Computational fluid dynamics (CFD) for fluidization is reaching maturity (R
oco, 1998). It has become common practice to compare time-averaged solid ve
locities and concentrations to measurements of fluxes and densities. The dy
namic behavior of the riser, however, has not been previously compared to e
xperiments. This article shows that the dynamics of solids flow in the rise
r is in the form of clusters, but the time-averaged particle concentrations
and fluxes give us the core-annular flow regime in agreement with measurem
ents. The computed clusters, which are essentially compressible gravity wav
es, produce major frequencies of density oscillations in agreement with mea
surements. The model and the CFD code compute granular temperature distribu
tions, agreeing qualitatively with data. For volume fraction around 3-4%, w
hich is the average particle concentration in the riser, the computed visco
sity agrees with our experimental measurements.