Oil spray and fluid film in rotating cylinder

kamillaa · May 8, 2014, 07:44

Hi,
I'm trying to simulate a rotating cylinder with an oil spraying the walls from the center. I tried a VOF-model in Star-CCM+, but I believe the spray part is unfit for that model (too diffusive), even though it might have worked to simulate the fluid film.
So now I'm trying to use the multiphase segregated flow model, and even though I've tried to simplify the model a lot (by excluding the heat flux and the conjugate heat transfer problem, by setting a gravitational vertical force instead of the rotation and using a very simple geometry), it just doesn't want to converge. I really don't understand what the problem is. I've been stuck for almost two months with this problem (master thesis), and I thought it was time to ask for some help since my supervisor can't help.
Thanks in advance!

ping · May 12, 2014, 10:44

you might get more answers if you post on the correct forum - you could move it and my answer maybe

the vof model will be good for this but it looks like the oil is going to want to atomise and if that is the case then the cell sizes in that area should be at least as small as the drops so the mesh might become pretty fine in the outer radial areas but if you rotate the mesh with the spray then only a conical volume will need to be finer

and when solving make sure you are using unsteady and with a timestep that keeps the courant number below 1

me3840 · May 12, 2014, 20:57

VOF will be far too expensive for this. As ping said, you need to have an extremely fine grid to resolve the particles.

This is best done with a lagrangian+fluid film approach. How thick is the film on the walls? If it's rather thin you can use fluid film. Otherwise I believe that lagrangian can be injected into a VOF phase.

kamillaa · May 14, 2014, 04:33

Hi,
thanks for the reply! I've actually set up a lagrangian multiphase injector spray with the fluid film model. However the fluid film doesn't converge. Maybe it doesn't work with high rotational speeds and periodic boundary conditions. I was in touch with the CD-adapco customer support, but their feedback didn't help very much unfortunately. Maybe the time step is too big... It's just that the simulation is already so slow, so it takes at least 10h for me to know if I should continue to runt he simulation or not (and it's never been okay so far).

Best regards,
Kamilla

me3840 · May 14, 2014, 23:57

The model assumes the film layer is 'thin'; that is, much thinner than the size of the first cell. Make sure that is true.

Otherwise the model should work. How many degrees are you rotating per timestep? Ideally it will be 1 or 0.5. Depending on the speed, size, and number of cores, 10h may not be unrealistic.

May 12, 2014, 10:44		#2
ping Senior Member Ping Join Date: Mar 2009 Posts: 556 Rep Power: 20	you might get more answers if you post on the correct forum - you could move it and my answer maybe the vof model will be good for this but it looks like the oil is going to want to atomise and if that is the case then the cell sizes in that area should be at least as small as the drops so the mesh might become pretty fine in the outer radial areas but if you rotate the mesh with the spray then only a conical volume will need to be finer and when solving make sure you are using unsteady and with a timestep that keeps the courant number below 1

May 12, 2014, 20:57		#3
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 25	VOF will be far too expensive for this. As ping said, you need to have an extremely fine grid to resolve the particles. This is best done with a lagrangian+fluid film approach. How thick is the film on the walls? If it's rather thin you can use fluid film. Otherwise I believe that lagrangian can be injected into a VOF phase.

May 14, 2014, 04:33		#4
kamillaa New Member Kamilla Join Date: Feb 2013 Posts: 7 Rep Power: 13	Hi, thanks for the reply! I've actually set up a lagrangian multiphase injector spray with the fluid film model. However the fluid film doesn't converge. Maybe it doesn't work with high rotational speeds and periodic boundary conditions. I was in touch with the CD-adapco customer support, but their feedback didn't help very much unfortunately. Maybe the time step is too big... It's just that the simulation is already so slow, so it takes at least 10h for me to know if I should continue to runt he simulation or not (and it's never been okay so far). Best regards, Kamilla

May 14, 2014, 23:57		#5
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 25	The model assumes the film layer is 'thin'; that is, much thinner than the size of the first cell. Make sure that is true. Otherwise the model should work. How many degrees are you rotating per timestep? Ideally it will be 1 or 0.5. Depending on the speed, size, and number of cores, 10h may not be unrealistic.