Application of an Artificial Intelligence Technique to Improve Purification in the Zone Refining Process

Abstract

A combined theoretical and experimental approach was undertaken to quantitatively determine the influence of a variable solute distribution coefficient, k, on impurity distribution in multipass purification by zone refining. Axial impurity profiles have been experimentally determined for a number of zone passes. It has been shown that the adoption of a variable-k approach in the simulation of impurity profiles during different zone passes is generally much closer to the experimental profiles than the usual adoption of a constant k. An artificial intelligence technique interacts with the numerical model to determine the best molten zone size in each pass in order to provide maximum purification.