You are on page 1of 4

CHIN. PHYS. LETT. Vol. 29, No.

4 (2012) 044701

*
Characterization of the Flow Separation of a Variable Camber Airfoil
YANG Wen-Chao(杨文超), WANG Hui(王辉), YANG Jian-Ting(杨剑挺), YANG Ji-Ming(杨基明)**
Department of Modern Mechanics, University of Science and Technology of China, Hefei 230027

(Received 25 August 2011)


An experimental investigation is carried out to study the flow separation behaviors of a variable camber airfoil.
The aerodynamic load measurements and related flow visualization show that there are two types of stalls caused
by the deformation on the camber: the leading-edge stall and the trailing-edge stall. Static measurements of
aerodynamic force show a drastic leading-edge stall, while the serial measurements on an airfoil with camber
deformation illustrate a trailing-edge stall and gradual bending-over on the aerodynamic coefficient curve. Under
flow separation circumstances, the flow structure is related not only to current boundary conditions, but also the
previous flow characteristics, so the quasi-steady aerodynamic characteristics are significantly distinct from those
of the static measurements.

PACS: 47.15. Cb, 47.32.Ff, 47.85. Gj DOI:10.1088/0256-307X/29/4/044701

The performance of low Reynolds-number airfoils ward trailing-flap deflections on the velocity and vor-
is greatly determined by the relatively feeble separa- ticity flow fields around an NACA 0015 airfoil sub-
tion resistance of the laminar boundary layer. Better jected to deep-stall oscillations. Lian and Shyy[8]
understanding and ultimately control of laminar sep- found that the self-excited flexible surface vibration
aration is therefore valuable for improving the flight can affect the separation and transition positions.
mechanics of low Reynolds-number aircraft, such as Stephan and Rolf[9] used a combined experimental and
micro air vehicles (MAVs). Hence, laminar separation computational study, which was presented for an air-
is a canonical topic in low speed flow, having been in- foil undergoing a combined pitching and plunging mo-
tensively investigated both from the viewpoint of fun- tion to see whether airfoil flexibility can have a posi-
damental fluid mechanics and in the context of the tive impact on the propulsive efficiency. These studies
aerodynamics of airfoils and wings.[1−3] have revealed the huge potential of airfoil deformation
Many approaches were implemented to describe on flow control. Moreover, aircraft morphing, espe-
the separation flow of low Reynolds-number airfoils, cially the variable camber airfoil, has the potential to
including numerical efforts, aerodynamic load mea- greatly improve system performance over an aircraft’s
surements and flow visualization methods. The early nominal operational envelope to allow a single aircraft
studies mainly concentrated on the structure of the to perform multiple missions efficiently, and even to
laminar separation bubble (LSB) and the hystere- significantly expand its operating envelope.[10] Thus
sis loop of aerodynamic load coefficients.[2,4] Most the effect of an airfoil’s camber on the characteristics
of these studies focused on parameters such as free- of laminar separation is a challenge for future aerody-
stream turbulence or angle of attack (AOA). Hoff- namics research. In this Letter, the camber effects on
mann Jon[4] found that the effects of free-stream tur- flow separation are investigated experimentally.
bulence can greatly affect the hysteresis loop of the Firstly, a particular type of variable camber air-
aerodynamic coefficient for a low Reynolds-number foil was designed, whose camber could be smoothly
airfoil. Serhiy et al.[5] studied the development of changed from zero to the desired extent. The base-
coherent structures in the separated shear layer and line airfoil was assumed to have an NACA 0015 pro-
wake of an airfoil in low Reynolds-number flows for file. A rigid D-spar, the primary load carrying mem-
a range of airfoil chord Reynolds numbers and AOAs, ber, extended from the leading edge of the airfoil to
experimentally. With the development of smart mate- 30 percent chord (Fig. 1(a)). The D-spar did not un-
rials in aircraft construction, the effect of airfoil shape dergo any deformation in the chordwise direction, and
on the characteristics of flow separation has recently consequently only the aft section of the D-spar can
received much more attention. David and Jamey[6] deform.[11] This aft section of the spar was specially
indicated that oscillating the curvature of the upper designed to have an internal substructure as shown
surface of an NACA 4415 airfoil has a pronounced ef- in Fig. 1(b). The substructure comprised five rows of
fect in reducing the degree of separation. Through the hinge-like units ranked serially along the chordwise
particle-image-velocimetry method, Gerontakos and direction. The design of the substructure was to sup-
Lee[7] studied the influence of both upward and down- port the flexible skin, which provides the aerodynamic

* Supported by the National Natural Science Foundation of China under Grant No 90816026, and the Knowledge Innovation

Project of Chinese Academy of Sciences (No KJCX-YW-L05).


** To whom correspondence should be addressed. Email: jmyang@ustc.edu.cn

© 2011 Chinese Physical Society and IOP Publishing Ltd

044701-1
CHIN. PHYS. LETT. Vol. 29, No. 4 (2012) 044701

shape and is capable of morphing. A servomotor was One side of the force balance was connected to the
fixed in the D-spar to offer power to pull the lower test model and the other side was fixed on a half arc
surface of the aft section into the D-spar to a certain drive, which can provide a variable angle of attack. In
extent, with the help of gearing. As part of the lower the experiments, a digital video camera (CASIO EX-
surface was pulled into the D-spar, the discrepancy be- F1) was employed to record the deformation and the
tween the area of the upper and lower surfaces came related flow field of the variable camber airfoil. The
up and the airfoil then became asymmetric, in other whole arrangement of the model and support is shown
words it had some camber. The D-spar was made by in Fig. 2.
acrylic materials and the skin of the aft section was
polypropylene plastics. The airfoil camber produced Serial measurement:
0.8
due to actuation loading can be quantified in differ-

Aerodynamic coefficients
ent ways, including vertical tip deflection or equiva-
lent flap angle.[11] In the present study the former is 0.6

used, so we use the deflection ratio 𝑑* = 𝑑/𝑐 to define


the camber, where 𝑐 is the airfoil chord and 𝑑 is the 0.4

trailing-edge deflection, as shown in Fig. 1(a).


0.2

Rigid D-spar Compliant section


(a) Undeformed
0.0
(baseline) airfoil
0 5 10 15 20

d (trailing-edge Angle of attack (deg)


Morphed deflection)
(cambered) airfoil
Fig. 3. Coefficients of lift and drag vs angle of attack.
c (chord)

(b) Servomotor Gear Scroll


In order to find the differences between the effects
of variable camber and the normal behaviors of a wing
pitching motion, some experiments on the variations
of aerodynamic coefficients versus AOA were carried
out as a first step, although they were not an empha-
Axis Hinge-like unit Skin
sis, except for the purpose of direct comparison. In
the tests there are two methods to obtain aerodynamic
Fig. 1. Construction of the variable camber airfoil: (a) lift and drag. Both were carried out on undeformed
basic design, (b) internal structure.
(Baseline) airfoil, so the results are similar to those of
the NACA 0015 airfoil. One process is to set the AOA
Flow Window of tunnel
Leading edge at a given angle at wind-off conditions, and then the
tunnel speed is increased to the test Reynolds num-
Force balance
ber. Once the flow settles down in the tunnel, the
Digital camera measurements can be made. This kind of measure-
D-spar
ment is described as the “static measure” in this study.
Morphing part The other routine is to set the model to 0∘ AOA un-
der the wind-off condition, and then the tunnel speed
is increased to the test Reynolds number. Once the
Half arc drive End disk
tunnel stabilized, AOA was increased to about 20∘ ,
while the data was collected at prescribed AOA, so
the measurement process was called the “serial mea-
Fig. 2. Sketch of the experimental setup.
sure”. After its maximum value, AOA was decreased
The aerodynamic load measurements, obtained via back to 0∘ without stopping the tunnel. As shown in
a five-component force balance, were carried out in a Fig. 3, the static measure underwent an increase in 𝐶𝑙
low-speed wind tunnel at the University of Science prior to its drastic stall at 𝛼𝑠𝑠 = 16.5∘ . On the other
and Technology of China. The variable camber air- hand, the curve for serial measures shows a significant
foil, with a chord of 200 mm and a span of 116 mm, hysteresis loop with a span of about 3∘ . There ap-
was used as the test model. The Reynolds number was pears to be a downward shift in the lift curve between
fixed at 1.1 × 105 so the wind velocity was 8.0 m/s and 16.5∘ and 19.5∘ . The similar hysteresis loop can also
measured with a hot-wire anemometry. The two ends be found in the 𝐶𝑑 curve. In addition, the results of
of the wing model were fitted with 30 cm diameter end the static measure obviously correspond to the lower
disks to reduce three-dimensional effects. The force side of the hysteresis loop obtained by changing AOA
balance was measured to an accuracy of ±0.04 N in at fixed wind speeds.
the lift direction and ±0.025 N in the drag direction. In the investigation on the effect of camber on
044701-2
CHIN. PHYS. LETT. Vol. 29, No. 4 (2012) 044701

flow separation, we used two similar methods to mea- process as shown in Fig. 4. (3) The camber 𝑑/𝑐 is set
sure the aerodynamic lift and drag. To better reveal to about 0.07 at the wind-off condition, and then the
the effect of camber on flow separation, we set the tunnel speed is increased to the test Reynolds num-
initial angle of attack (AOA) at the maximum lift ber. Once the tunnel is stabilized, the camber 𝑑/𝑐 is
AOA𝑙max = 16.5∘ for all the cases of the two meth- increased to about 0.13 and then decreased back to
ods. First we use static measure to study the drastic nearly 0, while collecting data at the prescribed cam-
stall camber. We increase 𝑑/𝑐 from 0 to 0.11 with a ber. (4) The motion is started like path 3, but the
step length of 0.01. As shown in Figs. 5(a) and 5(b), camber 𝑑/𝑐 of the airfoil is fixed at 0.13, as described
the maximum lift is obtained at the camber deflection in path 2 after the increased process. All of these
ratio 𝑑/𝑐 of 0.03, after which 𝐶𝑙 will drastically drop paths are shown in Fig. 4. Each step of camber 𝑑/𝑐
and 𝐶𝑑 will abruptly increase. deformation was about 0.007, and the time for each
camber deformation is about 0.5 s, while the airfoil
0.12 will subsequently stay in such a shape for 10 s. This
Path 1 kind of route is used to study the quasi-steady aero-
0.10 Path 2
Path 3 dynamic character, so each deformation is very small
Path 4
0.08 and the time for the whole route is large enough to
0.06
avoid the effect of deformation velocity.
Camber

0.04

0.02

0.00

0 100 200 300 400 500

Time (s)
(a)

Fig. 4. Camber vs time for four paths.

1.1 (a)

1.0

0.9
Static airfoil
0.8 Path 1 (b)
Path 2
0.7
Path 3 Fig. 6. Smoke wire visualization results at the maxi-
Path 4
0.6 mum camber (𝑑/𝑐 = 0.13) under an AOA of 14.5∘ and
0.5 𝑅𝑒 = 6.8 × 104 : (a) path 2, (b) path 4.
0.4

0.32
The aerodynamic force of each step was averaged
(b)
0.30
corresponding to the specific shape of the airfoil to
0.28
describe the relationship between aerodynamic forces
0.26
and camber. The results of the four paths are plotted
in Figs. 5(a) and 5(b). For path 1, the stall does not
0.24
Static airfoil
Path 1
completely happen before deformation, so the curves
0.22
Path 2 of 𝐶𝑙 and 𝐶𝑑 are similar to those of the static mea-
0.20 Path 3
Path 4 sure prior to the stall camber. However the curve
0.18
of path 1 doesn’t show a drastic drop in 𝐶𝑙 for fur-
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
ther camber deflection. Instead there is a gradual
Camber
bending-over of the lift curve at maximum lift. As for
Fig. 5. Aerodynamic coefficients vs camber: (a) coeffi- the explanation of the mechanism, similar phenomena
cient of lift, (b) coefficient of drag.
might be found on the stall characters between thinner
To investigate the quasi-steady aerodynamic char- and thicker airfoils referred by Anderson,[12] although
acteristics and the effect of different initial flow con- in the present case the corresponding parameter was
ditions, we designed four types of paths for the serial camber instead of AOA. We can therefore infer that
measures. (1) Camber 𝑑/𝑐 is set to 0 at the wind-off different stall types determine the distinction in the 𝐶𝑙
condition, and then the tunnel speed is increased to curve trend between the two measuring processes. For
the test Reynolds number. Once the tunnel is stabi- the static measure the stall is a kind of leading-edge
lized, the camber 𝑑/𝑐 is gradually increased to about stall. During this stall process, flow separation takes
0.13 and then decreased back to near 0, while data is place rather suddenly and abruptly over the entire top
collected at the prescribed camber. (2) The motion is surface of the airfoil, with the origin of this separa-
started like path 1, but the camber 𝑑/𝑐 of the airfoil is tion occurring at the leading edge, so the lift curve of
fixed at 0.13, more than 1 minute after the increased the static measure shown in Fig. 5(a) is rather sharp-
044701-3
CHIN. PHYS. LETT. Vol. 29, No. 4 (2012) 044701

peaked in the vicinity of 𝐶𝑙max , with a rapid decrease reattached flow is almost the same as that of the static
in 𝐶𝑙 near the stall. However, for path 1 the stall measure. This character is similar to the analogous re-
is a kind of trailing-edge stall. In this kind of stall sult on AOA as shown in Fig. 3. As expected, the lift
a progressive and gradual separation from the trail- and drag curves of path 4 show a leading-edge stall
ing edge toward the leading edge takes place as the throughout, which is close to the results in the static
camber is increased. Thus the trailing-edge stall is a measure. At the position of maximum camber, the
kind of “soft” one compared to the leading-edge stall. aerodynamic forces convey a stable stall character no
Moreover, no significant hysteresis loop was found in matter how long the airfoil keeps this shape. Although
either the life curve or drag curve of path 1. This is both paths 2 and 4 remain at the same maximum cam-
quite different from the curves based on the variation ber for a long time and have the same boundary condi-
of AOA. The baseline airfoil is close to NACA 0015, tions, their aerodynamic characters are totally differ-
which is a kind of thin airfoil defined by Anderson,[12] ent. Therefore under flow separation circumstances,
hence its lift curve for the camber’s variation in static the flow structure is related not only to current en-
measure is a kind of leading-edge stall. On the con- vironment but also to the previous flow characters,
trary, the stall of path 1 is a kind of trailing-edge stall. so the history of flow evolvement must be considered
This kind of stall is caused by a gradual movement of in the design of low Reynolds-number aircraft. This
separation from the trailing edge toward the leading conclusion is also supported with the help of flow vi-
edge, thus the process is reversible and has no obvious sualization, as shown in Figs. 6(a) and 6(b). A similar
hysteresis character. For path 2 both the lift curve technique on smoke wire by Liu et al.[13] was used to
and drag curve are similar to those of path 1, so it depict the flow separation structure. The incoming
can be concluded that the trailing-edge stall of path flow speed was 5 m/s and the AOA was set at 14.5∘
1 is a kind of quasi-steady phenomenon when the 𝑑/𝑐 (in the vicinity of the static-stall AOA in such a flow
reaches 0.13. In path 2 the trailing-edge of the air- speed). Figure 6(a) describes the flow of path 2 at
foil keeps a camber deflection of 26 mm for at least the maximum camber. Figure 6(b) depicts the flow at
1 minute from the axis of symmetry after a consec- the corresponding camber of path 4. From the flow
utive increase. In Figs. 5(a) and 5(b), both the lift visualization it is easy to tell the difference between
curve and the drag curve of path 2 have a group of the leading-edge stall and the trailing-edge stall, al-
dense points overlapping at the maximum position of though the two flow phenomena have the same bound-
the camber, which means that the aerodynamic force ary conditions and airfoil shapes. It is worth noting
remains stable after reaching its maximum camber. It that the aerodynamic forces between paths 3 and 4
is clearly seen that the flow around the airfoil will stay differ slightly during the stall stage. Since the history
at the above-mentioned steady state even if the static of flow evolvement can affect the separation character-
measure shows an obvious drop in the lift curve under istics and the leading-edge stall is extremely unstable,
the same airfoil shape. The results of path 2 prove slight variations in the wind tunnel’s starting proce-
that the difference in aerodynamic forces between the dures may lead to the different separation features and
static measure and the serial measures of path 1 does locations between various measurements.
not come from the velocity of deformation, because
even if the deformation stops at the maximum camber,
the aerodynamic forces show no noticeable distinction References
from the deformation process. It can therefore be con- [1] Lissaman P B S 1983 Ann. Rev. Fluid Mech. 15 223
cluded that the trailing-edge stall for paths 1 and 2 is [2] Mueller T J and DeLaurier J D 2003 Ann. Rev. Fluid Mech.
a kind of quasi-steady character which is independent 35 89
of deformation velocity. [3] Ishak A 2009 Chin. Phys. Lett. 26 034701
[4] Hoffmann Jon A 1991 AIAA J. 29 9
In path 3, the airfoil whose initial camber 𝑑/𝑐 was [5] Yarusevych S, Sullivan PE and Kawall JG 2009 J. Fluid
around 0.07, first experienced a leading-edge stall after Mech. 632 245
the tunnel speed was increased to the test Reynolds [6] Munday D and Jacoby J 2002 J. Aircraft 39 1
number. The airfoil then deformed to the maximum [7] Gerontakos P and Lee T 2008 Exp. Fluids 45 955
[8] Lian Y and Shyy W 2007 AIAA J. 45 1501
camber and ultimately recovered to an undeformed [9] Bansmer S, Radespiel R, Unger R, Haupt M and Horst P
airfoil under the test Reynolds number. Due to the 2010 AIAA J. 48 1959
leading-edge stall, the airfoil remained in poor aero- [10] Szodruch J and Hilbig R 1988 Prog. Aerospace Sci. 25 297
dynamic performance for a long time until its trailing- [11] Gandhi F and Anusonti-Inthra P 2008 Smart Mater. Struct.
17 015025
edge got back to the stall camber position, as shown [12] Anderson J D 2007 Fundamentals of Aerodynamics (New
in Figs. 5(a) and 5(b). Moreover, the position where York: McGraw-Hill Education)
the airfoil in path 3 transits from leading-edge stall to [13] Liu Y, Yang W C and Yang J M 2011 J. Visualization 14
141

044701-4

You might also like