Parallel Resolved Particle Questions

[jcs]

Good afternoon,

I have a few questions related to resolved decomposed particle modelling, I’d be very grateful for the community’s advice.

– Is it possible to have periodic boundaries with resolved decomposed particles? I add particles with this command:
creators::addResolvedSphere3D(particleSystem,center,radius,eps,rho,v,{true,true,false});
but my particles disappear at the x and y boundaries, rather than behaving periodically.

– I am initialising resolved particles at random locations next to my inlet, if the location is along the dividing line of two decomposed regions, the particle disappears. Is there a “parallel” method of initialising particles that I am not aware of, or could I ensure a single cuboid spans the entire inlet and therefore the new particles never span two cuboids?

– Due to the material properties, geometry, and flow rates for my models, I run into a lot of stability issues. Is there likely to be an implementation of resolved particles with a TRT (or MRT) collision operator in the future, or failing that, how big a task would it be to implement? I am a reasonably confident programmer, but new to C++

– Size – is there a recommended minimum size for resolved particles in relation to voxel size? And the same for subgrid particles. I understand that subgrid particles should be exactly that – smaller than the grid, but is there some guidance in terms of fraction of the size of a voxel they should be?

Kind regards,
James

[Christoph]

Dear James.

Did you define your particle system as SuperParticleSystem or just as ParticleSystem? This should solve your problem with creating particles on the boundary between blocks. The periodic boundary problem might also be affected by this, as the periodic boundary also requires communication between blocks.

Introducing a different collision operator is quite easy in OpenLB. These are defined in src/dynamics/collision.h, just check the other implementations for guidance. However, as the dynamics are cse-optimized, you will have to either remove the cse optimization or run it again after your implementation. The collsion operator is then used in porousBGKdynamics.h .

Regarding size of particles, a resolved particle should have a diameter of at least 3 cells.

Best regards,
Christoph

[Author]

Posts