SOL600 Results

[Alakin]

Hi I've made a Sol 600 analysis just to check if I could run it with no problem. It's a cantilever beam with linear material and BC but geometric nonlinearity. The problem comes when I import the results in Patran. The displacements values are exactly the same between the XDB file created by NASTRAN and the t16 file created by Marc, but the stress tensor of the xdb differs form the "stress, Cauchy" of the t16. The stresses of the xdb are about 10% lower than the stresses of the t16. Why do they differ so much if the xdb is just the translation of the t16? and which one should i trust?

Secondly I don't understand why, even though I created only one subcase in the bdf file I get 8 subcases in the results files with a time variable that goes form o to 1. I've always used the sol 106 to run non linear analysis and things are pretty different. What is a meaning of a time variable in a static analysis and why do I get all the intermediate steps whose results just occupy memory and are of no interest to me?

 

[franc11HD]

In the load cases parameter, you can change the incremental time step of marc. For example, if you ask for increments of 10%, Marc should make 10 times steps from 0 to 1 with each time 10% of the load. This is the way of MARC to divide the problem. When there is non-linear behavior, after each time step, MARC acutally recalculates the stiffness matrix that might have change due to those non-linearities. For linear static problems, the stiffness matrix do not change and therefore only is calculated once so only one time step is needed. Those intermediay time step could be usefull. If the problem is long to solve you can always import the T16 file and it will show all the increment that Marc has solved. This way, you can look at your structure and see if it behaves as you think it should and correct immediately if neededd instead of waiting the end of the solution.

You need to be carefull at some point also with all the time step results. If you only select the last time step (giving you the state of the structure after full loading) and click on animate, PAtran will interpolate the structure from 0 for time step 1, ignoring the "real state" of the structure in intermediate increments. As if you select all time steps and click animate, the real state of the structure will be displayed at each time step.

As for Stress tensors and cauchy stress tensors, I think Cauchy stress and XDB stress might be calculated differently. A look at the MSC theory manual should give you more insight on this matter. I think Nastran (XDB) do not have cauchy stresses available for results.

 

[Alakin]

fran11HD thanks for answering. So if I've understood correctly:
- the 'time' it shows is simply the percentage of the load, it doesn't have anything to do with the actual time (seconds) (to me this is an incorrect way to show it).
- there is no way to have only the last 'time step' in the result file but also all the intermidiate subcases, (sometimes it could be useful but sometimes it's just a waste of space especially if the model is large, it would be better if I could choose it)
- the stresses in the t16 and the stresses in the xdb are actually two different things so are both 'correct' in their own way

Thanks a lot