Bed density prediction during fluidized bed scaling up using machine learning

Abstract

During the process of amplifying a fluidized bed, phenomena such as an increase in bubble size and severe particle back-mixing will occur, resulting in poor operational stability and a reduction in the mineral particle separation effect. This paper, simulated the process of bed amplification and recorded the change of bed density. It was found that as the bed diameter increases, the density fluctuation of the bed gradually decreases. A larger bed diameter will reduce the density fluctuation range, and the bed will tend to be stable. The higher the height of the medium stacking, the more intense the density fluctuation. By comparing the experimental values of bed density with theoretical calculations and predicted values using machine learning methods, the variation process of traditional density models in predicting inaccuracies was revealed, and a new approach based on “dimensional analysis + machine learning” was proposed to predict bed density.