Reduced Order Modelling for Flow Control
September 15, 2008 — September 19, 2008
Coordinators:
- Marek Morzynski (Poznan Univ. of Technology, Poznan, Poland)
- Bernd R. Noack (D. Fluides, Therm., & Combustion, Poitiers, France)
- Gilead Tadmor (Northeastern University, Boston, Mass. , USA)
The aim of the course is to teach efficient strategies of feedback flow control using reduced order modelling and control theoretical methods. The contents range from the foundations to the state of the art. The foundations are exemplified for simple systems. The state of the art is condensed to best practices and illustrated for wake stabilization and other configurations. The considered audience are students, researchers and industrial engineers with background in fluid mechanics and with interest in reduced order modelling and modern flow control.
Reduced order modelling of actuated turbulent flows plays an increasing role in industry and academia, establishing a new discipline besides numerical and experimental methods. Low dimensional models may enable a quick exploration of actuation concepts before experiment or before expensive higher-fidelity simulations. In addition, these models can be considered as the test case of the physical understanding required for new control ideas. Last but not least, the low system dimension may enable the development of controllers employing powerful tools of control theory. The importance is emphasized by two recently established Multi University Research Initiatives on “Closed loop aerodynamic flow control” headed by CalTech and by GeorgiaTech, by the Collaborative Research Center (Sfb 557) “Control of complex turbulent shear flows” hosted at TU Berlin as well as many other similar research consortia and initiatives in Europe, USA and Asian countries.
The first group of lectures include best practices of low-dimensional modelling and associated control design. M. Morzyński elaborates how global stability analysis of the Navier-Stokes equation can be employed in modelling and flow control. The physical foundations and best practices of control-oriented low-dimensional Galerkin modelling are outlined by B.R. Noack, including theoretical and empirical approaches. G. Tadmor addresses critical enablers for observer and control design.
All these talks will focus on wake stabilization as one benchmark application.
The second group of talks emphasize applications in numerical simulations and experiments.
By definition, optimal flow control based on a Navier-Stokes discretization and based on complete knowledge of the flow at all times, defines an upper limit for achievable actuation opportunities. L. Cordier discusses numerical tractable strategies of optimal flow control using reduced order models as plants. In most experiments, only few sensor readings monitor the flow state and optimality is traded against online-capability and error tolerance (robustness).
S. Siegel outlines experimental wake stabilization results guided by CFD and using reduced-order models. In some cases, reduced order models are difficult to develop due to lack of illuminating flow data. R. King presents flow control strategies using black-box models and model-free approaches as an alternative. These strategies have been successfully employed in industrial applications.
All six speakers head research groups in the area of feedback flow control. They have long-term experience in low-dimensional modelling, in computational and experimental fluid dynamics as well as in control theory.