Investigation of Active Control of Boundary Layer Transition in Laminar Separation Bubbles

Abstract

The presented research addresses the dynamics of the transition process in laminar separation bubbles (LSBs). Two major objectives will be discussed in this work: an investigation of the natural transition process in an LSB forming in the flow along a flat plate when subjected to an adverse pressure gradient (APG) and active flow control (AFC) exploiting the inherent instabilities in the LSB shear layer to control and delay laminar to turbulent transition.Extensive characterization of the boundary conditions in the experiment indicate a low turbulence environment necessary for stability and transition research. Prior to characterization of the LSB, the boundary layer along the flat plate model is investigated in the absence of the APG. Development of Tollmien Schlichting (T-S) waves in the laminar boundary layer along the flat plate show linear growth for several tested amplitudes and frequencies. Comparison to linear stability theory (LST) calculations and accompanying direct numerical simulations (DNS) show reasonable agreement in disturbance profiles and downstream amplitude development. Differences to the numerical results increase with downstream distance and are attributed to a slight favorable pressure gradient and the onset of non-linear behavior around the first maximum in the disturbance amplitude profiles in the experiment. Results confirm adequate quality of the free-stream turbulence (FST) (Tu ≤ 0.035%) and velocity spectra in the Arizona Low Speed Wind Tunnel (ALSWT) for the subsequent LSB transition study. The flow around the displacement body, used to impose the favorable to adverse pressure gradient to the flat plate model, is investigated and flow control measures are installed to ensure smooth, attached flow along the surface of the NACA 643−618 airfoil. The resulting baseline matches the time-averaged LSB from DNS with low levels of random disturbances (Tu = 0.02%). Two major unsteady features are found in the experiment. Low frequency (6 Hz, St = 0.02) content in the shear layer is connected to a large scale ’flapping’ motion, leading to significant periodic change in the reattachment location, causing an expansion/ contraction of downstream half of the LSB. High frequency content is related to the inviscid Kelvin-Helmholtz instability, causing disturbance amplification along the separated shear layer. Finite disturbance growth leads to formation of two-dimensional vortical structures, followed by rapid breakdown to turbulence upstream of mean reattachment. The dominant frequency in the shear layer (centered at 250 Hz, St = 0.88) is found higher than in the DNS (185 Hz, St = 0.671). Linear stability calculations on the matching baseline show a broad peak of unstable frequencies, similar in shape to the experimental results, centered between the peak values found in experiment and DNS. AFC is applied upstream of laminar separation. Initial forcing further upstream in the favorable pressure gradient shows significant strengthening of the two-dimensional roller structures leading to significant reduction in bubble size. All tested AFC in the experiment was successful in suppressing the large scale, low frequency ’flapping’ in the LSB and led to a change in transition dynamics. Notable damping due to the reminder of the favorable pressure, reduces disturbance amplitudes far below the critical amplitude (Acr) suggested by secondary instability analysis (SIA) and rapid breakdown to turbulence similar to the baseline case is observed for all forcing amplitudes. Moving the actuator close to the onset of the adverse pressure gradient increases separation control authority and pressure data suggests the LSB is suppressed at high forcing amplitudes. Time resolved PIV data identifies significant amplitudes of two-dimensional roller structures, especially at low and intermediate amplitudes. Increased forcing amplitudes leads to three-dimensionality in the mean flow causing a peak valley formation predominantly in the streamwise velocity component. Spanwise periodic structures with a wavelength of λz = 1 and a frequency of half of the forced frequency, are observed in all tested cases, with decreasing amplitude at higher forcing amplitudes. Fourier amplitude development suggest an optimal set of forcing parameters at (6 kVpp). Results show a delay in secondary mode amplitude growth and signs of delay of transition in the experiment. This case is compared to numerical results and matches the wavelength of the primary and secondary instability with predictions from LST, SIA and DNS, confirming the same transition dynamics present in the experiment and numerical simulations. Addition of FST in the DNS significantly reduces the delay in transition found in the numerical simulations without FST, explaining the observed differences to the experiment.