Towards the Optimization of Flotation Circuits with Statistical Design of Experiments

"Flotation is a complex and highly multivariate process. Using the conventional OFf (one-factor-at-a-time) technique, the optimization of flotation circuits through laboratory or plant testwork is difficult, time-consuming and therefore, expensive and often results in marginal improvements. The optimal operating conditions can be more efficiently estimated when a systematic approach like SDE (Statistical Design of Experiments) is taken.The concentrator of Noranda's Brunswick Mining Division (near Bathurst, New-Brunswick) is well known for the complexity of its differential copper-lead-zinc flotation circuits. One of them, the chalcopyrite-galena separation (or CuSep) circuit, is impacted by more than 20 variables.Statistical Design of Experiments (SDE) techniques were applied at the laboratory scale to improve the performance of Brunswick's CuSep circuit. Screening tests were conducted using Fractional Factorial designs while optimization tests were carried out based on Response Surface Modelling techniques like the Central Composite design. Every series of tests, without exception, revealed new information and process insights which led to significant circuit performance improvements in the plant So far, the metallurgical gain may be as high as four percentage points of copper recovery (in the CuSep circuit). Potential cost savings of expensive consumables (steam, activated carbon) have also been identified but will require additional circuit modifications before being realized.IntroductionIn spite of its apparent simplicity, flotation is a complex three-phase (solid, liquid and gas) separation process. Fundamentally, it is a highly multivariate process. Industrial flotation circuits are commonly affected by 10 or more process variables."