Experimental investigation into the kinetics of Falcon UF concentration: Implications for fluid dynamic-based modelling

Abstract

Centrifugal gravity separators, such as the Falcon UltraFine (UF) concentrator, are the most common gravity concentration techniques used for fine particles processing. Hence, understanding the kinetics and separation mechanisms at play within these separators is of paramount interest. Recent research yielded a predictive physical model for the Falcon UF which however does not explain some results obtained with industrial ores. The Falcon UF kinetics have been investigated through the processing of fine-grained ores from the Altenberg tin deposit (Germany), the Tabuaço tungsten deposit (Portugal), a synthetic iron ore as well as results from previous studies on kaolin residues. Results have shown an evolution of Falcon UF performance with time/feed mass in contradiction with the stationary separation hypothesis on which the physical model was based. In terms of Falcon UF separation timing, four phases can be distinguished. First, upon initial feeding of the bowl, particles are trapped or rejected depending on their settling velocity. It yields a relatively ineffective selection according to density so that only ultrafine particles are ejected from the bowl, resulting in the quick growth of the concentrate bed. When the bed reaches a critical size, recovery and enrichment continue to increase through selective resuspension phenomenon that favours the concentration of dense particles and the ejection of larger particles. This way, the bed builds up while the content of concentrate bed surface evolves until resuspension balances the stream of dense material reaching the bed and recovery drops. The evolution of partition curves over time confirmed the low recovery of ultrafine particles during the whole operation but also showed a decrease of coarse particles recovery with time. It suggests that the second separation mechanism is less sensible to particle size compared to the first one and that size even has a negative impact on recovery. Furthermore, erosion figures in furrows are observed in the concentrate bed which may play locally an active role in the separation. These observations suggest that two separation mechanisms are at play. Firstly, differential particles settling within the flowing film which is already accounted for in the existing physical model. Secondly, resuspension of particles from the concentrate bed by the action of a lift force acting preferentially on coarse particles deposited at the surface of the bed and resulting in the rejection of coarser and lower-density particles. The addition of a lift force component to the existing model is discussed and a resuspension criterion is proposed as a guidance of the physics involved in this second separation mechanism. Future developments will require a dynamic model which would need to integrate the evolution of the concentrate bed content over time.