[lammps-users] Is LAMMPS fitted for pebble bed reactors?


I am a PhD student at UC Berkeley, and my project involves computing the motion and recirculation of a large number of spheres (pebbles) within a defined geometry, which is a nuclear reactor core, called a Pebble Bed Reactor.

The way the motion works is that graphite spheres (3-6 cm diameter, 100,000-500,000 of them) are inserted to load the core with a given packing fraction from the bottom of the geometry. They are buoyant and are also pushed up by a coolant flow. Once the core contains the correct packing fraction of spheres, we start the circulation. Basically, spheres are extracted with a constant extraction rate from the top, and other spheres are inserted at the same time from the bottom to keep a constant number in the core.

Therefore, I am wondering first if work on pebble bed reactors (models, parameters, etc.) is available somewhere. Then, if the following features are available with LAMMPS:

  1. Spheres insertion from a surface, or within a volume. In particular, is overlap between inserted spheres avoided?

  2. Drag/buoyancy in addition to spheres collision. I know this should work, but please let me know if it is more challenging than it seems.

  3. Spheres extraction with a constant rate. If I can just open the top surface, how to make the insertion match the extraction and control these two rates (variable step length?). Otherwise, can I extract (delete) some of the top spheres to match the desired rate?

  4. Slow physics: pebbles move by tens of centimeters only per day. I would like to output the geometry every several days. It means that the DEM must be very efficient in some way, but also accurately replicate the physics.

Help on this would allow me to go forward with motion and make an incredible step in this field. Up to now, I have been using the OpenFOAM DEM, but this tool is very limited.

Thank you in advance for your help.

Best regards,
Yves Robert

First off, I am not the most competent person to discuss DEM capabilities in LAMMPS, but your description sounds to me like you need a CFD - DEM coupling.
That doesn’t exist in LAMMPS directly, but you may search around and perhaps find some projects that have done that. I recall people on the mailing list mentioning it and also reporting results in talks at past LAMMPS workshops.

The most flexible DEM feature in LAMMPS is pair style granular. You should study its documentation to get a sense of what is possible.

I should probably also point out the LIGGGHTS software that was forked from an older version of LAMMPS several years ago that focuses on DEM and has many more features. They also have a CFD-DEM coupling with OpenFOAM. https://www.lammps.org/external.html#liggghts


Hello Dr. Kohlmeyer, and thank you for your reply.

Long-term calculations would need some CFD coupling indeed, as the coolant flow might have a non-negligible effect on the motion.
I will then add LIGGGHTS to the list of possible tools for the future.

However, for now, I will focus on “one-way coupling”, meaning that the flow can be accounted as a vertical drag force on the spheres.
In this context, I believe LAMMPS is able to do that. I will look at the “pair style granular” documentation.
In the meantime, if anybody has more details about my needs and how to fulfill them, please let me know.

Best regards,
Yves Robert