Dear Wolfgang thank you for your quick reply. All your suggestions are very useful. I have already done some debugging you proposed, but I will go back on my steps, and try again to do it more carefully.
Best Giovanni On Tuesday, April 5, 2022 at 4:49:30 PM UTC+2 Wolfgang Bangerth wrote: > On 4/5/22 07:21, [email protected] wrote: > > > > I tried with different mesh refinements and different time-step sizes, > but the > > analyses provide unsatisfactory results. By summarizing, the size of > active > > set reduces to zero after the first time step, and then increases. The > > convergence of the Newton solver slows down since the initial loops, > leading > > to loose convergence (for max loop reached). The evaluation of the > resulting > > contact force fails giving negative values. > > > > At the moment, I'm really stuck. Is there anybody who can give me some > hints? > > Giovanni, > short of finding someone who is (i) an expert in time dependent > plasticity, > (ii) has plenty of spare time to look through your program, you are left > with > debugging the problem yourself. Here is how I would approach things: > > * Simplify. Instead of looking at a whole load history, just consider two > steps. > * Pick a load history so that what step-42 produces corresponds to your > first > step. Make sure you get the same result. If it doesn't, you've just > figured > out that already your first load step has a problem. > * Then double the load in the second step. You know that that should > result in > a larger indentation, and a larger active set. > * If it doesn't, investigate why that is. One approach would be to > simplify > the set up so that instead of indenting a complex shape (or just a > sphere), > you impose a constant load across the top surface of the domain. You might > even be able to compute the solution for this problem by hand because of > the > symmetries of the deformation you expect. > > > It is not uncommon for a complex program to be wrong in the beginning -- > in > fact, I'd say that's how nearly every program starts out. The question is > whether you can build the mental tools to break things down into smaller > pieces that can be debugged more easily. Simplifying the situation to > something for which it is easier to reason about the behavior is a key > first > step in this process. > > Best > W. > > -- > ------------------------------------------------------------------------ > Wolfgang Bangerth email: [email protected] > www: http://www.math.colostate.edu/~bangerth/ > > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/5d3f2d6a-0c22-48b7-af1f-2bf9cc0f4998n%40googlegroups.com.
