1. Kevin Knowles , Peter Wilkins, Salman Ansari, Rafal Zbikowski
Integrated Computational and Experimental Studies of Flapping-wing Micro Air Vehicle Aerodynamics
Kevin Knowles , Peter Wilkins, Salman Ansari, Rafal Zbikowski
Department of Aerospace, Power and Sensors
Cranfield University
Defence Academy of the UK
Shrivenham, England
3rd Int Symp on Integrating CFD and Experiments in Aerodynamics, Colorado Springs, 2007

2. Outline Introduction Flapping-Wing Problem Aerodynamic Model
LEV stability
Conclusions
Knowles et al.

3. Micro Air Vehicles Defined as small flying vehicles with
Size/Weight: mm/50–100g
Endurance: 20–60min
Reasons for MAVs:
Existing UAVs limited by large size
Niche exists for MAVs – e.g. indoor flight, low altitude, man-portable
MAV Essential (Desirable) Attributes:
High efficiency
High manoeuvrability at low speeds
Vertical flight & hover capability
Sensor-carrying; autonomous
(Stealthy; durable)
Microgyro
Microsensors
Knowles et al.

4. Why insect-like flapping?
Insects are more manoeuvrable
Power requirement:
Insect – 70 W/kg maximum
Bird – 80 W/kg minimum
Aeroplane – 150 W/kg
Speeds:
Insects ~ 7mph
Birds ~ 15mph
Knowles et al.

5. Wing Kinematics – 1 Flapping Motion Key Phases sweeping heaving
pitching
Key Phases
Translational
downstroke
upstroke
Knowles et al.

6. Wing Kinematics – 1 Flapping Motion Key Phases sweeping heaving
pitching
Key Phases
Translational
downstroke
upstroke
Rotational
stroke reversal
high angle of attack
Knowles et al.

7. Wing Kinematics – 2
Knowles et al.

8. Mechanical Implementation
Knowles et al.

9. Generic insect wing kinematics
Three important differences when compared to conventional aircraft:
wings stop and start during flight
large wing-wake interactions
high angle of attack (45° or more)
Complex kinematics:
difficult to determine
difficult to understand
difficult to reproduce
Knowles et al. 9

10. Aerodynamics Key phenomena unsteady aerodynamics leading-edge vortex
apparent mass
Wagner effect
returning wake
leading-edge vortex
[Photo: Prenel et al 1997]
Knowles et al.

11. Aerodynamic Modelling – 1
Quasi-3D Model
2-D blade elements with
attached flow
separated flow
leading-edge vortex
trailing-edge wake
Convert to 3-D
radial chords
Robofly wing
Knowles et al.

12. Aerodynamic Modelling – 1
Quasi-3D Model
2-D blade elements with
attached flow
separated flow
leading-edge vortex
trailing-edge wake
Convert to 3-D
radial chords
cylindrical cross-planes
integrate along wing span
Knowles et al.

13. Aerodynamic Modelling – 2
Model Summary
6 DOF kinematics
circulation-based approach
inviscid model with viscosity introduced indirectly
numerical implementation by discrete vortex method
validated against experimental data
Knowles et al.

14. Flow Visualisation Output
Knowles et al.

15. Impulsively-started plate
Knowles et al.

16. Validation of Model
Knowles et al.

17. The leading-edge vortex (LEV)
Insect wings operate at high angles of attack (>45°), but no catastrophic stall
Instead, stable, lift-enhancing (~80%) LEV created
Flapping wing MAVs (FMAVs) need to retain stable LEV for efficiency
Why is the LEV stable? Is it due to a 3D effect?
Difficulties involved with ascertaining kinematics of insect flight – keeping insect still for observation. Aerodynamics (seeing what the flow is doing) even more difficult – flow visualisation (smoke) problems.
Four main aspects:
Wing stops/starts regularly in flight – doesn’t happen with either fixed or rotary wing aircraft where (we hope) wing only starts/stops when aircraft is on the ground
Wake is not ‘left behind’ but remains in close proximity to wing for relatively long time period.
Angle of attack regularly exceeds stall angle but no catastrophic stall – stable leading edge vortex
Many insects use specialised techniques to enhance lift – FMAVs may be able to use some of these.
Knowles et al. 17

18. 2D flows at low Re Re = 5 Re = 10 Knowles et al.
What effect does increasing Reynolds number have for 2D flows?
How is the flow affected by changes in section?
Explore LEV – look for spanwise flow. Is LEV stable at high Re? FMAVs likely to operate at Re = 15000ish.
Knowles et al. 18

19. Influence of Reynolds number
Basic shape of graphs are the same
LEV build up/shed then TEV then LEV. Cycle repeats.
Increased unsteadiness at higher Re.
α = 45°
Knowles et al. 19

20. 2D flows
Re = 500, α = 45°
Knowles et al.

21. Influence of Reynolds number
Basic shape of graphs are the same
LEV build up/shed then TEV then LEV. Cycle repeats.
Increased unsteadiness at higher Re.
α = 45°
Knowles et al. 21

22. Kelvin-Helmholtz instability at Re > 1000
Notice LEV breakdown at high Re.
Re 500
Re 5000
Knowles et al. 22

23. Secondary vortices Re = 1000 Re = 5000 Knowles et al.
Notice LEV breakdown at high Re.
Re = Re = 5000
Knowles et al. 23

24. 2D LEV Stability
For Re<25, vorticity is dissipated quickly and generated slowly – the LEV cannot grow large enough to become unstable
For Re>25, vorticity is generated quickly and dissipated slowly – the LEV grows beyond a stable size
In order to stabilise the LEV, vorticity must be extracted – spanwise flow is required for stability
Rotational flow produces conical LEV.
Knowles et al. 24

25. Structure of 3D LEV
Knowles et al.

26. Stable 3D LEV
Re = 120
Re = 500
Knowles et al.

27. Conclusions
LEV is unstable for 2D flows except at very low Reynolds numbers
Sweeping motion of 3D wing leads to conical LEV; leads to spanwise flow which extracts vorticity from LEV core and stabilises LEV.
3D LEV stable & lift-enhancing at high Reynolds numbers (>10 000) despite occurrence of Kelvin-Helmholtz instability.
Knowles et al.

28. Questions?
Knowles et al.