Efficient, Robust Design Tool for Open-Channel Flow

Abstract

The use of computational fluid dynamics to simulate physical phenomenon, such as flow through open channels, has become more prolific as computational resources and numerical algorithms have improved. With current algorithms, the designer defines the shape of the channel geometry, and the code calculates the fluid dynamic state within the channel. For these computational tools to be effectively used in the design process, the designer should be able to define the desired flow characteristics and the computational tool should generate the channel geometry that yields the closest match. In this paper, a finite-element simulation code, which solves the viscous, 2D, shallow water equations, is modified to become an efficient, robust design tool by inclusion of discrete sensitivity analysis and the Gauss-Newton optimization algorithm. This technique is demonstrated for design of supercritical flow channels with the objective of producing constant depth flow downstream of the channel transitions.