Yade

The whole view becomes very large at the end of a triaxial test

Asked by Leonard

Hi,

I am working on a triaxial compression test. Normally, the test ends with the view like Fig1 shows in [1], in which the model located at the middle of the view space. However, some times I got the situation like Fig2 shows in [1], in which the whole view space became very large so that the model looks very small.

Fig1 and Fig2 are just for illstration of my problem, the two figures came from different models. Actually, there are additional tiny particles in the model of Fig2, so I thought the reason why I got what Fig2 shows may because that some tiny particles have been ejected away from the main model. I am not sure about this because all my particles are generated in the rigid wall boundaries which means they would not leave out of this boundary. Particles will not penetrate the rigid wall I think.

Do you have any ideas for avoiding this problem?

Thanks
Leonard

[1]https://we.tl/t-fEElfCnqgZ

Question information

Language:
English Edit question
Status:
Solved
For:
Yade Edit question
Assignee:
No assignee Edit question
Solved by:
Jan Stránský
Solved:
Last query:
Last reply:
Revision history for this message
Jan Stránský (honzik) said : 		#1

Hello,

> the reason why I got what Fig2 shows may because that some tiny particles have been ejected away from the main model

sounds like the most probable reason...

> I am not sure about this

please make sure :-)
something like [1]

> all my particles are generated in the rigid wall boundaries which means they would not leave out of this boundary. Particles will not penetrate the rigid wall I think.

"generating particles in the rigid wall boundaries" is NO guarantee that "they would not leave out of this boundary" NOR that "particles will not penetrate the rigid wall".

It depends on many factors:
- what "rigid wall boundary" is. wall ? box? what thickness? ...
- what "penetrate" mean (the particles has some penetration by DEM definition, recall i.geom.penetrationDepth). If it means "go through", see below
- simulation itself (difficult to say without any code)
   - tiny particles has higher probability to jump across wall without contact detection
   - might be influenced by time step
- material parameters (mainly stiffness)
- ...

cheers
Jan

[1] https://yade-dem.org/doc/yade.utils.html#yade._utils.aabbExtrema

Revision history for this message
Leonard (z2521899293) said : 		#2

Hi Jan,
Thanks for your reply.

> check use aabbExtrema()
The initial size of my model is (vector3(0,0,0), vector3(0.07,0.14,0.07)) by using aabbExtrema(),
The aabbExtrema of Fig2 shows is (vector3(0,0,0.05),vector3(0.126,2.7,0.078)). So I think some particles go out of the walls.

> what "rigid wall boundary" is?
Sorry for having not clearly described it.
The walls I defined in my model are as that used in [1] for triaxial simulation:
mn,mx=Vector3(0,0,0),Vector3(0.07,0.14,0.07)
walls=aabbWalls([mn,mx],thickness=0,material='FrictMat')

 "penetrate" mean "go through" in my question.
According to your suggestion, I think the time step could be the reason.

And just a curiosity, From the results I got, the stress-strain response and volumetric response have not been affected by this situation. Does it mean that I can basically ignore this problem?

Thanks

[1]https://gitlab.com/yade-dev/trunk/blob/master/examples/triax-tutorial/script-session1.py

Revision history for this message
Best Jan Stránský (honzik) said : 		#3

> So I think some particles go out of the walls.

yes, the picture as well as numbers suggest this

> And just a curiosity, From the results I got, the stress-strain response and volumetric response have not been affected by this situation.

It depends on details. If one particle "flies away", then it is reasonable that it does not have much influence..

> Does it mean that I can basically ignore this problem?

It depends on details :-) and other circumstances, but most importantly it is up to you :-)

cheers
Jan

Revision history for this message
Leonard (z2521899293) said : 		#4

Thanks Jan Stránský, that solved my question.