Fluids

Fluids

SBES Shielding Function issues

    • JamesWright
      Subscriber

      I'm having some issues with applying the SBES turbulence model to my simulation. I'm trying to recreate the ERCOFTAC experiment found here. It includes turbulence statistics, of which I use the data at the -25mm station for the boundary conditions.


       


      The issue I'm running into is that the SBES shielding function doesn't go to into it's LES mode. Here's a time average plot of the simulation (taken over ~30 flow-through times). (Flow is going right to left)


      u2berggeist


       


      The turbulence data is experimentally based and is being interpolated correctly, so there shouldn't be any issues there.


       


      Could it be that this flow isn't suited to using an SBES turbulence model? Or should the shielding function always force the boundary layer to use RANS and the "free stream" to use LES regardless of it's separation status?


       


      Also, I know I probably didn't put enough info for the question, but I'm tired and just wanted to get it posted. Let me know if I need to explain any details.


       

    • DrAmine
      Ansys Employee

      Shielding Function is zero means that it is running in LES mode. So it offers the most superior shielding to keep your boundary layers in RANS regime even if refinement has been used. It would switch quick to pure LES mode whenever the mesh resolution is there.  SBES covers the entire boundary layer including the rapid growth area of the boundary layer due to the adverse pressure gradient which DDES won't reach safely. 


      As ANSYS Stuff we cannot give much more details here on open community and perhaps some others can chime in here.


       

    • JamesWright
      Subscriber

      So if I understand correctly, in order to have the LES switch occur, I probably need to refine my mesh so that shielding function has enough resolution to switch to a pure LES mode?

    • DrAmine
      Ansys Employee

      No I just commented on you picture. SBES will always try to keep the BL in RANS regime that is why we are using this strong shielding to avoid grid induced separation. 


      If you want a pure LES  ( still expensive) then with WALE or a wall modeled LES. The mesh requirements are very very high.

    • JamesWright
      Subscriber

      Ok, I gotcha.

      So my main question is why isn't the SBES Shielding function successfully restricting RANS to the BL? Because, as shown in the picture, it's using RANS for the far field turbulence model. The boundary layer height at the inlet is only around 4% of the actual pipe diameter (10mm in a 260mm diameter pipe), so there should be a significant portion that is in LES mode.


       


      To explain the simulation setup, flow is an axisymmetric 3D Structured mesh. The flow has a strong swirl component to it (max tangential velocity ~50% of mean axial velocity. See the data linked above for more details). The flow is going left to right. It goes through a length 100mm tube of 130mm diameter. It then enters a conical diffuser, and then into a plenum, where the walls of the plenum are atmospheric pressure boundary conditions.


       


      Other note: The only turbulence parameter that wasn't directly taken from the experimental results was the Specific Turbulence Dissipation Rate (ie. omega from k-omega SST). I used omega = sqrt(k) / L, where L is supposed to be the turbulence length scale. I used mixing length as the length scale, so L is 40% of the wall height (taken from Wilcox's Turbulence Modeling for CFD, pg. 29). Is this the correct way of taking the turbulence data in ERCOFTAC data and turning it into omega for the boundary conditions?

    • DrAmine
      Ansys Employee
      Check if you have sufficient mesh and temporal resolution. That are all the things you might influence. Based on precursor TABS you can make a proper evaluation of required resolution depending on how much you would try to resolve the eddies.
    • JamesWright
      Subscriber
      What do you mean by precursor TABS?
    • DrAmine
      Ansys Employee

      Auto correction. I meant precursor RANS. Your goal is to resolve more then 80% of cumulative turbulent kinetic energy meaning you require at least 5 cells for the integral scale.

Viewing 7 reply threads
  • You must be logged in to reply to this topic.