🚀 go-pugleaf

Many thanks to all those who posted their views about my following query on the net.

>This is a query for those who use commercially availabel Finite Element Packages.
>There is a glut of such packages available in the market. I am interested in 
>knowing which packages are prefered over others for whatever reason - price, support,
>flexibility, GUI, versatality, licensing, availability, hardware, speed etc.
>
>I will summarize the responses if there are many.
>

I would like a make few comments of my own regarding the query. Some netters did not
think highly of some package which apparently was most useful to others. I do not 
intend to be the cause of a flame war that might ensue, after the netters read this
summary. I emphasize again, PLEASE DO NOT START SOMETHING WHICH WAS NOT THE INTENT OF
THIS POST.

I also received replies from companies who tried to advertize their software. I have 
not included those in my summary.

I also understand the implication of my summary, which may or may not influence potential
users of making a decision reagarding the best package. I would like to add that the
netters who have replied to me, form a small spectrum of the infinite users who may
not have access to net. In conclusion, I will request the netters to read the posting by
David Lombard carefully.

Any more comments and suggestions are most welcome.

	Pushkar Tamhane
	Department of Aerospace Engineering and Engineering Mechanics
	University of Texas at Austin

Disclaimer: These are my opinions alone. The UT and my Dept. has nothing to do with it.

********************************************************************************************

rizzo@cbnewsf.att.com writes:

I prefer ANSYS, because:

1) It can do 90% of the analyses, 90% of the time.

2) It has the most extensive parametric modeling features available.

3) It is very well supported, and it is available on all the
   popular platforms.

Ansys has a GUI.  But, I neither use it nor recommend it, not
because it isn't any good, but because GUI's are impediments
to productivity!

If you want to know more, ask me specific questions.
Otherwise, I could go on for days. ;-)

  =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
  * Anthony R. Rizzo       |                                              *
  * The FEA Group          | There's no shortage of engineers in the USA. *
  * AT&T Bell Laboratories |                                              *
  * att.com!homxc!rizzo    | There's only a shortage of engineering.      *
  * (201) 386-2565         |                                              *
  =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

********************************************************************************************

natarsu@mse.ogi.edu writes:

	I'll give you some info. on ANSYS 4.4A (Ansys, IMO, is
the easiest/most versatile among the known s/w's).  I can't speak 
for other codes; hopefully you'll get enough feedback to make
a judicious decision.

	I'm running ANSYS on an IBM RS/6000, with 150MB of
internal memory + ~600 MB of external memory, and 16MB of RAM.
Have run structural (static) analyses using Ansys.  IMO, the
salient feature of ANSYS over other codes is the ease with
which one can develop the meshes (as per the users'
requirements), as well as the excellent post-processing
capabilities.  With ver. 4.4A, the parametric modeling
capabilities provide one  the flexibility of developing
user-defined models, with varying input variables. Basically
one develops "macros"/subroutine-like, to facilitate the
model-development.   Of course, the latest version of ANSYS,
namely Ansys 5.0 has far-out capabilities -> Boolean
variables, arrays, top-down or bottom-up model development,
"subtraction" of solid objects etc...!

	There's a lot more to say ... however, I don't 
wan't to sound like an Ansys rep. ! If you have particular
questions regarding your application [therma/structural/
transient etc] and need further info., drop me a line, and
I'll try to address them.

	Hope this info. is of help.

-suku.

---------------------------------------------------------------
N. Sukumar                   Home Phone: (503)-690-3097
Department of MS&E           Work  "   : (503)-690-1121 (X7451)
Oregon Graduate Institute      ****** BLAZERS IN 1993 ******
---------------------------------------------------------------
********************************************************************************************

hbudiman@Athena.MIT.EDU writes:

I like ABAQUS FEM package due to its versatility ; We can even use our own
elements if needed and include the subroutine in the input file.  The post-
processor is also very good. The pre-processor is not really good ; Creating
the model using programs like PATRAN, IDEAS, etc is preferable.  The license
fee is very nominal (750 $ / year for universities), but it doesn't provide any
supports.  However, so many research groups at MIT are using this package that
not having support from the company does not cause any problems, since we can just
discuss the problems with other grad students or professors.

ADINA is also available at MIT and the support is great (since Bathe is an MIT
professor and he is very accessible to answer any technical questions on ADINA).
ADINA (as far as I know) doesn't provide as much flexibility as
ABAQUS.  Again, the pre-processor for ADINA is not good.  I have no further
comments on the package since I only heard from my colleague who has used it.

I used another package called COSMOS/M.  I have only negative comments on that
package.  We bought the Macintosh version of the package and it doesn't do the
job really well. I have no comments on the performance of COSMOS/M on other
platforms.

Harry Budiman
Ph.D student
Aeronautics & Astronautics
MIT

The opinion expressed above is solely my own.  MIT is not responsible for any of
the comments above.
********************************************************************************************

<@ugw.utcs.utoronto.ca:sfysh@epas.utoronto.ca> writes:

I am also interested in FEM packages.  At work I use NISA from EMRC on
a 386 33Mhz PC.  I have no basis for comparison on speed, but in
general the package is not very user friendly, the documentation
assumes a very high level of knowledge, the post processing (ie output
of results) and documentation is poor, and I don't think that the
program is terrribly stable.

I talked with a consultant whose company works with many packages, he
perfers COSMOS on the PC and ANSYS on unix workstations.  We didn't
discus his reasons.

Since I don't read this newsgroup often, I'd appreciate if you could
please send me a copy of your summary.

Stephen Fysh (on Stephanie's account)
sfysh@epas.utoronto.ca

********************************************************************************************

bengel@msc.edu writes:

Here's what I know about FEA packages. Some of this is a bit old.

ANSYS  basically a static structures code, has the capibility to tie into
 FIDAP which is a good fluids code. I think ANSYS did about $20 million
 worth of business last year. Big usage on PCs and workstations.

NASTRAN similar to ANSYS, it had a near monoploy on Government work because
 QA required NASTRAN analysis. It also does about $20 million worth of business.
 I think this is the oldest of the bunch.

ABAQUS. This code was written in part with a grant from EPRI (A research
 consortum of the Electric Power industry.) I think EPRIs intention was to 
 model nuclear reactors and the possible disasters such as earthquakes.
 This is a good code for dynamic analysis.

MARC. Written by a person who worked on the ABAQUS project. Similar capibilities
 as ABAQUS but said to be better for sliding surfaces and rubber. 

DYNA: Originally written in government labs, this code is basically a 
 dynamics code. I think NASTRAN distributes it now. I don't have much
 knowledge of this code.

SAP and NONSAP. Public domain structures code. You get what you pay for.

STARDYNE: a series of programs to do static, thermal, and dynamic analysis.

Pre and post processing:

ANSYS has a good pre and post processor which works under a variety of 
graphics terminals. (X, Textronics, etc).

ABAQUS has some post processing ability.

Performance:

I run mostly on Cray machines. Ansys runs the fastest for what it does.
Many of the other codes run real slow.

Bob Engel
MARC has a companion program called MENTAT to do pre and post processing.

I think the other packages require a separate pre and post processor such
as PATRAN.
********************************************************************************************

dh4j+@andrew.cmu.edu writes:

I've used ANSYS, ABAQUS and MSC/NASTRAN, both out in industry.

NASTRAN sucks.

ANSYS is pretty good (been a while and I hear the newest versions are pretty
spiffy).

ABAQUS may not perfect, and it's not "friendly" all the time.  But of the three
I think it's be far the best.



David Hunt - Graduate Slave |     My mind is my own.      | Towards both a
Mechanical Engineering      | So are my ideas & opinions. | Palestinian and
Carnegie Mellon University  | <<<Use Golden Rule v2.0>>>  | Jewish homeland!
====T=H=E=R=E===I=S===N=O===G=O=D=========T=H=E=R=E===I=S===N=O===G=O=D=====
Email:  bluelobster+@cmu.edu    Working towards my "Piled Higher and Deeper"

"Out there is a fortune waiting to be had; do you think I'd let it go you're
mad - you got another think coming!" -- Judas Priest
********************************************************************************************

mgkst1+@pitt.edu writes:

MSC/Nastran is still the number one FE package, however, ANSYS has
been closing-in rapidly.  You might consider querying about
application/code use as opposed to just FE generally.  The code used
depends on the problem domain and the sophistication of the user.


-- 
Mike Koopman
Concurrent Technologies Corporation     phone:   +1-814-269-2637
1450 Scalp Avenue                       telefax: +1-814-269-2666
Johnstown, PA  15904                    e-mail:  koopman@server1.ctc.com
USA
********************************************************************************************

tafrakes@Athena.MIT.EDU writes:

I work in Cleveland now (OK, so I still have my old college account)
and we use mostly ABAQUS right now.  We often use PATRAN to create
meshes for ABAQUS, though.  A couple of our new clients use ANSYS, so
we may start doing more with that, now.  We also occasionally use
IDEAS.

I forgot to mention what we use them for: glass modelling (light bulbs
for GE, drinking glasses for Libbey, plate glass for Nippon, etc).  We
prefer the creep law features of ABAQUS to ANSYS for heat transfer and
fluid flow.  We also do crack analysis, but don't use FEM for this.

-Andy

-- 
T. Andy Frakes (tafrakes@athena.mit.edu)
   "Give a difficult task to a lazy person and he'll find an
    easy way to do it." -- perhaps Bartlett knows who said this,
    but I'm too lazy to ask him.
********************************************************************************************

dnl@macsch.com writes:

The most important aspect of selecting an FEA code (or any code or any 
hardware or any thing) is:

	Can it do the job I want to do?

	Can I get the results I want/expect?  With emphasis on I.

Each of the commercial codes has something different to offer.  Some are
exceedingly simple to use, but don't offer a lot of capability or
flexibility.  If you're looking for an educational package in a program 
that just want's to introduce basic concepts, your probably best served
by this class of code.  If you want to be able to do *any* problem, from 
simple statics or normal modes to explicit transient dynamics.  Well, 
you probably need to look at several packages.  At the risk of offending 
the hand the feeds me, no single code will satisfy every need of every user.

So your checklist should be:

     1)	What types of analyses are needed now?

     2)	What type of analyses may be needed later?

     3)	What are the constraints (i.e. available on your selected platforms,
	afforable, user interface, available experience, etc.)?

     4)	Pick some *representative* test cases and run them.  

After comparing your check list to the available codes, and considering
recommendations from source knowledgeable with the specific codes AND
your requirements, you *should* come to one of the following conculsions:

      -	there a single best system which does everything I want!
	An easy choice!  If only life were always so simple.

      -	there are several more-or-less equal systems, each does
	everything I want.

      -	there are several systems, none do everything I want, and I
	want them all.

      -	there are several systems, none do everything I want, but I'll
	pick the one that does the most important things.

      -	there are several systems, none do everything I want, and I
	don't want any of them.  (Sound familiar?)

      -	I'm sorry I asked the question.

Hopefully, you will not arrive at the latter two conclusions!
So, the questions:

	What types of analysis do you want to do?

	What are your assumptions about who will be using the system?

	What are your constraints?  Any and all!


Now, a LECTURE for the gentleman from CMU.

	Your final comment, "NASTRAN SUCKS", was UNPROFFESIONAL, CHILDISH,
	AND QUITE POSSIBLY ROOTED IN IGNORANCE.

	If one of the several NASTRAN codes did not satisfy your needs, so 
	be it, don't use it!


Regards,
DNL
******************************************************************************
*                         My comments are my own.                            *
******************************************************************************
* David N. Lombard, dnl@macsch.com                                           *
******************************************************************************
********************************************************************************************

budkee@unvax.union.edu writes:

I am at a school, but we don't have any such packages to my knowledge.
But over the summer I worked at the offices of a distributor
of ANSYS.  Version 5.0 was supposed to come out this summer, and 
my job would have been to make copies, but it didn't and was delayed.
As of now I don't know it's status.  But I do know that 4.4a was leaps 
and bounds above the prior versions and it always puzzeled me why 
people would continue to use the older versions when there were errors in
the software that had been corrected.  I know that GE uses the 
ANSYS all over the place, and not the others mentioned.  As does 
Textron Lycoming.  My office used it for consulting work.  And I know
RPI has it.  I don't think we have any computers capable of handling it well.
Also the Navy uses ANSYS.  The guys at work didn't like NASTRAN, but
I forget their reasons.  When I talked to a friend in the Navy who 
uses this occasionally he said that ANSYS was by far the best.
That's all I can add to the subject.
		
					Eric Budke
********************************************************************************************

pkbhatti@kopth004.delcoelect.com (Pardeep K. Bhatti) writes:

Here is one response from industry.

We have both MSC/NASTRAN and ABAQUS. We use NASTRAN primarily for static
analyses, and ABAQUS for nonlinear and dynamic analyses.

ABAQUS, in general, has better nonlinear analysis capabilities (nonlinear
material properties, gap and contact elements, automatic time-stepping,
etc.) It has an extensive material library (plasticity, creep, temperature
effects, etc.) It also has several different types of multi-point
constraints, which we find very useful.

NASTRAN, on the other hand, is more efficient at normal mode analysis (with
availability of several different algorithms).

>On our Campus, we had a choice of using ABAQUS and MSC/NASTRAN. But MSC/NASTRAN was
>removed due to low usage. It seems industry is more fond of this package. 

You may be right about this. However, engineers in my group seem to prefer
ABAQUS over NASTRAN, but this could be due to our application environment.
-- 
Pardeep K. Bhatti                         pkbhatti@kopth004.delcoelect.com
Powertrain Engineering Analysis
Delco Electronics Corporation
GM Hughes Electronics
********************************************************************************************