SIMULTANEOUS OPTIMIZATION OF FLOTATION COLUMN PERFORMANCE USING GENETIC EVOLUTIONARY ALGORITHM

Column flotation is a multivariable process. Its optimization guarantees the metallurgical yield of the process, expressed by the grade and recovery of the concentrate. The present work aimed at applying genetic algorithms (GAs) to optimize a pilot column flotation process which is characterized by being difficult to be optimized via conventional methods. A non-linear mathematical model was used to describe the dynamic behavior of the multivariable process. The solution of the optimization problem using conventional algorithms does not always lead to convergence because of the high dimensionality and non-linearity of the model. In order to deal with this process, the use of a genetic evolutionary algorithm is justified. In this way, GA was coupled with the multivariate non-linear regression (MNLR) of the column flotation metallurgical performance as a fitting function in order to optimize the column flotation process. Then, this kind of intelligent approach was verified by using mineral processing approaches such as Halbich's upgrading curve. The aim of the optimization through GAs was searching for the process inputs that maximize the productivity of copper in the Sarcheshmeh pilot plant. In this case, the simulation optimization problem was defined as finding the best values for the froth height, chemical reagent dosage, wash water, air flow rate, air holdup, and Cu grade in rougher and column feed streams. The results indicated that GA was a robust and powerful search method to find the best values of the flotation column model parameters that lead to more reliable simulation predictions at a reasonable time. Based on the grade-recovery Halbich upgrading curve, the MNLR model coupled with GA can be used for determination of the flotation optimum conditions.