Physica A 283 (2000) 218–222

www.elsevier.com/locate/physa

Surface uctuations in a slowly driven

granular system
N. Neronea; ∗ , M.A. Aguirrea , A. Calvoa , I. Ippolitob , D. Bideaub
a Grupo à Universidad de Buenos Aires,
de Medios Porosos, Facultad de Ingenieria,
Paseo ColÃon 850, CP 1063, Buenos Aires, Argentina
b Groupe de MatierÂ
e CondensÃee et MatÃeriaux, UMR 6626, UniversitÃe de Rennes 1 Campus de Beaulieu,
35042 Rennes Cedex, France

Abstract
We report an experiment on a granular packing: a box lled with glass beads is tilted very
slowly up to the maximum angle of stability where a big avalanche is produced. During the
build-up period many rearrangements occur on the free surface of the packing. Digital imaging
was used to study these rearrangements. The probability distribution of sizes for the observed
mass uctuations follow a power-law behavior, which is the signature of self-organized criticality.
However, this description breaks down in the limit of big rearrangements where inertia eects
are not negligible. c 2000 Elsevier Science B.V. All rights reserved.

PACS: 05.40.+j; 64.60.Ht; 05.70.Jk

Keywords: Granular matter; Self-organized criticality; Avalanches

1. Introduction

The concept of self-organized criticality (SOC) was developed by Bak et al. [1,2]
in order to describe the dynamics of some extended systems. These systems evolve
into a critical state, without tuning external parameters, where events of all sizes are
observed with no spatial and temporal correlations. The size distribution function of
those events (avalanches) follows a power law. Although, the prototypic example of
SOC is the sandpile cellular automata model the behavior of real piles has proven to
be a hard task to describe.
In this work, we present an experiment on a granular packing in a contained geome-
try: a box lled with a certain number of layers of glass beads is tilted very slowly up
∗Corresponding author.
E-mail address: nnerone@.uba.ar (N. Nerone)

0378-4371/00/$ - see front matter c 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 4 3 7 1 ( 0 0 ) 0 0 1 5 6 - 4
 N. Nerone et al. / Physica A 283 (2000) 218–222 219

Fig. 1. Semi-log plot of the relative size of events, s, as a function of the angle of inclination, Â, obtained
in one experiment. The dashed line corresponds to the exponential tting function s = A exp(Â=Âo ), where A
and Âo are constants.

to the threshold of instability where a big avalanche is produced. The avalanche takes
place at a critical angle called the maximum angle of stability, ÂM . This large sliding
decreases the slope of the free surface of the packing until a second critical angle
is reached: the angle of repose, ÂR . During the build up period many rearrangements
occur on the free surface of the packing. We have used digital imaging to study the
behavior of these rearrangements leading up to a large slide.
In a previous work [3] we have determined the existence of a critical number of
layers (Nc ∼ 13, for our system dimensions) above which the critical angles and the
mass displaced by the avalanche do not depend on the height of the packing. We
analyze here a 20-layers packing (above Nc ). The packing is obtained by lling up
with glass beads a box with a rough bottom. The dimensions of the box are: length
L=320 mm, width w=260 mm and the beads diameter is d=2:2±0:2 mm. Experiments
are performed under a controlled relative humidity of 50%. The box is inclined at a
◦
rate of 0:3 per minute. When the critical angle ÂM is reached a large avalanche is
detected. The avalanche mass, M , the maximum angle of stability, ÂM and the angle 
between the plane and the nal free surface are measured. Finally, the angle of repose
is computed as: ÂR = ÂM − . The size of the rearrangements is determined using a
CCD camera interfaced to a computer. In order to make a good determination of these
quantities, 10 experiments were performed in identical conditions.
In Fig. 1, the size of the rearrangements and the avalanche normalized with the
size of one layer, s, is plotted as a function of the angle of inclination of the box, Â.
Three dierent scales are observed: (1) Small rearrangements: involve less than 10%
of the grains of the free surface (s ¡ 0:1). No grains ow out of the box. The
220 N. Nerone et al. / Physica A 283 (2000) 218–222

Fig. 2. Log–log plot of the size distribution function, D(s). The solid line correspond to the tting function
D(s)˙s−2 .

measurements present a noisy structure. (2) Big rearrangements: they involve from
10% to almost the whole free surface and in very few occasions some grains ow out
of the box (0:1 ¡ s ¡ 1). (3) Avalanche: involves typically 13 layers and half of them
ow out of the box. This is the mechanism that resets the slope of the packing to the
angle of repose.
These three scales can also be identied in the plot of the size distribution function,
D(s), shown in Fig. 2. D(s) has a clear power-law behavior that last for almost two
decades (small rearrangements) characterized by an exponent  = 2:0 ± 0:4. This result
agrees with that obtained by Frette and co-workers [4] in experiments performed in a
pile of rise with elongated grains, where the inertia eect was negligible. However, for
the big rearrangements, an increase in the probability that deviates from the power-law
behavior is observed. The same result is obtained in the sandpile cellular-automata
model developed by Prado and Olami [5] in which inertia eects are taken into account.
Finally, note that no events are seen from s = 1 to the avalanche that involves 6 layers
(s = 6). The same result was observed by Rosendahl and co-workers [6] in experiments
performed in a conical pile.
Fig. 3 shows 10 measurements of the parameters that characterize the avalanche,
obtained in identical experimental conditions. The maximum angle of stability, ÂM ,
and  are plotted as a function of the relative size, s. We observe that avalanches do
not start exactly at the same ÂM . ÂM values uctuate within a certain range, typically
◦
3 , with corresponding uctuations in the relative avalanche size. Similar results were
obtained by Pouliquen and Evesque [7,8], who related the dispersion to the initial
random packing. The same uctuation is found for  values. However, within these
uctuations, a strong linear correlation is found between the relative size of a given
 N. Nerone et al. / Physica A 283 (2000) 218–222 221

Fig. 3. Measurements of the parameters that characterize the avalanche obtained in ten experiments performed
in identical conditions. The maximum angle of stability, ÂM , and  are plotted as a function of the relative
size of the avalanche, s. Solid lines correspond to linear regressions.

avalanche and the corresponding values of ÂM and √ , as can be seen in Fig. 3. Simple
geometrical considerations lead to:  ≈ tan() = ( 3d=L)s. Roughly, the same slope
is found for the variation of ÂM with s. Therefore, ÂR is almost constant. This result
shows that once the avalanche starts at some angle ÂM the process evolves displacing
out of the packing a number of layers such that the free surface always reaches the
same angle ÂR , independent of ÂM and s values. In other words, ÂR appears to be an
intrinsic parameter of the granular medium.
In Fig 1, it can also be observed that the relative size of the big rearrangements,
s, grows exponentially with Â. In addition, these big rearrangements are triggered at
constant intervals,
Â. This result can be qualitatively explained considering inertial
eects. A big rearrangement that is triggered at a higher angle is bigger because the
grains that are set in motion acquire more momentum. In consequence, they will only
be stopped after destabilizing a larger number of grains. Note that the same interval

 separates the last big rearrangement from the avalanche. This result reinforces the
idea that the precursor of an avalanche is a big rearrangement that has acquired enough
momentum to destabilize grains from layers below. Once this threshold is overcome
the event will grow aecting, on average, 13 layers.

2. Conclusions

The experiment proves that when inertia eects are negligible, the SOC predictions
are recovered: power-law distribution of sizes and noisy structure of events. Inertia
eects play an important role when big rearrangements are considered: above some
222 N. Nerone et al. / Physica A 283 (2000) 218–222

scale inertia eects destroy the power-law behavior observed in SOC models. Moreover,
the inertia eect introduces order and destroys the noisy structure presented by the small
rearrangements: big rearrangements are observed at regular intervals. At the maximum
angle of stability, the event acquires enough momentum to destabilize grains from
layers below. This avalanching process is nally stopped when the slope of the free
surface reaches the angle of repose. This angle is a characteristic parameter of the
system in terms of avalanches and appears to be essential to characterize the surface
ow properties of a granular medium.

Acknowledgements

This work was supported by Ecos-Sud A97-E03, PICS CNRS-CONICET 561 and
SECyT TI-07.

References

[1] P. Bak, C. Tang, K. Wiesenfeld, Phys. Rev. Lett. 59 (1987) 381.

[2] P. Bak, C. Tang, K. Wiesenfeld, Phys. Rev. A 38 (1988) 364.
[3] M.A. Aguirre, N. Nerone, A. Calvo, I. Ippolito, D. Bideau, Phys. Rev. E, in press.
[4] V. Frette, K. Christensen, A. Malthe-Srenssen, J. Feder, T. Jssang, P. Meakin, Nature 379 (1996) 49.
[5] C.P.C. Prado, Z. Olami, Phys. Rev. A 45 (1992) 665–669.
[6] J. Rosendahl, M. Vekic, J. Kelly, Phys. Rev. E 47 (1993) 1401–1404.
[7] O. Pouliquen, N. Renault, J. Phys. II France 6 (1996) 923–935.
[8] Evesque, D. Fargeix, P. Habib, M.P. Luong, P. Porion, Phys. Rev. E 47 (1993) 2326 –2332.