vehicle sound measure-
ment procedures. So with that challenge to the audience and the
panel I'd like to turn the session back over to Ernie Oddo.
Ernie Oddo
I'd like to amplify on one of Bill's statements. Currently in our
contract we are going to test vehicles in each one of these categories
that are on the board. He will also test, a minimum of three after-
market mufflers on each one o'c these vehicles, using the currently
most applicable total vehicle noise measuring procedure, such as
the SAE J-336 for trucks, for instance. Then we will take those
mufflers off the vehicle and test them using a"candidate" bench test
methodology. This is part of the test plan that is in the current
contract. Continuing then with the panel discussion I'd like to
flash on the board the questions that we gave to the panel at lunch
time to review. We'll.give the audience a chance to read the questions.
Then we "will flash a viewgraph on the screen showing a matrix of
transportation vehicles versus various muffler assessment approaches
we would like considered by the panel.
Cecil Sparks
Looks to me like it addresses itself to the evolution of the bench
test facility which will be used for actual predicted purposes, that
is to predict the sound level coming out of the thing which in essence
means we can then put a label on this muffler that >,ill define the
muffler, the exhaust system, the engine, the whole thing. In such
cases, it appears to me that your label's going to be bigger than
your muffler in the sense that if you consider all the possible
parametric variations involved you're going to include in the label,
including the testing facility, the wide variations and engine operating
conditions and the exhaust system, etc.. The approach inferred
then is one of predictive rather than just a bench test facility
that will say that this is a reasonable quality muffler and as such
will have to be a label of the system rather than the muffler itself
and while this kind of thing it seems to me is theoretically possible
504
-------
in that you can build a source simulator for any given engine to
cover a wide range of conditions put simulated exhaust systems,
etc. on it, seems like it would be much simpler just to test it on
the vehicle.
Ernie Oddo
I want to clarify one point there, by labeling we don't necessarily
mean physically sticking a label on a muffler. He have a much broader
description of labeling. Labeling could be just some identifying
numbers on the muffler similar to what is done today. The numbers
or letters would identify the manufacturer of a muffler which then
could be traced back to the manufacturer's catalog. The catalog which
most manufacturers currently issue would have all of the information
that you have discussed. This is just one alternative.
Cecil Sparks
The other alternative would be to categorize it in terms of the
inherent passive response characteristics of that particular
configuration but again you would need the same kind of information
we're talking about if your intent is merely to be able to predict
what the ultimate noise level at a given application will be rather
than say, okay this is a hospital type (stationary) muffler or something
like that.
Ernie Oddo
As an example, I would like to reiterate that which Doug Ralley from Donaldson
presented. That approach is similar to what we are talking about,
for trucks. In other words, Donaldson has all kinds of information
computerized on tab runs and in catalogs, which take into account
back pressure and all the other parameters that we discussed. Their
program considers the specific parameters such as engine back pressure,
pipe length, etc., and then indicates candidate mufflers, for that
application.
505
-------
Cecil Sparks
That was the point I was trying to make, do you do this with -a bench
test facility or do you use the actual installation?
Ernie Qddo
Okay, well that's the question we're posing here to the panel and to
the audience today, considering the broader definition of bench testing
which could be any of the categories up on the board.
Dr. Davies
I wanted to step back two steps first - I know Bill Roper said that
he wanted us to concentrate on technical performance data and that
communication of information concerned was of secondary importance,
well already we've seen you can't separate the tv/o, they're a combined
exercise. You can't really decide about the technical performance
data you're going to produce without taking the communication problem
with it so you can't divorce these. They're part of the same process
in the first place. The second point I'd like to make is that when
you come to a labeling procedure and we've heard the difficulties of
labeling muffler units on their own you really must look at the system
and all these other complications and that there isn't such a thing
as a good or a bad muffler, it just depends on how you use it. The
consumer and if you think of the consumer in a simple level, and
that's the housewife in her house, she has the same problem, she has
to buy a cooker and a dishwasher and various other things, and operat
these and get them to perform certain tasks, she makes a distinction,
she knows what she wants, and so I think that what you've really got
to do is to think of the two together, you've got to provide technical
information that's understandable. It can be complicated, I mean you
are going to look at the sales feature on some of this equipment, I
don't understand it. The housewife does. You don't get bugged up
on the technical problems too much, but you put the others on the
consumer to say, all right we've given you this information and it's
506
-------
up to you to make proper use; what he's got to be sure is that the
information isn't deliberately misleading. I think that's the first,
and secondly the information is sufficient for him to make a qualified
judgement. Well now, that's one part of the problem, the second part
of the problem _ that I'm horrified by this here table or matrix,
because it's quite clear, and I'd like to add another category to the
list, why we've got recreational vehicles there because they really
cause a lot of problems.
Ernie Oddo
They are not in our contract.
Dr. Davies
They are not in your contract? Then let's exclude these explicitly.
The second thing is that we have a very wide range of engine types and
I don't see how v/e can come to a simple and meaningful way, consumer
oriented way of describing the characteristics of these systems over
this big range. For two reasons, the guy is not going to be interested
if it isn't tailored to his requirements, lie's not goina to go through
five pages of data just to get the tv/o lines he's interested in.
So what you've got to do is to come up first with a clearly defined
classification system. It's not difficult, it's here, heavy trucks,
you might put light trucks and autos together, buses are a special
problem because buses are operated on the whole by corporations and
the corporations have the technical expertise to make decisions.
And then you've got the other problem, the snowmobile, the motorboat,
the motorcycle, the semi-recreational vehicle and also you've got the
ordinary driving car and also our washer or our cooker or whatever
we have at home, in our house. I think we have to produce a different
labeling system to suit the application and I think if you start in
that direction you- might make some progress.
507
-------
Ernie Oddo
First of all v;e do have recreational vehicles in our contract, motor-
cycles and motorboats are recreational but on your point I agree with
everything you said Professor Davies, concerning this matrix, we don't
in any way intend for muffler labeling information to be collected
in this format to be passed on to the consumer. He just present
this information in a matrix format for the panel's consideration;
as an easy way to keep in front of you all the various possibilities
that we would like you to consider. We realize, of course, that for
light trucks one or more assessment approaches could be used. For
heavy trucks or automobiles, the same thing holds true. The question
is, can we group the engine categories above and then use one of
these particular approaches to handle two or three or four of these
vehicle categories?
Prof. Davies
What I should have been clear in saying is that I think that as well
as this categorization you really ought to categorize the consumer or
the purchaser or whatever you'd like to call him. That after all
the fleet operator represents one category and he wants a different
sort of information than the individual operator or the private
individual. You might think again that you really have a different
labeling procedure for these three categories, because they are
different.
Ernie Oddo
That's true and that's why we try to separate the two issues - one
being the technical. We feel that once we have good technical
information obtained from a good bench test methodology, the trans-
mittal then of that data or information to the consumer, is another
problem, we recognize that. We are also open for suggestions on the
best way to transmit information to consumers, but I think the first
step has to be the technical question,, do we have a bench test
methodology that would give us good, valid, accurate data to do with
it what we want to do?
508
-------
10
Prof. Davies - It's this question of accuracy that bothers me. If
you'd left out that word I'd go along with everything you say. I
think you have to define what you mean by accurate. I think it's
got to be convincing. Convincing is a different overtone. The
consumer, the guy that's going to fit this to his car of tit this to
his truck has to convince himself that what he does has to comply
with regulations and he needs information that will convince him
that what he do.es is a sensible approach to solving the problem that
he's up against - regulations. That's what he wants. Accuracy really
doesn't come into it. He's going to depend on the certification
provided by the manufacturer.
Ernie Oddo
That's where we want to apply the word accuracy. Not really to the
consumer, we're really interested in the manufacturer guaranteeing that
his product when used on a certain vehicle is going to do what he
says it is going to do.
Dr. Brammer
I believe that the sort of question the consumer is probably going
to ask is something very simple such as, is this replacement muffler
equal to the one I have on my car or better, or is it worse, and if
these are the type of questions one wants to obtain answers for then
we're really talking about a relative measure of muffler performance;
we're not talking about an absolute measure and in terms of questions
that are posed here, this moves us more towards B than A perhaps and
also it enables us to, if we think about it, we can now start
running some form of test as yet undefined, in which we can replace
single components, compared with the original existing components,
and see the effect of them relative to the original muffler. I think
we have to think a little bit about the type of labeling
that will be used and the sort of questions that we
509
-------
11
want to answer otherwise I don't see how we're going to get started
on this particular problem but here's a notion that I think we could
usefully pursue. If one tries to answer questions of this type but
it does get away from a lot of these problems of predicting the noise
of the vehicle and things like this and the accuracy of the measurement
that was giving a lot of concern and rightfully so. If I had to rank
order one A or B of which I think is most important to the consumer
I think in terms of questions he's asking from a replacement piece
of equipment, I'd rank B above A at this point in time and let that
influence the choice of measurement technique that I would go for.
Ernie Oddo
Any comments?
Doug Rowley
Bill Roper laid down quite a stiff guideline for us and I think I'd
like to get them a little bit stiffer. Talk about this accuracy thing
and I'd like to ask, accuracy to do what? What are we really trving
to do? By that I mean what level are we trying to control overall
truck noise too? Then we can talk about whatever the exhaust system
has to do. Can you comment on that Bill? Can you follow the question.
In other words, somewhere along the line I'm trying to get someone from
EPA to tell me that you'd like to control the noise of the new 1978
trucks once they get in use, to some level. Then, when a fellow
starts looking for a replacement product he's got some guide lines.
Bill Roper
Okay, in response to your last question, you're right, the new medium
heavy truck standardsis one that applies to the date of manufacture
and we have an in-use standard for interstate motor carriers which
is 86 dBA for speed zones less than 35 mph, there is a gap so to speak
in the Federal program although not in some state programs, I understand,
as to the in-use level that would be applicable to the medium and heavy
truck, say that's manufactured at 83 dBA level beginning 1 January 78;
there is no Federal standards other than the 86 dBA pass by, now we
have under way right now a program at EPA developing the background
510
-------
12
information that will be necessary for revising the interstate motor-
carrier regulation with the intended purpose at this time of setting
that interstate motor carrier standard at a lower level which would
be equivalent for an in-use truck to the new truck standards. In other
words a truck that is manufactured to meet an 83 dBA newly manufactured
standard would then be required if operated by an interstate carrier
to meet some equivalent standard while in use. Now, it may be the same
level, it may be slightly different because there's a different measure-
ment methodology involved. But yes, we are addressing that now. In
regard to the labeling aspect I think we're talking about more than
just a label that identifies how close or how a product complies with
an existing standard because in some of the areas there may never
be Federal standards for those products. VJe're again focusing on the
information that describes the acoustical performance of that product
to the consumer so that he can consider noise as one of the elements
he thinks about in making that purchase decision. I guess I would
also want to talk about two different ways that you could look at
two different types of information that could be used for a basis for
labeling. One would be if you're comparing system A with system B
or system A with the original equipment, and that's such as you were
mentioning, a comparative type of information. The other would be
how does it compare with the total system or total vehicle performance;
in other words, given this exhaust system, how is it going to affect
total vehicle acoustic performance. There's really two different
approaches there from the EPA standpoint ;we are not locked into either
approach. We're looking for the one that makes the most sense. There
may be implications, depending on which kind of approach you take as
to what's available from a measurement standpoint,
to provide the tool to develop the data for labeling. That's one
element that I'd like to hear more comment on. Considering these
two general types of approach, to collect the necessary information
for labeling, which one has the necessary measurement tools commensurate
with it to provide the data, at this time?
511
-------
Cecil Sparks
You could go a couple of ways in that regard, again I think that if
you're trying to use a bench test facility or evolve one whereby you
can predict what this particular muffler will do on trucks X, Y and
Z, etc. you've got a pretty tough row to hoe. On the other hand if you
can evolve the system of labeling where you label the truck and the
muffler that this trick has, then, when a replacement muffler is used,
class G31 and 4X82 or something like this then in essence qualify your
mufflers for those various applications. Now that is something that
seems to me would be a practical approach. But again, perhaps you don't
neet a bench test facility to do this you could qualify the muffler then
as being original equipment or better. And then you put in your ov/ner's
manual which mufflers you can use, as possibilities.
Ernie Qddo
That would lock it into OEM only and how v/ould the replacement
manufacturer, for instance, comply.
Cecil Sparks
They'd just have to qualify their muffler for that application.
Ernie Oddo
Right, and that's what we're talking about. Qualify it how?
Cecil Sparks
On the vehicle.
Ernie Oddo
Okay, that's true, that is definitely one methodology that can be
employed and we know it will work if you test every one on the vehicle,
but we are looking for methodologies other than vehicle testing, to
supply performance data on mufflers.
512
-------
14
Cecil Sparks
The point is though* if you build a bench test facility whereby you're
able to predict this muffler's performance on this whole broad spectrum
of truck configurations you've got a horrendous job.
Ernie Oddo
That may be the -case.
Peter Cheng
I agree that while the best thing is to put a muffler on aach truck
model. We've got two problems here. First, even OEM truck manufacturers
cannot test the mufflers on every truck model. Say one particular
truck model, they may have 80 to 90 different combinations. Some of
them have a fan clutch some of them have different fans, some of them
have transmission boxes, etc. As to the second question, if
we are going to test the muffler on the truck who is going to do it?
Who's going to pick up the vehicle? There are so many aftermarket
truck muffler manufacturers. Do each one of them have the right to
ask OEM manufacturers to test mufflers on every one of the OEM truck
models?
Cecil Sparks
More people would have access to the trucks than they would have the
facility, I would think.
Peter Cheng
Well, from our experience it's very difficult to get a truck. Most
likely, we would like to test the muffler on the new truck because
the other noise sources were controlled when the truck is relatively
new. And usually, the dealers would not allow us to get the new
truck to test and another thing is that talking with some of the OEM
truck manufacturers when they want us to test some truck, especially
on back pressure, they would specify the truck must have gross vehicle
weight. We have to put say, a few thousand pounds at least on the
513
-------
15
truck and no local dealer or whoever would like to loan us a truck
by putting a few concrete blocks on it. I am looking at this problem
from the other aftermarket companies' point of view. We are also in the
OEM business and first of all, our experience again is limited to
heavy duty trucks. I don't know anything on snowmobiles, etc. The
heavy duty truck is differnt from the passenger car in one sense in
that the customer is more knowledgable than the general consumer.
It is a different type object. Second of all I am not saying that
they understand exactly what dBA is, etc. but at least everyone of
our distributors has a noise level meter they can somehow crank up
an engine and run some tests. And then, let me view the problem from
OEM market experience. I don't think there is 100% satisfactory
bench test method. Because of the pipe length, etc., but the SAE
test procedure mentioned by Mr. Larry Erickson this morning, I think
that's a good compromise between practicality and 100% accuracy.
And, we also have a lot of experience on judgement of whether the
muffler we sent out to our OEM customer will pass the drive-by test
or not. We have a very good idea if it will. We're just like Mr.
Doug Rowley said when he got 95% accuracy. I don't know whether I
would have 95% accuracy or 80% accuracy but I tend to agree with him
that there is some correlation between a bench test and drive-by test.
If we cannot get some kind of ball park feeling from our bench test
then the OEM truck muffler manufacturers simply would not be in the
business. We cannot send five mufflers for our customers to test and
for them to pick one. They are not going to do that. He send him
one sometimes at most two and we make our best judgement whether he
will test it or not, also, v/e do not send one muffler to one manufacturer.
We send a muffler to possibly a lot of manufacturers. And from our
experience if the muffler which we judge is a good muffler probably
will pass the test with a lot of our customers. On the other hand,
a bad muffler probably will not pass the test.
Ernie Oddo
Thank you very much, Peter
514
-------
16
Larry Eriksson
Well, I had a few general observations, a little bit over what Peter .
said and what Cecil said, looking at these vehicle categories one
observation I'd like to make most, is that many of these categories
are products that either have been or will shortly be governed by
some new product noise regulation by EPA. And I'd like to focus on
that a little bit. One of the observations I'd like to make is that
for those products I personally don't see the need for muffler labeling
for the new products and I think this is an important point. We're
talking about a truck or whatever that is already subject to a new
product regulation. I for one feel it would just add complexity to
also ask for a label on the particular muffler used on this piece of
OEM equipment. It's already meeting specification for the overall
vehicle. Accepting that.point then what that leaves is the aftermarket.
And, in terms of the aftermarket the only observation I can make is
if we are setting levels for overall vehicles, new products that are
as stringent and as accurately measured, etc. as we are for trucks,
buses, or what have you, it seems to me that any aftermarket evaluation
procedure measure ouqht to be at least comparable in accuracy. He
shouldn't give away an awful lot in terms of the aftermarket measurement
procedure. Essentially what we ought to be shooting for is something
that is more or less equivalent to OEM and the OEM unit that the OEM
equipment has. In the sense that we don't want to allow any degradation
of that product, that the EPA's proposed regs already have included
some aspects of not allowing any degradation. I frankly see the
requirement in the aftermarket ending up one way or another. Saying
in so many words it's going to be about like the OEM unit was, Accepting
that fact and the fact that you want an accurate test it seems to me
that you're going to be looking at an actual engine test of one sort
or another. Now, I agree with Peter, I think the SAE procedure that
we have worked up is probably not too bad a compromise, but whatever
you come up with I think it's going to have to be something very
similar to that in order to obtain the kind of accuracies to be
515
-------
17
consistent with the rest of the program. And, I have not seen
in any other presentations including my own that the other four
techniques listed here really provide accuracy that is at all comparable
to the rest of the noise program9 that is at all comparable to the type
of thing we can achieve in the SAE type procedure, with the type of
engine dynometer and real engine close to real system type of test,
Now, if you're still with me on that where that leaves me, is saying
that okay, we're going to do real engine testing, we're going to test
on something like and SAE test, but what about the multitude of
combinations. It's been stated, it seems to me if my observation is
correct, that there is no practical way to measure all combinations that
exist and so it strikes me that we're going to be in a situation
where some kind of certification that it meets is a preferable route
and then a test program would have to back up that certification.
The burden would be on the man who certifies it, to the muffler supplier
to have his engineering house in order sufficiently so he can certify
it and be reasonably confident that when he gets around to testing it
on an engine or when somebody else gets around to testing that particular
situation on an engine that within some tolerance it does in fact follow
what he said it would. So those are a bunch of observations which
are connected.
Ernie Oddo
Hith reference to Doug's comment before on the SAE procedure, on the
accuracy of that procedure, would you still consider the new SAE
procedure accurate enough for this purpose?
Larry Eriksson
I didn't really disagree that much with Doug, maybe it came out that
way I don't know. The procedure is a very good procedure. It's an
accurate procedure in a sense that certainly I think all of us in
the muffler end of things at least in this panel, are using, something
very similar to that procedure today in our muffler testing and it
certainly does correlate in an indirect sort of way with the kind of
measurements the vehicle manufacturer might be making. I guess I'm
516
-------
18
like Peter, I don't know what percent is exactly, the correlation, but
certainly we do supply units to our customers, and often times there
are no problems in terms of correlating our numbers with their numbers.
Occasionally, of course, there are, but there is certainly room for
improvement in that particular area. However, it seems to me it's
far-and-away, from a technical point of view, the best way to do it,
that we've found and usually, the correlation is quite satisfactory.
Ernie Oddo
Thank you, are there any comments?
Dr. Robin Alfredson
It seems to me that a fairly easy measurement to make in the laboratory
anyway is the measurement of transmission loss. And the question we
really have to work out is how good is transmission loss a measure of
performance on an actual vehicle. My guess is, and it's really only
a guess, that transmission loss is probably not too bad for the large
multi-cylinder engine situation. That's only an intuitive guess.
I believe in the single cylinder or two cylinder case
transmission loss is very unrealiable. I suppose on the average if
you're measuring transmission loss for a large multi-cylinder type
of vehicle that might give you an indication of the performance.
A little bit like having your feet in two buckets of water. Have
one foot in a bucket of water that's freezing cold and the other is
boiling hot, you can say on the average it's warm but it's hurting quite
a bit. I don't have any strong feelings, perhaps some of the manufacturers
might have. If you do have a good muffler, and I imagine that means
good in terms of transmission loss perhaps, can you be reasonably
certain on a large number of vehicles that on the whole it performs
well. My feeling is that probably with a larger multi-cylinder engine
that would be the case but certainly not with the smaller configurations.
517
-------
19
Doug Rowley
I'm not going to try to answer that question. I don't happen to agree
with that. I'd like to go back to Larry, Peter and Bill, obvtously
one of the reasons I wanted to know what your goals are is to establish
a point that we will be faced with replacing a product that is equivalent
to the original equipment and yesterday Bill Roper mentioned something
about replacing the exhaust pipe with an equivalent to the original
equipment. I think one thing we as part of the industry do not wish
to get into is placing a standard on the exhaust system. Really,
what we're trying to do is control overall truck noise, which, perhaps
exhaust noise is a very significant part. The question is, and it could
be a little bit ridiculous, are you going to put a standard on the
mechanical noise in the engine, intake noise, fan noise and etc. Well,
this is pretty much what I'm driving at, I do feel that if our catalogs
should say, as a guide to the user, that this is equivalent to the
original equipment, really to carry that on further, is there a need
for a specific type of evaluation method. Perhaps there is, but you're
coming up with an assessment. I could perhaps look at a product and
say well, yes based on a lot of experience that's going to be equivalent
to original equipment. Do you get what I'm driving at here Bill?
For instance, to meet the 83 dBA requirement we may have an exhaust
system that controls the exhaust noise to 80 dBA or in another case
we have to control the exhaust noise to 70 dBA. A vast difference
probably in the size, shape, weight and the cost of the exhaust
system. And really, when you get right down into the trucking business,
this is the name of the game. They just aet by with as little as they
can possibly use.
Bill Roper
I think in your comments you brought out one of the points I think
important. That is, knowing what the exhaust system will do on a
particular truck is vitally important to the person who is using that
truck. You mentioned the one case you sited. The one case might be
an 80 dBA muffler and the other case was a 70 dBA muffler to meet a
518
-------
20
particular desired level of total vehicle noise. So it's vital to
the user of those two trucks to know which muffler or which exhaust
system to apply. And I think that the general thrust of a labeling
program is just that. To provide to the purchaser of the product,
information that will allow him to evaluate the acoustical performance
of the product he is buying along with the other things; cost, or what-
ever. I don't know, I guess it's not true that in a general sense the
quieter muffler is always the most expensive, sometimes it isn't.
So he v/ould have the acoustical performance available along with other
information when he makes a decision. The other point you raised there,
is other components of the vehicle are important. As I recall, my opening
remarks pointed out a couple of things that are particular to the exhaust
system. That is, one, it's an important source of noise. Two, is that
it is replaced on a cyclic basis throughout the useful life of the
product so that it is something that a user later on in the life of
that vehicle will be replacing and if it is replaced with a system
that is acoustically louder it's just a louder source of noise in the
environment. Being in the noise control business we're concerned about
that, so it's for that reason too we are interested in coming up with
a way of defining the performance of replacement parts. Exhaust systems
fall into that particular category of a product that is in fact a
replacement part, to a total vehicle system.
Dwight Blaser
I think the one thing that baffles me a little bit on what seems to
be charged here of this three day symposium is that maybe it's the
next to the last line there on the screen, everything seerns to be
pointed toward characterizing the performance and we all seem to be
charged with which technique is the best to do that. In order to
decide which technique it seems to me, that first you have to define
which performance parameter are we going to use to characterize it.
Let's even limit it to the acoustic performance. I feel certain that
of all the bench tests, analytical techniques, all the on-vehicle
tests, they've all been carried out in a very systematic careful
manner, they're all relatively accurate for developing data which
519
-------
21
refers to a particular performance parameter. It looks to me like
what we really have to do is to define or decide which performance
parameter first then maybe we can back off and look at which technique,
if you wish, is the most accurate, to measure that parameter.
Ernie Oddo
Respectfully, I am asking you the question back again as a panel.
Taking into account that you've done a lot of research, a lot of
work in this area, and you're familiar with the important parameters,
which should or should not be included in any bench test methodology.
Can you eliminate as maybe less significant some of these parameters
to come up with a simpler bench test and still meet the objectives
here.
Dr. Davies
I don't really feel it's helpful to repeat what one's said but I think
a lot of things said in between on a remark I made earlier and a remark
I make now is along the same lines. The point is, if we're going to
get anywhere, that we've got to state some objectives very clearly
and this is what Doug Rowley said. We've heard about heavy trucks
mostly, in this discussion. That's only one part of the problem.
Now we know what the objectives are there. The operator has got a
tough job. To meet the noise requirement legislation. Because we
know the engine noise that's the carcass noise is so dominant, that's
one particular problem, and the methodology you want and the problem
that the muffler designer is facing is in one category. How if you
talk in terms o.f total environmental pollution, the private automobile,
the problem is quite different, that is an exhaust noise dominated
area, as far as the environment is concerned in general. That's very
much more difficult I think, the replacement problem, because there
are more replacements, that are going to happen in the life of the
auto. Secondly, the replacement's going to be made in a much more
arbitrary way. A private individual's going to put a replacement part
520
-------
22
on, that he can get cheapest and quickest to getting by, I mean that's
the answer for the average user. That's a different problem and if
you're going to try to come up with a methodology or a test procedure
I think you've got to look at each of these categories on the list and
in the matrix and say allright let's pick the parameter for that one
and place the methodology on that one and let's go on to the next one
and look at that and then you make progress.
Ernie Oddo
Good observation and if you'd li^ to continue that discussion --
Larry Eriksson
To carry on a little bit on what Dwight's comment was, which I think
I heartily agree with. It's very difficult I think to separate the
technical questions from questions of the objectives and what the EPA's
trying to accomplish, why they're undertaking this program in the first
place. I think you've got to get very specific about why this program
is being done. Specifically, what it's trying to respond to, what it
hopes to accomplish. I know with our own company there's one excellent
way to waste a lot of time and get a lot of wrong information and that
is one of the personnel in our company, whoever it might be, someone
from our sales group or engineering group walks over to some guy in
our research department and he asks some question of our research guy,
how do you do this? And unless he gets very specific about what he's
really going to do with that information and why he wants it in the
first place, chances are they're not going to talk the same language
at all, they're going to get a very strange answer. And the research
guy may be operating from a totally different point of view. I think
the only way v/e can work is you've got to have a person who's asking
the questions to give you all the background. What is he really looking
for? What is he trying to accomplish? And this has been lacking. I
have felt this is needed for us to have a better idea of exactly why
we're trying to do all this. Now, that's kind of a cop-out. Now part
2 is the SAE subcommittee to a certain extent answered that from their
521
-------
23
point of view. Their answer from their point of view was, v/e want the
sound pressure level produced by that exhaust system. Our subcommittee
had to deal with that, not from a government regulatory point of view
but from the point of view of a group of engineers trying to provide
some reasonable characterizations of exhaust systems so we had to answer
questions from that point of view. Regulatory agencies are something
else again. I have not heard that, from the EPA. He were looking
forward to the question session with EPA, because that was to be my
question.
Peter Cheng
I'm not trying to answer Larry's question to EPA for EPA but I imagine
one of the objectives in the muffler labeling proposal probably is
because there are many mufflers on the streets which are basically
tin cans. We can label mufflers in a very strict sense, put an A,
B, C, D on it or we can label the mufflers in a rather general in a
broad sense. That is, in the very first step the EPA would require
each aftermarket muffler manufacturer have a good test facility they
would have to know what they are doing. The EPA can somehow certify
their test or their test methodology. In addition, EPA would have to
to require the aftermarket muffler companies to report the test results
to their consumer. I personally believe that EPA should adopt these
two steps and then wait for awhile and then see whether there is indeed
a need to label the mufflers in a strict sense.
Dr. Seybert
We talked a lot about non technical things and perhaps I'm not quite
as familiar with the rest of the people in regard to some of these
questions. Robin Alfredson touched on something I don't think that we
have received a satisfactory answer for and that is, how can we use
a basic muffler descriptor such as transmission loss. Maybe not on
its own, but modified according to some particular configuration with
522
-------
24
exhaust pipe or tail pipe lengths and engine configuration, as a
descriptor. I don't think anybody has really demonstrated that this
cannot be done. If we do have a proper descriptor for each of the
subsystems of the overall exhaust systems. Certainly transmission loss
and insertion loss have a lack of correlation. Transmission loss with
more definitive information on the rest of the system may be an adequate
descriptor. We haven't proved that it isn't. That's one thing I v/ould
like to see pursued.
Prof. Davies
I've disagreed with Robin before so I'll disagree again. Can I refer
back on the three days past, to my original presentation
in which I pointed out that an outstanding problem, and this affects
the issue on technical accuracy, is that we don't really know how to
categorize the source and so we're really in the dark. You categorize
the source and you can then categorize the rest of the system. Fine,
if transmission loss is it. That's quite satisfactory, that's nice
as Charlie pointed out, it's invariable for a particular unit, that's
nice too, you can label it, as he said gold plate the label and shove
it on there. That's grand, vastly, but we're not in that position.
In fact I don't know that we ever will be because if you take the top
line operator it keeps these vehicles on the top line and all that
jazz then you're talking turkey. If you're taking the average user
and particularly, and we haven't talked about cars much in this
discussion, the average driver of a family car, he's not .going to
keep that in the shape that all the accurate measurements and every-
thing else are made in. And so, talking about one or 2 dB or high
accuracy or whatever is meaningless, it doesn't mean anything. Because
the source is not going to be anything like the OEM source the vehicle
was when the vehicle was categorized. It's going to be different. I
think you've got to go back to something that will provide the consumer
with the data rather like the truck operators are provided with data
by the equipment manufacturers and they make the decision which muffler
to buy and to put on their particular truck. It's their decision, in
523
-------
25
the long run. You provide the legal authority, the police or whatever,
with a test procedure like the 20 inch procedure for deciding whether
the individuals are complying with the lav/. And, we've heard about
the difficulties of providing a simple bench test procedure for that.
So that's what you've got to do and I think you've got to be specific.
But there's no way of stamping a label on a particular product and say
that's going to always be satisfactory. It's been said several times
and I agree, there's no such thing as a good muffler or a bad muffler
excluding the tin cans. Without saying where and how and why you.'re
using it.
Cecil Sparks
I just want to second that and the way your first question is worded
it says that the prediction has to be in a form of an actual noise level
so it can be added to the other noise levels from the o-ther vehicle
sources so we agree that some of these more erudite definitions of the
inherent muffler characteristics much more adequately characterize
muffler performance than something like insertion loss. My wife isn't
going to be able to use something like that and very few people will.
So it's more of an evaluation process of what you do with the data
after you get it more so than how you get the data.
Ernie Oddo
That's true, that's an important part of the contract. Would any of
the panel members like to comment on those two questions relative to
any other vehicles other than autos and trucks which is more or less what
we have dwelled on here.
Dr. Alfredson
I thought I'd just make a point here which really isn't very relevant
but the manner in which a vehicle is driven, can make quite a difference
to the amount of noise. This is particularly important for .the
recreation vehicles.
524
-------
26
Dr. Brammer
A comment on the small engine vehicles, I think the technique employing
some form of engines is highly preferable to those that don't, so
if you want a constant measure I would use one of those. I don't know
whether the panel agrees, but we've really v/andered around and I don't
think we've got anywhere. I think if one simplifies the question
perhaps in the way I suggested right in the beginning we might lead off
to some direction, that is of course assuming there is a need in some
way to control the production of mufflers which is what it boils down
to. Control the performance of mufflers I should say. This can be
either by some form of self certification that this muffler is better
or worse, backed up with some test procedure which could be used as
a method of arbitrating between a manufacturer perhaps that claims it
is equivalent to the existing one and perhaps a consumer or in
this case the regulation agency that claims the muffler is in fact
superior or inferior. All of the qualitative descriptions that I
have used will be turned into quantitative terms such as equivalent
could be for example +_ 5 dB of original equipment for example, and I
think that if we're going to make progress on these questions I'd like
to see us sort of direct the discussion a little bit, somewhere along
these lines.
Ernie Oddo
I don't know if panel members are familiar with the two testing Institutes
in France and Germany. The one in Germany I'm referring to is the TUV.
We've been in correspondence with Heinrich Gillet Company one of the
German manufacturers who makes mufflers for various vehicles. They
sent us a lot of information and data on these two Institutes that
do testing for the respective governments.
I believe they're not' government institutes or testing agencies but
they are certified by the governments in each one of the countries.
They do have a scheme and a process whereby if a company wants to sell
an aftermarket muffler, in either country he must submit that product
525
-------
27
to the appropriate testing Institute and that testing Institute uses
a standard bench test methodology to evaluate the mufflers. The test
is an A, B type comparison in which they compare the OEM muffler to
the replacement muffler as part of the methodology. We haven't
interpreted the articles fully yet, since we haven't had them fully
translated, lie just have selected paragraphs that have been translated,
There is an indication that they use a standard engine as part of the
test methodology. We will follow up on this information after this
symposium.
Cecil Sparks
But they're not taking that to predict noise level on any arbitrary
configuration that you have in mind thereafter. So I agree, that's
a reasonable approach. To qualify your muffler.
Ernie Oddo
Well, that's what we have to find out, what qualify means. We don't
have the-articles fully translated but if any of the panel members
are familiar with those testing methodologies and what they mean
we'd really appreciate hearing.
Prof. Davies
I don't know about these two but in England it's the Motor Industries
Research Assoc. and they do perform this function. And I can state
quite categorically they don't use a standard engine because I know
it doesn't work. They are certifying a product or a range of products
for a specific vehicle and that's the way they work. They provide
the certificate. I think also that from what I've heard in this
meeting, from all the manufacturers including the replacement manu-
facturers, they do provide some sort of certificate. And I think we're
getting hung up on technology. Can I get back to what I said in the
beginning, if you go to buy a washer or cooker or whatever that's
certified when you buy it. If you're going to buy a recreational
device like a high-fi system that's really certified, I really can't
understand what they put on the documentation but that's certified
all right. The manufacturer puts so much dope there, if he didn't
526
-------
28
comply you'd get him. You know for non-compliance, at least he's
responsible. Well, there's one point. After and secondly you go and
buy whatever junk you like and put it in your house but if that
doesn't meet city regulations that's your responsibility and it's
not the supplier's fault. So I'm saying, the route to1 follow is the
supplier, provides the certificate, and I think they're willing to
do this, and the user is responsible to seeing the compliance is
agreeable. Now, if the user's worried it's up to him to approach
the supplier and" say, look, if I use that product am I going to get
bombed. And he'll get an answer.
Peter Cheng
I would like to agree with Professor Davies and I would like to
amplify that point showing our extremes. In the State of Florida our
aftermarket customers would like to buy high performance mufflers
more so than many other states for the simple fact the State of
Florida has a rather strict enforcement.
Larry Eriksson
You mentioned other nroducts and I think it's probably obvious but I
think you should say for the record that there are a couple of other
things on these other products that are extremely important to consider,
the obvious one, particularly for motorcycles and snowmobiles is the
extremely strong connection between the sound level of the exhaust
system and the horsepower. Certainly the exhaust system is connected
with the power produced by the engine for all of these products but
snowmobiles and motorcycles is of such a different order of magnitude
consideration in my mind that that truly has to be considered separately.
The other one would be in the automobile area although we're not
involved in automobile mufflers it's certainly the case that as I've
been told by my friends in the industry there that subjective consider-
ations, and I think v/e're all aware of this in terms of automobile
mufflers, are at least as important as objective measurements and I
think that's fairly unique to automobiles and perhaps it does carry
over to some of the others but particularly so in automobiles that
527
-------
29
in terms of what's good or bad for the consumer a subjective character-
ization does play a pretty important role in terms of whether the
consumer finds this to be a satisfactory muffler and I assume this is
the kind of thing we're shooting at in terms of regulatory activities
as to somehow satisfy the consumer in terms of what he buys. So I
think the subjective aspect is going to have to be looked at if you're
out to do that for cars.
Ernie Oddo
At this point I believe we'll open up questions from the floor.
Don Whitney
I think I'd like to bring up a point that I don't think anybody at
this conference has said. Namely, that we already have labels on our
mufflers. We all have part numbers on them, those part numbers
refer back to catalogs, those catalogs go to the individual manufacturers,
the muffler manufacturers already know the performance of those mufflers
in relation to the performance of other mufflers that they themselves
have and they have a pretty darn good idea of what those mufflers do
already. I would like to add one other part with respect to the SAE
test, as I understand it in terms of an insertion loss test I really
don't agree particularly that insertion loss is the thing that we
want to measure. However, in terms of comparison of one muffler with
respect to another, I think it can do a pretty good job of telling us
equivalence on a system that truly duplicates whatever the vehicle
with its exhaust pipe lengths, tail pipe lengths, etc. do manage to
do. I think that we can ask a question here relative to the accuracy
point that's come up many times and I would like to turn the question
around instead of saying how good is the accuracy I'm more concerned
with how bad is the accuracy from the standpoint that it's fine to
say that a muffler is approximately equivalent to the muffler that
might have been on the equipment in the first place but I worry when
we say it's approximately equivalent. Is that accuracy good, do I
have to put in a standard deviation of 2 dB and then in order to
manufacture a replacement muffler and satisfy myself with some reasonable
confidence that my new muffler will be below or equivalent to, do I
have to design the new one to 5 dB below, or whatever. How bad is
528
-------
30
the correlation is of more concern than how good is the correlation.
I'd like to reiterate again it's been mentioned several times that
the performance of a muffler on a particular engine does in fact
affect the power but I'd like to also say that it does affect emissions
also since the pressure pulse is back on the engine will affect the
instantaneous pressure at the valves, etc. and as a result will affect
the emission characteristics. We're getting into a dual regulatory
situation where we've got a lot more than just sound levels to
consider. I think that's an extremely important thing. Just the
fact of possibly putting double testing in terms of requiring an original
manufacturer for the full vehicle which is what I'm involved in, a
double test, I would say that whenever double testing is involved it
ineffectively decreases the level to which we have to manufacture
trucks. Using trucks as an example simply because you have to meet
both standards therefore the total truck noise is lov/er. That might
be a desirable objective but I don't think that's the way to go about
it. I would like to say that while I don't necessarily endorse the
precise California procedure the J1169 SAE procedure for passenger
cars is a course filter, it's difficult to get down to precise levels
in terms of enforcement, however, it can do a job, it can do a real
job more than I think new truck or new passenger car regulations
will do, in the sense that those vehicles aren't really bad right now
the ones that are really causing the problem in the community are
the ones that don't have any mufflers, they have straight pipes, they
have modified systems, that type of thing is the thing that we really
need to get rid of and while the J1169 for passenger cars is a coarse
affair and we all agree it's coarse it's not a fine test it can do
a very effective job.
Nick Miller
I think we need to focus on the fact that as it's been mentioned, there
are two areas here of concern, I think, first those pieces of equipment
that are now subject to regulation as new equipment and those that
aren't. We're more familiar with those that are, so we'll address those.
529
-------
31
I think we need to remember that dun'nq the promulgation of the truck
regulation and all of the other new vehicle regulations both EPA and
the Industry were extremely careful to avoid any restrictions upon the
componentry that's used to meet the standards. The truck regulation
and the other regulations are overall performance standards and this
was the philosophy taken so that each manufacturer based on his
understanding of his market, could comply with those regulations most
economically. How, the concept of labeling a component is somewhat
akin to wearing suspenders with a belt. The vehicle regulation, .the
truck regulations, and the others that are patterned after that, has
tampering provisions which obligate the user to use equipment that will
not degrade his noise level. In addition, the proposed revisions to
in-use regulations will also provide some assurance that won't get out
of hand. I think what was going to happen is that obviously the
manufacturers are not going to provide equipment that will raise noise
levels and the aftermarket suppliers are going to be forced into that
position just to stay in business. I think this is a situation where
we can depend on the free enterprise system and along with the in-use
regulations to provide all the necessary policing that we need. So,
I think we have to look at the objectives that we had when we first
started looking at regulations for new products and stick with that
philosophy because I think it is a well formed one and I think it's
been fairly successful.
Ross Little
I have a comment more than a question. In sitting througn
this whole program, many of the sneakers appear to me really aren't
addressing what we need or what's needed out in the field. We need as
I see it, to identify the aftermarket exhaust system which when
installed degrades the noise level of the vehicle. We don't have
problems as a general rule, with new vehicles. So in the rating system
we need a relative noise level which correlates to a sound level ascribed
530
-------
32
to the vehicle when the vehicle is first delivered to the first user.
That can be the same test procedure or some other way of arriving at
it. Then these are the main things, there is a standard being proposed
here for labeling but someone eventually has to enforce it and if the
numbers aren't correctable or something that can be used, then the
enforcement goes down the drain, and there is no enforcement and the
whole program is lost.
Uayne Marcus - Motorcycle Industry Council
First off, in the regulations that are under consideration now labeling
regulations are naturally directly from the noise control act and I'll
read you one relative clause from that. Section 8, which says, "the
administrator shall by regulation require that notice be given to the
perspective user of the effectiveness, of the products effectiveness
in reducing noise." So this is what at least the Congress and the
President of the United States were looking for when this act was passed.
Now, in determining what the effectiveness in reducing noise is, in
my mind, we're looking not for a comparative number relative to an OEM
number. What we want is to know what is the reduction in noise from
a muffler, any muffler because certainly the OEM produces replacement
mufflers as well as aftermarket companies. Secondly, earlier in the
program today we learned that even OEM produced composite or universal
mufflers for older vehicles. The replacement muffler industry including
OEM replacement mufflers, is as far as motorcycles industry is concerned,
is from a labeling standpoint, this labeling regulation 204, should be
aimed at pre-effective date motorcycles, that is, motorcycles which
are produced prior to the effective date of the upcoming new motorcycle
and replacement exhaust regulations because I don't know if you're
familiar with it, if all of you are familiar with it, but as far as
motorcycles are concerned there are two such regulations which include
labeling provisions and which include noise provisions. The ones that
are coming up, very shortly will set noise level standards for motor-
cycles such as other types of vehicles already have on the books.
This one, that we're considering here is purely for the consumer's
information. Therefore, motorcycles which are produced after the
531
-------
33
effective date of this, soon-to-be-announced noise reduction regulation,
will be controlled. They will be controlled to a certain level of
noise emissions. It's the pre-effective date, the ones that are out
on the streets right now, those that have deteriorating mufflers on
them at present and those which have engines which have gone through
an extensive break-in period and have different source characteristics
than when they were originally produced. So what I'm interested in
is knowing how to look into and how to discover what the reduction
characteristics of an exhaust system are on these broken-in, presently-
on-the-street vehicles, not necessarily the vehicles that are going
to be regulated.
Martin Burke - John Deere
I have both questions of the panel as well as comments.
In the area of snowmobiles, snowmobiles have been regulated by States
for a number of years now, have a 78 dBA drive-by level per SAE J192.
As a result of this fairly stringent regulation snowmobile manufacturers
have had to put in unitized exhaust systems on the column in which
there is only a single connection between the engine and the exhaust
system that is a single flexible type connection. Earlier years we
used to see systems that had two or three joints in it and which you
could perhaps replace with various components. Since snowmobiles
are basically different between manufacturers, I guess I'm not currently
aware of an outside replacement market on snowmobiles other-than the
OEM supplying exact replacement parts. Which would I guess in the
case of our company, be identical to or better than the original ones,
and I say better than, it could be a case where we carried a model
through several years and because of the increase or reduction of
noise we've had to improve the exhaust system in those cases we have
replaced the older systems for repairs with the newer systems, flow
what decision does a customer have to make if he can only get one
system from one source for that machine.
53*2
-------
34
Unknown
I'd like to clarify one point, and that is that it's impractical to
put a noise level on an exhaust system. Uhere vehicles that are
manufactured to meet an overall vehicle regulation one manufacturer
may require more of the exhaust system than another does. And so
the only thing that makes any sense is to require equivalence to the
original system. And that's much easier to get than a number to begin
with and it's the only one that's going to make any sense to the consumer.
Frank Savage - Donaldson
I look at this thing and there are three parts to this whole question
here. One is the government which is responsible for setting the
standards and enforcing the standards and the manufacturer who makes
the particular product and has to and must stand behind that product
as far as performance is concerned. And then the consumer, and it
seems like what we're doing here is putting the entire load or the
responsibility for meeting noise regulations on the manufacturer or
the government. I think the consumer has an equal share in this
whole business here. I think that the muffler manufacturers can provide
a bench mark and I say bench mark because that eliminates the accuracy
type of question but at least it's a bench mark which he will certify,
that says that this product will work on these machines. You've got
to make sure that the consumer' has not taken this good quality muffler
off and replaced it with a tin can or a straight pipe. You've defeated
the purpose of course, of the silencer supplier or the program or in
the case of the heavy truck user, where the shell is still in good
condition but all of the internal parts are ignored, but you still
run it down the road. The second test that has been used widely is
the total vehicle noise test. Now, I'm not suggesting that all these
tests be run simultaneously by any one person but the total vehicle
noise test allows the final supplier of either the whole snowmobile or
the whole truck or the whole motorboat, integrate all its noise sources
to qualify through some procedure in his own facility. I think it's
been demonstrated a number of times that if you want to get a sound
533
-------
35
pressure level at some distance in trying to use a bench type test,
that you have to use the actual engine v/ith the actual system or a
system that is qualified to predict some sound pressure level at some
distances. There were at least two procedures given today by Larry
Peters, by a John Deere man where they had correlation with their
own bench testing to get them to fifty feet. Of course, these facilities
individually could be certified by EPA and then with published sound
levels a certificate could go out that certifies that the silencer was
tested in a facility certified by EPA. Ne already have a mechanism
that takes care of not relating accurate information and that's called
a guarantee. A man simply has to ask for a guarantee and if it doesn't
meet it let's say a truck muffler, if he buys one and takes it out to
Mr. Ross's test station and it doesn't pass the test he carries it back
and gets his money back. So, that allows all the test facilities to
date to go ahead and operate. We have dealt with the problem of muffler
labeling only in-ISMA, Industrial Silencer Manufacturing Association,
we have to deal with that because of the stationary source, seldom do
you know what the exhaust pipe length is or what the tail pipe length
is and in many cases the silencer is purchased and you really don't
know what the engine is. From my own experience, and I'm going to go
back to some of the things that Larry indicated and Mr. Blaser from
General Motors, if you want to talk apples to apples, a simple comparison
of mufflers, not relating it the in-use sound pressure levels, because
you cannot unless it's on the actual engine on the same source but if
you want something like the absorption coefficient, or transmission
loss class, what is it? - ASTM70 they give a laboratory test procedure
and clearly state that you'll get different numbers when you apply this
to the field. If you have to have some comparison, then you need to
look at broad-band noise. I prefer insertion loss with no tail pipe
and then an exhaust system, exhaust pipe that minimizes the effect
on any silencer that would be tested. And it would have to be tested
at an average flow rate for the mean end use, i.e. automobile exhaust
typically has much higher exhaust velocities than in the stationary
engine and it would have to be tested at some average or mean temperature
for the end use, this is particularly true for an engine exhaust versus
an engine intake.
534
-------
36
Ken
I'd like to make a comment on these procedures that, if they don't
include shell noise or pipe noise or leaks due to clamps or anything
like that, they aren't going to be accurate and we have to ask EPA
what accuracy we're looking for.
Ernie Oddo
Thank you. This is the last call for questions for the panel w,,ile
we have them up here. He will next go into the third part of our
program in which the EPA members will replace the panel members on
stage and we will open the session with questions from the floor.
Panel members, v/e thank you very much for your participation in this
symposium.
A reminder to everyone that we will be publishing proceedings of this
symposium in the very near future. Everyone who attended this symposium
certainly will receive a copy of the proceedings. A word to those
people who gave papers at the symposium, please send copies of your
paper, with art work to me at McDonnell Douglas in California. We
are assembling the proceedings for the EPA.
At this time, v/e will open this session for questions for the EPA
from the audience.
Bill Roper
Perhaps I should pick up on some of the questions that v/ere asked
earlier. The one from Larry Erickson about what is the objective of
the EPA labeling program? I think that at least the general objective
remains the same as it v/as spelled out in the Federal Register Notice,
the four points that we've put on the board, or the viewgraph a little
earlier, but I think specifically relating to exhaust systems, there's
535
-------
37
tv/o specific areas where we were looking for information at this meeting
and that was information on development of a statistic for a comparison
between two exhaust system or two mufflers; an A-B comparison with OEM
or whatever, a relative comparison between two systems. The other is a
statistic or approach for developing information, statistic information,
on comparison between a total vehicle level and the exhaust system.
Now those are two general categories of information that involve
different methodologies and can be used in different ways. And we've
had opinions expressed as which one is the better or the worse. I
think to be quite frank, in a government study effort such as we have
under way here that we may or may not lead to any type of regulation,
whether it be labeling or eventually a standard, a noise performance
standard, it would be in a sense dishonest on my part to say specifically
what is going to happen or what's not going to happen. We're collecting
information at this point, to define what the problem is and what the
possible solutions are -given the general objective providing information
to the consumer or user, in this case, exhaust system muffler, that
he can use in the purchase decision. I don't know if that's a satisfactory
answer Larry, but that's what I have to give you. Another point that
was raised by Nick Miller regarding the situation in the truck area.
Implying that there really wasn't a need for this kind of information
to be conveyed to the user, or purchaser of a muffler, I think he has
raised some good points; that is a good point in the truck area
I would limit it to that portion of the truck industry that
involves vehicles that are operated by interstate carriers. I think
that's fairly valid because in that area EPA does have the authority
to set in-use standards. There's only two areas where EPA has that
authority and that's for interstate motor carriers, or vehicles operated
by interstate motor carriers, and for interstate equipment and facilities
operated by interstate rail carriers, Section 17 and 18 in the floise
Control Act. So in those two areas and the railroad area we have not
set Section 6 new product standards that apply to those vehicles when
they are newly manufactured. We also have authority in the in-use
area and we have such standards. So there is a follow-through so-to-
speak on total vehicle, at least compliance requirements. But of course
536
-------
38
that would not hold true in every other product category that was
listed in the matrix. But again, I think I would go back and say that
it still remains important for the user to have the information
available to him that the muffler or the exhaust system that he's applying
to his truck will allow the total vehicle to meet a particular sound
level and quite frankly in looking at some of the material that has
been presented by Donaldson for example, where they I think, to a large
degree, are providing their customers with that type of information.
Now, one of the principal objectives of the EPA is the encouragement
of voluntary labeling which would describe the acoustical performance
of a product. Now we're encouraging that and if that occurs without
any Federal involvement, which is one of our other objectives that I
mentioned, minimal Federal involvement, I think that's what we're after,
which is a reduction in noise and if it can come about with voluntary
programs, that's fine. So, I've attempted to respond I guess to some
of your comments Nick and I think maybe this helps clarify for the
others some of the ramifications that are applicable on trucks but not
perhaps in other areas. With that I guess I'd open this session with
a call for questions from the floor.
Ed Halter - Burgess
You do have promulgated regulations, proposed regulations for air
compressors, that give a dB level that you have to check at four
or five points around the compressor and that is an overall level
including a prime mover which could be an engine, which undoubtedly
would have some kind of a muffler on it. And you've also required
the manufacturer of the air compressor to warranty it for the life of the unit,
service life be it four years, that the system would, noise wise,
maintain that level. It's required when it's manufactured. I would
assume then that the manufacturer is going to, if necessary replace
those acoustic components with equivalent acoustic components of the
same, I guess the same manufacturer, right? He would have to if he
installed these OEM parts and he's warranted this, if they had arty
problems or the customer ran a truck or damaged one of these components
they have to be replaced with the same item that was originally
manufactured. Is that correct?
537
-------
39
Bill Roper
It would have to be replaced with a comparable system component. Let
me go back a minute now. On the portable air compressor, when that
standard was promolugated it didn't include as I recall, what we call
the acoustical assurance period of some period of time when that product
would be required to continue to emit or meet the standard at which
it was designed to meet the standard at the date of manufacture. In
the later regulations that we recently imposed on wheel and crawler
tractors that the acoustical assurance period concept was involved.
But essentially, the maintenance instructions that are incorporated
in the standards require that the manufacturer identify those components
of the piece of equipment that are key noise control components that
if something happens to one of those components unless it's replaced
with an equivalent system it would not meet the standards. Essentially
identifying to the user, hey Iook9 here's a list of things you better
keep track of and maintain properly or you're not going to meet the
standard.
Ed Halter - Burgess
Isn't this essentially what you're addressing here with respect to
ground transportation. In other words, if you hold a muffler as part
of a package and you have to replace that muffler with the same type
muffler9 right? The easiest way to do that is replace it with the
same item, the same part number, the same manufacturer, you may have
to qualify other suppliers if you have a monopoly problem to produce
that same product.
Bill Roper
I think from our perspective v/e get into our general counsel informs
us, a constraint of trade situation, if we specifiythat it must be
OEM replacement. So we're looking at v/ays of identifying the performance
so that anyone who produces a product that meets that performance could
in fact sell it, have it applied to the piece of equipment and if that
gets back to what we're talking about today, and that way can be used
to characterize the performance, in this case of the exhaust system.
538
-------
40
Ed Halter - Burgess
But wouldn't the ultimate be that you had to qualify that on that
particular piece of equipment. In other words, if you're going to
replace this on a crawler tractor and you had a certain procedure to
checkout on a crawler tractor, you would then, any of the replacement
mufflers or components would be tested on the crawler tractor and
that same method.
Bill Roper
That's certainly one way it could be done. Probably the easiest
way it could be done at this time.
Ernie Qddo
Another thing I'd like to add here. Concerning exact replacements to
the OEM, we've met with the automobile manufacturers and other motor
vehicle manufacturers and have discussed consolidation of design.
A wide variation of many different designs result from continued
consolidation. The end result is a raft of mufflers that are still
so-called OEM equipment. You may find a wide tolerance there if you
would actually measure the performance of those aftermarket mufflers
and compare them with the OEM performance. There could be 3, 4 maybe
5 dB difference. That's the practical world.
Doug McBann - Ford Motor Co.
I'd like to clarify the statement that Ernie just made. From a
regulatory standpoing the aftermarket mufflers that we produce and sell
are equivalent to original equipment. The subjective levels have been
compromised in many cases.
539
-------
41
Bill Roper
Could I ask a question? It came up in the earlier session that in
the automotive area, looking at subjective levels was important.
Is that in regard to exterior, interior or both?
Doug McBann - Ford Motor Co.
Both.
Jim Moore - John Deere
The snowmobile industry currently has a voluntary total vehicle noise
labeling program and Martin Burke brought out the fact that there
currently exists no aftermarket in snowmobile exhaust systems. In
view of this, do you think it's necessary to label snowmobile exhaust
systems?
Bill Roper
I think the information we have been provided on snowmobiles certainly
puts them in a unique situation. I think, compared with some of these
other areas and that's certainly something we'll consider. Whether
there is a need or not in the snowmobile area. Again, I think I want
to go back to the point that we're really on a fact-finding mission
at this point in this particular area of exhaust emission performance
and this kind of information is very useful to us. I can't sit here
and say what the agency is going to decide to do on that particular
question because I don't know, but certainly that information would
raise a question of whether or not it's necessary on snov/mobiles.
Doug Rowley - Donaldson
I'd like to discuss this voluntary action a little bit Bill. I know
that Ross Little spent about a year and a half getting voluntary action
out in the State of California relative to controlling truck noise and
I'd like to ask the EPA the question, how you intend to get voluntary
action? Obviously, it must be through some enforcement pronram. Could
you touch on that a bit?
540
-------
42
Bill Roper
In response to that I think again of the EPA's standpoint we would be
looking at what's happening out in the country now. For example, is
there an effective voluntary compliance program now? As a result of
say State regulations. An awareness on the part of the manufacturer
that his product is noisy and is adversely affecting his sales and
causing him a harrassing problem because it's against state regulations
or whatever and that the industry say has gotten together and come up
with a test procedure and is voluntarily certifying or labeling or
whatever their product to meet a specific noise level. We would be
looking at what's happening today, and how that relates to reducing
noise from that particular product. I might go on further and site
some examples. In the snowmobile area which was mentioned earlier today,
there was a lot of concern in various snowbelt states for levels from
snowmobiles and there were laws passed and then there was response by
the snowmobile industry to do something about lowering their noise
levels. They did establish or agree amongst the association a procedure
that was acceptable to them to identify the noise performance of their
product and they have gone ahead and labeled. That's just one example,
there's perhaps others but from EPA's standpoint, I think as we move
into any area where there was labeling or setting standards we would
be assessing and looking at what's being done now with that product
and what's possible to be done. Again, I guess we are going to a
Section 6 regulatory study which many of you may be aware, the kind
of three pronged approach we take there and that is to look at what
technology is available, what's the cost of applying that technology
and what kinds of health and welfare benefits you get from applying
the various levels of technology. We in the standards and regulations
division are responsible for putting together the facts and coming
up with recommendations for the agency to make decisions on and so
again, our job is fact finding and certainly what's going on in the
industry as far as voluntary standards is an important factor that
would go into the arraying of information and generation of recommendations.
541
-------
43
Ernie Oddo
We have time for one or two more questions.
Ross Little - CHP
I have a comment on snowmobiles - To begin with, I don't know anything
about snowmobiles. We regulate them but we don't have many out in
California, fortunately. But I am hard pressed to believe that they're
as innocent and pure as they are making out to be, I beg your pardon,
but I know they race snowmobiles and if Hooker industries think they
can get another ounce of horsepower out of the snowmobile with an
unsilenced expansion chamber, that's what you're going to find on it.
And if they'll race with them, they'll also ride out in the woods
with them. They do motorcycles.
Bill Roper
That's the other side of the coin. We're looking and we're sensitive
to that side also. Although there appears to be some difference between
the snowmobile user as a general group and motorcycle users as a general
group based on the information we've seen so far.
Jim Moore
Just a slight rebuttal to what the gentlemen is saying. It is certainly
true, there are expansion chambers and stuff available but I don't
call those silenced exhaust systems, and in most states they are not
allowed to run except on the race track in a sanctioned race and in
today's racing rules, generally you could determine whether you're
going to race stock or race modified. If you race stock you're going
to have to have a system that meets the 78 dBA level. If you race
modified, and they are allowed in some areas, the manufacturer has
no control of that and nobody gives a dang about the sound level on
those machines, especially the guy racing or the people at the race track.
542
-------
44
Ernie Oddo
One last question.
Nick Miller - International Harvester
I think the point here is that whether the parts are labeled or whether
they're not labeled, has nothing to do with whether someone will modify
a vehicle no matter what it is. I think it's important as we address
the EPA's concern for voluntary program. Could we have the'rnatrix back
up on the board for just a second.
I think it's important to bring up at this point the areas where we
do have voluntary areas that have been successful. First of all, both
the auto and light trucks have been very successfully controlled in
California and some other localities on a voluntary basis by the
manufacturers. It's not new vehicles and well maintained vehicles in
any of those areas that are a problem, it's modified vehicles and only
enforcement will solve that problem, The heavy truck you alluded to
Bill is a matter there of the ICC regulation, motorcycles are just
about to be regulated and in the hearings that I've attended in the
various states and so on they have done a good job of bringing their
vehicles and aftermarket parts into compliance where they are regulated.
Snowmobiles we have noticed, have a special situation as you said,
buses you now have your thumb on and so I guess all I can see that's
there any major gain for is motorboats and I understand you're looking
at those, Bill
Bill Roper
Ue just started this year looking at those.
I might respond a little more to Nick's comment there. I'd add though,
that in the early stages on all of those products that we have regulatory
programs fairly downstream or have already set regulations that we did
look at what was going on from a voluntary standpoint in the early
stages of the study and I'd like to mention that in California and some
543
-------
45
of these other places, automobiles and light trucks, they did have
standards in effect in the late sixties or earlier seventies that set
standards in a sense did have a lot to do in bringing some of the noise
levels down. I also agree that it's the modified vehicles that are
a problem. That varies from category of vehicle to other categories
of vehicles on how big a problem it is. Particularly motorcycles seems
to be a big problem.
Uayne Marcus, - MIC
I'd like a clarification, I got the impression from listening to you
earlier that you're shooting for some form of comparative rating as
opposed to an absolute rating. I'm speaking of comparing the level
of an aftermarket exhaust system to an OEM exhaust system or comparing
an exhaust system to a total vehicle noise. Is this a misconception,
if not can you explain why you're shooting for comparative?
Bill Roper
I meant to convey the thought that we're looking at both of those.
We have not decided at this point whether one from our standpoint
is better than the other, but we did want to get comment and information
on the kinds of things that would be available to us as tools in
assessing the performance of an exhaust system by both approaches.
Does that answer your question?
Wayne Marcus
Yes
Ernie Oddo
Thank you very much. Is there a final comment you would like to make,
Bill, before we close the session?
544
-------
46
Bill Roper
I guess from EPA's standpoint I would like to again thank all of you
for participating in this symposium. This is, I think the first time
the EPA in the noise office has conducted this type of meeting with
the technical experts in an area this early in a study program and
as I think has been shown, in this afternoon's session there really
are no easy answers to some of the questions that we're faced with
attempting to collect information on and make recommendations to the
agency. There is difference of opinion and we're not surprised
by that, but I think it's been very constructive the last three days
to have the caliber of people that we've had at this meeting together
in discussing, I think quite frankly and openly, their opinions on this
subject and I heard a comment earlier this morning that even if there
were no specific recommendations that came out of this meeting, but
just the fact that a lot of ideas were thrown up, a lot of thoughts
have been discussed that some of the manufacturers of these products
may have picked up some ideas and we may get potentially some noise
quieting coming out of the ideas that were exchanged at this meeting.
After all, that's the business that we're really in is to make it a
little quieter out there in the environment and I think that's great
if we contributed toward doing that through this meeting; so again
I'd like to thank you all and wish you a safe journey home with one
thought too that I want to leave, and that is that tin's is in a sense
the beginning of what I hope will be a continuing dialog between
many of you and EPA as we move further along in this program, so
thank you.
54,5-
-------
LIST OF ATTENDEES
ALFREDSON, R. J. DR.
Monash University, Dept. of Mechanical Eng., Clayton Victoria,
Australia 3168
ARBIZZANI, RON, Dir. Muff. Eng.
Maremont Corp., 250 E. Kehoe Blvd., Carol Stream, 111. 60187
BASCOM ROGER, Chief Engineer
Harley-Davidson, 3700 H. Juneau, Milv/aukee, Misc. 53201
BAXA, DONALD E., Prof.
University of Uisconsin, 432 N. Lake St., Madison, Wise. 53706
BECK, JIM, Manager
Sun Electric, 6323 N. Avondale, Chicago, 111. 60631
BECKON, HEIR, Manager Sales
Donaldson Co., 1400 W. 94 St., Bloomington, Minnesota 55431
BELLING, ROCKY, R&D Manager
Mr. Gasket Co., Div. W.R. Grace, 4566 Spring Rd., Cleveland, Ohio 44131
BERIL, MARTIN, Manager Ace. Prod.
John Deere, 220 E. Lake St., Horicon, Wise. 53032
BLASER, DWIGHT A., Sr. Research Engineer
GM Research Labs, GM Technical Center, Warren, MI 48090
BLASS, JAROSLAV, Res. Mgr.
Kawasaki Motor Corp., Shakopee, Minn
BORTHWICK, JESSE 0., Noise Section Administrator
FT Dept. of Envir. Regulation, 2562 Executive Center Circle E.,
Tallahassee, FL 32301
BOTELER, KEN, V.P.
B&W Mufflers, 2415 S.W. 14th, Oklahoma City, Oklahoma
BRAMMER, A. J., Dr.
National Research Council of Canada, Montreal Road, Ottawa, Ont. KIAOSI
BRENNAN, KENNETH, Engr.
Hendrickson flfg., 8001 W. 47th St., Lyons, 111. 60534
BURKE, MARTIN
John Deere, 220 E. Lake St., Horicon, Wise. 53032
CAHILL, JOHN, Sales Manager
Stemco Mfg. Co., P.O. Box 1939, Longview, Texas 75601
CHENG, PETER, Sr. Engineer
Stemco, #9 Industrial Blvd., Longview, Texas 75601
547
-------
CRAGGS, TONY, Dr.
University of Alberta, Edmonton, Alberta, Canada T6G 2E1
CROCKER, M. J., Dr.
Purdue University, 'Ray W. Herrick Labs, W, Lafayette, Indiana 47907
DANNER, T. A., V.P. Engineering
Arvin Auto Div., Arvin Ind., Columbus, Indiana 47201
DAVIES, PETER, Prof.
I.S.V.R., University of Southampton, Southampton, England S095NH
DIKA, ROBERT, Noise Engineer
Chrysler Corp., Proving Ground, Chelsea, III
EARNSHAW, GINNY, Editor
Bureau of National Affairs, 1231 25th St., Washington, D.C. 20037
EDWARDS, SCOTT
EPA, Crystal! Mall, #2, Room 1102, Mail Code AH-471,
Arlington, VA 20460
ERICKSON, LARRY, V.P.
Nelson Industries, P.O. Box 428, Stoughton, Hisc. 53589
GIRVAN, MICHAEL J., Asst. Exec. Dir.
Motorcycle & Moped Ind. Council, 802-45 Richmond St., W. Toronto,
Ontario Canada M5H 1Z2
GOPLCN, GARY, Devel. Engr.
Nelson Muffler, Box 428, Stoughton, Wise. 53509
GROCK, RAY, President
Pipes Etc., 1632 W. 139th, Gardena, Calif.
HALL, JAMES R., V.P.
A P Parts Co., 1801 Spielbush Ave., Toledo, Ohio 43G94
HAAS, THOMAS G, Noise Control Engr.
J.I. Case Co., 700 State St. CL 124, Racine, Wise. 53404
HARTER, E. J., Chief R&.D Engr.
Burgess Industries, Burgess Manning Div., 8101 Carpenter Fwv.,
Dallas, Texas 75247
HENDRIX, DAVID, R., Manager C. Engr.
Riley-Beaird Inc., P.O. Box 31115, Shreveport, LA 71130
548
-------
HERBERT, MARK
University of Cincinnati, Mech. Eng., Hail Loc #72, Cincinnati,
Ohio 45221
HICKLING, ROBERT
General Motors Research, Harren, MI 48070
HORNETT, HARRY, Senior Engr.
McDonnell Douglas Astr. Co., 5301 Bolsa Ave., Huntington Beach, CA 92647
INAGAWA, MINEICHI
Mitsubishi Motors Co., 10-Ohkuracho, Nakaharaku, Kawasaki, Japan
IRVINE, GERALD, Mgr. Engr.
Scorpion Inc., P.O. Box 300, Crosby MM, 56441
JOHNSON, DUANE, Test Engr.
Caterpillar Tractor Co., 100 N.E. Adams, Peoria, ILL 61601
KERR, JAMES
USEPA, Washington, D.C. 20460
KICINSKI, KEN, Engr.
Nelson Muffler, Rt. 51, Stoughton, Misc., 53539
KILMER, ROGER D., Mech. Engr.
National Bureau of Standards, Bldg. 233, Room A149,
Washington, D.C. 20234
KONISHI, K., Director, Washington Office
JAMA 1050 17th St., N.H., Washington, D.C. 2003G
KOPEC, JOHN, Acoustical Liaison Engr.
Riverbank Acoustical Labs, 1512 Batavia Ave., Geneva, ILL. 60134
KRALL, ERIC G.
A B Volvo Truck Div., Dept. 26435, A.B. 29, Gothenburg, Sweden
LAI, PATRICK K.
A.C.S. Ltd., 114 Railside Road, Toronto, Ontario, Canada
LITTLE, ROSS A., Engr.
Calif. Highway Patrol, 255 1st Avenue., Sacramento, CA 95808
LOTZ, R. W.
Chrysler Corp., Detroit, Michigan
MC CORMICK, JAMES, St. Engr.
Walker Mfg. Co., 3901 Willis Rd., Grass Lake, Mich 49240
MC DONAGH, JAMES R.
Riker Mfg. Inc., 4901 Stickney Ave., Toledo, OH 43612
549
-------
MANN, ROY L., Engr.
J.I. Case, 700 State St., Racine, Wise. 53404
MARCUS, WAYNE, Technical Analyst
Motorcycle Industry Council, 4100 Birch St., Suite 101
[lev/port Beach, CA 92660
MARGOLIS, DONALD, Prof.
University of California, Dept. of M.E., Davis, CA 95616
MASON, BOB
US DOT/Transportation Systems Center, Kenaall Sq., Cambridge MA 02142
MILOS, JOSEPH, Q.C. Manager
M.D.I., 5310 W. 66th St., Chicago, 111.
MILLER, N.A., Staff Engr.
International Harvester, 2911 Meyer Road, Ft. Wayne, Ind. 46803
MOLLOY, CHARLES T., Dr.
USEPA, Washington, D.C.
MONDZYK, DARRYL J.
Muffler Dynamics Inc., 5310 W. 66th St., Chicago, 111., 60638
MOON, CHAS. L., Mgr Test
White Motor Corp., 35129 Curtis Blvd., Eastlake, Ohio 44094
MOORE, JIM, Engr
John Deere Co., Horicon, Wise. 53032
MORLEY, R. K., Supv.
Ford Motor Co., 21500 Oakwood, Dearborn, Mich 48124
MORSE, IVAN E., Prof.
University of Cincinnati, Mail Loc 72, Cincinnati, Ohio 45221
MUTH, ROY W.
International Snowmobile Industry Association, 1800 M. Street N.W.
Washington, D.C. 22036'
MC BAIN, W.D., Dev. Engr.
Oakwood Blvd., Dearborn, MI 48120
NAVARRE, GEORGE, Sales Manager
Riker flfg. Inc., 4901 Stickney, Toledo, Ohio 43612
NECHUATAL, MICHAEl
Illinois EPA, 2200 Churchill Rd., Springfield, 111. 62706
550
-------
NEMECELL, JACK, Sales Mgr.
Donaldson Co., Minneapolis, Minn
NICHOLS, JEAN
Bombardier Research Center, Valcourt, Quebec, Canada
NIEMOELLER, DON, Lab Manager
Arvin Automotive, 2505 N. Salisbury St., W. Lafayette, Ind. 47906
NOLEN, ROBERT K., Manager
Maremont Corp., 250 E. Kehoe, Carol Stream, 111. 60187
NORDIE, KENNETH D., Designer
Nelson Muffler, Box 428, Stoughton, Wise. 53589
ODDO, ERNEST T.
McDonnell Douglas Astr., 5301 Bolsa Ave., Huntington Beach, CA 92647
OLSON, DAVID A., Sr. Research Engr.
Nelson Industries, P.O. Box 420, Stoughton, Wise., 53589
PAGE, W.H., Res. Engr.
International Harvester, 75600 County Line Rd., Hinsdale, 111. 60521
PALAZZOLO, JOSEPH A., Dev. Engr.
Ford Motor Co., 20500 Oakwood, Dearborn, MI 48124
PARKER, ROBERT, Ac. Engr.
A M F Harley Davidson, Milwaukee, Wise.
PETRALATI, VIC, M.E.
USEPA, 401 M. St., S.W., Washington, D.C. 20057
PRAIDRA, NICK, ACCT. Mgr.
Donaldson, Co., Inc., P.O. Box 1299, Minneapolis, Minn 55440
REINHART, CHARLES, Test Engr.
Donaldson Co., Inc., P.O. Box 1299, Minneapolis, Minn_. 55440
RENNEK, J. N.
Cipon Industries, Inc., 22 Ikon St., Keydale, Ontario
RENZ, WILLIAM, Ct. Mgr.
Nissan Motor Corp., 18501 S. Figueroa St., Carson, CA 90243
ROBERTS, PETER, MGR.
Gidon Ind., Inc., 22 Iron Street, Rexdale, Ont. MGLS 5E2
ROBLEY, ELROY, Proj. Engr.
FWD Corp., Clintonville, Wise. 54929
551
-------
ROMINGER, D., Engr.
818 Sylver Ave., Englewood, Cliffs, fl.J. ,07632
ROPER, WM. E., Chief STB
USEPA, AW471, Std. & Reg., Div., Washington, D.C. 20460
RONCI, W.L., Director Exch. Sys. Eng. & Research
Walker Mfg., 3901 Hill is Road, Grass Lake, Michigan 49240
ROSA, B.A., Director Product Engr.
A P Parts Co., 543 Matzinger Road, Toledo, Ohio 43697
ROSS, DAVID, Res. Engr.
Arvin Ind., 2505 N. Salisbury, W. Lafayette, Ind. 47906
ROWLEY, DOUG, Ch. Engr.
Donaldson Co., Inc., 1400 W. 94th St., Minneapolis, Minn.
SCHEIDT, WAYNE, A., ilgr
Maremont Corp. 250 E. Kehoe Ave., Carol Stream, Illinois 60187
SCHMEICHEZ, STEVE, Pr. Engr.
Donaldson Co., Inc., 1400 West 94th St., Minneapolis, Minn.
SCHULTZ, DOUG, VP Engr.
Walker Mfg. Co., 1201 Mich. Blvd., Racine, Wise.- 53402
SEYBERT, ANDREW, F., Asst. Prof;
University of Kentucky, Lexington, KY 40506
SHAFFER, FRED, Proj. Engr.
The Flxible Co., 970 Pittsburgh Drive, Delaware, Ohio 43015
SHAUGHNESSY, JIM, Supv. Exh. Engr.
Hayes Albion Corp., 1999 Wildwood Ave., Jackson, Mich 49202
SMITH, WM. A., Prof.
University South Florida, College of Engineering, Tampa, FL 33620
SNECKENBERGER, JOHN E., Assoc. Prof.
West Virginia University, Mechanical Engineering, Morgantown, H.V. 26505
SUIDARICH, FRANK, Mgr.
Donaldson Co., Inc., 1400 W. 94th St., Bloomington, .Minn 55431
SPARKS, CECIL R., Director
Southwest Research Inst., P.O. Drawer 28510, San Antonio, Texas 78284
552
-------
STEPHENSON, DONALD, Sen. Research Engr.
Outboard Marine, P.O. Box 663, Milwaukee, Wise. 53201
STOCK, ARTHUR W., Chief Engr.
Crov/n Coach Corp., 2428 E. 12th St., Los Angeles, CA 90646
STURTEVANT, B. Prof.
Caltech, Pasadena, CA 91125
SVOBODA, ED, VP Engr
Midas International Corp., 5300 W. 73rd St., Bedford Park, 111 60638
THOMAS, J. W., Vice Pres.
Maremont Corp., 250 E. Kehoe Blvd., Carol Stream, 111. 60187
VAN DEMARK, RALPH, Ext. Dir.
A.E.S.M.C., 222 Cedar Lane, Teaneck, N.J. 07665
WHITNEY, DON R., Exec. Engr.
General Motors, G.M. tech Center, Warren, Mich 48090
WRIGHT, PETER, Pres.
Gidon Ind., Inc., 22 Iron Street, Rexdale, Ontario
YAMADA, MAKOTO, Staff Eng.
Toyoto Motor Co., 1099 Wall Street West, Lyndhurst, N.J. 07071
553
U.S. GOVERNMENT PRINTING OFFICE 1978 0-720-335/6122
------- |