Fluids

Fluids

Numerical instabilities in simulation of an airflow through a diffusor

    • simonhenn
      Subscriber

      Hello,

      I am simulating an airflow through a test stand for which I am using the k-omega SST model, SIMPLE solver and a pressure inlet and two pressure outlets (using a mass flow inlet turned out to be less stable). One part of the geometry is a diffusor with an opening angle of 8° so there probably aren't any flow instabilities occuring in the diffuser.

      However, in the simulation of the flow the diffuser seems to make problems since you can clearly see eddies forming and the residuals are oscillating. Here are some pictures of the geometry and the diffuser:


    • RK
      Ansys Employee
      Hi Simon, nWhat are your boundary conditions?n
    • simonhenn
      Subscriber
      Hi, nI use either a mass flow or a pressure inlet and two pressure outlets. Since I asked the question I have done some experiments and found that the diffuser is very likely the problem. nI meshed one diffuser with a circular inlet and outlet and one with rectangular inlets and outlets. Both inlets and outlets have the same area. The simulation of the circular diffuser works well, when I simulate the rectangular diffuser the residuals seem to converge at first but at some point backflow begins to occur and eddies form in the diffuser leading to the simulation becoming unstable. nHere are the pathlines from both simulations: nnThese are cuts through the meshes: nnWhat could possibly be the problem?.Thanks for your help!
    • Rob
      Ansys Employee
      It's hard to see what's happening to the flow in the square diffuser, but you may see slower regions in the corners as two walls will slow the flow up more than in the centre of a wall. Planes with contours and vectors will help there. nHowever, the transients you see in the original images is due to the jet passing into the open space: look up unconfined jets and coanda effects for the flow side of things and pitchfork bifurcation for a mathematical explanation. n
Viewing 3 reply threads
  • You must be logged in to reply to this topic.