Any ideas on how to share/collaborate a FEM with a colleague?

[Jetgirl8]

My company uses Femap as our pre- and post-processor. Does anyone have any ideas or creative solutions for sharing a finite element model with a colleage? This is not something that can be divided into submodels. I have already considered and pitched working 2 shifts to management. They would prefer if 2 people could be optimizing at the same time during the day due to shop workload issues.

 

[rb1957]

two people simultaneously editting a model ... "i don't think so"

i think the only way would be to divide into two super-elements, you can divide the model at a reasonably arbitary boundary (section plane), maybe at a "convenient" set of nodes ??

if you Had to, work on different groups of nodes/elements, export the bdf for each then combine the two altered sets ... but i feel really dirty even suggesting that approach ...

you could check this horrible approach by keeping a master bdf and comparing both altered bdfs to it ... nodes elements should change in only one or the other of the altered bdfs, but not both.

but then that approach would probably double the amount of time to change the model ... but rationality was never a convincing argument with mgmt.

 

[Jetgirl8]

Thanks. Just thought I'd poll to make sure there was some elegant trick I was missing out on.

 

[mlamping]

Could you provide a little more detail regarding the model? You state you can't submodel it, so that tells me there's no convenient boundary for you to segment the model between the two analysts. Is it a solid model? Is it an assembly? If it's a single solid component could you consider cutting it at a plane and use certain solver techniques to connect the two solids? I'm thinking you could use NX Nastran surface-surface glue capabilities or ABAQUS tied contact.

As far as sharing the data, you can start with edited geometry in the form of a PARASOLID or STEP file. Maybe there's a better solution depending on your CAD package. Both of you work independent on your half, then 2 Nastran input files could be read into the same FEMAP model for the analysis runs.

As you can see, I've got more questions than answers at this point. I'm mostly speculating and throwing out suggestions for a still unclear problem you're trying to solve. It is an interesting one though.

Regards,
Mark

Mark Lamping
CAE Technical Consultant
Siemens PLM Software

 

[Jetgirl8]

mlamping, you are correct in that there is no convenient boundary to segment the model. It is an assembly model and the global loads are shared across the structure - what can be removed and modeled separately has already been done. I really do not see a workable solution apart from the 2 shift proposal I made earlier. I considered the super-element approach, but again I struggle with an appropriate boundary plane. I don't really need a model review, I was just curious if there was a tip or trick that I was unaware of for finite element model sharing. I know there are CAD tips and tricks but I just couldn't seem to apply them to the FE world in my mind.

Thanks for your time.

 

[rb1957]

if you're just dividing up the model so two people can edit it at the same time, you can divide it anywhere ... draw a box on the model, pick "items entirely within", ... not nice, but better than anything else i can think of ...

 

[trainguy]

Can you define an interface of certain nodes that have specific numbering, and each analyst works his/her respective half of the model in 2 distinct Femap files?

Then, for example, 1 analyst uses only elements numbered 1 - 10 000, and the other uses elements numbered 20 000 to 30 000, such that there is no overlap.

Same for the nodes, except the common ones.

Then, you could export each of these halfs to 2 Nastran files, then, using a text editor, combine the Nastran files.

In theory, very doable, but you have to manipulate the Nastran text files.

You also have to each be working in consistent global coordinates, so you don't end up with an arm where your foot should be...

Once combined, you could import into a new Femap file and add loads, etc.

Hope it helps.

tg

 

[Jetgirl8]

rb1957 & trainguy -
Thank you both. I think those ideas are very workable solutions. I guess one bit of the puzzle that I have not mentioned is that one of the analysts is a very disorganized person, so I've been trying to avoid things that require work and rules and I know will be ignored. Again, thanks for all the comments and I think I've got a viable path forward that will hopefully minimize rework for all.

 

[rb1957]

debugging it (as i'm sure you'll have to) will be a nightmare !

 

[Jetgirl8]

Yes, that's the part I'm dreading.

 

[rb1957]

maybe get your colleague to do that part ?

 

[trainguy]

Jetgirl8,

You mean there are people in Aerospace who avoid things that require work?

Sounds like some of my Rail people...

tg