"Exceeded CPU time limit" problem

[bridgehl]

Hi everyone,
I am doing a contact transient analysis using Ansys 10. Most of the analyses are converged without any problems. However, there is one that is stopped in the middle without any errors or warnings prompted in *.err, in output file...

There is only a message in nohup.out (It is run in batch mode) file:

/opt/ansys_inc/v100/ansys/bin/ansys100[305]: 1511450 Exceeded CPU time limit

I checked the NCNV command and found that the default value of CPU time limit is infinitive. To be on the safe side, I set a very high value for this and re-run. Unfortunately, the problem is still there!!!

Could anyone help me on this problem. Thank you very much.

Regards,
Bridge

 

[cbrn]

Hi,
there is possibly one case in which to get this message, afaik: if, at a certain point of the equilibrium iterations, the force convergence value stabilizes at a constant level. In this case, for some tens of iterations the solver tries to update the force convergence criterium, but at a certain point also this value stabilizes, so the time needed to get a convergence becomes infinite (I call it a sort of asymptotical mis-convergence). If you set the NITER to, say, 100000, you may run into the asymptotic situation well before the stop due to NITER.
This kind of problem is most probably due to the contacts' setup. They are ill-conditioned, or participate of CPs which are not perfectly well-posed, or participate of CYCLIC symmetry together with CEs with a pilot node and the whole is not well-posed, or...

Regards

 

[bridgehl]

Hi cbrn,

Thank you very much for the valuable help. I set the NITER to a very large value by this command:
NCNV, , , 99999999, 99999999, 99999999
However, the problem is still there.

I will trace the problem according to your suggestion.

Regards,
Bridge