Learning Forum

[amnteno1]

Hello,

 

I have doubts regarding the force convergence plot. As far as I understood, we want the purpled line to be under the blue line. The purpled line is the difference of internal and external forces and I want it to be zero, or at least smaller than a reference value (the blue line). So far so right?

Now I did run a simulation and got this force convergence plot:

The purple line is oscillating, almost all the time above the blue line, reaching high values up to 800 N. I was expecting a convergence error, but was surprised, that I didn't get one. The question is, why didn't I get an error? Can I trust the results? Or should I get sceptical about my results, when seeing this plot? And - maybe a bit general - What is the reason for oscillation in the force convergence plot? Are there some "typical mistakes", that lead to such a plot?

 

Thank you,

 

Stephan

[peteroznewman]

The last iteration in a substep is when the purple line goes below the blue (cyan) line. Then the solve increments the load (or time) and does the first iteration in this new substep. It will plot a point that is generally above the blue line. This provides information on how to adjust the solution for the next iteration. This continues, driving the purple line lower until it gets below the blue line again, which means that the system is in equilibrium and this can be saves as a valid solution, for this increment of the load.

From that description, it should be obvious that the purple line is mostly going to be above the blue line and that the solution is always converged.

There are settings that force the solver to take smaller increments than it needs in order to get output at a finer resolution. When you set a minimum number of substeps to be a high number, you are forcing a small increment. That means after the solution has converged at one load increment, the next load increment is not far away and the first iteration may immediately be converged, the purple line remains below the blue line and there is no need for another iteration.

Sometimes the trend toward convergence looks bad and the solution control logic will see that and decide that it would be better to stop iterating at this load increment and try again with a smaller load increment. That is when you see a red line through the plot.

There are no mistakes in this plot, it is a normal, good plot.

[amnteno1]

Thank you, that was very helpful for the understanding what is happening behind the plot and what it actually tells and how it "thinks".