Convergece issues in Mesh Analysis

[DenisRhode3948]

I meet  a persistent convergence issue when tryng to launch analyses on a structure.

The context is the following : I am using a 30-days demonstration version of Autodesk Simulation Mechanical to verify whether it is possible to use it instead of the previously available ALGOR 2011.

Importing geometries from STEP files and adjusting them thanks to Inventor Fusion are unproblematic. Meshing the structure, whether using 3D bricks or plates is also carried out without any problems.

But when the mesh analysis routine is launched, this routine seems to run in an endless way, and on the small window that appears then, the progress  bar remains blocked at about 15%, while the "Stop" and "Terminate" buttons are not active (see attached screen copy). As a consequence, the only way to stop the routine is to use the Task  Manager. Moreover, this does not shut down files that were in use, and to start a clean new session, it is in fact necessary to reboot the computer.

This issue happened first on a very simple example ( a parallepipedic box whitout top). It has been solved, for this simple case, by installing SP2 : so I expected that  this would solve also the issue on reasonably more complex structures. Unfortunately, not !

The first attempt on a real study case, a parallepipedic structure with several holes, led again to the same situation. I expected it was a convergence issue, which I tried to  fix by rounding the angles of the holes ; no effect. I also suspected VCAD might be at stake, so I performed a new meshing using the legacy mesh engine : same result.

To complete the scope, I would like to add that the scenario based upon the legacy meshing tool couls be successfully exported to ALGOR 2011, which provided the expected results, whil expporting the VCAD-bas model to ALGOR produced the same undesirable effect as in Simulation 2013.

Has anybody an idea about the possible origin of this strage phenomenon ?

[DenisRhode3948]

In addition, I could see that the question of hyperthreading might be involved in this issue, in other terms, that what I interprete as a convergence issue coming from the software might actually be in fact just a very slow progress due to a conflict between threading strategies of Simulation Mechanical and of the computer itself.

However, I'd like to get some advice before undertaking BIOS changes that might affect the performance of other applicatons I let run on my brand new 64bits PC with Windows 7. So please... HELP !!!

[AstroJohnPE]

Hi Denis,

If I can read the language correctly, your image says the software is "Verifying the Model". This step of the process has nothing to do with the meshing. I cannot imagine how the meshing would cause it to hang.

To confirm this is true, go to "Mesh > Mesh > 3D Mesh Settings". Somewhere under "Options", there is an option to "Solid Mesh the model at the time of the analysis" or some similar terminology. Uncheck the box to indicate that you want to solid mesh when you generate the surface mesh. OK out of the options and create a new mesh. Does it complete?

Make an archive of the model in its current state: "File > Archive > Create".

Now go to "Analysis > Analysis > Check Model". Does it hang? Are you sure there is no pop-up dialog indicating there was a problem and giving you the option to view the warnings?

If all else fails, attach the archive file (.ach renamed to .zip) to your reply.

[DenisRhode3948 · Accepted solution]

Thanks a lot for the advice, which is no doubt the appropriate solution in many situations, but I checked inbetween the impact of inactivating the Hyperthread function of my Intel processor, and this fixes the issue. : mesh analysis, previously locked at about 15% progress, now runs through in less than 5s.

As a comment, I just find that it is a pity that, although Hyperthread proved incompatible on my computer with Autodesk Mechanical's mesh control protocol, and although this issue seems to occur frequently, this fact (and the solution) is not highlighted in the installing protocol and documentation. It would be worth writing this in large bold red characters to prevent nervous breakdowns among new users.

[AstroJohnPE]

Since you are familiar with the problem and solution, I suggest that YOU revise the wiki help to put your findings on the pages you think would be most appropriate. (I believe this is one reason they moved the Help to a wiki.)