OS Numerical Optimization: Finite-Dimensional Model Predictive Control for the Navier-Stokes Equations
Time
Monday, 5. August 2024
10:00 - 0:00
Location
G227a
Organizer
B. Azmi & S. Volkwein
Speaker:
Felix Sauer
On 5th August 2024 at 10:00, Felix Sauer will give a talk.
Abstract: Fluid dynamics and optimal control are active and demanding areas of research. In this talk, I will first showcase how I devised different methods for numerically solving the Navier-Stokes equations using finite element techniques. Then, I will break down how to solve optimal control problems based on the Navier-Stokes equations by computing adjoint-based gradients in both continuous and discrete settings, which can be used in gradient descent methods. For large time-horizons, I present the Model Predictive Control (MPC) framework, which aims to decompose a large optimal control problem into several smaller subproblems that can be dynamically tackled. To support the numerical foundation, I will present several examples that highlight the efficiency of the MPC approach, such as how MPC can be used to reduce turbulence in a channel flow.