I shared the pipe bending simulation I made for testing purposes. Breaks occur in 2D bending, which is normally made without a guide. I closed the second order because it took too much time. It gave more realistic results by reducing the initial time increment and min time increment values. It also breaks when the secondary order is open, but the simulation gives an error in the last steps, unfortunately. What effect does reducing these values have on the analysis?
Note: I will try 3D bending, there will be no breakage in this bending. https://app.box.com/s/q6nu8tinzu7e07lmte94lvzf51iomxbl
Second order of elements is important for accuracy and is usually needed, especially when tetrahedral elements are used (which is always the case with 3D solid meshes generated in PrePoMax). First order tetrahedral elements are rarely used because of their low accuracy - they are too stiff and bad for analyses involving bending.
When it comes to incrementation parameters, they affect the convergence. CalculiX uses them to control the size of time increments necessary in the Newton-Raphson method used by the implicit solver. The initial increment size is just a way of suggesting the solver which increment size it should start with. Reducing it can help achieve initial convergence in analyses involving nonlinearities. The minimum time increment is a value below which the solver cannot go when decreasing the increment size. Sometimes quite low increments are needed and you can set it to e.g. 1E-5 (default) or even lower (1E-6 or 1E-7). Don’t set it too low because if the solver has to use extremely low increments, it likely won’t converge anyway,
Hi @Ali, I did some similar simulations in the past, I suggest you to use hexa elements, they can be easily generated in Salome and then exported as .unv files to import in PrePoMax.
By the way, as is an long calculation, try to reduce the quantity of elements of your model, for example the tooling can be hollow, or better reduce to the minimal faces in contact with the tube. For the tube there is a long part that doesen´t see any deformation (beginin and end), so with hexa meshing you could setup a mesh densification only were the tube bends and use very coarse mesh in the ends.
Right, I forgot about this potential issue with CalculiX. I haven’t tested it yet but I will try to run some tests and I will likely report this on the CalculiX forum, notifying Guido.
Yes, that is what I first tried to do but it does not work. The allowed min. time step is 1E-5 * step time. So changing the step durations does not work. That is why I found that curious.
