Learning Forum

[Tim Dietl]

Hello everyone, I am not sure why I get the warning "out of core memory mode" with only 300 000 nodes. What could be the reason for that? Could the modeling be the problem?

Thanks in advance!

[Nanda]

Hello Tim,

Solving a small model can be done in-core, which means all in RAM. When the solver can't fit a large model entirely in RAM, it uses an out-of-core method that uses the HDD or SSD storage to hold parts of the solution while it is working. For 300,000 nodes, if the model has only solid elements, we have 0.9 million degree of freedom. So the solver needs around 13.5 GB RAM memory. Check for available memory in your system and see if it's failing this requirement. For speeding up your solutions, have a look at this article.

Hardware Tips to Accelerate Simulation | Ansys

Regards,

Nanda.

Guidelines for Posting on Ansys Learning Forum

How to access ANSYS help links

 

[Tim Dietl]

Hello Nanda, thank you for your answer. I have 30 GB of RAM so it should be possible to calculate it in core. Could the modeling be the problem?

[Nanda]

Hello Tim,

Looks a simple model to me, nothing complicated from observing it, looks like you have a bolt involved, is it pretensioned? If you wish to run your solution in core only. There's a way of forcing the solver to do this, you must use this following command. 

DSPOPTION,,INCORE

All the best,

Nanda.

[Tim Dietl]

Hello Nanda, yes it is not a complicated model. Yes, there is a bolt pretension of 95408N. Okay, thank you but I am still wondering why the model takes such a long time to be solved. Can symmetry parts have that problem in general?

[Nanda]

Hello Tim,

Ideally a symmetry model should take less time than full model, hope that APDL command was of help.

Regards,

Nanda.

[Tim Dietl]

Hello Nanda, yes the command helps but I am still wondering why 300 000 nodes are too much for 30GB RAM. It would be interesting to find some possible conclusion for this.

[Claudio Pedrazzi]

Does the model now compute incore, if you issue DSPOPTION,,INCORE  ?

Does the warning message "out of core memory mode" disappear?

Is the model linear elastic, small deformations?

To check, try to build another model (completely stand alone, do not copy from this one).

Make a simple block, held at the base.  Refine the mesh in order to have approximately the same number of nodes (better: the same number of DOF).  Apply one force on the side opposite the base.  Do you still get the message "out of core"?

This will eliminate the doubt that something is "special" in your model.  Neither symmetry, nor pretension should be the reason.

 

Hope this helps, Regards

Claudio

[Tim Dietl]

Hello Claudio,

so I tried to solve the model again with the command but I still get the out of core warning.

Yes, the model is linear elastic and only has small deformations.

I built the block with approximately the same amount of nodes and it is solved in core easily.

I noticed something else. Shouldn't have the bolt pretension and the working load (in the results of the bolt pretension) the same value if there's no working load?

Because this is not the case with this model if I use a finder mesh.

Thank you for your answer!

Tim