yade-users team mailing list archive

[Fu zuoguang <question250783@xxxxxxxxxxxxxxxxxxxxx>]

Question #250783 on Yade changed:
https://answers.launchpad.net/yade/+question/250783

    Status: Answered => Open

Fu zuoguang is still having a problem:
Dear Hongyang Alex Cheng:

    I am sorry for answering your question so late.Yes, you are
right,‘GlobalStiffnessTimeStepper’ has done very well in ensuring the
stability of simulations. My reason of choosing ‘dynDt’ has nothing to
do with controlling the stability but has close relationship with
postprocess.Take the tri-axial simulation in Yade as a simple example,
after the isotropic compression, the process of deviatoric shearing can
be carried out just by these two conditions:

(1)	Laternal stress which is equal to isotropic pressure.(always choose x and z axis)
(2)	Axial strain which is only determined by a certain strain rate.(always choose y axis)

And the final state of this simulation is always checked by a special
axial strain. It is obvious that the relationship between the unit
simulation step delta-t and the final axial strain can be written as
follows:

(Delta-t)*(total-steps)*(y-strainrate)=final axial strain      (1)

    The y-strainrate and the final axial strain are always treated as
the prediction of the simulation and are known. For this reason (Delta-t
)*(total-steps) should be a constant to establish the formula(1).

    In yade, a changed ‘timestepper’ can hardly provide us a certain
constant (total-steps) for controlling the total steps(cost) in
simulation with a given smallest particle size and a total number. So I
hope that there exists a constant delta-t to offer much help to not only
the stable of simulation but also the fully controlled number of steps.
In this condition, as the absolutely necessary portion of total result,
each step or each interval of steps in data can represent a specific
state into the whole process and then be as one of the specific
representations of all the simulation objectives. In the other hand, the
attributes of particles may not the same in different simulations. If
they can use the same delta-t as the unit step, each element of the two
result data output by ‘iterPeriod’ may have the same meaning. It can
bring convenience to comparison of the data derived from different
simulations. I think the chosen delta-t, which is smaller than
‘timestepper’ can meet what I want, even though it may bring a little
more cost of simulation.

    I am not very right and there may be a better way for my purpose.
Please tell me a little more.

Fu

-- 
You received this question notification because you are a member of
yade-users, which is an answer contact for Yade.