I couple of general questions regarding linear elasticity and convergence.
1) I though that the linear-elasticity was a liner-problem - Why does the solver do both a linear then non-linear solver? (this is probably my lack of understanding)
2) How does one interpret the convergence numbers (NRM, RELC) and the output from the linear solver. The way I look at it is if for example you have a 100N force on the end of a bar and it only converges to 1e-2 then the resultant deformation is correct within one sigma for a force 100+-1N?
Linear Elasticity, non-linear and convergence tollerance
-
- Posts: 4
- Joined: 23 Oct 2012, 21:41
- Antispam: Yes
-
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Linear Elasticity, non-linear and convergence tollerance
Hi
1) Generally many equation may be potentially non-linear as they could depend on some nonlinear material laws. So ElmerGUI will set the nonlinear keywords even though the equation would be linear only. You can of course set the number of nonlinear iterations to one then.
2) The accuracy relates only to the accuracy of the linear system Ax=b, not to the discretization error.
-Peter
1) Generally many equation may be potentially non-linear as they could depend on some nonlinear material laws. So ElmerGUI will set the nonlinear keywords even though the equation would be linear only. You can of course set the number of nonlinear iterations to one then.
2) The accuracy relates only to the accuracy of the linear system Ax=b, not to the discretization error.
-Peter