COMPUTER DESIGN PROCEDURE OF LIQUID-LIQUID PULSED PLATE EXTRACTION COLUMNS

Dispersed phase holdup is a major determinant of the mass transfer area and hydrodynamics in a column. None of the existing holdup correlations is satisfactory for design purposes. A new correlation which includes the effect of solute transport was developed in this work with an average error of 28.9% from experimental data.;The internal motion of drops correlates strongly with size. Small drops tend to behave as rigid particles with very little internal circulation; drops of intermediate size tend to assume laminar internal circulation; large drops have turbulent internal circulation.;Most previous methods for the design of extraction columns have treated the drops as having a single mass transfer mechanism. A design approach is proposed here that allows different mass transfer mechanisms for two size fractions of the dispersed drops. Rigid drop and turbulent internal circulation models are incorporated into a forward mixing model for the pulsed plate extraction column.;A key variable in the design is the transition drop size that separates the two mass transfer mechanisms. An empirical correlation is presented for the transition drop size as a function of the operating conditions, the geometric characteristics of the pulsed column, and the properties of the liquid system.;Backmixing impairs the performance of extraction columns. It decreases the mass transfer driving force by formation of concentration jumps at the stream inlet and by longitudinal transport of the solute. None of the existing correlations of backmixing for the continuous phase is adequate for design purposes. Dimensional analysis was used to develop a new correlation with an average error of 21.4% from experimental data.;A design approach is presented using a forward mixing model combined with the developed correlations for holdup, longitudinal mixing, and transition size. For all of the 299 sets of data that were found, the design method predicts the column stage number for specified recoveries with an average error of 19.2%.