CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Particle impaction on surface (https://www.cfd-online.com/Forums/cfx/94399-particle-impaction-surface.html)

 Julian K. November 14, 2011 18:47

Particle impaction on surface

Hi!

I am modelling the impaction of particles on a circular cylinder. My problem is, that the particles are not captured, although the distance of the particle's center to the cylinder's surface is less than 1/2*radius_particle. Thus my question:

When does CFX realize, that a particle hits a wall? When the particle's center coordinates coincide with the surface coordinates, or when the distance btw. particle center coordinates and surface coordinates is less than 1/2'particle_diameter?

FYI: I use Lagrangian particle tracking and the 'stick-to-wall' boundary condition for the cylinder surface.

 Julian K. November 16, 2011 09:44

I printed the tracking files and look at the final coordinates of captured particles. I found out, that this final coordiante lies on the collecting surface. Thus, CFX does not consider the particle's radius when checking for impaction (wall BC: Stick-to-Wall; Equation dependent with both restitution coeff. = 0 and wall roughness model with all dimensions set to 0 m).

This answers my first question, but my second question still remains, which is:

Who can I make CFX capture a particle, when the distance between particle centre and collecting surface is less than the particle's radius? Would I need to implement my own algorithm using a CEL? How would I do that? How would I start?

 ghorrocks November 16, 2011 19:48

As I said in your other thread, I do not think this can be done in CEL. You may be able to implement it in user fortran. Also have a look at the particle collision model in CFX, you may be able to adapt it to wall collisions also.

 juliom November 17, 2011 08:38

Please can you explain in more detail your work. You said you are using "stick to thw wall", so I think ytou are using a transiente model and you are calculating energy, right?
If not, you can just use the restitution coeffcient and thatīs it..

 Julian K. November 17, 2011 10:27

Quote:
 As I said in your other thread, I do not think this can be done in CEL. You may be able to implement it in user fortran. Also have a look at the particle collision model in CFX, you may be able to adapt it to wall collisions also.
I try to get in contact with Ansys and ask for advice. All I would need to do is to write an algorithem which calculates the distance between the cylinder's centre and the particle's centre. However, I have no experience with user fortran. Are there tutorials which explain what to do?

 ghorrocks November 17, 2011 17:01

Calculating that distance is not trivial - how do you find the closest wall? how do you calculate the normal distance?

User fortran is not thoroughly documented and few examples exist. ANSYS does run training courses on it. There is also some experience with user fortran on the forum (but not me, I have rarely used it).

 Julian K. November 17, 2011 17:25

Quote:
 Calculating that distance is not trivial - how do you find the closest wall? how do you calculate the normal distance?
I am simulating particle flow over a 2D circular cylinder. So I have the centre coordinates of the cylinder and the centre coordinates of the particle.

Quote:
 User fortran is not thoroughly documented and few examples exist. ANSYS does run training courses on it. There is also some experience with user fortran on the forum (but not me, I have rarely used it).
I will look into that. Thank you for the hints!

 ghorrocks November 17, 2011 17:36

If you can describe your wall analytically then it should be quite simple. But if you had to extract this from the mesh this would be a challenge.

 Julian K. November 17, 2011 17:49

Quote:
 If you can describe your wall analytically then it should be quite simple. But if you had to extract this from the mesh this would be a challenge.
I can describe the wall analytically. That's not the problem. For me it's more challenging to implement the algorithem with a user Fortran.

 Julian K. November 18, 2011 13:08

I opened a new thread for my problem: here

 Julian K. January 12, 2012 10:49

I found a way to solve my problem using a User Fortran. I posted the solution here.

 All times are GMT -4. The time now is 07:09.