Planejamento de sistemas de distribuição com representação de incertezas através de análise intervalar

Abstract

This work proposes an approach for optimal expansion planning of electrical energy distribution system with the modeling of uncertainties over load demand and wind-based distributed generation. The planning criterion is the minimization of costs related to investment, operation, carbon dioxide emission and reliability, as well as uncertainties over load demand and. The reliability cost is related to the energy not supplied, by using the ‘N-1’ criterion for network sections and evaluating the impact of isolated permanent faults through a restoration strategy, which depends directly on the number and location of protection and switching devices on the network. The problem is modeled as mixed-integer non-linear programming and solved through the bioinspired optimization meta-heuristic technique called Artificial Immune System. The planning decisions involve: expansion of existing substations and/or construction of new ones, retrofitting existing feeders by replacing cables and/or construction of new ones, installation of wind distributed generators in candidate buses, allocation of sectionalizing switches, network reconfiguration and defining the optimal geographical coordinates for the construction of new substations, as well as the optimal feeder routing between the substations and the load points. The new substations location and the construction of the connecting feeders take into account a geographical information system that imposes a set of physical routing constraints on the problem, such as the existence of natural obstacles and high-cost paths. For the uncertainties representation, two methods are applied and compared: an interval technique through an interval power flow and a scenariobased approach. Network constraints are considered, as the limits of current, voltage and power from substations, as well as the obtaining of radial and connected topology. The main contribution of the proposed interval methodology consists of handling the uncertainties in an efficient manner through a single step, instead of the several deterministic evaluations of the scenario-based approach. Numerical results are presented for well-known test systems and show the potentials of the proposed interval approach.