General Mechanical

General Mechanical

Issues using symmetry model with thermal expansion

    • Rashi

      I'm trying to simulate effect of thermal expansion and rotational inertia of three objects. 

      Below are the four objects, all contact surfaces used with frictional contact. Following are the load steps which are used.

      step 1 : room temperature with no rotation

      step 2 : 250 C temperature 13,000 rpm (around the circular symmetric axis).

      Other than object 2,3 other objects are with different material properties.

      thermal expansion is changes as following,

      C.T. object 2,3 < C.T. object 1 < C.T. object 3

      Since the model and the loading symmetric I have used 3D symmetric simulation around the symmetric axis shown in the image. 

      Also full 3D simulation has being done. 

      The issue is the "signed von mises" stress results varies in object 2,3 in symmetric plane along rotation axis. Altair hyperworks have being used to process results. Below are the results


      Top full 3D, below 3D symmetric

      As you can see in 3D symmetric simulation in the inner edge of the circle stress is positive, but in the full 3D simulation stress is negative.

      Can there be problematic results in the symmetric plane due to thermal expansion where symmetric model is used?

      Also I've checked the results using 2D axisymmetric simulation considering object 2 and 3 as single object still I get similar positive stress. 

      Furthermore it have to be noted that this positive stress cannot be physically explained. 


    • peteroznewman


      Please describe the 3D boundary conditions fully. Is a temperature applied at one end?

      Is there a thermal contact resistance at the contact surface between Object 2 and 3? Due to the split line and potential thermal contact resistance, you cannot use a 2D axisymmetric assumption on this problem.

      The plots you are showing don't look like axisymmetric plots, they look like a 2D plane strain or plane stress model was used, which is totally wrong for this problem.

Viewing 1 reply thread
  • You must be logged in to reply to this topic.