How to set up properly Restart From Another Analysis

[EngineerMickeyMouse]

Hi All, this question relates to Time-Step tab found under Advanced options of Analysis Parameters.

Analysis type: Nonlinear RIKS post buckling.

Purpose: Analysis A and B results to be joined together via Restart From Another Analysis to enable for load factor increment modification.

Problem description: Analysis A is set with following Time-Step Details: output interval 1, Interval to save restart data is last step only, then Analysis A is solved with properly converged solution. Analysis A is copied and renamed to Analysis B with smaller initial load factor increment set, Time-Step Details: Restart from another analysis, selecting ds.asd from Analysis A and restart from step 0 (last converged). After running solver, the error in the log pops up and solving is cancelled. The error at log says Can not find restart file and gives path line of Analysis B/ds.rnv.

Query: Please let me know how to solve this issue.

[John_Holtz]

Hi E.M.M.

I tried to reproduce your problem with a couple of models, but it was working okay for me.

Since Analysis B says that it is searching for the restart files in Analysis B (instead of Analysis A), my only thought is to double check the Analysis Parameters in the model for Analysis B. Check the text in the box for the "Name of model to restart from" to confirm that model B didn't get selected by mistake. You can click in the text box and use the arrow keys to scroll to the end of the text. (You have probably already confirmed it, in which case I do not have any theories as to what is going on.)

However, even if the restart would work, it may not work like you are expecting. One thing to keep in mind is that the "Initial load factor increment" is used only for the first time step. After the first step, the load increment is controlled automatically by the solver. (In the documentation, it says that the displacements are incremented on subsequent steps, with the amount of load required to produce that is calculated. That may be an over simplification of what the solver is really doing. I do not know the actual math that the solver is doing.)

So the question is this: if you do a restart, what load increment is the solver using? I could not deduce what the solver is doing in a restart from my test, but I can see that it is not using the "initial load factor increment" that I specified in the "analysis B" model. My guess is that the restart analysis is automatically updating the load increment.

[EngineerMickeyMouse]

Thanks for your input.

To solve the issue, could you please advise me any method to perform RIKS analysis in two stages, first stage for larger time step (or load increment) and second stage for a few more condensed time steps to catch properly stability reduction?