A numerical program has been developed tc, simulate an assembly of ine
lastic, frictional hard spheres inside a control volume undergoing a s
teady-stale rapid Couette Bow induced by the top and bottom bumpy wall
s. The bumpy walls are made of hemispheric particles fixed onto flat p
lates. The flow particles can collide with the wall particles and the
exposed flat areas of the walls. The macroscopic flow properties are f
ound to depend on a number of material and geometric properties of the
granules, the bumpy walls, and the control volume. These properties i
nclude the overall solids fraction of the system, the height of the sh
ear gap, the wall-particle concentration, the wall-particle distributi
on, the diameter ratio of the wall particle to the flow particle, the
coefficients of restitution, the friction coefficients, and the sticki
ng tangential restitution coefficients between the flow particles, the
wall particles, and the Rat wails. A parametric study is undertaken t
o examine the effect of some of the interesting factors identified abo
ve. A new definition for the slip velocity yields positive values cons
istently, and it represents a significant improvement over the previou
s ones. By exposing the flat areas of the bumpy walls for collisions,
the transfer of energy and momentum from the driving surfaces to tile
flow medium ran: be enhanced. Depending on the wall-particle distribut
ion, there exist optimal wall-particle concentrations at which the str
esses may be maximized or the slip velocities may be minimized. for he
mispheric wall particles arranged in an equilateral triangular lattice
, the optimal wall-particle area fraction for maximizing the stresses
is about 0.44 while the one for minimizing the slip velocity is about
0.36. The simulation results also show that there exists for gravity-f
ree Couette flow of inelastic, frictional spheres a critical solids fr
action of about 0.5 beyond which the stresses are found to decrease wi
th increasing solids concentration. In general, there is reasonable ag
reement between the simulation results for stresses and the experiment
al measurements. (C) 1996 American Institute of Physics.