A first principles-neural networks approach to model a vibrated fluidized bed dryer: Simulations and experimental results

This article presents the modeling and simulation of a batch pilot-scale vibrofluidized bed dryer. The model considers the effect of back-mixing by establishing interconnected drying zones. The model's equations consist of the mass and energy balances for each zone in the solid phase, while a complete mixing is assumed in the gas phase. The drying and heat transfer parameters are correlated with the operating conditions by means of three neural networks that have been adapted from data obtained experimentally. The system of algebraic-differential equations provides the solid's moisture content and temperature profiles as a function of time. The model was validated by experiments with turnip seeds. Good fit was obtained using only four drying zones.