Active Flow Control Using Transverse Travelling Waves Speaker: Aki Pakarinen Imperial College

Design Objectives A net skin friction reduction of 20% would save the airlines roughly $40m. per day worldwide on direct fuel costs. (2004 fuel prices and consumption) Rugged, reliable control system Surface mounted Generous net energy savings Open loop, always “on-design” Commercial operation

Control Schemes Large-scale longitudinal vortices “Wall turbulence manipulation by large-scale streamwise vortices,” Gaetano Iuso, Michele Onorato, Pier Giorgio Spazzini, Gaetano Maria Di Cicca, JFM (2002) Oscillating wall “On the effects of lateral wall oscillations on a turbulent boundary layer,” Pierre Ricco, Shengli Wu, Experimental Thermal and Fluid Science (2004)

Du, Symeonidis & Karniadakis (2002) “Drag reduction in wall-bounded turbulence via a transverse travelling wave” by Du, Symeonidis and Karniadakis, JFM 2002 DNS simulations, using a force resembling a transverse travelling wave, showed > 30% reduction in wall shear stress through the suppression of turbulence production. The forcing was confined to the viscous sublayer

Flow Visualisation At y + = 4. Blue indicates low-speed streaks and yellow-red high speed streaks. Plots of instantaneous streamwise velocity on both unactuated (top) and actuated (bottom) walls. Top plot shows characteristics 100 viscous unit spacing of low-speed/high-speed streak pairs. Controlled case shows distinct lack of streaks, which are replaced by a large region of low-speed fluid

Control Variables Frequency T + Strong dependence Force magnitude I Smaller “interaction parameter” showed better results Wavelength + The longer the better Penetration length  For Re  = 150, they found I T +   1 to produce C ƒ of 30%

Non-Ideal Waveforms (a)C ƒ of -30% (b)C ƒ increase (c)no change (d)C ƒ of ≈ -20% Initial simulations were done with an “ideal” sine wave. Further studies showed the feasibility of using piecewise approximations

Zhao et al. (2004) “Turbulent drag reduction by travelling wave of a flexible wall,” H. Zhao, J.-Z. Wu, J.-S. Luo, Fluid Dynamics Research (2004) DNS simulation showed 30% drag reduction, using actual wall motion. A pseudo-Stokes layer was seen, which has a large spanwise structure and is relatively homogenous in a streamwise direction and interferes with the turbulence regeneration cycle

Re Effects As Re increases, viscous sublayer becomes thinner. Power input decreases with decreasing actuation deflection. Therefore, as Re increases, control becomes “easier”, though the associated frequencies increase. Power required due to skin friction ~ Re x 17 /

Re Effects cont. Net power savings per unit areaPower savedPower spent Using simple boundary layer and structural mechanics approximations, an equation for the net power savings of monolith flow control device can be derived. Positive laboratory results should translate directly to higher Reynolds numbers, despite the requirement for higher frequencies

Into the Lab? Produce an accurate, transverse travelling wave (TTW) of < viscous sublayer height, with ability to control amplitude, wavelength and frequency… At lab conditions, these starting values are: wavelength λ: mm amplitude A: mm frequency ƒ: Hz length and width of actuated flat plate: to suit tunnel

Model Actuation Design parameters signify a definite challenge. Many options were considered and rejected: Springs Magnets Hydraulics Mechanical (piano keys, revolving disks, moving ridges) Not one could produce the frequency and the amplitude criteria. The search for a viable system continued until a very specific actuator was found

Model Actuation cont. Face ® International Thunder ® TH11R-3 pre-curved piezoceramic actuators Peak to peak response: 0.8mm Frequencies up to 100Hz ±500V operating voltage Teflon rail Perspex cylinder Actuator Perspex base plate Simply Supported Mounting Frequency Response 50V pp 3mm

Apparatus is capable of the following: Wavelength: 7mm – 105mm Frequency: up to 100hz Amplitude: up to 0.8mm Net savings? Possibly, but worst case scenario would be 11.5W spent to save 0.9W. Apparatus Capabilities 7mm Up to 105mm

Future Work Initial system validation (ongoing) Current set-up has frequency related issues. System evolution ongoing, with two analogue back-up designs ready. Complete system manufacture Testing Investigating effects of changing control variables (frequency, amplitude and wavelength) Including mass balance, hotwire and PIV measurements This work leads onto specification and design of the more practical surface, as presented by Kevin

Conclusion DNS has shown the possibility of significant drag reductions using transverse travelling waves confined to the viscous sublayer. An experimental set-up and further work has been suggested. Any questions?