Hello everyone, 1 When I use Y+/Y* Adaption to refine, how do I set the ¡°Max Allowed¡±? 2 When I use fluent to simulate 3D flow field in compressor, I find the streamline distribution on the suction surface exhibit differently if I employ different turbulent model(the k-epsilon Standard model, RNG model and Realizable model). In the results of the last two models, the streamline distribution deteriorate and don¡¯t accord with the actual status. What¡¯s the reason? And what should I do? Thank you!

hi,

For your first question, I believe you can set your "Max Allowed" around 60~120 (different book has different standard). By the way, from my experience, it is not good to refine your grid using this way. It will give you too many grids. The better way is to build a boundary layer from Gambit and set your first point next to the wall according to the Y+

I am not familiar with compressor. But you do have to choose a right turbulent model according to the specific problem you are working on. And please don't cheat Fluent as a 'black box", that will be very "dangerous" :-D

Good luck

Cindy

Thank you for your advice! But I don¡¯t know what the unit of Y+ and how to set the first point next to the wall according to the Y+. what is your idea? Thanks! raintung

raintung,

Y+ is the normalised wall distance and is described by the equation; y+=(yU*)/v, where y is the distance normal to the solid, U* is the wall friction velocity and v is the Kinematic viscosity.

Cindy,

Unfortunately the 'Black Box' approach seems to me to be the way that industry what's to go in, due to commercial pressures wanting quick reliable results - for example Nastran (FEA code) as developed such a tool in the form of visual Nastran, which is import the geometry and press a button. Therefore surely we most ensure that robust model are developed for this purpose.

Stephen