Sintered material

Asked by Guillaume Maîtrejean on 2013-09-27

Hello all,
I would like to perform calculations of sintered material (mainly metal) and I wonder if it is possible to allow some (controlled) initial penetration between spheres simulating sintered material. Of course, as it is an initial state, the penetration should not create any stress... So far I have not found any way of doing that easily (keeping in mind I am a fairly new user of Yade)... Is it possible ?

Thanks in advance

Question information

English Edit question
Yade Edit question
No assignee Edit question
Last query:
Last reply:

If you use Law2_ScGeom6D_CohFrictPhys_CohesionMoment, this line will
make current configuration an equilibrium configuration:

i.phys.unp = i.geom.un #plastic disp = current disp, i.e. the elastic
deformation is null, fn=0


Jérôme Duriez (jduriez) said : #2


Using JCFpm classes (designed initially for jointed rock, but if you do not introduce any joint, it describes simply a cohesive material) should probably do what you want.
Especially thanks to

Perhaps not so simple, depending how much you're "new", but surely possible (looking the code in JointedCohesiveFrictionalPM files may help)

Jérôme Duriez (jduriez) said : #3


I'm currently trying to achieve the same goal, following the advice concerning CohFrictPhys.unp, but I can not obtain a quiet sample...

I think this is due to some "chronological offset" : performing one step is necessary to get access to interactions (i.phys / i.geom..). During this step, spheres moved because of the initial overlap. Then, when i.geom.penetrationDepth is accessed (to set i.phys.unp), this penetrationDepth does not reflect the current state of the sample => the operation does not cancel the normal force.

Computing myself the current penetration depth to assign it to phys.unp allowed me to do the trick for 2 spheres, but not for a whole sample...

So, my question is in fact simple : did someone ever use Bruno's suggestion in practice ?

Jérôme Duriez (jduriez) said : #4

Finally, repeating (20 times in my case, I did not try to pick up the minimum value, but it is > 1) the affectation of phys.unp with a current penetration depth (thus # from geom.penetrationDepth) seems to be finally efficient...

(I also cancelled all bodies velocities, and i.phys.normal/shearForce)

>did someone ever use Bruno's suggestion in practice ?

I did. :)

I don't understand your last post. Does it need to repeat the same assignment many times?!...

Jérôme Duriez (jduriez) said : #6

I meant in my last post "to assign it once, making one time step, and then assign it a second time, and a second time step,". It sounds weird, but by this way I can have a sample at rest, that was not possible, with only one assignment.

> that was not possible, with only one assignment

For future reference: it is possible with a single assignment.

p.s. came back here after reading #1 in

Can you help with this problem?

Provide an answer of your own, or ask Guillaume Maîtrejean for more information if necessary.

To post a message you must log in.