Wing-gust encounters produce highly complex, vortex-dominated flows which are difficult to study experimentally, mainly due to the sensitivity of these flows to the forcing parameters and the challenges associated with generating repeatable gusts in a wind tunnel. Finite span and flexibility are known to affect the gust response of a wing, but there is a notable scarcity of experimental data relating to the high aspect ratios and low natural frequencies of wings relevant to civil and general aviation. This thesis presents a novel ‘oscillating fence’ gust-generator, which uses aerodynamic fences deployed sinusoidally from the wind tunnel walls to introduce an oscillating transverse component in the freestream. Unsteady gust angle was experimentally surveyed at reduced frequencies from k = 0 to k = 0.181. Further experiments investigated the gust aerodynamics of one rigid and two flexible wings of semi-aspect ratio sAR = 5.
Particle Image Velocimetry (PIV) of the test section was acquired, with no wing present, to studythe gust field. It revealed reasonably uniform gusts that were non-convective, but could approximate convective gusts at the low k considered herein. The small transverse velocities were difficult to quantify accurately with PIV, so an alternative calibration of gust angle was performed by measuring the unsteady lift of the rigid wing, at zero geometric angle of attack, in gusts at a range of frequencies. Theodorsen’s function was used to compute ‘effective’ gust angle profiles from the measured unstead lift, revealing distorted sinusoidal gusts that lagged and decayed with increasing k in a manner typical of a first-order system. The rig was capable of peak-to-peak gust amplitudes up to 6.2o at k = 0 which decayed to 3o at k = 0.181.
Lift of the rigid wing was measured along with the flow field at the mid-span in gusts of varying frequency. Geometric angles of attack of 0o, 5o, 10o and 12o were tested to cover flows ranging from attached to deeply stalled. The aerodynamic response was mainly dependent on the maximum effective angle of attack and reduced frequency. Attached flows behaved as expected, producing CL − eff loops with counter-clockwise hysteresis which aligned with the static lift curve, though some clockwise hysteresis was observed at 0 = 5o due to onset of stall. Marginal to moderate excess of the static stall angle caused periodic separation and reattachment at low frequency, but at higher frequencies the marginal cases became fully attached, while those with significant stall penetration became fully separated. Excess of the static stall angle caused smaller lift amplitudes at low frequency, but at higher frequency lift overshoot and clockwise hysteresis led to increased amplitude which surpassed attached flows by a substantial margin. Analysis of instantaneous flow fields revealed Leading-Edge Vortices (LEVs) which were responsible for the lift overshoot in separated flows. However, due to the low amplitude and frequency of the gusts, vortex lock-in did not occur, and cycle-to-cycle variations in the position of vortices caused these to be lost during phase-averaging.
Two flexible wings were purpose-built to have natural bending frequencies within the range of gust frequencies. Their lift and mid-span flow field were measured, while their structural dynamics were recorded by Digital Image Correlation (DIC). Under steady conditions the low flexibility wing experienced a post-stall rise in lift due to self-excited torsional vibration, which was seen to encourage flow reattachment. The high flexibility wing would twist under load and increase its effective incidence, leading to a lower stall angle and CLMAX when compared to the rigid wing. During gust encounters, the structural dynamics were mainly dependent on the gust frequency k and the wing’s natural bending frequency kn, and were insensitive to geometric angle of attack. For k < kn, wing deformation was approximately quasi-steady and had little effect on aerodynamics. For k kn resonance caused large bending oscillations which reduced the effective gust amplitude perceived by the wing. For k > kn the wing’s motion was damped. The normalised lift amplitude of the flexible wings was dependent on maximum effective angle of attack for k > 0, like that of the rigid wing. However, stalled cases experienced smaller amplitudes than on the rigid wing due to their lower stall hysteresis. For deeply stalled flows, lift amplitude and hysteresis was independent from geometric angle of attack. The flow fields behaved as per the rigid wing. However, during resonance, the effects of the gust on the flow field were reduced due to the aforementioned relief of the wing’s effective angle of attack.
The gust generator presented herein is the first of its kind and is capable of producing repeatable gusts at competitive amplitudes and frequencies. It has been used to carry out the first unified study of high-AR wings, both rigid and flexible, in gusts of varying frequency.
|Date of Award
|28 Apr 2021
|James Dyson Foundation
|Ismet Gursul (Supervisor) & David Cleaver (Supervisor)
- Fluid dynamics
- flow separation
- finite wing