UNCONFIRMED MINUTES OF A MEETINGS OF
SAE Fatigue Design and
Evaluation Committee
Spring 2001 meeting, 3 and 4
April, Doubletree Hotel, Romulus, Michigan
Component Test Division meeting
minutes
DIVISION
PROGRESS MEETING, Tuesday 3 April
Meeting minutes for October 2000 were
approved. We need a volunteer for
vice chairman of the division. Please
notify Paul Lubinski (812-341-2362, paul.lubinski(at)arvinmeritor.com) of
your interests.
Progress Report of ATV.
Ric Mousseau recently volunteered to be "czar" of the ATV
project. Ric gave a presentation
for this meeting on the specific status and plans of multi-body dynamic
modeling. The participants in the
dynamic model project include Ric Mousseau, Ali Fatemi, Nikhil Kulkarni and
Girish Markale (U. of Toledo grad. students), Will Mars, and John Hakala.
The objectives are to develop a simple model to predict ATV durability
loads, and to develop a human response/ driver model, which can supplement DADS
and ADAMS efforts. The fully
parametric model is built in AUTOSIM, a multi-body simulation code that
generates code to run very efficient simulations on Windows computers.
MSC (creators of AUTOSIM) have provided for free distribution of the ATV
model to FD&E members.
This rigid body model initially has
used revolute joints for the bushings but will later use force elements.
Linear force is assumed for the tires; only vertical forces are included
initially, but longitudinal and lateral forces will follow.
Shock and spring forces are defined linearly with coefficients, but data
tables to define nonlinear behavior will be used later.
The suspension has 9 DOF and is assumed rigid.
The model inputs are vehicle speed and left & right road profiles.
The outputs are vehicle component displacements, velocity, and
accelerations, including a time history format of these outputs.
Results have been generated for a
single impact event, consisting of impact with a 3m x 0.1m trapezoid-shaped bump
on the left side of the vehicle. The
outputs include time histories of upper and lower ball joint acceleration, and
upper and lower control arm restraining moment.
A basic ATV model
has been built in AUTOSIM, consisting of a rigid suspension and a four-post
simulation of vertical loading at the tires.
Outputs include forces on the tires, springs, shocks, and joints.
The model does not yet have steering.
The model is posted at fatigue.org, and animation files have been
completed.
The next steps are to add steering, to
account for driver dynamics, to generate additional acceleration and force
outputs, to add jounce-rebound stops, to add lookup tables for shock data, to
clean up parameter sets, and to improve tire modeling.
Improvements to the tire model can include use of Cooper’s data, use of
a lookup table for the tires’ lateral characteristics, and implementation of
an enveloping tire model.
Questions and answers at the end of the
presentation: Regarding publication of ATV work, the response was there would be
a special publication about a year from now. Regarding the idea of returning to the practice of a separate
day in the FD&E meeting devoted to the ATV, the response was that lots of
work is accomplished on personal agendas, and we would not want to create
another obstacle to such work.
Regarding the approach to use for a
driver model, based on previous published work, the response was that the model
would be based on UMTRI’s “path-following” approach.
There was also the question of whether the U. of Iowa’s multi-body
driving simulator could be of assistance for this project.
Technical presentation
"Considerations of Variability and Uncertainty in Component Life," by
Dan Lingenfelser.
“A single answer to a set of inputs is of limited value without an idea
of the variation.” There must be consideration of statistical aspects, such as
group-to-group variations from a single source and source-to-source variations.
Cyclic testing of bushings was shown as a case of the need to keep
variability and uncertainty effects separate.
Comparison tests are used for decision-making, such as whether to make a
part change for cost savings. We
need to have confidence in the data; this means answering questions such as who
ran the test and whether all equipment is calibrated.
There are other possible questions of uncertainty, such as: Do we have
the right model? How can the data
help manage risk? Should we use
load-vs.-life or stress-vs.-life? Some
sources of variability are loads and sequence effects, material properties
(hardness, fatigue and toughness), and residual stress.
A cumulative distribution function can be generated to quantify
variability, but what is the uncertainty? If
uncertainty is properly assessed, then it is possible to generate confidence
bands and look at sensitivities to determine where to spend time and money.
A single life prediction without estimates of uncertainties is of lower
value than a range of predictions (not a very precise range at that) with known
uncertainty. Uncertainty must be quantified in each step in the process.
Questions and answers at the end of the
presentation: Regarding environmental influence, Dan stated that it is another
uncertainty. Different uncertainty bands can be indicated for underwater
conditions, sand, mud, etc., as opposed to a more consistent environment.
Regarding accounting for corrosion, the response was that there was an
approach of trying to eliminate corrosion, rather than trying to calculate
corrosion’s effects.
There was a final comment from the
audience that there is a practice to put specifications on variables of
materials/processing such that we do what we can to achieve desired fatigue
properties, but the fatigue properties themselves are not specified directly.
Technical presentation on University of
Illinois ATV efforts, by Ge (Jerry) Wang.
The project objective is to develop methods using computer simulation for
accurate prediction of structural fatigue life.
The structure being studied in this project is the ATV frame and
suspension. The project plan was to
set up analytical boundary conditions like those of the test fixture, then to
perform multi-body dynamic analysis, static stress analysis, and fatigue
analysis, then to run tests after completion of the analyses.
The test fixture was fixed at the rear and right side, while the left
side was loaded in two directions. First
the shock absorbers were studied for their non-linear reactions.
A rigid body simulation incorporating the suspension showed much higher
natural frequencies (e.g. 83 Hz) than the excitation frequencies.
Flexible body analysis was also performed.
Reaction forces were determined from ADAMS. Multi-body elements such as
revolute joints and linear actuators were used for the boundary conditions of
the FE model. After three
high-stress locations of interest were found on the frame from the FE analysis,
fatigue analysis was performed with the assumption of class F2 from British
Standards 7608 for the weld joint properties.
Application of different load magnitude/ direction combinations led to
the generation of a fatigue “map” of lives vs. given combinations of
vertical and horizontal input force. The
next step is durability testing!
One final note from the division
progress meeting: a solid model of the lower left
control arm of the ATV is available on the fatigue.org web site.
COMBINED
PLANNING SESSION FOR COMPONENT TEST DIVISION AND ROAD LOAD DATA ACQUISITION
DIVISION, Wednesday 4 April
Recap of ATV status:
Original mission statement still seems to be valid. Who’s doing what: Ric Mousseau, Ge Wang, etc. have been
working on multi-body dynamic modeling; Dan Klann, Ge, and GM (Zhengxian?) have
been working on FE modeling (plus Ford solid modeling of control arm).
Questions from the floor were what are the tangible outputs, how can the
results be used toward education, what is the benchmark set of data, and what
should be published. Ric will review the project plan and will guide the project
to the point at which there will have been (at least) one attempt to complete
each step in the durability simulation process, so that there will be something
available for all interested experts to review.
Other initiatives of the Component Test
division:
no new ideas came forth.
RLDA division ASCII data format
initiative:
the group consensus is not to pursue this activity any further.
Other initiatives of the Road Load Data
Acquisition division: no new ideas came forth.
Statement of goals/objectives for the
committee and divisions, in preparation for FD&E’s possible move from
Technical Standards to Conferences/TOPTEC’s in the SAE hierarchy:
no suggestions came from the floor, so executive committee members feel
empowered to formulate this statement by themselves!
Reminder: both the
RLDA and Component Test divisions seek volunteers for technical presentations at
the fall meeting, and the Component Test division seeks a vice-chair.
Respectfully submitted,
Paul Lubinski, Chairperson, Component Test division