< Back to previous page


Optimization and control of turbulence with applications in boundary layers and wind energy.

A major challenge for the use of adjoint methods in turbulence-resolving methods such as direct numerical simulation (DNS) and large-eddy simulation (LES), is the chaotic nature of turbulence, which leads to adjoint sensitivity equations that naturally diverge in time. This makes it currently impossible to employ LES and DNS for design optimization, as this requires the time integration of turbulence over very long time horizons. Only relatively recently, a possible solution for the sensitivity analysis of ergodic chaotic systems has been proposed, based on shadowing. However, many challenges remain, in particular related to the conditioning of the method, and all work is currently focused on ‘simple’ dynamical systems. It is the ambition of the current proposal to significantly extend time horizons for sensitivity analysis and optimization of turbulence, with focus on turbulent boundary layers. To this end, we will combine shadowing-based methods, with preconditioned multiple shooting approaches and multigrid methods. We envisage application for optimization and optimal control of wind-farm boundary layers, aiming at improved energy extraction, as well as for design optimization of surface structures in turbulent boundary layers, aiming at reduced skin-friction drag.

Date:1 Oct 2009 →  Today
Keywords:optimization of turbulence, sensitivity of turbulent flows
Disciplines:Numerical modelling and design, Wind energy, Fluid mechanics and fluid dynamics