Heuristic Techniques and Evolutionary Algorithms in Microgrid Optimization Problems
Aykut Fatih Güven
Abstract
This study introduces an optimization model and evaluates the performance of various metaheuristic optimization techniques in solving the sizing problem for an on-grid hybrid renewable energy system (HRES), comprising solar photovoltaic (PV) panels, diesel generators (DG), wind turbines (WT), and battery storage systems (BSS). A university campus was selected as a case study to assess the effectiveness of the proposed HRES. The optimization focuses on minimizing the annual system cost (ACS), the levelized cost of energy (LCOE), and the total net present cost (TNPC), while ensuring that the annual load demand is met and the load power supply probability (LPSP) remains within acceptable limits. The decision variables include the sizes of the PV and WT sources, the power transfer capacity of the inverter to the load, and the capacities of the BSS and DG. The snake optimization algorithm (SOA) was employed to optimize the objective function, taking into account constraints including the maximum and minimum sizes of system components, the renewable energy fraction, and reliability measures such as LPSP. The performance evaluation of on-grid hybrid systems is based on their costs, reliability, and greenhouse gas emissions reduction. Detailed mathematical models are provided to estimate the power output of the hybrid system. The results indicate that the combination of PV, WT, and BSS offers the best performance in terms of ACS, LCOE, TNPC, and reliability indices. Furthermore, a comparison of SOA with other algorithms, including gray wolf optimization (GWO), reptile search algorithm (RSA), firefly algorithm (FFA), battle royale optimization (BRO), and ant colony optimization (ACO), shows SOA’s superior capabilities in system design, achieving lower costs and better reliability indices. The findings also reveal that TNPC, ACS, and LCOE are approximately 33% lower than those of a standalone hybrid system in an on-grid setup, with values reaching $3,340,600, $383,020, and $0.1342/kWh, respectively. Additionally, significant grid interaction was observed, with up to 801,040 kWh purchased and 1,105,800 kWh sold. The proposed system is economically viable, with an LCOE of $0.1511/kWh, below the commercial electricity rate of $0.35/kWh in Turkey. All analyses were conducted using MATLAB 2022b.