Small change in geometry makes answer completely wrong.

    • TryingMyBest1232


      I am relatively new to ANSYS Fluent so forgive me if I have made any stupid mistakes.

      I am trying to model a solar cyclone tower (SCT). This is identical to a solar updraft tower (SUT) except for the addition of swirl vanes to generate a cyclone. An SUT collects the sun's energy in a solar collector and heats air. This hot air rises through a central tower and can be used to turn turbines.

      I started by modelling the SUT to make sure that my setup was correct. With many previous studies available, I was able to predict the maximum updraft velocity to be 56.7m/s. The simulation value fell only a couple of meters per second away from this, leaving me confident that everything was working as expected.

      I moved on to testing the SCT. The setup was left exactly the same (duplicated) and the only change in the geometry was the addition of swirl vanes. Now fluent does not converge and the answer is all over the place. Every iteration has cells limited to T=5e3k and cells limited to T=1e0k. The velocities in the 'solution' are 6 times larger than the SUT, even though they should be fairly similar.

      Since both simulations have the same setup, I came to conclusion that the mesh must be the problem. So I tried everything I could to improve it. My experience is limited, but I found a lot of things I tried didn't help or made things worse. The mesh parameters seem reasonable, so I'm not sure if this is responsible or if it's something else.

      I would really appreciate some help on this, I've been stuck on it for over a week. I have attached images of the mesh as well as mesh parameters. I have also attached wbpz files for the SCT and SUT.



      Steady-state calculation.

      Heat is applied with solar race tracing and the DO radiation model

      inlet - pressure inlet 0Pa , 293k

      outlet - pressure outlet 0Pa, 293k

      glass collector - semi-transparent to solar rays. Thermal = mixed (coef = 10, temps = 293k)

      ground in collector - absorbs solar rays. Thermal = convection (coef = 10, temp = 293k)

      operating temperature - 293k

      operating density - 1.2 kg/m3

      viscosity - RNG k-epsilon

      coupling - SIMPLE

      gradient - Green-Gauss cell based

      pressure - PRESTO!

      All other methods - second order upwind


      Element Quality:

    • TryingMyBest1232
      The mesh shown here was generated with a multi zone sweep and inflation layers on the swirl vanes and tower. n
    • YasserSelima
      Try using Pseudo transient ... change all settings to first order until you get convergence, then set it back to higher ordernnIf you still can't get convergence, try running unsteady simulation n
Viewing 2 reply threads
  • You must be logged in to reply to this topic.