Nonlinear model predictive control of an oxy-fuel combustion boiler for regulating the excess oxygen percentage in the flue gas

In this paper, a nonlinear model predictive control (MPC) of a water-tube boiler system with oxy-fuel combustion process in an electric power plant is investigated. The control objectives are to maintain the water and pressure levels in the drum, the steam temperature in the secondary superheater, and the oxygen percentage in the flue gas. Upon the riser and drum models developed by Åstrom [5], mathematical (nonlinear) models of the primary and secondary superheaters, attemperator and oxy-fuel combustion process are first developed. The introduced seven state variables are the water volume, steam pressure, and steam volume in the drum, vapor fraction in the riser, temperatures in the primary and secondary superheaters, and attemperator temperature. The five input variables are the feed water flow, steam flow, fuel flow, oxygen flow, and spray flow. The four output variables are the steam pressure and water level in the drum, secondary superheater temperature, and excess of oxygen in the flue gas. Since the water and steam volumes in the drum, vapor fraction in the riser, and the temperatures in the primary superheater and attemperator cannot be directly measured, two nonlinear state estimators (extended Kalman filter and particle filter for comparison purpose) are designed. Simulation results of the designed nonlinear MPC algorithm are provided.