Using Abaqus mesh as input

[Stephen Thomas]

Hi,

As part of acquainting myself with MOOSE and FEA, I am trying to run example ex01 by replacing mug.e with my own Abaqus format mesh file (attached: anm.inp) by just modifying the ex01.i (attached). The tetrahedral mesh represents an irregular shaped particle embedded within a box. I am able to open this in cubit and the shapes are as expected. I assigned two opposite faces of the box as "top" nodeset and "bottom" nodeset.  However, I am facing two issues when I load it in MOOSE.
1) When I open the inp file in peacock, the box shape is distorted for some reason.

2) When I execute the program, it fails immediately due to the following error:       
ERROR: negative Jacobian: -18.3084 in element 1528

Could someone guide me in the right direction?

Thank you,
Stephen Thomas.

[Peterson, JW]

On Tue, Jul 22, 2014 at 2:26 PM, Stephen Thomas <stephen...@gmail.com> wrote:

Hi,

As part of acquainting myself with MOOSE and FEA, I am trying to run example ex01 by replacing mug.e with my own Abaqus format mesh file (attached: anm.inp) by just modifying the ex01.i (attached). The tetrahedral mesh represents an irregular shaped particle embedded within a box. I am able to open this in cubit and the shapes are as expected. I assigned two opposite faces of the box as "top" nodeset and "bottom" nodeset.  However, I am facing two issues when I load it in MOOSE.
1) When I open the inp file in peacock, the box shape is distorted for some reason.

Can you also send us screenshot of what it *should* look like?

 

 

2) When I execute the program, it fails immediately due to the following error:       
ERROR: negative Jacobian: -18.3084 in element 1528

Could someone guide me in the right direction?

This could be caused by a tetrahedron being read incorrectly from file and therefore getting inverted...  or it could be caused by your mesh generator generating a bad mesh.

I'll see if I can figure out anything more.

 

--
John

[Stephen Thomas]

On Tuesday, July 22, 2014 2:37:20 PM UTC-6, jw.peterson wrote:

On Tue, Jul 22, 2014 at 2:26 PM, Stephen Thomas <stephen...@gmail.com> wrote:

Hi,

As part of acquainting myself with MOOSE and FEA, I am trying to run example ex01 by replacing mug.e with my own Abaqus format mesh file (attached: anm.inp) by just modifying the ex01.i (attached). The tetrahedral mesh represents an irregular shaped particle embedded within a box. I am able to open this in cubit and the shapes are as expected. I assigned two opposite faces of the box as "top" nodeset and "bottom" nodeset.  However, I am facing two issues when I load it in MOOSE.
1) When I open the inp file in peacock, the box shape is distorted for some reason.

Can you also send us screenshot of what it *should* look like?

It should look like this: 

[Stephen Thomas]

I just noticed that a similar distorted shape shows up when I import my inp file into Cubit with the "Import Geometry" option enabled.

Now I should give you a short history about this situation. When I imported my original inp file (attached: anm_orig.inp) into Cubit, it looked fine:

But this mesh did not have the nodesets "top" and "bottom". I had modified the original mesh file to rearrange the nodes to group them into 3 nodesets: "top"(nodes on the top surface of the box), "bottom" (nodes on the bottom surface of the box") and "MeshNode" (rest of the nodes). After I did this rearrangement, the node ids are no longer in increasing order as in the original inp file. Could this be a problem?

[Peterson, JW]

On Tue, Jul 22, 2014 at 4:03 PM, Stephen Thomas <stephen...@u.boisestate.edu> wrote:

I just noticed that a similar distorted shape shows up when I import my inp file into Cubit with the "Import Geometry" option enabled.

Now I should give you a short history about this situation. When I imported my original inp file (attached: anm_orig.inp) into Cubit, it looked fine:

I guess I am not as familiar with this mesh/geometry as you are, but I don't see a big difference between the two images...

 

But this mesh did not have the nodesets "top" and "bottom". I had modified the original mesh file to rearrange the nodes to group them into 3 nodesets: "top"(nodes on the top surface of the box), "bottom" (nodes on the bottom surface of the box") and "MeshNode" (rest of the nodes). After I did this rearrangement, the node ids are no longer in increasing order as in the original inp file. Could this be a problem?

It's possible... I don't know if reading an arbitrary node numbering has ever been tested in the Abaqus reader...

--
John

[Peterson, JW]

On Tue, Jul 22, 2014 at 4:19 PM, Peterson, JW <jw.pe...@inl.gov> wrote:
>
>>
>> But this mesh did not have the nodesets "top" and "bottom". I had modified the original mesh file to rearrange the nodes to group them into 3 nodesets: "top"(nodes on the top surface of the box), "bottom" (nodes on the bottom surface of the box") and "MeshNode" (rest of the nodes). After I did this rearrangement, the node ids are no longer in increasing order as in the original inp file. Could this be a problem?
>
>
> It's possible... I don't know if reading an arbitrary node numbering has ever been tested in the Abaqus reader...

Here's the error I get when I try to read in your Abaqus Mesh from a
test program (not Peacock).

[0] ../src/mesh/abaqus_io.C, line 535, compiled Jul 18 2014 at 15:13:04
Unrecognized element type: *ELEMENT, ELSET=MESHTRI0, TYPE=S3R

I've never seen an Abaqus input file with elements of type "S3R", but
if they are just triangles (equivalent to "CPS3") we can probably
support them.

But if you are already using Cubit anyway, just export the mesh as
type "Exodus". That will hopefully work quite a bit better.

--
John

[Peterson, JW]

On Wed, Jul 23, 2014 at 8:46 AM, Peterson, JW <jw.pe...@inl.gov> wrote:
>
> But if you are already using Cubit anyway, just export the mesh as
> type "Exodus". That will hopefully work quite a bit better.

Yep, definitely try this, because even beyond the S3R issue, there is
something else wrong with the connectivity (possibly the node
numbering issue you mentioned) which is causing the problem you're
seeing.

--
John

[Peterson, JW]

So, the main problem is that your Abaqus file has multiple *NODE
sections, which our Abaqus reader is not set up to handle... I've only
ever seen Abaqus files with a single *NODE section up to now.

What happens is that the nodes from the "bottom" nodeset overwrite the
"top" nodeset, and then the nodes from the "meshNodeSet" overwrite
those. This explains why the final Mesh is actually missing a good
chunk of the nodes from the outside of the domain, and why the
connectivity is messed up. I've started issue #3587 on the MOOSE
github to continue the conversation, but I don't know when I'll get
time to actually work on it. Please let me know if you can export the
mesh from Cubit in another format that works better in the meantime.

--
John

[Gaston, Derek R]

By the way - when using Cubit the "Genesis" format is the same thing as "Exodus II"... Cubit uses different words in different places for some reason...

Derek

--
John

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.
Visit this group at http://groups.google.com/group/moose-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/CANj%2BmwCiaFgGG6OPhkdfUk4HoaoWz42u5gfV3wxU6rXwT%3D818w%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

[Stephen Thomas]

Hi John,

Thanks for looking into it.

How can I specify the "top" and "bottom" boundary conditions with a single *NODE section? 

Thank you,
Stephen.

[Peterson, JW]

On Thu, Jul 24, 2014 at 5:05 PM, Stephen Thomas
<stephen...@gmail.com> wrote:
> Hi John,
>
> Thanks for looking into it.
>
> How can I specify the "top" and "bottom" boundary conditions with a single
> *NODE section?

Well, the way I've seen it in the past is that the file will have 1
*NODE section with all the nodes, and then several *NSET sections
which organized the nodes into the different boundary condition node
sets.

I don't know if/how you can control this from within Abaqus itself.

--
John

[Stephen Thomas]

Hi John,

Will you be able to send me a sample input file where the above mentioned format is used?

Thank you,
Stephen Thomas.

[Stephen Thomas]

On Monday, July 28, 2014 1:25:25 PM UTC-6, Stephen Thomas wrote:

Hi John,

Will you be able to send me a sample input file where the above mentioned format is used?

Also, how do you guys normally debug MOOSE code? Is there a way to set breakpoints and step through code?

[Peterson, JW]

On Mon, Jul 28, 2014 at 3:16 PM, Stephen Thomas
<stephen...@u.boisestate.edu> wrote:
>
>
> On Monday, July 28, 2014 1:25:25 PM UTC-6, Stephen Thomas wrote:
>>
>> Hi John,
>>
>> Will you be able to send me a sample input file where the above mentioned
>> format is used?

Sent you one off-list.

> Also, how do you guys normally debug MOOSE code? Is there a way to set
> breakpoints and step through code?

I usually use gdb/lldb for gcc/clang compiled programs. The syntax is
a little different between the two, but for GCC you would launch it
with:

gdb --args <your program> <your program args>

To set a breakpoint I usually give the file and line number:

b foo.C:1234

There are lots of gdb/lldb tutorials available on the web.

--
John