Yade

Damping in Hertz-Mindlin contact law

Asked by Bettina Suhr

Dear all,

I simulate direct shear tests using the Hertz-Mindling contact law with damping (given coefficient of restitution). By chance I saw that there exist contacts where F_t > mu F_n, which violates the Mohr-Coulomb criterion.
When I looked at the source code (HertzMindlin.cpp lines 407-420) I found the reason: The tangential force is calculated according to Mindlin’s law. Then the Coulomb criterion is checked. If the contact is sliding, F_t=mu F_n is set. If the contact is not sliding, the damping term is added. So it can happen that a contact which was not sliding before damping force is added, violates F_t <= mu F_n after the damping part is added.
Is this intention or is it a bug? From my understanding of the paper of (Tsuji et at., 1992), first the damping force should be added and then the check for Mohr-Coloumb criterion should be applied.

Thanks,
Bettina

Question information

Language:
English Edit question
Status:
Solved
For:
Yade Edit question
Assignee:
No assignee Edit question
Solved by:
Bettina Suhr
Solved:
Last query:
Last reply:
Revision history for this message
lichangsheng (sheng0619) said : 		#1

it's not a bug. see the paper of (cundall,1979).

Revision history for this message
Bettina Suhr (bettina-suhr) said : 		#2

Hi lichangsheng,

thanks for you answer. I had a look at (cundall, 1979). There it says “The viscous damping in shear direction is not applied when sliding occurs”. Would you agree that in (tsuji, 1992) it is described differently? As I understand it, here shear force is calculated as sum of constitutive law and damping, and then the Mohr-Coulomb criterion is checked. Is the approach in the cundall paper better/more correct? Is it really intended that there exist cases, where Mohr-Coulomb criterion is violated (as I described above)?

Thanks,
Bettina

Revision history for this message
lichangsheng (sheng0619) said : 		#3

Hi Bettina,

I had a look at (tsuji, 1992). I think your understanding of the paper of (Tsuji et at., 1992) is correct. It is described differently in the paper of (tsuji, 1992). The difference between (tsuji, 1992) and (cundall, 1979) may be that when the contact damping should be addad, before (Tsuji et at., 1992) or after (cundall, 1979) the check for Mohr-Coloumb criterion.I have thought for a few days. In the source code of yade(HertzMindlin.cpp lines 407-420), I think Mohr-Coulomb criterion may be violated as you described,although it is hard to happen. However, I'm not really sure which approach is better/more correct today.
In the past, I also felt confused about this question. At present, I often adopt the local damping used in PFC2D (see PFC2D manual, Version 4.0 , Theory and Background). I will ask other people for this question. I will tell you, if I have a answer. I really hope some experts or you can give a reasonable interpretation.

Thanks,
LI Chang-Sheng

Revision history for this message
Bruno Chareyre (bruno-chareyre) said : 		#4

Both ways are consistent. If it considered that friction coefficient gives an upper bound of the elastic force then the fact that elastic+viscous force can be larger is consistent (this is the current situation).
If it is decided that the upper bound is an upper bound for the total force, then it should not exceed (not possible currently if I understand correctly).
There is no "best" assumption in my opinion, they are both constitutive assumptions and which one is closer to reality may depend on the materials.

Revision history for this message
Bettina Suhr (bettina-suhr) said : 		#5

Thank you, Li Chang-Sheng and Bruno.
Your answers made the whole topic much clearer for me.