Problem with 10 node tetrahedral on RS3.
Results from using 4 node tetrahedral are similar to 6 node triangle on RS2 (and to analitical calculations) , but results with 10 node tetrahedral are very different, and not logical. Time calculation for 3d model of tunnel with 4 node tetrahedral is 20 minutes, and for 10 node tetrahedral is 16 hours.
What could be the problem with my model?
The difference between the computing time is normal because in 10 node you have 6 more points in between two vertex to calculate forces and when you put that difference in a whole model thats a lot more calculations. The 4 node tetrahedral is a very light model when you compare with 10 node tetrahedral.
Erion just a question to try to figure out what is happening in your 10 node interpret result
Is there any message that the model didn't converge?
If that's so you could increase the number of iterations in the project settings, obviously it will increase your computing time but maybe you could reach to a reasonable and similar result as the 4 node mesh model.
Dear Bernardo.
Thank you for your response.
Model converges, there is no warning. Just results are not as expected. Deformation 7 to 8 times larger. Plastic zone larger.
Regards.
I have used a coarser mesh with tetrahedral 10 (tet10), and results are good. There were some bad elements in the first model (6 bad elements in 196186 total elements), and i think they were the problem.
Thank you, regards.
Thanks for submitting your solution Erion. I got curious about what could have been the problem in the modelling and was hoping for you or any other researcher to find the solution.
Regards
I have to say that is best to use tet10 elements, and try to play with mesh Gradiation Parameters and Mesh Customization. And always check mesh quality before analysing the model.
Thank you for your feedback Bernardo.
Regards.
How does the mesh quality and its elements affect the results generated within your model?
Hello Chikondi,
"Bad elements" are the ones which have high aspect ratio, like a "thin element". These elements can lead to inaccurate results, and the analysis may not converge.
In addition, you should use a finer mesh near zones when there is a rapid variation of strains and/or stresses, such as, the application point of the load, near corners, near cracks.
Regards
Makes sense now! Im having issues doing that at the moment. Do you mind explaining how I would do that using RS3? I tried following the tut. But somehow its not making my mesh fine.
Use these instructions:
https://www.rocscience.com/help/RS3/webhelp2/Mesh/Mesh_Tools/Mesh_Quality.htm
Hi
I do not understand how i should change the data in the metric quality to improve my mesh. Please explain?
what would happen if i change either values in the tables
- Badges
- Science topic
- Similar topics
- Filing