Workflow for Large Assemblies and Multiple Analysis Cases

[dnunezHSLXH]

I have an assembly model that is currently at ~208k elements/ ~406k nodes.  All contacts are bonded.  I have to run 11 analysis studies as specified from customer requirements.  The assembly file size is 1.3 GB.  It is painfully slow to open up NASTRAN (~41 minutes) for the first time and also very slow to switch between analysis (~10-20 minutes).  I apologize for not sharing the model.  I am hesitant to do so because of company policy. 

Does anyone know of an efficient workflow to deal with this scenario?  I am thinking of converting the assembly model to a solid body part to remove the contacts.  Not sure how much this would help.  I am also thinking of grouping the analysis files into multiple assemblies to increase file opening speed, but there are obvious downsides to that as well.

Any feedback is welcome, thank you in advance!

Laptop Specs:

Processor 11th Gen Intel(R) Core(TM) i7-11850H @ 2.50GHz
Installed RAM 48.0 GB (47.2 GB usable)
Edition Windows 11 Pro
Version 22H2
Installed on ‎10/‎11/‎2023
OS build 22621.3007
Experience Windows Feature Experience Pack 1000.22681.1000.0

[John_Holtz · Accepted solution]

Hi @dnunezHSLXH 

~208k elements/ ~406k nodes is not that large, but other aspects of the model may contribute to the speed.

1. How many part are in the model? The two numbers at the very bottom right corner of the Inventor interface indicate something, maybe the number of times all the parts are inserted in the model and the number of files used. For example, if you have 2 different plates and 50 identical bolts, the numbers would be 52 (=2+50) and 4 (= 2 part files for the plate + 1 part file for the bolt, ignoring nut and washer + 1 assembly file).
2. How many contact entries are in the model tree? If there are more than a few dozen, it is better to define one "Contacts > Solver" since it sounds like everything is bonded.
3. What analysis type are you using? If you are doing a linear static analysis, you only need one analysis with 11 subcases. That will reduce the size of the file (since there will be only 1 mesh instead of 11 meshes).
4. I sure hope your model is saved on the local hard disk (usually C: or D:) and not on a network drive. Network communication is slow. Also, you may not want to save the model in a folder that is automatically backed up (such as to OneDrive). In the old days, folders such as "Documents" and "Desktop" were just regular folders under "C:\users\username\Documents" and "Desktop". I have noticed lately that these folder are saved under "C:\users\username\OneDrive\Documents" which means they get backed up. Hopefully that does not interfere with the speed, and we have never known Windows or any other software to do something silly like that. 🙂 (Windows does "hide" the fact that Documents in under OneDrive unless you really dig into it.)
5. What version of Nastran are you using? 

Merging the separate CAD parts would reduce the number of contacts. That would be helpful if you have many contacts listed in the model tree and did not want to replace them with solver contact. (Deleting all those contacts will take a long time and may not be worth it at this point.) The disadvantage is if you need to hide some parts to see others. Naturally, you cannot hide a piece if it is merged with other pieces.

John

[dnunezHSLXH]

Hey @John_Holtz,

Thanks for the quick reply!

1. The numbers on the bottom right for me are 117 and 75.
2. There are a large number of contacts.  I'll try this out and report back.
3. Something tells me that this is the killer.  I don't know why I didn't think of that before.  I've got 1 normal modes, 9 direct transient response, 1 direct frequency response, and 1 vibration fatigue.  I'll see if I can group the transient response studies and report back.
4. I am working of my C:/ drive.  I can't imagine how much slower this would be off my network...
5. I'm working in Nastran 2024.