Evaluation of the Genetic Algorithms Technique for the Multi-objective Optimization of the Sizing of a Constant Section Surge Tanks and its Junction Branch in a Hydroelectric Power Plant

Abstract

This project presents the application of a genetic algorithm used in the improvement of a hydroelectric system (reservoir system, tunnel, upper equilibrium chimney, pressure pipe, and turbine), utilizing various hydrodynamic (transient flow, wave propagation speed, transient phenomena) and economic variables, to achieve optimization of the geometry of the upper equilibrium chimney and its connection branch. A program was developed using Matlab software by the method of characteristics to create a one-dimensional numerical model in a hydroelectric power plant, with results showing variations in piezometric head and flow during a transient event. By applying a code for genetic algorithms using the variables of the objective function and parameters to optimize, such as the diameter of the connection branch and the diameter of the equilibrium chimney, three solutions for the chimney design were obtained, each analyzed based on its geometry and construction cost. The final design meets technical requirements and integrates well with the environment. To verify the functionality of the program, an analysis, comparison, and validation were performed on 10 hydroelectric power plants, using their main parameters, including the diameter of the connection branch, construction cost per monetary unit, and oscillation, resulting in different solutions depending on the initial characteristics assigned to each of the analyzed hydraulic plants.