particle time-step in DPMFoam

Shuai_W · October 27, 2015, 23:14

Hi,
I use DPMFoam to simulate gas-solid flow in a CFB riser, particle number is 201000 with diameter 185um, density 2400kg/m3. The wide-depth-height of the geometry is 32mm-1.2mm-300mm with grids of 64-3-600. However, I have some questions about the solver DPMFoam.
1. How to determine particle time-step in DPMFoam, I can't find this parameter.
2. In my work, I use 16 processors and find the simulation so slow. I want to know if anyone who utilized DPMFoam to simulate gas-solid flow with or more than 200000 particles?
3. The difference of detection algorithm of particle collisions between in DPMFoam and in CFDEM? Which is the better?
Thank you!

Wang

jarpve · November 5, 2015, 17:58

I was wondering that somebody was working with DPMFoam right now. I have the same doubts and indeed I would like to know if you know which method is implemented in DPMFoam for tracking particles. is it the lagrangian Particle Source in Cell method (PSC)? or the MP-PSC? or an eulerian approach?.

Thank you.

ansubru · November 6, 2015, 08:52

Hi jarpve!!

From what i know, DPMFoam using lagrangian particle source in cell method.. however its not that straightforward as not all of the 'particulate phase forces' can be directly added as a source in-cell.. what i mean to say is that both drag and buoyancy are dependent on velocity and hence cannot be treated a complete source... for more info read this..
http://www.openfoam.org/mantisbt/fil...=1030&type=bug

Regards
ansubru

ansubru · November 6, 2015, 08:59

Hi Wang!!

I am currently simulating around 150000 particles in my case study (an extension of a fluidized bed) and the case is quite fast... I havent paralellized my application as i use a self modified version of the DPMFOam solver (with capability to handle vibrating/moving walls) which has not been tuned to tackle MPI right now.. However in serial mode, i can simulate upto 5-6 secs (in 10 hrs)... Not the most efficient, but it works

... In my opinion DPMFoam can be tuned and altered much more than DEM solution's software.. If your simulation is slow, maybe your paralellization is'nt great.. sometimes its better to run the code in serial mode first to see if there are some 'numerical instabilities' leading to longer computational times,

Regards
ansubru

Shuai_W · November 6, 2015, 09:03

Hi jarpve,
I validated the DPMFoam with experiments by Muller et al, the investigated object is a bubbling fluidized bed with 44mm width, 200mm heights and 10mm depth. The numerical results from DPMFoam showed good agreements with the experimental data. Collisions between particles are modeled used soft-sphere model, which is different from MP-PIC in software Barracuda (CPFD). I think DPMFoam is DEM method that likes CFDEM.

Regards
Wang

Shuai_W · November 6, 2015, 09:13

Hi ansubru,
Frankly speaking, the parallelization performance of DPMFoam is not as good as CFDEM, I have tested it. In DPMFoam, particle number < 300000, grid number < 200000, I think it can be used as an alternative to other DEM solution’s software. However, when particle number increases, I don’t think DPMFoam is a good choice (Using same processors).
In fact, I want to combine DPMFoam with coalChemicalFoam to simulate coal/biomass gasification or combustion in fluidized beds, where particle collisions are modeled by DEM, I am doing this work just now.

Best regards
Wang

jarpve · December 29, 2015, 11:52

Quote:

Originally Posted by ansubru

Hi jarpve!!

From what i know, DPMFoam using lagrangian particle source in cell method.. however its not that straightforward as not all of the 'particulate phase forces' can be directly added as a source in-cell.. what i mean to say is that both drag and buoyancy are dependent on velocity and hence cannot be treated a complete source... for more info read this..
http://www.openfoam.org/mantisbt/fil...=1030&type=bug

Regards
ansubru

Thank you.

Quote:

Originally Posted by Shuai_W

Hi jarpve,
I validated the DPMFoam with experiments by Muller et al, the investigated object is a bubbling fluidized bed with 44mm width, 200mm heights and 10mm depth. The numerical results from DPMFoam showed good agreements with the experimental data. Collisions between particles are modeled used soft-sphere model, which is different from MP-PIC in software Barracuda (CPFD). I think DPMFoam is DEM method that likes CFDEM.

Regards
Wang

Thank you Wang

Shuai_W · December 29, 2015, 20:01

Dear jarpve,
In recent days, I will develop a solver to simulate coal/wood gasification/combustion based on a CFD-DEM coupling approach (i.e. four-way coupling method), this work is based on coalChemistryFoam and DPMFoam, if you are interested in it, more information in this thread:
http://www.cfd-online.com/Forums/ope...tml#post577524

harmonica · February 14, 2018, 09:33

Hi ansubru
Do you have the rest of MPPICFoam tutorial?Can you give me a link?

thank you !

minzhang · May 24, 2018, 11:28

Quote:

Originally Posted by Shuai_W

Hi,
I use DPMFoam to simulate gas-solid flow in a CFB riser, particle number is 201000 with diameter 185um, density 2400kg/m3. The wide-depth-height of the geometry is 32mm-1.2mm-300mm with grids of 64-3-600. However, I have some questions about the solver DPMFoam.
1. How to determine particle time-step in DPMFoam, I can't find this parameter.
2. In my work, I use 16 processors and find the simulation so slow. I want to know if anyone who utilized DPMFoam to simulate gas-solid flow with or more than 200000 particles?
3. The difference of detection algorithm of particle collisions between in DPMFoam and in CFDEM? Which is the better?
Thank you!

Wang

Hello Shuai,

I am wondering whether you could give me some reference (more details) about the detection algorithm of particle collisions. I want to know the difference between CFDEM and DPMFoam.

minzhang · May 24, 2018, 11:32

Quote:

Originally Posted by harmonica

Hi ansubru
Do you have the rest of MPPICFoam tutorial?Can you give me a link?

thank you !

Hi,

Have you got the tutorial already? If you still need it, I could send to your email.

minzhang · May 24, 2018, 11:49

Quote:

Originally Posted by Shuai_W

Dear jarpve,
In recent days, I will develop a solver to simulate coal/wood gasification/combustion based on a CFD-DEM coupling approach (i.e. four-way coupling method), this work is based on coalChemistryFoam and DPMFoam, if you are interested in it, more information in this thread:
http://www.cfd-online.com/Forums/ope...tml#post577524

Hello Shuai,

This is Min. I think you are an expert on DPMFoam now.

I am wondering whether you have some ideas about the difference between CFDEM and DPMFoam now?

Thanks and best regards,
Min

October 27, 2015, 23:14	particle time-step in DPMFoam	#1
Shuai_W New Member Shuai Wang Join Date: Mar 2014 Posts: 26 Rep Power: 12	Hi, I use DPMFoam to simulate gas-solid flow in a CFB riser, particle number is 201000 with diameter 185um, density 2400kg/m3. The wide-depth-height of the geometry is 32mm-1.2mm-300mm with grids of 64-3-600. However, I have some questions about the solver DPMFoam. 1. How to determine particle time-step in DPMFoam, I can't find this parameter. 2. In my work, I use 16 processors and find the simulation so slow. I want to know if anyone who utilized DPMFoam to simulate gas-solid flow with or more than 200000 particles? 3. The difference of detection algorithm of particle collisions between in DPMFoam and in CFDEM? Which is the better? Thank you! Wang

November 6, 2015, 09:03	DPMFoam DEM	#5
Shuai_W New Member Shuai Wang Join Date: Mar 2014 Posts: 26 Rep Power: 12	Hi jarpve, I validated the DPMFoam with experiments by Muller et al, the investigated object is a bubbling fluidized bed with 44mm width, 200mm heights and 10mm depth. The numerical results from DPMFoam showed good agreements with the experimental data. Collisions between particles are modeled used soft-sphere model, which is different from MP-PIC in software Barracuda (CPFD). I think DPMFoam is DEM method that likes CFDEM. Regards Wang jarpve likes this.

November 6, 2015, 09:13	parallelization performance of DPMFoam	#6
Shuai_W New Member Shuai Wang Join Date: Mar 2014 Posts: 26 Rep Power: 12	Hi ansubru, Frankly speaking, the parallelization performance of DPMFoam is not as good as CFDEM, I have tested it. In DPMFoam, particle number < 300000, grid number < 200000, I think it can be used as an alternative to other DEM solution’s software. However, when particle number increases, I don’t think DPMFoam is a good choice (Using same processors). In fact, I want to combine DPMFoam with coalChemicalFoam to simulate coal/biomass gasification or combustion in fluidized beds, where particle collisions are modeled by DEM, I am doing this work just now. Best regards Wang

December 29, 2015, 20:01	four-way coupling method	#8
Shuai_W New Member Shuai Wang Join Date: Mar 2014 Posts: 26 Rep Power: 12	Dear jarpve, In recent days, I will develop a solver to simulate coal/wood gasification/combustion based on a CFD-DEM coupling approach (i.e. four-way coupling method), this work is based on coalChemistryFoam and DPMFoam, if you are interested in it, more information in this thread: http://www.cfd-online.com/Forums/ope...tml#post577524

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Floating point exception error	lpz_michele	OpenFOAM Running, Solving & CFD	53	October 19, 2015 02:50
time step directories naming issue	Andrea_85	OpenFOAM	3	April 3, 2014 08:38
InterFoam negative alpha	karasa03	OpenFOAM	7	December 12, 2013 03:41
Micro Scale Pore, icoFoam	gooya_kabir	OpenFOAM Running, Solving & CFD	2	November 2, 2013 13:58
Upgraded from Karmic Koala 9.10 to Lucid Lynx10.04.3	bookie56	OpenFOAM Installation	8	August 13, 2011 04:03

November 5, 2015, 17:58		#2
jarpve New Member Jesus Ramirez Join Date: Aug 2014 Location: Colombia Posts: 13 Rep Power: 11	I was wondering that somebody was working with DPMFoam right now. I have the same doubts and indeed I would like to know if you know which method is implemented in DPMFoam for tracking particles. is it the lagrangian Particle Source in Cell method (PSC)? or the MP-PSC? or an eulerian approach?. Thank you.

November 6, 2015, 08:52		#3
ansubru Member Ananda Kannan Join Date: Feb 2014 Location: Göteborg, Sweden Posts: 55 Rep Power: 12	Hi jarpve!! From what i know, DPMFoam using lagrangian particle source in cell method.. however its not that straightforward as not all of the 'particulate phase forces' can be directly added as a source in-cell.. what i mean to say is that both drag and buoyancy are dependent on velocity and hence cannot be treated a complete source... for more info read this.. http://www.openfoam.org/mantisbt/fil...=1030&type=bug Regards ansubru

November 6, 2015, 08:59		#4
ansubru Member Ananda Kannan Join Date: Feb 2014 Location: Göteborg, Sweden Posts: 55 Rep Power: 12	Hi Wang!! I am currently simulating around 150000 particles in my case study (an extension of a fluidized bed) and the case is quite fast... I havent paralellized my application as i use a self modified version of the DPMFOam solver (with capability to handle vibrating/moving walls) which has not been tuned to tackle MPI right now.. However in serial mode, i can simulate upto 5-6 secs (in 10 hrs)... Not the most efficient, but it works ... In my opinion DPMFoam can be tuned and altered much more than DEM solution's software.. If your simulation is slow, maybe your paralellization is'nt great.. sometimes its better to run the code in serial mode first to see if there are some 'numerical instabilities' leading to longer computational times, Regards ansubru

February 14, 2018, 09:33		#9
harmonica New Member peter Li Join Date: Feb 2018 Posts: 1 Rep Power: 0	Hi ansubru Do you have the rest of MPPICFoam tutorial?Can you give me a link? thank you !