![]() |
![]() |
In the Fig. fig:tflip_vglob and Fig. fig:tflip_vstripe we
display the distributions of horizontal velocities for the 2D Inclined
Plane Model with a bottom wall with stochastic vibrations. The first
Fig. fig:tflip_vglob, shows the distributions taken from all
over the system (i.e. counting particles independently of the height)
for different : the data collapse is obtained by rescaling the
velocities by
. The distributions are very different from
a Gaussian.
Instead, in Fig. fig:tflip_vstripe we show the velocity
distributions of particles contained in stripes at different heights
from the wall, again rescaled by
(their own variance)
in order to obtain the data collapse. It again appears that the
distributions are non-Gaussian and their broadening (that is the
granular temperature
) is height dependent. This dependence is
shown in Fig. fig:tflip_t.
The case of periodically vibrated wall is illustrated in
Fig. fig:vflip_v. One can see the distribution of horizontal
velocities in two different regimes: for a non-Gaussian
distribution is obtained, while a distribution close to a Gaussian
appears when
. This trend toward a Gaussian, as the angle
of inclination is raised up, reproduces exactly the experimental
observation of Kudrolli and Henry [132] (where the angle
of inclination of the plane was raised up from
to
). This phenomena can be explained as an effect of the
increase of the collision rate with the wall induced by the increase
of the force of gravity: an higher collision rate ``randomizes'' the
velocities in a more efficient way. If one accepts the analogy between
the vibrating wall and the heating bath used to drive the granular gas
in Chapter
3, the increase of collision rate with the wall is analogous
to the decrease of the ratio
(an increase of the
``heating rate''). In the Randomly Driven model the decrease of this
ratio corresponds to the transition from the non-Gaussian regime to
the Gaussian one.
In Fig. fig:vflip_v_tails we show a log-log plot of the tails
of the collapsed distributions presented in
Fig. fig:tflip_vstripe. An algebraic fit
is proposed.
![]() |
![]() |