I'm simulating a special elbow's pressure drop by Fluent. The results I got are about 60% of tested data. Your help is appreciated.

Problem description: Galvanized elbow for air system, elbow inlet and oulet openings are 25.5 inches, upstream duct and downstream duct openings are 40 inches, duct is 24 inches thick. Inlet flow velocity is 1750 fpm.

I built a 2D model with 8 equivalent diameters upstream and 10 equivalent diameters downstream duct. Inlet velocity and outlet pressure boundary conditions. I have tried standand k-e,realizable k-e, RNG k-e, RSM modles with standard wall functions,and standard k-e with near wall function, but all pressure drops just seem too small compare to tested data.

Addition simulation with a turning vane in the elbow has same problem, pressure drop looks too small in Fluent.

I am wondersing if the air flows is too bad to simulate in this elbow?

Thank you for your comments.

Tony

The answer to your problem is quite simple: by choosing to simulate your duct in a 2d environment, you neglected two very important aspects:

- the lateral walls of the duct, which contribute significantly to pressure losses by friction;

- three-dimensional effects generated by the same lateral-wall friction, that translates into additional turbulent-mixing losses.

So there are your missing 40 percents. The idea is that the flow is far from longitudinal symmetry, which is what 2d simulation is based on. And anyway, the friction on the lateral walls would still be missing... Take care next time when considering your modeling strategy! Sometimes 2d is not a good assumption.

Razvan

Thank you very much, Razvan. Tony