Need a book? Engineering books recommendations...

Return to index: [Subject] [Thread] [Date] [Author]

Re: [SEAOC] Computer Applications

[Subject Prev][Subject Next][Thread Prev][Thread Next]
>Marlou B. Rodriguez wrote:
> Therefore, the most usefull thing you could do would be to dump your
>Mac's and get some Intel chip computers.
As SEAOC's token liberal (;->) the last thing I want to do is rekindle an
OS war, but I use a Mac in my practice, which is primarily FEA, and I find
Macs to be far more productive than a Wintel platform with far less time
wasted on computer geekery. There are a number of good FEA packages that
run on a Mac. COSMOS/M, INERTIA and LapFEA are three commercial packages
which run on Macs. I use COSMOS, and I have some experience with MSC/Pal
which is a predecessor of LapFEA. COSMOS and INERTIA provide demonstation
versions that can only run small models (50 nodes or so) which are
excellent learning tools. MSC/Pal used to have such a demo; I don't know if
that carried over to LapFEA. The MacSciTech ftp archive
<ftp://ari.net/MacSciTech/> has a demo version of MACSAP/SAP95 and some
other oddments that may be of use. Depending on your needs, the MultiFrame
programs look pretty good, too.

To risk telling you your job (albeit from the perspective of 20 years in
the FEA biz) I think it's a waste of time to concentrate on anything but
the basic elements of FEA in an undergraduate environment. For this you
don't need large-scale software like STAAD or ANSYS, and the process of
learning such programs obviates any but the most trivial demo problems
anyway. Students should know some of the basics of matrix-based structure
analysis, but even more important to know how to validate results and fit
them into a design context. Wading through the input detail in one large
real world 3D analysis wastes time better spent completely dissecting the
results of many smaller problems to show techniques.

I've always thought a neat excercise would be to give students the results
of, say, 5 problems. Diagram the problems to show what is being done, then
tell them that some have input errors and their job is to find the mistakes
and re-run the problems and prove that their answer is correct. Another
interesting approach would be to computerize a student's first course in
structural analysis--basically re-teach the course with finite element
techniques, showing the productivity gains and pitfalls along with the
applicability of first principles.

Christopher Wright P.E.    |"They couldn't hit an elephant from
chrisw(--nospam--at)skypoint.com        | this dist--"   (last words of Gen.
Voice phone (612)933-6182  | John Sedgwick, Spotsylvania 1864)


...