Optimal Sizing and Power Management Strategies of Islanded Microgrids for Remote Electrification Systems

Over the past few years, electrification of remote communities with an efficient utilization of on-site energy resources has entered a new phase of evolution. However, the planning tools and studies for the remote microgrids are considered inadequate. Moreover, the existing techniques have not taken into account the impact of reactive power on component sizes. Thus, this thesis concentrates on optimal sizing design of an islanded microgrid (IMG), which is composed of renewable energy resources (RERs), battery energy storage system (BESS), and diesel generation system (DGS), for the purpose of electrifying off-grid communities. Owing to the utilization of both BESS and DGS, four power management strategies (PMSs) are modeled upon analyzing the impacts of reactive power to chronologically simulate the IMG. In this work, two single-objective optimization (SOO) and two multiobjective optimization (MOO) approaches are developed for determining the optimal component sizes in an IMG. Chronological simulation and an enumeration-based search technique are adopted in the first SOO approach. Then, an accelerated SOO approach is proposed by adopting an improved piecewise aggregate approximation (IPAA)-based time series and a genetic algorithm (GA). Next, an adaptive weighted sum (AWS) method, in conjunction with an enumeration search technique, is adopted in a bi-objective optimization approach. Finally, an elitist non-dominated sorting GA-II (NSGA-II) technique is proposed for MOO of the IMG by introducing three objective functions. The enumeration-based SOO approach ensures a global optimum, determines the optimal sizes and PMSs simultaneously, and offers a realistic solution. The accelerated SOO approach significantly reduces the central processing unit (CPU) time without largely deviating the life cycle cost (LCC). The bi-objective optimal sizing approach generates a large number of evenly spread trade-off solutions both in regular and uneven regions upon adopting the LCC and renewable energy penetration (REP) as the objective functions. Using the MOO approach, one can produce a diversified set of Pareto optimal solutions, for both the component sizes and PMSs, at a reduced computational effort. The effectiveness of the proposed approaches is demonstrated by simulation studies in the MATLAB/Simulink software environment