Ind. Eng. Chem. Process Des. Dev. 1981, 20, 533-538
533
Solids Mixing in a Fluidized Bed with Horizontal Tubes Oto Sltnal CSIRO, Division of Mineral Engineering, P.O. Box 3 1.2, Clayton, Victoria 3168. Australla
A model of solids mixing in a fluidized bed containing horizontal tubes is developed from new data on mixing of magnetic tracer in a 1.2 X 1.2 m square fluidized bed of 700-pm partlcles of silica sand. In the model the dense phase conslsts of an upward moving drift phase and two downflowing phases, the one at the vessel wall parallel with the tubes moving faster than the other. Solids can exchange between the phases. Of the nine parameters appearing in the governing difference-differential equations four have dominant influence on the concentration histories: the velocities of the two down-flowing phases, the solids exchange coefficient, and the volumetric fraction of the drift phase. The model interpretation of the mixing data at gas velocities less than 1 m/s revealed low circulation of solids and high solids exchange rate for the drift phase, both relevant for the heat and gas mass transfer rates in the bed.
Introduction Mixing of solids is of particular importance in large-scale atmospheric fluidized bed combustors (AFBC). For the efficient operation a high thermal load (about 3 MW per square meter of bed) and a small air pressure drop through the bed (less than 50 cm of water) require high fluidization velocities (up to about 3 m/s) and shallow beds (about 1-1.5 m) almost filled with the heat exchanger tubes. Particle sizes of the bed materials-refractory sand, lime or agglomerated ash, and coal-must be about 1 mm or larger to avoid excessive entrainment. Feeding the coal into a bed of several hundred square meters of area is an intricate and costly operation (Biswas and Baley, 1978). Early studies of solids mixing in the fluidized beds related to the experimental measurement of the heat transfer rate (Leva and Grummer, 1952; Lewis et al., 1962) and to the catalyst mixing patterns in commercial catalytic cracking units (Singer et al., 1957; May, 1959). The models of axial dispersion of solids using constant dispersion coefficients (Gilliland and Mason, 1949) were not in satisfactory agreement with the data (Tailby and Cocquerel, 1961). May (1959) recognized the reason for this being the presence of gross circulation pattern in the bed. More successful models were based on the solids turnover (Leva and Grummer, 1952; Talmor and Benenati, 1963) or on the convection flow of solids in the regions of rising and descending streams with an exchange of solids between the regions (Bailie, 1967; van Deemter, 1967). The rising streams in the circulation models corresponded to the bubble wakes and clouds and the descending stream to the rest of dense phase (Kunii and Levenspiel, 1968; Potter, 1971). Marsheck and Gomezplata (1965) showed by the direct measurement of solids trajectories that, generally, particles flow up in those areas where the bubbles rise and flow down in the surrounding regions. This correlation of the bubble and particle flow patterns has been since well established (Miyauchi and Morooka, 1969; Ohki and Shirai, 1975; Werther, 1976). Sizeable work has been done to elucidate the solids mixing mechanisms in fluidized beds of small (less than about 300 pm) particles (Lewis et al., 1962; Rowe et al., 1965; Merry and Davidson, 1973; Whitehead et al., 1976). Talmor and Benenati (1963) recognized early that the particulate solids size has a strong influence on both the mechanism and the intensity of mixing. Abraham and Resnick (1974) showed experimentally that for about 0.50-mm particles the total solids volume moved upward by bubbles via two mechanisms-wakes and drift-equal 0196-43051aiii i20-0533$01 .2wo
Table I. Properties of Bed Materials silica sand particle density, kg/m3 bulk density, kg/m3 size distribution, 5% wt >1170 pm 1170-830 830-590 590-420 420-300
