yade-users team mailing list archive

[Bruno Chareyre <question658629@xxxxxxxxxxxxxxxxxxxxx>]

Question #658629 on Yade changed:
https://answers.launchpad.net/yade/+question/658629

Bruno Chareyre proposed the following answer:
I must point out (sorry to enter this discussion late) that you seem to make your life complex for no good, for the sole reason that you want a circular cross-section.
Proof: after 29 messages it is still not solved ;)
More precisely:
-  pack.inCylinder(...) + randomDensePack will give extremely large porosity along the curved wall, which implies another equilibration stage, and still a highly heterogeneous micro-structure in the end.
- you have no automatic servo-control for the applied stress, so you replace it with an applied force on top plate, which is not the best choice in terms of stability (then you'll need to have artificial damping, never a good thing).
- calculating the radial stress is less easy (unless you are only interested in the sample-scale average)
- you cannot use the fluid modules (be it 1 or 2-phase flow) because they assume a square box.

On the other hand, all the above problems disappear if you use TriaxialStressController as in many examples. You can even find example scripts for oedometer in both 1 and 2-phase flow.
Since there is nothing that distinguish a square cross section from a circular cross section in an oedometer (from a mechanical point of view), I would seriously consider throwing away the idea of a "realistic" mold shape unless there is _very_ strong reason for it. "Being like in experiment" is not a good reason IMO.

Bruno

-- 
You received this question notification because your team yade-users is
an answer contact for Yade.