In a steady state run it is simply the longest path length (in terms of time) that the solver will integrate.

Usually you use the default value (or put a guess in) and if the particle tracks are terminated early you make it longer.

Thanks a lot Glenn. That clears it up and the issue with early termination of the particle tracks in my steady state run is resolved now.

One more question, would a transient run require the specification of this parameter? Or would the particles be tracked at each timestep until the specfied time(as would be the logical expectation)?

I have not done transient particle tracking for a long time and am not familiar with the available solver options.

cfx particle tracking
 

Dear Ghorrock,

I am running steady state simulation with particle tracking of a free impeller. The problem is I am finding different efficiency for same conditions when I run with or without one-way coupling particle tracking.

The difference I found is due to the definition of the boundary in cfx pre.
The difference is only due to the "Moving (hub+ shroud) and Blade" boundary (Attached figure name: 11) being segregated in this case. When they are segregated I put restitution coeff. in moving boundary as 1. So, the particle (if collides) bounces back. And for the blade i put rest. coeff. 0 to select it as outlet region in the post process.

If I do not segregate moving and blade domain, i cannot put the restitution coeff. value in blade and that makes me unable to select blade as the outlet boundary in post.

The difference in efficiency is due to the "torque" as i have to add?? (since the blade lies inside the "moving" [hub+shroud] boundary in my view) the torque by both moving (hub+shroud) and blade domain. Previously, for higher efficiency I combined blade+hub+shroud as "Moving boundary"(figure number 12). This gives me a different torque with higher efficiency. But I can only put the restitution coefficient in moving boundary as a whole (not in the blade) in this case. That again prevents me from selecting blade as outlet region in post since the restitution coeff. is supplied for whole moving boundary (hub+shroud+blade). Is there a way to solve this problem? i hope you understand why I had to change the setup for creating unique boundary for blade. Is there another way to do this that allows me to track particle hitting the blade?

Please post your CCL file, and an image of your geometry so we know what boundary is what face.

cfx particle tracking
 

Dear Ghorrocks,

Thank you for the reply. I have uploaded the ccl file in zip and the images. The images are for the cases where moving (blade+shroud) and blade domain is separated. This is the case where I am getting not confirming efficiency because I did not use (moving+blade) domain as a whole. The boundary faces are indicated in the name while the arrow will point out the inlet and outlet.

Some comments:
* You have alternate rotation model on. Are you sure you need it here? I suspect not.
* You have 0 reference pressure and 1 MPa outlet pressure. This is asking for trouble from round off errors. You should use 1 MPa reference pressure and 0 outlet pressure.
* You don't seem to be using a turbulent dispersion force. Note this force already includes factors to control where it applies - see the documentation for details.

What do you mean by making the blade an outlet boundary (eg "select blade as the outlet boundary in post")? Isn't it a wall?

On top of that:
You are injecting 2925.6 [kg s^-1] of sand. This is approximately 1 m3/s. In only a single segment of your pump.....
Is this correct? Looking at the size of the ruler in Pre, I don't think so.
If you think it is correct, then what is your volume fraction and is one-way-coupling the appropriate model? Is Lagrangian Particle tracking the correct model?

Thanking you all
 

Dear Gert and Ghorrocks,

It took some time but your suggestions worked out. Thank you for making this community great again :)

Hello all, i have gone through the comments in this forum. I have a doubt, is it possible to have number of positions greater than the number of real particles?. Does number of positions mean particle per second or particles per timestep?. My doubts are with respect to transient lagrangian particle tracking.

Thanks and regards
Ganesh

To be sure I would just try it. It is always best to work these things out for yourself.

But I don't think there is any problem with having more positions than real particles. Keep in mind the population of modelled particles are intended to model the population of real particles, so that means that if you have more positions than real particles, but your positions correctly reproduce the distribution of real particles then your model should be fine.

Hello Glen,
Thank you for your time and information.