Constructive heuristic approaches based on graph theory and mathematical modeling for distribution system reconfiguration

• Grant number: 26/03455-9
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: May 01, 2026
• End date: July 31, 2026
• Field of knowledge: Engineering - Electrical Engineering - Power Systems
• Principal Investigator: Leonardo Henrique Faria Macedo Possagnolo
• Grantee: Gustavo Manfredini Cerqueira
• Supervisor: Joao Andre Pinto Soares
• Host Institution: Faculdade de Engenharia e Ciências (FEC). Universidade Estadual Paulista (UNESP). Campus de Rosana. Rosana , SP, Brazil
• Institution abroad: Instituto Superior de Engenharia do Porto (ISEP), Portugal
• Associated to the scholarship: 24/13699-7 - Analysis and evaluation of radiality constraints in the reconfiguration of electrical distribution systems, BP.IC

Abstract

The reconfiguration of electrical distribution systems is an essential technique for optimizing network operation, aiming to reduce technical losses and improve power supply quality. To ensure protection coordination and operational simplicity, these networks must operate radially, which characterizes the system's topology as a spanning tree. Continuing the evaluation of mathematical optimization models to guarantee this radiality constraint, this research project, to be developed at the Instituto Superior de Engenharia do Porto (ISEP), proposes the development of generalized constructive heuristic algorithms for the reconfiguration problem. The core methodology is divided into two approaches. The first is based on the iterative resolution of a mathematical model, where the transition between topologies is controlled by limiting the maximum number of switches that can be simultaneously opened and closed from the current system configuration. The second approach proposes the development of constructive heuristic variants structured upon the classic graph theory algorithms of Kruskal, Prim, and Boruvka. To apply these methods to the reality of electrical networks, branch weights will be estimated from the solution of power flow problems for meshed systems. Additionally, these three graph algorithms will be evaluated in solving the classic minimum spanning tree problem. The performance and computational effort of these constructive heuristic techniques will be critically compared with the results obtained from the exact mathematical formulations analyzed in previous stages of the research. As a result, it is expected to consolidate highly efficient and scalable methods capable of solving the reconfiguration problem in large-scale systems with reduced computational effort. (AU)