转贴：general discussion on features/capabilities of ABAQUS

linqus

from yahoo abaqus group


It all depends on what you'll be using it for. For instance, in you will be
doing single body contact (a deformable body against a rigid body/surface)
several major FEA codes will fit the bill. If you're doing multi-body
contact, the list narrows down quite at bit. For the latter, I would not go
beyond ABAQUS, LS-Dyna, Adina, MARC and (deferring to BenZ) ANSYS. I'm
assuming that you'll be working in the realm of solids; if you want to do
Fluid-Structure Interaction (FSI), things are get more complicated.

If you want to make a table of attributes that will help you separate the
better codes from the lesser codes, I would suggest looking at

1. element technology
2. material laws (constitutive models) available and
3. solvers, particularly if you'll be doing a lot of nonlinear
analyses.

I have intentionally excluded CAD Import/Meshing because (a) I think all FEA
codes can import CAD geometry one way or another, so there's not much
differentiation there and (b) when it comes to meshing, everyone can do a
good 2D mesh or an "extrudable" 3D mesh, a passable Tet-mesh, and most
cannot easily/consistently do a good Hex-Mesh (except for maybe TrueGrid,
but that's another lecture!)

One thing that is quite useful about ABAQUS is that it does not limit your
elements to a particular material law except for obvious situations (e.g.
you wouldn't want to have a no-tension material law with a membrane
element). I have heard that in some codes you have to use shell element
type X for materials (A1, A2,..) and shell element type Y for material (B1,
B2...). As for solvers, I know that ABAQUS and MARC offer a variety of
solvers that can be quite useful for very large problems. In passing,
I recall that years ago some vendors would say that wavefront solvers were
all you ever needed, which was not the case at all.

Another deciding factor would be to compare the size of the technical R&D
staff at each company. I know that ABAQUS has several hundred people
involved in R&D in Rhode Island; indeed, I would guess that their solid
mechanics technical staff is larger than that of any university Mechanics
department in the US (and probably the world).

When using ABAQUS, I have found that it actually does what it's supposed to
(and advertised) to do, and it is QA'd quite well. I have a strong
preference for using ABAQUS/Explicit for heavy-duty 3D contact problems, and
it has worked very well for me in the past. In my current job, I use the
superelastic material model almost exclusively; in my previous job, I used
elastomeric constitutive models (Mooney-Rivlin, Ogden) almost exclusively.
The superelastic model used by ABAQUS is based on publicly available
research; the caveat is that the originam model can have numerical
instabilities which ABAQUS seems to have resolved.

Finally, you have to make a honest assessment of the technical expertise
within your organization *and* that of the FEA vendor's support staff. If
you and/or your engineers are not "hard core" when it comes to FEA, make
sure that the vendor's support staff is technically superior. In either
case, there are way too many "demo jocks" and people with minimal product
knowledge working for some FEA vendors/distributors (and don't even get me
started on staff turnover issue), although I must say that ABAQUS is not one
of them!

Good luck!

Milton DeHerrera