Due to the many inputs, outputs and constraints, the control problem is well-suited to solution by a multi-variable model-predictive control (MPC) (ABB’s “Advanced Predict & Control – APC”).
- Multi-variable MPC models “know” the dynamic behavior and the interactions between process quantities, by learning from a short history of a range of boiler operations (steady state and ramping).
- MPC controls all variables simultaneously, which is different from traditional base controls (e.g. PID control), where each controller has one input and one output.
- MPC controller, when active, sends set points to the base controls and can be independently activated or deactivated.
- The advantages:
o Greatly simplified control structure
o Much easier tuning
o Smoother transition from one constraint to another
o Superior performance over base controls during load and fuel quality changes. |