Comparative Methods for Solving Optimal Power Flow in Distribution Networks Considering Distributed Generators: Metaheuristics vs. Convex Optimization

Abstract

Objective: This article presents an analysis of different optimization methodologies, which aims to make an objective comparison between metaheuristic and convex optimization methods in distribution networks, focusing on the inclusion of distributed generation (DG). The MATLAB software is used as a tool for implementation and obtaining results. The objective was to determine the optimal size of the DGs to be integrated into the networks, with the purpose of reducing the active power losses (objective function). Methodology: Based on the specialized literature, the methodologies are selected, and the bases and conditions for the implementation of the optimization techniques are determined. In the case of second-order cone programming (SOCP), the relaxation of the nonlinear optimal power flow (OPF) problem is performed in order to use convex optimization. Then, the structures of each technique are established and applied in the MATLAB software. Due to the iterative nature of metaheuristic methods, the data corresponding to 100 compilations for each algorithm are collected. Finally, by means of a statistical analysis, the optimal solutions for the objective function in each methodology are determined, and, with these results, the different methods applied to the networks are compared. Results: By analyzing 33- and 69-node systems, it is demonstrated that metaheuristic methods are able to effectively size DGs in distribution systems and yield good results that are similar and comparable to SOCP regarding the OPF problem. Genetic algorithms (GA) showed the best results for the studied implementation, even surpassing the SOCP. Conclusions: Metaheuristic methods proved to be algorithms with a high computational efficiency and are suitable for real-time applications if implemented in distribution systems with well-defined conditions. These techniques provide innovative ideas because they are not rigid algorithms, which makes them very versatile methods that can be adapted to any combinatorial optimization problem and software, yielding results even at the convex optimization level.