Despacho estocástico para sistemas elétricos de distribuição com geração fotovoltaica

Abstract

This thesis proposes a new stochastic intertemporal economic dispatch of generation and demand for small electricity power systems with photovoltaic generation. The model is formulated in the context of smart-grids, which allows the control technology and communication is available to the system. The dispatch strategy proposed comprises thepre-dispatch for a day ahead; followed by a classification process of possible solutions and a dispatch model for one hour ahead. The pre-dispatch involves generation and demand schedule and is formulated by a stochastic, non-linear, mixed-integer programming model. In this step, an inequality constraint, called flexibility constraint, is included in the model in order to ensure the availability of resources with response time compatible with the variability of photovoltaic generation. It is assumed in this work that these flexible resources can be represented by peak generators and flexible loads which participate in load reductionprograms. The model includes the non-linear network representation, unit commitment restrictions of thermal generators, and a preference model of some consumers. On the other hand, the classification process of solutions, coming from the day ahead dispatch, is based on conditioned value at risk concept, which evaluates the probability of high costs for one solution of pre-dispatch. The last phase of strategy is the stochastic dispatch of flexible resources, for one hour ahead, which checks the system's ability to maintain the balance between demand and generation considering the variation of photovoltaic generation. Ahybrid meta-heuristics, developed in this work, is applied in order to access the pre-dispatch model results. Test systems from Brazilian utilities, with 192 bars and with 31, are used to illustrate the potential of proposed dispatch strategy. The results indicate that proposedmanagement model ensured the availability of flexible resources. Therefore, the main contributions of this thesis are the intertemporal stochastic pre-dispatch model that considers the prediction error of photovoltaic generation; the management model of the flexibleresources through a flexibility restriction; the demand dispatch model which observes the consumption preferences; the development of hybrid meta-heuristic dedicated to predispatch solution; the definition of a solutions classification process; and a stochastic dispatch model for one hour ahead.