Cantilever beam deflection- best parameter

[adindar]

Dear all,
I'm very much pleased running examples in quoFEM. As come to create my own problem I'm facing a simple but hard issue 

I'm trying to obtain the best value in a simple cantilever tip deflection problem using 1) direct formula 2) openseespy. I include the files for first approach to this message and a pdf screenshots file. 

I'm facing the same error message  " No dakotaTab.out file - dakota failed .. possibly no QoI " in the UI. It seems that I'm missing very fundamental but tricky point in the json file. I hope you can let me find where I'm failing.

Best regards

++Ahmet

[Sang-ri]

Hi Ahmet,

Thank you for posting the question and sharing all the necessary files! I was able to test the example on my laptops, but they all ran successfully.

In the screenshot, in the folder "tmp.SimCenter", I see that the size of "quoFEMScalarCalibrationData.cal" is zero, while it should have contained -0.04. That means quoFEM is failing to copy "calibration.csv" to this folder for some reason.

I additionally tested several possible scenarios, and the same kind of error was observed only when "calibration.csv" was empty, which I assume is not the case for your problem.

To help us identify in which step the workflow has failed, can you please share a screenshot showing the contents of the folder "templatedir" inside tmp.SimCenter?

Best,
Sang-ri

[adindar]

Dear Sang-ri,
I appreciate your reply. I understand that you were able to run the files successfully without any improvement or modification. This is interesting to me. I hope I can figure out the problem causing that.

I'm attaching the screenshot of the templatedir to the message. An interesting point is that tmpSimCenter.params file is zero byte in size which I presume that it should not be empty.

I've been mingling with this problem, perhaps, I've already become selective in perception.

Best regards

++Ahmet

[Sang-ri]

Hi Ahmet,

Indeed it is strange that the tmpSimCenter.params file is also empty.

One possibility is that it could be related to the onedrive sync. We were able to reproduce the error (i.e. the two empty files) with an "online-only" sync option and offline environment, which I again presume is not your case . But it might be a signal that the issue is related to the fact the input files are located in the onedrive folder.

Can you please try locating the files in a different directory and let us know the results?

Best,
Sang-ri

[adindar]

Dear Sang-ri,
I've tested two case where I moved from OneDrive folder to Desktop of the computer.

1. Cantilever beam deflection.
It still fails but error message is "21:14:50 - Error Running Dakota: Could not spawn; error code 0 (No error)" instead of "no dakota...". It seems quoFEM would solve if I had constructed problem better.

Details can be found in the attached file named as "2022_0713-AAD-quoFEM-MessageDocumentation.pdf"

2. Material test- best fit.

I was facing the same "no dakota..." error when I tried another problem in which a test data was used in the determination of yield force. It was again in one of the OneDrive folder. Having moved from OneDrive to Desktop, contrary to the above case, I ran quoFEM successfully.

Details can be found in the attached file named as "2022_0713-AAD-quoFEM-Success_Documentation.pdf". I'm also enclosing all files used in the second case.

I've one question and one conclusion in mind and a recommendation about the UI.

Question: Why I failed "Cantilever beam deflection" even if I moved all the files to Desktop but Sang-ri succeeded?

Conclusion: Considering the Material test- best fit case (second case above), OneDrive is not the best folder to keep the input files (main, params, input.json).

I'll be pleased if you can lead me answering my question and comment on my conclusion.

Recommendation: I believe there should be a reset button in the UI. If one tries to run different cases, text boxes (input script, postprocess script, RVs , EDPs) are messing up with irrelevant input text. I've found that exit and start quoFEM after every run is a practical solution

Best regards.

++Ahmet

[Sang-ri]

Hi Ahmet,

Great to know that one of the examples ran successfully and that, supposedly cloud folder was causing the error. I believe your conclusion is correct. I appreciate your investigation.

For the cantilever model, it looks like having "constant" variables in the RV tab is causing the error. We will update the source code to fix this bug. For now, as a workaround, please try removing the two constant variables from the RV panel such that only H is shown (params.py and any other scripts or inputs do not need to be modified, and the two variables will automatically be considered constant). I believe this was exactly what was shown in the screenshot of your first post.

Thank you also for your suggestion!! We actually implemented this feature very recently; in the menubar, please try out the edit->clear inputs button. It should work for FEM/RV/QoI tabs, but maybe not for the UQ/RES tabs in the current version.

We are planning for another release later this week (v3.1.0). We will include the fix of the above bug (i.e. support constant variables option for calibration). From your previous posts, we have also noticed that saving/loading JSON input was not smooth as it should be, so we improved this as well. We appreciate your contribution!

Best,
Sang-ri

[adindar]

Dear Sang-ri,
I did not notice clear commands in the edit menu. It worked for all except UQ. I think I will stick "exit and start" approach.

Regarding the cantilever beam problem, I kept only one RV with limit and initial values. And it worked (details are in the attached file : 2022_0714-AAD-quoFEM-Cantilever_Beam_Success_Documentation.pdf)

I appreciate your recommendations in solving my problem. Writing a small report about the experience,  I intend to add the report and the input files of the two cases to the post titled "Share quoFEM examples here (user-to-user file sharing)" for documentation.

Your efforts in the development of quoFEM is highly appreciated. It seems I'll provide many cases and feedbacks to quoFEM forum in the following weeks. I see quoFEM as an incredible tool in my current studies.

Best regards from Istanbul.

++Ahmet

[Sang-ri]

Hi Ahmet,

Glad the solution worked! Thank you for sharing the details. We greatly appreciate your warm interest and support. Your examples and feedback would certainly be of tremendous help to us and future users of quoFEM.

Best,
Sang-ri