Physic interaction betweem particles and polyhedra

Hi Jonathan,

currently you can have only one instance (thus one set of parameters) of
p2_FrictMat_FrictMat_MindlinPhys. If needed, it could be modified to
distinguish between sphere-sphere and sphere-box interactions, but it would
need some coding and probably would be a bit hacky..

Do you have more boxes in your simulation? If no, I would use box instead
of polyhedra.

cheers
Jan

2016-05-04 16:52 GMT+02:00 Jonathan Pergoli <
<email address hidden>>:

> New question #293247 on Yade:
> https://answers.launchpad.net/yade/+question/293247
>
> Hi everybody!
>
> I'm trying to simulate a box falling on a cylindrical pack of spheres. The
> spheres have coefficient of restitution = .1 and the box has a COR = .6.
> For the spheres I have used Ip2_FrictMat_FrictMat_MindlinPhys(en=.1,es=.1)
> . As regard the box, I would like to use the same but if I write:
> ...
>
> [p2_FrictMat_FrictMat_MindlinPhys(en=.1,es=.1),p2_FrictMat_FrictMat_MindlinPhys(en=.6,es=.6)]
> the former properties are overwritten by the latter.
>
> The box is made using utils.polyhedron().
>
> I have also used Ip2_FrictMat_PolyhedraMat_CundallStrack(), but I cannot
> set the COR.
>
> Can someone help me,please?
>
> Thank you :)
>
> --
> You received this question notification because your team yade-users is
> an answer contact for Yade.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-users
> Post to : <email address hidden>
> Unsubscribe : https://launchpad.net/~yade-users
> More help : https://help.launchpad.net/ListHelp
>

Hi Jan,

How can be modified?

I have only one box.

instead of using utils.polyhedron() using utils.box() :-)
+ changing some Ig2 and Law2 fomr Polyhedra to Box
cheers
Jan

2016-05-04 18:13 GMT+02:00 Jonathan Pergoli <
<email address hidden>>:

> Question #293247 on Yade changed:
> https://answers.launchpad.net/yade/+question/293247
>
> Jonathan Pergoli posted a new comment:
> Hi Jan,
>
> How can be modified?
>
> I have only one box.
>
> --
> You received this question notification because your team yade-users is
> an answer contact for Yade.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-users
> Post to : <email address hidden>
> Unsubscribe : https://launchpad.net/~yade-users
> More help : https://help.launchpad.net/ListHelp
>

Thanks Jan, but I mean how can i modifie Ip2_FrictMat_FrictMat_MindlinPhys in order to distinguish between sphere-sphere and sphere box? :-)

I'm using box now instead of polyhedra but the problem is the same.

P.S. I'm new to YADE :-)

Hello,

You mix here "geometry" [1] and "physics" [2] of YADE interactions. In order to describe interactions in DEM, you need to know:

- what is the relative movement between two Discrete Elements (i.e. YADE bodies), that's what's inside the "geometry" of the interaction, often denoted i.geom

- how forces are associated to these relative displacements, i.e. which contact law, that's what's inside the "physics" of the interaction: i.phys

Ip2_FrictMat_FrictMat_MindlinPhys, like all "Ip2_" plays a role in describing the physics of the interaction (computing e.g. the stifness parameters of the interaction from the material parameters of the bodies). It has most often nothing to do with bodies shape !
Indeed, you can still use Ip2_FrictMat_FrictMat_MindlinPhys in your simulations whatever your shapes are, the only condition is that your bodies are of "FrictMat" type.
(I'm not 100% correct here, because of https://github.com/yade/trunk/blob/6d4a0d1689c6bad2dbdd6e27d1653418f7fc6498/pkg/dem/HertzMindlin.cpp#L43, but if you're new to YADE you may disregard this parenthesis that won't have any effect in your case...)

The bodies shapes enter into picture through the "geom". Because the way e.g. overlap is computed depends whether we face overlap between spheres, or between a sphere and a box, or between a sphere and a facet.

Your problem is thus (as Jan said: he told you about "Ig2", not "Ip2") to use in your simulation the correct "Ig2_". Look here [3] (the names do help !) to pick the right one.

Once this will be done, you will have a new kind of i.geom in your simulation. For this reason, you may also have to update your "Law2_.." (as Jan suggested again) which, in the end, has to consider both i.geom and i.phys to do his job (computing and applying interaction forces)

Again, you will have to choose the correct Law2 (from [4], there is a lot..., but again the names do help) to be sure it can handle the i.geom and i.phys you have in your simulation.

[1] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.Interaction.geom, see https://yade-dem.org/doc/yade.wrapper.html#igeom for all possible choices
[2] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.Interaction.phys, see https://yade-dem.org/doc/yade.wrapper.html#iphys for all possible choices
[3] https://yade-dem.org/doc/yade.wrapper.html#igeomfunctor
[4] https://yade-dem.org/doc/yade.wrapper.html#lawfunctor

Hi Jérome,

you mean this:
O.engines=[
 ForceResetter(),
 InsertionSortCollider([Bo1_Sphere_Aabb(),Bo1_Facet_Aabb(),Bo1_Box_Aabb()]),
 InteractionLoop(
  [Ig2_Sphere_Sphere_ScGeom(),Ig2_Facet_Sphere_ScGeom(),Ig2_Box_Sphere_ScGeom()],
  [Ip2_FrictMat_FrictMat_MindlinPhys(en=.55,es=.55,krot=3)],
  [Law2_ScGeom_MindlinPhys_Mindlin(includeMoment=True)]
 ),
 NewtonIntegrator(gravity=(0,0,-2.5e-4),damping=.2)
]
?

Yes, this is what I meant (in case your non-spherical bodies are all of the same type, either Facet or Box, you may remove either Bo1_Box_Aabb() ; Ig2_Box_Sphere_ScGeom() or Bo1_Facet_Aabb() ; Ig2_Facet_Sphere_ScGeom())

This being said, I realize my lengthy answer may not be directly related to your initial problem, I am sorry...

I now understand you have different material parameters (these en,es) for different bodies. Usually, YADE handles this at the bodies "material" properties level [1].

For instance, simulations with "FrictPhys" handle like a charm different friction angles for sphere-sphere and sphere-box interactions. We just put different values for the frictionAngle attribute of the sphere or box bodies' Material [2], and one unique Ip2 (Ip2_FrictMat_FrictMat_FrictPhys) does the job giving different local friction angles for sphere-sphere and sphere-box interactions.

Here, the design of Ip2_FrictMat_FrictMat_MindlinPhys is flawed in my point of view (I'm not used to MindlinPhys, though) if these en,es,krot are material parameters that can be different among bodies. Because including these parameters in Ip2_FrictMat_FrictMat_MindlinPhys prevents playing with different parameters the same way as frictionAngle within FrictPhys (hence your problem)..

I see two possible solutions to your problem, both involving modifying the source code

- either modifying the source code of Ip2_FrictMat_FrictMat_MindlinPhys to detect the shape of interacting bodies and assign distinct en,es in accordance. From my point of view, this would not conform the classical YADE design I exposed in #6, I agree thus with Jan it would be "hacky"

- removing these en,es from Ip2_FrictMat_FrictMat_MindlinPhys to insert them in a (new ?) material class. Which would make more sense, from my point of view.

Further discussion would require other people more used to MindlinPhys than me. Hope it still helps....

[1] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.Body.material
[2] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.FrictMat.frictionAngle

What about this:
gravel=FrictMat(density=1700,frictionAngle=.9188,label="gravel")
O.materials.append(gravel)
MLI=FrictMat(density=1280,frictionAngle=0.1489,label="MLI")
O.materials.append(MLI)
...
O.engines=[

 ForceResetter(),
 InsertionSortCollider([Bo1_Sphere_Aabb(),Bo1_Box_Aabb()]),
 InteractionLoop(
  [Ig2_Sphere_Sphere_ScGeom(),Ig2_Box_Sphere_ScGeom()],
  [Ip2_FrictMat_FrictMat_MindlinPhys(en=MatchMaker(matches=((gravel,MLI,.6),(gravel,gravel,.55))))],
  [Law2_ScGeom_MindlinPhys_Mindlin(includeMoment=True)]
 ),
 NewtonIntegrator(gravity=(0,0,-9.81),damping=0.1)
]

It gives me this error:
No registered converter was able to produce a C++ rvalue of type Eigen::Matrix<double, 3, 1, 0, 3, 1> from this Python object of type tuple

Sorry I wrote the wrong code. I did this:
O.engines=[

 ForceResetter(),
 InsertionSortCollider([Bo1_Sphere_Aabb(),Bo1_Box_Aabb()]),
 InteractionLoop(
  [Ig2_Sphere_Sphere_ScGeom(),Ig2_Box_Sphere_ScGeom()],
  [Ip2_FrictMat_FrictMat_MindlinPhys(en=MatchMaker(matches=((gravel,MLI,.6),(gravel,gravel,.55))),es=MatchMaker(matches=((gravel,MLI,.6),(gravel,gravel,.55))),krot=MatchMaker(matches=((gravel,MLI,.15),(gravel,gravel,3))))],
  [Law2_ScGeom_MindlinPhys_Mindlin(includeMoment=True)]
 ),
 NewtonIntegrator(gravity=(0,0,-9.81),damping=0.1)
]

In this way the simulation distinguishes between en and es, but for the rolling friction coefficient krot it doesn't work.

Now I can distinguish between interactions:

def rollingfriction():
 for i in sphere:
  try:
   O.interactions[m,sphere[i]].phys.kr=.15
   print "Interaction between",m,sphere[i]
  except:
   print "No interaction exists",m,sphere[i]

Thank you guys for helping me. :)