[ClemensK]

Hi everyone,
I'm doing a steady state simulation (CFX R19.2) on a centrifugal fan with frozen rotor frame change. To avoid that dust adheres to the blade walls, water is injected at two positions. The water droplets are tracked in an Lagrangian manner as particles with fixed diameter (one-way-coupling). In the attached pictures you can see the particle paths, until the particles first hit the blade. Up from this point I want that all droplets stick to the wall and the water (particles) should be transported outwards (due to wall shear stress and centrifugal forces) to the blade tip, where the particles disengage from the blade tip and are centrifuged outwards. (In these pictures the particles rebound from the wall. I also tried verly low restitution coefficients, but that does not help either. )
For me it is important to know, how long a particle needs from the impact to the tip of the blade as a wall film.
Does anyone know how to model this problem? Is this even possible with the Lagrangian formulation?




In an easier simulation (backward facing step) I tried the wall film model. As the wall film model is only possible in a transient formulation, I took the steady state solution as inital condtion of the transient one. Water is injected as a cone. In CFD-Post I just see the particle hitting the wall, but from there it seems that the particles are sticked to the wall and no wall film is developed. How can I visualize the water film in CFD-Post?

[Gert-Jan]

When the particles hit the fan blade, these need to be converted in to a continous Eulerian phase, which allows transport over the surface to a large diameter. This conversion cannot be done in CFX, but is possible in Fluent, if I'm right.
So, you started with the wrong package. You are lucky though, since you can import a CFX mesh into Fluent, so a lot of work can be reused.
For the correct settings in Fluent to setup such a cse, I cannot help you. I try to avoid Fluent as much as possible. But there are tutorials available. Definitely.

[ClemensK]

Thanks for your answer.



Isn't there a possibility to get a Lagrangian wall film? I think, I already read something about this. Why can I choose the wall film option in the boundary condition of the wall, even tough this not possible?? This is strange.

I want to avoid the Eulerian formulation due to computational cost.


That's funny. I chose CFX, because it seems, that it is the better choice for turbomachinery applications. I also tried it with fluent at first, but it was not possible to import a mesh from Ansys Mesher and Turbogrid simultaniously in Workbench.

[Gert-Jan]

I have never had a question like this, so never had to think about it. However, I have seen animations on things like this in Star-CCM+ and, I think, Fluent.

A wall film is continuous liquid sticked to a surface where surface tension comes into play as well. You cannot model this with lagrangian particles that are hard spheres. So better use a Eulerian model that is expensive than a lagrangian model that is unrealistc. Maybe start with a single segment of your fan instead of full 3D......

And you are right that CFX is better for turbomachinery, but it is lacking some physical models, like what you need.

[ghorrocks]

CFX has a wall film model. At least on the current version it does (V21.2). It works with the Lagrangian particle model so that when particles hit a wall they build up a wall film, and the wall film can gain and lose mass and energy. So I think this model will do what you want. Have a look in the documentation.

Not sure if there is a tutorial example using it. If not look on the ANSYS Customer webpage for one or contact ANSYS support for an example case.