I am working on simulations for centrifugal fans with air foil blades. The predicted overall performance of the fan(e.g. total pressure rise, brake horse power, total efficiency) is much off from the test results. The error is about 10%, which seems big. Any advice on this kind of case to improve the simulation accuracy?

I would say this is special for ail foil blades. When the blades is single thickness curved plates, I can get very close results (predicted value is within 3-5% error comparing to test value)

Thanks,

Bian

Hello Bian, have you ever tried to test first only the airfoil blade ? then you test the entire fan ? A very hard thing to get the value in airfoil blades is the drag. For this simulations is good to use hex and O-grid type. And define in a very good way your boundary layer. These problens are not easy get to a very good accuracy, because it is specific and need a good computer as you get near and near to the test (experimental) results. Thanks. I hope this will help you a little bit.

Thanks, Jonas.

I did not test the airfoil blade first. Actually, it is hard to define the attack angle of the flow, which is varying along the blade height.

It seems airfoil needs much better surface and near wall mesh to get better simulation results, in which Hex and O-grid can help. Unfortunately, I can only generate Tetra and Prism meshes.

I think the best way is to simualte the airfoil blade itself first with good mesh. After getting resonable results, then put the blades into the fan wheel and keeping the meshes near around the blade. Is there an easy and better way, or a shortcut?

Bian

Have you done mesh refinement study? What turbulence model are you using - SST?

Bian,

I have been working on simulating fan performance as well and would like to share thoughts/questions. could you contact me at 845-679-1583. Thanks!

Giovan