PFV with clumps?

Asked by Christian Jakob on 2014-07-23

Hi,

I am thinking about a simulation with clumps and PFV method. I know clumps are not compatible with PFV, yet.
I simulate a sand at microscale and want to have a more realistic grain shape using clumps (maybe 2 or more overlapping members). I see there is the first problem in triangulation algorithm. But it should be possible to use clump center and equivalent radius instead of sphere center and sphere radius for triangulation. I think this is ok as long as clumps are nearly spherical, right?
The next problem is the calculation of fluid volume inside the cells. But isn't it possible to use clump centers and equivalent radius at this stage as well? I know the volumes will not be exactly, but I could live with a small error as long as calculation stays stable...

What do you think about this? Did I miss something, that makes it completely impossible?

Regards,

Christian

Question information

Language:
English Edit question
Status:
Solved
For:
Yade Edit question
Assignee:
No assignee Edit question
Solved by:
Bruno Chareyre
Solved:
2014-07-24
Last query:
2014-07-24
Last reply:
2014-07-23

Hi Christian.
Your suggestion is possible. It needs to skip clump members when
building the triangulation, and insert clumps as spheres. I don't
foresee big difficulties to implement that.

Another possibilty is the opposite, skip clumps and build the
triangulation only with clump members (precondition: clump members don't
overlap too much). This is almost what would happen now without changing
the code too much (it would only need to explicitely skip clumps in the
triangulation, +1 line of code). Consequence of this: there will be
fluid flow through the clumps... Not good.
However, the tetrahedra inside one clump could be just excluded from the
calculation. The fluid would be computed only in between the clumps. It
would give probably more fluid DOFs than your suggestion and it needs
more implementation. The benefit would be a more realistic model of the
pore space geometry.

Cheers.
Bruno

Christian Jakob (jakob-ifgt) said : #2

Thanks Bruno Chareyre, that solved my question.