Architecture and control of large power networks with distributed generation

摘要

The architecture of the UK's passive power network has taken over one hundred years to evolve through a process of demand and technology led development. In the early years of electrical power, distribution systems were islands of distributed generation, often of different voltages and frequencies. Increasing demand for electrical power and the need to reduce distribution costs eventually led to the standardisation of frequency and voltages and to the connection of the island systems into a large network. Today's power networks are characterised by their rigid hierarchical structure and unidirectional power flows. The threat of climate change is driving the demand for the use of more renewable energy. For electricity production, this is achieved through generation using more wind, biomass, tidal and solar energy. This type of generation is often referred to as Distributed Generation (DG) because it is not a centralised facility connected to the high voltage transmission grid but a distributed source connected to the lower voltage distribution network. The connection of DG to the distribution network significantly alters the power flow throughout the network, and costly network reinforcement is often necessary. The advancement in the control of electrical power has largely been facilitated by the development of semiconductor power electronic devices and has led to the application of "Flexible Alternating Current Transmission Systems (FACTS), which include such devices as "Static Var Compensators" (SVC) and Static Compensators (STATCOM), for the control of network voltages and power flows. Providing a secure power network is a demanding task, but as network complexity is expected to grow with the connection of high levels of DG, so the problem of integration, not just connection, of each successive generator becomes more protracted. A fundamental change to the network architecture may eventually become necessary, and a new, more active network architecture, perhaps based on power cells containing local generation, energy storage and loads, has been proposed by some researchers. The results of an historic review of the growth of power networks, largely in the UK, forms the basis of a case to replace the conventional power transformer with an Active Transformer that will provide a more controllable, flexible and robust DG connection and (i) will facilitate greater network management and business opportunities, and new power flow control features. The Active Transformer design is based on an a.c. link system and an a.c.-a.c. highfrequency direct resonant converter. This thesis describes a model of the converter, built in MATLAB and Simulink®, and used to explore control of the converters. The converter model was then used to construct a model of the Active Transformer, consisting of a resonant, supply-side converter, a high frequency transformer and a resonant, load-side converter. This was then used to demonstrate control of bi-directional power flow and power factor control at the Grid and Distribution Network connections. Issues of robustness and sensitivity to parameter change are discussed, both for the uncompensated and compensated converters used in the Active Transformer. The application of robust H∞ control scheme proposed and compared to a current PI control scheme to prove its efficacy.