Offset-free Nonlinear Model Predictive Control of A Drum-boiler Pilot Plant

Abstract

In this paper, we present the offset-free Nonlinear Model Predictive Controller (NMPC) for the control of drum-boiler pilot-plant at College of Engineering Pune (COEP), India. We consider a laboratory-scale pilot plant available in the process control lab designed to generate steam up to 0.0083 kg/s at the pressure 0.4 MPa. To design an NMPC, we developed a Multi-Input Multi-Output (MIMO) nonlinear model of the pilot plant. The developed model is based on the first principle (mass and energy balance) equations where the thermodynamic properties and its partial differentiation values for the desired operating region are calculated by IAPWS-IF97 and transformation rule. The developed model is validated through several experiments on the pilot plant and the same is used as a prediction model for NMPC. Further, we designed an Extended Kalman Filter (EKF) for offset-free NMPC (disturbance modeling and state estimation) to handle any disturbances in the output measurements. The designed offset-free NMPC strategy is then implemented for the nonlinear MIMO model of the boiler to control the desired pressure by regulating three inputs. Closed-loop performance of the offset-free NMPC is evaluated and compared with NMPC without state estimation. Presented results show that the designed offset-free NMPC performs well and successfully handle disturbances in the output measurements.