Optimization and design of isolated microgrid systems with the daily energy balance method

Abstract

A state-of-the art alternative or supplement to the legacy electric power delivery system – the electric power grid – is the microgrid, a decentralized, local group of interconnected loads and distributed power sources that can operate in conjunction with or independently of the main grid, in a cooperative group or as stand-alone entities. The focus of this dissertation is the optimization and design of stand-alone, also called islanded or isolated, microgrids. Microgrids typically include various power sources such as engine generators, solar panels, wind turbines, fuel cells, and energy storage devices such as batteries. Given the high variability of power demand of loads, and the high variability of generated power by different types of energy sources; considering the completely different cost structures for different microgrid components, a major technical challenge is designing an optimal microgrid system, that is, determining the optimal combination of components for a reliable, stable, cost-effective and environmentally friendly microgrid system. Virtually all currently existing optimization approaches use extensive simulations to attempt to select configurations that match the generated and consumed power for each hour over the course of the project lifetime. This approach requires detailed system data, is computationally inefficient and slow; is not easily adaptable to different systems, different consumer profiles and different locations; requires expensive and complex software to operate; and in general is not very suitable for making practical engineering decisions. The novel method presented in this dissertation is uniquely different from all existing approaches in that it does not attempt to match power quantities. Instead, this method aims to achieve a balance between generated and consumed energy over the course of a day. To determine the amount of energy that can be generated by different sources, new and revised equations were developed that calculate the expected amount of energy generated by solar and wind power systems. In addition, new equations were developed to calculate the expected amount of energy that needs to be stored in batteries for reliable operation. A Generalized Reduced Gradient (GRG) and a genetic algorithm (GA) were then used to determine the optimum combination of components that minimize cost and maximize reliability. Unlike some other approaches, this method allows a high degree of customization. Various power profiles and various components can be selected, and multi-objective optimization operators produce optimization results according to the preferences of the user. The equations and results are easily verifiable and provide a high degree of confidence in the obtained solutions. This method is much more computationally efficient than other methods, and can be implemented on widely available computation platforms such as MS Excel. In short, this method is much more engineer friendly compared to all other approaches. To verify the validity of this method, the developed models and equations were tested with real-life production data of operational generator, wind, and solar power systems. The optimization algorithms were then run with power demand data of real-life small, medium, and large individual users, as well as with demand data for a community of multiple users. For all systems, the resulting configurations were then bench-marked to configurations determined by a specialty software tool called HOMER. In all cases the obtained solutions by the new method were superior in terms of reliability and economic as well as environmental costs, hereby confirming the validity of this method as well as its superiority to competing approaches.