Falcon concentrators are commonly used in the mining industry to separate ores. Recent research yielded a predictive model for their separation performance in the form an analytical expression, based on a mechanistic understanding of the separation physics in these devices. While giving good results for lab-scale concentrators, in the form it is given, this model is not ready for industrial applications because of two limitations: its strong dependance on high quality measurements of size-density distributions, and its sensitivity to the discretization of data. These two shortcomings are solved by representing the distributions using continuous polynomials functions, in order to produce a robust model that makes it possible to validate the predictions for industrial scale Falcon bowls. (C) 2014 Elsevier Ltd. All rights reserved.