Experiments in Fluids (2006) 41: 319–326

DOI 10.1007/s00348-006-0153-6

R ES E AR C H A RT I C L E

S. Burgmann Æ C. Brücker Æ W. Schröder

Scanning PIV measurements of a laminar separation bubble

Received: 10 November 2005 / Revised: 11 April 2006 / Accepted: 11 April 2006 / Published online: 11 May 2006

Springer-Verlag 2006

Abstract Scanning PIV is applied to a laminar separation deployed slat. Especially due to the growing research in
bubble to investigate the spanwise structure and the field of small unmanned air vehicles (UAV), which
dynamics of the roll-up of vortices within the bubble. The are typically cruising at relatively low Reynolds num-
laminar flow separation with turbulent reattachment is bers, laminar airfoil flows become of more and more
studied on the suction side of an airfoil SD7003 at Rey- interest. Usually, the tendency for flow separation under
nolds numbers of 20,000–60,000. The flow is recorded such conditions is more critical than for large-scale air-
with a CMOS high-speed camera in successive light-sheet foils, degrading massively the performance and stability
planes over a time span of 1–2 s to resolve the temporal of the UAV (McMichael and Francis 1997).
evolution of the flow in the different planes. The results For airfoils at low Reynolds numbers a locally con-
show the quasi-periodic development of large vortex- fined flow separation with a turbulent reattachment
rolls at the downstream end of the separation bubble, closing the separated region can be observed (O’Meara
which have a convex structure and an extension of and Mueller 1987). The laminar boundary layer detaches
10–20% chord length in the spanwise direction. These due to an adverse pressure gradient. Kelvin–Helmholtz
vortices possess an irregular spanwise pattern. The evo- instabilities occur, leading to a roll-up of the separated
lution process of an exemplary vortex structure is shown shear layer such that vortex structures evolve in the
in detail starting from small disturbances within the vicinity of the reattachment region. Usually the flow of a
separation bubble transforming into a compact vortex at separation bubble exhibits a complex temporal behavior
the downstream end of the separation bubble. As the and mutual interaction with the mean flow, as shown in
vortex grows in size and strength it reaches a critical state numerical studies, e.g., by Pauley et al. (1990) and
that leads to an abrupt burst of the vortex with a large Maucher et al. (1994) as well as in experimental inves-
ejection of fluid into the mean flow. tigations, e.g., by Gaster (1966) or Watmuff (1999).
To control a laminar separation bubble a good
physical understanding is essential, which requires a
detailed knowledge of the time-dependent three-dimen-
sional flow evolution within the separation bubble as
1 Introduction well as in the upstream and downstream regions. Con-
sidering time-resolved and three-dimensional studies on
Low Reynolds number flows in the order of O(104–105) a laminar separation on the suction side of an airfoil
are of major concern for some significant aerodynamic there are hardly any publications in the literature. Only
applications, e.g., turbomachine blades or flow around a recently, some thorough experimental investigations
have been carried out using hot-wire probes (Watmuff
S. Burgmann (&) Æ W. Schröder 1999), laser-Doppler-anemometry and particle-image-
Department of Fluid Mechanics, Institute of Aerodynamics, velocimetry (Lang et al. 2004; Hain and Kähler 2005).
RWTH Aachen University, Wüllnerstrasse zw. 5 u. 7, Watmuff (1999) and Lang et al. (2004) investigated the
52062 Aachen, Germany
Tel.: +49-241-8095569 separation bubble that emerges on a flat plate due to an
Fax: +49-241-8092257 adverse pressure gradient while Hain and Kähler (2005)
E-mail: s.burgmann@aia.rwth-aachen.de analyzed the separation bubble that develops on the
URL: www.aia.rwth-aachen.de suction side of an SD7003 airfoil. Only the results given
C. Brücker
by Lang et al. and Hain and Kähler are able to give an
Institute of Mechanics and Fluid Dynamics, TU Freiberg, impression of the three-dimensional flow structures in
Lampadiusstraße 4, 09596 Freiberg, Germany time and space by means of flow visualization and
320

phase-averaged PIV-measurements and conventional of 0.0851c the blockage effect can be neglected. To
PIV-measurements, respectively. So the spatial structure characterize the spatial and temporal evolution in the
of the separation bubble on an airfoil is not completely separation region, the standard time-resolved PIV and
investigated and to fully understand the dynamic the SPIV method were applied in selected regions of the
behavior of the laminar separation bubble on an airfoil separation bubble. For all measurements, images were
time-resolved and three-dimensional measurements are a recorded with a CMOS high-speed camera at frame rates
must. Therefore we use in this work time-resolved PIV between 308 and 922 Hz and a resolution of
and scanning PIV (SPIV) (Brücker 1997) to capture the 1,024 · 512 pixels. The light-sheet illumination system
spatial and temporal evolution of the vortex dynamics at consisted of ten adjustable laser diodes each of which
the downstream end of the separation bubble. Regard- had a power of 50 W in continuous mode at a wave-
ing the application of PIV, this is a straightforward task length of 805 nm (Fig. 2). All diodes could be adjusted
due to the small aspect ratio and the high-velocity gra- and positioned independently. A special electronic con-
dients within this region (Raffel et al. 1998). In this pa- trol allowed to pulse the laser diodes in any desired se-
per, we focus on the three-dimensional and temporal quence or simultaneously. The frame capture signal was
resolution of a laminar separation bubble, which is synchronized with the pulse control such that each im-
generated at the suction side of a laminar airfoil SD7003 age in the sequence corresponded to a pulse of a certain
at low Reynolds numbers and at moderate angles of laser diode.
attack between 4 and 8. The paper is organized as First, time-resolved 2C-PIV measurements were
follows. First, the experimental method is described. taken in a single light sheet positioned in the flow
Subsequently, the flow field is analyzed based on PIV direction and perpendicular to the airfoil surface to
measurements in differently oriented cross-sections. In detect the local streamwise and spanwise expansion and
the final section, an overall discussion leading to a the dynamic behavior of the separation bubble. For
qualitative model of the separation structure is presented these measurements only one single laser diode in pulsed
and a conclusion is given. mode was used. The flow field around the airfoil was
covered with three slightly overlapping measurement
regions. To resolve the dynamic behavior an enlarged
2 Experimental method area of the reattachment region was investigated. Next,
SPIV measurements were done using different orienta-
PIV measurements of the laminar separation on the tions of the scanning light sheets. Firstly, the scanning
suction side of the SD7003 airfoil were performed in a planes were orientated parallel to the main flow and
Göttingen type water tunnel with a cross-section of perpendicular to the airfoil surface and secondly, the
300 mm · 250 mm (Fig. 1) at Reynolds numbers based light sheets were located parallel to the airfoil surface. In
on the chord length (c = 200 mm) of Rec = 20,000– the first experiments, the flow was scanned in the span-
60,000 and angles of attack of a = 4–8. This type of wise direction starting from the midplane of the airfoil
airfoil was chosen since a long-stretched laminar sepa- model in five parallel planes with a spacing of
ration bubble occurs over a broad range of angles of Dz = 2.5 mm yielding a total scanning width of
attack at Reynolds numbers below 100,000 (Selig et al. Dscan = 10 mm. In the second measurements, three
1995). The airfoil with a transparent segment in the scanning planes were used with Dz = 1.5 mm and
spanwise center was mounted between the upper and Dscan = 4.5 mm. The illuminated two-dimensional area
lower wall of the tunnel to avoid disturbances of the is moving stepwise through the measurement volume,
wing tip vortices with the separation bubble (Bastedo
and Mueller 1986). Due to the relatively small thickness

Fig. 1 Sketch of the water tunnel with the model and the laser sheet Fig. 2 Photograph of laser sheet scanning system; rack for five
scanning system laser diodes
 321

starting with the first plane after each scanning-cycle. than the maximum bubble thickness dbub = O (4 mm).
The PIV post-processing was done with a cross-corre- Furthermore, the streamwise extent of the total separa-
lation of images of the same plane, i.e., one image is tion region is in the order of O(10d0). The size of the
correlated with the subsequent fifth and third image, separation bubble is evidenced by the mean velocity
respectively. Since the scanning time is proportional to distribution and the calculated streamlines in the sepa-
the number of desired planes, the maximum flow speed ration region (Fig. 3). The locations, where separation
has to be adjusted such that the particles remain within starts and reattachment occurs, are cross-checked by
the interrogation windows in the correlated images that investigating the particle movements in the original
are images of the same plane. In other words, this limits particle images. The lower part of Fig. 3 displays the
either the number of planes or the flow speed. In addi- contours of the normalized Reynolds shear stresses
tion, the limited depth of focus of the recording lens, above a critical value of 0.001. This value is chosen
which restricts the total scanning width, has to be taken in the literature, e.g., by Ol et al. (2005), to locate the
into account. In our case, we chose a long focal lens onset of transition in the separated shear layer. In
(f = 105 mm) and relatively large polystyrol tracer our investigation the onset occurs approximately near
particles with a mean diameter of 20 lm. x/c = 0.37.
In general, the PIV data were post-processed using an Due to the shear layer roll-up in the outer region of
adaptive cross-correlation starting with 32 · 32 pixels the separation bubble, regions of concentrated vorticity
interrogation windows with window shifting and win- are formed, which interact with the downstream part
dow reduction, which leads to 16 · 16 pixel interroga- near the reattachment point. In addition, via the pres-
tion windows with 50% overlap, i.e., 128 · 64 vectors sure distribution this process has a feedback on the
per image. separation bubble itself leading to a complex mutual
The following table shows in detail the features for interaction, i.e., a temporal change of the length and
each measurement (Table 1). thickness of the separation bubble and an influence on
the vortex size and strength. Figure 4 displays a series of
vector fields over a time span of 2.2 s for u¥ = 10 cm/s
3 Results and a = 6. The roll-up process starts with a small
disturbance upstream in the separation bubble (Fig. 4 at
In this section, we focus on the total flow around the t0). At increasing time, the small disturbance grows and
airfoil such that the location of the separation bubble vorticity accumulates until a large vortex is formed,
can be detected in the first part. Next, the two-dimen- which is located approximately O(4d0) upstream of the
sional time-resolved PIV results of the flow in the reat- reattachment line, e.g., vortices no. 2 and no. 3 in Fig. 4
tachment region within the separation bubble are at t0 + Dt and t0 + 2Dt. This phenomenon of vortex
shown. Finally, the results of the SPIV measurements in formation and drift was also observed by Watmuff
the reattachment region are presented using several (1999) and calculated by Pauley et al. (1990) defined as
planes in the spanwise and wall-normal direction. ‘‘vortex shedding.’’ The vortex travels downstream at
In the following, exemplary results are shown for a approximately 60% of the freestream velocity u¥, which
freestream velocity of u¥ = 10 cm/s and an angle of is of the same size as the calculated drift speed of 65%
attack of a = 6. Under such conditions the separation given by Pauley et al. (1990). The vortex continues its
bubble is located between x/c = 0.25 and 0.55, where drift until it reaches a critical condition, which forces the
the surface of the suction side of the airfoil is approxi- vortex to burst. This bursting process is seen as an
mately planar. Although larger freestream velocities are abrupt strong vertical fluid ejection into the mean flow,
possible the study focuses on the lower freestream see vortex no. 1 and no. 3 at t0 + Dt and t0 + 5Dt.
velocity of u¥ = 10 cm/s to enable a larger number of Occasionally, some vortices pair like vortex no. 3 with
planes in the SPIV measurements. A global picture of no. 4 (t0 + 5Dt) and burst right after pairing. The
the mean flow over the airfoil is given in Fig. 3, which is bursting is a consequence of a sometimes only slight
based on the average of 1,000 images in three slightly decrease of the downstream speed of the vortex, the
overlapping recording regions. simultaneous accumulation of vortex strength, and the
The boundary layer thickness near the separation is growing size of the vortex until a critical condition is
in the order of d0 = O(6 mm), which is slightly larger reached, which is assumed to be defined, among other

Table 1 Technical details of the

PIV-measurements Measurement In-plane Measurement Vectors Particle-image Particle-image
case resolution volume per diameter density per
(mm/pixel) (mm3) plot 32 · 32 pixels

Global view 0.075 – 23,000 2–3 pixels 11

Detail view 0.033 – 8,192 3–4 pixels 8
Spanwise scan 0.033 33.8 · 16.9 · 10.0 5 · 8,192 3–4 pixels 8
Wall-normal scan 0.185 189.4 · 94.7 · 4.5 3 · 8,192 1–3 pixels 20
322

Fig. 3 Contours of mean

streamwise velocity with
streamlines (top) and Reynolds
shear stresses -u¢v¢/u2¥ (bottom)
for u¥ = 10 cm/s and a = 6;
the separation and
reattachment points are derived
from the near-wall streamlines

criteria, by the ratio of the tangential and normal x/c = 0.45, 0.47, 0.5 and 0.52. The data are taken from
momentum flux. The burst itself looks as if the vortex the time-resolved PIV results. The red color indicates a
packet as a whole turns around the reattachment point relatively large v-component away from the surface, that
in a tumble-like motion. A vortex loop with a detach- is an ejection and the blue color indicates motion to-
ment of the vortex from the airfoil surface as observed wards the surface wall, called sweep. The structures in
by Watmuff (1999) was not found. the strip charts correspond to the vertical fluid motion of
A visualization of the dynamic behavior in this region the vortex passing the corresponding locations A–D.
is shown in Fig. 5 by means of strip charts, which dis- They are to a certain extent related to the vector plots in
play the spatio-temporal reconstruction of the magni- Fig. 4. It turned out that the velocity disturbances
tude of the vertical velocity component along a extend far into the outer flow region. The temporal
line normal to the surface at the streamwise locations reconstruction shows a number of four vortices starting

Fig. 4 Evolution process from

t0 to t0 + 5Dt of finally
bursting large-scale vortices;
time interval between pictures
Dt200 ms. Vortices are
indicated by streamlines.
Vectors are drawn with uniform
length except in (a) where
proportional vector length is
used
 323

sometimes abruptly. The latter one is called a burst,

when the flow is ejected into the outer field. Such a
bursting event can be clearly seen in Fig. 5, position C.
When you compare positions B and C in Fig. 5 for the
vortices no. 1 and no. 4 the rapid change from yellow
into deep red evidences the burst phenomenon.
A three-dimensional impression of the vortical
structures is obtained using the SPIV technique within a
small area of the separation region. The planes are ori-
entated to reconstruct the spanwise structure. Figure 6
displays vector plots of the flow structure in five suc-
cessive planes at two times of the vortex drift. The time
distance between the first and the last image in these
vector plots is about 9.9 ms which leads to a scanning
speed of 1.84 m/s. This speed suffices for analyzing
vortices with a drift speed of 0.6 u¥, since in this study
the freestream velocity is u¥ = 0.1 m/s. The results
show a definite variation of the vortex structure in the
spanwise direction. While in scan no. 1 the rear planes
(4, 5) show a large circular vortex, there is no evident
circular vortex pattern in the front planes (1, 2). Rather
only a vertical motion is visible. In addition, the region
of the vertical motion shifts from the front plane to the
rear plane, where it is part of the circular vortex, which
is, however, located further upstream. Scan no. 2 shows
a circular vortex in the typical bursting stage with a
pronounced vertical motion in the rear planes (4, 5). In
the front plane (1) a region of stronger backflow can be
seen, which does not seem to be related to the vortex
structure in the rear planes, i.e., it belongs to another
vortical structure. Note that this strong variation of the
flow structure takes place in the range of O(2d0) in the
spanwise direction, i.e., over a distance of approximately
10 mm. As can be seen in the scanning results the
bursting vortex can be considered a secondary part of
the separation bubble. In fact, another rotating structure
can be observed upstream of the bursting vortex,
Fig. 5 Strip charts of the spatio-temporal reconstruction of the
magnitude of the vertical velocity component
defining the first or inner separation region. It can be
conjectured that the drifting vortices form a convex or
c-like structure in the streamwise and spanwise directions
with an almost regular pattern over the whole recording and therefore, the front scanning planes in scan no. 1 do
period of 2.2 s (Fig. 5, position A). From the number of not cut the vortex core, such that only the roll-up part of
images from vortex no. 1 through vortex no. 4 and the the vortex can be seen. Regarding the backflow region in
recording frequency of 461 Hz the vortex roll-up fre- scan no. 2 the spanwise extension of the convex vortices
quency can be obtained, which is approximately 2.6 Hz seems to be limited due to other vortices in their vicinity.
for u¥ = 10 cm/s and a = 6. The vortex roll-up fre- Due to the limited depth of focus of the SPIV the
quency denotes the temporal formation of vortices at the total scanning width is limited. Therefore, additional
end of the separation bubble. The vortices possess a SPIV measurements with horizontal planes parallel to
downstream velocity which is equal to roughly 60% of the airfoil surface were taken to visualize the spanwise
the freestream velocity. When they propagate from the extension of these vortical structures. In Fig. 7 the
upstream location of the separation bubble the intensity lowest scan plane, which is located 0.25 mm above the
of vertical motion increases as evidenced in Fig. 5 by the surface, is shown. This image, being a raw vector plot,
change of the color from bright to deep red and bright to gives a good impression of the vortical structures and
deep blue, respectively. Most of the vortices keep their their spanwise extent. It displays the velocity field, which
downstream speed (no. 2 and no. 4), whereas some is not additionally smoothed to not falsify the strong
vortices decelerate, leading to vortex-pairing as can be gradients that can also be detected in the planes per-
seen for vortex no. 3 and no. 4. This phenomenon of loss pendicular to the surface. Pixel locking did not occur in
of downstream speed needs detailed investigation. The any of our measurements. To better illustrate the flow
exchange with the outer flow is either continuously or the color code from blue to red is used to emphasize the
324

Fig. 6 Vortical structures in

five subsequent scanning planes
for two scans with indicated
vortex centerline; vectors are
drawn with uniform length

upstream (blue) and downstream (red) part of the flow to the vortex circulation. There are in general two sub-
pattern. Furthermore, it can be stated that the upstream sequent blue regions in the downstream direction, indi-
orientated vectors form the footprint of the clockwise cating two subsequent vortex structures. As can be
rotating vortices. As already conjectured from the SPIV clearly seen, there is no regular pattern of the vortex
measurements with planes perpendicular to the surface, ‘‘half-moons’’ in the spanwise direction. These vortical
the vortical structures seem to possess a convex or c-like structures seem to interact with and disturb each other.
structure when considered in plan view. This structure is The small bright blue vectors in the left part of Fig. 7
indicated with dotted blue lines in Fig. 7. The spanwise indicate the footprint of the main recirculation area.
extent of the vortex ‘‘half-moons’’ is in the range of 3–7 These patterns of curved velocity distributions described
d0. Further downstream of the vortex structures a region above were also found by Hain and Kähler (2005) and
of relatively high downstream velocity can be observed. are interpreted as vortex-’’rollers’’ drifting downstream.
This can be interpreted as a sweep process where fluid Lang et al. (2004) showed phase-locked PIV results in a
from the main flow is transported towards the wall due measurement plane parallel to the surface of a flat plate.
The existence of fully counter-rotating vortices observed
in their PIV measurements could not be approved.
In Fig. 8 three subsequent planes parallel to the
surface are displayed. The upper planes are only partly
shown to better illustrate the overall flow pattern.
Backflow regions are dominant in the lowest scan plane,
while the mean flow direction prevails in the uppermost
scan plane. The small regions of backward or low speed
flow in the upper planes indicate overlapping or pairing
vortices a part of which, e.g., the vortex tips, drifts away
from the wall.

Fig. 7 Footprint of vortical structures in the lowest scan plane Fig. 8 Scanning planes parallel to the airfoil surface; upper planes
close to the airfoil surface are cut for better visibility of vortical structures
 325

Fig. 9 Sketch of vortex

footprint and curved vortical
structures in the reattachment
region

Based on these results a sketch of the vortical struc- mean flow. This behavior is similar to the ejection events
tures is drawn that appear in the reattachment zone. As in turbulent boundary layer flow. The developed ejec-
can be seen in Fig. 9 this region can be divided into two tion/sweep model gives a rough impression of the vortex
major parts: the main recirculation area and the sub- structure and the resulting fluid motion near the vortex
sequent vortex zone with two vortex regions in the core and the wall. In this model backflow regions with a
downstream direction. The vortices in these regions convex pattern represent the footprint of the c-like
show a convex or c-like structure. Downstream of these vortex near the airfoil surface. The mechanism to gen-
structures mass from the outer flow impinges upon the erate these convex structures is assumed to be linked
wall. The second vortex region is the area where vortex with a fragmentation process of the initial vortex roll
bursting can happen due to a critical size and strength of due to vorticity differences in the spanwise direction. As
the vortex. The reattachment line is located on the lee- observed in our results, the vortex generated within the
ward side of this second vortex region. separation bubble on the one hand increases in size and
strength and on the other hand travels downstream and
simultaneously reduces its downstream drift speed. The
4 Discussion initial drift speed was found to be 60% of the freestream
velocity u¥, which is in satisfactory agreement compared
The aforementioned results showed the generating and to 65% determined by Pauley el al. (1990). However, the
bursting processes of vortices within the locally sepa- low speed state is unstable and develops a critical con-
rated flow on the suction side of a SD7003 airfoil. The dition, which is assumed to be dominated by the
SPIV measurement technique was used to resolve the momentum ratio, determined by the accumulation of
dynamic vortex behavior and to analyze the spatial vortex strength. The vortex rotates as a whole structure
extension of the vortex structures within and near the around the reattachment point in the downstream
reattachment of the separation bubble. The vortices are direction. This process leads to a strong ejection of low
formed within the separation bubble as a consequence of speed fluid into the main flow and acts as a local flow
the shear layer roll-up due to Kelvin–Helmholtz insta- disturbance. This effect is similar to a flow over a
bilities and peel away from the main recirculation re- hemispherical bump on the surface yielding the defor-
gion, agreeing with the results of Pauley et al. (1990) and mation of the vorticity line and influencing the upstream
Watmuff (1999). These vortices form rather three- roll-up of vorticity. This phenomenon of vortex bursting
dimensional structures so called half-moons than two- is not assumed to be the same as the formation of a
dimensional spanwise rolls. The reconstructed spanwise vortex loop as suggested by Watmuff (1999). Counter-
shape evidences convex, i.e., c-like vortices with the rotating vortex pairs as observed by Lang et al. (2004)
lateral arms pointing in the downstream direction. The were not found in our measurements. This leads to the
SPIV measurements confirm that the extension in the assumption that the curvature of the airfoil surface has a
spanwise direction can be as small as 10% chord, i.e., distinct influence on the development of vortical struc-
3d0. In addition, there is no regular pattern of the tures compared to those emerging in a separation bubble
arrangement of these vortices in the spanwise direction, due to an adverse pressure gradient on a flat plate. In
as has also been shown by Hain and Kähler (2005). conclusion, the dynamics of the local separation bubble
Sometimes, the vortices start as large structures and later shows a complex interaction of different phenomena.
decompose into smaller units. One distinct feature of The sequence of events seems to start with a first insta-
these vortices is their tendency to burst abruptly causing bility related to the vortex roll-up in the laminar sepa-
a strong vertical fluid motion from the wall into the ration region generating a convex pattern, a growth of
326

the vortex strength and size until the burst of the vortex Maucher U, Rist U, Wagner S (1994) Direct numerical simulations
with feedback on the upstream flow field occurs. of airfoil separation bubbles. In: Computational fluid dynamics,
2nd ECCOMAS conference, 5–8 September 1994, Wiley,
Stuttgart, Germany, pp. 471–477
Acknowledgment This work was funded by the Deutsche Fors- McMichael JM, Francis MS (1997) Micro air vehicles—toward a
chungsgemeinschaft within the special focus research program new dimension in flight. URL: http://www.darpa.mil/tto/MAV/
‘‘SPP 1147 Bildgebende Messverfahren für die Strömungsanalyse’’ mav_auvsi.html, August 1997
(Experimental Visualization Methods for Flow Analysis) under the Ol MV, Hanff E, McAuliffe B, Scholz U, Kaehler C (2005) Com-
contract SCHR 309/25-1. parison of laminar separation bubble measurements on a low
Reynolds number airfoil in three facilities. In: AIAA Paper
2005–5149, 35th AIAA fluid dynamics conference and exhibit,
References Toronto, Ontario, 6–9 June 2005
O’Meara MM, Mueller TJ (1987) Laminar separation bubble
characteristics on an airfoil at low Reynolds numbers. AIAA J
Bastedo WG, Mueller TJ (1986) Spanwise variation of laminar 25(8):1033–1041
separation bubble on wings at low Reynolds numbers. J Airc Pauley LL, Moin P, Reynolds WC (1990) The structure of two-
23(9):687–694 dimensional separation. J Fluid Mech 220:397–411
Brücker Ch (1997) 3-D scanning PIV applied to an air flow in a Raffel M, Willert C, Kompenhans J (1998) Particle image veloci-
motored engine using digital high-speed video. Meas Sci metry: a practical guide. Experimental fluid mechanics.
Technol 8:1480–1492 Springer, Berlin Heidelberg New York
Gaster M (1966) The structure and behaviour of laminar separa- Rist U (2002) On instabilities and transition in laminar separation
tion bubbles. AGARD CP-4, flow separation, Part 2, pp. 813– bubbles. In: Proceedings of CEAS aerospace aerodynamics re-
854 search conference, Cambridge, UK, 10–12 June 2002
Hain R, Kähler CJ (2005) Advanced evaluation of time-resolved Selig MS, Guglielmo JJ, Broeren AP, Giguère P (1995) Summary of
PIV image sequences. In: 6th international symposium on low-speed airfoil data, vol. 1. SoarTech Publications, Virginia
particle image velocimetry, Pasadena, California, 21–23 Sep- Beach, 292 pp
tember 2005 Watmuff JH (1999) Evolution of a wave packet into vortex loops in
Lang M, Rist U, Wagner S (2004) Investigations on disturbance a laminar separation bubble. J Fluid Mech 397:119–170
amplification in a laminar separation bubble by means of LDA
and PIV. Exp Fluids 36(1):43–52