Hallo all

My question is: How does the cfd-programs deal with the velocity near the wall, when the flow is turblent and not parallel to the surface?

When the flow is parallel the programs either use some kind of law-of the wall or uses these new two-layer models. But what if the flow for exampel is over a backward facing step? (the point where the flow divides and a part goes in the recirculation zone just after the expansion and the main part flows down thorugh the channel). Or a jet hitting a wall??

Regards

Bo

Dear Bo,

by using wall functions, you have to split the mean velocity vector (at the node closest to the wall) in its vertical and parallel component. The component parallel to the wall is the quantity you have to use in the wall function approach. For recirculating flows the normalized quantity in the u+ definition is formulated as: u+=v_par / u_tau = v_par / (c_mue**0.25 * sqrt(k)) (1). This is done to avoid singularity at separation and reattachment points where the wall shear stress is zero. In eqn. (1) the assumption u_tau = c_mue**0.25 * sqrt(k) has been introduced which is valid for flows in local equilibrium.

When you resolve the boundary layer with a two layer model ,as far as the velocity is concerned, you can treat the velocity at the cell closest to the wall like the velocity at the other grid points.

Regards,

Dietmar

Is the wall function also used for the zero-equation models as well, or it is just for the one or two equation models?

Thanks

MJ