Variable shell thickness

[ChrisBay]

Hello everyone,
i have two parts (solid and shell(2D planar)). The parts are connected with the shell to solid coupling.
The Solid Part has an ellipse as a cross section. Now i want to continue this cross section for the 2D planar shell-> that means a variable thickness for the shell. Therefore i created an analytical field with the function for an ellipse -> sqrt ( (1 - (pow ( Z ,2 ) /27225) )* 100). But in the results, Abaqus makes everytime a rectangular cross section.
I tried also other functions as sinus, linear or quadratic. But i always get an rectangular section.
I have created a new coordinate. But nothing works. I didn't find in any forum a solution for this problem.
Has anyone an idea, how i can solve this problem and get an ellipse cross section for the shell??

I uploaded a picture, that you can see the CAD. As you can see in the picture, i tried a lot of different coordinates and functions.

I have really no more idea. I hope you can help me!

Thanks for your help,

Chris

I have the abaqus student version 2017.

 

[FEA way]

Spatially varying shell thickness definition can be element-based (Shell thickness --> Element distribution) or node-based (Shell thickness --> Nodal distribution). In both cases you can choose either analytical field or discrete field. There are two types of analytical fields: expression and mapped. The one that you use is expression field but you may have to use mapped field instead.

The function seems incorrect, I think that it should define thickness based on X coordinate.

 

[ChrisBay]

Thanks you very much. I tried it with the mapped method and it works.
I have no idea why it doesn't work with the expression. I also tried it with the thickness defined on X coordinate and a lot of other methods but that didn't work either.

 

[FEA way]

It’s possible that Abaqus doesn’t support variable shell thickness defined using expression field (function). On the other hand, in such case no thickness should be assigned to shell and thus an error should appear right after the Analysis Input File Processor starts. What’s more, shell rendering wouldn’t work at all in such case. But from what you said in the first post it seems that somehow thickness was assigned to the shell (just not with proper distribution).

Can you share the .cae file of this analysis (the one with thickness defined using function) here ? I would take a look and try to find out why it doesn’t work as expected.