Learning Forum

[henriquegpo]

Dear community,

Due to the large amount of data coming from the simulation, I am now running Ansys remotely in batch mode inside a cluster. I can make it work relatively good with the general command from the Mechanical APDL manual:

ansys232 -b -p productvar < inputname > outputname 2>&1

being productvar = mapdl or ansys in my case.

The thing is that I can only run one input file at a time, making the simulation of interconnected systems very tedious and more time consuming that it should be. For example, if I want to run a thermal analysis followed by a structural one, I need first to prepare the input for the thermal, run the batch mode in the cluster, wait for the result file, upload it to my PC and open it in ansys, get the resulting temperature load from it, generate the input file for the structural and only then run it again in the cluster.

My question is, is there a way of running (in batch mode) such an analysis at once, with a single input file and in a way that Ansys will automatically get the thermal results and use it as input for the structural?

I'd appreciate any hints!

Best regards,
Henrique

[danielshaw]

You can only access one input file from the command line, but why do you need two input files?  An input file can have multiple solves.  Why can't you do the thermal analysis followed by the structural analysis within the same input file?  See Example 3.4 in the MAPDL Coupled-Field Analysis Guide.

[henriquegpo]

 

Hello Daniel,

Thanks for your response.

I should have said, but I am an Ansys beginner, specially using MAPDL. I have mainly used it to apply some settings withing the Mechanical interface. I didn’t know about the possibility of using a single input file with multiple solves.

So what I was actually doing now is setting up my model in Mechanical and, from there, generating the input data for the batch job submission. I have now checked the example you mentioned and could not directly get what I should do, but in the meanwhile I came across with the System Coupling system wihtin workbench, and I believe it may a be a good starting point. So far I could not even connect it properly in the Workbench (steady state thermal + static structural), but I am on the process of figuring that out.

Is this indeed a good direction? I have the impression that, with System Coupling, I will be able to connect both my thermal and structural analysis there and create a single input file. Is that correct?

Cheers,
Henrique

 

[danielshaw]

If you need to do a thermal-structural analysis, I recommend linking a Mechanical structural system to a Mechanical thermal system on Workbench project schematic and solving them in Mechanical.  That method is the simplest to setup, but you cannot use it to create one unified MAPDL input file. You can create a MAPDL input file from the thermal system and another input file from the structural system, but you cannot create one MAPDL input file that includes both systems.  You could attempt to use a text editor to merge the 2 single system input files into 1 unified MAPDL input file, but I suspect that some commands will overlap between the files.  You will probably need to know what lines to remove from the structural system input file.

[henriquegpo]

Yes, linking the two systems in Workbench and solving them in Mechanical is indeed the way I am used to work. My issue is that, when working with larger models, the generated files (model, temporary, results) can become quite large and unfortunately my PC sometimes is not able to deal with that. And as my company has a HPC cluster, I would like to use it. It has virtually no storage limitation and much higher computing capabilities.

So just to make sure, the only way to get one input file with multiple solves is via full APDL scripting? There is no way of preparing the models in mechanical using its GUI and simply generate an common input file? I saw this System Coupling system in Workbench. Isn't it possible to use it to get to what I need?

[danielshaw]

The System Coupling systems do direct coupled-field analyses using the coupled-field (22x) elements.  So, yes if you used direct coupling with the 22x elements, you could create 1 input file.  However, there are disadvantages (and some advantages) of using the direct coupled-field elements for thermal-structural analysis.  Most users prefer using the sequential coupled-field approach (see the MAPDL Coupled-Field Analysis Guide) for thermal-structural analysis.  I assumed that you were using the sequential approach.