Hi,
I encountered a problem during the data extraction process:
If I rerun a model without any changes, sometimes I will encounter a convergence issue at a phase where no convergence issue occurs the first time I run the model. It can be resolved by turning off arc-length control (which is not abnormal for my model, I had to do it at several phases), but the final results will inevitably be slightly different. Despite the difference is small, it does make one feel a bit uncertain about the results. Is it a known issue and can someone please provide some explanations?
Many thanks,
Dear Teng,
If you encounter convergence issue sometimes I would advise you to check what causes them. You can check our Calculation Kernel Convergence log article that explains how to read our log file and try and understand what causes the problem.
In general "small difference" is a very subjective term. I would care on judging the difference on engineering levels and not on purely mathematic differences.
There are no known issues on convergence problems with a calculation. That can occur as we perform numerical analyses. Each time you get an error there are possible explanations, which are described on our Reference manual Appendix.
Should you need further assistance, please submit a service request. Then, one of our support engineers can help you in detail: https://apps.bentley.com/srmanager/ProductSupport