Boiling in Granular material using MD simulation

Submitted by hamid350 on Mon, 03/12/2012 - 19:11

Dear LIGGGHTS users,
I am very new to LIGGGHTS. I want to use this code for Hybrid modeling of boiling and evaporation of Liquid Ar in Microporous structures. My porous structure is a packed bed with liquid Ar inside it. The size of particles are different and they are spherical. I should use continuum model for solid particles and pure Molecular Simulation for liquid and vapor phase of Ar around the particles. The main reason that I am going to use LIGGGHTS for my problem is that I know that this code can calculate the conduction heat transfer between fixed/moving particles. Therefore, the computational cost of simulation will become very low because It is not necessary to use MD inside the particles instead I use continuum model for particles. I have some questions, Would you please answer them?

1- Can I simultaneously use fixed granular particle inside the computational domain and atoms of liquid Ar around the particles? Is it possible to use two different atom_style in the simulation, i.e, atomic style for MD region and granular style for particles?

2- Is LIGGGHTS able to solve my problem? Please note that I have both vapor and liquid in MD region.

3- My problem is phase change problem with mass transfer, Is LIGGGHTS stable for such problems?

4- In LIGGGHTS can I apply constant heat flux at bottom of computational domain as well as at some parts of porous matrix?

5- Can I use the capability of LAMMPS in LIGGGHTS?

Thank you very much for your helps in advance,
Hamid

Forums:

Philippe's picture

Philippe | Tue, 03/13/2012 - 09:11

Hello Hamid,

while modelling of granular materials and Ar atoms might be possible using LIGGGHTS, you will very likely not be able to resolve both scales with an atomistic description as the Ar atoms are several orders of magnitude smaller than the particles that form your packed bed. For such a simulation, I would recommend that you resolve the liquid and vapour with a continuum method and simulate your bed with LIGGGHTS, very probably CFDEM is able to do this.

concerning 5: yes, LAMMPS features can be used in LIGGGHTS, only sometimes the commands are named differently.

regards
Philippe

yangdm | Tue, 03/13/2012 - 15:00

This looks very interesting.
Do you mean that atomic interaction and granular contact are basically compatible in one model in LIGGGHTS, and the only problem is the requirement of super compuatational capability for the atomic part of the model? I was just wondering if the granular particle number is much larger than that of atoms, then this hybrid MD-granular model would become more feasible, e.g., one thousand atoms and one million granular particles?

Philippe's picture

Philippe | Tue, 03/13/2012 - 15:40

To perform such simulations, you would have to write your own interactions between - say - granular particles and Lennard-Jones atoms. Also, things like neighbor list build may be very inefficient for large size differences of atoms. And the movement of granular particles will take place on totally different scales, so you might run into numerical troubles with timesteps, force magnitudes etc. I can imagine that your plans are feasible, but they are very ambitious and will require a lot of work.

hamid350 | Tue, 03/13/2012 - 18:41

Hello,
Thank you very much for your responses to my questions.
The main scope of my research is to capture the exact location and shape of liquid vapor interface using pure MD simulation. Therefore, I cannot use continuum model for liquid and vapor phases. I can use continuum model just for core of nanoparticles but for Ar atoms and a layer of atoms on surface of each solid particles I should use MD simulation. I think there is another way to solve this problem but I have some other questions about it, Would you please answer them?
I am not familiar with CFDEM. Can I use LIGGGHTS for MD region (liquid Ar and a layer of atoms on surface of each particles) and CFDEM for continuum modeling of core of particles? As you indicated before, LAMMPS features can be used in LIGGGHTS, Therefore, modeling MD region is possible in LIGGGHTS. But I am not sure about the continuum modeling in CFDDEM.
1- Is the coupling method used in CFDDEM just useful for coupling granular models and CFD? can I couple atomic models with CFD using CFDDEM?
2- Can the continuum part of CFDEM solve the 3D heat conduction in core of nanoparticles?
If the answer of above question be positive, I can easily couple MD and continuum region in LIGGGHTS. The advantage of this approach is that the coupling between MD region and Continuum model (core of nanoparticles) is implemented in LIGGGHTS and CFDEM. Is this approach suitable?
Thanks again for your helps in advance,
Hamid

ckloss's picture

ckloss | Tue, 03/13/2012 - 23:29

Hi Hamid,

I am no expert in MD so I am not the best person to comment on if the approach is physically feasible but I would at least be cautious - I suggest you talk to your supervisor.

What I can say is that what you intend to do will be quite some work from the programming point of view.

>>can I couple atomic models with CFD using CFDDEM?
yes, but not out-of-the-box, you will have to code

>>2- Can the continuum part of CFDEM solve the 3D heat conduction in core of nanoparticles?
that's not a technical question, but a physical one - again I suggest you talk to your supervisor.

>>1- Can I simultaneously use fixed granular particle inside the computational domain and atoms of liquid Ar around the particles?
Yes - see atom_style hybrid

>>2- Is LIGGGHTS able to solve my problem?
see above

>>3- My problem is phase change problem with mass transfer, Is LIGGGHTS stable for such problems?
don't know what you refer to with that

>>4- In LIGGGHTS can I apply constant heat flux at bottom of computational domain
yes, see fix mesh/gran doc

>>as well as at some parts of porous matrix?
yes

Cheers, Christoph