yade-dev team mailing list archive

[Chiara Modenese <c.modenese@xxxxxxxxx>]

On 8 July 2011 07:12, Christian Jakob <806944@xxxxxxxxxxxxxxxxxx> wrote:

> Zitat von Chiara Modenese <c.modenese@xxxxxxxxx>:
>
> > So the problem is not the tension on or off. Can you check that the
> damping
> > coefficients are defined in the same way for both codes? That could be
> the
> > cause of the difference then.
> >
>
>
> As far as I know the viscous damping slows the particles down, that means
> that the relative velocity of the two contacting particles influences the
> damping force D_n = c_n * abs(v_n). In PFC manual there are two formulas for
> the normal force F_n = k_n * u_n and the damping force, that I joined
> together :
>
> F_n_damped = F_n - D_n
> F_n_damped = k_n * u_n - c_n * abs(v_n)
>
> k_n - normal stiffness
> u_n - overlap (positive if particles overlap)
> c_n - normal damping constant
> v_n - relative velocity at the contact
>
> (see page 1-9 and 1-26 in Itasca manual Theory and Background PFC 4.0)
>
> c_n = beta_n * c_crit = beta_n * 2 * sqrt(m * k_n)
>
> beta_n - critical damping ratio
> c_crit - critical damping constant
> m - average mass of the two contacting spheres
>
This may be different in Yade. If one of the two bodies is fixed, like in
your case, then only the mass of the moving body is considered. I do not
know how is PFC in this regard, please check it out.

>
> So with the command "damp viscous normal visc_damp_normal notension on"
> I can set beta_n. This should be the same as in YADE, when I set
> "ContactModel.betan=visc_damp_normal" .... or not ?!
>
Again, this notension on means that Fn >= 0 (I assume it because I do not
really use PFC). In Yade, at the moment, this option is not enforced in case
of viscous damping. As I said, it does not have any influence for
quasi-static conditions but it matters in dynamic situations.

>
>
> > In case 2 and 3 we should get the same result, I think. Am not sure what
> > could be the cause.
> > Hint 1) Can you try to compare the numbers step by step, having exactly
> the
> > same configuration and time step?
>
>
> I have the same configuration in both programs, same step width, same
> properties, same constants, as you can see when you are looking into the
> scripts.
>
Do the test just at the contact, step by step. This way you simplify your
test. Really, just take two balls and follow the normal contact behaviour,
step by step. There may be some small differences in the way the law is
implemented in PFC, I do not know if they follow an incremental formulation
or not.

>
>
> > Hint 2) Can you try to do the same test but with the simple linear
> elastic
> > law and no damping at all?
> > Also I would just suggest to do a much simpler test where you just study
> the
> > contact between two balls and record velocities, positions etc in every
> time
> > step. Let me know.
>
>
> I will do so, but it takes some time ...
>
OK, let me know.

>
>
> > Bruno, what do you think about the above numbers?
> >
> >>
> >> ...hope it helps ...
> >>
> >> I do not understand where the attractive force (in PFC notension off)
> comes
> >> from, but the influence on the results is very low (see case1).
> >>
> > The attractive force comes from the damper.
>
>
> Ok, but can you explain what the damper does? Is there a difference in the
> formula (see above)?
>
Please keep in mind that the coefficient of viscous damping you are using
now is calculated with reference to a linear spring-dashpot system, which is
only an approximation for a non linear elastic law such as the Hertzian one.
The formula above are correct for a linear elastic spring. In case of Hertz
law I take the current tangential stiffness to compute the critical damping
coefficients.

>
>
> > Christian, did you write the model or you are taking the one already
> > implemented in PFC? In the last case, I can have a look at the manual and
> > see what sort of viscous damping they implement. Let me know.
>
>
> I used the Hertz model, that is implemented in the standard PFC package
> (command "model hertz" in line 36 in two-spheres-jumping.dat)
>
OK.

>
> Christian
>
> --
> You received this bug notification because you are a member of Yade
> developers, which is the registrant for Yade.
> https://bugs.launchpad.net/bugs/806944
>
> Title:
>  different behavoir of Hertz model while comparing PFC and YADE
>
> Status in Yet Another Dynamic Engine:
>  New
>
> Bug description:
>  Hello,
>
>  During verifiing my model I compared output from PFC 4.0 with the
>  output from YADE (bzr2877 on Debian Squeeze 64bit). There I found a
>  different behavior between these two programs. I dont know if this is
>  a bug or not, but I hope someone can explain the different behavior.
>
>  The model itself is very simple. There are two spheres, an upper one and a
> lower one. The lower one is fixed and the upper one is falling down to the
> lower one, collides and jumps back. The value of interest is the flyback
> height of the upper particle.
>  In PFC I measure higher values of flyback height, then in YADE (see
> log-files in the attachment).
>
>  Can someone tell me why there is a difference in the flyback height?
>  (input parameters are the same for both programs ... see
> caller_two-spheres-jumping.dat for PFC and two-spheres.py for YADE)
>
>  Regards,
>
>  Christian Jakob
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/yade/+bug/806944/+subscriptions
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev
> More help   : https://help.launchpad.net/ListHelp
>

-- 
You received this bug notification because you are a member of Yade
developers, which is the registrant for Yade.
https://bugs.launchpad.net/bugs/806944

Title:
  different behavoir of Hertz model while comparing PFC and YADE

Status in Yet Another Dynamic Engine:
  New

Bug description:
  Hello,

  During verifiing my model I compared output from PFC 4.0 with the
  output from YADE (bzr2877 on Debian Squeeze 64bit). There I found a
  different behavior between these two programs. I dont know if this is
  a bug or not, but I hope someone can explain the different behavior.

  The model itself is very simple. There are two spheres, an upper one and a lower one. The lower one is fixed and the upper one is falling down to the lower one, collides and jumps back. The value of interest is the flyback height of the upper particle.
  In PFC I measure higher values of flyback height, then in YADE (see log-files in the attachment).

  Can someone tell me why there is a difference in the flyback height?
  (input parameters are the same for both programs ... see caller_two-spheres-jumping.dat for PFC and two-spheres.py for YADE)

  Regards,

  Christian Jakob

To manage notifications about this bug go to:
https://bugs.launchpad.net/yade/+bug/806944/+subscriptions