Modeling and Model Predictive Control of a Conveyor-Belt Dryer: Applied to the Drying of Fish Feed

Abstract

Moisture control of industrial processes is often difficult and complex. The product composition and moisture content changes as disturbances affect the upstream production processes. When there are changes in the product content, simple control configurations may not work due to changes in the drying characteristics of the product. This thesis presents a general conveyor-belt dryer model describing a six-zone multiple-pass dryer, accounting for the falling rate drying period, input disturbances, conveyor-belts with different belt speeds and product bed heights. In addition, a description is presented of how linear input dynamics can be included in the dryer model. Furthermore, open loop simulations are performed in order to investigate the behavior of the model. The model is linearized and reduced, in order to be utilized in a model-based control solution (MPC), where stability and feasibility is ensured through an algorithm based on known techniques within the field of model predictive control (e.g. infinite horizon optimalization, target optimalization routine, soft constraints and a Kalman filter). Several closed loop simulation examples are presented, illustrating reference steps and disturbance rejection. Furthermore, a modeling error is introduced in order to investigate the limitations imposed by model uncertainty. Finally, a basic control solution (PI control) is compared to the model-based control.