Automation of the Hourly Contracting of Electric Power in the National Interconnected System

The main objective of this work has been to develop a methodology that allows for short-term operative programming, predicting the hourly demand of electric power in the National Interconnected System (NIS). It is considering the operational, economic and contractual restrictions for the dispatch of power, in order to reduce the purchase costs of power and energy of the ´Administración Nacional de Electricidad´ (ANDE) from Paraguay. For this aim, the influence of different factors on the demand for electricity in Paraguay has been analyzed, with the scheme of identifying those with the greatest impact, to be used later in a demand forecast model. Considering the complexity of the demand series, this forecast has been made with Artificial Neural Networks (ANN). Together with the operating conditions of the NIS, which impose physical and contractual restrictions, considering associated dispatch costs and the system load curve, estimated through the ANN, an optimization model has been developed, implemented through Mixed Integer Linear Programming (MILP), whose objective function is to reduce daily costs by dispatching the plants that supply the NIS. Particular instances were verified to corroborate the correct functioning of the MILP model. Subsequently, the integration of the ANN with the MILP model was carried out, simulating various scenarios to verify the robustness of the proposed methodology, providing encouraging results in relation to the accuracy of the forecast of the demand, the feasibility of the prescribed dispensing and the reduction of the costs for purchasing power and energy.