Resumo:
There is a constant search for greater availability of energy resources since energy is a vital
item for the development of a country. Projections for electricity demand until 2030 point to an
increase, in which 20% of the total electricity generation will come from combined cycle power
plants with natural gas, considered one of the most developed technologies for electricity
production. Given the relevance of natural gas fuel and combined cycle power plants in the
electrical and energy matrix, it is important to use available resources in the most efficient way.
In this Thesis, three optimization techniques were analyzed (genetic algorithm, particle swarm
and simulated annealing) applied to a combined cycle power plant in steady state, under design
and off-design conditions, to perform a multi-objective optimization. The proposed method was
previously applied in the CGAM cogeneration system for validation and later applied in the
combined cycle power plant. Thermodynamic, traditional and advanced exergetic, and
economic analyzes were employed. The GateCycleTM software together with the CycleLink
tool were used to simulate the energy system model in design and in the partial loads of 90%,
80%, 70%, 60%, 50% and 40%. The minimized objective functions were the electricity cost
and the total unavoidable exergy destruction rate. The decision variables were air compressor
pressure ratio, isentropic air compressor efficiency, isentropic gas turbine efficiency, exhaust
gas temperature at the inlet of the gas turbine, mass flow of fuel to the duct burner, isentropic
efficiency of the steam turbine and isentropic pump efficiency. The modeFRONTIERTM
software was used to apply the three optimization techniques in order to evaluate the objective
functions. The two energy systems were optimized with each of the optimization techniques in
all analyzed conditions. There was no unanimous technique in performance under all
conditions. In the design condition, the combined cycle power plant presented an electricity
cost 11.11% lower and an inevitable destroyed exergy 23.48% lower, when applying the
particle swarm technique in the search for the best values of the decision variables, being the
algorithm of better performance in this specific condition. In all other conditions an optimal
solution was also obtained.
