[raptorishungry]

Hello Foamers!

I need help in post processing of my results from Lagrangian solver. Is there anyway to backtrack each and every particle from its final position all the way to it's inlet position? (some particle ID ?) as the cloud positions store only the positions of each particle and the cell ID associated with the positions, is there anyway I could identify the particle individually?

Please let me know if you can think of any work around. Thank you.

[guin]

I guess that you are looking for the particle property called "origId", which is unique for each particle.

[raptorishungry]

Hello Rodrigo, thank you for your quick reply. I just have a couple of more questions, please let me know your views.

1. The order in which the items appear in each file under <timestep>/lagrangian/kinematicCloud/ is the same?. For example, the first entry in "origID" and the first entry to "positions" correspond to the same particle?

2. The "active" in <timestep>/lagrangian/kinematicCloud/, has values of 0 or 1. Does 0 mean the particle is at rest?

3. Is it possible to backtrack particles between different computational runs across two geometries? For example, if a particle is travelling from one part of my geometry (simulated first) to another part (simulated after), is it possible to backtrack the particle to it's inlet in the first part geomtery?

Thank you once again!

[guin]

Originally Posted by raptorishungry

Hello Rodrigo, thank you for your quick reply. I just have a couple of more questions, please let me know your views.

1. The order in which the items appear in each file under <timestep>/lagrangian/kinematicCloud/ is the same?. For example, the first entry in "origID" and the first entry to "positions" correspond to the same particle?

2. The "active" in <timestep>/lagrangian/kinematicCloud/, has values of 0 or 1. Does 0 mean the particle is at rest?

3. Is it possible to backtrack particles between different computational runs across two geometries? For example, if a particle is travelling from one part of my geometry (simulated first) to another part (simulated after), is it possible to backtrack the particle to it's inlet in the first part geomtery?

Thank you once again!

1. In principle, yes. But not necessarily over the time . For instance, if particles are being injected or leaving the domain the nth position may or may not correspond to the same particle as in previous timestep. You can find the new position by following the "origId".

2. ... yes, on may see it this way. Take a look at https://github.com/OpenFOAM/OpenFOAM...el.C#L317-L331

3. I do not understand well the situation, but I guess that you would need to manually figure out the correspondences between the last stage of the initial simulation and the initial stage of the 2nd one... May a combined check for origId and origProc be enough for your purpose? https://cpp.openfoam.org/dev/classFoam_1_1particle.html

[raptorishungry]

Originally Posted by guin

1. In principle, yes. But not necessarily over the time . For instance, if particles are being injected or leaving the domain the nth position may or may not correspond to the same particle as in previous timestep. You can find the new position by following the "origId".

2. ... yes, on may see it this way. Take a look at https://github.com/OpenFOAM/OpenFOAM...el.C#L317-L331

3. I do not understand well the situation, but I guess that you would need to manually figure out the correspondences between the last stage of the initial simulation and the initial stage of the 2nd one... May a combined check for origId and origProc be enough for your purpose? https://cpp.openfoam.org/dev/classFoam_1_1particle.html

1. Yes, that makes sense. I thought that would be the case. Thank you so much for clarifying!

2. Thank you so much! that clarifies my concern. I tracked some of the particles with "active" status as 0, through their origId, and noticed that they had a non-zero velocity under <timestep>/lagrangian/kinematicCloud/U. Is this the velocity of the particle just before its deposition in the patch that has "stick"?

3. I apologize for being a bit vague. I have two channels that are connected to each other and aligned according the gravity vector. As it was too cumbersome to simulate them together (around 10 million elements), I split the geometry and thought of simulating the second portion after the first.

So, the same particle would have a different origId when I inject it for the second run. As you said, I can manually monitor the particle's new origId, but is it the only way to do it?. Also, the "origProcId" in every <timestep>/lagrangian/kinematicCloud/ shows the total number of particles and the number of processors handling them. How do you suggest I may use them for particle tracking?

Once again, I appreciate your help. Thank you for your inputs.

[guin]

2. Thank you so much! that clarifies my concern. I tracked some of the particles with "active" status as 0, through their origId, and noticed that they had a non-zero velocity under <timestep>/lagrangian/kinematicCloud/U. Is this the velocity of the particle just before its deposition in the patch that has "stick"?

I didn't check the code for it, but if this value does't vary with the time most probably it is.

3. I apologize for being a bit vague. I have two channels that are connected to each other and aligned according the gravity vector. As it was too cumbersome to simulate them together (around 10 million elements), I split the geometry and thought of simulating the second portion after the first.

So, the same particle would have a different origId when I inject it for the second run. As you said, I can manually monitor the particle's new origId, but is it the only way to do it?. Also, the "origProcId" in every <timestep>/lagrangian/kinematicCloud/ shows the total number of particles and the number of processors handling them. How do you suggest I may use them for particle tracking?

Hmmm... to be honest, I would have to ply a bit trial-and-error to answer this question.


Expert user's suggestion: If the answer is worth to invest a day on the seek, why don't you resemble your simulation workflow with two easy 2D-geometries (e.g. 10x10x1 cells each) and a low number of particles passing from one to the other (e.g. less than 10)? this will surely solve your questions much better than any guesswork from my side.

[raptorishungry]

Originally Posted by guin

I didn't check the code for it, but if this value does't vary with the time most probably it is.




Hmmm... to be honest, I would have to ply a bit trial-and-error to answer this question.


Expert user's suggestion: If the answer is worth to invest a day on the seek, why don't you resemble your simulation workflow with two easy 2D-geometries (e.g. 10x10x1 cells each) and a low number of particles passing from one to the other (e.g. less than 10)? this will surely solve your questions much better than any guesswork from my side.

1. Yes, I checked the code and the values of U don't vary with time for the "inactive" particles. So I assume it's the velocity just before it got deposited. Don't know why it is stored though.

2. That is better than guessing, true. I will try to run it on a simple geometry, and see if I can come up with some method to couple the particle flow between them.

Thank you very much for your help Rodrigo! I shall post my results if I'm successful