DBD Plasma Actuator
Dielectric barrier discharge (DBD) plasma actuators are a novel and rapidly growing technology for flow control. The high voltage creates a flow of ionized air on the upper surface of the airfoil, which imparts momentum into the flow without any moving parts. I'm really fascinated by the behavior of airfoils near stall in the laminar regime, my tubercle work has only made me more interested in understanding the flow characteristics here. I plan to study the aerodynamics of a DBD plasma actuator applied to a NACA 0012 airfoil section across varying turbulence intensities, both computationally and experimentally. My goal is to take away more information about DBD actuators as a whole, elucidate their near and post stall performance, as well as learn more about the laminar separation bubble and boundary layer interactions in general.
In order to do that though, I need to make a wind tunnel and accurate lift arm to conduct my experiments with. I'm basing my sting off of an existing design, but the wind tunnel is my original construction. The wind tunnel construction has been the single bar to my progress, and it has been for two years, but there's no better time to hit the ground running.
The tunnel
The wind tunnel I designed is about 2.2 m long by 0.9 m wide. A powerful fan or squirrel cage blower is going to pull air through the system. The suction configuration is ideal for the laminar flow and controlled turbulence conditions as it works off a pressure differential, rather than turbulent, forced air through a system. It also allows me to add or subtract screens from the inlet to vary my TI effectively.
The test section is 200 × 400 × 500 mm in clear acrylic, with one removable panel for airfoil access. Its 200 mm width is exactly the airfoil span, so the section spans wall to wall and the flow stays nominally two-dimensional. Upstream, a 900 × 400 mm settling chamber holds a honeycomb followed by various mesh screens at 40 mm spacing. The contraction is two-dimensional, 900 mm down to 200 mm over an 800 mm length for a ratio of 4.5, with side walls following a fifth-order Bell–Mehta polynomial so slope and curvature both go to zero at each end, which avoids the boundary-layer separation a naive curve would introduce. Thank you to Claude for figuring out the contraction for me.
Design velocity is 10 m/s, giving Re ≈ 67,000 on a 100 mm chord, with fan headroom to 15 m/s. The blower is variable-speed, mounted on a separate vibration-isolated stand and coupled to the tunnel through a flexible joint. Fan vibration reads as force on the load cells, and the signals being measured are small enough that it would matter. Exhaust ducts outdoors, since DBD ionization produces ozone.
Instrumentation
An external balance carries four load cells in a cross, two for lift and two for drag, on 20 × 20 mm aluminum sting isolated from the tunnel. The airfoil connects through a single hollow rod passing through a bushed wall slot, which also routes the actuator leads away from the sensors.
An Arduino reads two HX711 24-bit ADCs, one per cell pair, and streams timestamped CSV at about 5 Hz. An IR remote handles tare, calibrate, and collect. Calibration is blocked while collection is running so a stray button press cannot corrupt a run partway through. It uses a 500 g reference mass and averages 30 samples per cell to get a newtons-per-count scale factor.
The actuator
Two copper-tape electrodes separated by a Kapton dielectric, one exposed on the suction surface and one encapsulated underneath. At around 15 kV AC the air near the exposed edge ionizes, and the moving charge drags neutral air along with it. The result is a thin wall jet running tangentially over the surface. It sits at the leading edge, following Bouremel et al. (2013), who report the best combined result there across lift, drag, and stall delay. Chord 100 mm, span 200 mm.
The HV AC is generated using a ZVS flyback transformer, the same kind used in CRT televisions. This creates a controllable, somewhat stable, and somewhat safe output without spending a kidney on a HV power supply. The voltage and output need to be measured, I'm not entirely sure how I'm going to do that without an oscilloscope, but it will all work out, hopefully.
Simulation
OpenFOAM, LM Transition SST, with the actuator represented as a distributed body force rather than resolved electrodes, following Abdollahzadeh et al. (2016). Intended as a validation companion to the experimental data rather than a primary result. In progress. I plan on conducting both 2D and 3D simulations using Suzen-Hwang body forces.
What I have not solved
A 15 kV AC supply sitting inches from an unshielded HX711 puts noise into the force signal, and it only does it when the actuator is on. That is the same condition I am testing. So an on/off force difference would be measuring the power supply as well as the plasma. I need to compare how much is interference by comparing on vs off, as well as shielding the setup as best I can.
The TAL220 cells are 98 N full scale. The forces I care about are around 1.2 N total, and the drag signal is 0.024–0.12 N, which falls inside the HX711 noise floor of roughly ±0.02–0.05 N. As built, the balance cannot resolve drag. Fixing it means smaller cells, about 100–300 g for drag and 500 g to 1 kg for lift, plus reprinted mounts to fit the smaller body.
Voltage is set at roughly 15 kV, but frequency, duty cycle, and waveform shape are not. Typical DBD operation spans 1–10 kHz, and the choice changes actuator authority substantially. The ZVS is slightly tunable; oscilloscope verification is imperative.
One of my 10kg load cells failed; I bought a new one already.
To use turbulence intensity as a controlled variable I have to measure it accurately, which means a fast differential-pressure transducer or a hot-wire probe. I have neither; estimation is the enemy in this case.
Status
Tunnel design complete and locked, and fabrication is underway. Force balance in prototype form, pending the replacement cell and the load-cell resize. Actuator designed, materials specified, not yet assembled. OpenFOAM is not started per-se, rather, I have been figuring out the workflow and how I am going to balance computation and mesh fidelity, the same as my ANSYS tubercle work, just more demanding. The plan is to build the wind tunnel in sections, curved contraction section first because I don't know what material I am going to use yet. The actuator is going to be built side by side with the tunnel.
Tools
OpenFOAM, Autodesk Fusion 360, Arduino.