radial stress vs thickness . help please

[mvp23]

Could anyone shed some light on the correct method to obtain the radial stress values once the coordinate transformation has been performed from rectangular to cylindrical coordinate system.

I have got 2 cylinders in contact and I'm interested in obtaining the radial stress as a function of thickness of the cylinders.

Currently , to help in seeding the edges I partition the 'cells' of cylinder horizontally (along the x-z plane). When this is done I specify the edge seed (for number of thickness elements ) by selecting the edges created on the thickness due to partitioning , followed by a global seed.
Though this method helps to add in a few thickness elements , I am convinced that it isn't the right way to do it. The reason being - the radial stress values I obtain by this method aren't accurate and then when I try to obtain CPRESS it says 'No Value ' , which (I learnt from this forum ) means the nodes may be off the surface. However when I do not select the partitioned edge to put in a few thickness elements I am able to obtain CPRESS values.

Hence I am forced to consider specifying a global seed and mesh this super fine so that I am able to obtain 3-4 thickness elements on each cylinder. The downside to this is that I'm looking at an overall mesh of about 12 hundred thousand nodes!!!

If anyone could help me with this please ?

 

[corus]

If the geometry is symmetric then use 2D axisymmetric elements instead of full blown XYZ geometry. If not, then try and use symmetry where it exists to reduce the model size. Including a picture of the model might help to decide which is best.

There's no reason why partitioning should affext whether or not you get CPRESS values, providing you have selected the contact surfaces correctly. If you have selected the surfaces before you made the partition then you'll find that the surfaces you expected to be in contact have moved. Whenever you make partitions after you've defined surface interactions, or loads/restraints, it's always best to go back and check that those surfaces and loadings are where you expect them to be.



Tara

http://tinyurl.com/4ydjg7m

 

[mvp23]

I use the same contact definitions when I run the model with & without thickness elements. Yet I’m able to see CPRESS values when I run the model without thickness elements. Now this led led me into believing that meshing the thickness by selecting the edge induced by partitioning was causing problems.

Removing the thickness elements – I’m able to get CPRESS values , but S11 values along the thickness can’t be obtained as I do not have any nodes along the thickness.

Adding the thickness elements – Unable to obtain CPRESS values , but I’m able to obtain S11 values .

I could send you my cae file , if that would work better ? Also , I have a picture attached .
Thanks

 

[mvp23]

And yes , I created the parts , partitioned them , assigned sections , created surfaces and finally did sectional assignments (in that order )

I'm not sure if I can use 2d axisymmetric elements as I am running a fully coupled thermal stress analysis. I use these coupled temp disp elements of type C3D8T.
The contact is based on a finite sliding , surface-surface discretization method .

 

[DanStro]

You can do a fully-coupled thermal analysis with 2d axisymmetric elements.

Sorry I can't help you your real problem though.