HYDRODYNAMIC MODEL CALIBRATION IN AN URBAN CHANNEL USING MULTIOBJECTIVE GENETIC ALGORITHMS

Abstract

This paper aims to perform calibration of a hydrodynamic model for open channels using evolutionary algorithms multi-objectives from water level measurements. The method was applied in an urban channel with gabion coating with bottom macro-roughness. An event was used to calibrate parameters and three events for validation. Two objective functions were used in the optimization, the Nash-Sutcliffe efficiency coefficient and the error of the peak quota. The NSGA-II optimization algorithm presented excellent results, obtaining Nash-Sutcliffe coefficients higher than 0.96, and errors of maximum dimension less than 0.40 cm for both calibration and validation of every event. The value of the roughness was found to be 0.0905 for the path with obstacles and 0.0277 for the part with gabion. Keywords: Multi-objective optimization; Saint-Venant equations; Unsteady flow; NSGA-II; Rating curve.