Learning Forum

[lordofthethings]

Hello everyone,

I want to perform a turbulent air flow analysis through this "S" shaped pipe. The pressure drop across the tube is to be evaluated.

The first image just shows an example of the routing of the part, but the final geometry has quite lot of "Corrugates" as shown in the second image.

My question: How do I simplify the simulation setup in order to make it computationally efficient? As of now the real part has millions of nodes and elements. I see some potential to exploit the symmetry of the geometry, but I am not sure how valid the final results would be.

For example, I tried to perform a turbulent flow simulation on just the 2D cross section of the part in Fluent, but the final distribution of velocity etc was very very low compared to the full-3D.

Any suggestions/ ideas are welcome. Thank you.

[Rob]

It'll depend on the flow too. Run the first geometry and look at the results: are they symmetrical? Re the second, which way is the flow going?

[aitor.amatriain]

For approximate studies (estimation of mass flow rate and pressure loss), you could consider each bend as a pipe with a certain equivalent length; for example, for 90┬║ smooth bends, the equivalent pipe length is around 50 bend diameters (there is much literature available about this topic). Then, you could simulate the equivalent straight pipe with a 2D-axisymmetric model (or more simply, use the 1D momentum equation with Darcy-Weisbach + Colebrrok-White equations for the pressure loss, thus giving an analytic solution).
In any case, as points out, the error induced by the simplifications depends on the flow. In your particular case, there will be a value of Re from which the flow is separated in a region around the bends (and this flow is far from simple).