Load Frequency Model Predictive Control of a Large-Scale Multi-Source Power System

Abstract

With increased interests in affordable energy resources, a cleaner environment, and sustainability, more objectives and operational obligations have been introduced to recent power plant control systems. This paper presents a verified load frequency model predictive control (MPC) that aims to satisfy the load demand of three practical generation technologies, which are wind energy systems, clean coal supercritical (SC) power plants, and dual-fuel gas turbines (GTs). Simplified state-space models for the two thermal units were constructed by concepts of subspace identification, whereas the individual wind turbine integration was implicated by the Hammerstein–Wiener (HW) model and then augmented from the output to simulate the effect of a wind farm, assuming similar power harvesting from all turbines in the farm. A practical strategy of control was then suggested, which was as follows: with a changing load demand, the available harvested wind energy must be fully admitted to the network to cover part of the load demand with the free energy, and the resultant load signal will then be instructed to the MPCs designed for the coal and gas units for the coordination of generation. The load signal, after being penetrated by wind, has more transients and faster changes, and needs a more sophisticated control in order to follow the load demand of the flexible coal and gas units. Furthermore, as the level of wind penetration increases, the power system frequency excursions are higher. The simulation results show an acceptable performance for linear MPCs embedded to the GT and coal units, with around a 90 MW share of wind without exceeding the safe restrictions of the plants and allowable reasonable frequency excursions. The complete simulation framework can be used to facilitate wind energy penetration in such power systems and train the operators and future engineers with subsequent power system frequency simulation studies.