[Torsten]

Hi,

I'm going to simulate the varnish particle trajectory starting from a rotary sprayer to a steel plate and use a structured cubic mesh (not cylindrical), because we want to couple the flow results with matlab and therefor "nice" cartesian coordiantes are required.

I have three different types of boundarys:
1st: the inlet, startet from the bottom of the rotary sprayer head (the circle shape in the picture)
2nd: the bottom (steel plate), declared as a no slip wall
and 3rd: the remaining 4 sides, declared as an opening.

And those sides are the problem within my cubic geometry! I can't use a direction normal to the boundary. That produces a flow like shown in the picture "vector field" 1 to 3.
On the other hand, I'm not able to define all three directions manually. The x- and y-direction (radial and tangential direction, respectively) should easily be done with an expression that depands on the x and y coordinats. But we don't know the z-direction a priori.

To cut a long story short: Does somebody know a way, how to implement the boundary condition for my case. (As mentioned before, the cubic geometry is unfortunately necessary).

Thanks in advance for your help,
Torsten


PS: I'm using the Ansys Workbench and CFX version 14.5

[ghorrocks]

Please attach your CCL or your output file.

[Torsten]

I attached the ccl.

Here are some further informations about my model:
We establish the rotaion by domain rotation and holding the inlet stationary. One quick mistake in my previous post, the inlet is an opening too. Because beneath the rotary sprayer head is a backflow.
At a first step, we only want to generate the pure fluid flow (air). If that produces reliable values, we are going to add the particles in the flow!

[ghorrocks]

I recommend you do a simpler simulation with just a single phase to start off with. Make sure you can model air flowing through a rotating domain without weird flow features. You may need to use the "Alternate Rotation Model" in the rotating domain options to reduce spurious flows from rotation. When the single phase simulation is working then return to the multiphase simulation.

[Torsten]

Thank you for your quick response.

I will try the "Alternate Rotation Model" out, maybe this will help me.

Regarding the single phase start off: I set the volume fraction of air to 1 and zero for the particle repectively. Thus I'm considering in my simulations only a single phase. Or does even the declaration of a second fluid (particle in my case) have an impact on the simulation. If yes, I will delete the particle definition in my default domain.