On the identification of fluiddynamical phenomena in pulsed extraction columns by means of radiotracers

A radiotracer measuring technique is presented, by means of which the convective fluid velocity and the intensity of mixing in each one of the two fluid phases, as well as the dispersed phase holdup can be determined and monitored on-line without disturbing the process. A mathematical treatment of the experimental RTD data based on the linear systems theory for the determination of the mixing parameters leads to integral equations whose solution represents an ill-posed problem. The usual approach for the direct solution of the mixing problem by means of the numerical optimization of a parametrical model leads to interpretation errors. The reasons for this are discussed in the present report. An alternative FFT method for the solution of the convolution equation is presented. Through this method, together with a non-parametrical statistical test, it is possible to determine the limitations of frequently used axial dispersion models for the design of pulsed extraction columns. This model fails in the description of the experimentally observed fluiddynamical behaviour of the dispersed phase. An alternative stochastical mixing model is suggested which fits the RTD data of both phases. On the basis of a statistical design, the main effects and interactions of the total flow rate, phase ratio, pulse amplitude and frequency on the afore mentioned fluiddynamical variables for the 3N HNO3/30% TBP - kerosene system are analyzed with two of the most common plate geometries (sieve and nozzle plates) found in the reprocessing of nuclear fuels. The experiments show a marked effect of the plate geometry on the analyzed variables. They refute also the admissibility of the generalized practice of representing the dispersed phase holdup and the intensity of mixing depending on the product of pulse amplitude and frequency as a determining factor of the fluid dynamics in a pulsed extraction column. (orig./HP)