Fluids

Fluids

Poor Shock Resolution

    • mmmaada00
      Subscriber

      I have simulating a oblique shockwaves in a channel with a terminating shock with the aim of analys8inig the total pressure loss. However, the resolution of the shock(s) in the solution is very poor (I have included an image below).

      I believe the mesh is sufficiently fine and has a high cell count with a good element quality. I am also using a density based solver.

      Could anybody offer help as to what I may do to fix this issue?

    • Murari Iyengar
      Ansys Employee
      Hi Can you post a screenshot of the image? Do not attach the image as we cannot open any attachments/links. You can also take a look at the "Courses" at the top of the page. There are a few Supersonic Flow tutorials which might be helpful for you.
    • mmmaada00
      Subscriber
      Apologies, here is a screenshot of the image
    • Rob
      Ansys Employee
      That looks pretty weird. Can you replot with node values off and also plot Mach Number?
    • mmmaada00
      Subscriber

    • Rob
      Ansys Employee
      Can you display mesh only at that location?
    • mmmaada00
      Subscriber
      This is the mesh in the region shown. I have tried reducing the element size even further but it takes a very long time to mesh.
    • Rob
      Ansys Employee
      Thanks, I was double checking for a discontinuity (interface) as I don't think resolution is necessarily to blame for the result. Is the solution well converged, and is it steady or transient?
    • mmmaada00
      Subscriber
      It is a steady simulation. The solution is also struggling to converge fully with most residuals around 1e-3 despite using higher order discretization. I should add that it is implicit and using AUSM flux type
    • Rob
      Ansys Employee
      Try using the Higher Order Term Relaxation and drop the Courant Number. I suspect the problem is convergence related. Should the shock be stable? I'm not an expert in shocks as in chemical engineering if it's going that quickly it's gone wrong....
    • mmmaada00
      Subscriber
      Weirdly, using high order term relaxation actually worsens convergence. The shock is and should be stable so I am a bit perplexed on how to achieve better convergence
Viewing 10 reply threads
  • You must be logged in to reply to this topic.