Control of Canal Flow by Adjoint Sensitivity Method

Abstract

A new method for control of unsteady flow in open channels is presented. Control is achieved using an iterative approach, the Broyden, Fletcher, Goldfarb, and Shanno quasi-Newton method that utilizes an adjoint sensitivity method to efficiently compute gradient information. The adjoint equations are derived from the differential form of the full shallow-water equations in one dimension and the resulting sensitivies allow a wide class of flows to be controlled. In addition, the adjoint sensitivity method permits both scheduled and adaptive flow control to be achieved. Application of the method to scheduled flow control is presented.