CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   y+ (https://www.cfd-online.com/Forums/main/2687-y.html)

 Raza Mirza October 12, 2000 19:48

y+

Hello all,

I am simulating flow (largerly internal flow) that goes through a duct. Then at the end of the duct it passes through a grille. This grille has grille bars whose size is of the order of a few millimeters.

I am using tetrahedral elements in the mesh throughout the computational domain. to resuove the geometry of the grille bar nicely I put elements on the order of less than a mm. Although I have not cheked it, but I am pretty certain I am violating the y+ criteria required by the turbulence model (standard k-epsilon with wall functions).

Fluid is air, and I am solving steady state, incompressible flow (no heat transfer).

Question: Is y+ constraint important here (it is like flow through a row of parallel tiny cylinders). I am thinking that y+ is important for the consideration of skin friction, however here the major portion of the pressure loss comes from the form drag. Then, is y+ important.

raza.

 allan October 20, 2000 17:40

Re: y+

Is the flow across the grill that important that you have to model it? Usually in situations like this a porous baffle can be used, with the porosity set to a value that it gives a pressure drop similar to the measured one across the grid. Using this approach removes the need for Y+ considerations completely, or if the grid is near the inlet then model your domain from the downstream side of the grid and set the length scale boundary condition to 10% of the grid spacing and a suitable value for turbulence intensity (I assume you have some measurements).

allan

 Raza Mirza October 22, 2000 14:31

Re: y+

I had thought about using a porous medium approach. But I do not know if that is appropriate. I do not know if the grille is important or not. I have not run the case with and without the grille and it seems it does not affect the results much (the grille is not really offering much resistance to the flow). So may be I should just ignore the grille for the initial iterations!

By the way, the grille is not near the inlet, rather it is in the middle of the computational domain.