Optimal load cycling of large steam turbines

The availability of inexpensive base load power from independent power producers and nuclear plants has caused electric utilities to cycle their large steam turbines in an effort to reduce overall generation cost. In this paper, the authors present a novel method of controlling a turbine during load cycling. The approach combines heuristic turbine control rules with a rigorous optimal control algorithm. Fuzzy logic encodes these heuristic rules and is used to scan over long term load profile information to prioritize system performance priorities based on both present and future cycling duty. A model predictive control algorithm utilizes these priorities in the form of cost function weights to determine short term optimal setpoint trajectories for boiler pressure and turbine governor valve position. This hybrid fuzzy logic model predictive controller yields both excellent load tracking and minimal long term operational costs.<>