Learning Forum

[Vicky]

Dear sir,

I am woking in a turbulent flow simulation, using the k-epsilon and standard wall function. I am confused with what the ANSYS manual indicates for the Standard wall function: "The lower limit always lies in the order of y* ~15. Below this limit, wall functions will typically deteriorate and the accuracy of the solutions cannot be maintained".

I understand that the y+ must stay above 30 to be able to work with this wall function, however, the software provides me with a y+ of 33.6 and y* of 33.5, but when I calculate the y+ I get 83 as a result. Could you please help to understand why that is?

Thank you.

 

[Essence]

Hello,

The y+ value depends on the geometry, velocity of the flow, viscosity and density of the fluid. Therefore, if any one of the factors is changing in your domain, then this change will be reflected on the y+ value also.

[Vicky]

 

 

 

[Essence]

Hello Vicky,

Could you please re-post the question from your second reply? The message box looks blank to me.

[Vicky]

 

Hello, excuse me, I wanted to ask if, when doing the two-way coupling with the ROCKY DEM software, I should also consider that the first cell must have a y* greater than 15.
In my research I am analyzing everything related to collisions with the walls of the pipes, so analyzing everything that happens close to the wall is important.
However, I have this doubt since in the ROCKY manual they mention that we should have cells larger than the diameter of the particle. I would appreciate it if you could help me with this question; thank you

 

 

[Rob]

The age old problem of y+ and particles. The answer is yes, in theory the particles must be smaller than the cells and you need a y+ to suit the wall function. It's why multiphase remains easy to set up but more difficult to understand, and why I always need to be careful with answers. 

For DPM (in Fluent) the problem is wall collision is currently only checked in the near wall cell. So, if the near wall mesh is very fine particles can get trapped in the viscous sublayer and don't hit the wall. However, as the particle is bigger than the cell it should have hit the wall.... 

In Rocky the particle will hit the wall, but the porous term that's fed back into Fluent gets messy as it's intended for particles that are smaller than the cell. 

So, the engineering bit. Do you care most about particles or the flow? Do you have sufficient particles in Rocky that the odd cell not being quite right is a problem? Show some images as it's easier for me to comment on those without bending rules. 

Numerical stuff. There's a semi-resolved model in Rocky that will allow bit particles and that does some refinement on the Fluent mesh (adaption) to mitigate big particles. It works OK for water etc but read the maths before looking at anything much more viscous. I'd leave it for now unless you REALLY need it, it's fairly new and is seeing a fair bit of work.