[Tobi]

Hi all,


I never investigated too much into the Lagrangian solver but it seems interesting and used in a variety of engineering problems such as DeNOx systems, separators and so on. Nevertheless, during the Covid-19 time, I am also investigating into a simple case. Until now, I only used the uncoupled solvers but here, I do have flow-field-particle coupling. Thus, I use DPMFoam. However, going through the MPPIC foam description vs. DPM foam description, the main different I see is that MPPIC uses some model for the particle collision rather than calculating them directly. I guess this is of main interest if we do have a high particle loaded flow as the collision calculation takes much computational time. Furthermore, I would say that if we model more particles in one parcel (nParticles > 1 in the kinematicCloudDict), the MPPIC is even better.

Can someone confirm my intention of the two solvers? Otherwise, it seems that the solvers are more or less similar.



However, another question. The nParticle is the number of particles in one parcel. As we do solve for parcels in foam, it is clear that setting the value of 1 represents that one parcel is one particle. However, if we say 1000, then one parcel includes 1000 particles.


Thus, in which form does it influence the system and does the following assumption holds (without digging into the code myself):

• Are these 1000 particles of same size or is there a size distribution inside a parcel?
• The weight of one parcel will probably change which interacts with other models (such as gravity etc). However, is it really similar if we would have 1000 single particles for which the gravity acts on each particle compared to 1 parcel on which acts the gravity on the accumulated weight. Probably the assumption must hold, that a lot of particles has to be inside the domain (the question might be what is "a lot", 5%, 10%, 50% ?)

Furthermore, assuming that we do have a particle distribution we want to separate while having 1000 particles in one parcel; again, does this parcel has a size distribution inside; probably not. Thus, if we have a separation of parcels which are removed from the domain at one patch (acting force might be the gravitational force), can we count the size distribution we get on these patch? If one parcel represents only one specific size class, then things would be easier to calculate. However,

• for one-size classed parcels, would we be able to calculate the amount of particles and size class that enter one specific patch, e.g. a size-distribution and amount of particles using foam libraries
• for multi-size classed parcels, would we also be able to calculate the amount of particles and their size distribution on a specific patch e.g. a size-distribution and amount of particles using foam libraries

I am also wondering about other keywords such as:

• coupled (to the flow field or one-way vs. two way --> parcels do influence the flow field)
• transient (shouldn't be the Lagragian calculation be always transient?)
• ...
• ...
  I can check them in the source code of course.

Thanks in advance,
Tobi

[liliu]

Hi, Tobi,

I am currently investigating the applications of the MPPICFoam and DPMFoam solvers, so wonder if you have had a look at the code for those questions in your thread and could share at here. Thank you.

[Taendyr]

Hi,

I'd like to simulate a mix af gas and particles. MPPIC seemed adequat to me. The only thing is that I need a compressible solver for the gas part. Does MPPIC handle this ? I didn't find the answer

[jetface]

This is a nice paper that talks about the different gas-solid modeling approaches.

I am familiar with the MP-PIC model used in OpenFOAM 5 through 8. I haven't used DPMFoam but I believe it is actually the Dense Discrete Phase Model (DDPM) since it does seem to handle collisions. MP-PIC is a newer method that describes the particles as groups (parcels) using a Particle Density Function (PDF) to improve computational efficiency. It uses different collision models because of that. This is my understanding anyways.

I'd recommend giving that paper a read. It describes things much better than I can.

And I believe both use an incompressible solver for the Eulerian portion of the model (the PIMPLE solver essentially). You may want to look at the compressible solver you want to use and compare it to the MP-PIC or DPM solver to see which will be easier to modify. I've written my own MP-PIC solver and I started with PIMPLE and added the Lagrangian part. Wasn't too difficult. So starting with the compressible solver might be the way to go.

[vonboett]

I merged DPMFoam into debrisInterMixing (interMixingFoam based debris flow solver) and interFoam to account for gravel collisions and it worked fine. I validated the code using the experiments of Fessler et al. (1999) for 1.1 % mass concentration.
As I could not find out about the questions of particle handling within parcels, I use 1 particle per parcel and the particle size distribution as defined by the material probe from the sieve curve using sizeDistribution{ type general; generalDistribution{ distribution( (sizeClass share));}}.

[Zurciram]

Hi,
You mentioned that you are familiar with compressible flows coupled with particles injection so I decide to write you. I have been trying to simulate a supersonic biphasic flow through de Laval nozzle with particle injection with a very low volume fraction without reaction. I have read and test with different solvers but my knowledge of OpenFOAM is not still enough. Could you give some advice or some tutorial guides to focus on this task because right now I feel some lost. I´d really appreciate it. Thanks in advance!.

[leeseungnan@163.com]

Originally Posted by Taendyr

Hi,

I'd like to simulate a mix af gas and particles. MPPIC seemed adequat to me. The only thing is that I need a compressible solver for the gas part. Does MPPIC handle this ? I didn't find the answer

Hi, Do you want to use MPPICFoam to couple with the compressible gas solver? Have you achieved it? Which compressible solver is used?

[leeseungnan@163.com]

Hi, Do you want to use MPPICFoam to couple with the compressible gas solver? Have you achieved it? Which compressible solver is used?

[Desimuser1]

how does MPPIC foam recognise particle smaller than the mesh size and also track particles inside the mesh size element

[Desimuser1]

i think MPPIC foam is for transient incompressible calculations.
i think you can use coal chemistry, reacting parcel, spray, spray dynamic mesh solvers for it

[Tobi]

Originally Posted by Desimuser1

how does MPPIC foam recognise particle smaller than the mesh size and also track particles inside the mesh size element

Because it is still a Lagrange solver. As far as I got it, only the collisions are calculated differently - but I am not sure about that. However, regarding the particles themself -> lagrange -> not mesh dependent.


But I never went through the theory ...

[Desimuser1]

Originally Posted by Tobi

Because it is still a Lagrange solver. As far as I got it, only the collisions are calculated differently - but I am not sure about that. However, regarding the particles themself -> lagrange -> not mesh dependent.


But I never went through the theory ...

thank you toby
been a fan

[rezaeimahdi]

The problem arises when you want to use two-way coupling.

If the particle size is big compared to the mesh size, you will get a wrong particle velocity, wrong fluid-particle forces (mostly drag) and wrong volume fraction.

These are field quantities based on the CFD mesh and are obtained from the locations, the sizes, and the velocities of each individual particles as well as the fluid forces acting on them.

So, in the literature, it mentioned that particle size should be 3-5 times smaller than mesh size. (I found even 25 times smaller in some cases)

[s.amirzadeh]

Originally Posted by rezaeimahdi

The problem arises when you want to use two-way coupling.

If the particle size is big compared to the mesh size, you will get a wrong particle velocity, wrong fluid-particle forces (mostly drag) and wrong volume fraction.

These are field quantities based on the CFD mesh and are obtained from the locations, the sizes, and the velocities of each individual particles as well as the fluid forces acting on them.

So, in the literature, it mentioned that particle size should be 3-5 times smaller than mesh size. (I found even 25 times smaller in some cases)

Hi Reza,

thanks for the information provided. I am trying to simulate a bubbly flow using discrete phase modeling. So far, I was using MPPICFoam and tonight I found out, as you said, wrong answers will be produced if particles sizes are big. mine have the same size as mesh elements! Do you know how I may be able to do such simulation (DPM with particles having same size as mesh elements) using OpenFOAM?