Back to company website

You are not logged in. Please login or register.

Forum updated

We have recently updated our forums system, and we have cleared all spam topics and users. In case we accidentally deleted your account, please register again. If you miss a topic you have posted before, please let us know.

Post new reply

Post new reply

Compose and post your new reply

You may use: BBCode Smilies

All fields with bold label must be completed before the form is submitted.

Required information for guests


Required information

Topic review (newest first)

5 Post by rjcosta

Dear Mr. Pryl,
Once again, thank you very much.
Regards
Ricardo

4 Post by dpryl

In postpeak, the results are not much precise anyway (think of many factors that influence localization that are not modelled, e.g., aggregate shapes and positions), and a tight error tolerance would not make much sense.

For pre-peak, several steps with large errors (over 10%) in a row clearly mean the results after are more or less useless (even if convergence is reached in some following steps!), and one has to check the boundary conditions, material parameters, and try to change the step length (typically, shorten, but sometimes a longer step with a larger iteration limit helps to overcome a problematic point, too).

On the other hand, a single step with an error of 2-3% definitely does not make the following results invalid.

3 Post by rjcosta

Dear Mr. Pryl,
Thank you very much for your instantaneous help.
One more question: when you say that the “After step† multiplier equal to one is usually too strict you mean that if I don’t get convergence in the end of a step this is irrelevant for the following steps?
Regards,
Ricardo

2 Post by dpryl

Hello Ricardo,
yes, this can be activated in the Conditional Break Criteria tab in Run - Solution Parameters - Edit. You can set the analysis to stop if the convergence criteria at the end of a step (or iteration) exceed some multiples of the allowed convergence errors. Setting the "After step" multiplier to 1 (as you suggest) is usually too strict, especially if you need any results in postpeak. The multiplier for stopping "After iteration" need to be set quite high (e.g., 10 000).
Regards, Dobromil

1 Post by rjcosta

Hi!
My name is Ricardo and I am a beginner in ATENA so I have a beginner question.
I am using ATENA 3D Eng.
I was following ATENA Tutorial 3D and in the post-processing phase I notice that the error stream (analysis progress information option) in analysis step 24 didn’t get convergence in 40 iterations. However ATENA didn’t stop the analysis. This is a nonlinear analysis so I thought it make no sense to continue to step 25.
If what I am thinking is right (please correct me if I am wrong), it would be useful to have an option that stops analysis when convergence criteria aren’t met in maximum number of iterations. This way it wouldn’t be necessary to check all the analysis step error messages and ATENA wouldn’t have to do unnecessary steps. Does ATENA have that option?
Many thanks,
Ricardo