Fluids

Fluids

Compressible Flow Steam Ejector Error Divergence Detected in AMG Solver

    • Rahadyan
      Subscriber

      Hi everyone, I am IAN

      thank you for your time to read my question,

      I am trying to simulate a flow of steam ejector as a 2d Axisymetric based on this paper https://www.sciencedirect.com/science/article/pii/S1290072912002098, but I have an error "divergence in AMG solver" in the begining of calculation

      currently, I am using Ansys R19.1

      I read about this error in ansys forum (https://forum.ansys.com/discussion/23920/problem-with-tutorial-in-fluent-divergence-amg-epsilon-for-ic-engine-cold-flow-simulation/p1) but still doesn't work

      I checked my model,

      the average orthogonal quality ,skewness quality, and aspect ratio is pretty nice, but I have problem with the small curvature of nozzle throat, so the max point of skewness occur in that area

    • DrAmine
      Ansys Employee
      Is it possible to ramp the pressure at outlet down by starting from high value? Also think about using transient solver at first.
    • Rob
      Ansys Employee
      The other thing to do is patch in the fast moving flow from the high pressure section. Using all pressure boundaries tends to make convergence more difficult and ejector flow is complex in that it's an entrainment flow system.
      What Mach Number are you expecting in the throat? Pressure based solver is good to around Mach 2-3 so you may find that a better choice.
    • Rahadyan
      Subscriber



      hi , thank you so much Rob for your answer
      I really want to say sorry that I miss the notification, unfortunately, I haven't solved it.
      the Mach number will be unity in the throat and will increase up to 1300m/s at the nozzle outlet (based on paper). so the density-based model will be sufficient for this (based on paper)
      I don't really know why, but if my mesh is coarse (2000 element) the simulation goes well.
      but when I increase up until 40000 like the paper said, the hot spot "excessive temperature" occur
      would you like to give me an advice, thank you so much
    • Rob
      Ansys Employee
      With a coarse mesh you may be diffusing some of the peak temperature away to the solver is less accurate but marginally more stable. As you resolve the flow you may be seeing a more accurate solution and picking up more flow features. Are you increasing resolution in the entire domain or just where the flow gets more interesting?
    • Rahadyan
      Subscriber
      Hi Rob, Thank you so much for your question
      I am very happy that my simulation works.
      I guess it is due to the transition between each mesh. I do a Bias to ensure the transition very smooth
      thank you so much
Viewing 5 reply threads
  • You must be logged in to reply to this topic.