Need help with normal-tangential displacements for elasticity.

General discussion about Elmer
Post Reply
skoushik
Posts: 20
Joined: 22 Jan 2018, 01:55
Antispam: Yes

Need help with normal-tangential displacements for elasticity.

Post by skoushik »

We are using Elmer for nonlinear (presently) 2D static simulations.
For prescribing displacements normal to a surface, we used the keywords and suggested in the Elmer Models manual as :

Code: Select all

!Displacing radially inward for 2% radius:
Boundary Condition 1
  Target Nodes(1) = Integer 9
  Normal-Tangential Displacement = True
  Displacement 1 = -0.05
End
Node no. 9 is located at 60deg from cartesian 1 direction, and the output generated for the above does not correspond to this. The prescribed displacement was rather observed along cartesian 1 direction, showing no effect of keywords.
Plz mention corrections (if any) or suggestions for solution. Thanks in advance.
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Need help with normal-tangential displacements for elasticity.

Post by raback »

Hi

The N-T coordinate system works in the following way

1) Tag the N-T nodes
2) Calculate the normal vector for these nodes and let them determine local coordinate system
3) Assembly the finite element system using local coordinates
4) Solve the linear system and revert to cartesian coordinate system

Thus, the N-T system is only applicable to elements that have a normal vector. It is not implemented for single nodes as they cannot be associated with any normal vector.

If you know the desired displacement in all coordinate directions the N-T system is not really needed. The intended use for it is, for example, when normal component is fixed but the tangential one is free to move.

-Peter
skoushik
Posts: 20
Joined: 22 Jan 2018, 01:55
Antispam: Yes

Re: Need help with normal-tangential displacements for elasticity.

Post by skoushik »

Thanks for that raback. I see that (as in our own problem,) the element edges or surfaces could be discontinuous at nodes, and one normal vector could not be computed. Hope I got it right..?
On the other hand, we are able to apply displacements along cartesian directions in this particular case which we used for verification. But when we are required to apply displacements at arbitrary points normal to the surface, thought this could come handy.

Thanks.
Post Reply