Abaqus vs Marc vs Ansys
LESAFFRE-Xavier
My company is looking for a proven and efficient fea code for structural
non-linear analysis (thermoplastics or elastomers or metals with contact,
visco-elastictity, creep,...). We are considering Abaqus, Marc and Ansys (we
import CAD models from Pro/E using IGES).
I am very interested in comments from people using these softwares dayly :
great features, efficient and less-efficient possibilities, advantages vs
limitations...
I am convinced all these fea codes are great. Each of them certainly has its
own areas of excellence, and I would like to be aware of advantages &
limitations in order to make the best decision according to our specific
needs.
Thank you in advance, and please keep it moderated !
Pascal L'Heritier
Development engineer
WABCO France
Philipp Pott
I've been working with Marc over 4 years now, seen a little bit of Ansys
and my best friend is learning Abaqus now.
- Ansys is confusing with all its windows poping up, some friends from
the automobile industry say it calculates wrong and it doesn't have many
features Marc and Abaqus already have. Ansys is not a good choice for
highly non-linear jobs
- Marc seems to be at the edge, although I never believed it myself. But
the capabilities in plastics and elastomers combined with the contact,
adaptive meshing and the rezoning made it the number one choice for
non-linear jobs like car door rubber-analysis. But Marc has been bought
by MSC, how sell Nastran etc. As far as I found out, Marc is going to be
integrated in Nastran and will not be supported any more. If you tend to
use Marc, consider of using a meshing tool like Hypermesh in addition.
- the guys how program Abaqus used to work at Marc earlier and were
about 1-2 years behind Marc. But since Marc didn't approve in the last
year, Abaqus caught up in the non-linear sector. A friend of mine has to
learn Abaqus now and is amazed how easy it is. He is a long term Marc
user, too. Since Abaqus is also capable of doing explicite calculations,
it is a good tool even for crash analysis. But I would suggest to use
Hypermesh for Preprocessing and Animator for Postprosessing.
I hope this helps
Philipp Pott
Materials Science Department
University of Erlangen
Germany
Martin Liddle
<po...@ww.uni-erlangen.de> writes
>- Ansys is confusing with all its windows poping up, some friends from
>the automobile industry say it calculates wrong and it doesn't have many
>features Marc and Abaqus already have. Ansys is not a good choice for
>highly non-linear jobs
comments to this response. I have 20 years experience with ANSYS, I
know a little about ABAQUS and virtually nothing about MARC. Most
people find the standard ANSYS user interface easy to get to grips with
and productive. (This is based on a number of years experience of
running Introductory training courses). However if you don't like it
you can turn it off and type commands from the command line or indeed it
is user programmable so that you can adapt it to your specific
preferences. My experience is that ANSYS does not calculate
incorrectly. Clearly with any tool you can use it in an inappropriate
way and get questionable answers. ANSYS are very strong on Quality
Assurance and their error notification procedures are, I think, the best
in the industry. Have your friends reported their suspect answers
through the correct channels? I think it is true that historically the
non-linear capability was not the strongest in the industry. However a
lot of new features have been added over the last few versions. The
contact capability is now very strong and certainly comparable with
ABAQUS. My experience would suggest that, for contact problems, ABAQUS
solves a bit faster but ANSYS produces higher quality results and is
more robust. A number of new material laws have been added in the
latest version. The major limitation of ANSYS for large strain problems
is that their is currently no regridding capability so that once your
elements have become highly deformed you can't go any further.
Other strengths of ANSYS include a powerful multi physics capability
allowing you to analyse a wide range of physical phenomema and couple
them together, a competent pre processor with good mesh generation and
able to accept geometry from most CAD systems (except perhaps CATIA), a
wide range of solvers, a competent post processor and perhaps most
importantly a built in programming language (APDL) allowing you to
extend the capabilities in many directions.
If your interests are solely in large strain non-linear analysis then
ABAQUS would be a sound choice. If you want to tackle a wider range of
problems then ANSYS has much to recommend it.
--
Martin Liddle, Tynemouth Computer Services, 27 Garforth Close,
Cramlington, Northumberland, England, NE23 6EW.
Phone: 01670-712624. Fax: 01670-717324.
Antonio L. Negrón
I have experience in both ANSYS and ABAQUS, none in MARC. Here are some of
the things I have found......
> - Ansys is confusing with all its windows poping up,
I will have to respectfully disagree with you on your comments. The ANSYS
GUI is extremely simple. As a matter of fact, it is made in such a way that
menu selections are located in the order in which you typically use them
when you perform an analysis and hence is very intuitive. I do not see how
a GUI system can be confusing, unless you are unfamiliar with GUI's and feel
uncomfortable with anything that is not console based.
> it doesn't have many features Marc and Abaqus already have. Ansys is not a
good choice for
> highly non-linear jobs
Could you name a few? The same argument can be made against ABAQUS. Here are
some of the things you can't do with ABAQUS :
1- Pre/post process (you need a pre imposed pre/post, extra $$, usually
a lot of $$)
2- Multiphysics analysis (i.e. the thermal stresses causes on a
structure due to an external flow)
3- CFD analysis
4- Parametric FEA models (this is an extremely powerful feature, when
used to its full potential)
5- P-method AND H-method
> - Marc seems to be at the edge, although I never believed it myself. But
> the capabilities in plastics and elastomers combined with the contact,
> adaptive meshing and the rezoning made it the number one choice for
> non-linear jobs like car door rubber-analysis.
It is also important to note that all the things you mentioned above can
also be done in ANSYS. As a matter of fact, ANSYS has an extremely easy way
of setting up contact pairs, UNLIKE ABAQUS.
> - the guys how program Abaqus used to work at Marc earlier and were
> about 1-2 years behind Marc. But since Marc didn't approve in the last
> year, Abaqus caught up in the non-linear sector.
When I went to the ABAQUS class, I was told there by one of the instructors
something that sound similar to what you are saying here!
>A friend of mine has to learn Abaqus now and is amazed how easy it is. He
is a long term Marc
> user, too.
Here I will disagree with you again. Having had a few years experience with
ANSYS when I went to take the ABAQUS class I was shocked by the fact that
THE WHOLE WEEK of the class was based on how to write the commands correctly
in a ridiculous and obsolete "input deck", as in deck of punched cards (that
is what they used to call a program in the 70's). ABAQUS is not trivial to
learn, less trivial to use. It takes ridiculously long to setup "real life
problems" (I'm not talking about beam or 2D problems) and there are too many
things where one can screw up and end up wasting you precious time (such as
comas and spaces in the wrong place).
>Since Abaqus is also capable of doing explicite calculations,
> it is a good tool even for crash analysis.
The only way you can do explicit analysis in ABAQUS is by purchasing an
extra license called ABAQUS/Explicit. It is NOT included in regular ABAQUS,
extra $$ to shell out. If you needed to do explicit, ANSYS also has an
explicit solver for extra $$$
Xavier, my honest opinion: If productivity is an issue for you as it is
mostly the case in industry and not in educational institutions (i.e. you
need the right results but you also need them ASAP, you're not doing science
experiments..) seriously consider ANSYS. It allows you to directly import
Pro/E models (no IGES BS to worry about) and has a VERY powerful mesher with
local mesh refinement capabilities. You can also do P-method solutions which
can be really effective for some problems. None of these things can be done
with ABAQUS. You can do optimizations by setting up parametric models and
using the APDL (ANSYS Parametric Design Language). It is a lot easier than
it sounds.
Last but not least, look at the TOTAL cost of each solution. ANSYS might
seem more expensive than ABAQUS, but when you look at the extra money you
spend in post/process. meshers, geom translators etc., it might even be
cheaper...
Just my 2 cents.... Hope it helps....
--
Antonio L. Negrón
Mechanical Engineer
Naval Surface Warfare Center, Dahlgren Laboratory
Launcher Systems Branch, Code G21
negr...@nswc.navy.mil
antonio...@mciworld.com
Philipp Pott
looks like I upset some old ansys-users...
But if you compare these programs and don't know Marc, it's just
ridiculous to make any comments.
I've seen problems ansys-guys have on a ansys workshop which could
easily be solved with other programs like Marc. These were
mesh-refinement/remeshing, dynamically moving mesh, low speed of the
macro-language.
I would be interested in hearing more about the problems you have to
solve. There is no general purpose multiphysics do everything
best/cheapest program, because then there wouldn't be so much programs
around. I found out, that you loose more money trying to solve the
problem with the wrong program then buying the right program - even
though I'm still at the university.
Therefore I would suggest that the original sender (Pascal or Xavier)
contributes some more details and then everybody tells how he would
solve this problem with his software.
Or we just leave it as it is: using fe-software is also a question of
personal taste and in 80% has nothing to do with the real capabilities.
Philipp Pott
Material Science Department
University Erlangen
Germany
Martin Liddle
<po...@ww.uni-erlangen.de> writes
>But if you compare these programs and don't know Marc, it's just
>ridiculous to make any comments.
>
making comments about it. I made it clear that I know very little about
MARC and refrained from making any comments.
>I've seen problems ansys-guys have on a ansys workshop which could
>easily be solved with other programs like Marc. These were
>mesh-refinement/remeshing, dynamically moving mesh, low speed of the
>macro-language.
>
Again I can only comment on my experience. I'd say the mesh-refinement
capability in ANSYS works well. I don't understand what you mean by a
dynamically moving mesh, could you give an example? I have certainly
worked on problems like analysis of electrical machines where the mesh
associated with the rotor is moved relative to the stator during a
transient analysis. There is no problem in doing this in ANSYS. The
macro language is interpreted and so isn't fast. This often isn't a
problem. However it does have vector operations and once you get your
head around using the vector operations (rather than coding do-loops)
the speed increases by a factor of 100 and is quite acceptable.
>I would be interested in hearing more about the problems you have to
>solve. There is no general purpose multiphysics do everything
>best/cheapest program, because then there wouldn't be so much programs
>around. I found out, that you loose more money trying to solve the
>problem with the wrong program then buying the right program - even
>though I'm still at the university.
>
In a University environment I'd agree whole heartedly with what you say.
Many vendors offer cut price or free versions to Universities.
Unfortunately in industry it isn't quite like that. Software costs
money. Managers don't understand the technical arguments (you do
understand that Dilbert is a mirror on the real world not a cartoon
strip) and won't sanction spending money on different software. You are
forced into using what you have to solve the problem. Hence having a
code like ANSYS with a wide range of capability (even though there are
better codes for specific problems) is not a bad survival strategy.
>Therefore I would suggest that the original sender (Pascal or Xavier)
>contributes some more details and then everybody tells how he would
>solve this problem with his software.
>
I'd agree with that.
>Or we just leave it as it is: using fe-software is also a question of
>personal taste and in 80% has nothing to do with the real capabilities.
>
There is a lot of truth in that but as one invests a substantial amount
of time in learning a tool like an FE code, it is still worth making the
most informed initial choice that you can.
LESAFFRE-Xavier
Well, your comments are very helpful to me, especially since you have used
all 3 softwares (Patran/Marc, Abaqus, Ansys).
Some comments :
- MSC tells us Marc is added to their list of FE softwares (like
Nastran, Dytran,...) as an independent software. MSC Patran will be the
general Pre/Post, with the whole capabilities of the current Marc pre/post
(Mentat) in the next year.
- Ansys highlights its new capabilities with contact problems.
- Marc highlights it does not have to define contact pairs, only
contacting bodies. Provides local mesh refinement while solving.
- Abaqus is a little busy with other people, I guess. And it does not
seem to provide a user-friendly, efficient Pre/Post yet.
- We deal with structural models (mechanics & materials), so we are not
interested in multiphysics.
- User-friendliness is not overlooked, since it means money (especially
with part-time users) : console commands is not a premium ! But it is mainly
a subjective point.
We don't believe a benchmark test is workable thru this news forum... Let's
give it to these 3 companies, and see what they can do !
Thank you for your comments !
Best regards,
Amicalement,
Pascal L'Heritier
WABCO France
http://www.wabco-auto.com
LESAFFRE-Xavier a écrit dans le message <80s3er$j9l$1...@wanadoo.fr>...
jwhi...@my-deja.com
feedback that I have seen in this thread.
Having recently been to an MSC demo featuring MARC, my understanding is
that you still have to define contact pairs with MARC, just not contact
surfaces. Contact bodies are defined instead. This does simplify
matters some.
Secondly, ABAQUS does have relatively new pre- and post-Processor tools.
See http://www.hks.com/products/p_abviewer.html (ABAQUS viewer) and
http://www.hks.com/products/p_abcae.html (ABAQUS/CAE) for details.
Both are developed in-house and look quite promising.
And lastly, there have been a number of comments regarding the
complexity of syntax in ABAQUS input files. I have worked with a few
different codes, and I see no real distinction [with respect to
complexity] in this area. If ABAQUS syntax seems more difficult than
that of other comparable products, perhaps it is because it has more
features/options...? I believe that the syntax is stressed in the
introductory courses because a number of third-party preprocessors used
by ABAQUS customers typically are not up to speed with such evolving
features, thus the user must manually modify the input files to obtain
the desired and available physics. It is difficult to modify something
that you do not comprehend, right?
My $0.02.
In article <813bmi$3p8$1...@wanadoo.fr>,
Sent via Deja.com http://www.deja.com/
Before you buy.
DANNY LEVINE
>
> - Abaqus is a little busy with other people, I guess. And it does not
> seem to provide a user-friendly, efficient Pre/Post yet.
IF you run on Win/NT, HP-Unix or SGI-Unix, you might want to look at
ABAQUS/CAE a relatively new pre- and post-processor that is
marketed by HKS. I've worked with the post-processing (contained
in a module called ABAQUS/Viewer) and find it quite easy to use,
althought it does not yet contain all of the functions of the previous
program ABAQUS/post.
Dan Levine
jape.a...@gmail.com
jonmor...@gmail.com
My experience has shown me that large errors come mainly from lack of material characterization ,loading and boundary conditions. These physics of the problem are more important.