Fluids

Fluids

How to obtain good convergence in a multiphase cavitation simulation?

    • José Mantovani
      Subscriber

      Hello everyone!


      I make a cavitation simulation through a venturi tube. I dont have much experience with this type of simulation, and I want know how do to get a good convergence. I used the Schnerr-Sauer cavitation model with Realizable k-epsilon turbulence model. I create a unstructured mesh with inflation layer at walls. Below have a image with flow solution in 500 iterations and the controls settings. 



      I noticed a behavior of the solution and scribbled up the graph. How can I improve convergence? Perhaps performing a simulation without the multiphase model and obtaining convergence to later connect the cavitation model and simulate? Can anyone give me tips on this?


      I did not turn on the energy equation, but I'm trying to simulate a part of a system in which the venturi has, and in the actual experiment there is a temperature increase throughout the experiment, if I turn on the energy equation I'm going to get that temperature increase ?


      Also, for this type of simulation is there a better mesh type? Structured or unstructured?


      Another thing I noticed is that when opening the results in the CFD Post I did not get data like pressure and vapor fraction, this error occurred to me before when I simulated cavitation in orifice plates according to the tutorial of Raef Kobeissi, creating another file and doing everything again it's works, I do not know why this error/bug occurs.


      Thanks for the help!


      Mantovani. 

    • DrAmine
      Ansys Employee

      Use another implicit under-relaxation (Pseudo-Transient) + altering some explicit under-relaxation factors (especially for vaporization and density)+small time scales (very conservative). 


       


      Structured grids are always preferred whenever the flow is depicting a main flow direction or when you are trying to explicitly resolve some features /scales.. (less diffusive than other mesh topologies).


       


      Please track other issues in different threads.

    • José Mantovani
      Subscriber

      Very thanks Abenhadj, I try to make this and soon I post here some results or possible new doubts about the numerical solution. I will do a structured mesh and use the Pseudo-Transient method with very small time scales. 


      When you say about relaxation factors, should I go changing over the calculation or use a small value for every calculation?


      Thanks again.


      Regards,


      Mantovani. 

    • DrAmine
      Ansys Employee

      URF's are case dependent but do not influence the final results but might influence the way to final results. Stick first of all to pseudo-transient approach. 

    • José Mantovani
      Subscriber

      Okay Abenhadj, I will try this now.


      Thanks,


      Mantovani.

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