Cluster
Figure 2 displays the boxplots of the local cluster data, specifically
1188 datasets from the fast fluidized bed 16,17 and
378 datasets from the turbulent one 20. The fiber
optic probe was used for cluster detection, with each measurement
lasting 30 s and cluster identification enabled using wavelet
decomposition. Cluster probability is the fraction of time instances
whereby cluster is detected, and is bounded between 0 and 1, while
cluster duration and cluster frequency reflects respectively cluster
size and number of clusters per unit time. Using a two-sample t-test,
the null hypotheses were rejected for all three characteristics,
indicating the data from the fast and turbulent fluidized beds were from
populations with different means within the 95% confidence level.
Relative to the fast fluidized bed, the turbulent bed withUg values of an order-of-magnitude lower gave
higher cluster probability, lower duration, and much higher frequency.
In particular, the higher cluster probability for the turbulent bed was
tied to the significantly higher number of clusters (i.e., cluster
frequency). The lower cluster probability and frequency in the fast
fluidization regime are tied to lower bed densities than the turbulent
bed, since by definition the former is classified under lean-phase
fluidization while the latter dense-phase fluidization1,2,9. The higher cluster duration in fast
fluidization is presumed to be due to the less chaotic hydrodynamics
relative to the turbulent bed, because the latter is dominated by bubble
break-up 2 and is known to exhibit higher standard
deviations of local voidage fluctuations 11. Another
study postulated that cluster sizes adjust to balance the gas drag force
and gravity, thus smaller clusters were observed to exist in the
turbulent bed compared to the fast one 13, which
agrees with the comparison here.