Learning Forum

[raju.chowdhury]

Hello,

I want to set slip boundary conditions at the gas-liquid interface in my DPM model. Can you suggest to me how I can do that?

 

Also, my job is terminating and getting the following message while running it in HPC.

"  iter  continuity  x-velocity  y-velocity  z-velocity     time/iter
!29253 solution is converged
 29253  7.8724e-04  1.0817e-06  1.9034e-06  1.1926e-06  0:00:07   28

Advancing DPM injections ....
number tracked = 90, escaped = 1

Writing "bubble_par-0.010.dat"...

Done.

Open existing project file for writing: bubble_particle.flprj

Error at host: 
failed to open file

===============Message from the Cortex Process================================

Compute processes interrupted. Processing can be resumed.

==============================================================================

Error: 
failed to open file

Error Object: #f

"

This is only happening when dpm model is on. Without dpm model, the job is running nicely. I believe I need to add few commands in my journal script. But I do not have any idea what I need to add when the dpm model is ON.

Any useful suggestion from anyone will be highly appreciated.

[Rob]

The DPM model doesn't see the free surface as such, it sees the cell fluid properties. So, unless you want to start some (very complex) UDF coding your particles will behave exactly as the flow in that cell forces them too. 

The flpj file is a way for Fluent to keep data files together for a transient run. Can you check disc/file access permissions as it should work fine. 

[raju.chowdhury]

How can I check disc/file access permissions? 

[Rob]

Rename the flpj file and try to run again. Where are you saving the files to? Ie your own PC, random location on a network etc. 

[raju.chowdhury]

I followed your instruction but still not working. Please note that my other simulations are working fine. Only this simulation where I am using dpm model, shows that error.

[Rob]

What else are you triggering at the 10th iteration/timestep? DPM on it's own shouldn't cause problems, and the error points towards something disc related. 

[raju.chowdhury]

Thanks, Rob. The previous problem of crushing the job in HPC has been identified and resolved.

I have the following queries:

(1) I am injecting particles from a single point. I expect to release the particle in one line (the blue line marked in the below image). Why the particles are distributed throughout the entire domain over time? How can I fix the issue? 

(2) Is there any way to implement when a particle reach (or touch) the gas-liquid interface, the particle will either trap or slip?

Thanks.

[Rob]

As I've not seen the injection, and don't know how the flow is behaving I can't comment on why you're not getting a single stream of particles. 

If you want to trap particles at the free surface you'd need to write a UDF to terminate particles at they hit the free surface or (easier to set up but potentially more computationally expensive) look at DPM-to-VOF. 

[raju.chowdhury]

It is just a bubble rising due to buoyancy in water.

[Rob]

And what are the DPM particles? 

[raju.chowdhury]

Particles with a density 2500 kg/m3.

[Rob]

Size?

[raju.chowdhury]

68 micrometer

[Rob]

Maybe add some images of the flow field (vectors), contours of phase etc.  Tea leaves only get me so far.... 

[raju.chowdhury]

[Rob]

What's the settling velocity for the particles? Flow isn't all "down" and depending on 3d effects you could have enough disturbance in the flow to explain the above. What is the particle velocity? 

[raju.chowdhury]

-0.014 m/s

[Rob]

And if you plot the vertical velocity component on the plane, how much is going up sufficiently quickly to cause particles to rise? Clipping the range is often a useful tool to see what's going on. 

[raju.chowdhury]

I have found the water velocity is higher at some positions compared to the particle's velocity. That could be one reason to float the particles instead of going down.