Hi all, I'm an Industrial Design student currently working on some static analyses.
One thing I have to do all the time, is waiting on the results. Meanwhile, I'm checking out facebook, having a break ánd I can't help to take a look at the 'Convergence Plot' also while using the FFEplus solver.
It looks great, but is there any useful information displayed for me? I'm just curious, and actually not a Simulation- or FEA-expert. Thanks!
It may be worth your time to learn and understand what convergence means. It will help you understand the difference between good analysis and poor analysis. Maybe (just maybe) you can rise above the herd of poor analysts.
Hi Roland, I do understand what convergence means in the context of both FEA and CFD. I used to train Simulation courses when I was working for a SolidWorks VAR. I understand when (and why) results diverge, as well as what methods can be used to help result converge.
However, the particular convergence plot that Willem is inquiring about (when using the FFEPlus solver) is actually of no use to users, and that's not coming from me, that's coming directly from the SolidWorks Knowledge Base. See the screenshot below:
Kind regards,
Prasad Bhonsule
It's useful. Helpful, even. How is it not useful to be able to see that your analysis is converging? Beats the heck out of sitting for minutes or hours wondering if your ship will come in.
p.s. Prasad--I have no doubts about your competence. That comment was directed at the OP.
Frankly, Sw Simulation makes it too easy to get well-packaged "results" that are usually poorly understood and just plain wrong. Most analyses seem to be thorough, complete answers to the wrong questions.
While those 'pretty graphs' don't really mean anything to the FEA analyst they can give you some indication as to how the solve is going. If everything is heading downward toward the limit line then I let it run. If, as I have seen some times, it is oscillating up and down then usually the solve seems to fail and it is better (IMHO) to abort the run and find out what was done wrong on the input.
Do NOT confuse this with 'convergence' which is an indicator of whether you have reached a reasonable solution as Prasad has indicated above.
Good point, James. Convergence has two different meanings in different scopes.
One is for a given solution to converge and yield a solution. The other is for strain values to "converge" toward a finite value as the mesh is refined and run repeatedly.
Both are important to know. The second, more so. I've seen more than a few failures due to would-be analysts failing to refine models to get converged results.
I think it should be noted that, strictly speaking, a converged solution does not mean you have an accurate solution. Generally speaking, all accurate solutions are converged, but not all converged solutions are accurate.
Absolutely, the "convergence graph" is merely some indication of the solver/computer progress and has nothing to do with a "converged" FEA solution.
Hi Willem, the convergence plot when you are using the FFEPlus solver is actually of no use to typical users, it plots the convergence of some parameters that are internal to the solver.
Kind regards,
Prasad Bhonsule