# thermal stress for multilayer structure

Member

Hi, I would like two simulate the thermal stress a triple layer structure (triple layer pipe). In each structure the thermal and mechanical properties of the materials are different.

When I build the geometry, I have put all the solids in a single part.

I solve the thermal problem and the results look reasonable.

But when I try to solve the thermal stress problem, a large concentration of thermal stress exists in the interface between materials. I know it may due to the mismatch of thermal expansion coefficient. But even if there is a very small misdismatch, the thermal stress in the interface is large. I want to know if there is should be some setups in ANSYS workbench mechanical?

Then I put the same material in the each layer of the tube. And the thermal stress result is not the same as that in the single layer tube with the same total thickness and same material. Is this logical?

I have attached the project. In the geometry, I set steel to all the solid. Constant heat flux and constant temperture BCs are set for thermal problem. The constraint of displacement is put on the two end of the tube, which says that no displacement is allowed in the X and Y direction but movement in Z direction is allowable.

The weird thing is that the thermal stress in the center is quite small.  Is the triple layer structure that has the same material in each layer not the same as a single layer structure?

My final goal is to simulate a steel-copper-steel structure.

• Member
edited February 19

I use the ANSYS 18.1

• Member
edited February 19

Case 1. Three layer, each layer 1 mm thick, they are are stainless steel.

Case 2. Single layer of 3 mm, stainless steel.

Why these two cases show different thermal stress distribution?

• Member
edited February 20

This should be a 2D model, either Axisymmetric (best) or Plane Stress (thermal) and Plane Strain (structural)

If it is Plane Stress & Strain, there should be 2 planes of symmetry and be a 1/4 circle.

• Member
edited February 20

Thank you sir.

Actually, the real heat flux distribution in my project is not uniform. So in the three dimensional space, the temperature distribution is also not uniform.

Does the bonded connection between two materials assume an abrupt change of properties in the interface? I know it is logical that the thermal stress is large. But is there any settings in workbench that give a gradual change of properties in the interface between two materials?

Really appreciate for you help.

• Member
edited February 20

Is the heat flux uniform around the circumference, but non-uniform along the axis?  If so, that is an axisymmetric load.

There is no bonded contact in your 3D model, you are using Shared Topology to connect each layer with shared nodes.

For the 3D model you have, create a Cylindrical Coordinate System at the origin.

Change your Displacement BC to be a single edge that leave the Radial (X-axis) direction free.

The other end has no BC and is free to grow in the Z direction with the thermal expansion.

• Member
edited February 20

It is a simplied model for the test. I use it to see the thermal stress in the interface of two materials.

In the real project, the flux and temperature is calculated by ray-tracing method and CFD (). They are non-uniform in three dimension as could be seen in the attached figure.

My questions are that

1. Does the Shared Topology consider the interface between two material just like the bonded connection?

2. In my projected, the two materials are connected by 3D printing, so it may be seen as bonded. Does the bonded connection between two materials assume an abrupt change of properties in the interface?

3. Is there any settings in workbench that give a gradual change of properties in the interface between two materials?