Multivariable generalized predictive control of a boiler system

An application of a multiloop generalized predictive control (GPC) scheme to achieve self-tuning control of superheat pressure and steam temperatures in a 200 MW power station drum boiler are presented. Controllers have been designed and evaluated using a detailed nonlinear boiler model which is well established and validated. Results illustrating the performance of the plant with GPC are presented and compared with conventional PI control. The results show that substantial improvements in control can be achieved with the GPC. Steam pressure and temperature variations are greatly reduced, without offsets, and with less controller activity.>