pair_colloid radius issues

Dear Lammps users,

I encountered a problem using pair_colloid. It appeared while using particles of diameter 0.02µm (only on pair) at a distance of 1µm – it gives the error large/large overlapping. Looking for the reason I found out, that pair_colloid is using:

d1_one = force->numeric(FLERR,arg[4]

d2_one = force->numeric(FLERR,arg[5]

which gives a diameter of 1µm in my case.

both are equivalent “diameters”

Is that correct? If yes, what is the reason?

I would expect it to use the real particle diameters, like

*radius = atom->radius

(see pair_dmt.cpp)

Can I just change the radii like pair_dmt.cpp – or will I get trouble when doing this?

Lammps-5Jun19

Thank you so much

Henrik

Dear Lammps users,

I encountered a problem using pair_colloid. It appeared while using particles of diameter 0.02µm (only on pair) at a distance of 1µm – it gives the error large/large overlapping. Looking for the reason I found out, that pair_colloid is using:

d1_one = force->numeric(FLERR,arg[4]

d2_one = force->numeric(FLERR,arg[5]

which gives a diameter of 1µm in my case.

both are equivalent “diameters”

Is that correct? If yes, what is the reason?

from the documentation:

D1 and d2 are particle diameters, so that d1 = 2a1 and d2 = 2a2 in the formulas above. Both d1 and d2 must be values >= 0. If d1 > 0 and d2 > 0, then the pair interacts via the colloid-colloid formula above. If d1 = 0 and d2 = 0, then the pair interacts via the solvent-solvent formula. I.e. a d value of 0 is a Lennard-Jones particle of size sigma. If either d1 = 0 or d2 = 0 and the other is larger, then the pair interacts via the colloid-solvent formula.

thus pair style colloid does not require (and in fact ignores) the extended properties from atom style sphere.
thus the radius of particles is coupled to the atom type. thus a generalized polydisperse system is not possible.

I would expect it to use the real particle diameters, like

*radius = atom->radius

(see pair_dmt.cpp)

there is no pair_dmt.cpp file in the LAMMPS distribution.

Can I just change the radii like pair_dmt.cpp – or will I get trouble when doing this?

as pointed out above. you must set the radius via pair_coeff and it is coupled to the atom type. if you modify the source, you would need to make significant changes. Just think about the different computation of the prefactors, that depend on the radii. but that is up to you to figure out. when you hack the source, you are on your own. :wink:
but then again, if you would be successful, i would expect, that there might be other LAMMPS users that would be happy to have such a colloid/sphere pair style in the LAMMPS distribution. so we would appreciate if this would be contributed to the LAMMPS distribution.

axel.

I assume Henrik is referring to the pair_dmt that I sent out into the mailing list two years ago ( https://sourceforge.net/p/lammps/mailman/message/36116721/ ).

The source files could use some clear up. I put it out there since the DMT and JKR models for colloid contacts being worked on at Sandia have been lingering on the verge of being released for >5 years. To be fair though, I got 4 publications out of using that pair style. I’m sure there might be more at this point.

Apparently, its been some time since I’ve looked in on the lammps distribution. I was pleasantly surprised to find that a version of DMT is not included, minus the short-ranged attractive well.

Over the past several years of passively looking at the list every so often, I have seen interest in colloids with interactions beyond contact(vdW/double layers) and granular interactions come up time and time again. I usually chime in for better or worse.

So some questions are: Is there any interest in including a short ranged attractive well/colloid interaction in the distribution? Should it be apart of the granular code? Should it be a simple extension like my pair_dmt to be used with hybrid overlay, and as pair_lubricate is?

There are other things to consider as well for computational speed in dry systems (well shape - usually doesn’t impact much), but those can be discussed offline if there is interest.

Regards,
Eric Murphy, PhD
Multiphase Flow Scientist
Mechanical Engineer
[email protected]…24… | 563-449-6661

LinkedIn | ResearchGate | Google Scholar

In the recent pair granular style (not gran/hooke, hertz, etc), there
are options for dmt and jkr in combination with the more common
normal, tangential forces. If these are differences between what this
would mean in a granular system versus a colloid system, then it
might make sense to have options like these in pair colloid.
Otherwise probably not?

Steve

Seems like a sound idea to incorporate these into the colloid pair style. It really isn’t that difficult to extend pair colloid to work with granular particles, particularly with the DMT model. I don’t believe it makes sense for JKR to have large-large interactions, but nonetheless interactions with solvent molecules could be still important for some folks.

I’d be happy to incorporate these for use in the distributed, but it’d probably be good to have a discussion about some particular details of possible models. There’s a lot of functionality/options that could be added for folks, but I don’t want to muddle it up with too many options.

Henrik,

If you have a hold of my old pair_dmt.cpp and h files, feel free to use them. I cannot recall if I retained the use of small-small and large-small interactions in it. The basic difference between the large-large interactions in pair_colloid and pair_dmt are that dmt can be used with granular interactions to do the repulsion of colloids. Rather than relying on the LJ potential to repel spheres (assuming complete rigidity and ) as in pair_colloid, in pair_dmt the solid mechanics of the colloids/granules are actually what repels the spheres.

It also makes the a1,a2>>sigma assumption, so the complicated U_A is just replaced with -Acc*(a1*a2)/(a1+a2)/(6 * (r-(a1+a2) + sigma). As you noted you can set the radius in this so that it treats a typical asperity rather than the radius of the entire colloid, since the vdW force is relatively short-ranged on the scale of a micron.

Do not use it with the granular interaction with dmt on, as that constant attractive force at contact is already in my pair_dmt.