Thomas Duriez

Closed-loop separation control over a sharp edge ramp using genetic programming

Abstract We experimentally perform open and closed-loop control of a separating turbulent boundary layer downstream from a sharp edge ramp. The turbulent boundary layer just above the separation point has a Reynolds number

From natural to forced counter-rotating streamwise vortices in boundary layers

Re_{\theta }\approx 3500

Base Flow Modification by Streamwise Vortices: Application to the Control of Separated Flows

Reθ≈3500 based on momentum thickness. The goal of the control is to mitigate separation and early re-attachment. The forcing employs a spanwise array of active vortex generators. The flow state is monitored with skin-friction sensors downstream of the actuators. The feedback control law is obtained using model-free genetic programming control (GPC) (Gautier et al. in J Fluid Mech 770:442–457, 2015). The resulting flow is assessed using the momentum coefficient, pressure distribution and skin friction over the ramp and stereo PIV. The PIV yields vector field statistics, e.g. shear layer growth, the back-flow area and vortex region. GPC is benchmarked against the best periodic forcing. While open-loop control achieves separation reduction by locking-on the shedding mode, GPC gives rise to similar benefits by accelerating the shear layer growth. Moreover, GPC uses less actuation energy.

Taming Nonlinear Dynamics with MLC

A novel approach, based on Non-Linear analysis of the flow, is proposed to study the modification of flat-plate boundary layer by bluff-body Vortex Generators (VGs). Small cylinders in a flat-plate boundary layer are used to create a set of counter-rotating streamwise vortices (CRSVs) which modify the global properties of the boundary layer. From the harmonic modulation and the mean flow distortion of the global mode (or zeroth mode), one can define some physical properties which can be used to quantify the efficiency of a given set of VGs. A first parametric study for a given Reynolds number shows a clear dependance of these properties on the spacing between the VGs. Some of these properties should be useful, for instance, to choose the right parameters to control a separated flow without having to run a full and tedious parametric study.

Streakline-based closed-loop control of a bluff body flow

Counter-rotating streamwise vortices (CRSVs) are ubiquitous in Fluid Mechanics, and especially in boundary layer and shear layer ows. A short review is presented showing how the CRSVs are naturally generated in boundary layer ows and how they can be articially generated in the purpose of ecient ow control experiments.

Methods of Linear Control Theory

We propose a novel closed-loop control strategy of turbulent flows using machine learning methods in a model-free manner. This strategy, called Machine Learning Control (MLC), allows – for the first time – to detect and exploit all enabling nonlinear actuation mechanisms in an un-supervised automatic manner. In this communication, we focus on MLC applications for in-time control of experimental shear flows and demonstrate how it outperforms state-of-the-art control. In particular, MLC is applied to three different experimental closed-loop control setups: (1) the TUCOROM mixing layer tunnel, (2) the Gortler PMMH water tunnel with a backward facing step, and (3) the LML Boundary Layer wind tunnel with a separating turbulent boundary layer. In all three cases, MLC finds a control which yields a significantly better performance with respect to the given cost functional as compared to the best previously tested open-loop actuation. We foresee numerous potential applications to most nonlinear multiple-input multiple-output (MIMO) flow control problems, particularly in experiments. In particular, the model-free architecture of MLC enables its application to a large class of complex nonlinear systems in all areas of science.

Machine Learning Control (MLC)

We first demonstrate the potential of cylindrical vortex generators to control the separation over a smoothly curved ramp in a low velocity hydrodynamic channel. We then focus on the influence of a row of four cylindrical vortex generators on a spatially growing flat plate boundary layer. Using two-component PIV measurements, we show how the boundary layer is modulated by counter-rotating streamwise vortices. We also analyze the 3D velocity field using non-linear perturbations to emphasize a clear modification of the base flow. This base flow modification, together with spanwise modulation, can explain the delay of the separation of the boundary layer over the curved ramp.

MLC Tactics and Strategy

We investigate the application of machine learning control (MLC) to the stabilization of a nonlinear dynamical system. This plant features a frequently observed frequency crosstalk between actuation and unstable dynamics. MLC explores and exploits this frequency crosstalk as an enabling actuation mechanism. MLC-based feedback is benchmarked against open- and closed-loop forcing derived from Kryloff-Bogoliubov approximation.