| The issue of electric power systems under persistent and high stress conditions and possible changes to electric power systems to deal with this issue is the subject of this dissertation. The stresses considered here are not the single event type of disruptions that occur as a result of a hurricane or other extreme weather event or the large blackouts that result from a particular set of circumstances. Instead the focus is on conditions that cause systematic and long-term performance degradation of the system such as under investment in infrastructure, poor maintenance, and military conflict.; While it has long been recognized that persistent stresses such as conflict and war can have a large impact on electric power systems, there have been few systematic analyses of the problem. Distributed generation (power generated by small units close to the demand) has been a proposed solution. However, quantitative assessments of the impact of widespread DG deployment on overall system reliability have also been lacking. The first goal of this research was to model and quantify the reliability and economic differences between centralized and distributed energy systems for providing electricity and heat, particularly under stress conditions. This goal was met through the development of Monte Carlo reliability simulations, applied to different system network topologies. The results of those models show significant potential improvements in energy delivery with distributed systems.; The second goal was to determine the impact of heterogeneity of local loads on the desired level of decentralization of the system and the impact of decentralization on the network requirements. This goal was met through a combination of Monte Carlo simulations applied to systems with differentiated and non-coincident loads and an optimal power flow applied to a more realistic network topology. The results of those models show the potential for improvements when loads are non-coincident and microgrids can share power as well as the fact that the power sharing may be largely limited to local clusters of micro-grids. This research also showed the need for incorporation of stress in power systems modeling and a method for characterizing stress. |
