VOF-DPM particle bounce

I built VOF-DPM model of rotatory flow with particle sedimentation. VOF part gives me very good results, but I have a problem with DPM part of the model. For some reason particles bounce off the wall and escape fluid, I attached picture with particle tracks. Particle time is correct, but their location is outside the fluid (green isosurface VOF=0.5).

I tried with and without interaction with continuous phase at DPM window; with and without virtual mass force, pressure gradient force and even two-way turbulence coupling (it's not suitable) and physical models - those were in various combinations. I set wall DPM reflection coefficients parameters as Normal/ Tangent as 0.3/1; 0.3/0; 0/0 and left blank polynominal. I also tried DEM collision model and collision partner has same value configurations. With DEM convergence was not so good and it is not necessary. I found that DPM is more suitable for solid particles and DEM is better for droplets.
In solution methods, I tried pressure-velocity coupling simple and coupled. I always use Second order upwind for discretization.

Is there something I'm missing? My domain is filled with air initially, and the mixture of fluid and particles enters the domain, why particle still bounce off the fluid?
I would be very grateful for any suggestion.


  • DrAmineDrAmine GermanyForum Coordinator
    edited April 2019

    Please insert your screenshot here. ANSYS Stuff do not download attachments. Moreover list all models included in your case. Bear in mind that particles are transported by the bulk and do not see any interface

  • OracleOracle Member
    edited April 2019

    I have three phases: two continuous (air an wort) and one dispersed (residue). Air is present at the beginning in the tank to create a free surface with wort. Thus for these two previously I used for two-phase model VOF. My current goal is to include dispersed phase suspended in the wort. So for air-wort I use VOF (implicid); turbulence is modeled by k-omega SST; residue is injected at the inlet, but it bounces. I followed set ups from tutorials, youtube videos and I created my own modifications.

    So is my display incorrect? Should I monitor particles separately from isosuface?

    Particle time and velocity are correct with the time and velocity of simulation and wort.

    Kind regards,


  • RobRob UKForum Coordinator
    edited April 2019

    Particles will interact with the wall based on settings there, they'll also (typically) pass into a second phase without slowing down other than due to the change in density & viscosity. 

    Read up on the wall film models: I suspect that's what you're after? 

  • OracleOracle Member
    edited April 2019

    No, I am not modeling wall film. Wort is supposed to fill the tank and particles should flow with wort suspended in it and collect at the bottom. 

    I am looking for a way to keep particles in the wort.


  • DrAmineDrAmine GermanyForum Coordinator
    edited April 2019

    Particles are transported with bulk and do not see the free surface. If you want to limit the particles to be only in a certain phase you need to use UDF's. Alternative is to use beta feature with Eulerian Model but here again you require some froces / treatment to let the particles bounce or disappear if they are approaching the interface.


    Alternative is to use a scalar which get transported with "Wort"

  • OracleOracle Member
    edited April 2019

    Thank you for suggestions. From what I read UDF could be the solution.

    Thank you again for time and help!

  • MaxxMaxx Member
    edited June 9

    I'm implementing a VOF-DPM model with wo phases (water and air) and solid DPM particles. However I would like the DPM particles flowing only in the water phase, but they go to air too. From the previous comments, you had a similar problem and the solution for that was a UDF. I looked for examples of UDFs for this case, but I do not found anyone. Could you please provide me a example of UDF to limit DPM flow in only one phase or some explanation on how to do that?

    I thank you in advance

  • OracleOracle Member
    edited June 11

    Hello Maxx,

    I have given up on writing a UDF. I have consulted this problem with few people and I was advised to try DDPM instead of writing UDF. Maybe that will work for you.

    However, in my case turbulence and time/physical scale were too mych for DDPM to handle. Time step was so small calculations were unacceptably long. 

    I wish you good luck!

Sign In or Register to comment.