I am trying to analyse a large structure with multiple loading scenarios (independent - different loading directions), while considering geometric non-linearity. Using different steps is not sufficient, since each step starts off with the geometry shaped as it was at the end of the previous step.
I tried creating a dummy step after each loading step just to deactivate all the loads from the previous step, but (I think) the non-linearity causes some distortion to remain even with all the loads deactivated.
One obvious solution is to run each step in a separate analysis, but the model is quite large and the input file processing takes about about as much time as it takes the solver to solve a step, which would immediately double the time required to obtain the entire solution and there would also be redundant information in the result files (mesh data will be repeated in each result file). So creating a separate analysis for each step is something that I am trying to avoid for the moment.