Directional heat flux varies ?

[mvp23]

I'm trying to analyze the thermal stresses induced in a pair of concentric cylinders due to a radially outward heat flux. I model them as being in perfect contact and there is a unifom heat flux load that is applied to the inner surface of the inner cylinder and the heat flow is radially outwards. There is coolant flow along the outer tube to carry away the heat. I use a sequentially coupled thermal stress analysis to simulate the same. Once I run the thermal model , I find that

- HFL1 and HFL2 ( transformed into cylindrical coords ) are non uniform. NT11 and HFL magnitude remain constant on the inner and outer tubes( which is expected) I used the quadratic heat transfer elements for the thermal part and quadratic reduced integration 3d stress elements for the mechanical part.

Contact conditions used :

surface to surface , finite sliding , surface smoothing on and adjust only to remove overclosure box checked
interaction properties specified - thermal conductance , frictionless tangential behavior , normal behavior would be hard contact and default constraint enforcement method ( penalty for surf surf finite sliding ) .

So when I use my thermal model to find the stress values , I get non uniform CPRESS , S22 (hoop stress in transformed coords) while the Von Mises stress remains constant.
I'm guessing the problem lies with the non uniform direction heat fluxes (HFL1,HFL2) which rub off on the S22 values and I get more of a sinusoidal profile when i plot hoop stress against the angle . I know this isn't right since the load I specify is uniform and I'd expect more of a linear behavior.

Any suggestions would be helpful .

 

[mvp23]

I'm attaching the cae file where i have the thermal part setup with the contact.

 

[mvp23]

and here is the jnl file . I've removed the mesh to reduce the file size .