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

SAE TECHNICAL 2000-01-3279

PAPER SERIES E

Nonlinear Finite Element Study of the Windup

Geometry of a Parabolic Front
Suspension Leaf Spring
Felipe Nogueira, Ricardo R. Teixeira, Oscar M. Ueda and Edson A. Yokoyama
Volkswagen Caminhões e Ônibus

IX Congresso e Exposição Internacionais

da Tecnologia da Mobilidade
São Paulo, Brasil
3 a 5 de outubro, 2000

AV. PAULISTA, 2073 - HORSA II - CJ. 2001 - CEP 01311-940 - SÃO PAULO – SP
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

The appearance of the ISSN code at the bottom of this page indicates SAE’s
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
Department.

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 2000 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.
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

2000-01-3279

Nonlinear Finite Element Study of the Windup Geometry of a

Parabolic Front Suspension Leaf Spring
Felipe Nogueira, Ricardo R. Teixeira, Oscar M. Ueda and Edson A. Yokoyama
Volkswagen Caminhões e Ônibus

Copyright © 2000 Society of Automotive Engineer, Inc

ABSTRACT Mathematically, windup stiffness and angle of a

symmetrical spring under brake load are defined as shown
Windup behavior is particularly dangerous in a in equations 1 and 2, respectively:
commercial vehicle, since it can generate undesired lateral
movements. Methods previously applied to determine this kL2
behavior were not cost nor time efficient. A new proposed ω=
method that applies CAD/CAE tools was presented in a 4
previous paper, but that work still relied on a correction (1)
factor that is based on a experimental test. Therefore, the
present study is geared at eliminating these experimental T
tests, through the use of a suitable mathematical model and V ≈ tan (V ) =
nonlinear analysis.
0.25kL2
(2)
BACKGROUND
Where k is the vertical rate of the spring, L is the overall
The torque in the vertical longitudinal plane of leaf length, T is the applied torque in each wheel of the front
springs, generated either by drive or brake forces acting on suspensions, ω is given in kgf.mm/radians and V is given in
a plane above or below the spring seat, is the so called radians [1, 2 and 3].
windup torque. Figure 1 shows the typical S-shaped windup
deformation in a commercial vehicle suspension [1].
Windup is particularly dangerous in commercial
vehicles since the suspension not only acts as an isolating
mechanism, but also as an attaching linkage that affects the
steering geometry. This could result in an undesirable
lateral movement of the vehicle due to shifts in steering arm
displacements [2].
This lateral movement comes into place if the path
followed by the steering arm under windup differs
significantly from the path followed under only Figure 1: Typical "S"-shaped windup deformation in leaf
rebound/jounce loads. Figure 2, in the last page, shows springs [1].
schematically how windup can effect the steering geometry,
inducing vehicle lateral movement under brake torque. A quick inspection of Equations (1) and (2)
That figure portrays a leaf spring, in perspective indicates that spring length is the variable that renders more
view, subjected to jounce/rebound and windup. If windup is sensitivity to the windup behavior. The longer the spring,
sufficiently large, the path followed by the steering arm the smaller the windup behavior is. The other variable that
shifts towards the left (or right depending on the design), could effect wind up is vertical rate, which is also closely
resulting in involuntary and dangerous vehicle lateral related to the spring length.
movement. There are acceptable limits to this lateral Although these linear estimates shown in
movement, so a proper design should be constrained within Equations (1) and (2) are error prone, they provide
these limits. appropriate accuracy for most cases. There are, however,
Several ways to determine the windup geometry in situations under which the windup behavior cannot be
the design stage of a project, as discussed in references [1] ignored, such as springs with small vertical rates. In such
and [2], do exist. However, most of them are neither cost cases, proper attention to windup behavior must be given.
efficient nor accurate enough. So trial and error must be
employed in order to arrive at a suitable design, implying
that several mechanical prototypes must be built and tested
[2].
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

PREVIOUSLY PROPOSED METHOD would also carry a level of uncertainty to the CAD
environment.
Reference [2] proposed a new method based on Therefore, the main objective of this paper is to
CAD-CAE tools that included a linear finite element present the development of a mathematical model, applied
analysis coupled with an optimization in the CAD in a nonlinear FE analysis, that could improve this
environment. This proposed method relied on an initial test methodology. This became mandatory in order to eliminate
to estimate the windup behavior of a similar leaf spring to the need of the initial experimental test coupled with the
the one being developed. Then, a linear finite element (FE) correction factor. The aim is to produce a correct initial
analysis was conducted, and a correction factor to the FE impression of the windup behavior, so that the optimization
analysis would be determined based on the comparison of can become less dependent on designer expertise. The
the FE analysis and the experimental test [2]. The primary modified flow of actions is shown schematically on Figure
assumption here is that the error between linear FE analysis 4, below:
and experimental data is small.
Once this initial windup behavior was understood, Certification
the design would be optimized in the CAD environment. and
Basically, this optimization consisted in altering spring validation
length in order to assure that the path of the steering arm
under windup would not differ significantly form the path
under rebound/jounce. If necessary, other elements of the
spring, such as spring eye, could also be altered in order to YES
achieve proper windup behavior. Figure 3 shows
schematically the actions required in this methodology.
Is result
satisfactory ?
Certification NO
Windup geometry
and behavior
validation
CAD optimization of Analytical + Non
YES steering/suspension Linear CAE

Is result
satisfactory ?
NO Correct Windup
Behavior

CAD optimization of Analytical +

steering/suspension Linear CAE

Figure 4: Schematic representation of present methodology.

Correction
Factor to adjust
angle Using the nonlinear FE analysis, much of the lack
of accuracy in the methodology can be overcome. In
addition to that, the method would become completely
Figure 3: Schematic representation of previous virtual, eliminating the need of an initial test to assess
methodology. windup behavior of a similar leaf spring.
The following sections explain the nonlinear
This optimization is as good as the initial windup analysis and the models constructed. It also presents
behavior provided to the CAD environment, and very comparisons of linear and nonlinear FE studies and
dependent on the designer expertise. So if the first estimate validation data.
is not close to a real windup behavior, all work performed MATHEMATICAL MODEL
under CAD environment is compromised.
Since small windup behavior can go unnoticed, the First, it is necessary to realize that the changes in
accuracy of the method was sufficient to arrive at initial geometry, due to the applied loads, are sufficiently large so
values for the CAD optimization. However, the major that the load x deflection characteristics of the structure is
drawback of this proposed method is that it still relied on an also altered. That is, the state of equilibrium of the structure
initial experimental test to determine the windup behavior is affected by the changes in geometry. This is due to:
of a similar leaf spring.
Moreover, the correction factor determined later • Large displacements; and
on in a linear FE study was not as small as predicted, and it • Changes in strength.
was applied to a range of similar designs. Additional
corrections were also not performed, since that would Therefore, geometric non-linearity must be taken
demand additional test, thus the initial windup behavior into account, if the study attempts to achieve proper
accuracy. The accuracy target was set at 5%, since better
accuracy would not provide significant improvements in the
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

overall method, and since the experimental results to which But this could bring an error to the model. Since
the FE analysis would be compared also carried an error. the thickness is equally distributed above and below the
An iterative solution scheme is necessary due to additional element neutral line, a misplacement of the neutral line
terms that must be considered in the global stiffness matrix. could occur in the model. So proper mesh construction was
A shell model was constructed with careful carried out to guarantee that the tapering occurred at the
attention being given to details, as shown on Figures 5 and leaf´s compression side.
6. The following features were included: The solid model used in linear analysis could have
been used. However, solid models usually require more
• Spring shackle; computational effort, both in cost and time. So the shell
• Contact between leafs, where applicable; and element was first studied and, if the results were not
• Clips in both front and rear portions of the spring. accurate, a solid model would also be studied in the
nonlinear FE analysis.
The prototype parabolic two-leaf spring modeled
had the following properties: VALIDATION EXPERIMENTS CONDUCTED
• Length L: 2240 mm; and
• Vertical rate k: 11 kgf/mm2. In order to explain the comparisons made at later
sections, the data from validation experiments is presented
here prior to the results of the FE analysis. Later sections
will discuss the results thoroughly.
Experimental analysis conducted in a prototype
spring were performed and discussed in reference [2]. One
of the experimental analyses consisted in applying four
strain gages to the tension surface, as shown on Figure 7 on
the next page, of the first leaf and submitting the vehicle to
a previously defined "panic-stop" braking test [2].
The Wind-up behavior was then inferred from the
deformation data on each sensor, since these deformations
describe two tangent arcs through which the spring
deforms. The angle of the spring seat (tangent to both arcs)
and a horizontal line is the windup angle. This is based on
the following relation of curvature and deformation for a
prismatic beam:
e
ε =−
Figure 5: Detail of shell model - front spring eye. 2⋅ ρ

Elements with tapering in thickness were applied, (3)

due to the parabolic nature of the spring under study. The
element thickness in the shell model was obtained through Where ε is the beam deformation on a given location, e is
the use of a "data surface", a software option which allows the beam thickness on that given location and ρ is the
the thickness to vary over the element according to a user- resulting local radius of curvature [2] [4]. From this study,
determined function. By doing so, the proper thickness the windup angle on the spring seat was experimentally
tapering can be achieved in a shell model. determined to be 3.17°.
Figure 6: Detail of shell model - rear end and shackle.

Figure 7: Strain gage positioning [2].

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

RESULTS FROM SHELL NONLINEAR MODELS This virtual loop acts integrating CAD and CAE
tools to optimize windup suspension and drive geometry.
Linear mathematical models with solid elements Furthermore, there is no longer the need of an initial test to
could only predict a windup angle at the spring seat of 5.2°, estimate the windup behavior and the correction factor to be
which is clearly not accurate (see reference [2] for details). applied to the FE analysis.
The non-linear model, as expected, provided a much better Since the methodology has already been proven in
representation of the windup behavior in parabolic leaf previous works, it is safe to say that the only test that must
springs. be conducted is the final validation test. This eliminates the
Table 1 shows the comparison between the windup need of intermediary prototype tests, reducing costs and
angle predicted in the linear and nonlinear FE models, and a shrinking development time.
comparison to the experimental results, as obtained in the Figure 9 shows a deformed plot of the spring under
test mentioned in the previous section. This comparison is windup given by the shell nonlinear model. The
made in the form of a percent error having the experimental displacements are exaggerated for clarity, but it is possible
result as benchmark. It should be reminded that the to see that the peculiar "S-shaped" deformation pattern is
accuracy target of the FE analysis was set in 5%. well represented. The curvatures extracted form this
deformed model can then be fed in the CAD environment
for the correct initial windup behavior.
Model Windup Angle % Error
Experimental Value 3.17 Benchmark
Linear FE 5.17 63.1%
Non-Linear Shell FE 3.05 3.7% Deformed

Table 1: Comparison of windup angle values predicted and

measured in the spring seat region.

It is clear that the computational effort involved in

the linear solid model FE calculation is considerably
smaller than the effort needed in a non-linear FE analysis.
But Table 1 and Figure 8 show that the accuracy of the non-
linear FE study does make up for the extra computational Undeformed
effort.
In Figure 8, a graphical comparison of both linear
and nonlinear answers is shown. The nonlinear curve Figure 9: Deformed nonlinear model (not to scale).
portrays the stiffening behavior of the spring due to the
geometry non-linearities. From that graph it is possible to Therefore, if a non-linear model is used, the
see that the changes in geometry alter the vertical rate of the windup behavior can be accurately predicted, thus
spring as well as the effective length, thus altering eliminating the need of a first experimental test and the use
significantly the windup behavior. Linear FE analysis can of correction factors. So, the entire optimization loop
only predict the correct windup behavior for small loads.
6,00
becomes virtual, and the only tests that still must be
performed are the certification tests.
5,00

CONCLUSIONS
Windup Angle (degre)

4,00

3,00 This paper discusses an improvement to a method

applied in the determination and design of front suspension
2,00 coupled to steering geometry under windup effect.
Previously, a linear FE study was applied along with a
1,00 correction factor in order to feed an optimization of the
suspension geometry in the CAD environment.
0,00
Although this method is a clear improvement on
0 0,2 0,4 0,6 0,8 1
Time Step
the trial and error approach usually taken, it still relied in an
initial test. This test intended to obtain a determination of
Nonlinear Linear
the windup characteristics of a spring similar to the one
under study.
It was believed that the linear FE would have to
Figure 8: Comparison of linear and nonlinear results. suffer only small corrections to provide results comparable
to the initial test, which was not true. The correction factor
It is also clear that the nonlinear shell model does used for the spring under study was close to 65%, rendering
provide the required accuracy level, with relative low the method unreliable if a new design would have to be
additional computational effort. Thus, a true virtual studied.
optimization loop can be achieved.
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

So this paper comes in as an attempt to develop a Using CAD-CAE Tools; 1o Colloquium de Suspensões
mathematical model, coupled with a nonlinear analysis, that - SAE/Caxias do Sul; 2000.
would provide better accuracy at relatively modest
3. Hoesch Hohenlimburg AG; Warmgeformte Federn –
computational costs. By doing so, the need of the initial test
Konstruktion und Fertigung; 52. IAA in Frankfurt;
would vanish, once the correct windup behavior would be
1987.
obtained, and passed on to the CAD environment.
A shell model was proposed instead of a solid 4. Dally, James W.; Riley, William F.; Experimental
element model used in the linear FE analysis. The results Stress Analysis; 3rd Edition; 1991.
from this model were then compared to the solid model
linear FE study and experimental results from validation
tests.
Table 1 and Figure 8 show this comparison in a
very straight and forward way. From the table it is possible
to see that the error from the nonlinear study is within the
preset target of a 5% maximum error. This error could be
made smaller if additional physical phenomena are taken
into account, such as bushing stiffness and spring shackles.
However, the gain in accuracy would not be as
significant. The bushing stiffness would have to be
determined experimentally, what would be time consuming.
Furthermore, the experimental validation tests that were
used as benchmark also carry an error inherent of the
approach taken. Thus the additional effort would not
compensate the gains in accuracy.
Figure 8 shows why the linear FE study cannot
provide accurate results. The geometry changes induce a
very noticeable stiffening behavior of the system that is not
captured in the linear study. In other words, as the load
increases, less deformation is taking place than the one
calculated by the linear FE.

Figure 9 shows a deformed plot of the spring

under windup. It is possible to see that the peculiar ”S-
shaped” deformation is clearly captured in the nonlinear
study. And this was another drawback of the methodology
presented in reference [2]. Since the linear FE study could
not capture the true deformed shape, the method still relied
on the expertise of the CAD operator for its success.
By portraying more accurately the spring
curvatures that give rise to the “S-shaped” deformation, the
methodology becomes less dependent on the expertise of
the CAD operator. This impels the optimization loop
further towards the virtualization.
Moreover, since the methodology has already been
successfully applied previously, it is safe to say that a true
virtual optimization loop can be achieved in this fashion.
By following the steps shown on Figure 4, one can analyze
a spring under windup and arrive at a design that is not
sensible to it.
So the main goal of this work and the goal of
reference [2] has been achieved. A virtual optimization loop
for the design of windup-free front suspensions that can be
achieved without any additional tests.

BIBLIOGRAPHY

1. SAE, Inc.; Manual on Design and Application of Leaf

Springs – SAE HS J788; SAE; Warrendale, Pa; 1982.
2. Nogueira, F.; et al; Theoretical-Experimental
Development of Steering Geometry under Windup
Torque in a Parabolic Leaf Spring Front Suspension
 Downloaded from SAE International by University of Michigan, Sunday, July 29, 2018

Normal path under

rebound/jounce

Pitman arm displacement path

Vehicle lateral movement

resulting from noticeable
windup behavior.

Path of of jounce/rebound
under windup influence.

Figure 2: Schematic representation of windup effect.