Different values of reactions ver8.4 and ver8.6

[user2705]

Dear,

I am switching to a version 8.6 and I tried running the same example I am currently working on in the version 8.4. However, I don't get the same values of nodal reactions.

Attached, I am sending you the output files for both versions and also the input file of the problem.

Update: The output files cannot be attached due to its size, so I sent it through the official e-mail address.

Best regards

[Prof. S. Govindjee]

 I see a few difference in the outputs that you sent, but they may be just reporting issues.  (1) the minimum element size in the two meshes are different (search h-minimum).  This probably can be ignored since the nodes and elements look to be the same.

(2) After you fix the mesh with CHECk, the output of OPTI the Number of tangent terms is different and (3) the Average column height is also different.   I wonder if the optimizer is doing the same thing or again this is a reporting issue.

(3) In version 8.4 on the very first convergence test your current energy norm is reported as a negative number.  This may be an issue with 8.4.  8.6 reports it as a positive number which makes more sense.  I see that your output DISPlacements look to be the same (on load steps 1 & 2)

(4) What worries me more is that your reactions from REAC,ALL have reversed signs.   See e.g. dof 1 on node 95.

Can you look at your problem and determine which version of the code is producing the correct reactions at the nodes?
I suspect there is a difference in the way that CHECk functions between the two versions.

[Prof. R.L. Taylor]

One check would be to see what the output file says after the CHECk command.  It should report any element that is misnumbered. 

[Prof. S. Govindjee]

Every element in the mesh is misnumbered.

[user2705]

Dear professors,

Thank you for the prompt reply.

I have tried a similar example in which I numbered elements correctly, hence check function is not activated, and I get the same results with both versions.

In this example where check is needed, I simplified the example where now I impose vertical displacement equal to 1 on top boundary with bottom boundary restrained. With 8.6 I get total reaction at the bottom -5.16E+06, and with 8.4 I get 5.16E+06. It seems that version 8.6 is producing correct results. Also, all nodal reactions have reversed signs.

Attached is the file with the results.

If it is due to the check function, for further work, I will pay attention to the node numbering in my files for mesh generation.

Best regards and thank you very much