New methodology for calibration of hydrodynamic models in curved open-channel flow

Abstract

This paper evaluates a new methodology for calibration of hydrodynamic models based on the theory of statistical design of experiments. An Eulerian-Eulerian hydrodynamic homogeneous model, integrated by the commercial software CFX Ansys Inc., is used to perform the numerical experiments. For the screening step, the fractional factorial experimental design 27-2 was used, followed by a Draper-Lin design of second order to find the optimum point in the calibration. A new method is introduced to generate the level of points to the center and to carry out the test of lack of fit. In this work, we develop a validated methodology for the calibration of deterministic hydrodynamic models with several factors, suggesting a second-order regression model for forecasting the optimum point of the simulations, with acceptable accuracy in predicting the response variable.