Drag reduction of a bluff body using adaptive control methods

A classical actuator is used to control the drag exerted on a bluff body at large Reynolds number (Re=20000). The geometry is similar to a backward-facing step whose separation point is modified using a rotating cylinder at the edge. The slow fluctuations of the total drag are directly measured by means of strain gauges. As shown by visualizations, the actuator delays the separation point. The size of the low-pressure region behind the body is decreased and the drag reduced. It is found that the faster the rotation of the cylinder, the lower the drag. In a first study, the goal of the control is for the system to reach a drag consign predetermined by the experimentalist. The control loop is closed with a proportional integral correction. This adaptive method is shown to be efficient and robust in spite of the large fluctuations of the drag. In the second method, the system finds itself its optimal set point. It is defined as the lowest cost of global energy consumption of the system (drag reduction versus energy used by the actuator). For this purpose, an extremum seeking control method is applied in order to deal with the large background noise due to turbulence. It consists in a synchronous detection of the response measured in the drag measurements to a modulation of the actuator. The phase shift and amplitude of the modulation estimate the local gradient of the total energy function. With this gradient estimation, the system goes to the minimum of global power consumption by itself. The system is found to be also robust and reacts successfully to changes of the external mean flow. This experiment attests to the real efficiency of local active control in reducing autonomously the global energy consumption of a system under turbulent flow.