Sensor-Less Predictive Drying Control of Pneumatic Conveying Batch Dryers

Abstract

This paper presents predictive drying control design for a lab-scaled industrial pneumatic conveying dryer (PDC) involves with continuous/ batch processing of powder materials. The model predictive control (MPC) is an established method for drying control in various drying applications, such as fluidized bed dryers, rotary dryer, infrared dryer, timber dryers, baker’s yeast dryers and so on. But the predictive control of PDCs have not been studied in the literature, however, these dryers are widely used in food, agriculture, and chemical industries, particularly suitable for batch processing of fine grained materials. The unavailability of any suitable control oriented first principle’s model of these dryers make the predictive control design and implementation issues more challenging. Existing control methods for similar drying applications use outlet material moisture as a main control variable, online measurement, of which is difficult, costly, and unreliable due to the involvement of materials in granular/powder form. In the present contribution, an innovative control oriented model of the dryer is derived from first principle’s encompassing a soft sensor-based online powder-moisture measurement procedure replacing the physical moisture sensors. The proposed physical sensor-less powder moisture control strategy stands on the traditional two-layer predictive control paradigm involving detection of an economically best operating point for batch dryer operation by optimizing the various process economic objectives followed by employing a suitable state space MPC (SSMPC) law for steering the process to operate at economically best operating point. The developed control strategy has been implemented and tested under practical settings and shown its effectiveness in improving the drying performance and product quality compared with an inbuilt auto-tuned proportional integral plus derivative controller of Honeywell make HC900 programmable logic controller.