Hydrodynamic behaviour of three-phase (gas—liquid—solid) airlift reactors

Abstract

The hydrodynamic behaviour of three-phase (gas—liquid—solid) airlift (TPAL) reactors have been studied using internal-loop reactors and the air—water—glass beads system. A hydrodynamic model has been developed which considers the reactor to be comprised of riser and downcomer regions. The uneven distribution of the solid particles between these regions leads to the effective densities of the liquid—solid mixtures being different between these regions. This effect is used to explain reactor stalling at low superficial gas velocities, and also to predict the effect of increasing concentration of solid particles and increasing solid particle settling velocities on reactor performance. The model can be used to predict both critical gas velocity for solids suspension and recirculating liquid velocity in the reactor. This hydrodynamic model has been tested experimentally using TPAL reactors ranging in diameters from 0.11 to 0.155 m, and in height from 1.0 to 2.0 m. The effects of superficial gas velocity, solid particle loading, particle diameter, draft tube diameter and draft tube height on the liquid recirculation velocity in the riser section ( U LR ) and critical gas velocity at stalling ( U Gcrit ) were investigated experimentally. U LR was found to increase with draft tube height and to decrease with draft tube diameter. Increasing solids loading or settling velocity increased U Gcrit . The set of conditions under which the TPAL reactors stalled were found to be the same whether approached via a decreasing gas velocity and a fixed quantity of solids, or a increasing quantity of solids and a fixed gas velocity. The hydrodynamics model predicted U LR well (within 30%) and U Gerit well (within 10%) when simplifying assumptions regarding the head loss coefficients and downcomer gas holdup were made. At gas velocities above 0.03 ms −1, the assumption that the gas holdup in the downcomer region was negligible (made in the absence of suitable information on this parameter) became invalid, and the hydrodynamic model predicted too-high recirculation velocities.