Monitoramento em tempo real e simulação de controle da umidade de pós produzidos pela secagem de pastas em leito de jorro

Abstract

The direct control of the product moisture content in dryers would be more interesting. However, as it happens with other processes, it is a concentration control, which is commonly dependent on slow or expensive procedures for industrial purposes. One way to reduce these costs is by using other sensors, whose response is faster, which have a lower cost and which can supply the required information for a physical-mathematical model which represents the process. In this context, the aim of this work was to use a virtual sensor for the online measurement of the moisture of the powder produced on a spouted bed used for drying of pastes. In order to do so, firstly water evaporation and whole milk drying experiments were carried out, for the determination of the dynamic behavior of the analyzed drying process. Afterwards, a hybrid CST/neural model was satisfactorily fit to experimental data for posterior use as a part of the virtual sensor. The sensor was able to estimate the powder moisture content, even when the dryer was submitted to open-loop disturbances. The virtual sensor was then used as a part of a simulated inferential controller for the powder moisture content. A classic PI controller was used. The controlled variable was the powder moisture content (inferred by the hybrid neural model) and the manipulated variables were those related to heat transfer. Two similar situations were simulated: one which uses the air inlet temperature as the manipulated variable and other which uses the heating rate as the manipulated variable, the latter being the manipulated variable in the real dryer. In these simulations, the dryer was submitted to simulated set point tracking (servo controller) and disturbance rejection (regulatory controller). Both simulations resulted on good closed-loop performances, without offsets, overshoots or oscillations, attaining to the set point condition within a short time interval.