Convergence Tolerance

[aphilic]

Hi everyone,

I want to ask a question about convergence tolerance. As much as I know, the program determines the convergence when force approaches to the predicted value as close as %0.5 and displacement %5. I also heard something but im not sure yet, which is that, Ansys determines a factor, whose value is different in every simulation, in front of the previous values. When i make a simulation, how can i be sure of the exact value of the convergence tolerance?

My real problem is that, I want to reduce the tolerance of convergence to %0.05 for the displacement values. How can i set it?

One more question, do you have any idea why i cannot solve over 1000 substeps in total when I try to solve 10 loadsteps, which have 200 substeps individually, in a row even though i enter this command when the program is on:

/CONFIG,NRES,10000

Thank you in advance.

Can

 

[doatis]

Long story short...the convergence tolerance that is calculated per substep is complicated. There are two methodologies that are used, but both operate on the residuals between the applied and internal load vectors. These two are L1, which takes the norm of the vector residual, or L2, wich takes the SRSS of the vector residual (default). Depending if you have solution control on or off, convergence will be either .5% or .1% of your {F}.

Look up the help for cnvtol. It's been a while since I messed around with those settings, so maybe someone else could comment on it further.

I don't know why you'd have any problem with the number of substeps, assuming your nsubst command is entered correctly.

Hope this helps,
Doug