The increasing feed-in of electrical power from renewable energy sources has a far-reaching impact on the entire energy supply. Most notably, conventional power plants need to adopt a more flexible mode of operation due to the increasingly fluctuating residual load. The notion of operational flexibility embraces several aspects; one of the most crucial of these is fast and precise load cycling operation. The focus of this thesis is on combined cycle gas turbine plants as this technology is widely used all over the world due to its high efficiencies and low specific emissions.
Among the multitude of approaches towards increased operational flexibility of both new builds and existing units, improvements of the instrumentation and control system (I&C) are particularly advantageous. They require lower investments than constructional modifications and can be implemented and commissioned quickly. The extended unit control concept that is described in this thesis falls into this category. It aims at enabling faster load changes while allowing enhanced control performance. Furthermore, disturbances in subordinated control loops will be limited to a minimum, resulting in smoother overall process behaviour. From a control engineering perspective, load changes of power plants are best handled by feedforward control. Therefore, within the scope of this thesis, a feedforward control path is added to the feedback control structure of the existing unit control system of combined cycle gas turbine plants. This feedforward control path consists of two major components:
At the core of this feedforward control path is a model-based feedforward control algorithm that has been calculated following the methodology of flatness based control. Knowledge of the dynamic behaviour of the power plant process is taken into account within the control algorithm by means of a dynamic model. The feedforward control enables faster load changes, leads to improved load tracking and thereby reduces feedback control action. Furthermore, the model based character of the feedforward control leads to improved coordination of all control variables which is relevant for multiple-input multiple-output type systems in the sense that undesired interaction between controlled variables are effectively reduced.
The second crucial component that is closely related to the feedforward control algorithm is a new type of trajectory planning. Trajectory planning takes place before the control itself and consists of the planning of suitable set-point trajectories based on the target values of the controlled variables. Ansatz-functions for set-point trajectories that provide adequate degrees of freedom are required in order to take the process dynamics into account, as was done in the design of the feedforward control algorithm. Moreover, certain boundary conditions, imposed by the feedforward control, need to be considered. Both conditions are met by so-called Bézier-Curves which are additionally beneficial due to their simplicity and numerical stability.
This thesis describes both the extended unit control concept in detail as well as the results of numerical simulations based on a nonlinear process model. The comparison between the extended unit control concept and the state-of-the-art one is particularly stressed in this context. Moreover, the impact of the new trajectory planning strategy is emphasized. This comprises both an optimal trajectory planning study for load changes as well as predictive online trajectory planning. The latter is necessary in order to also cover the provision of so-called Frequency Restoration Reserve (FRR) by the combined cycle gas turbine plant. The simulation results reveal the considerable foreseen improvements in terms of control performance as well as the positive impact of suitable set-point trajectories on the required control effort.