You are on page 1of 11

SAE TECHNICAL 2008-36-0575


Advanced Modeling of Aircraft Interior Noise using the Hybrid

FE-SEA method
Vincent Cotoni
Bryce Gardner
Julio A. Cordioli

James Carneal
Chris Fuller
Virginia Polytechnic Institute

2008 SAE BRASIL Noise and Vibration Conference
Florianpolis, SC, Brazil
March 30th to April 01st, 2008

AV. PAULISTA, 2073 - HORSA II - CJ. 1003 - CEP 01311-940 - SO PAULO SP


Advanced Modeling of Aircraft Interior Noise using the Hybrid

FE-SEA method
Vincent Cotoni
Bryce Gardner
Julio A. Cordioli

James Carneal
Chris Fuller
Virginia Polytechnic Institute

Copyright 2008 Society of Automotive Engineers, Inc

ABSTRACT vibration, it is useful to be able to predict and quantify the

Noise transmission paths in an aircraft include, in various sound transmission paths. This can prove difficult
many cases, both components with a few modes and others as the system may require many degrees of freedom to
with a high modal density. The components with few describe the response, and the response itself may be
modes display a long wavelength behavior and are usually sensitive to variations in manufacturing processes, material
modeled using the Finite Element Method (FEM). On the variability and environmental conditions. These problems
other hand, components with many modes show a short become increasingly severe as the frequency of interest
wavelength behavior and suit the application of the increases due to the reduction in wavelength of the system
Statistical Energy Analysis (SEA). An example of this kind deformations. Therefore, standard deterministic approaches
of transmission path is given by the vibration transmission to structural-acoustic dynamics such as finite elements (FE)
from the fuselage to the floor panels through the floor and boundary elements (BEM) are confined to low
beams. The fuselage and the floor panels possess a high frequencies.
modal density while the floor beams are considerably stiff
and display a small number of modes. The prediction of the The Statistical Energy Analysis (SEA) method
vibro-acoustic response of such systems is commonly overcomes the above difficulties by the use of space,
called the mid frequency problem and, until recently, was frequency and ensemble averaged quantities such as energy
difficult to be handled with traditional modeling [1]. This allows the modelling of some components of a
approaches. A Hybrid method that rigorously couples SEA system and the power exchanged between those
and FEM has been recently proposed. In this work, components. SEA is now a proven method for creating an
different applications of the Hybrid FE-SEA method to efficient system level model of the various transmission
aircraft interior problems are investigated and validated paths in a vibro-acoustic system. Two areas for improving
against experimental results. It is seen that the Hybrid FE- the accuracy of SEA for aerospace structures have been
SEA method can considerably improve the predictions of identified and addressed recently: (i) the need to compute
mid frequency problems. It is also demonstrated that the coupling loss factors between SEA subsystems connected
method can be successfully used to improve SEA models at complex junctions, and (ii) the need to model complex
by including some details to the model that affect the vibro- ribbed panels as SEA subsystem. While modern SEA codes
acoustic performance of the system. contain a large library of different subsystems and
junctions, applications are sometimes encountered where a
INTRODUCTION standard formulation doesnt exist for a given type of
construction. In some instances it is possible to estimate the
The transmission of noise and vibration in aerospace subsystem properties from test or local FE models;
structures often involves a combination of air-borne however, generic algorithms that can include an arbitrarily
transmission through acoustic fluids and/or structure- complicated geometry are desirable.
borne transmission through structural components. In
order to guide design changes that minimize the noise and
In this paper a recently developed Hybrid method is
used which provides a rigorous way to couple FE and SEA
components of a system [2,3,4,5]. The Hybrid method
enables the introduction of FE details into a standard SEA
analysis. This is typically performed at complex junctions
in order to capture the physics of the transmission. FE can
also be introduced to describe complex loaded components
and obtain a more accurate estimation of the power input
into the rest of the structure represented as SEA Figure 1. Stowage bin assembly connected to the
subsystems. fuselage section, and test setup for bin-frame-tie rod
The second development reported in this paper is the
periodic SEA subsystem, based on the use of periodic SEA MODEL OF THE BIN The bin is made of two
structure theory [6,7]. A FE model is created of a unit cell side panels, a door panel, and a top and bottom panels. All
and analytical expressions are used to obtain the SEA four panels are descried using singly-curved shell SEA
properties of a larger panel comprised of a large number of subsystems in the software package VA One [8]. The
such cells. The approach provides an efficient and accurate model is shown in Figure 2, and it comprises five line
way to model arbitrarily complex sections in SEA that are junctions and six point junctions. The structure was tested
difficult to model using traditional formulations. in isolation for intermediate validation and to measure the
damping loss factors (using the decay rate method [1]).
In what follows, three applications of these recent
extensions of SEA are presented and validated against
experimental results.


This section describes the process for building a

model of structure-borne noise transmission from a
fuselage attachment point to the stowage bin through a
connecting frame.


Figure 2. Hybrid FE-SEA model of the bin-frame-tie
frame and connecting tie rods shown in Figure 1 are
rod assembly in VA One.
included in the analysis that targets the frequency range
from 50 to 1600 Hz. The overall dimensions of the bin and
frame are of the order of 65x20. The frame is of
shown in Figure 1, the frame structure connecting the bin to
rectangular shape and made of connected aluminum beams
the fuselage is made of beams and several small add-on
with different cross sections. Small add-on brackets are
brackets used to connect the beams together or to the tie
used to attach the several rods and the bin. The bin whose
rods and bin. The tie rod that was driven in the tests is
door has been removed is made of four curved or
attached in the middle of the Z-section beam via a bracket.
corrugated panels: the bottom part is made of a honeycomb
All frame components are made of aluminum.
sandwich, while both sides and the top panels are in
aluminum. There are four connections between the bin and
the frame. An FE model of the frame was built based on simple
geometrical measurements and weighting (see Figure 2).
The FE model is made of independent components
An approximate analysis of the dynamic properties of
described with CQUAD4 and CTRIA3 elements and
each component shows that: (i) the tie rod and frame are
connected via CRIGD1 elements (rigid elements from
stiff structures that are not expected to have too many
COSMIC Nastran). The tie rod and screws in the four
modes in the frequency range of interest, so that SEA might
brackets connecting to the bin where described with
not be accurate for those; (ii) Alternatively, the bin panels
CBAR. The complete model comprises 3783 nodes.
have enough modes to be adequately described by SEA. It
is thus decided to build a Hybrid model where the frames
and tie rods are described with FE, and the bin with SEA. HYBRID FE-SEA MODEL OF THE ASSEMBLY
The frame and bin were assembled and suspended with
elastic strings from two tie rods, as shown in Figure 1. It
can be seen that there are two connections per side beam:
the connection with the yellow bracket is with the side
panel only, while the other one is with the side and bottom -30
panels. The assembly was tested using an impact hammer
and 14 accelerometers. As the ultimate goal is to measure -40
the transmission from the tip of the tie rod to the bin, LSide
impact test were performed with loading the rod tip. -50

The Hybrid model of complete assembly is made of -60

the SEA model of the bin and the FE model of the frame.
As shown in Figure 2, those are connected through four -70
Hybrid point junctions representing the actual physical
connections between the bin and the frame. The dark arrow
shows the location and direction of the excitation at the end
of the tie rod.
2 3
PREDICTIONS VERSUS TESTS For the loading 10 10
in the axial directions at the tip of the tie rod, the measured Frequency (Hz)
and predicted RMS velocity responses at a point of the Figure 4. Energy responses of the four panels of the bin.
frame are shown in Figure 3. The responses are averaged Experimental (solid) and predicted (dotted) response,
over the 1/3rd octave bands from 50 to 1600 Hz. The when loading at the end of the tie rod.
energy response of the four panels of the bin is shown in
Figure 4. The Hybrid model predicts the band-averaged
energy or RMS velocity response of the frame and bin STRUCTURE-BORNE TRANSMISSION BETWEEN
within about 5 dB over most of the frequency range. SIDEWALL AND FLOOR PANELS

-10 This section is concerned with the noise transmission

between sidewall and floor panels of an aircraft. Those
-20 panels exhibit a high modal density and are connected
through some complicated and stiff beams, so that neither
-30 FE nor SEA alone is very well suited for the analysis. As a
approach to enhance SEA models, the Hybrid FE-SEA
method is used to couple FE and SEA descriptions of the
various components.
sidewall and floor panels were extracted from the Boeing
737 section (Figure 5). The sidewall panels above and
below the floor are 63 wide and respectively 16.5 and
-70 24 high: they are made of similar skin-stringer-frame
2 3
10 10 arrangement, with two stringers and three frames. The floor
Frequency (Hz) is split in two unequal panels made of honeycomb
Figure 3. RMS velocity response at a point of the frame. sandwich, with dimensions 51.7517.5 and 1116.25.
Experimental (solid) and predicted (dotted) response,
when loading at the end of the tie rod. The sidewall and floor panels are connected through
three stiff I-section beams. A plate with holes next to the
skin connects the beams with the floor and skin. The seat
track is parallel to the skin and connects the beams and the
floor. The floor panels are directly connected to the beams,
along the edges of the panels.

The structure was suspended from two beams with elastic

strings, and tested using an impact hammer and 14
accelerometers. The impact locations were scattered on the
bottom sidewall panel (7 points), the top sidewall panel (7
points), the floor panel (4 points), and on the top of each I-
section beam. The acceleration response was measured at 5
points on each sidewall panel and 4 points on the floor
while the ribbed formulation predicts zero modes in the
bands centered at 64 and 128 Hz.

Figure 6. FE model of a single cell of the sidewall panel

used in the periodic SEA subsystem formulation.
Figure 5. Sidewall and floor panels connected through 3
some beams, and details of the frames. Ribbed panel SEA
Periodic SEA
floor panels in the fuselage section have different area, but 10

are made of the same sandwich material with honeycomb

aramid core and two identical fiberglass faces with
thickness 0.02. An SEA model of each floor panel was N 1
built using the sandwich formulation in VA One [8]. The
standard fiberglass material properties available in VA One
where used for the faces and the effective material
properties of a honeycomb construction were obtained from 10

the cell size and foil thickness, and the assumed material
properties of the foil.
2 3
Frequency (Hz)
This section describes the process for building a Statistical
Energy Analysis (SEA) model of the dynamic response of Figure 7. Modes in third octave bands - 32 to 1600 Hz.
the sidewall panels of the aircraft fuselage section. It can
sometime prove difficult to model complex ribbed The predicted distribution of RMS velocity per unit
structures with SEA, and a new periodic theory developed energy of the panel is shown on Figure 8 for the band
to enhance the SEA models is used here. centered at 1600 Hz. It can be seen that the response of the
structure is not homogeneous at this frequency, as scatter of
An FE model of a single periodic cell was created as the response goes to more than 12 dB.
shown in Figure 6, and was used to compute the modal
density, power input from point forces, and velocity
response per unit energy (using the algorithms described in
[6,7]). The FE model is made of 1856 CQUAD4 shell
elements and comprises 1941 nodes.

The periodic formulation for the SEA subsystem was

validated against tests on a simpler structure comprising the
skin and stringer. The modes in band of a simply stiffened
panel were computed using the periodic theory and the
standard VA One ribbed panel formulation, and then
compared to test data obtained on the panel in isolation (no
frames). The agreement in Figure 7 is satisfactory in the
Figure 8. Distribution of RMS velocity over the periodic
frequency range below 600 Hz where the experimental data
cell per unit energy of the panel at 1600 Hz.
are reliable (above 600 Hz, phase mismatch made the drive
point impedance measurements inaccurate). Both the ribbed
panel and the periodic formulations give good results, but
sidewall and floor panels are connected through stiff beam
the periodic section does not predict bands with no modes
components, and it was decided to model those with FE. from the lowest frequencies where discrepancies can be
Three stiff beams below the floor have an I-section 5.5 seen. This may be due to the limited number of modes in
high and 1.875 wide, with wall thickness between 0.08 the panels below 200 Hz. The SEA prediction is fairly good
and 1/8. The beams have holes and reinforcements. The as well, although it seems that the transmission to and from
rail track is 63 long, with an I-section (1 high, 1.75 the floor is systematically under-estimated. It can also be
wide) and the rail on the top of it. It is bolted to the floor. seen that the ribbed panel formulation predicts some bands
The component between the floor and the sidewalls is a with no resonant modes, and this explain the missing data
corrugated plate with holes, with width varying from 5.4 on the pure SEA prediction below 200 Hz. The periodic
to 7, and thickness about 0.033. formulation for the ribbed panel does not predict such stop
The FE subsystems of the hybrid model shown in Figure 9
(left) were built with all connections assumed perfect. It is Mean square velocity, loading Top
made of CQUAD4 finite elements, and contains 6874
nodes. Standard aluminum properties were used. The VA
One inbuilt FE solver was used to extract 426 modes below -30
2700 Hz.




2 3
Figure 9. Hybrid FE-SEA model (left) and pure SEA 10 10
model (right) of the sidewall and floor panels assembly. Frequency (Hz)

HYBRID FE-SEA MODEL OF THE ASSEMBLY Figure 10. Velocity response of sidewall and floor panels
The Hybrid model of the complete assembly is made of the to rain-on-the-roof loading on the top panel.
SEA model of the sidewall and floor panels and the FE Experimental (solid) and predicted by Hybrid (dotted).
model of the beams (Figure 9 left). There are thus four SEA
subsystems and one FE subsystem, connected through Mean square velocity, loading Top
Hybrid line junctions (blue lines). -20

As a reference, a full SEA model of the sidewall and -30

floor panels assembly was built in VA One (Figure 9 right).
The sidewall and the floor panels were kept identical to the -40
ones in the Hybrid model. A lot of details have been
removed from the floor beams to keep one flat plate per I-
section beam; a flat plate connects the skin and the floor -50
(with thickness equal to the averaged value for this part),
and a flat plate describes the seat track. A quick look at the -60
number of modes in band for each SEA subsystem
indicates that almost all subsystem may be correctly -70
described with SEA above 800 Hz.


2 3
10 10
rain-on-the-roof loading on the top sidewall panel, the Frequency (Hz)
measured mean square velocity responses of the panels are
compared with Hybrid prediction in Figure 10. The same Figure 11. Velocity response of the sidewall and floor
test data are present in Figure 11, but the predictions are panels to rain-on-the-roof loading on the top panel.
from the pure SEA model. Experimental (solid) and predicted by SEA (dotted).

The agreement between tests and Hybrid predictions For the forces applied on the I-section beams below
is fairly satisfactory (within 5 dB and right trends) apart the floor, the measured and predicted mean square velocity
responses of the panels are shown in Figure 12 for the updating), and to the models assumption that the sidewall
Hybrid results. Here again, the agreement between tests and and floor panels carry a diffuse field. The first limitation is
Hybrid predictions is satisfactory above 200 Hz. The SEA likely to become more significant as frequency increases
predictions are shown in Figure 13 and are not good, as the while the second one is more present at low frequencies
SEA model of the I-section beam is not able to capture the (below 200 Hz). The second limitation also applies to the
right power input into the system. It can be checked that the SEA model with a larger frequency span as stiffer
number of modes of the beams in the flexure becomes components are described with SEA.
adequate at around 800 Hz (more than 3 modes per
frequency band). This is where the SEA prediction Considering the complexity of the structure, the
becomes more accurate (Figure 13). available amount of information, and the absence of any
forms of updating/tuning, those results are fairly
Mean square velocity, loading Beams Y satisfactory. In particular, they show that an SEA model
-20 with structure-borne transmission can be significantly
improved in two ways: (i) use of the Hybrid method, that
-30 allows to better capture the power input into stiff
components and the transmission through stiff and/or
-40 complicated parts; and (ii) use of the periodic formulation
for the SEA subsystem of the sidewall ribbed panels, where
accurate SEA characteristics can be obtained for
complicated geometries with fair robustness with regard to
the perfect periodicity.

This section briefly describes the Hybrid FE-SEA

3 modeling of a fuselage section made of a ribbed skin panel
Frequency (Hz) and two trim panels connected together through three
frames and some isolation mounts. In the frequency range
Figure 12. Velocity response of the sidewall and floor of interest (below 2000 Hz), the skin and trim panels have a
panels to a unit point force applied on the floor beams. high modal density, while the frames are fairly stiff with
Experimental (solid) and predicted by Hybrid (dotted). few modes. The first point is a restriction for the FE
method due to the resulting excessive number of degrees of
Mean square velocity, loading Beams Y
freedom required and the increasing sensitivity of the
response. The second point is a restriction for SEA which
has low frequency limitations due to the underlying
-30 assumptions.

-40 The benefit of using a Hybrid modeling technique for

this problem is thus twofold: (i) more details can be added
to the description of the junctions (the rubber mounts in
particular); and (ii) the above restrictions of SEA or FE
alone are addressed by the Hybrid FE-SEA method. The
-60 frames and isolation mounts are modeled deterministically
with FE, while the skin-stringer and trim panels are
-70 modeled statistically with SEA. The SEA model of the
ribbed skin is obtained from the periodic SEA subsystem
2 3
10 10
VIBRATION TEST SETUP A fuselage panel was
Figure 13. Velocity response of the sidewall and floor extracted from a Boeing 737 section. It is connected to two
panels to a unit point force applied on the floor beams. trim panels through rubber mounts, as shown in Figure 14.
Experimental (solid) and predicted by SEA (dotted). The overall dimensions of the panel are 38.563 and the
radius of curvature was estimated from measurement to be
Overall, some of the limitations of the Hybrid model 74. The panel comprises 5 stringers and 3 frames. All
may be due to the rough FE description of the beams components are made of aluminum. The panel was covered
(partly due to the lack of exact information, and real with 2 thick melamine foam, inserted between the
stringers against the skin so that direct acoustic
transmission was minimized.

The trim panels are made of 1/32 thick aluminum,

and are curved and corrugated. The overall dimensions are
5819 and the main radius of curvature was estimated
from measurement to be 70. A window with dimensions
1512 is in the middle of the panel. The window is made
of 0.22 thick Plexiglas and is connected to the trim panel
with eight screws through a plastic joint.

Five rubber mounts connect the top of each frame to

the sides of the trim panels. As shown in Figure 14, the
mounts are fixed to the frames with two bolts and to the
trim panels with two washers.
Figure 15. SEA model of the trim panel and window
radiating into a semi-infinite fluid.

The SEA prediction of the modes in band is compared

in Figure 16 with test results on the components in
isolation. The agreement is good for both components in
the frequency range below 800 Hz where the experimental
data are reliable (the drop in the experimental modes in
band after 1000 Hz is not physical). There are less than 3
modes per band in the trim panel below 100 Hz, and it
Figure 14. Sidewall and trim panels connected through
might thus be expected that the SEA model will be of
frames and isolation mounts. Front view of the
limited accuracy below that frequency.
structure in the anechoic chamber (left); and details of
the connections and rubber mount (right).

The structure assembly was suspended in an anechoic

chamber and tested using an impact hammer. The impacts
were located on the exterior side of the skin at 12 points
away from stringers, frames and edges. The acceleration
response was measured at 3 locations on the skin, 3
locations on the trim panel, and 2 locations on the
windows. The transfer functions were measured over the
range [0, 2000] Hz. The measured acceleration transfer
functions were post-processed to get the space-averaged
mean square velocity response of the panels, which was
further averaged over the 12 loading locations and over
third octave frequency bands.


WINDOW Based on the geometry and weight
measurements, an SEA model of the trim panel and
window was built (Figure 15). The trim panel was
described using the singly-curved shell physical property of Figure 16. Number of modes of the trim panel and
VA One, and standard aluminum material properties were window in the third octave bands.
employed, with the mass updated to fit the measured mass
( = 2800 kg/m3, E = 71 GPa, = 0.33). The window was FE MODEL OF THE FRAMES AND RUBBER
modeled as a flat plate with standard Plexiglas material MOUNTS As part of the Hybrid model of the assembled
properties ( = 1500 kg/m3, E = 4.59 GPa, = 0.35). structure, the frames were described with FE since they do
not have many modes in the frequency range of interest.
The rubber mounts were also included with FE to allow a
detailed description of the isolation effect. Those models
were built based on geometrical and weight measurements,
as well as simple characterization tests (first natural measured and predicted response of the skin, trim panels
frequencies of a frame in isolation, and of the rubber mount and windows are shown in Figure 18, 19 and 20,
mounted on a shaking table and supporting a rigid mass). In respectively. The predicted response is plotted with light
all cases, the standard values of material (aluminum and blue lines and the other two curves are two separately
rubber) were initially used and then updated to fit the groups of experimental data. The overall agreement
characterization tests. between tests and Hybrid predictions is satisfactory (levels
within 5 dB and right trends) even at the lowest
The FE model of the rubber mount was merged into frequencies. The response level of the skin component is
the FE model of the frames by enforcing compatibility underestimated, and it is unclear if this could be due to the
between four nodes of the frame and the base of the mount small number of accelerometers used to measure the skin
(thus assuming perfect connection between the response in the experiments. The response of the trim panel
components). The overall view of the frame and five and window is correctly estimated, showing that the FE
isolators and a detailed view of the point connections description of the frames and rubber mounts is able to
between them are shown in Figure 17. The careful capture the significant isolation effect (more than 30 dB at
modeling of the mounts, and in particular of the mass added higher frequency).
by the washers, bolts and screw (20.6 g per mount, when
the mass of the mount is 5.7 g), is a critical part of the
model building, as the isolation effect of the mounts was
seen to be quite sensitive to these details.

Figure 17. Hybrid FE-SEA model of the sidewall and

trim panels assembly in VA One. Figure 18 - Velocity response of the skin to a rain-on-
the-roof excitation on the skin. Predicted response
HYBRID FE-SEA MODEL OF THE SIDEWALL (light blue) and two experimental results.
The Hybrid model of the complete assembly is made of the
SEA models of the trim panel, window and skin-stringer
panel and the FE model of the frames and rubber mounts
(Figure 17). There are seven SEA subsystems (each trim
panel is split in two). The SEA subsystem for the ribbed
skin panels was characterized using the periodic theory: the
modal density, engineering unit to energy ratio, and the
power input from point force have been overridden with the
predictions of the periodic theory.

As shown on the right-hand side of Figure 17, FE and

SEA subsystems are connected through hybrid point
junctions (blue dots). There is one point junction at the base
of each bracket, connecting the frames to the skin; there is
one point junction connecting the washer of each rubber
mount to the trim panels. The windows and trim panels are
connected with an SEA line junction.


the-roof excitation of the skin of the fuselage panel, the
Figure 19 - Velocity response of the trim panels to a REFERENCES
rain-on-the-roof excitation on the skin. Predicted
response (light blue) and two experimental results. [1] R.J. LYON, R.G. DEJONG 1995 Theory and
Application of Statistical Energy Analysis, 2nd Edition,

[2] P.J. SHORTER, R.S. LANGLEY 2004 J.

Acoust. Soc. Am. 117, 85-95. On the reciprocity
relationship between direct field radiation and diffuse
reverberant loading.

[3] P.J. SHORTER, R.S. LANGLEY 2005 J. Sound

Vib. 288, 669-699. Vibro-acoustic analysis of complex


Proceedings of ISMA conference, Leuven, Belgium.
Modeling of structural sound transmission in train
structures using hybrid FE-SEA and EFM analysis


2006 Proceedings of Inter-Noise, Honolulu, Hawaii, US.
Using the Hybrid FE-SEA method to predict structure-
borne noise in a car body-in-white.
Figure 20 - Velocity response of the windows to a rain-
on-the-roof excitation on the skin. Predicted response
(light blue) and two experimental results.
2006 Submitted J. Sound Vib. Characterization of periodic
panels for statistical energy analysis using periodic theory
and finite elements.
[7] V. COTONI, R.S. LANGLEY 2006 Proceedings
This paper has discussed the application of the Hybrid
of ISMA conference, Leuven, Belgium. Advanced
FE-SEA method and the periodic SEA subsystems to
modeling of periodic structures in SEA
various structural-borne transmission problems. The
examples demonstrate that the new methods can provide
[8] VA One 2007 Users Guide, July 2007. The ESI
improvement to existing SEA models. For the problems
considered, the methods are typically several orders of
magnitude faster than a purely deterministic analysis,
although they are more demanding than pure SEA. The
methods are well suited to the analysis and design of
structural-acoustic systems of practical interest.


This work was funded by the NASA Langley

Research Center, Structural Acoustics Branch, Hampton
VI, USA under SBIR phase II contract NNL06AA04C.
The appearance of the ISSN code at the bottom of this page indicates SAEs
consent that copies of the paper may be made for personal or internal use of
specific clients. This consent is given on the condition however, that the copier pay a
$ 7.00 per article copy fee through the Copyright Clearance Center, Inc. Operations
Center, 222 Rosewood Drive, Danvers, MA 01923 for copying beyond that permitted
by Sections 107 or 108 of U.S. Copyright Law. This consent does not extend to
other kinds of copying such as copying for general distribution, for advertising or
promotional purposes, for creating new collective works, or for resale.

SAE routinely stocks printed papers for a period of three years following date of
publication. Direct your orders to SAE Customer Sales and Satisfaction Department.

Quantity reprint rates can be obtained from the Customer Sales and Satisfaction

To request permission to reprint a technical paper or permission to use copyrighted

SAE publications in other works, contact the SAE Publications Group.

All SAE papers, standards, and selected

books are abstracted and indexed in the
Global Mobility Database.

No part of this publication may be reproduced in any form, in an electronic retrieval

system or otherwise, without the prior written permission of the publisher.

ISSN 0148-7191
Copyright 2008 Society of Automotive Engineers, Inc.

Positions and opinions advanced in this paper are those of the author(s) and not
necessarily those of SAE. The author is solely responsible for the content of the
paper. A process is available by which discussions will be printed with the paper if it
is published in SAE Transactions. For permission to publish this paper in full or in
part, contact the SAE Publications Group.

Persons wishing to submit papers to be considered for presentation or publication through SAE
should send the manuscript or a 300 word abstract of a proposed manuscript to: Secretary,
Engineering Meetings Board, SAE.