General Mechanical

General Mechanical

Topics relate to Mechanical Enterprise, Motion, Additive Print and more

Coupling thermal and structural analysis

    • Christian Neuer

      Hi everyone, 

      I'm trying to simulate deformation due to thermal expansion in multilayered fire-resistant glasses. At the moment my model consists of two glass panes with an intumescent interlayer in the middle. (There is also a steel plate in the model, where the glass is lying on, which serves as a heat source via its body temperature)

      My workbench looks like this:

      And at the moment I calculate the thermal part first and after that I calculate the mechanical part. And as you can see, for now the simulation works, but there is still a big difference to real world experiments and thats why I want to couple the simulation step by step if this is possible.

      First, the expansion of the intumescent material is huge, it's greater than 10x (1.4mm at the start, 14.8mm at the end of the experiment). Also, the material properties change with temperature. Finally, when I look at the end result of the the structural simulation, the edges of the glass lost contact (bent upwards) sometime during the heating process, but this wasn't considered in the thermal analysis.

      For all these reasons, I want to couple the two systems step by step, so after the first thermal time step is completed I want to give the result to the structural system, then use the result (which is mainly the change geometry) as input for the next time step of the thermal system and so on.

      I also just watched this video:

      but it seems, while there is a different way of connecting the 2 systems, the way shown in the video also does not allow a data transfer in both directions.

      I would be very happy if someone could help me, thank you!

    • Sampat Kumar
      Ansys Employee

      Hi Christian,
      Thanks for the detailed explanation.
      What type of contact is given between the glass and intumescent interlayer?
      Yes, you can perform the two-step method, where you perform the thermal analysis in the first step and solve the structural analysis in the second step. For doing that, you should enter the number of steps 2.
      Coupling is a good way to simulate this model,but have you thought about the external data to transfer the heat from the first model and use it for the second model for the structural analysis?

      Please let me know. 


    • Christian Neuer

      Hi Sampat,

      thank you for your fast reply. I will attach a few pictures to further illustrate my problem:

      1) The contacts are bonded for both glass/intumescent contacts and frictional for the glass/steel contact (since the glass probe is lying on the hotplate in this experiment) (friction coefficient 0.2)

      contacts in the model and material assignments

      2) a project overview with a temperature distribution at random timestep:

      project overview and temperature distribution in thermal system

      3) shows the last time step in the structural analysis with the intumescent fully expanded:

      last time step of structural analysis with directional deformation shown

      At the moment, as I said in my initial post, I always complete the thermal analysis first, then I use the temperatures calculated at each timestep in the structural analysis via “Imported Load” and perform the structural analysis. The way I try to compensate for the expansion in the thermal analysis is that I use a adapted λ (I divide the λ that the intumescent material would have at a certain temperature by the expansion rate).

      So but what I want is as follows:

      1. Calculate time step 1 in the thermal system
      2. transfer the result to the structural system, calculate the time step there
      3. transfer the structural result back to the thermal system (if the intumescent has expanded, then the way the heat has to go through is longer, therefore the λ has no longer to be adapted)
      4. repeat until it is done


      Can I do that somehow?



Viewing 2 reply threads
  • You must be logged in to reply to this topic.