still working with my soil deposit total stress analysis. The model is stable in the plastic phase.
I am trying to evaluate bearing capacity of a strip footing for the moment (I hope to move on to the dynamic analysis later).
To evaluate the bearing capacity, I use displacement increment at all my footing's node (which are connected to the soil deposit).
But I do have a question about efficiency of the calculation, that is, is there any thumb rules to know the meshing refinement I need (size of the element) versus the incremental displacement (dU) I do apply to the model?
Right now, I want to get to 20 mm displacement under the footing and evaluate the stress in the element (I hope 20 mmm is enough to get to the generalised failure). So my maxU (maximum displacement) is set to 0.02 m and my dU I defined as a percentage of the maximum displacement. To have convergence, I currently have to use dU = [expr 0.0000025*$maxU] (that's very small, and it's taking the whole night to complete (It need 4-5 iteration to attain convergence criteria which is displacement incr 1.e-6). The meshing are 0.05 m in both X and Y direction.
First question : Does this level of strain increment make any sense (is it too low, or is it okay?). I don't have experience with FE so I cannot validate the value expect that it seems to be working...
Second question : I thought that, if I refine my mesh to 0.025 I would have more precision. So then I simply increased the mesh number in my loop to match this dimension and I didn't touch the dU. Well, that didn't work! It was worse, convergence was now very long to attain...and at very small displacement, it collapsed. So if I increase the mesh refinement, I also have to reduce the dU ??
Last question.. Sometime, when I realise the program will take forever, I want to stop it.. but after a night of doing calculation, I would still like to the recorded value.. how can you stop it without crashing the program (like crlt + c .. which invariably make it crash and my loose all data) ? (ctrl+s does simply stop the execution but am still not able to really stop it and close the recorder).
And also, when the program is done a large model... I type in "wipe" to close the recorder and obtain data... The damn thing crashed and I lost all !! Is there a way to prevent that or do I have to live with it?
Thanks for your time and sorry if the questions are a little stupid.. hopefully not too much
