How does the TIE interpolation works?

[Nicolaf]

Hello:

I´m simulating the welding of a plate using a sequentially coupled aproach with DS4 and S8RI5 shell elements. In order to acuratelly represent the stress and temperature field through the thickness within the weld line and its sorroundings i need to use more than 9 section points, but beyond that area 5 section points would be enough. I´ve tried to use a TIE MPC to fulfill this need but the solution yields an unrealistic stress field in the tie´s sorroundings and discontinuities in the temperature field . Does anyone know how to solve tihs problem? or how does the abaqus interpolates the displacements and temperature fields from the master surface to the slave nodes through the thickness?

Thanks, Nicolaf

 

[JayTung]

I think "Tie function" is to share the same displacement amongst points. The unrealistic stress field may be attributed to the mesh size and your boundary conditions.

 

[corus]

You should really only use the tie function away from areas of interest as these 'discontinuities' often occur, particularly when you have dissimilar meshes. The tie function will use a linear interpolation between nodes in order to equate displacements, but often linear methods are inaccurate. Try and use similar mesh densities between the two surfaces to reduce the discontinuity, or use a continuous mesh across the surface boundaries.

 

[Nicolaf]

Thanks for the insigths they were very usefull. For the records, i´ll try it using same elements type and size across the tie (my real need is to reduce the number of section points).
The other problem is that i´m using shell elements in which each section point is associated with one DOF (at least this is the case in the thermal analisys). I`ve been running some tests and came this conclusion: The tie only works for one DoF, for example: lets assume im tying one node from parte 1 to one node from part 2 and that both parts have 5 section points. I can tie toghether NT11 from part 1 to NT11 from part 2 or NT11 from part 1 with NT15 from part 2 (in general, Top with top, top with bottom or bottom with bottom). BUT, i cannot tie NT11 (part 1) with NT11 (part 2); NT12(part 1) with NT12(part 2);...;NT15(part 1) with NT15(part2) all at once. I guess the tie doesn´t work that way, so, the question: how can i to that?? there is any MPC or interaction that work that way??

 

[corus]

Perhaps there's no need to tie the internal points as the temperatures through the section points will be calculated knowing the surface values?

Personally I'd change the modelling method you're using and opt for a 2 or 3D solid element representation if you want to gain any confidence in the results.