MODEL PREDICTIVE MASTER CONTROL OF ONCE-THROUGH CFB UNIT

摘要

Metso Automation is providing control system for PKE Lagisza power station in Poland. This plant will be the first supercritical CFB unit in the world. As the CFB combustion process is dynamically slower than the traditional PCF-combustion and the once-through process does not have the steam storage capacity that the drum boilers have, the combination is challenging from the control response point of view. Prompted by this challenge we decided to develop a completely new master controller for this type of plant.rnTo help us develop and test the new controller, we built a hybrid simulator of the plant. Structure of the simulator is based on simple physical models, but the dynamics are based on identified dynamics of compatible sub-processes. As the actual plant has not been built yet, we used data from an existing CFB-boiler for combustion dynamics and data from another once-trough boiler for water-steam cycle dynamics. The simulator principle and modeling approach are briefly explained.rnThe developed controller is based on model predictive control theory. Taking into account our long experience with the traditional controls we aimed to combine the best of both approaches. This paper describes the controller structure and the selected manipulated and controlled variables. The goals for main operating parameters are derived from plant operation mode and process design parameters.rnResults of various simulation tests are presented and the results compared with the dynamical requirements of the plant. From these results we draw conclusions on applicability of this controller for the real CFB once-through unit.