DPM convergence and particle size

sirsammie72 · April 23, 2020, 11:55

New to CFD:

I recently ran a simulation with 1um particles and converged fine. When I increased to 5 and 10 um the simulation no longer converges. I am wondering if Eulerian time steps should be changed with particle size, or DPM time step settings?

Or any thoughts on what is causing this?

I am using LES and DPM steady particle tracking.

Thank you

vinerm · April 24, 2020, 09:48

Since you are using LES, what is the mesh size that you have? 5 micron is not a very big size. What are the volume fraction and mass fraction values for the particles in the domain?

sirsammie72 · April 24, 2020, 11:57

Mesh size is 0.004m, not yet optimized or confirmed for independence. I think it may have to do with particle drag, and must switch to unsteady tracking and give separate particle time step

vinerm · April 24, 2020, 12:04

What is the Re that you are trying to resolve? 4 mm is too large for LES. If there is no specific requirement to use LES, use RANS, i.e., $k-\varepsilon$ or $k-\omega$ . LES would require much finer mesh if you want to resolve, say, at least 92-95% of energy containing eddies. Determine your eddy length scale and then compare it against your mesh size.

Anshs · May 13, 2020, 03:45

Quote:

Originally Posted by vinerm

What is the Re that you are trying to resolve? 4 mm is too large for LES. If there is no specific requirement to use LES, use RANS, i.e., $k-\varepsilon$ or $k-\omega$ . LES would require much finer mesh if you want to resolve, say, at least 92-95% of energy containing eddies. Determine your eddy length scale and then compare it against your mesh size.

But DPM requires mesh size to be greater than particle size.

And how to determine the eddy length scale?

vinerm · May 14, 2020, 08:58

Yes, it is required that particles should not occupy more than 10% of the volume, locally as well as globally for DPM to be valid. However, LES has its own mesh requirement. So, if your mesh is not fine enough, use RANS. If it is fine enough for LES but large enough for DPM, then it should not pose any challenge at all. Secondly, determine Stokes number value for particles that you are using. You may not even require two-way coupling for particles.

For length scales, you can have a look at

https://en.wikipedia.org/wiki/Taylor_microscale

April 23, 2020, 11:55	DPM convergence and particle size	#1
sirsammie72 New Member CFDSim72 Join Date: Apr 2020 Posts: 17 Rep Power: 6	New to CFD: I recently ran a simulation with 1um particles and converged fine. When I increased to 5 and 10 um the simulation no longer converges. I am wondering if Eulerian time steps should be changed with particle size, or DPM time step settings? Or any thoughts on what is causing this? I am using LES and DPM steady particle tracking. Thank you

April 24, 2020, 09:48	Particle Size	#2
vinerm Senior Member Vinerm Join Date: Jun 2009 Location: Nederland Posts: 2,946 Blog Entries: 1 Rep Power: 36	Since you are using LES, what is the mesh size that you have? 5 micron is not a very big size. What are the volume fraction and mass fraction values for the particles in the domain? __________________ Regards, Vinerm PM to be used if and only if you do not want something to be shared publicly. PM is considered to be of the least priority.

April 24, 2020, 12:04	Mesh Size	#4
vinerm Senior Member Vinerm Join Date: Jun 2009 Location: Nederland Posts: 2,946 Blog Entries: 1 Rep Power: 36	What is the Re that you are trying to resolve? 4 mm is too large for LES. If there is no specific requirement to use LES, use RANS, i.e., $k-\varepsilon$ or $k-\omega$ . LES would require much finer mesh if you want to resolve, say, at least 92-95% of energy containing eddies. Determine your eddy length scale and then compare it against your mesh size. __________________ Regards, Vinerm PM to be used if and only if you do not want something to be shared publicly. PM is considered to be of the least priority.

May 14, 2020, 08:58	DPM and Eddy Length Scale	#6
vinerm Senior Member Vinerm Join Date: Jun 2009 Location: Nederland Posts: 2,946 Blog Entries: 1 Rep Power: 36	Yes, it is required that particles should not occupy more than 10% of the volume, locally as well as globally for DPM to be valid. However, LES has its own mesh requirement. So, if your mesh is not fine enough, use RANS. If it is fine enough for LES but large enough for DPM, then it should not pose any challenge at all. Secondly, determine Stokes number value for particles that you are using. You may not even require two-way coupling for particles. For length scales, you can have a look at https://en.wikipedia.org/wiki/Taylor_microscale __________________ Regards, Vinerm PM to be used if and only if you do not want something to be shared publicly. PM is considered to be of the least priority.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
particles leave domain	Steffen595	CFX	9	March 7, 2016 17:19
Check particle impaction with User Fortran	Julian K.	CFX	3	January 12, 2012 10:46
cfx particle convergence problem	bharath	CFX	0	November 11, 2009 09:34
Steady state DPM - Convergence and particle tracks	Sandilya Garimella	FLUENT	2	March 28, 2008 16:53
convergence in particle transport model	Jose Tinoco	CFX	0	March 3, 2003 04:10

April 24, 2020, 11:57		#3
sirsammie72 New Member CFDSim72 Join Date: Apr 2020 Posts: 17 Rep Power: 6	Mesh size is 0.004m, not yet optimized or confirmed for independence. I think it may have to do with particle drag, and must switch to unsteady tracking and give separate particle time step