Fluids

Fluids

Topics related to Fluent, CFX, Turbogrid and more.

Converging a periodic case / Setting BC correctly

    • atulsingh92
      Subscriber

      I have a periodic unit of a tube. Surface is impressed with a helical groove.

      Mesh is above 0.1 orthogonal quality, magenta and green region is wall, which later merges in the run, instead of the mesh

    • SVV
      Ansys Employee
      Hi Lower convergence of energy might be the cause of higher temperature at outlet. How do you define the periodic condition and inlet mass flow condition? You need to bring down the residuals of energy.
    • atulsingh92
      Subscriber
      I have mentioned that with the small sketch in black.
      inlet and outlets are interfaces, then made into non-conformal periodic boundary condition. Massflow inlet is defined for water at 70 degrees C.
      Also, like i said, have lowered the under-relaxation for turb-viscosity, which didn't help.
      So how can I bring down these residuals?
    • atulsingh92
      Subscriber
      ?

    • Rob
      Forum Moderator
      What reference values are you using for HTC? That may explain the negative, the outlet temperature is likely convergence related. What's the cell & mesh quality like?
    • SVV
      Ansys Employee
      Hi How can you set both mass flow inlet and periodic condition on a single face? Please correct me if I am missing out something.
      Make sure your cell aspect ratio is less than 300, though sometimes slightly higher values are also fine. Its always recommended to have y+ ~ 1 with more layers at boundary layer.
    • atulsingh92
      Subscriber
      Mesh quality is as indicated in the very first image , i.e. orhto quality approx 0.15 ish.
      Regarding reference temperature, I am not sure what the correct value should be. I know that my bulk inlet is at 70C and wall is at 28C, and have a mass flow rate for a certain reynolds.
      I tried various reference temperature as follows
      22.85 was randomly chosen, this is when the htc was negative
      49 was chosen as a mean of bulk and wall temperature. Here the residual do converge, however, i am unsure if the htc is correct. (Exp result for this bc of 70 C bulk and 28C wall should yeild htc of 2592.03)
      30 was chosen becuase I thought ref temperature could be the free stream temperature around the tube wall, so would be slightly higher than wall temperature of 28. This didn't converge
      70 was assumed becuase, again, I thought, ref temperature for internal flow could be of bulk temperature, as that would be the free stream temperature as athe temperature converges. However, this too didn't converge.
      ( Not sure if it is important , but I also got turbulent viscosity greater than 1e5 in so and so cells in many of the meshes that I tested. I corrected this with initializing epsilon to 0.01 and k to 100 from another question on this forum. However, my guess is this should not have appeared as the flow is of internal tube with 13mm diameter tube, that too a periodic unit. MEsh was already at min ortho of 0.1, while points out in that question that scale is important and should be corrected with changing the limits.)
      /forum/discussion/3648/turbulent-viscosity-limited-to-viscosity-ratio-of-1-000000e-05-in-5-cells

      So, the question that I would need help with is, what do I keep my reference temperature, if wall is 28C and bulk is 70C, for a periodic tube, with mesh of min 0.1 orthogonal quality?
    • atulsingh92
      Subscriber
      That's just how periodic conditions are set in fluent (2019R2), one needs either mass flow rate or pressure gradient as input, and then inlet temperature of fluid , only then periodic bc is fully defined. (Wall temperature is defined in wall bc)
      Yes, I am using a aspect ration boundary layer generation, which already yields a y+ of roughy 0.4 or 0.5 to me.
      Also, would you have any other suggestion for me, regarding the reference temperature that I set for various of my cases in the table above?
      Could use some help related to it.

    • Rob
      Forum Moderator
      Reference temperature (in the Reference Values) shouldn't alter convergence. Since you have a heat flux and a bulk volume temperature I'd calculate manually, that way you can compare with your experiments: HTC is a comparison value, what you're really after is a heat flux as that's what you will measure in reality.
    • atulsingh92
      Subscriber
      you mean manually calculate the reference temperature, or manually calculate the htc, from heat flux and bulk temperature?
      Also, I am still left with the question of what should i enter as my ref temperature, as that should be the input.
      Or do you mean, that I don't observe htc and nusselt from surface-integrals as they are ref temperature dependent, just observe the converged heat flux (q), and manually calculate htc, form h = q/(Twall - Tbulk)?
    • Rob
      Forum Moderator
      Manually calculate the HTC. The reference value temperature (not the one on the periodic boundary) is one you choose to compare results. You still need a dT for the equation but that's for you to decide: the heat flux per unit area might be a better comparison?
Viewing 10 reply threads
  • The topic ‘Converging a periodic case / Setting BC correctly’ is closed to new replies.