Multi-Objective Optimization of Thermodynamic Power Cycles Using an Evolutionary Algorithm

Abstract

Pressurized water nuclear power stations are characterized by thermal efficiencies of about 31 to 35% which are much lower than those that are in general obtained by fossil fuels plants. It has been demonstrated that the effect of superheating the steam before entering into a low-pressure turbine makes it possible to increase the cycle efficiency by about 16% but it can affect the overall generated power. In this work, an optimization study is applied to the secondary circuit of Gentilly-2 nuclear power station, where the resulting steam flow distribution required to obtain superheated steam, increases the available energy at the entrance of low-pressure turbine but decreases the total work produced by the high-pressure stage. Further, the Gentilly-2 secondary loop regenerates part of the thermal energy in feed-water preheaters, therefore, the problem consists in determining the best fractions of extracted steam that permit increasing cycle efficiency without decreasing the power produced by the station. The simultaneous improvement of the power and the plant efficiency constitutes a multi-objective optimization problem where two objective functions compete one to the other. It is obvious that such an optimization task does not have a unique solution, but rather a set of optimal solutions that consists of a compromise among the objectives imposed to the problem. In this work, we present an innovative technique based on a genetic algorithm that consists of finding optimal values that converge towards Pareto's optimal front.