Scheduling maneuvers for the restoration of electric power distribution networks

• Autor(a) principal: André Luiz Maravilha Silva
• Data de Publicação: 2018
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/63981
• topic: Energy utilities; Electricity; Distribution networks; Engenharia elétrica; Energia elétrica - Distribuição; Heurística

During a fault in a power distribution network, energy utilities can change the network topology to reconnect all or at least a portion of disconnected clients, then minimizing the area affected by the fault. These changes in the network are defined by a restoration plan that specifies a set of switches to be maneuvered. The faster energy utilitiesreconnectdisconnectedclients, thelighterthepenaltiesappliedtothem. Then, the energy utilities have a tight time frame to define the restoration plan before dispatching maintenance teams to perform the required maneuvers. Besides, the total time needed to perform the maneuvers has to be considered when determining the restoration plan, since the new topology that restores/minimizes the affected clients willbefullyoperationalonlyafterthemaneuversarecompleted. Althoughtheproblem of restoring power distribution networks is widely studied in the literature, no study has considered both the existence of multiple maintenance teams working in parallel and the time taken by the teams to move between locations where the maneuverable switches are located. Ignoring these characteristics results in inefficient restoration plans, taking longer than expected. In this work, we address the problem of providing a better estimation of the time to perform the restoration plan by modeling the assignment and sequencing of maneuver operations as a scheduling problem that minimizes the makespan, i.e., the total time required to complete all maneuver operations. Furthermore, we present specific heuristics for its solution that are fast enough to be incorporated into existing restoration algorithms without compromising their performance, since they already need to perform other time-consuming routines, e.g., power flow algorithms. Computational experiments with different fault scenarios showed that incorporating the proposed strategy in a restoration algorithm led to more efficient restoration plans.