I am interested in creating a wall boundary condition that can switch from no-slip to free-slip depending on temperature. Is this possible with cfx 5.7?

Any advice would be appreciated. Malcolm

Hi,

In CFX, the B.C. type cannot be changed: you cannot switch from no-slip to free-slip, as you cannot, for example, switch from inlet to wall.

But there may be other ways to implement a simmilar behaviour. You may define the walls as free-slip, create a subdomain (1 element thick) next to the walls and define a source of momentum as:<font color="blue">Sm=-C*U*step(T-Ts)</font> Sm will be equal to -C*U for T above Ts, and 0 for T below Ts. If you set C to a large value, the source term -C*U will provoque a zero velocity at the walls when T>Ts. While the for T below Ts, the wall will be a free-slip wall.

Are you performing a steady-state or a transient simulation? And why would a wall switch from no-slip to free-slip?

Regards,

Rui

Hi Rui,

Thanks for the advice.

Currently I'm performing a steady-state simulation.

I want to switch the wall from no-slip to free-slip because I am modeling solidification. At some points liquid is in contacting with the mold while at other locations a 'moving solid' is present. I am using a single phase model where the moving solid is simulated using temperature dependent momentum sources.

Malcolm

you can set a no slip wall with a wall velocity function of the temperature

e.g. wall velocity u= step(T-T*) * YOUR_velocity

if T<T* ---> step(T-T*)= 0 ---> u=0 ---> wall no slip if T>T* ---> step(T-T*)= 1 ---> u=YOUR_velocity ---> wall moving

Hi,

That's interesting. But how would you define YOUR_velocity to simulate a free-slip wall? Do you think it should be the tangencial component of the fluid velocity, Vt, at the wall? Do you think that defining YOUR_Velocity as: YOUR_velocity = Vt = V - Vn = V - (V dot n) * n would work?Are the variables Normal X, Normal Y and Normal Z available in CFX-Pre and Solver? Or are they only avaialble in Post?

Rui

Hi ,
Just more description about <font color="blue">step()</font> function :

1. if a < b then step(a-b)=0
2. if a = b then step(a-b)=0.5
3. if a > b then step(a-b)=1

Normal X, Normal Y and Normal Z are NOT available in CFX-Pre... Maybe the problem is that you have a curved wall? If you have a plane wall you can use obviously the velocity component u,v,w.

Hi,

But would you define YOUR_velocity = (u,v,w), or in case a wall is perpendicular to the y-axis would you define it as (u,0,w)?

Have you ever tried to implement this wall definition? Are there problems in obtaining convergence?

Regards,

Rui

Never tried Rui, also because defining a velocity of the wall equal to the velocity of the flow is not really like having a free slip.....I mean...is quite good but I don't think is the same...it depemnds from the geometry of the problem...

but I've got no other ideas......so this is the best setup ...

Hi,

Probably you're rigth when you say "defining a velocity of the wall equal to the velocity of the flow is not really like having a free slip".However if you define the wall velocity (which corresponds to the hybrid value of velocity) equal to the tangencial component of the flow velocity (which corresponds to the conservative value of the tangencial component of the velocity - the mean velocity of the control volume next to the wall) it will result in zero shear stress --> free-slip condition

But it sounds like it will lead to convergence difficulties, as the flow velocity will "push" the wall and the wall velocity will "push" the flow, but I'm not sure.

I think it would worth experimenting. But it´s a shame you cannot use Normal X, Y and Z in CFX-Solver.

Regards,

Rui

I agree when you say that the wall with a tangential velocity should push the flow and that is not physically right. I agree also that is a shame you cannot use Normal X, Y and Z , and in my case massFlow and massFlowAve too.

Regards stex