Search
Close this search box.
Notifications
Clear all

High aspect ratio elements with Hybrid Model UMAT

2 Posts
2 Users
0 Reactions
579 Views
Posts: 9
Topic starter
(@hohaigang)
Active Member
Joined: 15 years ago

Hello Jorgen and all,

I ran a simulation fine in plane strain using the hybrid model, and then transitioned to a 3-D model using the same mesh. I am modeling deformation at a notch tip under creep conditions, so the elements get very fine there. However, the overall specimen is fairly thick, so these fine elements have a very high aspect ratio (~2/900), since I only used 3 elements through the thickness. This puts the number of elements at around 3000, and reducing the mesh to get the elements to an acceptable aspect ratio will likely increase the number of elements to an unsatisfactory level.

The code does not converge in 3-D, but results in an error that states that the FORCE is zero everywhere in the model. This looks like a jacobian error to me, and I recall that the Hybrid Model uses a proprietary jacobian solver. The 3-D model solves linear elasticity just fine.

Is it the jacobian computation that is the problem?

How high of an aspect ratio is acceptable for 3-D elements?

Should I turn down the force iteration tolerance and let it ride, or will this just result in nonsense?

Will running in Explicit mitigate these issues?

-Jevan

1 Reply
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

If you post or email me the inp-file then I will take a look...

-Jorgen

1 Reply
Share: