Hi, I am having trouble getting my nonlinear analysis to converge (it is of a tensile membrane structure with pressure uplift). Is there any documentation/suggestions for how to set up these models? Where could I find more info about the load stepping component and the adaptive force control? Also, is there any way to control the size of the model? My formfinding model only has ~2000 gps compared to ~40000 for the nonlinear analysis. Thanks in advance.

the amount of gps is most probably related to the refinement settings. The geometry is refined twice for the second analysis. If you want to use the same parametrization as in the formfinding, just set the parameters of refinement to 0 or 1. If the parametrization is coarser, it will be easier to find convergence as well.

The settings for the nonlinear analysis are dependent on the structure. However, you can control the load application factor by the global and local load curves. You can split the load application into smaller steps. More time steps, less risk of nonconvergence. If you know the critical point, you can apply the load slower in the critical domain than the rest. Therefore, a pseudo time is used. You give time steps and respective load multiplication factors. E.g. T = 0,6,10 and V=0,0.1,1 means that 10% of the load are applied in 1,667% load increments and the remaining 90% in 22,5% increments. The full load is applied in load step 10.The adaptive force control can be used if crictical points (bifurcation) are unknown. If no convergence is reached within the set number of iteration, it tries to divide the loadsteps in half. This bisection is done as often as the given number.

Thank you Anna that is very helpful, I was able to get convergence. Some follow up questions:

- Is there any documentation of the membrane element? Specifically, how does it handle compressive state/is there any sort of wrinkling model? Or is it just a plane stress element?
- What is the difference between setting 0 and 1 for the refinement?
- For the formfinding to analysis pipeline, I currently set up a whole new model by extracting and sorting the deformed geometry. This is a bit tedious and I just wanted to make sure there isn’t an automated way to do this.

Thanks again!

- There is a brief description on the carat wiki page Users:General FEM Analysis/Elements Reference/Membrane NURBS - Carat++ Public Wiki also useful for understanding and manipulating the input text for the solver. Just plane stress, no special wrinkling model. It will be possible to apply prestress on a shell in the upcoming version.
- There is no difference
- You can also use the DE output of the deformed model component. It provides all elements with deformed geometry (without inner stresses) and the chosen refinement state. If you’re not changing the refinement or other element properties, this is the way to go. Supports have to be reset. Note that if weak supports were applied, the supports may move a little bit if the penalty factor was not high enough. So reusing the components of the formfinding analysis is possible but has to be checked. Remark: this is only feasible for formfinding → analysis and not a construction stage analysis since stress states resulting from deformation are lost. They can only be considered with the reference element component.