Please I am curious how to select the best convergence for my model. Currently, when I use model solve convergence 1 which was stated as a more rigorous way in the FLAC3D 9.0 tutorial by ITASCA and it does not solve, then I use model solve cyc (a certain number). I use trial and error and observe when the difference in the mechanical ratio is infinitesimal. However, after converging to a certain level it could jump up a little and start converging again. I usually select the cycle number before the jumps occur. See the snippet for reference.
Any comment on the jump and if it is right to select the cycle number before the small jumps even though it will still converge more later?
There are many possible reasons why a model may not “solve”, many of which I cannot exclude based on the limited information you have provided.
“Solve” in the context of the itasca command means reaching a somewhat stable state. Given your image, a possible (and maybe even probable) reason might be, that the system simply is not stable, e.g. some part is constantly failing.
Usually, modeling starts at some “stable” and properly initialized (stresses and boundary conditions) setup, which quickly reaches equilibrium and is then the starting point for the actual excavations etc. to be investigated. Therefore it is very important to make sure to start from an initial equilibrium and not just say “Eh, 1000 steps should be fine. Let’s move on.”
Again, I can’t say much more without more information. If it is sensitive, you should better turn directly to the Itasca support.
