Use of Statistical Methods of Experimental Design in Optimizing Hydrometallurgical Processes

Abstract

Many hydrometallurgical processes take place under conditions where use of the theory of the thermodynamics of aqueous solutions is difficult or practically impossible. This can make it necessary to try to optimize the working of the process by trial and error during its development. While working on attempts to use a computerized Databank system¹ to increase the possibility of using thermodynamic information as a guide to process development, it was decided to investigate how the use of statistical methods of experimental design could be employed for this purpose also. Dr. A. J. Monhemius had developed an interest in that subject while working in Brazil and guided Dr. M. J. Pinto Dias² in the early stages of her work on its application to the Jarosite process. Dr. Styliani Agatzini³ took the theoretical treatment further during her work on the Goethite process and subsequently the method has been used by many workers in the Hydrometallurgy Group. All are greatly indebted to Dr. A. C. Atkinson of the Department of Mathematics, Imperial College for the special course of lectures which he gave for them and for his assistance and advice over a long period. The role of the present author has been to watch skeptically the attempts of a young, very gifted group of workers to apply rigorously mathematical methods to the design of hydrometallurgical processes. It seems now that they have been successful and any questions on statistical matters should be directed to them.