[rahulksoni]

I want to perform moving mesh cfd simulation with discrete granular phase. I guess DPMDymFoam is for that purpose as I have seen over internet. But is it distributed with any version of openfoam. How to get it? Will it need separate compilation.

[Yann]

Which version of OpenFOAM do you use?



I think DPMDyMFoam is available since OpenFOAM-5 / OpenFOAM-v1706 and above.

[rahulksoni]

As such I have the installation of all 3x, 5x and 6 versions. But in none of them, I find the tutorial for solver MPPICDyMFoam or DPMDyMFoam. You can check here:
https://github.com/OpenFOAM/OpenFOAM...als/lagrangian
https://github.com/OpenFOAM/OpenFOAM...als/lagrangian

We can see there is no mention of DPMDyMFoam or MPPICDyMFoam. Or it is like that there is no tutorial provided for these solvers and we need to manually make the case for same.

OMG, I have checked the commands DPMDyMFoam and MPPICDyMFoam working in OpenFOAM5.x installation. Now, my question is how to set up the case in absence of a tutorial case. Can anyone help me out?

[Yann]

Hello Rahul,

As you have seen, the solvers are implemented, there are just no tutorials for it.

The Lagrangian approach and dynamic mesh capabilities are 2 different things. So you can start from the DPMFoam and MPPICFoam tutorials to see how the case is set up.

The moving mesh set up will be defined in constant/dynamicMeshDict and you can investigate it using tutorials from other DyM solvers.

I suggest you to start with DPMFoam or MPPICFoam tutorials to see if it is suitable for your own case. Then, if possible, modify this tutorial to adapt it to your case and once you made it work, add the moving mesh with the DyM version of the solver.

[rahulksoni]

Thanks a lot Yann for the help. I will do as suggested and update you with developments. Two more things (a) I can see the difference between DPMFoam and MPPICFoam (collision resolution by multiohase particle in cell) over internet. But I am still not very clear about difference in them. Secondly, can I define size of particles in these solvers and if yes then how.

Just to brief, I have a cylinder of approx 2 Litre volume. Which is completely filled with fluid and large number of 5 mm balls. At last there is a stirrer inside it which rotates. Can you suggest me which solver would be best for this situation?

Thanks in advance.

[Yann]

The major difference seems to be that particle-particle interactions are directly solved in DPMFoam while they are modeled in MPPICFoam. Have a look at the initial release notes here : https://openfoam.org/release/2-3-0/dpm/

Since there is no direct resolution of particle-particle interactions in MPPICFoam, I think it is more suitable to cases with large amount of particles in very dense particle flows where direct resolution could be very time consuming.
I don't have much experience with lagrangian solvers though, so maybe somebody else can give you better advice about which solver to choose.

About the particle size, you can define it in constant/kinematicCloudProperties:

        sizeDistribution
        {
            type        normal;
            normalDistribution
            {
                expectation 100e-6;
                variance 25e-6;
                minValue 20e-6;
                maxValue 180e-6;
            }
        }

You can either use a fixed size or define a distribution law.

Generally speaking, all the particle properties are defined in the kinematicCloudProperties file. As you can see, there are a lot of sub-models and options so it can get quite complex to set up.
This is also why I was suggesting to start with tutorials and slowly increase complexity.

[rahulksoni]

So, as suggested I developed a case in an incremental manner i.e. changes step by step.
The simulation runs with RAS turbulence. It works well when there is only one particle. However, if I increase the number of particles say 100 then starts diverging. Please see the attached file for same. Can someone tell what changes do I need to do to handle the situation?

I am also attaching the case directory at https://drive.google.com/drive/folde...OJ?usp=sharing so that one can check what wrong I am doing.
PS: The case already has block and snappy meshed geometry.

[Yann]

Hello Rahul,

There is something (or probably several things) wrong with your mesh. There are cells which shouldn't be here inside the propeller tip, which is probably due to problems with your STL file (surface not closed)

How did you create your STL files?

You can use the surfaceCheck function to check your STL and get some information about potential issues with it.


Yann

[rahulksoni]

I have created blockMesh then used snappyHexMesh -overwrite followed by renumberMesh -overwrite to create the mesh. Surely, I will check the water tightness of stl and let you know.