Discussion:
Number of nodes/elements in assembly?
j***@gmx.net
2007-01-08 12:14:27 UTC
Permalink
Hi,
$Subject says it all:

How can I find total number of nodes/elements in assembly?
I know I can query instance mesh, but I do not want to
check all 50 instances and then summ it up, I would like
to know just total number of nodes/elements to estimate
problem size and its memory requirements...

Is it somehow possible to find this info in Abaqus/CAE?

Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
Gurmeet Cheema
2007-01-08 19:11:53 UTC
Permalink
I think that the only way to do this by adding nodes for each part.

Gurmeet

***@gmx.net wrote:
Hi,
$Subject says it all:

How can I find total number of nodes/elements in assembly?
I know I can query instance mesh, but I do not want to
check all 50 instances and then summ it up, I would like
to know just total number of nodes/elements to estimate
problem size and its memory requirements...

Is it somehow possible to find this info in Abaqus/CAE?

Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer


</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Jarry
2007-01-09 18:10:41 UTC
Permalink
Post by Gurmeet Cheema
I think that the only way to do this by adding nodes for each part.
Great idea, welcome in the 21-st century! Why the heck abaqus
can not do this???
-----------------------------------------------------
Post by Gurmeet Cheema
You can create a node set, let's call it "ALLN", in the assembly
module. Once created, you can see a message like this in the message
area of your CAE: The set 'ALLN' has been created (43025 nodes).
Nice, but I want to have this information always when I need
it to check how number of nodes/elements is rising as I mesh
parts, and have it always updated (e.g. when I remesh some part,
or add some instance in assembly)...
-----------------------------------------------------
Post by Gurmeet Cheema
Why don't you let ABAQUS do it for you? Do an "abaqus datacheck"
So when I want to know, how many nodes/elements I have in assembly,
I must define some load-steps and jobs, and let it run for a couple
of minutes, and then check *.dat file, in order to get two numbers.
Really "practical" solution. That's what I call "abaqus-style"...
-----------------------------------------------------

I remember days, when I was working with different fem-software,
where all I had to do to get this info was to use command AFLIST...

But thanks to all, who repplied. I can not believe Abaqus inc.
did not make this important info available...

Jarry
Gonzalez, Diego
2007-01-09 20:43:26 UTC
Permalink
Hi all,

My doubt is related with the generation of a composite model. I would
like to embed a sort of 'particles' in different positions of a cube.
Given that the partition technique applied on the cube is not an option,
does anybody know how to embed a set of previously meshed particles in a
bigger structure (a cube, for instance).

Thanks in advance,

Diego.



[Non-text portions of this message have been removed]
Fernando
2007-01-10 16:29:33 UTC
Permalink
Diego,
Post by Gonzalez, Diego
My doubt is related with the generation of a composite model. I would
like to embed a sort of 'particles' in different positions of a cube.
Given that the partition technique applied on the cube is not an
option, does anybody know how to embed a set of previously meshed
particles in a bigger structure (a cube, for instance).
Someone was trying to do something similar as what you are asking
about two months ago, with fiber-reinforced materials. Check the
archive for that thread.

Fernando
BenZ
2007-01-11 22:28:22 UTC
Permalink
You can do that with I-Deas, the technique is not very complicated.
But with Abaqus/CAE, no ideas !

BenZ.
Post by Gonzalez, Diego
Hi all,
My doubt is related with the generation of a composite model. I would
like to embed a sort of 'particles' in different positions of a cube.
Given that the partition technique applied on the cube is not an option,
does anybody know how to embed a set of previously meshed
particles in a
Post by Gonzalez, Diego
bigger structure (a cube, for instance).
Thanks in advance,
Diego.
[Non-text portions of this message have been removed]
Ryan S
2007-01-12 18:17:10 UTC
Permalink
Maybe you can assemble the cube and the particles separately in the assembly
module, and constrain the particles to the cube using the "Embedded region"
constraint.?
Post by BenZ
You can do that with I-Deas, the technique is not very complicated.
But with Abaqus/CAE, no ideas !
BenZ.
Post by Gonzalez, Diego
Hi all,
My doubt is related with the generation of a composite model. I
would
Post by Gonzalez, Diego
like to embed a sort of 'particles' in different positions of a
cube.
Post by Gonzalez, Diego
Given that the partition technique applied on the cube is not an
option,
Post by Gonzalez, Diego
does anybody know how to embed a set of previously meshed
particles in a
Post by Gonzalez, Diego
bigger structure (a cube, for instance).
Thanks in advance,
Diego.
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
Ryan S
2007-01-10 15:23:45 UTC
Permalink
Jarry,

You can write a little macro that will do this for you. The following script
adds up and displays the number of nodes and elements in your model. It is
approximate, since it does not account for reference nodes etc.

This approach is what you call 'abaqus-style', i.e. it requires some user
fiddling, but once you learn how to, the possibilities are endless. Plugins
and customizations are more sophisticated alternative solutions.

Stick the following script in your abaqusMacros.py file, and run it as a
macro:

from abaqus import *
from abaqusConstants import *
import __main__
def MeshStats():
import part
import assembly
modelName = session.sessionState[session.currentViewportName
]['modelName']
assy = mdb.models[modelName].rootAssembly
nNodes=0
nElems=0
print '%12s %6s %6s'%('INSTANCE','NODES','ELEMENTS')
for iName in assy.instances.keys():
nn=len(assy.instances[iName].nodes)
ne=len(assy.instances[iName].elements)
print '%12s %6i %6i'%(iName,nn, ne)
nNodes = nNodes + nn
nElems = nElems + ne
print '%12s %6s %6s'%('------------','------','------')
print '%12s %6i(N) %6i(E)'%('TOTAL',nNodes,nElems)
Post by Jarry
Post by Gurmeet Cheema
I think that the only way to do this by adding nodes for each part.
Great idea, welcome in the 21-st century! Why the heck abaqus
can not do this???
-----------------------------------------------------
Post by Gurmeet Cheema
You can create a node set, let's call it "ALLN", in the assembly
module. Once created, you can see a message like this in the message
area of your CAE: The set 'ALLN' has been created (43025 nodes).
Nice, but I want to have this information always when I need
it to check how number of nodes/elements is rising as I mesh
parts, and have it always updated (e.g. when I remesh some part,
or add some instance in assembly)...
-----------------------------------------------------
Post by Gurmeet Cheema
Why don't you let ABAQUS do it for you? Do an "abaqus datacheck"
So when I want to know, how many nodes/elements I have in assembly,
I must define some load-steps and jobs, and let it run for a couple
of minutes, and then check *.dat file, in order to get two numbers.
Really "practical" solution. That's what I call "abaqus-style"...
-----------------------------------------------------
I remember days, when I was working with different fem-software,
where all I had to do to get this info was to use command AFLIST...
But thanks to all, who repplied. I can not believe Abaqus inc.
did not make this important info available...
Jarry
[Non-text portions of this message have been removed]
Fernando
2007-01-10 16:27:39 UTC
Permalink
Jarry,
Post by Jarry
Great idea, welcome in the 21-st century! Why the heck abaqus
can not do this???
[...]
So when I want to know, how many nodes/elements I have in assembly,
I must define some load-steps and jobs, and let it run for a couple
of minutes, and then check *.dat file, in order to get two numbers.
Really "practical" solution. That's what I call "abaqus-style"...
OK, no need to get sarcastic, we are all trying to help you, remember.
From the original question it was not clear whether you needed this as
you are modeling (not it is :-) )

The answer to your (updated) question is easy: use a simple (I would
say trivial) python script to get the answer, something like this
(warning: this is pseudocode!):

nel = 0;
nn = 0;
for x in model.parts()
*** nel = nel + count(x.elements());
*** nn = nn + count(x.nodes());
end

*** = appropriate indents for python

The count() function returns an integer saying how many of those are,
assuming that parts(), elements() and nodes() returns a list with all
of them. Once you have the script working you have 2 options to apply it:
a) store it somewhere (your home folder is a good candidate) and run
the script from the File menu each time you need it
b) integrate it into the "custom tools" of the user interface. This is
a little more involved, but worth it if you spend time using CAE,
which you seem to...

Hope this helps,
Fernando
Thomas Zander
2007-01-11 07:51:54 UTC
Permalink
Hi,
Post by Fernando
Post by Jarry
Really "practical" solution. That's what I call "abaqus-style"...
The answer to your (updated) question is easy: use a simple (I would
say trivial) python script to get the answer,
i followed the discussion because i am in the process of deciding wether it is
sensible to switch from patran to abaqus cae. My impression is, that abaqus
cae is far away from the input file, whereas patran is rather close to it: In
Patran only few clicks are necessary to count node/element numbers, to find
out which node/element belongs to which groups etc. etc. In abaqus you need
to programm scipts for this?

I think abaqus cae was programmed with the idea, that the user should have to
know as less as possible about FEM or at least about nodes and elements. Is
my impression correct, that when you are based on nodes and elements and not
on geometry, abaqus is not well suited?

Thomas
Gurmeet Cheema
2007-01-11 21:46:35 UTC
Permalink
Thomas Zander <***@charite.de> wrote: I think abaqus cae was programmed with the idea, that the user should have to know as less as possible about FEM or at least about nodes and elements. Is my impression correct, that when you are based on nodes and elements and not on geometry, abaqus is not well suited?

I would not agree with above statement. I have been using ABAQUS CAE for a while and am quite happy with it. It gives you the option of working with the goemetry or the nodes/elements at your choice. If you want to work with nodes/elements only you can convert your part to orphan mesh part. However some times it is simpler to apply loads/boundary conditions on geomentry.

However I have also missed this feature (counting number of nodes in an assemlby) a few times. But I would only classify this as just one more feature to be added. I do not believe that any generalization from this is correct. ABAQUS CAE comes with lot of convenient and attractive features.

Gurmeet S. Cheema

Thomas Zander <***@charite.de> wrote:

Hi,
Post by Fernando
Post by Jarry
Really "practical" solution. That's what I call "abaqus-style"...
The answer to your (updated) question is easy: use a simple (I would
say trivial) python script to get the answer,
i followed the discussion because i am in the process of deciding wether it is
sensible to switch from patran to abaqus cae. My impression is, that abaqus
cae is far away from the input file, whereas patran is rather close to it: In
Patran only few clicks are necessary to count node/element numbers, to find
out which node/element belongs to which groups etc. etc. In abaqus you need
to programm scipts for this?

I think abaqus cae was programmed with the idea, that the user should have to
know as less as possible about FEM or at least about nodes and elements. Is
my impression correct, that when you are based on nodes and elements and not
on geometry, abaqus is not well suited?

Thomas


</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Jarry
2007-01-11 17:43:37 UTC
Permalink
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you, remember.
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...

For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina, cosmos/m,
just to name a few), now 2 years I'm using abaqus, and my feelsings
are... mixed. That's the best, what I can say...

My 1st impression was: Wow, what a perfect "cad-like" pre/processor
(abaqus/cae)! How is this abaqus transparent, and user-friendly!
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!

Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...

Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!

Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!

What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!

What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!

...etc, etc, etc, etc...

I do not know... Maybe I'm really too demanging...

Jarry
--
_______________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
BenZ
2007-01-11 22:54:41 UTC
Permalink
Hiii mates, very interesting debate !

Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact pairs
by reading the generated job (coming out from I-Deas or Patran) ! It
was a better technique that syntax checking the model then reviewing
surfaces one by one then pairs as a second step.

To me, /CAE has two very weak points :
1/The "selection power" is zero, nothing compared to other proggys !
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).

To Thomas : don't please don't forget Patran. Patran was Abaqus/Pre
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.

Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export module,
then you'll be happy :)

My 2cents of the day.

BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina, cosmos/m,
just to name a few), now 2 years I'm using abaqus, and my feelsings
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like" pre/processor
(abaqus/cae)! How is this abaqus transparent, and user-friendly!
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
_______________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
Gurmeet Cheema
2007-01-12 13:52:41 UTC
Permalink
BenZ <***@yahoo.fr> wrote:
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).

I do not understand this statement. It takes only a couple of clicks to create an orphan mesh part and it works flawlessly. I wonder if BenZ has worked with the latest version of ABAQUS CAE.

Gurmeet S. Cheema

BenZ <***@yahoo.fr> wrote:
Hiii mates, very interesting debate !

Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact pairs
by reading the generated job (coming out from I-Deas or Patran) ! It
was a better technique that syntax checking the model then reviewing
surfaces one by one then pairs as a second step.

To me, /CAE has two very weak points :
1/The "selection power" is zero, nothing compared to other proggys !
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).

To Thomas : don't please don't forget Patran. Patran was Abaqus/Pre
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.

Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export module,
then you'll be happy :)

My 2cents of the day.

BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina, cosmos/m,
just to name a few), now 2 years I'm using abaqus, and my feelsings
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like" pre/processor
(abaqus/cae)! How is this abaqus transparent, and user-friendly!
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
_______________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
BenZ
2007-01-12 16:56:38 UTC
Permalink
Ah OK great, you are right, I have stopped working with /CAE since
many years... I use only the visualisation module. I was used to
Abaqus/Post hey (small tears) :)

BenZ.
Post by BenZ
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
I do not understand this statement. It takes only a couple of
clicks to create an orphan mesh part and it works flawlessly. I
wonder if BenZ has worked with the latest version of ABAQUS CAE.
Post by BenZ
Gurmeet S. Cheema
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact pairs
by reading the generated job (coming out from I-Deas or Patran) ! It
was a better technique that syntax checking the model then
reviewing
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by BenZ
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export module,
then you'll be happy :)
My 2cents of the day.
BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-friendly!
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
_______________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
http://groups.yahoo.com/group/abaqus
Yahoo! Groups Links
---------------------------------
Have a burning question? Go to Yahoo! Answers and get answers from
real people who know.
Post by BenZ
[Non-text portions of this message have been removed]
Ryan S
2007-01-12 18:12:08 UTC
Permalink
Thought I should pass on replying, but there's so much controversial stuff
in some of the messages. Here goes.

Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.

Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try and guess.
Also, I hope these statements are being made after using the latest
available version of the product.

Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's the whole
point of CAE. There are cases where you do, and that sucks, I agree.

Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get product XYZ, it is
the best*. No product is always much better at everything.

The one thing I will say in positive for CAE is itself a generalization :-)
: It generally does a better job at providing abaqus coverage than any other
pre-processor (particularly for the newer features in abaqus). And the
post-processor is fast.

Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact pairs
by reading the generated job (coming out from I-Deas or Patran) ! It
was a better technique that syntax checking the model then reviewing
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other proggys !
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was Abaqus/Pre
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export module,
then you'll be happy :)
My 2cents of the day.
BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina, cosmos/m,
just to name a few), now 2 years I'm using abaqus, and my feelsings
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like" pre/processor
(abaqus/cae)! How is this abaqus transparent, and user-friendly!
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
[Non-text portions of this message have been removed]
BenZ
2007-01-12 22:52:33 UTC
Permalink
You have your point of view, well ! But I must answer on what I mean
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort and
create group with nodes, elements... different types of entity.
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that Abaqus/CAE
allows that. One has to go into the .inp. More generally, those who
never go into the .inp do not really know what is computed, it is a
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this in /CAE if
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes number.
Abaqus has this lack : it relies on labels of elements or nodes to
define a FEM.

BenZ.
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try and guess.
Also, I hope these statements are being made after using the latest
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's the whole
point of CAE. There are cases where you do, and that sucks, I
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get product XYZ, it is
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus coverage than any other
pre-processor (particularly for the newer features in abaqus). And the
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
[Non-text portions of this message have been removed]
Gurmeet Cheema
2007-01-13 16:19:35 UTC
Permalink
In ABAQUS CAE since you can pick geometrical surfaces, it is easy to apply bolt pretension (I am assuming a 3-d bolt). One has to create a cross-sectional surface using partition tools. Then that surface can be picked for applying pretension. Even if you want to work with an orphan mesh above can be done. The way to do this is to partition the part and create the internal cross-sectional surface. Then when the meshing is done, the cross-sectional surface is retained in the mesh (all the relevent nodes can be picked with angle).

Again working with geometry it would be easy to pick any nodes for kinematic coupling. All you have to do is pick the corresponding geometrical entity; volume, surface, edge etc.

Basic facility provided in ABAQUS CAE is the ability to work with geometry. For those who can get used to it, it is a lot faster. I think it is the wave of the future.

In ABAQUS CAE one rarely needs to work with the node label.

BenZ did I answer some of the concerns raised in your e-mail?

Gurmeet S. Cheema
BenZ <***@yahoo.fr> wrote:
You have your point of view, well ! But I must answer on what I mean
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort and
create group with nodes, elements... different types of entity.
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that Abaqus/CAE
allows that. One has to go into the .inp. More generally, those who
never go into the .inp do not really know what is computed, it is a
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this in /CAE if
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes number.
Abaqus has this lack : it relies on labels of elements or nodes to
define a FEM.

BenZ.
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try and guess.
Also, I hope these statements are being made after using the latest
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's the whole
point of CAE. There are cases where you do, and that sucks, I
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get product XYZ, it is
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus coverage than any other
pre-processor (particularly for the newer features in abaqus). And the
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only one
thing when I was calculation engineer : checking the contact
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between CAD
and FEM. On the other hand, I think that I-Deas will remain the most
powerfull .inp generator because FEM is prioritized, but with a very
good CAD module. Patran is better at meshing with hex 3D volumes.
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is easy
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
[Non-text portions of this message have been removed]
</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Ryan S
2007-01-14 16:46:00 UTC
Permalink
Ben,

If you're trying to the equivalent of the *pre-tension, there is the bolt
load feature. Kinematic coupling accepts nset/elset. Your point about Abaqus
needing node/element numbers is incorrect: most keywords accept
nsets/elsets, so you don't have to know the node/element numbers in your
model.

Maybe your example was a bad one, but I think I see your point. Cae should
do better about the lack of control of numbering and selection of mesh
entities.

Ryan
Post by BenZ
You have your point of view, well ! But I must answer on what I mean
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort and
create group with nodes, elements... different types of entity.
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that Abaqus/CAE
allows that. One has to go into the .inp. More generally, those who
never go into the .inp do not really know what is computed, it is a
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this in /CAE if
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes number.
Abaqus has this lack : it relies on labels of elements or nodes to
define a FEM.
BenZ.
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try and
guess.
Post by Ryan S
Also, I hope these statements are being made after using the latest
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's the
whole
Post by Ryan S
point of CAE. There are cases where you do, and that sucks, I
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get product
XYZ, it is
Post by Ryan S
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus coverage than
any other
Post by Ryan S
pre-processor (particularly for the newer features in abaqus). And
the
Post by Ryan S
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only
one
Post by Ryan S
Post by BenZ
thing when I was calculation engineer : checking the contact
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is typically
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an input
from /CAE with the default options will generate an awful .inp.
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise between
CAD
Post by Ryan S
Post by BenZ
and FEM. On the other hand, I think that I-Deas will remain the
most
Post by Ryan S
Post by BenZ
powerfull .inp generator because FEM is prioritized, but with a
very
Post by Ryan S
Post by BenZ
good CAD module. Patran is better at meshing with hex 3D volumes.
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
ps.com>, Jarry
Post by Ryan S
Post by BenZ
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation engineer,
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is
easy
Post by Ryan S
Post by BenZ
Post by Jarry
to get number of nodes/elements/dof in model. Not in abaqus,
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3 times
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
BenZ
2007-01-14 21:57:27 UTC
Permalink
Now I'm convinced that those who has never used Ansys cannot see
what we are talking about when comparing both technologies.

Thx, I know that Abaqus is based on ELSET/NSET. But one HAS TO
DEFINE THEM at a moment, nope ?! How ? By giving labels.
In Ansys, the selection power makes that the FEM is not very
dependent on the numerotation... Try it, you'll see what I'm talking
about.


BenZ.
Post by Ryan S
Ben,
If you're trying to the equivalent of the *pre-tension, there is the bolt
load feature. Kinematic coupling accepts nset/elset. Your point about Abaqus
needing node/element numbers is incorrect: most keywords accept
nsets/elsets, so you don't have to know the node/element numbers in your
model.
Maybe your example was a bad one, but I think I see your point. Cae should
do better about the lack of control of numbering and selection of mesh
entities.
Ryan
Post by BenZ
You have your point of view, well ! But I must answer on what I mean
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort and
create group with nodes, elements... different types of entity.
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that Abaqus/CAE
allows that. One has to go into the .inp. More generally, those who
never go into the .inp do not really know what is computed, it is a
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this in /CAE if
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes
number.
Post by Ryan S
Post by BenZ
Abaqus has this lack : it relies on labels of elements or nodes to
define a FEM.
BenZ.
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try and
guess.
Post by Ryan S
Also, I hope these statements are being made after using the latest
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's the
whole
Post by Ryan S
point of CAE. There are cases where you do, and that sucks, I
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get
product
Post by Ryan S
Post by BenZ
XYZ, it is
Post by Ryan S
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus coverage than
any other
Post by Ryan S
pre-processor (particularly for the newer features in abaqus). And
the
Post by Ryan S
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very weak
and cannot be used for professional FEA. It was useful for only
one
Post by Ryan S
Post by BenZ
thing when I was calculation engineer : checking the contact
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is
typically
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an
input
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
from /CAE with the default options will generate an
awful .inp.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Dissociating the mesh from the geometry in order to work on it is
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise
between
Post by Ryan S
Post by BenZ
CAD
Post by Ryan S
Post by BenZ
and FEM. On the other hand, I think that I-Deas will remain the
most
Post by Ryan S
Post by BenZ
powerfull .inp generator because FEM is prioritized, but
with a
Post by Ryan S
Post by BenZ
very
Post by Ryan S
Post by BenZ
good CAD module. Patran is better at meshing with hex 3D
volumes.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
40yahoogroups.com><ABAQUS%40yahoogrou
Post by Ryan S
Post by BenZ
ps.com>, Jarry
Post by Ryan S
Post by BenZ
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help you,
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation
engineer,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How could
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems like
this: in every fem-software I have been working with, it is
easy
Post by Ryan S
Post by BenZ
Post by Jarry
to get number of nodes/elements/dof in model. Not in
abaqus,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3
times
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-
lists!
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
Everything else is considered to be spam and therefore
deleted.
Post by Ryan S
Post by BenZ
Post by Ryan S
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
Ryan S
2007-01-15 14:44:16 UTC
Permalink
Ok, maybe I oversimplified. Let me try again:

Let's say you want to create a pressure load on a geometry face. You go:
Create Load -> Pressure -> Pick the face you want -> Put a magnitude in the
load dialog box ->OK.

See? No need to create a set or surface. No need to know about node/element
numbers. Cae creates the nset (or surface) on the face for you when you
created the load. In fact, your part can be totally unmeshed when you create
the pressure. And, if you remesh or partition the face, Cae recomputes the
(unseen) nset or surface.

Of course, if you created a nset or surface up front, you could use that in
your load instead of picking. But that's your choice.

Maybe ansys selection is superior. I wish I could see for myself how the
selection works, so there wouldn't be all this bandwidth wastage.

Ryan.
Post by BenZ
Now I'm convinced that those who has never used Ansys cannot see
what we are talking about when comparing both technologies.
Thx, I know that Abaqus is based on ELSET/NSET. But one HAS TO
DEFINE THEM at a moment, nope ?! How ? By giving labels.
In Ansys, the selection power makes that the FEM is not very
dependent on the numerotation... Try it, you'll see what I'm talking
about.
BenZ.
Post by Ryan S
Ben,
If you're trying to the equivalent of the *pre-tension, there is
the bolt
Post by Ryan S
load feature. Kinematic coupling accepts nset/elset. Your point
about Abaqus
Post by Ryan S
needing node/element numbers is incorrect: most keywords accept
nsets/elsets, so you don't have to know the node/element numbers
in your
Post by Ryan S
model.
Maybe your example was a bad one, but I think I see your point.
Cae should
Post by Ryan S
do better about the lack of control of numbering and selection of
mesh
Post by Ryan S
entities.
Ryan
Post by BenZ
You have your point of view, well ! But I must answer on what
I mean
Post by Ryan S
Post by BenZ
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort
and
Post by Ryan S
Post by BenZ
create group with nodes, elements... different types of entity.
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that Abaqus/CAE
allows that. One has to go into the .inp. More generally, those
who
Post by Ryan S
Post by BenZ
never go into the .inp do not really know what is computed, it
is a
Post by Ryan S
Post by BenZ
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this in /CAE if
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes
number.
Post by Ryan S
Post by BenZ
Abaqus has this lack : it relies on labels of elements or nodes
to
Post by Ryan S
Post by BenZ
define a FEM.
BenZ.
ps.com>, "Ryan
S" <
Post by Ryan S
Post by BenZ
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try
and
Post by Ryan S
Post by BenZ
guess.
Post by Ryan S
Also, I hope these statements are being made after using the
latest
Post by Ryan S
Post by BenZ
Post by Ryan S
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's
the
Post by Ryan S
Post by BenZ
whole
Post by Ryan S
point of CAE. There are cases where you do, and that sucks, I
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get
product
Post by Ryan S
Post by BenZ
XYZ, it is
Post by Ryan S
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus coverage
than
Post by Ryan S
Post by BenZ
any other
Post by Ryan S
pre-processor (particularly for the newer features in abaqus).
And
Post by Ryan S
Post by BenZ
the
Post by Ryan S
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very
weak
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
and cannot be used for professional FEA. It was useful for
only
Post by Ryan S
Post by BenZ
one
Post by Ryan S
Post by BenZ
thing when I was calculation engineer : checking the contact
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to other
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is
typically
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an
input
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
from /CAE with the default options will generate an
awful .inp.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Dissociating the mesh from the geometry in order to work on
it is
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with Abaqus.
Moreover, it is very powerful and make a good compromise
between
Post by Ryan S
Post by BenZ
CAD
Post by Ryan S
Post by BenZ
and FEM. On the other hand, I think that I-Deas will remain
the
Post by Ryan S
Post by BenZ
most
Post by Ryan S
Post by BenZ
powerfull .inp generator because FEM is prioritized, but
with a
Post by Ryan S
Post by BenZ
very
Post by Ryan S
Post by BenZ
good CAD module. Patran is better at meshing with hex 3D
volumes.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Jarry : manage to get I-Deas with in the simulation module
the 'Design' and 'meshing' options, with the Abaqus export
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
40yahoogroups.com><ABAQUS%40yahoogrou
Post by Ryan S
Post by BenZ
ps.com>, Jarry
Post by Ryan S
Post by BenZ
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help
you,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I expected
a "little" more...
For about 15 years I've been working as a calculation
engineer,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
with various fem-softwares (ansys, nastran, marc, adina,
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How
could
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems
like
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
this: in every fem-software I have been working with, it is
easy
Post by Ryan S
Post by BenZ
Post by Jarry
to get number of nodes/elements/dof in model. Not in
abaqus,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3
times
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are any)!
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them again!
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
This mailbox accepts e-mails only from selected mailing-
lists!
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
Everything else is considered to be spam and therefore
deleted.
Post by Ryan S
Post by BenZ
Post by Ryan S
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
BenZ
2007-01-15 21:37:15 UTC
Permalink
This is not a wastage. This is very interesting, from many points of
views...

BenZ.
Post by Ryan S
Create Load -> Pressure -> Pick the face you want -> Put a
magnitude in the
Post by Ryan S
load dialog box ->OK.
See? No need to create a set or surface. No need to know about
node/element
Post by Ryan S
numbers. Cae creates the nset (or surface) on the face for you
when you
Post by Ryan S
created the load. In fact, your part can be totally unmeshed when you create
the pressure. And, if you remesh or partition the face, Cae
recomputes the
Post by Ryan S
(unseen) nset or surface.
Of course, if you created a nset or surface up front, you could use that in
your load instead of picking. But that's your choice.
Maybe ansys selection is superior. I wish I could see for myself how the
selection works, so there wouldn't be all this bandwidth wastage.
Ryan.
Post by BenZ
Now I'm convinced that those who has never used Ansys cannot see
what we are talking about when comparing both technologies.
Thx, I know that Abaqus is based on ELSET/NSET. But one HAS TO
DEFINE THEM at a moment, nope ?! How ? By giving labels.
In Ansys, the selection power makes that the FEM is not very
dependent on the numerotation... Try it, you'll see what I'm
talking
Post by Ryan S
Post by BenZ
about.
BenZ.
Post by Ryan S
Ben,
If you're trying to the equivalent of the *pre-tension, there is
the bolt
Post by Ryan S
load feature. Kinematic coupling accepts nset/elset. Your point
about Abaqus
Post by Ryan S
needing node/element numbers is incorrect: most keywords accept
nsets/elsets, so you don't have to know the node/element
numbers
Post by Ryan S
Post by BenZ
in your
Post by Ryan S
model.
Maybe your example was a bad one, but I think I see your point.
Cae should
Post by Ryan S
do better about the lack of control of numbering and selection of
mesh
Post by Ryan S
entities.
Ryan
Post by BenZ
You have your point of view, well ! But I must answer on what
I mean
Post by Ryan S
Post by BenZ
by "selection power" because it is very împortant.
Selection power is the inverse of the difficulty to select, sort
and
Post by Ryan S
Post by BenZ
create group with nodes, elements... different types of
entity.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Strongly related to the initial question of Jarry !
Just an example : a screw pretensionned modeled by beams and some
KINEMATIC COUPLING.
1/To do the pretension, I'm still not convinced that
Abaqus/CAE
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
allows that. One has to go into the .inp. More generally, those
who
Post by Ryan S
Post by BenZ
never go into the .inp do not really know what is computed, it
is a
Post by Ryan S
Post by BenZ
necessary step of any calculation engineer to me.
2/to create the kinematic coupling, how do you do this
in /CAE if
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
you don't know the node labels ? For instance, in Ansys you can
select all nodes within a sphere, a cylinder... Then create the
rigid elements without the need to know the elements/nodes
number.
Post by Ryan S
Post by BenZ
Abaqus has this lack : it relies on labels of elements or nodes
to
Post by Ryan S
Post by BenZ
define a FEM.
BenZ.
40yahoogroups.com><ABAQUS%40yahoogrou
Post by Ryan S
Post by BenZ
ps.com>, "Ryan
S" <
Post by Ryan S
Post by BenZ
Post by Ryan S
Thought I should pass on replying, but there's so much
controversial stuff
Post by Ryan S
in some of the messages. Here goes.
Quote: *Very weak and cannot be used for professional FEA*
Obviously untrue. It's not so black-and-white. Ever.
Quote: *Selection power is zero*.
I don't understand what this means, and I'm not going to try
and
Post by Ryan S
Post by BenZ
guess.
Post by Ryan S
Also, I hope these statements are being made after using the
latest
Post by Ryan S
Post by BenZ
Post by Ryan S
available version of the product.
Quote: *... generates awful input files...*
You're not supposed to have to look at the input files. That's
the
Post by Ryan S
Post by BenZ
whole
Post by Ryan S
point of CAE. There are cases where you do, and that
sucks, I
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
agree.
Post by Ryan S
Quote: *... manage to get I-Deas ...*
More generalization. I am wary of advice that goes: *Get
product
Post by Ryan S
Post by BenZ
XYZ, it is
Post by Ryan S
the best*. No product is always much better at everything.
The one thing I will say in positive for CAE is itself a
generalization :-)
Post by Ryan S
: It generally does a better job at providing abaqus
coverage
Post by Ryan S
Post by BenZ
than
Post by Ryan S
Post by BenZ
any other
Post by Ryan S
pre-processor (particularly for the newer features in
abaqus).
Post by Ryan S
Post by BenZ
And
Post by Ryan S
Post by BenZ
the
Post by Ryan S
post-processor is fast.
Ryan.
Post by BenZ
Hiii mates, very interesting debate !
Please BE sarcastic Jarry, I do agree with you. /CAE is very
weak
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
and cannot be used for professional FEA. It was useful for
only
Post by Ryan S
Post by BenZ
one
Post by Ryan S
Post by BenZ
thing when I was calculation engineer : checking the
contact
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
pairs
Post by Ryan S
Post by BenZ
by reading the generated job (coming out from I-Deas or
Patran) ! It
Post by Ryan S
Post by BenZ
was a better technique that syntax checking the model then
reviewing
Post by Ryan S
Post by BenZ
surfaces one by one then pairs as a second step.
1/The "selection power" is zero, nothing compared to
other
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
proggys !
Post by Ryan S
Post by BenZ
The best at this task is Ansys. The problem you have is
typically
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
connected to this issue.
2/It is TOO MUCH strongly based on geometry ! Writing an
input
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
from /CAE with the default options will generate an
awful .inp.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Dissociating the mesh from the geometry in order to work on
it is
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
not easy (I do not want to say impossible lol !).
To Thomas : don't please don't forget Patran. Patran was
Abaqus/Pre
Post by Ryan S
Post by BenZ
some years ago, it is very well suited to work with
Abaqus.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Moreover, it is very powerful and make a good compromise
between
Post by Ryan S
Post by BenZ
CAD
Post by Ryan S
Post by BenZ
and FEM. On the other hand, I think that I-Deas will
remain
Post by Ryan S
Post by BenZ
the
Post by Ryan S
Post by BenZ
most
Post by Ryan S
Post by BenZ
powerfull .inp generator because FEM is prioritized, but
with a
Post by Ryan S
Post by BenZ
very
Post by Ryan S
Post by BenZ
good CAD module. Patran is better at meshing with hex 3D
volumes.
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Jarry : manage to get I-Deas with in the simulation
module
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
the 'Design' and 'meshing' options, with the Abaqus
export
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
module,
Post by Ryan S
Post by BenZ
then you'll be happy :)
My 2cents of the day.
BenZ.
40yahoogroups.com><ABAQUS%40yahoogrou
Post by Ryan S
Post by BenZ
ps.com>, Jarry
Post by Ryan S
Post by BenZ
Post by Jarry
Post by Fernando
OK, no need to get sarcastic, we are all trying to help
you,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
remember.
Post by Jarry
OK, I appologize. But frankly, I'm a little frustrated and
disapointed by abaqus. When I switched to it, I
expected
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
a "little" more...
For about 15 years I've been working as a calculation
engineer,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
with various fem-softwares (ansys, nastran, marc,
adina,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
cosmos/m,
Post by Ryan S
Post by BenZ
Post by Jarry
just to name a few), now 2 years I'm using abaqus, and my
feelsings
Post by Ryan S
Post by BenZ
Post by Jarry
are... mixed. That's the best, what I can say...
My 1st impression was: Wow, what a perfect "cad-like"
pre/processor
Post by Ryan S
Post by BenZ
Post by Jarry
(abaqus/cae)! How is this abaqus transparent, and user-
friendly!
Post by Ryan S
Post by BenZ
Post by Jarry
Much better, than patran, mentat or prep7/workbench! How
could
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
I only live without it!
Then suddenly I lost my enthusiasm, when I hit problems
like
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
this: in every fem-software I have been working with, it is
easy
Post by Ryan S
Post by BenZ
Post by Jarry
to get number of nodes/elements/dof in model. Not in
abaqus,
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
here you must learn python and write some script/macro!
Or wait till you assembly is complete, start datacheck!
Only then you are allowed to find that your model is 3
times
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
bigger then what you can solve...
Or do you want to have cut-view of your part/assembly?
No way, only in visualisation/results!
Maybe you want to know the length of the shortest edge in
your assembly? No-no you smart ass, select some length,
and we show you all that are shorter (if there are
any)!
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
What? You want 100% reliable backward compatibility?
Do not be ridiculous! Who does care, if your abaqus-6.5
models can be really transformed into 6.6! Do them
again!
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
What? Free hexa-mesher is what you want? Forget about it!
You need some meshing-practising, use mapped-meshing!
...etc, etc, etc, etc...
I do not know... Maybe I'm really too demanging...
Jarry
--
__________________________________________________________
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
This mailbox accepts e-mails only from selected
mailing-
Post by Ryan S
Post by BenZ
lists!
Post by Ryan S
Post by BenZ
Post by Ryan S
Post by BenZ
Post by Jarry
Everything else is considered to be spam and therefore
deleted.
Post by Ryan S
Post by BenZ
Post by Ryan S
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
Fernando
2007-01-12 19:17:00 UTC
Permalink
Post by BenZ
Hiii mates, very interesting debate !
Indeed, this is one of the most interesting threads in MONTHS!

My 2 cents here...

I fully agree with BenZ and others, in that CAE is WAY behind other
preprocessors. I have used a variety of them (the only big name
missing in my list is Patran, but I only hear good things about it),
and CAE is hands down the one that needs most work. In the end, it all
comes down to what you want to be able to do:

- I-DEAS is my darling for hand-meshing and hand-modifying meshes. Its
support for ABAQUS input files is also excellent (you can import an
*inp and modify it in I-DEAS!).
- For automatic meshing, I liked the way ANSYS handled the
geometry/mesh association, but you can't export ANSYS meshes to ABAQUS...
- I loved the integration between MARC and MENTAT for automatic mesh
refinement, it was kind of cool back when I used it to see the results
and the refined mesh as the simulation progressed...
- At the time I used Hypermesh, it had the best automatic mesher, it
was extremely fast and plain fantastic, once you got used to the
interface. The surface mesher did wonders, too. I follow its
developments (although I don't use it anymore), and now it can do mesh
morphing, which is a *REALLY REALLY REALLY COOL* thing, especially for
industry clients who do iterations and small modifications for their
designs: invest time in a really good mesh, then use it for all
iterations of the design, Hypermesh takes care of the modifications by
applying the old mesh to the new geometry!
- The fastest automatic, hex-dominant mesher around is called Harpoon.
You won't believe it until you have seen that baby in action. Its
support for geometry formats is kind of limited, and you will have to
get used to the paradigm, but it is extremely simple to use (the
manual is something like 20 pages, no need to say more). ABAQUS export
support is not the strongest, but the developers are very quick to
respond to demands and incorporate user-requested features, at least
for format improvements.
- The best automatic mesher on Earth is called CUBIT, a meshing
research code by Sandia National Labs. I am not sure it is available
for industrial customers though. It is the only code I have seen that
can hex-mesh nearly anything you feed it (ACIS geometry, just like
CAE, but on Windows it also accepts the Pro/Engineer geometry engine).
The paradigm is also part/assembly based, as CATIA, I-DEAS or CAE, and
the variety of meshers and smoothers is unmatched. Manual mesh editing
is a little limited, though, because the program was not designed with
that in mind.

In summary, there is no one-size-fits-all solution!

Regarding hex vs 2nd order tets... what can I say, as always it
depends! Tets have the advantage that you can easily mesh *any* volume
with them, unlike hexes (probably you can, it is only a matter of
effort and problem complexity). The problem with tets is that they
cause denser stiffness matrices (i.e. longer solver times) and are
more expensive to compute and postprocess than linear hexes. In the
end it is a matter of taste (BenZ likes 2nd order tets, I prefer hexes
:-D)

The purchase of ABAQUS by Dassault/CATIA is mixed news, I think: on
one hand, CAE has been and will continue to dramatically improve in
the geometry handling and modeling department, which is an extreme
necessity given the industry trend to do as much as possible with
computers in order to cut costs. Meshing and CAD are two different
things, however, and our problem is that an engineer needs both. CAE
has been concentrating in improving CAD, while meshing is still quite
primitive in my opinion. Other integrated packages (I-DEAS, ANSYS)
have had integrated meshing for ages, and it just shows. Give ABAQUS
some time, and maybe we look at a very different picture... did you
remember CAE from version 6.0 (6 years ago)? Enough said

Fernando
Cheng-Kong Wu
2007-01-12 02:24:30 UTC
Permalink
Jarry,

I did not read the whole story, and I only used
ABAQUS/CAE intensively for less than 1.5 years. I do
have some comments:

1. After you mesh a part, you should be able to see
how many elements you have. (Do a element quality
check). After you assembled them, you still can check
the number of element (again, element quality check).

2. If you want to know the shortest length of the
element, you can check it using "Verify Mesh" in the
Mesh module, and pay attention to the message show in
the window.

3. ABAQUS/CAE improved a lot in Hex mesh, and I think
it's better than I-DEAS and Hypermesh (I am not
familair with other software). I think it's really
powerful, but as you mentioned, the analyst needs a
lot of practice. I used other Hex mesher before, it
also require a lot of learning, and I think all Hex
mesher requires a lot of learning.

ABAQUS/CAE still have a lot of drawbacks (for example,
renumber the node or element IDs), but I think they
make good improvement in the past 2-3 years. Please
keep in mind that no single mesher can be the best in
all category, at least for now.

Best Regards,
Cheng-Kong



____________________________________________________________________________________
Want to start your own business?
Learn how on Yahoo! Small Business.
http://smallbusiness.yahoo.com/r-index
BenZ
2007-01-12 11:50:14 UTC
Permalink
Your point of view is interesting, but what you say depends on the
meshing techniques used. For instance, to me, advanced hex mesh can
only be manually done. With extrusions, hand-made modifications...
Beware, I mean quality meshes with warp below 16 and good aspect
ratio !
And at this task, I-Deas is very good.

BenZ.
Post by Fernando
Jarry,
I did not read the whole story, and I only used
ABAQUS/CAE intensively for less than 1.5 years. I do
1. After you mesh a part, you should be able to see
how many elements you have. (Do a element quality
check). After you assembled them, you still can check
the number of element (again, element quality check).
2. If you want to know the shortest length of the
element, you can check it using "Verify Mesh" in the
Mesh module, and pay attention to the message show in
the window.
3. ABAQUS/CAE improved a lot in Hex mesh, and I think
it's better than I-DEAS and Hypermesh (I am not
familair with other software). I think it's really
powerful, but as you mentioned, the analyst needs a
lot of practice. I used other Hex mesher before, it
also require a lot of learning, and I think all Hex
mesher requires a lot of learning.
ABAQUS/CAE still have a lot of drawbacks (for example,
renumber the node or element IDs), but I think they
make good improvement in the past 2-3 years. Please
keep in mind that no single mesher can be the best in
all category, at least for now.
Best Regards,
Cheng-Kong
_____________________________________________________________________
_______________
Post by Fernando
Want to start your own business?
Learn how on Yahoo! Small Business.
http://smallbusiness.yahoo.com/r-index
j***@gmx.net
2007-01-12 13:56:39 UTC
Permalink
For instance, to me, advanced hex mesh can only be manually done.
I understand your point of view, but consider this:
In school, or maybe small research projects, you can spend
days or weeks with meshing, until you are satisfied.

In common industry applications you do not have that time!
I usually get model(s) prepared in cad, I try to modify/simpify
it in cad, export, load into abaqus. For meshing I have hardly
1 day, and that is way too short for manually made hex meshing
even for simple parts (like crankshaft or piston - casting).

Industry needs robust and reliable free mesher. Or even
better, robust and reliable free hex-dominant mesher,
because element/node ratio is much better with hex-mesh
(not to talk about higher precision of hexa-elements).

BYT, do you know what is the lowest number of tet-elements
you need to mesh a cube space? 5!
But for meshing one tetrahedron-space, you need just single
hexahedral element. Degenerated, of course...

Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
BenZ
2007-01-12 16:51:10 UTC
Permalink
I do agree Jarry !

But sometimes, the clients want hex meshes and not 2nd order tetra
meshes. Why ? I still wonder... I have two theories :
-they make a competition between them in order to show the
beautifullest mesh,
-more seriously, they want FEM that are not too big in terms of dof,
as you said.

In France, big clients in the automotive industry request this kind
of mesh. So it is an industrial reality. And yes it can takes weeks,
but this is the willing of the client ;)
Beware, this kind of mesh can seriously damage your brain, so I
recommend using 2nd order tetras all the time !!!!

BenZ.
Post by j***@gmx.net
For instance, to me, advanced hex mesh can only be manually done.
In school, or maybe small research projects, you can spend
days or weeks with meshing, until you are satisfied.
In common industry applications you do not have that time!
I usually get model(s) prepared in cad, I try to modify/simpify
it in cad, export, load into abaqus. For meshing I have hardly
1 day, and that is way too short for manually made hex meshing
even for simple parts (like crankshaft or piston - casting).
Industry needs robust and reliable free mesher. Or even
better, robust and reliable free hex-dominant mesher,
because element/node ratio is much better with hex-mesh
(not to talk about higher precision of hexa-elements).
BYT, do you know what is the lowest number of tet-elements
you need to mesh a cube space? 5!
But for meshing one tetrahedron-space, you need just single
hexahedral element. Degenerated, of course...
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
Lieb, James C.
2007-01-12 18:22:05 UTC
Permalink
As far as using tetras all the time, there are classes of geometries that do not lend themselves well to tet meshes. Geometries such as pressure vessels that have a large diameter compared to wall thickness. When these geometries need to be analyzed using solid elements (like for transient thermal loading around a nozzle or other penetrations/attachments) the tetra meshes become entirely impractical. Multiple hex elements can be used through the thickness to get accurate through wall stress gradients, but they can have large aspect ratios when the gradients in the axial and hoop directions are not as significant (e.g. .25" thick by 10" in other directions). Tets cannot be generated with these kinds of aspect ratios. I created a demonstration model for a geometry such as this and it required over 1000 times more tet elements than hex elements to achieve the same number of elements through the wall thickness needed for accurate stress result. The computer required to solve a half million degree of freedom hex model is not that big a deal anymore, but to analyze the same geometry with tets need a computer than can solve over 5-10 million degrees of freedom.



From: ***@yahoogroups.com [mailto:***@yahoogroups.com] On Behalf Of BenZ
Sent: Friday, January 12, 2007 11:51 AM
To: ***@yahoogroups.com
Subject: [ABAQUS] Re: Number of nodes/elements in assembly?



I do agree Jarry !

But sometimes, the clients want hex meshes and not 2nd order tetra
meshes. Why ? I still wonder... I have two theories :
-they make a competition between them in order to show the
beautifullest mesh,
-more seriously, they want FEM that are not too big in terms of dof,
as you said.

In France, big clients in the automotive industry request this kind
of mesh. So it is an industrial reality. And yes it can takes weeks,
but this is the willing of the client ;)
Beware, this kind of mesh can seriously damage your brain, so I
recommend using 2nd order tetras all the time !!!!

BenZ.
Post by j***@gmx.net
For instance, to me, advanced hex mesh can only be manually done.
In school, or maybe small research projects, you can spend
days or weeks with meshing, until you are satisfied.
In common industry applications you do not have that time!
I usually get model(s) prepared in cad, I try to modify/simpify
it in cad, export, load into abaqus. For meshing I have hardly
1 day, and that is way too short for manually made hex meshing
even for simple parts (like crankshaft or piston - casting).
Industry needs robust and reliable free mesher. Or even
better, robust and reliable free hex-dominant mesher,
because element/node ratio is much better with hex-mesh
(not to talk about higher precision of hexa-elements).
BYT, do you know what is the lowest number of tet-elements
you need to mesh a cube space? 5!
But for meshing one tetrahedron-space, you need just single
hexahedral element. Degenerated, of course...
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
[Non-text portions of this message have been removed]
Cheng-Kong Wu
2007-01-12 15:31:44 UTC
Permalink
Benz,
From my experience and from what I learned from other,
the quality checks in I-DEAS and Hypermesh are far
better than ABAQUS/CAE. But to mesh a part that can
run using ABAQUS shouldn be fairly easy in CAE.

It took a lot of practice to get good quality hex mesh
using sweep mesh, and you need to know how to
partition the part. Even you can make you part in
yellow or green color does not mean you will get
usable hex mesh. Perform virtual topology first, then
do the cut helps a lot.

CAE still needs a lot of improvement in meshing, but
it did create good quality mesh. And some of the parts
I hex mesh in CAE, I think it's not easy or possible
to mesh in either I-DEAS or Hypermesh.

CAE lack of replace node function like Hypermesh,
otherwise it's easier to improve mesh quality.

Regards,
Cheng-Kong



____________________________________________________________________________________
Finding fabulous fares is fun.
Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains.
http://farechase.yahoo.com/promo-generic-14795097
j***@gmx.net
2007-01-12 12:13:27 UTC
Permalink
Post by Cheng-Kong Wu
1. After you mesh a part, you should be able to see
how many elements you have. (Do a element quality
check). After you assembled them, you still can check
the number of element (again, element quality check).
True, but not for assembly. It is evaluated per instance.
So you do not get total number of elements/nodes/dof in
assembly. Of course, I can go through that listing,
and add those 50 numbers together manually... :-)
Post by Cheng-Kong Wu
2. If you want to know the shortest length of the
element, you can check it using "Verify Mesh"
You can select elements shorter than..., and you get
info like "10 elements shorter, the shortest is...".
But all 10 elements get highlighted, so it is not
easy to find the shortest one. You must decrease that
limit more, and more, and more, till only 1 element
remains. Then you can highlight it, and find where
this particular shortest element is...
Post by Cheng-Kong Wu
3. ABAQUS/CAE improved a lot in Hex mesh
It might be true, but I think abaqus mesher is horrible!
It can not even do compatible elements between cells with
structural and free meshes without help of tie-constraint.

BTW, have you ever tried some hex-dominant automatic-mesher?
(that is a mesher, with which you can mesh the whole engine
block casting per one click, without partitioning, and you
get 99.8% regular hexahedral elements, and only ~0.2%
degenerated elements - but compatible! - like pyramid,
tetrahedral, triangular prism, etc.).

Believe me, if you did, you'd never like abaqus-mesher...

Similar situation is in abaqus-solvers: abaqus is probably
very strong in nonlinear routines, but comparing to other
products (e.g. ansys/pcg, cosmos/ffe) abaqus solvers are
very slow, and memory-greedy (I did comparison)...

And I think things get worse after Dassault Systemes
acquired Abaqus (imho the 2nd worst news in cae-industry,
after "msc/softvare took-over mdi/adams"). Abaqus will be
more and more "cad-like" product (for cad-engineers, who
do not know a lot about fem) with less emphasis on abaqus
as a specialised fem-calculation tool for proffesionals...

Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
BenZ
2007-01-12 16:55:00 UTC
Permalink
Once again, I do agree with you on the Abaqus performances compared
to Ansys.

But I disagree on the sad vision of the future you have about
abaqus ! Dassault has bought Abaqus to implement it in Catia as
Simulia. But Abaqus will still be developed and will remain reserved
to advanced engineers. To me, Simulia is the answer or Abaqus to
Workbench made by Ansys. And you know that Workbench lives on one
side, and that Ansys is still developped on the other side. I think
it'll be the same way with Simulia and the "real" Abaqus. I hope :)

BenZ.
Post by j***@gmx.net
Post by Cheng-Kong Wu
1. After you mesh a part, you should be able to see
how many elements you have. (Do a element quality
check). After you assembled them, you still can check
the number of element (again, element quality check).
True, but not for assembly. It is evaluated per instance.
So you do not get total number of elements/nodes/dof in
assembly. Of course, I can go through that listing,
and add those 50 numbers together manually... :-)
Post by Cheng-Kong Wu
2. If you want to know the shortest length of the
element, you can check it using "Verify Mesh"
You can select elements shorter than..., and you get
info like "10 elements shorter, the shortest is...".
But all 10 elements get highlighted, so it is not
easy to find the shortest one. You must decrease that
limit more, and more, and more, till only 1 element
remains. Then you can highlight it, and find where
this particular shortest element is...
Post by Cheng-Kong Wu
3. ABAQUS/CAE improved a lot in Hex mesh
It might be true, but I think abaqus mesher is horrible!
It can not even do compatible elements between cells with
structural and free meshes without help of tie-constraint.
BTW, have you ever tried some hex-dominant automatic-mesher?
(that is a mesher, with which you can mesh the whole engine
block casting per one click, without partitioning, and you
get 99.8% regular hexahedral elements, and only ~0.2%
degenerated elements - but compatible! - like pyramid,
tetrahedral, triangular prism, etc.).
Believe me, if you did, you'd never like abaqus-mesher...
Similar situation is in abaqus-solvers: abaqus is probably
very strong in nonlinear routines, but comparing to other
products (e.g. ansys/pcg, cosmos/ffe) abaqus solvers are
very slow, and memory-greedy (I did comparison)...
And I think things get worse after Dassault Systemes
acquired Abaqus (imho the 2nd worst news in cae-industry,
after "msc/softvare took-over mdi/adams"). Abaqus will be
more and more "cad-like" product (for cad-engineers, who
do not know a lot about fem) with less emphasis on abaqus
as a specialised fem-calculation tool for proffesionals...
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
Gurmeet Cheema
2007-01-12 18:57:40 UTC
Permalink
ABAQUS CAE mesher is not the best. But in the latest version v.6.6.-1 they have improved the mesher quite a bit. The partitioning tool is makes it easier to customize the mesh. I use both Hypermesh and ABQUS CAE for meshing. I think at this time Hypermesh is better, but ABQUS CAE is comming along.

I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the geometry. But comparing the two is like comparing apples and oranges. Workbench was designed to be a linear tool. And later changes have come as an after thought. ABAQUS CAE was originally designed to be the front end for full ABAQUS power. Therefore ABAQUS CAE is much much more powerful than Workbench. I think it is very unlikely that Workbench will catch up in the near future.

Gurmeet S. Cheema

BenZ <***@yahoo.fr> wrote:
Once again, I do agree with you on the Abaqus performances compared
to Ansys.

But I disagree on the sad vision of the future you have about
abaqus ! Dassault has bought Abaqus to implement it in Catia as
Simulia. But Abaqus will still be developed and will remain reserved
to advanced engineers. To me, Simulia is the answer or Abaqus to
Workbench made by Ansys. And you know that Workbench lives on one
side, and that Ansys is still developped on the other side. I think
it'll be the same way with Simulia and the "real" Abaqus. I hope :)

BenZ.
Post by j***@gmx.net
Post by Cheng-Kong Wu
1. After you mesh a part, you should be able to see
how many elements you have. (Do a element quality
check). After you assembled them, you still can check
the number of element (again, element quality check).
True, but not for assembly. It is evaluated per instance.
So you do not get total number of elements/nodes/dof in
assembly. Of course, I can go through that listing,
and add those 50 numbers together manually... :-)
Post by Cheng-Kong Wu
2. If you want to know the shortest length of the
element, you can check it using "Verify Mesh"
You can select elements shorter than..., and you get
info like "10 elements shorter, the shortest is...".
But all 10 elements get highlighted, so it is not
easy to find the shortest one. You must decrease that
limit more, and more, and more, till only 1 element
remains. Then you can highlight it, and find where
this particular shortest element is...
Post by Cheng-Kong Wu
3. ABAQUS/CAE improved a lot in Hex mesh
It might be true, but I think abaqus mesher is horrible!
It can not even do compatible elements between cells with
structural and free meshes without help of tie-constraint.
BTW, have you ever tried some hex-dominant automatic-mesher?
(that is a mesher, with which you can mesh the whole engine
block casting per one click, without partitioning, and you
get 99.8% regular hexahedral elements, and only ~0.2%
degenerated elements - but compatible! - like pyramid,
tetrahedral, triangular prism, etc.).
Believe me, if you did, you'd never like abaqus-mesher...
Similar situation is in abaqus-solvers: abaqus is probably
very strong in nonlinear routines, but comparing to other
products (e.g. ansys/pcg, cosmos/ffe) abaqus solvers are
very slow, and memory-greedy (I did comparison)...
And I think things get worse after Dassault Systemes
acquired Abaqus (imho the 2nd worst news in cae-industry,
after "msc/softvare took-over mdi/adams"). Abaqus will be
more and more "cad-like" product (for cad-engineers, who
do not know a lot about fem) with less emphasis on abaqus
as a specialised fem-calculation tool for proffesionals...
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Jarry
2007-01-12 20:37:26 UTC
Permalink
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importing and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line-area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry
Gurmeet Cheema
2007-01-13 17:30:39 UTC
Permalink
Jarry wrote:,"The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends "

I used to work with ANSYS (mostly using the workbench interface) before I switched to ABAQUS. At the time I did a comparison of two softwares and also talked to some users. I did not come across any user who said ANSYS is a better software than ABAQUS. I did find some who went the other way.

I do not believe one can make a statement that ANSYS solvers are faster. I think ABAQUS has a more extensive element library. I stand behind my earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access only limited numbers of features of ANSYS. On the other hand ABAQUS/CAE can access lot more features of ABAQUS.

In terms of usefulness of using geomtry for FEA model building, it is a matter of choice and experience. For those who have already invested 10 years of their life learning intricate features of the software, it is probably better to continue the old way. However for new enterants into the field, using geometry to apply loads, boundary conditions etc. is so much faster. Also if you change the mesh, you can still preserve these loads and boundary conditions.

I have not heard of the software called PREP7. Is this software still marketed by ANSYS?

In general having worked with both ANSYS and ABAQUS, I am happier in my experience with ABAQUS. I spent some trying to learn the so called classical interface of ANSYS and found it to be mess. Also ANSYS practice of having their distributors do the tech. support did not work for me. I found the distributors were a lot more busier doing consulting work and did not have sufficient time to support the product.

Gurmeet S. Cheema
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importing and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line-area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry



</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Jarry
2007-01-13 18:59:54 UTC
Permalink
I did not come across any user who said ANSYS is a better than ABAQUS
Neither did I. I think no one can say that, but it is possible
to name features, in which one is better then the other...
I do not believe one can make a statement that ANSYS solvers are faster.
Oh yes, can! Just prepare the same model (e.g. in patran, with
the same structural meshing for both ansys and abaqus), let it solve
in abaqus and ansys and compare time. But understand me correctly:
As I wrote, I noticed abaqus is more stable and faster when solving
nonlinear problems (needs less iteration sub-steps, converges faster).
But when it comes to linear solution (or 1 iteration step of nonlinear)
ansys/pcg is definitely faster, and less memory-hungry...
I think ABAQUS has a more extensive element library.
IIRC, ansys has more than 100 (I can find it) various element types.
Do you still think abaqus has a more extensive element library?
earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access
only limited numbers of features of ANSYS.
I do agree. That's why there is still older alternative interface
(prep7). Up to now, you can not make in Workbench everything, what
you can make with Prep7.
I have not heard of the software called PREP7. Is this software still marketed by ANSYS?
It is not stand-alone software, it is an integral part of ansys.
Prep7 is simply pre-processor (something like abaqus/cae, or patran).
May I ask how long you've been working with ansys? (me 15 years)
I think not long, if you never heard of Prep7 or Post26...
classical interface of ANSYS and found it to be mess.
AFAIK, what you call "classical interface" is Prep7. And it has
two modification. The first is so called "multi-windows" (like
Pro/E has), the second is "single-window" (like solidworks).
But the fact is, both Prep7-interface are badly structured.

Modern interface is definitely plus-point for abaqus/cae...

Jarry
--
_______________________________________________________________
This mailbox accepts e-mails only from selected mailing-lists!
Everything else is considered to be spam and therefore deleted.
Richard
2007-01-18 21:41:11 UTC
Permalink
Post by Jarry
Post by Gurmeet Cheema
I think ABAQUS has a more extensive element library.
IIRC, ansys has more than 100 (I can find it) various element types.
Do you still think abaqus has a more extensive element library?
Number of elements in a library is not a good indicator of product
quality or reliability. Having an extra 10 (or even 100) elements
available will not help if you cannot get convergence and your boss
wants a solution !

Best regards
Richard
ps: I just counted... Abaqus has 498 different elements !

PhilipS
2007-01-08 20:33:15 UTC
Permalink
Hi,
You can create a node set, let's call it "ALLN", in the assembly
module. Once created, you can see a message like this in the message
area of your CAE:
The set 'ALLN' has been created (43025 nodes).
So you will know that you have 43025 nodes overall. Same thing for
elements.

Alternatively, you can write a plug-in using python to know how many
elements or nodes you have in the assembly. This is a better choice
because even if you have native mesh, it can be helpful. For nodes
for example, if you model name is MN:

x=0
for Name in mdb.models['MN'].rootAssembly.instances.keys():
******x=x+len(mdb.models['MN'].rootAssembly.instance[Name].nodes)

print 'Number of nodes:', x

Please note that you should replace ****** with spaces. In other
word, in python format, you should tabify after "for".

There should be some easier ways ...
Regards
Post by j***@gmx.net
Hi,
How can I find total number of nodes/elements in assembly?
I know I can query instance mesh, but I do not want to
check all 50 instances and then summ it up, I would like
to know just total number of nodes/elements to estimate
problem size and its memory requirements...
Is it somehow possible to find this info in Abaqus/CAE?
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
Badel Pierre
2007-01-09 07:12:24 UTC
Permalink
Hello;
I think you could run a datacheck of your analysis. Then you can find
the information at the end of the .dat file.
It looks like the easiest way to me,
Pierre
Post by PhilipS
Hi,
You can create a node set, let's call it "ALLN", in the assembly
module. Once created, you can see a message like this in the message
The set 'ALLN' has been created (43025 nodes).
So you will know that you have 43025 nodes overall. Same thing for
elements.
Alternatively, you can write a plug-in using python to know how many
elements or nodes you have in the assembly. This is a better choice
because even if you have native mesh, it can be helpful. For nodes
x=0
******x=x+len(mdb.models['MN'].rootAssembly.instance[Name].nodes)
print 'Number of nodes:', x
Please note that you should replace ****** with spaces. In other
word, in python format, you should tabify after "for".
There should be some easier ways ...
Regards
Post by j***@gmx.net
Hi,
How can I find total number of nodes/elements in assembly?
I know I can query instance mesh, but I do not want to
check all 50 instances and then summ it up, I would like
to know just total number of nodes/elements to estimate
problem size and its memory requirements...
Is it somehow possible to find this info in Abaqus/CAE?
Jarry
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
<http://www.gmx.net/de/go/smartsurfer>
--
- Pierre Badel -
Doctorant - LaMCoS

INSA de Lyon - Bât. Jacquard
Avenue Jean Capelle Tél: (33) 4 72 43 83 08
69621 Villeurbanne Fax: (33) 4 72 43 85 25
http://lamcos.insa-lyon.fr









[Non-text portions of this message have been removed]
Fernando
2007-01-09 09:10:11 UTC
Permalink
Jarry,
Post by j***@gmx.net
How can I find total number of nodes/elements in assembly?
I know I can query instance mesh, but I do not want to
check all 50 instances and then summ it up, I would like
to know just total number of nodes/elements to estimate
problem size and its memory requirements...
Is it somehow possible to find this info in Abaqus/CAE?
Why don't you let ABAQUS do it for you? Do an "abaqus datacheck" and
check the resulting *.dat file, the memory requirements for the
simulation with the current setup are given there (you can influence
the memory requirements by changing the value of
standard_memory_policy in the file abaqus_v6.env (changes for this
simulation should be done in a copy of abaqus_v6.env placed in the
directory where the simulation runs)

Hope this helps,
Fernando
Maziar Mahzari
2007-01-14 11:18:08 UTC
Permalink
As an old fashioned person I don't like beautiful and fancy graphical interfaces. I prefer to interact through a text file and just check the model in the graphical interface. I wonder if you have taken advantage of APDL (programming language of ANSYS) but it is very powerful. I always enjoy mixing the power of programming with FE analysis. I’ve developed some macros which have saved me lots of time and in some cases I just changed some parameters and reused the macro.
I tried to do the same tricks with Python in /CAE but at this stage it’s not that powerful. Although I think /CAE has a firm foundation for development and its concepts are properly devised, but still not comparable to APDL.
Regarding to your comments on geometry concept, you should know that ANSYS/prep7/solution/post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.
Still depending on the problem (and the accessibility) you have to choose between them.
M. Mahzari

----- Original Message ----
From: Gurmeet Cheema <***@yahoo.com>
To: ***@yahoogroups.com
Sent: Saturday, 13 January, 2007 9:00:39 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Jarry wrote:,"The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends "

I used to work with ANSYS (mostly using the workbench interface) before I switched to ABAQUS. At the time I did a comparison of two softwares and also talked to some users. I did not come across any user who said ANSYS is a better software than ABAQUS. I did find some who went the other way.

I do not believe one can make a statement that ANSYS solvers are faster. I think ABAQUS has a more extensive element library. I stand behind my earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access only limited numbers of features of ANSYS. On the other hand ABAQUS/CAE can access lot more features of ABAQUS.

In terms of usefulness of using geomtry for FEA model building, it is a matter of choice and experience. For those who have already invested 10 years of their life learning intricate features of the software, it is probably better to continue the old way. However for new enterants into the field, using geometry to apply loads, boundary conditions etc. is so much faster. Also if you change the mesh, you can still preserve these loads and boundary conditions.

I have not heard of the software called PREP7. Is this software still marketed by ANSYS?

In general having worked with both ANSYS and ABAQUS, I am happier in my experience with ABAQUS. I spent some trying to learn the so called classical interface of ANSYS and found it to be mess. Also ANSYS practice of having their distributors do the tech. support did not work for me. I found the distributors were a lot more busier doing consulting work and did not have sufficient time to support the product.

Gurmeet S. Cheema
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importin g and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line- area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry

</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Gurmeet Cheema
2007-01-14 17:55:19 UTC
Permalink
Maziar Mahzari <***@yahoo.com> wrote;"Regarding to your comments on geometry concept, you should know that ANSYS/prep7/solution/post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.Still depending on the problem (and the accessibility) you have to choose between them."

ABAQUS/CAE gives one the option of working with the geometry or mesh alone. If one wants to work with mesh only (elements or nodes only) one can create orphan mesh parts. These are parts that are without any geometry. Therefore above statement about ABAQUS/CAE is not correct.

Gurmeet S. Cheema
Maziar Mahzari <***@yahoo.com> wrote:
As an old fashioned person I don't like beautiful and fancy graphical interfaces. I prefer to interact through a text file and just check the model in the graphical interface. I wonder if you have taken advantage of APDL (programming language of ANSYS) but it is very powerful. I always enjoy mixing the power of programming with FE analysis. I’ve developed some macros which have saved me lots of time and in some cases I just changed some parameters and reused the macro.
I tried to do the same tricks with Python in /CAE but at this stage it’s not that powerful. Although I think /CAE has a firm foundation for development and its concepts are properly devised, but still not comparable to APDL.
Regarding to your comments on geometry concept, you should know that ANSYS/prep7/solution/post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.
Still depending on the problem (and the accessibility) you have to choose between them.
M. Mahzari

----- Original Message ----
From: Gurmeet Cheema
To: ***@yahoogroups.com
Sent: Saturday, 13 January, 2007 9:00:39 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Jarry wrote:,"The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends "

I used to work with ANSYS (mostly using the workbench interface) before I switched to ABAQUS. At the time I did a comparison of two softwares and also talked to some users. I did not come across any user who said ANSYS is a better software than ABAQUS. I did find some who went the other way.

I do not believe one can make a statement that ANSYS solvers are faster. I think ABAQUS has a more extensive element library. I stand behind my earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access only limited numbers of features of ANSYS. On the other hand ABAQUS/CAE can access lot more features of ABAQUS.

In terms of usefulness of using geomtry for FEA model building, it is a matter of choice and experience. For those who have already invested 10 years of their life learning intricate features of the software, it is probably better to continue the old way. However for new enterants into the field, using geometry to apply loads, boundary conditions etc. is so much faster. Also if you change the mesh, you can still preserve these loads and boundary conditions.

I have not heard of the software called PREP7. Is this software still marketed by ANSYS?

In general having worked with both ANSYS and ABAQUS, I am happier in my experience with ABAQUS. I spent some trying to learn the so called classical interface of ANSYS and found it to be mess. Also ANSYS practice of having their distributors do the tech. support did not work for me. I found the distributors were a lot more busier doing consulting work and did not have sufficient time to support the product.

Gurmeet S. Cheema
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importin g and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line- area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry

</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Maziar Mahzari
2007-01-15 08:13:23 UTC
Permalink
Let me ask you some questions,
1-How do you make an orphan mesh (from another mesh generator? or another CAE session? That's not a good idea if you want to have all your model with MODEL and HISTORY data in a single script text file. That's a luxuray you have while you work with ANSYS/APDL, even parametrically)
2-Let say you have an orphan mesh in CAE. How do your define node sets based on some geometry rules (e.g. located on a flat surface, or using different arbitrary coordinate system) or attached to a Face (orphan meshes don't know about the geometry objects) in the python script?
3-How do you define a surface (which is the most important concept in ABAQUS) based on the node sets, element sets, location or a Face of the model in an orphan mesh inside a python script?
4-Assume you want to load a node of the model based on its location (x,y,z), how do you find its number and put it in your python script (ANSYS/APDL has a function node(x,y,z) which gets the node number)?
Note that I use a script to generate the model and frequently import it just to check. No feedback from CAE/GUI, just python programming.
M. Mahzari




----- Original Message ----
From: Gurmeet Cheema <***@yahoo.com>
To: ***@yahoogroups.com
Sent: Sunday, 14 January, 2007 9:25:19 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Maziar Mahzari <maziarmahzari@ yahoo.com> wrote;"Regarding to your comments on geometry concept, you should know that ANSYS/prep7/ solution/ post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.Still depending on the problem (and the accessibility) you have to choose between them."

ABAQUS/CAE gives one the option of working with the geometry or mesh alone. If one wants to work with mesh only (elements or nodes only) one can create orphan mesh parts. These are parts that are without any geometry. Therefore above statement about ABAQUS/CAE is not correct.

Gurmeet S. Cheema
Maziar Mahzari <maziarmahzari@ yahoo.com> wrote:
As an old fashioned person I don't like beautiful and fancy graphical interfaces. I prefer to interact through a text file and just check the model in the graphical interface. I wonder if you have taken advantage of APDL (programming language of ANSYS) but it is very powerful. I always enjoy mixing the power of programming with FE analysis. I’ve developed some macros which have saved me lots of time and in some cases I just changed some parameters and reused the macro.
I tried to do the same tricks with Python in /CAE but at this stage it’s not that powerful. Although I think /CAE has a firm foundation for development and its concepts are properly devised, but still not comparable to APDL.
Regarding to your comments on geometry concept, you should know that ANSYS/prep7/ solution/ post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.
Still depending on the problem (and the accessibility) you have to choose between them.
M. Mahzari

----- Original Message ----
From: Gurmeet Cheema
To: ***@yahoogroups. com
Sent: Saturday, 13 January, 2007 9:00:39 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Jarry wrote:,"The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends "

I used to work with ANSYS (mostly using the workbench interface) before I switched to ABAQUS. At the time I did a comparison of two softwares and also talked to some users. I did not come across any user who said ANSYS is a better software than ABAQUS. I did find some who went the other way.

I do not believe one can make a statement that ANSYS solvers are faster. I think ABAQUS has a more extensive element library. I stand behind my earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access only limited numbers of features of ANSYS. On the other hand ABAQUS/CAE can access lot more features of ABAQUS.

In terms of usefulness of using geomtry for FEA model building, it is a matter of choice and experience. For those who have already invested 10 years of their life learning intricate features of the software, it is probably better to continue the old way. However for new enterants into the field, using geometry to apply loads, boundary conditions etc. is so much faster. Also if you change the mesh, you can still preserve these loads and boundary conditions.

I have not heard of the software called PREP7. Is this software still marketed by ANSYS?

In general having worked with both ANSYS and ABAQUS, I am happier in my experience with ABAQUS. I spent some trying to learn the so called classical interface of ANSYS and found it to be mess. Also ANSYS practice of having their distributors do the tech. support did not work for me. I found the distributors were a lot more busier doing consulting work and did not have sufficient time to support the product.

Gurmeet S. Cheema
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importin g and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line- area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry

</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Gurmeet Cheema
2007-01-15 19:56:54 UTC
Permalink
Maziar,

You have raised some good points. Let me start off by saying that I had some limited experience with PREP7 and I remember that it had very extensive selection tools, which were not dependent on geometry.

1. Many times I have made the mesh in Hypermesh and imported it into CAE. Then I finish rest of modelling (loads, bcs etc.) inside CAE, because I find it easier. However I agree with you , once a mesh is imported into CAE (or an orphan mesh is created), not many changes can be made in the mesh inside CAE. The only change that I remember is you can change a linear mesh to quadratic or visa versa.

2. CAE does have a tool called 'selection by angle', which would let one pick the nodes/elements on a surface/face (you can define a small enough angle that only the nodes on the surface are picked). You can also pick nodes by window which is a laborious method of picking nodes. However I imagine the selections in PREP7 are better. The idea in CAE is that one decides about the node sets required in advance before the orphan mesh is created. Then one would create suitable partitions in geometry before meshing the part. Once the partitions are created, meshing process takes into account the partitioning surfaces and makes sure that no elements cross the partition boundary. For meshing purposes partition is almost like a separate part, except that the nodes on the boundary are connected to nodes on both sides of the boundary.

3. For picking nodes surfaces in CAE please refer to my comments in point 2. I can not answer your question about python scripting as I do not have any experience in python scripting. I am going for a scripting class next week and I may be able to answer your question after that.

4. In a modeler based on CAE geometry there is no need to identify a node based on co-ordinates. Nodes are identified based on geometric features (vertex, center of a line, distance from a vertex etc.). This is also the paradigm for modern CAD systems and makes picking a lot easier and faster.

I guess I understand your earlier statements better now. I need to restate my position slightly. If one wants to work with only FEA mesh for subsequent modelling (applying lods, bcs, picking node sets etc.) using the the geometry only for creating the mesh, PREP7 may work out to be better. However if one wants to use geometry for not only meshing but also for subsequent modelling (loads, bcs , node sets etc.) CAE is faster and better. For new users CAE is a lot easie to learn than PREP7. Workbench is not an alternative to CAE. Workbench is only a shadow of CAE.

Gurmeet S. Cheema
Ohio USA

Maziar Mahzari <***@yahoo.com> wrote:
Let me ask you some questions,
1-How do you make an orphan mesh (from another mesh generator? or another CAE session? That's not a good idea if you want to have all your model with MODEL and HISTORY data in a single script text file. That's a luxuray you have while you work with ANSYS/APDL, even parametrically)
2-Let say you have an orphan mesh in CAE. How do your define node sets based on some geometry rules (e.g. located on a flat surface, or using different arbitrary coordinate system) or attached to a Face (orphan meshes don't know about the geometry objects) in the python script?
3-How do you define a surface (which is the most important concept in ABAQUS) based on the node sets, element sets, location or a Face of the model in an orphan mesh inside a python script?
4-Assume you want to load a node of the model based on its location (x,y,z), how do you find its number and put it in your python script (ANSYS/APDL has a function node(x,y,z) which gets the node number)?
Note that I use a script to generate the model and frequently import it just to check. No feedback from CAE/GUI, just python programming.
M. Mahzari




----- Original Message ----
From: Gurmeet Cheema
To: ***@yahoogroups.com
Sent: Sunday, 14 January, 2007 9:25:19 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Maziar Mahzari wrote;"Regarding to your comments on geometry concept, you should know that ANSYS/prep7/ solution/ post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.Still depending on the problem (and the accessibility) you have to choose between them."

ABAQUS/CAE gives one the option of working with the geometry or mesh alone. If one wants to work with mesh only (elements or nodes only) one can create orphan mesh parts. These are parts that are without any geometry. Therefore above statement about ABAQUS/CAE is not correct.

Gurmeet S. Cheema
Maziar Mahzari wrote:
As an old fashioned person I don't like beautiful and fancy graphical interfaces. I prefer to interact through a text file and just check the model in the graphical interface. I wonder if you have taken advantage of APDL (programming language of ANSYS) but it is very powerful. I always enjoy mixing the power of programming with FE analysis. I’ve developed some macros which have saved me lots of time and in some cases I just changed some parameters and reused the macro.
I tried to do the same tricks with Python in /CAE but at this stage it’s not that powerful. Although I think /CAE has a firm foundation for development and its concepts are properly devised, but still not comparable to APDL.
Regarding to your comments on geometry concept, you should know that ANSYS/prep7/ solution/ post1/post26 ... (not talking about workbench) has control on both geometry (keypoint, line, area and volume) and finite element model (nodes and elements) where /CAE Python does not give you control on FE model. Having control on entities (geometric or FE) based on their number, location etc. is definitely stronger than .findAt() function in /CAE.
Still depending on the problem (and the accessibility) you have to choose between them.
M. Mahzari

----- Original Message ----
From: Gurmeet Cheema
To: ***@yahoogroups. com
Sent: Saturday, 13 January, 2007 9:00:39 PM
Subject: Re: [ABAQUS] Re: Number of nodes/elements in assembly?

Jarry wrote:,"The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends "

I used to work with ANSYS (mostly using the workbench interface) before I switched to ABAQUS. At the time I did a comparison of two softwares and also talked to some users. I did not come across any user who said ANSYS is a better software than ABAQUS. I did find some who went the other way.

I do not believe one can make a statement that ANSYS solvers are faster. I think ABAQUS has a more extensive element library. I stand behind my earlier statements on ABAQUS/CAE vs. Workbench. Work bench can access only limited numbers of features of ANSYS. On the other hand ABAQUS/CAE can access lot more features of ABAQUS.

In terms of usefulness of using geomtry for FEA model building, it is a matter of choice and experience. For those who have already invested 10 years of their life learning intricate features of the software, it is probably better to continue the old way. However for new enterants into the field, using geometry to apply loads, boundary conditions etc. is so much faster. Also if you change the mesh, you can still preserve these loads and boundary conditions.

I have not heard of the software called PREP7. Is this software still marketed by ANSYS?

In general having worked with both ANSYS and ABAQUS, I am happier in my experience with ABAQUS. I spent some trying to learn the so called classical interface of ANSYS and found it to be mess. Also ANSYS practice of having their distributors do the tech. support did not work for me. I found the distributors were a lot more busier doing consulting work and did not have sufficient time to support the product.

Gurmeet S. Cheema
Post by Gurmeet Cheema
I have used both the ANSYS Workbench and ABAQUS CAE. Both work with the
geometry. But comparing the two is like comparing apples and oranges.
Workbench was designed to be a linear tool. And later changes have come
as an after thought. ABAQUS CAE was originally designed to be the front
end for full ABAQUS power. Therefore ABAQUS CAE is much much more
powerful than Workbench. I think it is very unlikely that Workbench will
catch up in the near future.
Workbench is something like next-gen "cad-like" pre-processor
for Ansys, but apart from that, there is still old good Prep7.
And frankly, abaqus/cae can not be compared to Prep7 concerning
meshing capabilities (I could name a lot of meshing-features
abaqus/cae can not handle, but why? those who worked with both
of them know it)...

On the other side, in geometry processing/importin g and modelling
abaqus/cae is much-much better than old-fashioned bottom-up
"point-line- area-volume" prep7.

The question is, what is more important for us? Perfect geometry
handling, modern interface and better geometry core of abaqus,
or robust mesher, extensive element library and fast solvers
of ansys? It depends...

BTW, Workbench is neither linear, nor nonlinear tool. It is
simply new pre/post-processor (something like abaqus/cae).
Behind it, there is still ansys/solver.

Jarry

</body>
<!--~-|**|PrettyHtmlStart|**|-~-->
<head>
<style type="text/css">
<!--
#ygrp-mlmsg {font-size:13px; font-family: arial,helvetica,clean,sans-serif;*font-size:small;*font:x-small;}
#ygrp-mlmsg table {font-size:inherit;font:100%;}
#ygrp-mlmsg select, input, textarea {font:99% arial,helvetica,clean,sans-serif;}
#ygrp-mlmsg pre, code {font:115% monospace;*font-size:100%;}
#ygrp-mlmsg * {line-height:1.22em;}
#ygrp-text{
font-family: Georgia;
}
#ygrp-text p{
margin: 0 0 1em 0;
}
#ygrp-tpmsgs{
font-family: Arial;
clear: both;
}
#ygrp-vitnav{
padding-top: 10px;
font-family: Verdana;
font-size: 77%;
margin: 0;
}
#ygrp-vitnav a{
padding: 0 1px;
}
#ygrp-actbar{
clear: both;
margin: 25px 0;
white-space:nowrap;
color: #666;
text-align: right;
}
#ygrp-actbar .left{
float: left;
white-space:nowrap;
}
.bld{font-weight:bold;}
#ygrp-grft{
font-family: Verdana;
font-size: 77%;
padding: 15px 0;
}
#ygrp-ft{
font-family: verdana;
font-size: 77%;
border-top: 1px solid #666;
padding: 5px 0;
}
#ygrp-mlmsg #logo{
padding-bottom: 10px;
}

#ygrp-vital{
background-color: #e0ecee;
margin-bottom: 20px;
padding: 2px 0 8px 8px;
}
#ygrp-vital #vithd{
font-size: 77%;
font-family: Verdana;
font-weight: bold;
color: #333;
text-transform: uppercase;
}
#ygrp-vital ul{
padding: 0;
margin: 2px 0;
}
#ygrp-vital ul li{
list-style-type: none;
clear: both;
border: 1px solid #e0ecee;
}
#ygrp-vital ul li .ct{
font-weight: bold;
color: #ff7900;
float: right;
width: 2em;
text-align:right;
padding-right: .5em;
}
#ygrp-vital ul li .cat{
font-weight: bold;
}
#ygrp-vital a {
text-decoration: none;
}

#ygrp-vital a:hover{
text-decoration: underline;
}

#ygrp-sponsor #hd{
color: #999;
font-size: 77%;
}
#ygrp-sponsor #ov{
padding: 6px 13px;
background-color: #e0ecee;
margin-bottom: 20px;
}
#ygrp-sponsor #ov ul{
padding: 0 0 0 8px;
margin: 0;
}
#ygrp-sponsor #ov li{
list-style-type: square;
padding: 6px 0;
font-size: 77%;
}
#ygrp-sponsor #ov li a{
text-decoration: none;
font-size: 130%;
}
#ygrp-sponsor #nc {
background-color: #eee;
margin-bottom: 20px;
padding: 0 8px;
}
#ygrp-sponsor .ad{
padding: 8px 0;
}
#ygrp-sponsor .ad #hd1{
font-family: Arial;
font-weight: bold;
color: #628c2a;
font-size: 100%;
line-height: 122%;
}
#ygrp-sponsor .ad a{
text-decoration: none;
}
#ygrp-sponsor .ad a:hover{
text-decoration: underline;
}
#ygrp-sponsor .ad p{
margin: 0;
}
o {font-size: 0; }
.MsoNormal {
margin: 0 0 0 0;
}
#ygrp-text tt{
font-size: 120%;
}
blockquote{margin: 0 0 0 4px;}
.replbq {margin:4}
-->
</style>
</head>
<!--~-|**|PrettyHtmlEnd|**|-~-->
</html><!--End group email -->
Continue reading on narkive:
Loading...