Big static job: how to reduce memory swapping ?

[PamCrash]

Hi

although I have 100 GB of of memory available I am not able to run a linear static analysis (20e+6 DOFs) in a reasonable time. It seems that the 100 GB are not enough, 200 GB might be necessary to avoid memory swapping. Anyone an idea how I can overcome this paging problem without buying further memory ? Maybe there is a better solver available than the default solver ?

Thanks in advance
Pam

 

[johnhors]

I doubt if there are many people around in your league! I believe that Ansys holds the world record for FE model size solved at about 120e+6 DOF's.

 

[corus]

Depends what you mean by a 'reasonale' time. I have had much smaller problems, but because of the transoent nature of the problem, have ran for weeks. I'd think about reducing the size of the model or using sub models for areas that require detailed examination, or just put the kettle on and wait for it to solve.

corus

 

[PamCrash]

Reasonable means approx. two days. The size of the model cannot be reduced further (kind of molecular model) and the problem is not transient; linear static only.

Thanks for your comments. Seems we have to buy more memory.

 

[gurmeet2003]

I have 32GB RAM. I try to keep my model size below 5 million DOF.

Gurmeet

 

[PamCrash]

Good idea

Thanks Gurmeet