Modelo matemático para otimização multiobjetivo do despacho econômico ambiental de usinas térmicas usando o NSGA-II

Abstract

One of the priority tasks for thermoelectric plants is to supply the requested energy demand, ensuring the lowest possible cost. This task is more important in the Northern Region of Brazil, especially in the Industrial Hub of Manaus (PIM) and in the city itself, where a large part of this energy is supplied by Thermoelectric Power Plants (UTE). The selection of generators and their work regime is known as the Economic Dispatch (DE). The essential objective of ED is to operate UTEs by meeting demand at the lowest possible cost of fuel. However, the worldwide concern about pollution caused by fossil fuels in recent times to minimize fuel costs can not be considered the only objective to be achieved in the UTEs and limiting the emission of pollutants has become another primary objective. Thus, the Environmental Economic Dispatch (DEA) appears, which seeks not to reduce costs, but also emissions. To solve the optimization of this task there are several deterministic as well as heuristic methods. One of the most used methods according to the literature is the Genetic Algorithm of Non-dominated Classification, NSGA-II, considering two objective functions, a function of fuel cost and another quantity function. In this thesis, the proposed solution has the following contributions: it develops a new and unprecedented function to evaluate the environmental contamination produced by the UTEs that, in addition to minimizing the amount of pollutants, takes into account the influence of pollutants more harmful to the environment. This function, called the Emissions Index, is applied to the engines of two UTEs in the city of Manaus with satisfactory results. The Emissions Index and the traditional fuel cost function is optimized using the NSGA-II, determining optimal solutions for output power in several characteristic and non-characteristic scenarios of the plants, and can be applied to any thermoelectric plant. In order to analyze the viability of the solution proposed by this thesis, a set of ten thermal generating units of a UTE of the city of Manaus and the IEEE 118-bar System were used as case studies, demonstrating the robustness of the proposal in what refer to the solution presented. These results were significant considering the Emissions Index and using the optimization procedure of the non-dominated classification algorithm II (NSGA-II). This new DEA methodology enables specialists in the area to reduce costs and generate generation planning.