Adaptive Control of Flow and Mass Transport by Multi-Sensor Arrays
A
method is presenting that can be potentially used to arrest and eliminate a toxic
plume or a water wave. The
fundamental hypothesis is that there exist boundary mechanisms that, if
activated at the right place and time, they can effectively eliminate the
threat with minimum delay.
Furthermore, the method relies on multiple sensor arrays to detect and
rapidly transmit current information about the plume or wave to the model. Adaptive control is then achieved by history matching of
appropriate boundary controls. The
execution can be made in real time by solving the corresponding adjoint
equations. Sensitivity expressions
relating the objective function response to perturbations in boundary controls
are derived for nonlinear flow equations using an adjoint sensitivity formalism
that does not require the assumption of compact support of the adjoint
variables or strong dissipation of the flow variables. The approximate Riemann invariants of
the adjoint problem are shown to transmit sensitivity information in the
reverse time direction along the characteristic paths of the basic problem. The use of these approximate Riemann
invariants gives rise to non-reflective inflow and outflow boundary conditions
for the adjoint problem that permit sensitivities to be accurately determined. Examples are presented for the
mitigation of catastrophic waves and toxic plumes.
Friday, March 28, 2003
3:30 – 4:30 p.m.