Low-order models for flow control
Feedback flow control needs a model that is accurate enough to capture unsteady flow and small enough to run in a controller. This work builds such models from data and uses them for control.
- Capability
- Machine learning and data, controls
- Method
- Double POD, adaptive control, CFD
- Published
- Journal of Fluid Mechanics, 2008; AIAA Journal, 2014
Challenge, method, outcome
Challenge
A flow controller needs a model of unsteady, nonlinear flow and of how actuators change it.
Method
Double POD builds modes from transient data, tested on the flow behind a cylinder. The approach was applied to a slender forebody at high angle of attack, where a model identified from open-loop forcing drove a feedback law, and plasma actuators were studied in experiments.
Outcome
Double POD follows the changing flow modes through both limit-cycle growth and forced transients (J. Fluid Mechanics, 2008). A side-force set point was tracked in closed-loop simulation of the forebody (AIAA Journal, 2014), and the ogive wake was characterized experimentally (2013).
Source: J. Fluid Mechanics, 2008; AIAA Journal, 2014; Int. J. Flow Control, 2013
Published work
- 2008Low-dimensional modelling of a transient cylinder wake using double proper orthogonal decomposition
- 2014Closed-Loop Flow Control of a Forebody at a High Incidence Angle
- 2018Feedback Flow Control: A Heuristic Approach
- 2013Open-Loop Dynamics of the Asymmetric Vortex Wake behind a von Kármán Ogive at High Incidence