[MOAB-dev] mbconvert from abaqus to h5m

[Jacob Hochhalter]

Hi Paul,

Ok, thanks.  It is no problem for me to add the *PARTS and *INSTANCE 
construct to the inp files from now on.

For me, the more important thing is to support common 3D element types.  
After looking through ReadABAQUS.cpp I see that there are five 3D 
element types supported:  "DC3D8", "DCC3D8", "C3D8R", "C3D4", and 
"DC3D4".  However, "DC3D8", "DCC3D8", and "C3D8R" are all the same 
element topology, 8-noded hexahedra (all set to MBHEX), and "C3D4" and 
"DC3D4" are both 4-noded tetrahedra (all set to MBTET).  "C3D8" is the 
abaqus default for an 8-noded hexahedron, and it could be supported by 
replicating what is already done for the other varieties of 8-noded 
hexahedra.

I saw that the patran and exodus converters have a good layout for 
supporting more element types.  Could that layout be used for abaqus, in 
which case we would just need to define the element connectivity mapping 
from abaqus-to-moab (as is done in exodus_order.h and patran_order.h)?  
If so, I can provide those mappings for the abaqus connectivities.

Iulian, in response to your previous message, yes, I will be reading the 
converted files into a finite element code which uses moab as it's input 
format for the mesh.

Thanks,
- Jake

On 04/22/2015 10:28 AM, Paul Wilson wrote:
> Hello Jacob,
>
> Since I was the main author of this (nearly 6 years ago), I'll add 
> some comments.
>
> The use case I was working always had PARTs defined separately from 
> ASSEMBLYs that were composed of INSTANCEs of those PARTs.  I hope it 
> won't be hard to support your use case, which doesn't seem to use any 
> of those constructs.
>
> I was able to load it successfully without any changes to MOAB by 
> adding PART and ASSEMBLY information, as in the attached file.
>
> For future reference, it also only currently supports a subset of the 
> element types and metadata.  It will definitely support the linear 
> tets that are in your sample, but does not support many higher order 
> elements.
>
> Paul
>
> On 04/21/2015 05:59 PM, Grindeanu, Iulian R. wrote:
>> thanks for reporting the problem.
>> this reader is not tested properly
>>
>> moab is looking for *PART or *ASSEMBLY keywords.
>>
>> Even if I modified the file, it is not read correctly
>>
>> I am working on it
>>
>> I will make a pull request
>>
>> Thanks,
>> Iulian
>>
>>
>> ________________________________________
>> From: moab-dev-bounces at mcs.anl.gov [moab-dev-bounces at mcs.anl.gov] on 
>> behalf of Jacob Hochhalter [Jacob.D.Hochhalter at nasa.gov]
>> Sent: Tuesday, April 21, 2015 12:56 PM
>> To: moab-dev at mcs.anl.gov
>> Subject: [MOAB-dev] mbconvert from abaqus to h5m
>>
>> Hello,
>>
>> I would like to use the mbconvert tool to translate abaqus '.inp' files
>> to '.h5m' files.  I have tried the following command:
>>
>> mbconvert -g tet4.inp tet4.h5m
>>
>> on moab 4.6 and 4.8 (tet4.inp is attached), but both return '.h5m' files
>> without any mesh data.
>>
>> 'mbconvert -l' returns:
>>
>> Format  Description                    Read  Write  File Name Suffixes
>> ------  -----------------------------  ----  ----- ------------------
>> ...
>> Abaqus mesh  ABAQUS INP mesh format    yes     no  abq
>> ...
>>
>> which makes me think I should be able to run the conversion.
>>
>> Am I missing something, or is this not yet functional?
>>
>> Thanks,
>> - Jake
>