How to select a proper model to model breakage of brittle material

Hi Xavier,

I am also working on the modelling of breakage of rock, and have been investigating that question for a bit of time.

I have found that CohFrictMat and JCFpmMat are quite similar, meaning they have the same expression of contact stiffnesses, however JCFpmMat has a few built in functions that come in handy :
the "equilibrium distance" is automatically set when creating a pack of spheres, while for the moment you still have to set it manually when you write a script using CohFrict Mat, although I think someone might be implementing that precise feature.

With JCFpmMat the number of shear-broken or tensile broken bonds is tracked automatically, so that's quite cool as well, because you can easily track bond failure. Don't think such a function is implemented in CohFrictMat.

However it is not clear to me how you handle plastic behavior with JCFpm, while it is a bit more clearer with CohFrictMat.

Oh, and in JCFpm you can introduce weak planes, that could represent foliated plans in a rock for example, depends which kind of rock you want to model, but it seems like a good way to introduce macro-scale defect, and maybe microscale.

CpmMat is a bit different, because it was designed for concrete material. The parameters used to describe the model are eps and sigma limit, whereas in the two other models you work with shearCohesion and tensileCohesion In the examples on CPMMat you can find a uniaxial test on a sample that is a good starting point for some calibration model.

As far as I know, none of those contact law include viscous damping though....

Hope that was of any help,
Cheers

Hi loiseaurare,

Thanks for your reply. I have tried them, and found that CohFrictMat got the best simulation results. But I have no idea on selecting some parameters such as alphaKr and alphaKtw.

Besides, some sub-particles overlapping with their adjacent particles, I wonder whether a 'clump' can be used to simulate breakage and impact crushing.

Xavier

Hello Xavier,

Really ? I have had so many troubles with the CohFrictMat, I prefer the JCFPMmat, I find it a lot more stable. What kind of simulation have you implemented ?

I think, but I'm not an expert here, that those parameter are related to bending and twisting moments, I guess they depend on the type of material you are using, and how you would like to describe interactions of particles with one another. So I think it also depends on the scale of your particles.

So, a clump is unbreakable, so I would say it is not the best to model crushing or breakage^^ You could however use clumps to model irregularly shaped particles. I think I saw another thread on the launchpad of a guy modelling crushing of particles with agglomerates, can't find it right now, but you should try and have look in the topics !

Thanks loiseaurare,

I mean 'Clump' is a packing with a small overlap between adjacent particles, because any packing using randomDensePack or regularHexa always got many voids. I tried a method posted in this forum, that is radius expansion method. I got a satisfied packing with this approach.
Besides, I am working on the effect of particle shape on crushing, thus many real rock shape will be used in my simulations.

Tri-axial compression only compress a cubic packing. Can you tell me how did you got a packing.

Thanks,

Xavier

Hi Xavier,

In [1], you should find a way of making packings with differents shapes, if I understood well what you are after ?
I think the main idea is, you've got you're randomDensePack, and comparing it with the predicates it return true or false according to the positions of the spheres. I think randomDensePack does perform compression on the packing...

I' m not sure I answered you question, did I understand what you were after?

[1] https://yade-dem.org/doc/search.html?q=predicate&check_keywords=yes&area=default