Reply-To: Stallion-feedback-2@lb.bcentral.com
From: "Aircraft Designs, Inc." <Stallion@lb.bcentral.com>
Subject: Design Classes
AIRCRAFT DESIGNS CLASSES IN MONTEREY, CA. Sign up now by calling
831-649-6212 while space is still available.
AIRCRAFT DESIGN using FINITE ELEMENT METHODS
January 12 and 13, 2002
The latest aircraft design technology, materials, and computer programs
are discussed. Run: Performance, C.G. A.C., Stability calculations, V-n
diagram, Wing Sizing, Fuselage Plot, Fuselage Loads, and the ALGOR finite
element analysis program. Each student will receive a copy of the full
blown ALGOR fea program, 200 nodes, that can do statics, dynamics, and
composites, and all new aircraft design programs. Fee: $495 includes lunch
and all course material.
HANDS-ON FLUTTER ANALYSIS
Saturday and Sunday, January 26-27, 2002
Solve subsonic aerodynamic flutter problems using a PC. Learn the steps
for solving flutter problems Each participant will receive a full blown
copy of ALGOR program, 200 nodes. Fee: $580 includes lunch and all course
material.
Marc Wiese
Bill I am really interested in your take on this. 200 nodes does not sound
like much.
Paul Lamar
I have never personally used Algore (except as an inept VP), but have looked
reasonably closely at most of the "smaller" FEA packages out there from a
professional interest standpoint, and with the view towards both getting some
version for some of our engineers at work, and possibly for personal consulting
work.
You're right, 200 nodes isn't much, but it depends on what you are going to
attempt. If you want to analyze the E-shaft or a housing, forget it. If
you want
to look at a moderately complex tube frame assy (fuselage, engine mount, tail
feathers, etc) it could be useful. Algor is capable, but this is their
"cheapo" version
which is intentionally limited to keep the price down. So, trusses and beams,
useful, a few shells (monocoques) - not too bad, but pretty limited. Solids
(castings
and forgings, etc) getting somewhere near useless, altho not totally, you
can do some
simple brackets with a reasonably coarse mesh and get something useful, if not
great.
I'm not sure what the post-processing capability is for Algor, seems like
they had a pre processor that was pretty slick, probably the post processor,too.
I expect you need to buy a ( $$) pre- post package to create your model and
then make color contour plots of stress or temp and show the deformed shape.
You CAN create simpler models (with a 200 node limit the model WILL be
pretty simple) by hand - I did it for several years in the bad old days, but
it is lot
of VERY nitpicky scut work. Modern pre-processors are SO wonderful, except
when they shit themselves and then you want to beat the programmer to death.
A love-hate relationship. Checking out the results without a post processor to
make pretty pix is fairly bleak. You CAN read the stress values for each
element,
possibly export the data to a spreadsheet and do some sorting and possibly XY
plotting based on element numbers, etc. But pretty lame compared to a real
graphics post processor. Algor has been madly expanding their capabilities in
the last few years, so perhaps this is included now, but I don't think so.
Price the
pre-post processor as they are nearly mandatory for anything complex. Come
to think of it, if you have only 100 beams, perhaps sorting by stress and then
reading the number wouldn't be TOO bad.
SO - keep the model simple, do hand model creation and look at the results
in the raw
numerical form, and then it is useful but a bit painful to use. The code is
capable,
and they do have an innovative automatic mesher, but I expect it is an
additional $2500 to
$5000. PCs today are actually pretty damned capable in this field. My home
system
is a 1GHz with 256Mb of memory and a 30 Gb hard drive. This was a supercomputer
10-12 years ago, no exaggeration. The bus speeds suck by comparison but the
CPU, mem and disk speeds are pretty decent, I've never benchmarked it, but
I'd bet it will
run about 1/3 to 1/2 the speed of our '89 vintage $1 million supercomputer,
possibly
better.
Depends on what you want to do. Another issue is the GIGO problem. I have been
working on a new type of very difficult dynamic shock response analysis at
work and
since I wasn't familiar with that particular analysis type (been using
ABAQUS for
about 15 years, tho) I made a small test problem and ran the new dynamic method
against a theoretically equivalent shock pulse with a direct dynamic
(explicit) analysis.
The results were crap, near zero correlation, which is what I was afraid of.
I now have
the two input decks at the vendor and they are very interested in why this
doesn't work,
and we'll get it sorted out eventually, BUT . . . . These are power tools
and you can
cut off you arm with a power tool if you don't know how to use it safely.
That's why
they pay me the big bucks, because usually I can tell shit from shinola, at
least in an
FEA analysis. Ya'll be careful out there. ;-)
Bill
The AirCraft Rotary Engine NewsLetter. Powered by Linux.
ACRE NL web site.
http://home.earthlink.net/~rotaryeng/ http://www.linux.org