Study of the Performance of a Large Mechanically-Agitated Extraction Column

Abstract

The literature relating to mechanically-agitated extraction columns and relevant phenomena eg. droplet break-up and coalescence, axial mixing, single drop mass transfer mechanisms and flooding/phase-inversion, and related mathematical models have been reviewed. An experimental study was performed of the hydrodynamic and mass transfer characteristics of a 0.45m diameter, 4.3m high Rotating Disc Contactor. The system used was Clairsol-350 (dispersed)-acetone-de-ionised water. Hydrodynamics were studied with mutually-saturated phases and in the presence of mass transfer. Drop size distributions under all conditions were best described by the Mugele-Evans Upper-Limit function. However with mass transfer from the dispersed to continuous phase d35 was, on average, approximately 22.8% greater and hold-up was 57% less than for non-mass transfer conditions. Therefore data obtained under non-mass transfer conditions should be applied with caution in extractor design. The phenomenon of flooding, rather than phase inversion, was the best criteria for limiting capacity. Operation at 80% of flooding, resulted in increased hold-ups and higher volumetric overall mass transfer coefficients (Ka)exp. Increasing rotor speed, up to a maximum at which a haze was generated, resulted in an increase in overall mass transfer coefficient (Ka). To calculate the overall mass transfer coefficient (Kexp) the mean driving force was calculated from the experimentally-determined concentration profile along the column using Simpson's Rule. Theoretical overall mass transfer coefficients were calculated using the drop size distribution diagram to determine the volume or area fraction of stagnant, circulating and oscillating drops in the sample population in conjunction with single drop mass transfer correlations. Both approaches predicted considerably lower mass transfer rates than were obtained experimentally. This was attributed to the drop swarm, in reality, oscillating due to drop interaction and because local drop velocities were much higher than the slip velocity used to calculate Kcal. Assuming that all the drops were oscillating, with mass transfer from the dispersed to continuous phase, the Rose and Kintner correlation gave the best approximation to the overall experimental mass transfer coefficient. A novel method is proposed for calculation of axial mixing coefficients based on the experimentally-determined concentration profile in the continuous phase.