1 Reply Latest reply on Aug 15, 2008 2:13 PM by Bill McEachern

    Meshing very long ducts


      We started recently to use Floworks to analyse thermal profiles on bus ducts and we are running into problems and problems. The tech support is not really helping us a lot because they have never worked with such geometry. Maybe somebody can help us here... We took the 2 days training but I would really like to have a "Meshing for Dummies" book on hand right now....

      Here is a brief description of the geometry and the setup:

      - The bus consist of 2 concentric cylinders. The internal cylinder has 14.5" O.D. X 0.6" thk. and the external 31" x 0.25" thk.
      - The total lenght is approximately 250 feet.
      -Air is going in by one of the tube and returning by the 2 others.
      - This is an external analysis and the computational domain has been reduced to just include the full geometry (it was way too large)
      - For this run, we had around 2.5 millions elements and it took around 100 hours to converge.

      Problems we are running into :

      - Flow trajectories going out of the model. SW Tech support said that this is a bug in the post-processor and they are working on this. We first tought that it would be a coarse mesh issue but they said that this is probably not the root cause.

      - Mass flow rates do not balance between the inlet and the outlets (30% lost approx). Tech support said that it might be a problem with the pst-processor, but we are measuring these rates with surface goals (refer to screenshot013), which are calculated by the solver, not the post-processor. These goals are measured on the internal surface of the inlet and outlet lids. Can it be again a meshing issue ? Since the mass flow rates do not balance, I do not trust my temperatures.... If really 30% of my air is lost somewhere, it will certainly affect the results...

      Meshing is the complex part in this kind of project. Does it exist somewhere a good guide or manual that explains the meshing ?

      FW help is not really easy to understand and the examples / tutorials are not similar to our application...

      Any help or ideas are welcome....

      Thanks in advance

        • Meshing very long ducts
          Bill McEachern
          try something shorter to remove the aspect ratio issue - like something in the 10's of feet and not hundreds and see if you get some sensible numbers. The other thing I would do is run it as an internal analysis and apply a heat transfer coefficient to the outer walls and get that to work with shorter geometry. Then you should extend the geometry and adjust the meshing to suit. You can apply mesh controls on the pipes themselves. Also in the summary you might want to see how many irregular cells you have though with the advent of the thin wall optimization this hasn't been much of an issue. Thin wall optimization should always be checked. High aspect ratio problems can be a challenge with any code. Once you find what sort of mesh works on something shorter you will know what is needed for the whole thing at theast from a cross sectional point of view. good luck