Soft-sensor model design for control of a virtual conveyor-belt dryer of mate leaves (Ilex paraguariensis)

The paper presents an empirical soft-sensor model to control the moisture content of mate leaves in the discharge of a virtual conveyor-belt dryer. The empirical model is applied to infer the unmeasured control variable based on virtual transient readings of temperature in the solid phase of the dryer. The inferential control uses a proportional–integral–derivative (PID) controller, which manipulates the conveyor speed to keep the outlet moisture content of mate leaves between 0.024 and 0.034 (dry basis). The parameters of the soft-sensor model were tuned on a set of around 2000 experimental data of surface temperature and moisture content of mate leaves. These results emerged from batch drying experiments at between 70 and 120 °C for up to 9600 s using leaves with initial moisture content varying from 0.14 to 1.7 (dry basis). Open- and closed-loop responses of moisture content and temperature of the mate leaves were obtained by simulation with a dynamic model. It was represented by a system of two partial differential equations derived from energy and mass balance for water in the solid phase of the dryer. Closed-loop simulations in the presence of arbitrary process disturbances confirmed the reliability of the inferential designed controller.