yade-users team mailing list archive

[Chiara Modenese <question159394@xxxxxxxxxxxxxxxxxxxxx>]

Question #159394 on Yade changed:
https://answers.launchpad.net/yade/+question/159394

    Status: Answered => Open

Chiara Modenese is still having a problem:
On 31 May 2011 08:41, Chareyre <question159394@xxxxxxxxxxxxxxxxxxxxx>
wrote:

> Your question #159394 on Yade changed:
> https://answers.launchpad.net/yade/+question/159394
>
>    Status: Open => Answered
>
> Chareyre proposed the following answer:
>
> > This equation is very similar to the servo-control we use with rigid
> > boundaries; with the difference that there we also have a gain parameter
> > (let me call it that way, for me it does not matter if it is split into
> two
> > parameters other than one). So, why here the formula is correct but with
> > rigid walls we need the gain parameter?
> For rigid walls (or periodic && !dynCell), it reads
> displacement=stress_error/stiffness in both cases (a bit more
> sophisticated in fact, but this is the basic idea).
> For dynCell, it reads acceleration=stress_error/mass (idem).
> I don't see what you mean by "we need a gain". Since a gain is any
> factor that enter a servo-control, there is basically a gain in all
> three cases.
>
Ok, you are right to say that there is a gain in all cases. I was making a
comparison between this control in the periodic

strain_rate=1/scene->dt*(goal[axis]-stress[axis])*cellArea[axis]/(stiff[axis]>0?stiff[axis]:1.);

and this control in triaxial test

previousTranslation[wall] = (1-wallDamping)*translation*normal[wall] +
0.8*previousTranslation[wall];

To me they are conceptually similar (leaving the 0.8 coeff aside), with the
difference that there is no wallDamping in periodic case. Maybe I am
misunderstanding...

>
> >
> > It is not good to have two methods if one of them does not work (at least
> to
> > my experience, but I would be happy to hear from others that I am wrong).
> Am
> > referring here to what you call "static" method (in PeriIso file). If you
> > ever try to use it, you will see that it does not properly work. This is
> why
> > my question about gain parameter.
>
> I think Vaclav and Jan used it successfully, with a different
> constitutive models than yours.
>
Vaclav did not use it and Jan wrote his own controller, but I let him add a
comment on all of this (could not you add your opinion, Jan?).


> I know Jan also tried to adjust the stiffness during iterations, instead
> of just predicting. So in principle it could work for many constitutive
> laws.
> I can't comment more personally, I never tried.
> > Mass is not mass, given its current unit of measure which can be derived
> > from the equation. As I see it, it is really a gain parameter (and I
> still
> > do not see the advantage to use 1/mass when this could simply be called
> > gain, unless you have a method which estimates the mass parameter from a
> > physical point of view, do you?).
> No, I don't have a clear derivation yet. I'm only guessing it needs to
> include lengths somewhere. It's an opened question for me (and if I read
> this book correctly, it's not explained either
> http://www.iste.co.uk/index.php?f=a&ACTION=View&id=384). I suggest you
> keep this question in mind for Radjaï in Alert school. ;-)
>
No worries, I have a list of questions ready to ask :-) Hopefully it will be
interesting.

>
> Still, it is an "inertia" (if you don't like mass...) because it acts on
> acceleration, while a stiffness would act on position. Stiffness and
> mass are all "gains" if they appear in a servo-control loop.
>
Definitely.

> In the sense you use it here, gain is a very vague concept. Strictly
> speaking, the unit of gain is dB and it is a function of frequency...
>
Maybe am not precise when I use this naming, but it is because I am used to
read about it in such a way.

Cheers.
Chiara

>
> Bruno
>
> --
> If this answers your question, please go to the following page to let us
> know that it is solved:
> https://answers.launchpad.net/yade/+question/159394/+confirm?answer_id=2
>
> If you still need help, you can reply to this email or go to the
> following page to enter your feedback:
> https://answers.launchpad.net/yade/+question/159394
>
> You received this question notification because you asked the question.
>

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