Fluids

Fluids

Floating point error after running for 11seconds

    • friday.abolorunke
      Subscriber

       

      Description.

      I am simulating a two-phase multicompoent flow with a porous and open region.

      VOF model, Species transport, Laminar flow. UDF for interfacial mass transfer.

      Average mesh quaility: 0.99998, sqewness 0.0105. I dont think the issue is with the mesh. Also, the UDF is fine. I have used it to solve another problem without porous media and this float point issue did not occur. No dynamic mesh.

       

      Problem:

      The simulation suddenly diverges after running for a flow time of about 11s. Even though the solution converges each time step before suddenly diverging. 

      I think the issue is from the liquid-species. Please has anyone encountered this before? what can I do? See picture attached.

       

       

      Thank you

       

    • Rob
      Ansys Employee

      How does the flow look a bit before divergence? Ie has a free surface position suddenly shifted, or mass transfer altered. 

    • friday.abolorunke
      Subscriber

       

      Hi Rob,

      The condition for the mass transfer to begin has not been met, so the interfacial mass transfer is zero. I don't think there is a change in the flow field. I am running the simulation in batch mode without saving contour plots or data files during the simulation.

       

    • Rob
      Ansys Employee

      No data saves? What are you going to post process? How has the time step altered over the last 10-50 time steps? Back flow may trigger problems, but it can also be an indicator that something else is going on. 

      • friday.abolorunke
        Subscriber

        The time step size did change much for the last 50 time steps. Between 1.3755e-3 - 1.44e-3. The max time step is 35 and the simulation consistently converges around 10th - 8th time step. 

         

        I did not save the data file because I have a different model that runs without issues, except it has significant reverse flow that leads to mass loss at the inlet. I made some changes to the shape of the inlet to try to eliminate of reduce reverse flow.  

    • Rob
      Ansys Employee

      OK. Unfortunately you'll need to re-run to about 10.5s and then save more data from there. Adaptive time stepping can trigger instabilities if the oscillation interacts badly with the flow time scales. 

      • friday.abolorunke
        Subscriber

        I will do that and let you know what the flow field looks like. Thank you.

Viewing 4 reply threads
  • You must be logged in to reply to this topic.