You may find the following FEA resources useful:
As regards software with a specifically educational orientation, you could
perhaps take a look at Felt (see http://felt.sourceforge.net /) or Felipe
(see http://www.brunel.ac.uk/~blstmbr/felipe /); although I personally have
no experience with either of these, you can readily evaluate these as free /
demo versions are available.
Trust this helps.
I'd recommend starting with 2D truss or frame analysis. The basic
steps are all there of defining nodes, elements (members), boundary
conditions, loads, running the solver, post processing. But it is not
overwhelming like a full FEM program can be and it is less likely you
will make a GIGO error which is very easy to do with FEA. Also, as
you have probably seen, free frame analysis programs abound from the
(Framework and Fastframe both seem pretty good. The step-by-step
organization in Fastframe makes it a better learning tool, I think.)
(In a truss, all members are straight and loaded in pure tension or
compression due to pinned joints, and ~all members form triangles. In
a frame, members can be bent and joints can exert torque on the
members, creating internal bending moments as well as
tension/compression. Trusses generally have the best potential for
high stiffness/weight ratios; frames are easier to build. Any frame
analysis program can analyze a truss, not vice versa.)
After you get the hang of it, I'd recommend creating some models to
match textbook examples to verify results. Study some Statics and
Mechanics/Strength of Materials texts to understand the mechanics
(Beer and Johnston are good for both). No GIGO.
All that said, the worst pitfall of FEA programs is that users don't
understand both the mechanics and the program well enough to get
meaningful output. The color-shaded stress distributions may be
impressive, but 1) are the stresses meaningful, 2) if they are, what
do they mean, and 3) was FEA even the appropriate tool for
analyzing/optimizing that design? E.g., for 1), if an unrealistic
boundary condition is specified, stresses can be off by a factor of
10, easy. For 2), lets say you get a peak principal tensile stress of
30ksi and your material has a yield stress of 60ksi; is that a
sufficient factor of safety (it depends)? Is the material notch
sensitive or will it allow local yielding w/o global failure? ... For
3), 100 FEA iterations on a bad design will result in a slightly less
bad design, although the user may believe it is 100x better; zero FEA
iterations on a good design will result in a good design, and for that
one needs to understand the fundamental mechanics. All this is to
point out that FEA is just a tool. But if you do understand both the
mechanics and the program, it can be an invaluable one.
Hope this helps, David
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.