MODELLING OF AN INTELLIGENT ENERGY MANAGEMENT SYSTEM AND HIERARCHICAL POWER SHEDULIG ALGORITHIM FOR A SOLAR MICROGRID

Abstract

In Nigeria, where many regions experience significantly low electrification rates, microgrids have been implemented to provide electricity to rural communities. There is a pressing need to enhance and redesign the conventional energy management system, which necessitates the development of an energy management strategy aimed at optimizing the power exchange with the grid profile. This thesis presents an optimal configuration algorithm for the allocation of energy resources within a microgrid, utilizing data obtained from a microgrid situated in Nigeria. When compared to an existing design utilized in the experimental setup of this study, the newly developed algorithm demonstrated superior performance. Four distinct scenarios were analyzed; the simulation results indicated that in scenario one, the system operated without any costs, as all energy consumption was effectively met by the solar panel. In scenario two, the yearly projection revealed a total savings of N1,232,400 compared to the existing design, which incurred a daily operational cost of N27,600. For scenario three, the yearly savings amounted to N1,029,600 when juxtaposed with the existing design that had an operating cost of N9,900. Additionally, the system was evaluated to assess the extent of deviation between the supplied power and the load demand. The simulation results indicated that at 100 seconds, the error percentage for the existing design peaked at 7%, while the developed algorithm's error approached zero. The system's responsiveness to transitions between different scenarios was also analyzed, revealing that the developed algorithm reacted to changes in just 83 milliseconds. Furthermore, an assessment of the battery state of charge (SoC) indicated that the batteries managed by the developed algorithm consistently maintained a level above the acceptable threshold of 33%, in contrast to the existing design, which fell below this threshold during simulations.