[lr103476]

Dear all,

Currently, I am studying the capabilities of using OpenFOAM for tracking lagrangian particles (kinematic clouds) and their interaction with a rotating object, such as propellors.

To combine a kinematicCloud and pimpleDymFoam, we modified the code, which is fairly straightforward, kinematicCloud.evolve() should do the trick.

I am currently solving the mixerVesselAMI2D tutorial by specifying particle injection outside of the AMI region. Everything start fine, until at some instance suddenly some particles are going through the wall and a bunch of particles are accelerating to reach velocities larger than the rotating regions, which is not physical.

Did anyone do something similar (resolve particle clouds combined with AMI regions) and encounter similar problems?

Thanks in advance!

Cheers, Frank

[thomasArk47]

I have worked on particle cloud + AMI a long time ago. I don't remember everything today

Ok, to start, are you running your cases in sequential or parallel mode? The last one currently don't implement particles crossing the AMI interfaces.

[lr103476]

Originally Posted by thomasArk47

I have worked on particle cloud + AMI a long time ago. I don't remember everything today

Ok, to start, are you running your cases in sequential or parallel mode? The last one currently don't implement particles crossing the AMI interfaces.

Hi Thomas,

Thanks for your reply. Well, to start with i have implemented the evolve cloud routine in pimpleDyMFoam of openfoam-3.0.x and running in serial only as a start. Even in serial when particles cross the ami interface some particles get lost and some particles are going to get wild and they will escape the domain.This all occurs for a simple mixerAMI2d tutorial.....

The GGI implementation of foam extend works well in parallel but it does not support particles to cross the GGI, all particles crossing are removed from the simulation.

It seems that there are several severe bugs in the official ESI release of Openfoam related to AMI and particle clouds.....

Regards, Frank

[Momo1805]

Hi there,

are there any news on this topic? I am currently simulating particle injection wiht sprayFoam using a moving rotor with AMI interfaces in OF-3.0.x which works fine but I really need to decompose the case.

My question:

Is there a possibility of doing both particle tracking with GGI/AMI interfaces together with particle tracking?

Thanks a lot and greets

Moritz

[CFDfun]

Dear all,
I coupled the pimpleDyMFoam with kinematicCloud lib. in order to track the particle moving inside stirring tank. UNfortunately, I faced the same problem which you had. The interaction between particle and rotating wall does not work well in OpenFOAM. Also the particles goes through the wall at some time steps and they are not able to recognize that there is a wall here. That's why, I would like to ask if you find a solution for this problem.

Best regards,
Amjad

[yuanlee2011]

Originally Posted by CFDfun

Dear all,
I coupled the pimpleDyMFoam with kinematicCloud lib. in order to track the particle moving inside stirring tank. UNfortunately, I faced the same problem which you had. The interaction between particle and rotating wall does not work well in OpenFOAM. Also the particles goes through the wall at some time steps and they are not able to recognize that there is a wall here. That's why, I would like to ask if you find a solution for this problem.

Best regards,
Amjad

did you solve this problem？

[yuanlee2011]

Originally Posted by lr103476

Dear all,

Currently, I am studying the capabilities of using OpenFOAM for tracking lagrangian particles (kinematic clouds) and their interaction with a rotating object, such as propellors.

To combine a kinematicCloud and pimpleDymFoam, we modified the code, which is fairly straightforward, kinematicCloud.evolve() should do the trick.

I am currently solving the mixerVesselAMI2D tutorial by specifying particle injection outside of the AMI region. Everything start fine, until at some instance suddenly some particles are going through the wall and a bunch of particles are accelerating to reach velocities larger than the rotating regions, which is not physical.

Did anyone do something similar (resolve particle clouds combined with AMI regions) and encounter similar problems?

Thanks in advance!

Cheers, Frank

How is it going on about the problem？

[Q.E.D.]

I'm struggling with the same problem...

I was able to successfully run a case for a certain configuration of particles + mesh + solver settings. Then I wanted to run the same configuration for particles with a higher density, but the error mentioned above appeared again...



PS. OpenFoam Version 5.0

[yuanlee2011]

Originally Posted by Q.E.D.

I'm struggling with the same problem...

I was able to successfully run a case for a certain configuration of particles + mesh + solver settings. Then I wanted to run the same configuration for particles with a higher density, but the error mentioned above appeared again...



PS. OpenFoam Version 5.0

did you use some certain solver directly or make some changes? can you share it for me?

[Q.E.D.]

Hi yuanlee2011,

first I have to emphasize that I'm definitely not an expert in lagrangian simulations. Maybe I was just lucky to find a physical solution for a certain configuration. For my calculations I used OF-5.0 and the standard DPMDyMFoam. I made the following observations: (Anyone who is more skilled than me is invited to correct or to explain these points)

• Parallel calculations are currently not working correctly. (Even with this solution: Running AMI case in parallel)
• I encountered certain situations, where the solver just stopped at specific times without giving any error message at all. In this case I restarted the calculation from a previously saved time step with a lower CFL-Number "to get past" this specific critical time step slowly.
• The size of your particles in relation to your geometry seems to be important. (Which is quite reasonable) In my case I wasn't able to calculate particle clouds with a particle diameter of greater than 10µm, where the smallest geometric feature had a size of ~1.2mm.
• I used a uniform distribution for the particle diameters with sizes between 1µm and 10µm, which worked quite well for me. But: you have be sure that your particle injection rate is high enough, because otherwise the particle diameters are not randomly distributed within the specified range. Therefore I used manual injection and generated particle positions in matlab.

I'm sorry, but I cannot share my case with you, because this was a industrial project. But I would recommend to start with a simplified case to get a feeling for what is possible with the solver.

Good Luck
Q.E.D.