Design of supervisory cascade model predictive control for industrial boilers

Abstract

This paper aims to improve the performance of boiler by applying cascade model predictive control to multiple control loops including water level, drum pressure and stream temperature. Since proportional-integral (PI) controllers do not provide satisfactory responses, model predictive control (MPC) is used to improve the performance of boiler in a framework of supervisory cascade control. We consider a linear time invariant (LTI) model of boiler system in which inputs are feedwater mass rate, fuel mass rate and attemperator spray mass rate, and outputs are water level, drum pressure and stream temperature. We present design of cascade control which PI controllers act as inner loop control and MPC acts as outer loop supervisory controller to improve performance of boiler. The design compares two control techniques, namely, standalone MPC and supervisory cascade MPC. We simulate step responses and compare performance in terms of maximum overshoot, settling time, and steady state error as well as bound of control inputs. Numerical results reveal that supervisory cascade MPC gives better performances than standalone MPC in all aspects.