Fluids

Fluids

Compressible Fluid – Floating point exception – solution diverging

    • destroy
      Subscriber

      Hello,

      Im performing FSI simulation (Fluent+Mechanical+System Coupling) and I tried to use a compressible density model for Water - to check whether it helps with stabilizing solution. However, I am getting a diverged solution. This does not happen when I use a constant density. I am using the same settings as default. I also tried to use a Density Exponent equal 1.0 - with the same unsuccesfull result.

      My settings are in attached image. In Fluent I am using an operating pressure (Physics->operating conditions) equal 101325 Pa, and Referrence pressure (set in Refference values) equal 0 Pa.

      What I am doing wrong?

    • Karthik R
      Administrator
      Hello:
      Based on Tait's relation, have you estimated and compared the value of density with that of regular liquid water at 1 atm? You should be able to find the relationship here in the link below.
      8.3.4. Compressible Liquid Density Method (ansys.com)
      Karthik
    • destroy
      Subscriber
      I used the default values from the Fluent database which I believe are proper..
      According to the equation - when p=1 atm, K/K_0 = 1 and rho = rho_0 - but I dont know how that would help me
    • Rob
      Ansys Employee
      What did you set the boundary conditions as?
    • destroy
      Subscriber
      velocity inlet, which changes with time
      three pressure-outlets, with 0 Pa gauge pressure
    • Karthik R
      Administrator
      When does your simulation diverge? Right at the beginning or some time in the middle? Is this a transient run? Can you share some convergence plots?
      Karthik
    • destroy
      Subscriber
      OK I checked that this variable density works with the fluent standalone, with the same Fluent case. Works even when the motion is present (imposed via UDF cg_motion)
      It does not work when I use a coupling of Fluent and Mechanical. Dont know why.

      it diverges with the second coupling iteration, that is, if you are not familiar with System Coupling - after the second Fluent solving. Residuals are pretty good after the first Fluent solving. Then Mechanical calculates displacements and Fluent works again and diverges. This not happens with the constant density.
    • Rob
      Ansys Employee
      If you check the displacement in both cases how different is it?
    • destroy
      Subscriber
      similiar, very small - as this is the beginning of the timestep, small inlet velocity
    • Karthik R
      Administrator
      Hello Are you positive that it is not the dynamic mesh that is causing the issue? Which residual is causing this divergence?
      Karthik
    • destroy
      Subscriber
      Yeah, Im sure, the dynamic mesh works well when the constant density is used, I even tested it in Preview mesh, using define_grid_motion
      Every residual diverges
Viewing 10 reply threads
  • You must be logged in to reply to this topic.