Author(s): N. D. Katopodes
Linked Author(s):
Keywords: Daptive Control; Adjoint sensitivity; Parameter identification; Data assimilation
Abstract: A mathematical method is developed for the adaptive control of flow and the transport of mass in a channel. The method is based on real-time information provided by a sensor array installed at appropriate locations and capable of monitoring instantaneous changes in the depth and velocity of flow or the concentration of an arbitrary number of species in solution or suspension. The method also utilizes concurrent hydrodynamic and quality information provided by a simulation model capable of forecasting the dependent variables based on a set of initial conditions of the system. Once a condition requiring control action is detected by the sensor array, the control model provides a set of commands to a number of actuators capable of modifying the flow or transport. The model requires continuous assimilation of data from the sensors to adjust the background, so the error between its current state and sensor measurements is minimized. The model must also perform numerous forecast simulations to determine the optimum set of actuator commands necessary to control the flow. Both errors are optimized in time and space by use of the adjoint equations. For advection-dominated flow and transport, the sensitivity signals are contaminated by boundary reflections. Non-reflective boundary conditions are introduced for the adjoint equations that allow the accurate processing of signals for multiple sensors and actuators. Examples are presented for discharge control, containment of solute plumes and reduction of sediment deposition.
Year: 2003