Controlling the behaviour of turbulent flows poses an outstanding challenge with a tremendous transversal impact on the efficiency and environmental footprint of numerous industrial processes, among which aeronautics and mobility stand out. A scientific barrier hindering efficient closed-loop control of turbulent flows is their strongly nonlinear chaotic multiscale behaviour. Nonetheless, they often present dominant features evolving on low-dimensional attractors, which can be exploited to obtain compact surrogate models for the dynamics.
Model Predictive Control (MPC) is a technique based on iterative optimization of control actions applied on compact models of a dynamical system. MPC provides a versatile framework in which advanced tools from applied mathematics and statistics can be employed to identify such patterns in turbulent flows, reduce the dimensionality of the problem, and offer compact low-rank dynamical models for control optimization. The main objective of the project PREDATOR-CM-UC3M is the development of novel efficient MPC techniques for turbulent flows, which need to be highly robust to measurement noise and model uncertainty due to dynamics truncation.
The expertise of the research team in the areas Aerospace Engineering and in Statistics contributes the multidisciplinary underpinnings to achieve the proposed goal. Advances in MPC are expected to have an impact on the application side, allowing the design of more proficient controllers, with the corresponding benefit in terms of increased efficiency and reduced environmental footprint. On the other hand, the challenges posed by turbulent flows will foster innovation in applied statistics and mathematics, triggering the development of novel robust tools for control of nonlinear high-dimensional chaotic system, which are ubiquitous in industry and nature.