Академический Документы
Профессиональный Документы
Культура Документы
OF GRAVITY FLOWS
ABSTRACT
We study the flow under gravity of a granular model system submitted to shear in a rotating
cylinder. The system is confined in a vertical 2D geometry which allows visualisation of the bulk
and direct measurements of the velocity and density fields. We establish the existence of scaling
properties displayed by velocity and density profiles for a large range of different flow rates.
INTRODUCTION
525
Mat. Res. Soc. Symp. Proc. Vol. 367 01995 Materials Research Society
behaviour [8] 9] [10]. In our experiments we deal with bidimensional model system of grains,
consequently we are able to get in situ velocity measurements and we test the existence of some
general flowing behaviour.
EXPERIMENTAL RESULTS
In the experiment reported here, we use a rotating cylinder which is a device initially designed
by Franklin and Johanson [II]: it is an hollow duralumin cylinder (20 cm diameter) with glass
windows, rotating around its horizontal axis at a constant speed fl. The rotating cylinder is a very
convenient set-up to accumulate data over long time, because the heap is continuously supplied
with new particles upstream. In the present case, the container is partly filled with monodisperse
metal spheres (d=1.5 mm) confined between two vertical boundaries separated by one bead
diameter. The rotation speed is driven by a quartz clock, and is varied from 5 to 20 r.p.m which
correspond to the steady continuous flow regime.
Figure 1.Bidimensional flow of steel spheres (d = 1.5 Figure 2. Computer reconstructed image
mm). White trails correspond to the beads corresponding to Fig. 1
displacement during the opening time of the shutter.(
At = 1/250 s).
On Fig. 1 we present a zoomed snapshot of the central region of the rotating cylinder. Each
photo looks like a collection of bright trails due to the displacement of each grain during the
opening time of the shutter. The inclined flow compares well with a flow of particles on an inclined
plane, provided that the flux of matter and of momentum, brought up by the rotation process, is
negligible in comparison to the surface gravity flow. This condition seems to be satisfied on the
photos, since for this range of shutter time, the bottom bed of beads looks motionless. In the region
where particles have attained their limiting velocity, the balance between shear stress and gravity
yields:
0
xz= (mgsinO/d2)J'z+00 p(h) dh (2).
The shear stress Yxz integrates all mechanisms leading to loss of momentum, which namely are the
diffusion of momentum due to collisions, the solid friction and the inelasticity. On Fig.2 we show a
computer reconstructed image of the photo presented in Fig. 1. We designed computer routines
suited to analyse images of the flow and calculate displacement and density fields.
526
Now we present the results of a standard analysis where, for a fixed position in the laboratory
reference frame and for a given rotation velocity, we average velocities and volume fractions
over hundreds of snapshots. Fig.3 and Fig. 4, display the volume fraction profiles and the
velocity profiles obtained for typical flow rate magnitudes corresponding to 500 particles/s
(case a), 800 particles/s (case b ) and 1500 particles (case c)/s. Note that the flux is
externally driven by adjusting the rotation speed of the drum. Comparing Fig.3. and Fig.4, we
notice that the flowing layer has a finite thickness (typically around 10 beads diameter) which
depends on the flux. The motion appears frozen for volume concentrations of the order of Pc
=-0.8, which is clearly smaller than the volume concentration of the 2D triangular compact
lattice (Pmax = 0.91). This gives an experimental approximation of the Reynolds dilatancy
threshold. Furthermore, a remarkable feature is that, in the average, the velocity gradient
looks weakly dependent on the flowing layer depth, and is roughly of same order of
magnitude: VV = 24 s-1.
8.2 £6.4
A 0 0.6 g e.2 0.4 6.6 6.6 1.6
0
A 4 A
1 ,- O
i -. s a• ,' -6.6. 4 1
a A
t4
4 4
Figure 3 . Velocity profile of the 2d flow of steel beads Figure 4 Corresponding volume fraction profiles
(d = 1.5 mm) (averaged on 100 snapshots)) [1i(case a (same experimental conditions)
), A(case b ), 0 (case c),see text.]
We claim that the previous analysis has a major drawback. The averaging procedure ignores
that for 2D experiments, fluctuations are of large amplitudes. Consequently, the standard
averaging procedure mixes spatial domains with different physical significance and may
smooth out essential pieces of information. Now we describe an alternative analysis
procedure based on the hypothesis that every snapshot represents some mechanical steady
state controlled by three variables: VV the velocity gradient, J the particle flux, 0 the
instantaneous flowing angle and pc the maximum volume fraction of the flowing layer. These
parameters are not necessarily independent but can be measured independently for each
snapshot. There is also an implicit hypotheses on the existence of a time scale separations
between the process controlling the shearing and the process controlling the fluctuations. For
each picture, VV, J and 0 are measured as well as zc the maximal depth of the flowing layer.
527
A flow depth scale H is computed: H = [J/VV/pc] 1 /2. On fig 5 and 6 we display the rescaled
velocity and density profiles for a large collection of snapshots at three rotation velocities.
%
Io - '.6 13
#a
1.6' jA 4Q.L.
0. .
V(2)
"V
., A
REFERENCES:
lJaeger,H.M. and Nagel,S.R. Science, 255: 1523 (1992)
2 Rajchenbach,J., Cl6ment,E., Duran J. and Mazozi T.in Scale Invariance,Interfacesand non-
equilibrium dynamics, A.McKane Editor, Plenum, New York (1994)
3 Rajchenbach,J. Phys.Rev.Lett.65: 2221 (1990).
4 Reynolds,O Phil.Mag. Ser.5,20 : 469 (1885)
5 Bagnold,R.A. Proc.Roy.Soc.A225: 49 (1954)
6 Jenkins J.T. and .Savage S.B, J.Fluid.Mech. 130, 187, (1983),
7 Lun C.K.K. Savage S.B., Jeffrey D.J., and Chepurniy N., J.Fluid.Mech. 140, 223, (1984),
8 Savage,S.B. J. Fluid. Mech. 92: 53 (1979).
9 Hanes,D.M. and Inman,D.L. J.Fluid Mech. 50 : 357 (1985).
10 Drake,T.G. J. of Geoph. Res. 95, B6: 8681 (1990).
11 Franklin,F.C. and Johanson,L.N. Chem.Eng.Sci. 4, 119 (1959)
528