Fatigue Design and Evaluation Committee |
FD&E HomeSearchEvents CalendarShort CoursesMinutes, Presentations, AgendasContact InfoATV ProjectWeld ProjectBolt ProjectDivisionsData OnlineLinksSteering Commitee
|
UNCONFIRMED MINUTES OF A MEETINGS OF SAE Fatigue Design and Evaluation Committee Component Test Division Holiday
Inn - Urbana, IL (11-12 October 2000) Wednesday,
11 October - Divisional Progress Meeting Raj
Thakkar began the meeting at about 10:30 a.m.
The minutes from the meeting held 4 April 2000 in Cedar Falls, IA were
approved. ATV
multi-body simulation: The first technical presentation for the division was by
Ric Mousseau of the University of Toledo, on progress in multi-body dynamic
simulation of the ATV. The model is
now running and just needs some parameter "tweaking."
The rigid-body, hard-constraint model is run by software called "Autosim,"
which is not the simulator itself, but instead a generator of source code for
the simulation. The model outputs
acceleration, velocity, displacement, and force time histories.
The road inputs are displacements on the simulated tires, which are
modeled as springs. Data related to
the ATV model can be freely downloaded via www.mime.end.utoledo.edu/people/faculty/rmousseau/Atv/ATV.html.
"Zip" files of the model input, the executable, and related
output plotting utilities are available at this site.
Ric is starting to document simulator usage and will add those documents
to the noted web site. Ric went
through one simulation example, specifically the ATV going over a trapezoidal
bump. The steps in creating the input file of parameters (time
steps, component masses and stiffness, vehicle speed, bump geometry, etc.), in
specifying output plotting, in executing the simulation, and in extracting the
output were noted. The loose ends
needing to be tied up for the ATV model are as follows: adding shock data and
jounce/rebound stop info, adding steering, adding capability for additional
force and acceleration outputs, improving the tire model, and interfacing the
model with the multi-body program "Carsim-Ed." Empirical
dynamic modeling: The next presentation was by Tom Renner of MTS, on Empirical
Dynamic Modeling (EDM) for non-linear suspension components.
EDM is a "black-box" model, a trained neural net that uses
random excitation on the component to build a model of the response.
The foremost application of EDM is representation of a component like a
bushing or tire in multi-body models (i.e. the kind solved by ADAMS). The excitation input parameters (amplitudes, frequencies,
etc.) should be within the "bounds" of what the part would actually
experience. Examples were presented
for a shock absorber, a suspension bushing, and a tire. In each case, EDM showed well-correlated representation of
force vs. displacement in terms of both hysteresis (which the polynomial
"white-box" model is not good at capturing) and non-linearity (which
the FRF "white-box" model is not good at capturing).
Someone in the audience asked how well EDM would handle "completely
different signals," which could probably be taken as meaning either model
response to excitation significantly different from the content used to build
the model, or the ability to simulate a part that is significantly different
than the one on which the model is based. Tom
responded that this issue needed further investigation.
MTS' Andrew Barber has led the development of the EDM method.
Further details about EDM can be found in several publications, including
SAE paper 2000-01-1625. Cold
expansion of holes: Jude Restis from Fatigue Technology, Inc. gave a
presentation called "Fatigue Life Enhancement Using Cold Expansion."
(cold expansion abbreviated Cx in the remainder of these notes) Cx has primarily aerospace applications, but has had, and can
have other applications as well, such as train rail end bolt holes, bridge
fastener retro-fits, and biomedical structures (like nails in femur
reinforcements). The typical
mandrel process for Cx of a hole introduces a zone of compressive residual
stress ~ one hole diameter wide; this compressive residual stress is a key to
improved fatigue performance. S-N
studies show Cx performance not far below that of parent material without a
hole; this is especially true if a fastener is inserted in the hole with a fair
amount of interference fit. Derivative
processes include rivet-less nut plates and high-interference bushing
installation. Press-fit
problem: The division's last presentation was by Ric Leist from Cessna,
regarding press-fit fatigue of lugs (the aircraft kind, not what's holding the
wheel rims on your car). Simulation
of the press-fit joints is a nasty contact mechanics problem, addressing the
fact that loading of a given direction relaxes the interference fit on one
"side" of the hole. Models
of this problem have involved lots of empirical data for different lug
configurations. The models have
worked well for some lug materials (steel and titanium), but no so well for
others (aluminum). The presentation
was effectively a plea for assistance for additional data or ideas that could
help such problem-solving. Thursday
12 October - Divisional Planning Session ATV:
Anyone with data or progress to report should give their information to Raj
Thakkar. He will follow up with
those persons regarding next steps. The
remaining discussion about the ATV focused on control arm testing.
Should the testing be conducted on an arm by itself or on a corner
assembly? The web site wheel load
data may be analyzed to determine test time histories with spindle loads.
Because the road load data still exist in the "spinning wheel"
form, the histories would still need to be resolved into forces.
The web site information has a spreadsheet including instructions for
making such calculations. It may be
possible to use old SAE cantilever fixtures at Ford or Deere to aid testing.
We may consider running uniaxial tests, which would create an opportunity
to choose from more facilities for testing.
Regardless of the test plan we choose, WE STILL NEED PARTS.
The money and parts that the U. of Toledo had have been used up. Variability
and uncertainty studies: Charlie Sieck will find a well-populated material data
set and make it available to any committee members interested in making
variability and uncertainty calculations "round-robin" style.
Chin-Chan Chu may also contribute to this effort the multiaxial data used
for publication AE-14. Charlie will
make a presentation on this and related matters at the next meeting.
Anyone interested in helping with studies on variability and uncertainty
(perhaps forming a task group?) is encourage to contact Charlie or Paul Lubinski. If
anyone is interested in volunteering wherever the project tasks require help or
if there are any questions, please contact Paul Lubinski or Raj Thakkar as soon
as possible. Respectfully
submitted, Paul
Lubinski Chairman,
Component Testing Division SAE Fatigue Design and Evaluation Committee Fatigue Life Prediction Division Unconfirmed
Minutes Holiday
Inn - Urbana, IL (11-12 October 2000) Wednesday,
11 October - Divisional Progress Meeting April
2000 minutes were approved. C.C.
Chu mentioned that for whatever reasons, people seem to be unusually
enthusiastic about presenting their work results this time.
As a result, she had scheduled 3 talks in this 1 hr. session.
In order to not steal too much time from later sessions, only 1 burning
question will be allowed after each talk. Further
discussion about the subject was encouraged to carry on during breaks.
Furthermore, some work update about the Waterloo web site would be
presented by A. Conle in the planning session. "A
Stress Based Fatigue Prediction Concept and Its Application to Vehicle
Components," by Helmut Dannbauer, FEMFAT. Mr.
Dannbauer presented a macroscopic phenomena based fatigue approach taken by
FEMFAT. Unlike the common approach
most of the FD&E members take, where we select a damage parameter such as
SWT's, calculate the SWT-value for a damage event based on the stress analysis
result, and use a UNIQUE SWT versus life curve to assess fatigue damage for the
event, FEMFAT uses the stress amplitude as the damage parameter, and let the S-N
curve of the material to vary with loading conditions. The conditions influencing the S-N curve, based on their
experimental data as well as experience, include stress gradient, surface
roughness, mean stress, etc. They
do use the Neuber correction method to obtain the local stress.
Mr. Dannbauer then showed several successful examples using this
approach. "Thermal
Fatigue of Automotive Components," by V. Ogarevic, Ncode. Dr.
Ogarevic gave a concise overview of the complex nature of thermal fatigue
problems of automotive components. He
pointed out that it is well known that both the rate sensitivity of
stress-strain behavior and the S-N curve vary with temperature.
The temperature dependence can be relatively easy to determine.
The interaction between thermal and mechanical stresses however is much
more difficult to test and model. "Crack
Energy Density as A Fatigue Life Parameter for Rubber Subjected to Multiaxial
Conditions," by W.V. Mars, Cooper Tire. Mr.
Mars reported that for rubbers fatigue failure prediction based on his new
'crack energy density' (CED) parameter seems to better correlate with
experimental results in particular the selection of failure plane.
The crack energy density is the inner product of the traction and the
strain increment on a potential failure plane.
He also succeeded in using a phase-dependent (CED)-life curve to predict
fatigue life for various amount of multiaxiality. Thursday
12 October - Divisional Planning Session Chu
mentioned that some overload test data had been posted by J. Bonnen on site
http://fde.uwaterloo.ca. R.
Chernenkoff will pose his mean stress tests on the site soon. Those who had promised to test their fatigue method, Deere,
Cat, Ford should go see whether the information posted is enough.
We expect to report the prediction results next meeting.
The FEMFAT team indicates that they are eager to participate as well. Al
Conle gave a short tour for his new web tool page, also on the "fde.uwaterloo.ca"
page. One can use the tool
to feed in rainflow counting results and predict fatigue life by choosing one of
the 3 materials of which Al has supplied digitized strain-life curve. We did have a roomful of blank faces staring back at us.
It is assumed that people need time to go try it out and digest the
concept a bit. Chu
then closed the session by outlining the next meeting's likely agenda -- life
prediction capability comparison for problems involving large mean stresses;
more tools on the waterloo web site; and possible talks on subjects that may
interest both the ground vehicle and the aerospace industry. Respectfully
submitted, C.C.
Chu Chairman,
Fatigue Life Prediction Division SAE Fatigue Design and Evaluation Committee Meeting
Road Load Data Division Unconfirmed
Minutes Holiday
Inn - Urbana, IL (October 11-12, 2000) Wednesday,
October 11 - Divisional Progress Meeting 1.
Ray Thompson moderated the meeting. 2.
Minutes from previous meeting were approved. 3.
Mark Earley presented a review of the load measurement project on the
ATV. Several test runs have been
made, many over discrete objects for model correlation purposes.
The data, along with photos and video of the test, have been placed on
the web server. 4.
Chakrapani Vallurupalli gave a technical presentation entitled
"Functional Virtual Prototyping in Fatigue, Design and Evaluation
Process." 5.
Ray Thompson lead a discussion about the web site www.fatigue.org. This
is the site maintained by Christoph Leser.
The committee member's general consensus is that the above site
complements the official SAE web site. The
fatigue.org site is easier to get to (no password required) and easier to post
information. Any updates or suggestions about the www.fatigue.org web site
should be directed to christoph.leser(at)mts.com. Thursday,
October 12 - Divisional Planning Meeting Ray
Thompson chaired the meeting. Load
data collected on the ATV was discussed. Mark
Earley had collected additional data while operating over discrete objects and
test tracks. The time history data
has been placed on the web site in dac file format.
There are 1886 data files using about 1 GB of disk space.
Mark will do some data reduction (noise spike replacement, drift removal,
etc.). He will also do the
conversion of the wheel forces and moments from rotating coordinates to
non-rotating coordinates. Run
descriptions, channel descriptions, vehicle configuration details, photos and
video have been posted on the web site. An
Excel spread sheet documents the details. All
the information can be found
at www.fatigue.org/data. There
was some discussion about the process of data reduction of the ATV data.
Charlie Sieck and other individuals would like to do this procedure as
well. This would include
statistical analysis of multi-pass data. The
raw data will be sent to them upon request to Mark Earley.
It is desired that anyone participating in this exercise present their
results at the next meeting. Ric
Mousseau has plans to use some of the discrete bump data to correlate his
dynamic model. Description of the
test track profile and all discrete objects is included in the Excel spread
sheet posted on the web. ASCII
data file format was not discussed since the individuals working on this were
not present. The
meeting was adjourned. Respectfully
submitted, Ray
Thompson Vice-Chairman,
Component Testing Division SAE Fatigue Design and Evaluation Committee Meeting
Structural Analysis Division Unconfirmed
Minutes Holiday
Inn - Urbana, IL (October 11-12, 2000) Wednesday,
October 11 - Divisional Progress Meeting Dan
Lingenfelser substituted for Zheng Xian Bai and Mary Wickham as chairman.
Two technical presentations were provided.
Jeff Sundermeyer presented work he was doing on "Ground-Interface
Load Inference" and Mark Barkley presented his work on Spotweld Fatigue
Under Multiaxial Loading Conditions. (overheads
attached) Structural
Analysis Division Planning Session Oct. 12,2000 Dan
Lingenfelser substituted for Zheng Xian Bai and Mary Wickham as chairman of the
planning session, John Hakala kept notes..
Attending:
John Hakala, Dan Lingenfelser, Dietmar Peishammer, Jin Qian, Ed Lu, Ge Wang,
Nishit Dey, Jerry Green, Ric Mousseau, Hong-Yan Zhang, Michelle Wegsheid,
Charlie Sieck, and Darrell Socie. Dan
showed a slide summarizing the Structural Analysis Division planning session
from the April meeting.
Dan Klann, Todd Vest and ZhengXian Bai were to exercise the ATV model
that was on the website. The website model was built two years ago and gives
warnings. The model evidently still
needs some cleanup and expertise applied. Students
exercising the model would end up with similar problems. The
question asked was "Is it a waste of time to run the model without the
refinements?" John
Hakala will send Dan Klann an e-mail to determine if he has done anything to
clean up the model. Jin
Qian and Hong -Yan Zhang from John Deere T.C.
will also contact Dan Klann and work on refining the ATV model. MultiBody
Dynamics - Ric Mousseau from the University of Toledo has completed the basic
model without bushings. Engine and frame are an integral model. Ric needs to
update the tire model, and set up to back calculate the loads.
Darrell Socie mentioned there is quite a bit of variation from ATV1 to
ATV2, with different parts on similar vehicles. We need to compare to the John
Deere measured data. Ric will work
with Darrell's post doctorate student. Ric's
objective is to sanction the dataset with representative C.G. locations, mass
and moment of inertia, determine how simple can you model the vehicle, and still
obtain reasonable results. John Hakala will work on a sensitivity study with
Ric's model. Ge
Wang mentioned an ADAMS model and
will compare to Gary Mauritzon's results, and determine shortcomings. Charlie
Sieck mentioned the possibility of running a ProMechanica model similar to Ric's
MBD model and generate some feedback about modifications. We
need to demonstrate the ability to predict loads. Does the engine mass
and stiffness make a difference? Does the luggage rack make a difference?
|
