Fluids

Fluids

Eulerian-Eulerian two phase flow Simulation in double pipe heat exchanger

    • Monta
      Subscriber

      Hello everyone, 

      I am simulating the two phase liquid-gas (bubbly) flow in a double pipe HE. The 2Phase flow streams inside the inner pipe and is cooled by water. I activated the multiphase mode and defined my phases. In the inlet of the pipe is a mass flow rate for each phase  and volume fraction for gas phase (oxygen) defined. In addition a bubble diameter of 0.2 mm is given. The gas bubbles are considered dispersed in the continuous liquid phase. A calculation using a mixture model was carried out. It converged. In the next step I wanted to move to the eulerian-eulerian model. This time the simulation is diverging. Modeling the interphase drag force with tomiyama model did not converge. So I tried another model (Schiller-Naumann) but the convergence is only achieved at the first time step (which is 2E-4s). Before starting any iteration at the second time step a floating point exception error occures. I checked the ansys manual and found that it is recommended to start with mixture model and then switch to eulerian. This did not solve the issue. 

      I would appreciate any suggestions or recommendations on this issue.

      Best regards,

      Monta

    • Prashanth
      Ansys Employee

      Hey

      Can you check if the same case works if you initialize and run using Eulerian from the start (using generic models like Schiller-Naumann etc). Also See if reducing time step helps. If it helps, try turning on the custom models (for your case as recommended) one by one and see how it behaves.

    • Monta
      Subscriber

       

      Thanks for replying! I tried already initializing and running using Eulerian from the start and did not help that’s why I tried starting with mixture model. Time step was 2E-3s and was reduced to 2E-4s. 

       

    • Prashanth
      Ansys Employee

      Try switching off surface tension and other taxing models to see if it improves.

    • Prashanth
      Ansys Employee

      Also check if the phase and turb BCs are set properly ? cross check using contours after initializing. Also, how is the mesh quality ? is the size small enough ? I see that the gas phase diameter is 0.2 mm. 

    • Monta
      Subscriber

       

      I set the time step to 10E-6 s, which is very very small. It delivers now a CFL < 250. Convergence lasts  for about 200 time steps and then diverges :/ Is there any tipps regarding meshing (1mm element size) and bubble diameter. Another thing, is it actually common to run eulerian-eulerian with  small time steps otherwise it diverges?  

       

    • Prashanth
      Ansys Employee

      Can you show something to look at. An image of the mesh and the residual behaviour just before divergence would be nice.

    • Monta
      Subscriber

      I attached a front view of the mesh and the residuals until floating point error occured.

    • Prashanth
      Ansys Employee

      Are you using Eulerian multifluid VOF model with explicit scheme ?

    • Monta
      Subscriber

      No Multifluid VOF just explicit sheme. 

    • Prashanth
      Ansys Employee

      I was wondering how the convergence behaviour was so good when you ran with mixture model. Did you consider slip velocity when you ran with mixture model ?

    • Monta
      Subscriber

      slip velocity was considered right! 

    • Prashanth
      Ansys Employee

      I see the the two phase flow is flowing at the core. You can try one case without inflation layers at the core and check if that helps with the sudden rise in the courant number. Basically simplify the mesh and model so that you can isolate the problem.

Viewing 12 reply threads
  • You must be logged in to reply to this topic.