Professional Documents
Culture Documents
Editors
W. Schröder/Aachen
K. Fujii/Kanagawa
W. Haase/München
E.H. Hirschel/München
B. van Leer/Ann Arbor
M.A. Leschziner/London
M. Pandolfi/Torino
J. Periaux/Paris
A. Rizzi/Stockholm
B. Roux/Marseille
Y. Shokin/Novosibirsk
Hermann Schlichting –
100 Years
Scientific Colloquium
Celebrating the Anniversary of His Birthday,
Braunschweig, Germany 2007
Rolf Radespiel
Cord-Christian Rossow
Benjamin Winfried Brinkmann
(Editors)
ABC
http://www.gereby.com [ wappen schwarz linksbündig ]
Prof. Dr.-Ing. Rolf Radespiel Dipl.-Ing. Benjamin Winfried Brinkmann
Institut für Strömungsmechanik Institut für Strömungsmechanik
Technische Universität Braunschweig Technische Universität Braunschweig
Bienroder Weg 3 Bienroder Weg 3
38106 Braunschweig 38106 Braunschweig
Germany Germany
E-mail: benjamin.brinkmann@
Prof. Dr.-Ing. habil. Cord-Christian Rossow tu-braunschweig.de
Institut für
Aerodynamik und Strömungstechnik
Deutsches Zentrum für
Luft- und Raumfahrt
Lilienthalplatz 7
38108 Braunschweig
Germany
DOI 10.1007/978-3-540-95998-4
c 2009 Springer-Verlag Berlin Heidelberg
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or
parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in
its current version, and permission for use must always be obtained from Springer. Violations are liable for
prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
Typeset & Cover Design: Scientific Publishing Services Pvt. Ltd., Chennai, India.
Braunschweig R. Radespiel
October 2008 C.-C. Rossow
Table of Contents
D. Hummel
Hermann Schlichting was born on September 22, 1907 in Balje (District Stade, North-
ern Lower Saxony, Germany). After his years at High School, he studied from 1926
to 1930 Mathematics, Physics and Applied Mechanics at the Universities of Jena,
Wien and Göttingen. In 1930 he wrote his thesis “Über das ebene Windschattenprob-
lem” and he got his PhD under Prof. Ludwig Prandtl at the University Göttingen, and
in the same year he passed the government examination for teachers in Mathematics
and Physics.
In 1931 Dr. Hermann Schlichting joined the Kaiser-Wilhelm-Institut für
Strömungsforschung in Göttingen and worked under Prof. Ludwig Prandtl mainly on
viscous flows and boundary layer theory. In 1935 he entered the Dornier Aircraft
Company in Friedrichshafen, and the main topic of his work became now the
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 1–2.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
2 D. Hummel
aerodynamics of airplanes. He planned the Dornier Low Speed Wind Tunnel, which
came into operation after a short construction period.
At the age of 30 years, Dr. Hermann Schlichting received 1937 the call of Tech-
nische Hochschule Braunschweig to become full professor for Flight Mechanics and
director of the Institute of Aerodynamics. At that time a new Centre for Education in
Aeronautics “Luftfahrtlehrzentrum” was under construction at the airport Braun-
schweig-Waggum, and the 1.2 m Low Speed Wind Tunnel of the Institute of Aerody-
namics has been built under its new director. During the Second World War Prof.
Schlichting worked on basic problems in airplane aerodynamics such as wing-
fuselage interference including swept wings, and he continued his former activities in
viscous flows and boundary layer theory.
After the destructions due to the Second World War Prof. Schlichting’s university
institute was renamed 1945 as Institute of Fluid Mechanics. Since research and educa-
tion in aerodynamics of airplanes was forbidden, Prof. Schlichting worked in new
fields such as aerodynamics of road vehicles and in turbomachinery, but after some
time he was also back in airplane aerodynamics. In 1951 the first edition of his book
“Grenzschicht-Theorie” appeared in Germany.
In 1953 the Deutsche Forschungsanstalt für Luftfahrt (DFL) was reorganized at the
airport Braunschweig-Waggum, and in addition to his duties at the TH Braunschweig
Prof. Schlichting became the director of the DFL-Institute of Aerodynamics, nowa-
days DLR-Institut für Aerodynamik und Strömungstechnik. In 1956 he didn’t follow
a call by the TH Munich, but in 1957 he accepted a third duty as director of the Aero-
dynamische Versuchsanstalt (AVA) in Göttingen.
Prof. Schlichting was an excellent academic lecturer. His famous books “Gren-
zschicht-Theorie” (Boundary Layer Theory) and “Aerodynamik des Flugzeuges”
(Aerodynamics of the Airplane, together with E. Truckenbrodt) appeared in 11 Edi-
tions and were translated into 5 foreign languages. These two books enabled a huge
number of students and scientists worldwide to get insight in modern fluid mechanics.
Prof. Schlichting guided more than 100 candidates for a doctor’s degree (Dr.-Ing.)
and 10 academic lecturers (venia legendi) during the preparation of their theses. Apart
from the already mentioned books his own scientific work is documented in more
than 100 distinguished publications. (Details may be taken from Hummel, D.: Das
wissenschaftliche Lebenswerk von Hermann Schlichting. Z. Flugwiss. Weltraumfor-
schung 8, 1984, 1–17).
Prof. Schlichting received a large number of awards, and he was honoured by vari-
ous international institutions to deliver memorial lectures, and he used them to dis-
seminate scientific results produced in his institutes. He retired from all his duties in
1975, but his lectures at the university finally ended 1978. He continued to work on
the last Edition of his book “Boundary Layer Theory”.
He died on 15 June 1982 in Göttingen.
Hermann Schlichting and the Boundary-Layer Theory
K. Gersten
Ruhr-Universität Bochum
Institut für Thermo-und Fluiddynamik
D-44780 Bochum, Germany
Summary
After a short summary of the main features of boundary-layer theory Hermann
Schlichting’s contributions to boundary-layer theory are described. They refer to
almost all areas of boundary-layer theory. Particularly, the following topics are worth
mentioning: laminar-turbulent transition, boundary layers with continuous suction or
blowing, boundary layers in cascade flows of turbomachines, boundary layers on
rough surfaces, laminar airfoils, heat transfer, automobile aerodynamics. A detailed
discussion is devoted to Tollmien-Schlichting waves, to the conception of “equivalent
roughness” and to the classic and well-known book “Boundary-Layer Theory”, which
are and will always be connected with the name of Hermann Schlichting.
1 Boundary-Layer Theory
The boundary-layer theory has been developed by Ludwig Prandtl, who presented the
basic conception of this theory for the first time in the year 1904 at the International
Mathematical Congress in Heidelberg, [1]. Hence, this theory is in the meantime more
than hundred years old. Nevertheless, it is still of great up-to-date interest, [2].
The flow past bodies is governed by the no-slip condition at the body surface, i.e.
the fluid “sticks” to the body surface due to intermolecular forces.
Figure 1 shows the flow past a symmetric airfoil at zero incidence in a water chan-
nel. Aluminium particles have been sprinkled on the surface of the water to make the
streamlines visible. The length of each particle streak is proportional to the flow ve-
locity. It can be seen, that in the vicinity of the wall there is a thin layer of retarded
flow called boundary layer.
The thickness of this boundary layer depends on the following parameters: free
stream velocity V, body length L, density ρ and viscosity µ of the fluid. The viscosity
is a physical property of the fluid describing the friction (tangential) forces in the fluid
flow. The English physicist Osborn Reynolds, [3], found out that the flow is not af-
fected by these four parameters independently, but rather only by a combination of
them which is dimensionless and called Reynolds number
ρV L V L
Re = = , (1)
µ ν
where ν = µ / ρ is the kinematic viscosity. The Reynolds number is the most impor-
tant dimensionless parameter to characterize the viscosity effects in fluid flows. The
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 3–17.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
4 K. Gersten
Fig. 1. Symmetric airfoil at zero incidence in a water channel. Little aluminium particles sprin-
kled on the water surface make the streamlines visible. The length of each particle streak is
proportional to the flow velocity.
Reynolds number increases when the viscosity and hence the viscosity effects become
smaller. The higher the Reynolds number is, the thinner is the boundary layer.
The conception of Prandtl’s boundary layer, therefore, implies that flow fields at
high Reynolds numbers can be divided into two unequally large regions. In the bulk
of the flow field the viscosity can be neglected, and the flow corresponds to the invis-
cid limiting solution. This region is called the inviscid outer flow. The second region
is the very thin boundary layer at the wall where the viscosity must be taken into
account. For the thin boundary layer the very complex general flow equations could
be simplified by Prandtl in such a way that they could be solved in the years after
1904. The thickness of the boundary layer is growing downstream along the wall. The
boundary layer leaves the body at the trailing edge and forms the wake flow past the
body. It could happen that the boundary layer will separate before it reaches the trail-
ing edge, but such a situation is usually not desirable.
The most important characteristics of the boundary layer (incompressible and two-
dimensional) are:
1. The pressure distribution at the wall and in the boundary layer is “impressed” by
the inviscid outer flow. It may, therefore, be regarded as a known function as far
as the boundary-layer flow is concerned.
2. The boundary layer can be laminar or turbulent (see Figure 2).
3. The boundary layer behaves differently depending on whether the pressure is
decreasing (front part of the airfoil) or increasing (rear part of the airfoil). The
latter case is more important: the laminar boundary layer tends to transition into
turbulence, laminar as well as turbulent boundary layers show the tendency to
separation. However the turbulent boundary layer can resist the positive pressure
gradient much longer than the laminar one.
4. The boundary-layer development determines the aerodynamic design:
One important design conception for airfoils or wings and for cascade profiles in
turbomachines is to avoid boundary-layer separation over the body surface.
Since the laminar boundary layer leads to smaller friction forces than the turbu-
lent one, a laminar boundary layer all the way to the trailing edge of the airfoil
would be optimal. Unfortunately, the laminar boundary layer would separate at
very small pressure increase. Therefore, an appropriate airfoil geometry, i.e. an
Hermann Schlichting and the Boundary-Layer Theory 5
Fig. 2. The Reynolds dye experiment, [3]. Flow in water made visible by injection of a dye.
(a) Laminar flow: A straight coloured filament forms and moves with the flow. (b) Turbulent
flow: The coloured filament carries out highly irregular transverse motions leading to complete
scattering and mixing of the colour.
Already in the year 1882 Osborn Reynolds, [3], formulated the conjecture that the
laminar flow, always a possible solution of the equations of motion, becomes unstable
above a certain limit (i.e. the indifference Reynolds number) and changes into turbu-
lent flow. In the following decades Reynolds and other well-known scientists, like
Lord Rayleigh, Sommerfeld, Heisenberg, tried to develop a mathematical stability
theory, but at first without success. It was only at about 1930 that the original aim,
namely the theoretical calculation of indifference Reynolds numbers, was achieved
satisfactorily by Prandtl’s co-workers Walter Tollmien and Hermann Schlichting.
Their theoretical investigations assume that the laminar flow is acted on by some
small disturbances which could come from irregularities in the outer flow or from the
wall roughness. The theory follows the rate of change of such perturbations superim-
posed on the laminar boundary layer. The decisive question here is whether the dis-
turbances die away or grow in time. If the disturbances die away in time, the basic
flow is considered to be stable; if they grow, the basic flow is unstable and may even-
tually go over into the turbulent state. For the first time Tollmien and Schlichting
calculated the stability boundaries and the features of the perturbed flow. Since the
perturbations are waves, they are called Tollmien-Schlichting waves (in short: TS
waves).
Figure 3 is the fundamental diagram of the stability theory according to Tollmien,
[39]. It shows the curve of neutral stability (indifference) for the boundary layer on a
Hermann Schlichting and the Boundary-Layer Theory 7
flat plate at zero incidence. Each point in the diagram represents a particular TS wave.
Its velocity component parallel to the wall is given by:
u' ( x, y , t ) = e β i t sin ⎡α x − β r t − -
ϑ ( y )⎤ , (2)
⎢⎣ ⎥⎦
where β i is the amplification factor, β r the frequency and -(y) the amplitude distribu-
tion. Each TS wave is fixed by the three parameters α , β i , β r . The parameter β i deter-
mines whether the wave is amplified and unstable ( β i >0) or damped and stable ( β i <0).
The area inside the curve refers to unstable waves, the one outside to stable waves.
Therefore, all waves at low Reynolds numbers are stable. Above the so-called indiffer-
ence Reynolds number (Re = 520) there are always some waves which are unstable.
Tollmien’s theory has been extended by Schlichting in the following respects:
(a) Unstable waves
Schlichting extended Tollmien’s diagram by determining also the curves β i = const.
for unstable waves, as it is shown in Figure 4, [7–9].
(b) Amplitude distribution
In order to obtain closer insight into the mechanism of the disturbances, Schlichting
determined the amplitude distribution - ( y ) for some neutral waves, [9]. With these
Fig. 3. Dependence of the curve of neutral stability on the Reynolds number for the boundary
layer on a flat plate at zero incidence, after W. Tollmien, [39]
α = 2πλ: wave number
λ: wavelength
δ1: displacement thickness of the boundary layer
U∞: outer flow velocity
υ: kinematic viscosity
8 K. Gersten
Fig. 4. Curves of constant βi for the unstable Tollmien-Schlichting waves for the boundary layer on
a flat plate, after H. Schlichting, [7]. The streamline pattern for the perturbed flow referring to point
I is given in Figure 5. The curve of neutral stability (βi = 0) differs slightly from the curve shown in
Figure 3, because in the meantime Schlichting had worked out more precise results.
it was possible to compute the streamline pattern of the perturbed flow for neutral
wave disturbances. An example of this is given in Figure 5. The streamlines and the
velocity distribution refer to the neutral wave indicated by I in Figure 3.
At around the same time, investigations of the plate boundary layer have been car-
ried out by Prandtl in Göttingen and these yielded at least a certain qualitative confir-
mation of the stability theory. Figure 6 shows the appearance of turbulence from an
initially long wave length perturbation. The similarity of this photo with Schlichting’s
theoretical streamline pattern of a neutral perturbation in Figure 5 is irrefutable.
The quantitative verification of the stability theory took more than a decade. It was
then provided by H.L. Dryden, [41], assisted by G.B. Schubauer and H.K. Skramstad,
[42], of the National Bureau of Standards in Washington, USA. Not until a new spe-
cial wind tunnel with extremely low turbulence intensity has been built for the inves-
tigations the stability theory could be convincingly verified by experiments. The
results, confidential during the war, have not been known to the public until 1947.
Fig. 5. Streamline pattern and velocity distributions for a neutral wave in the boundary layer on
a flat plate at zero incidence (perturbation I in Figure 4), after H. Schlichting, [13]
U(y): Basic flow velocity
U(y) + u'(x,y,t): Perturbed flow velocity
Re1 = U∞δ1υ = 893: Reynolds number
λ = 40 δ1: Perturbation wavelength
cr = 0.35 U∞: Wave propagation velocity
Hermann Schlichting and the Boundary-Layer Theory 9
Fig. 6. Flow along a flat plate in a water channel, appearance of turbulence originating from a
disturbance of long wavelength, after L. Prandtl, [40]. The camera is moving with the flow so
that the same group of vortices always remains in the picture. The flow is made visible by
aluminium dust scattered on the water surface.
The indifference point, where the laminar boundary layer becomes unstable, cannot
be the location, where already turbulence appears. There is a certain length of the
transition zone necessary for the development of the flow from the indifference point
all the way to the point of completed transition. Schlichting found for the first time
that this transition length depends essentially on the magnitude of amplification of the
unstable TS waves, [7]. When a particular TS wave with fixed α , β i , β r is considered
in its development in the boundary layer downstream, it is equivalent to moving along
a straight line in the stability diagram (Figure 4, dotted line) due to the increase of the
boundary-layer thickness downstream. Hence, the wave becomes unstable only tem-
porarily, it leaves the unstable area further downstream and becomes stable again.
10 K. Gersten
Fig. 7. Indifference curves of Tollmien-Schlichting waves for laminar boundary layers with
pressure gradient, after H.Schlichting and A. Ulrich, [11]
δ dp
Λ=− Λ > 0 : Pressure decrease downstream Λ < 0 : Pressure increase
µ U dx
e∫
β1 dt (3)
= e 9 = 8103.
This discovery was confirmed at about the same time by I.L. van Ingen, [44], and
later by N.A. Jaffe et al., [45]. By using Eq. (3) it is possible to determine the location
of the point of completed transition. This method based on calculating amplification
rates of unstable TS waves is called e9-method. It is still today a common method in
engineering practice to determine points of completed transition, see D. Arnal, [46].
The problems of laminar-turbulent transition and the origin of turbulence are after
more than hundred years of scientific research still not fully understood today, see
Hermann Schlichting and the Boundary-Layer Theory 11
Fig. 8. Positions of indifference points on a airfoil, after H. Schlichting, [13]. St: Stagnation
point and S: Laminar separation point.
surveys on the state of the art, [2, 47–49]. The sketch of the transition region in
Figure 9 shows how complex the flow is even for the simple flat-plate boundary layer.
According to this sketch the boundary-layer flow goes through the following stages,
starting with the leading edge:
1. Stable laminar flow
2. Unstable Tollmien-Schlichting waves (primary stability theory)
3. Three-dimensional waves and vortex formation (secondary stability theory)
4. Vortex decay
5. Turbulent spots
6. Fully turbulent flow
The complex process of transition is almost always beginning with the TS waves. The excep-
tion is the so-called bypass transition, where the stage of TS waves is skipped because of high
intensity of turbulence in the outer flow. These cases can be found in turbomachines.
Fig. 9. Idealized sketch of the transition zone in the boundary layer on a flat plate at zero
incidence, see [50]
12 K. Gersten
Fig. 10. Resistance diagram for turbulent flow past rough flat plate at zero incidence, after L.
Prandtl and H. Schlichting, [23]
c D = 2 D / ρ U ∞2 L B : drag coefficient
D: drag of plate (one side)
L: length of plate
B: span of plate
kS: sand grain roughness
- - - - - border-line between flows dependent on roughness
and viscosity (left) and fully rough flows (right)
Hermann Schlichting and the Boundary-Layer Theory 13
the turbulent flow past flat plates at zero incidence in Figure 10. At high Reynolds
numbers the drag coefficient becomes constant, i.e. independent of viscosity. With
increasing Reynolds number the thickness of the boundary layer and hence the thick-
ness of the sublayer would decrease, whereby the latter eventually disappears among
the roughness elements. This region is called fully rough.
Schlichting has investigated a great variety of rough surfaces and arranged these
roughnesses on a scale of standard roughness. He adopted Nikuradse’s sand rough-
ness as such a standard roughness and defined the equivalent sand roughness. Any
arbitrary roughness can be related to an equivalent sand roughness. By definition a
flat plate with this equivalent sand roughness has the same friction drag in the fully
rough region as the plate with the given roughness. Only one single experiment is
sufficient to determine the equivalent sand roughness.
The equivalent sand roughness is of particular interest for technical or natural
roughness, where the geometric height is statistically random and one parameter is
sufficient to characterize the roughness effect on the flow. For an unknown technical
roughness it seems to be sufficient to determine the arithmetic mean roughness Ra by
using mechanical profilometers or Laser interferometry, because experiments led to
an approximate proportionality between the technical roughness and the geometric
roughness: k tech = 3.5 Ra , see K. Gersten, [52].
An example shall illustrate the conception of equivalent sand roughness. The photo
in Figure 11 has been taken on February 6, 1958 at the airport München-Riem. In the
airplane the English soccer team Manchester United was travelling. This team had
won an important match in Belgrade and was on its way back to London with an
intermediate stop in Munich. A few minutes after this photo was taken the take-off of
the airplane ended in a terrible disaster. Seven soccer player lost their life. One of the
greatest soccer player and coach in England survived: Bobby Charlton, much later
ennobled by the Queen: Sir Robert Charlton. What happened to the airplane? The
airplane did not lift off at the end of the runway although it had reached the required
lift-off speed. What was the explanation for this failure? Snow had fallen during the
whole intermediate stop, and the temperature was about the freezing point.
Fig. 11. Photo of the airplane carrying the soccer team Manchester United on the runway of the
airport München-Riem just before take-off on February 6, 1958
14 K. Gersten
Table 1 shows an overview of the unique story of this classic and well-known book.
Since its first edition (in German) in 1951 it has been revised and brought up to the
newest state regularly. In particular, the English translation led quickly to a world-
wide dissemination of the book. Once, Professor John Weske (USA) called Schlicht-
ing’s book the “Bible of the aerodynamicists”.
Hermann Schlichting and the Boundary-Layer Theory 15
The last edition brought out by Schlichting was the eighth German edition. It came
out in 1982, the year when Schlichting died. Schlichting has made an all-out effort,
supported by his colleague and friend Friedrich Riegels, to bring this edition to com-
pletion. Shortly later both scientists died, their days of death only about a week apart.
Professor Jürgen Zierep, who was at that time scientific supervisor of the editor
G. Braun in Karlsruhe, suggested that I should take care of Schliching’s book after his
death. I was delighted and I do it to express my thanks and tribute of admiration to my
great mentor and scientist Hermann Schlichting.
References
[1] Prandtl, L.: Über Flüssigkeitsbewegungen bei sehr kleiner Reibung. In: Verhandlg. III.
Internat. Math. Kongr. Heidelberg, pp. 484–491 (1904)
[2] Meier, G.E.A., Sreenivasan, K.R., Heinemann, H.-J. (eds.): IUTAM Symposium on One
Hundred Years of Boundary Layer Research. Springer, Heidelberg (2006)
[3] Reynolds, O.: An experimental investigation of the circumstances which determine
whether the motion of water shall be direct or sinious, and of the law of resistance in par-
allel channels. Phil. Trans. Roy. Soc. London A 174, 935–982 (1883)
[4] Über das ebene Windschattenproblem. Dissertation Universität Göttingen, Ing. Archiv 1,
533–571 (1930)
[5] Entstehung der Turbulenz. In: Flügge, S. (eds.) Handbuch der Physik, Band VIII/1:
Strömungsmechanik I, pp. 351–450. Springer, Heidelberg (1959)
[6] Über die Theorie der Turbulenzentstehung. Zusammenfassender Bericht. Forschg. Ing.
Wesen 16, 65–78 (1950)
[7] Zur Entstehung der Turbulenz bei der Plattenströmung. Nachr. Ges. Wiss., Math. -Phys.
Klasse, 181–208 (1933)
[8] Berechnung der Anfachung kleiner Störungen bei der Plattenströmung. Z. angew. Math.
Mech. 13, 171–182 (1933)
16 K. Gersten
[9] Amplitudenverteilung und Energiebilanz der kleinen Störungen bei der Plattenströmung.
Nachr. Ges. Wiss. Göttingen, Math. -Phys. Klasse, 47–78 (1935)
[10] Über die theoretische Berechnung der kritischen Reynoldsschen Zahl einer Reibungss-
chicht in beschleunigter und verzögerter Strömung. Jahrbuch, d. dtsch. Luftfahrtforsch., I,
97–112 (1940)
[11] Zur Berechnung des Umschlages laminar/turbulent (gemeinsam mit A. Ulrich), Lilien-
thal-Gesellschaft f. Luftfahrtforsch. Bericht S 10, 75–135 (1940)
[12] Über die Berechnung der Umschlagstelle laminar/turbulent“ (gemeinsam mit A. Ulrich),
Lilienthal-Gesellschaft f. Luftfahrtforsch., Bericht Nr. 141, 17–26 (1941)
[13] Zur Berechnung des Umschlages laminar/turbulent (gemeinsam mit A.Ulrich), Jahrbuch,
d. dtsch. Luftfahrtforsch. I, 8–35 (1942)
[14] Berechnung der laminaren Grenzschicht mit Absaugung. Lilienthal-Gesellschaft f. Luft-
fahrtforsch., Bericht Nr. 141, 14–16 (1941)
[15] Die Grenzschicht mit Absaugung und Ausblasen. Luftfahrtforsch. 19, 179–181 (1942)
[16] Die Grenzschicht an der ebenen Platte mit Absaugung und Ausblasen. Luftfahrtforsch,
19, 293–301 (1942)
[17] Exakte Lösungen für die laminare Grenzschicht mit Absaugung und Ausblasen (gemein-
sam mit K. Bußmann), Schriften der Deutschen Akademie der Luftfahrtforsch. Bd. 7 B,
Heft 2, 25–69 (1943)
[18] Die Beeinflussung der Grenzschicht durch Absaugung und Ausblasen. Jahrbuch 1943/44
der Deutschen Akademie der Luftfahrtforsch., 90–108 (1943-1944)
[19] Ein Näherungsverfahren zur Berechnung der laminaren Reibungsschicht mit Absaugung.
Ing. Archiv 16, 201–220 (1948); English translation: NACA TM 1216 (1949)
[20] L’aspiration de la couche limite en technique aéronautique. Technique et Science Aéro-
nautiques 6, 149–161 (1956); Deutsche Version: Jahrbuch 1956 d. WGL, 19–28 (1957)
[21] Einige neuere Ergebnisse über Grenzschichtbeeinflussung. Advances in Aeronautical
Sciences (1958), vol. II, pp. 563–586. Pergamon Press, London (1959)
[22] Experimentelle Untersuchungen zum Rauhigkeitsproblem. Ing. Archiv 7, 1–34 (1936)
[23] Prandtl, L., Schlichting, H.W., Reederei: Das Widerstandsgesetz rauher Platten, pp. 1–4
(1934)
[24] Einfluß der Turbulenz und der Reynoldsschen Zahl auf die Tragflügeleigenschaften.
Ringbuch der Luftfahrttechnik. Bd. I, Berlin (1935)
[25] Messungen an Laminarprofilen (gemeinsam mit Bußmann, K.). Lilienthal-Gesellschaft f.
Luftfahrtforsch., Bericht Nr. 149, 17–20 (1942)
[26] Über die theoretische Berechnung der Strömungsverluste eines ebenen Schaufelgitters
(gemeinsam mit Scholz, N.). Ing. Archiv 19, 42–65 (1951)
[27] Ergebnisse und Probleme von Gitteruntersuchungen. Z. Flugwiss. 1, 109–122, auch VDI-
Bericht 3, 7–21 (1954); English version: Journ. Aeron. Sci. 21, 163–178 (1954)
[28] Anwendung der Grenzschichttheorie auf Strömungsprobleme der Turbomaschinen. Sie-
mens-Zeitschrift 33, 429–438 (1959); English version: Basic Eng. 81, 543–551 (1959)
[29] A survey of some research investigations on boundary layers and heat transfer. J. Appl.
Mech. 38, 289–300 (1971)
[30] Einige exakte Lösungen für die Temperaturverteilung in einer laminaren Strömung. Z.
angew. Math. Mech. 31, 78–83 (1951)
[31] Der Wärmeübergang an einer längsangeströmten ebenen Platte mit veränderlicher Wand-
temperatur. Forschg. Ing. Wesen 17, 1–8 (1951)
[32] Das wissenschaftliche Lebenswerk von Hermann Schlichting. Hummel, D.Z. Flugwiss.
Weltraumforsch. 8, 1–17 (1984)
Hermann Schlichting and the Boundary-Layer Theory 17
H. Körner*
Summary
This paper gives a survey of the work of Hermann Schlichting in the field of airplane-
aerodynamics. It includes his scientific work as well as his commitment in building
up test facilities. It furthermore describes his role in the reconstruction of aerodynam-
ics research after the war, when he became one of the key-persons. Under his leader-
ship the reconstruction of windtunnels and the establishment of new priority fields in
research took place. Subsequently, he introduced this new orientation of German
aerodynamics research to the international scientific community.
1 An Approach
Being a student of mechanical engineering at the „Technische Hochschule“ (TH)
Braunschweig you first saw Hermann Schlichting in his lecture “Hydromechanics”.
This happened to me in 1960. Figure 1 shows him in this lecture a dozen years earlier.
*
Direktor i. R.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 18–38.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 19
Fig. 3. Scheme of the circuit of the 1.2 m wind tunnel of the Institute of Aerodynamics at the
TH Braunschweig [1]
This 1.2 m windtunnel played an important role for his airplane research in the
years that followed. His main research topics using this facility have been:
• Interference of wing, fuselage and empennage;
• High lift behaviour of swept wings;
• Laminar airfoils.
Interference [4–7]: This was a field which had not been dealt with in a systematic way
up to that time. Schlichting and his collaborators did investigations at low speed in
symmetric and asymmetric (yaw) conditions. He used sets of wings, fuselages and
empennages which he could combine in different ways. A partial survey of the mod-
els used gives Figure 4, where all the wings are depicted. These wings could be
Fig. 5. Roll-derivative due to yaw – Influence of the vertical wing position [7]
22 H. Körner
wings. The spanwise distribution of circulation shows nicely how the local Clmax shifts
outboard with growing sweep angle. In the centre section the lift diminishes. This
behaviour is typical for swept wings and has an important influence on the high-lift
behaviour and the pitching-moment of these wings. At that time, there were no theo-
retical methods available to predict this behaviour, therefore this finding was of
utmost importance.
Laminar airfoils [10–15]: This is a topic that has a high actuality today again. The
pending climate changes force aircraft designers to reduce the fuel consumption of
future vehicles in a drastic way. The laminar wing is one of the most important means
to achieve this. Based on his stability investigations of laminar flow [10] Schlichting
in 1940 published a paper entitled „Zur Berechnung des Umschlages laminar / turbu-
lent“ [11, 12]. This work was opening new perspectives concerning the design of low-
drag airfoils. Figure 8 shows a result of his investigations, for which he was awarded
with the Lilienthal-prize of the Lilienthal-Gesellschaft.
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 23
Fig. 7. Spanwise normalized circulation distribution of wings with different sweep [7]
Schlichting’s work was a first important step in the design of laminar airfoils using
theoretical methods. Nowadays sailplanes and a number of business-aircraft fly with
laminar wings. For commercial airplanes this is still an important challenge for the
future.
Schlichting also did experimental investigations on laminar airfoils using his wind-
tunnel and the A 1 facility at LFA-Völkenrode (Figure 9) [13].
He furthermore investigated the effect of suction on the behaviour of laminar
boundary layers on airfoils and proved with his calculations the positive effect of
suction on the downstream shift of the transition (Figure 10).
His investigations showed the potential of suction through a porous surface
[14, 15]. Thus, he provided important contributions to “Laminar Flow Control”
(LFC).
24 H. Körner
Fig. 8. Influence of the position of maximum thickness on the transition location [11]
Fig. 10. Boundary layer on a symmetric Joukowsky airfoil (t/c = 0.15) at CL = 0 with homoge-
neous suction and different suction rates. [15]
swept wing and at that time head of the LFA Institute of Gasdynamics – was con-
nected to Schlichting’s institute as part-time lecturer on gasdynamics. Schlichting vice
versa gave a lecture series on boundary layer theory at LFA which was very well
received. The material he compiled, became the core of his famous textbook on
boundary layer theory.
Most of the scientific investigations, which have been conducted at Schlichting’s
institute, have been sponsored by the „Reichsluftfahrtministerium“. These investiga-
tions had basic character and no direct connection to war-airplanes or weapons. Spon-
sorship by the government on the other side allowed him and his colleagues the
mathematicians Iglisch and Rehbock and the mechanics specialist Schäfer to continue
26 H. Körner
work through the war period. Thus, these TH lecturers could continue their research
work and their lectures at TH Braunschweig until 1945 although the circumstances,
under which they worked, became very tough. The day before the US-army entered
Braunschweig, Erich Truckenbrodt had his final exam for his doctoral graduation
with Schlichting as supervisor.
After the collapse of Germany in 1945 the development of airplanes and aeronautical
research stopped totally and were not allowed anymore. Aeronautic research- and
development facilities were dismounted or destroyed. This also happened to the 1.2 m
wind- tunnel of the institute. The TH furthermore gave up the site at Waggum airport
and Schlichting returned with his institute – now named “Institute of Fluid Mechan-
ics” – into the location Bienroder Weg, the former site of the institute. Although the
research activities stopped, the Allied Forces were highly interested in the results of
German aeronautical research and gave order to leading German scientists to summa-
rize the results in monographs. H. Schlichting got the order to summarize the results
in the field aerodynamic interference of parts of the airplane [7]. The major part of the
contents of this monograph originated from his institute.
Since aeronautical research was not allowed Schlichting looked for other fields and
engaged himself in cascade-aerodynamics for turbines and compressors and in vehicle
aerodynamics.
Fortunately, the old 1.3 m windtunnel at Bienroder Weg which was built around
1930 still existed. The now starting investigations in vehicle aerodynamics were con-
ducted in this windtunnel [16]. A major activity was for the Volkswagen Company.
The VW beetle and the VW van have been tested with respect to performance as well
as stability. As a consequence of these investigations, the aerodynamic drag of the
VW van could be reduced from CD = 0.75 to 0.45.
Another example for the tests conducted at that time was the world-record motor-
bike of NSU. This motorbike (Figure 11) has been tested with respect to low drag and
acceptable lateral stability. In 1951 the NSU-test driver Wilhelm Herz achieved a new
speed world record for motorcycles. This had a big impact on the public and was seen
as an indication of growing upswing in Germany.
While after 1945 activities at the large aeronautical research establishments DVL,
DFL-LFA and AVA completely stopped, the university institutes were allowed to
continue with not yet completed projects. Thus in the years to follow a number of
doctoral thesises in the field of aeronautics were completed under the guidance of
Schlichting e.g. the thesises of Scholz, Mayer, Ksoll and Söhne. E. Truckenbrodt
developed his lifting surface theory and published it 1953.
With the formation of the „Bundesrepublik Deutschland“ it became obvious that
Germany would enter the aeronautics field again. This fostered intense efforts to es-
tablish just one research establishment for aeronautics as successor of the former
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 27
Fig. 11. Windtunnel models of motorbike and vehicle for speed record tests [16]
DVL, DFL-LFA and AVA. Hermann Blenk, who was one of the main actors in this
field, engaged also Schlichting. But these activities failed and the former research
establishments have been reactivated in 1953. In this process Schlichting was co-
initiator for the re-establishment of the DFL in Braunschweig which has been located
on the site of the „Luftfahrtlehrzentrum“ Waggum. Schlichting became director of the
Institut of Aerodynamics, one of the six institutes of the new DFL.
Subsequently Schlichting played a key-role in the construction of new windtunnels
[17]. Since all windtunnels had been dismounted or destroyed a new start from the
scratch was necessary. Schlichting became „Obmann“ of a group of personalities in
aerodynamics who cared for the planning of new experimental facilities. This group
produced a „Denkschrift über den Ausbau der deutschen Luftfahrtforschung, Fachge-
biet Aerodynamik und Gasdynamik“ (Memorandum on the reconstruction of German
aeronauctical research, section aerodynamics and gasdynamics), which proposed a 90
million DM1 program. The federal administration provided a reduced sum of 30 mil-
lion DM. Subsequently a number of windtunnels were built in Braunschweig, Göttin-
gen and Köln/Porz-Wahn [18, 19]. Most of these windtunnels are active still
1
45 million EUR.
28 H. Körner
today and have become very capable facilities. One of these is the AVA “Transonic
Windtunnel Göttingen” (Figure 12) with an installed power of 12 MW.
Under Schlichting’s supervision this windtunnel has been extended to a powerful
facility especially for investigations in basic and missile aerodynamics, covering a
Mach-range from 0.4 to 2.0. Another facility built within this program is the DFL
“Low Speed Windtunnel Braunschweig”, nowadays DNW-NWB (Figure 13). This
windtunnel was designed and built by Scholz, Hubert and Trienes in the 60ies and run
by R. Göthert, Schlichting’s first doctoral graduate. Nowadays this facility is one of
the best medium size facilities and has a high reputation.
In retrospect, one may ask: why didn’t they design and build larger facilities? Fa-
cilities as they existed before 1945? The answer is: the funds for the new windtunnels
were restricted and there was another argument which Schlichting expresses very
clear in one of his publications. What Germany needed most, was the education of a
new generation of aeronautical scientists and engineers. The old generation was
dispersed or dead. Only a few had survived and served the community.
The building up of a new generation had to be performed on a somewhat
broader basis and for this it would be preferable to have several medium size fa-
cilities at different places. Looking back from today this obviously was the right
choice. Thus, a stable aeronautics community has been built up again. As to the
large facilities, it has successfully proven, to construct them in European coopera-
tion several years later, see DNW and ETW. These windtunnels operate in a
successful and efficient way.
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 29
Fig. 13. DFL Low Speed Windtunnel Braunschweig, today DNW-NWB [19]
The restart of aircraft industry and aeronautical research in Germany in 1953 and
with it the restart of lectures at the THs unveiled the lack of a proper textbook in air-
plane aerodynamics. The last comprehensive survey had been given by Fuchs and
Hopf twenty years ago. Furthermore the knowledge in this field expanded rapidly in
these years. The „Springer-Verlag“ therefore asked Schlichting whether he would
engage himself to produce a state of the art textbook. Schlichting agreed and together
with E. Truckenbrodt he compiled the two-volume textbook „Aerodynamik des Flug-
zeuges“ [20]. This book has become a standard textbook of airplane-aerodynamics in
the years to follow (Figure 14).
In this book the authors first describe the basics of incompressible and compressi-
ble flow. This is followed by boundary-layer theory, Schlichting’s speciality. The
next chapters deal with airfoil- and wing-theory in subsonic and supersonic flow.
Finally they describe the aerodynamics of fuselage and empennage including interfer-
ence with the wing. This chapter is mainly based on Schlichting’s monograph on this
subject. There are two editions of this book in German and translations in English
[21], Japanese and Chinese.
30 H. Körner
Fig. 14. Front pages of the textbook „Aerodynamik des Flugzeuges“ [20]
1956 another position in research was offered to Schlichting. The board of direc-
tors of the „Aerodynamische Versuchsanstalt Göttingen“ (AVA) offered him the
position of the director. Schlichting accepted. With this he became successor of Al-
bert Betz and took over the position, which Ludwig Prandtl held over decades. This is
obviously the peak of his career. He now was at the same time
• full professor at TH Braunschweig and director of the TH-Institute of Fluid Me-
chanics
• director of the DFL Institute of Aerodynamics in Braunschweig
• director of the AVA in Göttingen
Subsequently his centre of life moved from Braunschweig to Göttingen, without ne-
glecting Braunschweig. This place he considered his scientific home. For the daily
work he proceeded that way being present three days a week in Braunschweig – this
was in general from Monday to Wednesday. In these days he also held his lectures.
Thursday to Saturday he was in his Göttingen bureau.
Of course he was supported by section / division heads and these were in the early six-
ties Arabindo Das, Rudolf Göthert, Klaus Gersten, Fred Thomas, Friedrich-Wilhelm
Riegels, Hubert Ludwieg, Walter Wuest and Klaus Jacob. Hermann Schlichting pursued
the further development of his institutes with great commentment.
With the foundation of the united research-establishment DFVLR he became
member of the board of directors of DFVLR responsible for the centre Göttingen.
Furthermore he supervised the research-area fluid mechanics in DFVLR.
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 31
In Schlichting’s opinion DFVLR was not the proper frame for aeronautics and
space research in Germany. He advocated a continuation of the long-standing inde-
pendency of AVA and was backed by all AVA staff. On the other side the German
Federal Parliament insisted on the unification of DVL, DFL and AVA and impended
a budget stop if this would not be achieved. This made the unification unavoidable.
Fig. 15. Influence of a boundary layer fence on the spanwise load distribution and the
downwash [25]
32 H. Körner
These two fields form a continuation of own earlier activities. These were added by:
3. delta wing aerodynamics
4. flexible wing aerodynamics
5. unsteady aerodynamic coefficients and derivatives.
And in the high speed area
6. transonic wings
7. supersonic wings
8. hypersonic configurations.
Subsequently some scientific results of collaborators of Schlichting are presented. The
Figures are taken from survey papers of Schlichting.
Figure 15 shows the high lift behaviour of a swept wing in clean condition as ana-
lysed by Das [25]. On swept wings the separation of the boundary layer starts at the
outer part of the wing. This detrimental behavior can be changed, introducing bound-
ary layer fences on the upper side. The flow stays attached and first separation occurs
at a higher angle of attack. Das also gives an explanation for this phenomenon. Vanes
have been widely used on early swept wing airplanes.
Another result of the high-lift research is shown in Figure 16 and deals with blow-
ing [27]. In the 50ies and 60ies a number of STOL and VTOL projects have been
pursued in Germany. Such types of aircraft need extremely high lift capabilities in the
take-off and landing phase. Thomas did investigations on the augmentation of lift by
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 33
internal blowing. His investigations show that a ∆CL of 6 can be achieved. He fur-
thermore shows that lower rates of blowing are more efficient. This can nicely be seen
from the diagram. Nowadays again this technique has appeal, since the conventional
Fowler-flap-systems of transport airplanes have become very complicated and heavy
and produce a lot of noise in take-off and landing conditions.
An alternate procedure to keep the flow attached under high lift conditions is
boundary-layer suction, which has to be used in areas where the boundary layer tends
to separate [24]. These suction experiments on a flapped airfoil by Wuest show, that
with a relatively low rate of suction a substantial gain in lift can be achieved
(Figure 17).
Another priority area in Schlichting’s institutes was delta-wing aerodynamics [28].
Gersten developed a nonlinear lifting-surface theory for small aspect ratio wings. In
his model of singularities the trailing vortices leave the surface with an upward kink
of the size α/2. This is in contrast to linear theory where the whole vortex system
stays planar. As Figure 18 shows, this theory reproduces the experimental results over
a large α-region. When the angle of attack becomes very high, discrepancies occur
which originate from vortex break-down on the upper side of the wing. Vortex break-
down was the subject of Hummel who engaged himself intensively with this
phenomenon.
Flexible Wing
M. Azmeh, K. Gersten, H.-W. Hucho
Aerodynamics
Unsteady
H. Otto, E. Schmidt
Aerodynamics
This is only a small part of the activities in Schlichting’s institutes. A detailed re-
sume cannot be given here.
Back to Hermann Schlichting himself: He now has become an influential member
of a number of scientific associations and affected the further development of the
scientific field of airplane aerodynamics. Among these memberships I would give the
highest rang to his involvement in AGARD (Advisory Group for Aerospace Research
Hermann Schlichting’s Work and Wake in Airplane-Aerodynamics 35
Fig. 18. Singularity model for planar and nonplanar lifting surface method and theoretical
results compared with experiment [28]
36 H. Körner
and Development, a subgroup of NATO). In the 50ies and 60ies AGARD was the
most important scientific community in the western world. Schlichting was member
of the AGARD-Fluid Dynamics Panel for twenty years and its first German chairman
in the period 1961 to 1963. This was an important step for the integration of German
aeronautial science in the free world. AGARD awarded him 1980 with its highest
honour, the v. Kármán-medal for his scientific work and his contributions to interna-
tional cooperation. This is only one of many honours he received. A resume is given
in Table 2.
The most important of them are the honorary doctorate of TH Munich, the Prandtl-
Ring of DGLR, the Großes Verdienstkreuz des Verdienstordens der Bundesrepublik
and as already mentioned the v. Kármán-medal.
Reaching his retirement age, Schlichting withdraw 1975 from his positions at
DFVLR and TU Braunschweig. He continued teaching at Braunschweig until 1978.
He died at Göttingen in 1982.
5 Schlichting Today
Secondly, there is something very specific and this is Schlichting’s work in the
field of transition laminar / turbulent which is of utmost importance for the design of
laminar wings (NLF and LFC). Damping Tollmien-Schlichting waves is nowadays an
important part of the design of low-drag airfoils, wings and configurations. Today the
laminar wing is the great challenge we are facing.
And third: Nowadays the Institute of Aerodynamics and Flow Technology at
Braunschweig and Göttingen of DLR is the centre of aerodynamic research in Ger-
many and one of the leading institutes in the world. Schlichting has constructed the
fundament for it. It was therefore decided to name the new large main building of the
institute „Hermann Schlichting Haus“ (Figure 19).
Acknowledgement
I would like to thank D. Hummel and H.U. Meier for supporting me in the search of
documents. A comprehensive account of the scientific life of H. Schlichting and the
TU-Institute of Fluid-Mechanics is compiled by D. Hummel [1, 31].
References
[1] Hummel, D.: Das Institut für Strömungsmechanik der TU Braunschweig. Appelhans-
Verlag, Braunschweig (2005)
[2] Tragflügeltheorie bei Überschallgeschwindigkeit. Luftfahrtforsch., vol. 13, pp. 320–395
(1936)
[3] Der Windkanal der Dornier-Werke. Luftfahrtforsch., vol. 15, pp. 95–100 (1938)
[4] Neuere Beiträge der Forschung zur aerodynamischen Flügelgestaltung (Umriß, Ver-
windung, Rumpfeinfluß), Lilienthal-Gesellschaft f. Luftfahrtforsch. Bericht, vol. 117, p.
29 (1939)
[5] Über den Einfluß von Flügel und Rumpf auf das Seitenleitwerk (gem. mit W. Frenz).
Jahrbuch, d. dtsch. Luftfahrtforsch., pp. I 300–I 314 (1941)
[6] Systematische Sechskomponentenmessungen über die gegenseitige Beeinflussung von
Flügel, Rumpf und Leitwerk (gem. mit W. Frenz). Jahrbuch, d. dtsch. Luftfahrtforsch.
Vorabdruck: TB Bd. 11, Heft 6 (1943)
[7] Monographie: Aerodynamik der gegenseitigen Beeinflussung der Flugzeugteile (Interfer-
enz). Bericht 46 / 5 des Inst. f. Strömungsmechanik der TH Braunschweig (1946)
[8] Meier, H.-U. (ed.): Die Pfeilflügelentwicklung in Deutschland bis 1945. Bernhard &
Graefe Verlag, Bonn (2006)
[9] Calculation of the influence of a body on the position of the aerodynamic centre of air-
craft with swept wings. ARC RM 2582 (1952)
[10] Über die theoretische Berechnung der kritischen Reynoldsschen Zahl einer Reibungss-
chicht in beschleunigter und verzögerter Strömung. Jahrbuch, d. dtsch. Luftfahrtfor-
schung. S I 97–S I 112 (1940)
[11] Zur Berechnung des Umschlages laminar / turbulent (gem. mit A. Ulrich). Lilienthal-
Gesellschaft f. Luftfahrtforsch. Ber., vol. S 10, pp. 75–135 (1940)
[12] Über die Berechnung des Umschlages laminar / turbulent (gem. mit A. Ulrich). Lilien-
thal-Gesellschaft f. Luftfahrtforsch., Ber. Nr. 141, pp. 17–26 (1941)
[13] Messungen an Laminarprofilen (gem. mit K. Bußmann). Lilienthal-Gesellschaft f. Luft-
fahrtforsch. Ber. Nr. 149, pp. 17–20 (1942)
38 H. Körner
Summary
1 Introduction
Aircraft concepts are always driven by the requirements of the desired mission. A
different purpose for the use of the aircraft consequently results in a different design.
This is illustrated in Figure 1, where examples of civil and military transport aircraft
and a fighter aircraft are shown. Depending on the intended purpose, different contra-
dicting requirements have to be fulfilled, e.g. efficient cruise speed and high cargo
capability in combination with short take-off and landing field lengths, or high speed
and agility combined with variable payload demands. Furthermore, the design task is
severely aggravated by the physical conditions under which aircraft have to operate.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 39–58.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
40 C.-C. Rossow and N. Kroll
transonic flow
Į shock wave & separation
angle of attack
buffet boundary
attached flow
dy
V dī
Į ī
b=2·s
c/4
c
r
where W = [ρ , ρu , ρv, ρE ]T represents the vector of conservative variables, F is the
r
flux-density tensor, and Vol, S, and n denote volume, surface, and outward facing
normal of the control volume. The flux density tensor F may be split into an invis-
cid, convective part Fc and a viscous part Fv :
F = Fc + Fv , (2)
where Fc and Fv are given by
r ⎡ 0 ⎤
⎡ ρq ⎤ ⎢ r r ⎥
⎢ r r⎥
⎢ τ xx i x + τ xy i y ⎥
ρu q + p i r r
Fc = ⎢ r r ⎥ , Fv = ⎢
x
τ xy i x + τ yy i y ⎥, (3)
⎢ ρvq + pi y ⎥ ⎢ ⎥
⎢ r ⎥ r
⎢⎛⎜ uτ + vτ + k
∂T ⎞ ⎛ ∂T ⎞ ⎥
r
⎣⎢ ρ ⎦⎥ ⎟ix + ⎜⎜ uτ xy + vτ yy + k ⎟i y
∂y ⎟⎠ ⎥⎦
H q ⎢⎝ xx xy
⎣ ∂x ⎠ ⎝
r r r
q is the velocity vector with Cartesian components u, v, and ix , i y denote the unit
vectors in direction of the Cartesian coordinates x and y. ρ , p, H , T represent density,
44 C.-C. Rossow and N. Kroll
pressure, total specific enthalpy, and temperature, k is the coefficient of thermal heat
conductivity, and τ xx ,τ yy ,τ xy are the viscous stress tensor components.
In order to close the system given by eq. (1), the equation of state
p / ρ = R ⋅T (4)
is used with R as specific gas constant. For turbulent flows the system of eq. (1) is
extended by an additional set of equations modeling the influence of turbulence to
properly account for the flow physics under consideration.
The origins in establishing the system of eq. (1) date back to the 18th and 19th cen-
turies, when significant work was done trying to mathematically describe the motion
of fluids. In 1755 Leonhard Euler proposed a set of differential equations [7], which
describe the conservation of mass and momentum for an inviscid fluid. More than 50
years later, Claude Navier in 1822 and George Stokes in 1845 independently intro-
duced viscous transport into these equations. Subsequently extended by including the
energy equation and being cast into the so-called conservation form to allow for cap-
turing of discontinuities, the differential mathematical equations being proposed
nearly 200 years ago form the basis of numerical simulation today. In honor of the
work of Navier and Stokes, the system of eq. (1) is called the system of Navier-Stokes
equations in conservation form, and neglecting the viscous part Fv of the flux density
tensor in eq. (1), the conservative form of the Euler equations is obtained.
Research on numerical methods was already performed in the first half of the 20th
century, when e.g. Courant, Friedrichs, and Levy defined a condition for certain algo-
rithms for solving partial differential equations to be convergent [8]. The advance-
ment in computer technology in the second half of the 20th century considerably
accelerated the development of numerical flow simulation, when in the 1950s and
1960s at the Los Alamos National Laboratory in the U.S. the world’s then largest
computational resources were created [9].
For aerodynamics, the development of Computational Fluid Dynamics was especially
driven by the interest in transonic flows, when in the early days of commercial jet avia-
tion the transonic drag rise phenomenon received particular attention. In the 1970s,
methods for solving the nonlinear potential equation were developed by e.g. Murman and
Cole [10], and matured by Jameson to the well-known code FLO-22 [11].
The solution of the Euler equations became a central focus of CFD research in the
1980s. Central spatial discetization schemes with artificial dissipation [12] as well as
upwind schemes were developed [13], [14], [15], and multigrid was introduced as a
means for convergence acceleration even suitable for hyperbolic equations [16]. At
the DLR Institute for Aerodynamics and Flow Technology in Braunschweig, solution
of the Euler equations started with a scheme following Jameson, Schmidt, and Turkel
[12], when a basic code was generously provided by the Dornier Company. Figure 8
gives an example of basic investigations carried out in Ref. [17], and Figure 9 repre-
sents the solution of the Euler equations for a transport aircraft configuration obtained
with a predecessor of the DLR-FLOWer code [18].
Computational Mesh for NACA 0012 Pressure Distribution with Mesh Refinement
The 1990s were the decade of advancements in solving the Navier-Stokes equa-
tions. Methods using structured meshes matured [19]; however the treatment of com-
plex configurations proved to be a major obstacle. The use of hybrid meshes [20]
combined with efficient solvers [21] addressed this problem. Early activities at the
DLR Institute for Aerodynamics and Flow Technology with respect to solving the
two-dimensional Navier-Stokes equations are represented by Figure 10 from [22].
Fig. 10. Solution of the Navier-Stokes equations on structured meshes, Ref. [22]
Fig. 11. Solution of the Navier-Stokes equations on unstructured meshes, Ref. [26]
Numerical Simulation – Complementing Theory and Experiment 47
The basic code was then extended to efficiently solve the three-dimensional Na-
vier-Stokes equations with multigrid [23], and the status of structured Euler- and
Navier-Stokes solvers at DLR was documented in [24]. At the DLR Institute for
Aerodynamics and Flow Technology in Göttingen, the development of the DLR-TAU
code was consequently initiated to allow for flow computation around complex ge-
ometries [25], and Figure 11 gives an example of applying the TAU code to accurate
drag prediction for engine-airframe interference effects [26].
From the mid of the 1990s, the CFD research and development activities at the DLR
Institute of Aerodynamics and Flow Technology were mainly embedded in the Ger-
man national program MEGAFLOW, which combined CFD development activities
from DLR, universities and aircraft industry. Initiated under the leadership of DLR,
the project goal was the development and validation of a reliable and efficient nu-
merical tool for the aerodynamic simulation of complete aircraft meeting require-
ments of industrial implementations [27], [28]. The MEGAFLOW software system
includes the block-structured Navier-Stokes code FLOWer and the unstructured Na-
vier-Stokes code TAU. Both codes have reached a high level of maturity and they are
intensively used by DLR and the German aerospace industry in the design process of
new aircraft. As follow-on activity, from 2004 to 2007 the project MEGADESIGN
was set up which focuses on the development and enhancement of efficient numerical
methods for shape design and optimization [29], [30]. In the following, examples of
flow computations for transport aircraft, military aircraft, and helicopter will be given
to demonstrate the current status of CFD development at DLR.
One key aspect of the development of a new transport aircraft is the design of an
efficient high-lift system for take-off and landing conditions. A possibility to increase
maximum achievable lift coefficients is the usage of small delta-wing like plates on
the engine nacelles. These so-called nacelle strakes generate vortices which run above
the wing, Figure 12, and influence wing and slat pressure distributions to shift flow
separation to higher angles of attack.
At cruise flight conditions the strakes should not produce any significant additional
drag. Investigations based on hybrid grid RANS solutions using the DLR-TAU soft-
ware have shown that the vortical flow phenomena induced by the nacelle strakes and
the qualitative influence on lift and drag may be captured by the computation [31], see
the comparison in Figure 13.
In order to however quantitatively predict the lift increment due to the strakes, spe-
cial care must be taken generating and adapting the grid with and without strakes [32],
since such computations require the resolution of very different geometrical length
scales: on the one hand the flow around the whole aircraft with deployed high-lift
system needs to be adequatly modeled, and on the other hand geometrical details of
less than one percent of wing-span and their influence on the overall flow field have
to captured, Figure 14. Within the EU-project EUROLIFT II it was demonstrated that
lift increments due to the nacelle strakes can be predicted within reasonable bounds
[33].
Another aspect addressed in the EUROLIFT II project was the effect of wing de-
formation on the aerodynamic behavior at high-lift conditions. Investigations were
carried out for the DLR-F11 configuration with both, take-off and landing settings for
the high-lift system [33]. For numerical flow simulation, the DLR-TAU code was
employed in coupled mode with a structural Finite Element Model (FEM), where the
FEM models were created by the Leichtwerk Company [34] with the commercial
software package ANSYS [35]. Computations were carried out for on-flow conditions
of M ∞ = 0.176, Re = 15,000,000 . Figure 15 shows the comparison of rigid and de-
formed wing shapes for take-off and landing configurations.
For the take-off configuration, aeroelastic equilibrium was computed for three dif-
ferent angles of attack. The maximum deflection in z direction ranges from about 22
mm for α = 7° up to 32 mm for α = 18.5° . In relation to the model wing half span of
1400 mm, this results in a maximum deflection of 1.6% for α = 7.0° and 2.3% for
α = 18.5° . For the landing configuration at α = 16.5° the maximum deflection was
computed to 34.6 mm (2.5%). Although the maximum deformation of the wing is
small even close to the wing tip, the change in the pressure distribution is clearly
visible, as displayed in Figure 16 for take-off and landing configuration.
take-off landing
Fig. 15. Deformation of DLR-F11 high-lift wing in take-off and landing configuration
cp
-2
-2
-1
0 0
Fig. 16. Pressure distributions of DLR-F11 high-lift wing with and without deformation
50 C.-C. Rossow and N. Kroll
Within the DLR internal project SikMa (Simulation of Complex Maneuvers) a mul-
tidisciplinary simulation environment was developed to predict the unsteady aerody-
namic behavior of a free flying elastic combat aircraft [36]. The numerical tool
combines time-accurate aerodynamic, aeroelastic and flight mechanics calculations,
and contributions from the DLR Institutes for Aeorelasticity, Flight Systems, and
Aerodynamics and Flow Technology were involved.
Computations were carried out for X-31 military configurations. Figure 17 illus-
trates the complexity of the vortex flow topology over the clean wing and fuselage at
an angle of attack α = 18° . A qualitative comparison between measured and computed
surface pressure distributions over the X-31 wing including control devices at
α = 16° and Re=2.07x106 is shown in Figure 18.
The main footprints of the vortices as well as their locations are predicted quite
well by the computation. The multidisciplinary simulation capability is demonstrated
by computation of a free-to-roll maneuver of the elastic X-31 configuration around
the longitudinal axis. A generic structural model containing the full aircraft wind
Fig. 17. Flow field around X-31 Fig. 18. Comparison of pressure distribution
tunnel configuration with fuselage, wings, and control surfaces was created. To simu-
late the free-to-roll maneuver around the longitudinal axis a simplified one degree of
freedom flight mechanic model was used, meaning that the acceleration of the model
depends only on the aerodynamic rolling moment and the moment of inertia around
the x-axis of the finite element model. The simulated maneuver starts from a roll
angle of 45° and a pitch angle of 15°. During the maneuver unsteady non-symmetric
loads on the aircraft lead to non-symmetric deformation of the structure.
Figure 19 shows the deflected FE model with generic stiffness and corresponding
transformed aerodynamic forces. Figure 20 compares the rolling angle for the rigid
and elastic model during a real time of 0.5s. The deflection of the elastic model due to
the aerodynamic forces induces a higher angle of attack and leads to a higher rolling
moment and a higher rolling frequency. This causes a higher aerodynamic damping of
the system, which has the effect of bringing the system to a trimmed position within a
shorter period of time than observed with the rigid model.
Numerical Simulation – Complementing Theory and Experiment 51
4.3 Helicopter
At DLR large effort is devoted to the enhancement of the MEGAFLOW software for
helicopter applications. The aerodynamics of helicopters present a formidable chal-
lenge for numerical simulation, since accurate prediction inherently requires the capa-
bility for unsteady, multi-disciplinary simulation. The numerical simulation of the
flowfield around helicopters has to deal with the different moving reference frames
for main and tail rotor, and with the fixed coordinate frame of the fuselage. In forward
flight, the rotor has to be trimmed by a corresponding module for flight mechanics,
and the elasticity of the blades has to be accounted for by appropriate fluid-structure
coupling. Furthermore, geometrical details like the empennage and landing-skids have
to be modeled, and at the bluff rear end of the fuselage strakes are often used to con-
trol separation.
Fig. 19. FE model with deformation and corre- Fig. 20. Evolution of rolling angle
sponding force vectors
-5
-4
-3
-2
cp
-1
2
-1 0 1 2 3
x
Fig. 21. Chimera grid for Bo 105 Fig. 22. Fuselage pressure distribution
flow fields, the DLR-FLOWer code with Chimera technique was used, and Figure 21
gives a view of the corresponding computational grid which consisted of 480 blocks
and 12 million cells. The rotor blades were still assumed to be rigid [37].
Figure 22 gives a comparison of experimental and computational results for the
pressure distribution on the fuselage. It was found to be essential to incorporate the
sting mount and landing-skids in the numerical simulation to achieve sufficient
agreement.
In Figure 23 the computed topology of the vortices generated by main and tail ro-
tors are displayed. Figure 24 shows a comparison of unsteady pressure on the tail
rotor at 80% radius. As can be seen from the figure, the computation accurately cap-
tures the interaction of vortices from the main rotor with the tail rotor at every fifth
revolution of the tail rotor.
In the follow-on EU-project GOAHEAD, the flow around the NH 90 helicopter is
computed with trimmed and elastic main rotor [38]. The corresponding computational
grid is shown in Figure 25, and Figure 26 displays the computed unsteady surface
pressure distribution on the complete helicopter. Experiments for this configuration
are planned for 2008 in the DNW-LLF.
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0 180 360 540 720 900 1080 1260 1440 1620 1800
psitr
Fig. 23. Rotor vortices of Bo 105 Fig. 24. Tail rotor pressure distribution
5 Two Challenges
Two main challenges may be identified for future development of numerical flow
simulation. The first challenge addresses the extension of the range of applicability
for CFD methods. Usually, today’s methods are applied well within the flight enve-
lope, where the aircraft operates under the conditions it was designed for. However, to
further reduce risk and cost in aircraft development, CFD methods need to be able to
reliably predict flow phenomena outside the regular flight envelope. Moreover, the
exact location of the boundary of the flight envelope, i.e. where aircraft operation
starts to become critical, is of utmost importance for a successful design, and it should
be accurately predicted. Within the flight envelope the flow stays mainly attached.
However, outside of the flight envelope the flow is governed significantly by severe
flow separation and is inherently unsteady, see also Figure 2. The CFD codes nowa-
days in routine use solve the Reynolds Averaged Navier-Stokes (RANS) equations,
where the influence of turbulence is modeled by eddy viscosity turbulence models.
Such models were mainly derived under the assumption of a steady main flow field,
and their application to unsteady flows with separation showed only limited success.
In Large Eddy Simulation (LES), more turbulent scales are resolved with the intention
to better account for the influence of turbulence. However, the enormous computa-
tional resources required for resolving the turbulent length and time scales for high
Reynolds number flows renders the use of LES prohibitive for the foreseeable future.
A practical alternative may be the use of hybrid RANS/LES methods which are suit-
able for technically relevant high Reynolds number flows. These methods are still
limited in application, and considerably more experience is required to bring these
methods to a level acceptable in an industrial design process. It should be noted that
such methods by definition require the solution of the unsteady equations, and accel-
eration techniques which exploit the existence of a final steady state may not be used.
The first challenge for future advancement in Computational Fluid Dynamics is there-
fore the one which accompanied numerical flow simulation from the very beginning:
the adequate simulation of transition and turbulence for engineering applications.
The second challenge is directly concerned with the cost of applying CFD methods
in the design process. To determine the static and dynamic loads required for struc-
tural design, hundreds of thousands of aerodynamic load cases need to be evaluated.
In Figure 27, the evolution of cost for experimental and numerical simulation is
sketched as a function of the number of simulations. For experimental simulation, the
initial cost is considerable due to model design and manufacture. However, since the
wind tunnel is very efficient in mass data production, the increase in cost with number
of simulations is only moderate. For numerical simulation, the opposite trend is ob-
served: initial cost is moderate, but repetitive simulation of cases with varying pa-
rameters results in a steep increase in cost. To make further inroads of CFD into the
industrial design process, cost has to be substantially decreased; e.g. by employing
more sophisticated meshing strategies, and by improvements in algorithms and hard-
ware. It should be noted that advancements in efficiency are also made in experimen-
tal wind tunnel (WT) simulation by means of rapid prototyping and modern
measurement techniques, as additionally indicated in Figure 27.
54 C.-C. Rossow and N. Kroll
improvements
algorithms & hardware
cost
future WT ?
future CFD ?
meshing
strategies
6 Perspectives
In the mid 1960s Gordon Moore, the co-founder of Intel, made the observation that com-
puter power, measured by the number of transistors that could fit onto a chip, doubled
once every 1.5 years [39]. This so-called Moore’s law performed extremely well over
more than the preceding 30 years, see Figure 28. The tremendous progress achieved in
numerical flow simulation over the last decades would not have been possible without
this increase of computational power. It may be arguable whether Moore’s law will still
hold for the next 30 years, as indicated by the extrapolation in Figure 28.
Accepting this assumption, in the table also displayed in Figure 28 the compute
power available on a typical 8 hours working day is extrapolated over the next 30
years, assuming a doubling of available compute power every two years only. In 30
years compute power would increase by a factor of 33000; and a compute job which
today would require 30 years is then run in only 8 hours. This scenario provides an
enormous potential for future numerical simulation.
A decisive step into the direction of future advancement in numerical simulation was
taken in May 2007 by DLR, Airbus, and the federal state of Niedersachsen, when
these three partners founded the Center for Computer Applications in AeroSpace
Science and Engineering (C2A2S2E). The goal of C2A2S2E is to establish an interdis-
ciplinary center of excellence for numerical aircraft simulation. The new center will
develop numerical methods and processes for highly accurate, multi-disciplinary
simulations of aircraft in an industrial context.
The research and development to be conducted in C2A2S2E addresses the following
four topics:
• real-time simulations of aircraft in flight based on the unsteady Navier-Stokes
equations coupled with a finite-element description of the aircraft structure and
the flight dynamic control laws
• calculation of the aerodynamic loads of the aircraft across the flight envelope,
including extreme flight conditions
• numerical prediction of aircraft performance and handling qualities prior to the
first flight
• certification prior to aircraft production based on numerical data
C2A2S2E consists of three major components, the first one being a major new aero-
space simulation center at the Research Airport Braunschweig on the premises of
DLR. The center provides a goal-oriented research environment dedicated to promot-
ing integrated research, development, and industrial application activities. Latest
developments and results in key areas of numerical simulation technology, such as
physical modeling, numerical methods and information technology, will be trans-
ferred to industrial partners. Scientists and engineers will work collaboratively in this
environment with a long-term perspective.
The second component is a campus environment that brings together world-
renowned experts and guest scientists to stimulate top-level research in the field of
numerical simulation.
The final component is the availability of professionally managed and operated
high-end computer and visualization hardware that meet the growing demand of
industry and applied research for computational power.
56 C.-C. Rossow and N. Kroll
It is expected that through the interdisciplinary research and exploitation of the ex-
pected growth in computational power, C2A2S2E will speed up numerical simulation
by a factor of 105–106 over the next 15 years. This, together with improved modeling
fidelity and a broader spectrum of applications, will accelerate the striving towards
the Future Simulation Concept of Airbus [40], and the vision of digital flight [41].
References
[1] Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeugs, vol. 2, p. 144. Springer,
Heidelberg (1969)
[2] Thomas, F.: Die Ermittlung der Schüttelgrenzen an Tragflügeln im transsonischen
Geschwindigkeitsbereich. Habilitationsschrift TH Braunschweig 1966; WGLR Jb, pp.
126–144 (1966)
[3] Raffel, M., Willert, C., Wereley, S., Kompenhans, J.: Particle Imange Velocimetry. In:
Experimental Fluid Mechanics, 2nd edn. Springer, Heidelberg (2007)
[4] Prandtl, L.: Tragflügeltheorie, I. Mitteilung. Nachrichten der Kgl. Ges. Wiss. Göttingen,
Math. -Phys. Klasse, pp. 151–177 (1918)
[5] Shannon, R.E.: Systems simulation – the art and science. Prentice Hall, Englewood Cliffs
(1975)
[6] Jackson, J.A., Bates, R.L.: Glossary of Geology, 3rd edn. American Geological Institute
(1987)
[7] Euler, L.: Institutiones Calculi Differentialis, vol. I (1755)
[8] Courant, R., Friedrichs, K., Lewy, H.: Über die partiellen Differenzengleichungen der
mathematischen Physik. Mathematische Annalen 100, 33–74 (1928)
[9] Johnson, N.L.: The Legacy and Future of CFD at Los Alamo. In: Proc. of 1996 Canadian
CFD Conference, Ottawa, June 3–4, Los Alamos National Laboratory Report LA-UR-96-
1426 (1996)
[10] Murman, E.M., Cole, J.: Calculation of Plane, Steady, Transonic Flows Past Profiles.
AIAA J. 9, 114–121 (1971)
[11] Jameson, A., Caughey, D.A.: Numerical Calculation of the Flow past a Swept Wing,
ERDA Research and Development Report COO-3077-140, Courant Institute of Mathe-
matical Sciences, New York University (1977)
[12] Jameson, A., Schmidt, W., Turkel, E.: Numerical Solutions of the Euler Equations by Fi-
nite Volume Methods using Runge-Kutta Time-Stepping Schemes, AIAA-Paper 81-1259
(1981)
[13] Roe, P.L.: Approximate Riemann Solvers, Parameter Vectors and Difference Schemes. J.
Comput. Phys. 43, 357–372 (1981)
[14] Steger, J., Warming, R.: Flux Vector Splitting of the Inviscid Gas Dynamic Equations
with Applications to Finite Difference Methods. J. Comput. Phys. 40, 263–293 (1981)
[15] Leer, v.B.: Flux Splitting for the Euler Equations. In: Proc. of 8th Int. Conf. on Num.
Meth. in Fluid Dynamics, pp. 507–512. Springer, Heidelberg (1982)
[16] Jameson, A.: Solution of the Euler Equations by a Multigrid Method. Appl. Math. And
Comp. 13, 327–356 (1983)
[17] Kroll, N., Jain, R.K.: Solution of the Two-Dimensional Euler Equations – Experience
with a Finite Volume Code, DFVLR-FB 129-87/11 (1987)
[18] Rossow, C.-C., Ronzheimer, A.: Investigation of Interference Phenomena of Modern
Wing-Mounted High-Bypass-Ratio Engines by the Solution of the Euler Equations,
AGARD CP 498 (1992)
Numerical Simulation – Complementing Theory and Experiment 57
[19] Rogers, S., Roth, K., Nash, S., Baker, M., Slotnick, J., Whitlock, M.: Advances in Over-
set CFD Processes Applied to Subsonic High-Lift Aircraft, AIAA-Paper 2000-4217
(2000)
[20] Kallinderis, Y., Khawaja, A., McMorris, H.: Hybrid Prismatic/Tetrahedral Grid Genera-
tion for Viscous Flows Complex Geometries. AIAA J. 34, 291–298 (1996)
[21] Mavriplis, D.J.: Three-Dimensional Multigrid Reynolds-Averaged Navier-Stokes Solver
for Unstructured Meshes. AIAA J. 33, 445–453 (1995)
[22] Radespiel, R., Rossow, C.-C.: A Cell Vertex Finite Volume Scheme for the Two-
Dimensional Navier-Stokes Equations, DFVLR-IB 129-87/40 (1987)
[23] Radespiel, R., Rossow, C.-C., Swanson, R.C.: An Efficient Cell-Vertex Multigrid
Scheme for the Three-Dimensional Navier-Stokes Equations. AIAA J. 28(8), 1464–1472
(1990)
[24] Kroll, N., Radespiel, R., Rossow, C.-C.: Accurate and Efficient Flow Solvers for 3D Ap-
plications on Structured Meshes. AGARD Report R-807, pp. 4.1–4.59 (1995)
[25] Gerhold, T., Friedrich, O., Evans, J., Galle, M.: Calculation of Complex Three-
Dimensional Configurations Employing the DLR-TAU Code, AIAA-Paper 97-0167
(1997)
[26] Brodersen, O.: Drag Prediction of Engine-Airframe Interference Effects Using Unstruc-
tured Navier-Stokes Calculations. AIAA J. 39(6), 927–959 (2002)
[27] Kroll, N., Fassbender, J.K. (eds.): MEGAFLOW – Numerical Flow Simulation for Air-
craft Design. Notes on Numerical Fluid Mechanics and Multidisciplinary Design
(NNFM), vol. 89. Springer, Heidelberg; Closing Presentation DLR Project
MEGAFLOW, Braunschweig, Germany (2002)
[28] Rossow, C.-C., Kroll, N., Schwamborn, D.: The MEGAFLOW-Project – Numerical Flow
Simulation for Aircraft. In: DiBucchianico, A., Mattheij, R.M.M., Peletier, M.A. (eds.)
Progress in Industrial Mathematics at ECMI 2004, pp. 3–33. Springer, New York (2005)
[29] Kroll, N., Gauger, N., Brezillon, J., Becker, K., Schulz, V.: Ongoing Activities in Shape
Optimization Within The German Project MEGADESIGN. In: Proc. of ECCOMAS
2004, Jyväskylä, Finland, July 24–28 (2004) (on CD)
[30] Kroll, N., Becker, K., Rieger, H., Thiele, F.: Ongoing Activities in Shape Optimization
with the German Project MEGADESIGN. In: Proc. of Annual Meeting of the Interna-
tional Council of the Aeronautical Sciences, ICAS-2006-3.11.1, Hamburg, Germany
(2006)
[31] Brodersen, O., Wild, J., Melber-Wilkending, S., Lekemark, L.: DLR-IB 129-2003-34,
DLR Braunschweig (2003)
[32] Brodersen, O., Wild, J.: DLR-IB 124-2004-18 (2004)
[33] Geyr, v.H., Schade, N., Burg, v.d.J.W., Eliasson, P., Esquieu, S.: CFD Prediction of
Maximum Lift Effects on Realistic High-Lift-Commercial-Aircraft-Configurations within
the European Project EUROLIFT II, AIAA-Paper 2007-4299 (2007)
[34] http://www.leichtwerk.de
[35] http://www.ansys.com
[36] Schütte, A., Einarsson, G., Raichle, A., Schöning, B., Orlt, M., Neumann, J., Arnold, J.,
Mönnich, W., Forkert, T.: Numerical Simulation of Maneuvering Aircraft by Aerody-
namics, Flight Mechanics and Structural Mechanics Coupling, AIAA-Paper 2007-1070
(2007)
[37] Schwarz, T., Khier, W., Raddatz, J.: Simulation of Unsteady Flow Field around a Com-
plete Helicopter with a Structured RANS Solver. In: Resch, M., Bönisch, T., Benkert, K.,
Furui, T., Seo, Y., Bez, W. (eds.) High Performance Computing on Vector Systems, pp.
125–137. Springer, Heidelberg (2006)
58 C.-C. Rossow and N. Kroll
[38] Khier, W., Dietz, M., Schwarz, T., Wagner, S.: Trimmed CFD Simulation of a Complete
Helicopter Configuration. In: 33rd European Rotorcraft Form, Kazan, Russia, September
11–13 (2007)
[39] Moore, G.E.: Cramming more Components onto Integrated Circuits. Electronica 34
(1965)
[40] Klenner, J., Becker, K., Cross, M., Kroll, N.: Future Simulation Concept., In: CEAS Con-
ference Berlin, Session Numerical Simulation, Paper No. 1 (September 2007)
[41] Salas, M.D.: Digital Flight: The Last CFD Aeronautical Grand Challenge. J. Scient.
Comp. 28(213) (2006)
Modern Wind Tunnel Techniques for Unsteady
Testing – Development of Dynamic Test Rigs
A. Bergmann
Summary
A survey is given about the capabilities of dynamic wind tunnel testing in Germany.
The survey is based on an overview of the historical development including works
from the beginning of the 1960s, therewith from the recommencement of the German
wind tunnels after World War II. This information has so far been available only in
internal reports and in German language. In a second part a review of new develop-
ments of dynamic testing capabilities at the German-Dutch Wind Tunnels DNW is
presented.
1 Introduction
In most fluid dynamics applications the natural phenomenon “unsteady flow” occurs.
Its variety is large and includes e.g. periodic flows from oscillating bodies, unsteady
flows due to a stability problem, turbulence, flow separation, vortex breakdown, flows
from maneuvering configurations and so on. The capability to increase the knowledge
about unsteady aerodynamics and to control complex unsteady flows opens beneficial
improvements of aircraft. It is well recognized that aerodynamic research has always
been and will be increasingly concerned with unsteady testing to increase the under-
standing of the physics as well as to provide suited data sets for the validation of the
upcoming CFD-methods.
Generally, it can be distinguished between the interest in predicting the behavior of
a maneuvering aircraft on the one hand and the interest in analyzing and understand-
ing the unsteady flow itself on the other hand. In the first case, the prediction of the
flight qualities is matter of concern as well as the prediction of aerodynamic induced
loads e.g. during a maneuver. So the forces and moments against the motion and its
variations are of interest and have to be determined. This requires special techniques
and rigs in the wind tunnel environment and a survey of their development and de-
scription especially with respect to the DNW-NWB wind tunnel is given in the
present article.
However, in the second case the measurement and prediction of the flow parame-
ters themselves is of interest and this requires certain experimental measurement
techniques like field methods. Their description is subject of an extra article in this
volume given by C. Kähler.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 59–77.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
60 A. Bergmann
For the prediction of the dynamic behavior of aircraft the derivatives of the aero-
dynamic forces and moments with respect to their momenta ( α& , β& , p, q, r) and their
control surface deflections (η, ξ, τ) are of interest. These flight-mechanical derivatives
can be distinguished into static derivatives and dynamic derivatives. Such parameters
can be obtained through dynamic experiments in a wind tunnel or other types of fa-
cilities or can be extracted from full-scale flight tests, cp. [1, 2]. As the results of
flight tests are obtained in a rather very late phase of aircraft development the predic-
tion from wind tunnel experiments is important up to today, in spite of the lack of data
at high Reynolds Numbers.
A general survey of literature about the determination of dynamic stability deriva-
tives is given in Refs. [2–4]. The present article contains further information, espe-
cially regarding the historical development of the experimental capabilities in German
research facilities mainly from extracted information from the AGARD/RTO publica-
tions mentioned in [3] and from internal reports available in German only.
only minor influence on the dynamic derivatives if the frequency is adequately large
enough, see [5]. The data scatter increases only when the oscillation frequency is set
too low.
In [6] a simplified compilation of different mechanisms of the oscillatory type is
given and this is displayed in Table 1, where the benefits and disadvantages are given.
The listing distinguishes between free oscillations and forced oscillations whereas the
latter is divided in elastic excitation and rigid excitation. Each type shows certain
characteristics so that benefits and disadvantages have to be balanced carefully
against the requirements. The simplest system uses a free oscillating model with the
major restriction that only aerodynamically stable configurations can be studied. This
is also the case for elastically suspended models with forced elastic excitation. Further
on, the excitation frequency f0 is given by the characteristics of the used springs, and
investigations against the reduced frequency ω* are only possible by varying the
onflow velocity U∞ so that the Reynolds Number cannot be kept constant with these
types and the influence of the amplitude cannot be studied at all. For a systematic
analysis of the impact of amplitude and frequency the excitation has to be rigid to
enable the adjustment of frequency and amplitude separately, but this leads to systems
which are much more complex.
3 Historical Development
The determination of dynamic derivatives in Braunschweig and Göttingen has a long
tradition. In 1972, a national working group was constituted, consisting of DFVLR
and the universities of Bochum and Darmstadt, representing aeronautical research,
62 A. Bergmann
Fig. 1. Thompson Rig with light-weight models in open and closed test section of NWB, taken
from [12]
and of the companies Dornier, MBB, and VFW-Fokker, representing the national
aeronautical industry. It was the objective to obtain reliable workbenches for the
determination of dynamic derivatives in experienced, high-quality production runs.
Already before that time some activities regarding the development of such re-
search facilities had taken place in Germany. As after the end of World War II the
aeronautical research was forbidden for several years it was not before the early 1960s
that the construction and implementation of new wind tunnel facilities could be com-
pleted. Concurrently first unsteady pressure measurement tests on oscillating wings
and later on half models were carried out at AVA in Göttingen and the development
of a two-component derivative balance according NASA-standard was accomplished
at DVL in Köln. German industry developed a first coning rig at „Entwicklungsring
Süd“ in München [7]. This was all done to catch up with the international state of the
art and in this regard likewise a collaboration between DFL Braunschweig and Royal
Aircraft Establishment [RAE] Bedford was agreed to replicate one of at that time
most modern test beds for the experimental determination of dynamic derivatives, the
Thompson-Rig [8–10]. This rig was reproduced in detail, hazarding the consequences
that the Thompson-Rig, primarily designed for investigations in trans- and supersonic
flows, was too imprecise for the low speed regime [11]. In 1968, the test bed was well
established by systematic tests on four cropped delta wings in the closed test section
of NWB, see Figure 1. This was by order of the Ministry of Defense and it was the
very first time for the parallel determination of all dynamic derivatives of longitudinal
motion in Germany [12].
The working principle was according to the method of forced oscillation with ex-
ternal elastic excitation. In this method, the model is suspended elastically on certain
springs with carefully chosen eigenfrequencies. For small power consumption the
exciter works in resonance frequency. As the resonance frequency for the different
modes for heave and pitch are different, distinct modes can be effected, of course with
superposition of a small share of the other modes respectively. However, the range of
Modern Wind Tunnel Techniques for Unsteady Testing 63
possible operations was relatively small. The maximum values for pitch amplitudes of
approximately ±1° and for heave ±0.01 m were relatively poor, but the achieved accu-
racy for the shift error in phase with max. ±0.3° was already within today’s standards.
The measured signals were the acceleration of the model and the frequency and power
of the exciter. The appropriate data acquisition system was developed at DFL and the
design and manufacturing of the wind tunnel models took place in the DFL-workshop
situated in Braunschweig-Kralenriede. To meet the requirements for the models to be
as stiff and as light as possible an integral monocoque construction method using
reinforced plastics was applied, keeping the weight between 30 kg and 50 kg for a
typical 1 m span slender delta wing with its lowest eigenfrequency well above 12 Hz.
The capabilities in Germany for dynamic testing trace back to the activities of a work-
ing group constituted in 1972. The objective was to provide facilities with the ability
to determine reliable aerodynamic derivatives in high quality production runs. This
was coordinated by the Ministry of Research and Development and the management
was delegated to Prof. X. Hafer from Darmstadt University. A comprehensive survey
of the devices is given in [6] and for detailed information about the measuring tech-
nique and the appropriate evaluation method to obtain the dynamic derivatives refer
to [13]. Additional information regarding a dynamic balance for transonic wind
tunnels can be obtained from [14, 15].
The programme comprised five projects altogether, which are briefly described in
the following.
Spring Elements
Fig. 2a. MFD, Programme Management v.d. Decken, Dornier, Subcontract: DFVLR Göttingen,
taken from [6]
Fig. 2b. Detail of the internal pitch spring element with sensors for heave and pitch acceleration [12]
Fig. 3a. MOD in the open test section of NWB, Programme Management v.d. Decken, Dornier,
Subcontract: VFW-Fokker, TU Darmstadt, DFVLR Göttingen
Fig. 3b. General arrangement of the internal motion unit with a six-component DMS Balance,
driven by a push rod
• For usage in 3 m low speed wind tunnels, first Köln, later Braunschweig
• Support of the model by rigid rear-sting
• Principle of forced steady rolling/coning with rigid excitation
• Internal balance
Fig. 4a. RTD in the closed test section of NWB, Programme Management H. Schulze, MBB,
Subcontract: DFVLR Köln and Braunschweig, Universities of Bochum and Darmstadt
Fig. 4b. General arrangement of RTD cart in rigging position and in the test section
Project 5
• Free oscillatory derivative balance (Freioszillierende Derivativwaage FROD)
• Developed by DFVLR Köln-Porz [21]
Modern Wind Tunnel Techniques for Unsteady Testing 67
Fig. 5b. Detail of the internal pitch element with balance [14]
with the MFD-system a lot of know-how was gained for the realization of the TRAD
for transonic application.
Recently launched research programmes with new AGARD/RTO working groups,
e.g. see [22, 23], showed a changing interest in dynamic testing to provide more ex-
perimental datasets for CFD validation. This was also the motivation for resuming
that kind of testing in DNW-NWB, after two decades of decommission. Within the
project MEGAFLUG the contributions of the airplane components were to be quanti-
fied more precisely compared to methods used so far and the first test started in the
end of the 1990s using the available MOD, being still ready for operation.
Concurrently at DERA a new six-degree-of-freedom dynamic test rig mechanism
was suggested for the usage in the 13” × 9” wind tunnel based on the mechanism of a
Stewart platform [24] with the background to investigate maneuvering characteristics
of future combat aircraft configurations.
The idea to use parallel kinematics is promising because of their good dynamic
properties. Already in 1949 Gough started with activities of designing and developing
a robotic manipulator with joined legs [25] and hereupon Stewart described a six-
component parallel kinematic in 1965 widely used for flight simulators [26]. But only
more than 20 years later their full dynamic potential could be used at last with the
increased computational power. At DERA the above mentioned strategy to develop a
parallel mechanism for wind tunnel tests was followed up to realize a novel six-
degree-of-freedom motion rig mechanism [27].
With serial kinematic structures, as depicted e.g. in Figure 4, the number of DoF
is achieved by serial arrangement of the corresponding number of linear and rotative
axes. So the bottom-most axis of movement has to carry the weight of all those lying
above it and this leads to a contradiction between the requirement for high stiffness
and high dynamics. For that reason most existing dynamic wind tunnel rigs are lim-
ited to a small (one or two, maximum three) degree-of-freedom motion. This results
in the measurement of combined aerodynamic derivatives, e.g. a pitching motion
gives a combination of α& and q derivatives while yawing motion gives a combina-
tion of β& and r derivatives. At small angles of attack and small amplitudes linear
aerodynamic characteristics can be assumed and these derivatives can be separated
by obtaining α& from (translational) plunging or β& from (sidestroke) motion ob-
tained with rotation of the model through 90° in roll. But this assumption is not valid
for high angles of attack. In addition, large amplitude motions with the described
single-degree-of-freedom oscillatory rigs are not possible so that these devices can-
not represent the typical aircraft motions during maneuvering flight. Further, for
non-linear flight regimes some coupling may exist between longitudinal and lateral
aerodynamics.
From that background a mechanism for continued motion of wind tunnel models
up to six DoF is utterly in demand. At ONERA the Sacso system [28] was developed
quite recently for the small LS vertical wind tunnel in Lille. The model is held by a
carbon fiber beam which is suspended in the flow by nine wires, see Figure 6. The
wires are motorized to realize six DoF trajectories imposed on the model. That limits
the size of the models to typically 1 m length. Nevertheless, it is possible to simulate
free-flight tests.
Modern Wind Tunnel Techniques for Unsteady Testing 69
Fig. 6b. Vertical wind tunnel in Lille equipped with the SACSO system [28]
For a new approach at DNW-NWB the minimal eigenvalue of the stiffness tensor
as an indicator for the assessment of the design was chosen. In an optimization proce-
dure several thousand design candidates with different design variables were com-
pared with respect to stiffness and with a penalty term regarding exceptionally high
actuator acceleration. To achieve constant stiffness on the required workspace the
stiffness has to be analyzed at several distinct positions, distributed carefully in the
workspace. The basic concept with rods with constant length and linear guiding rails
according [29] was chosen as the most promising concept regarding cost effective-
ness, dynamic capabilities, and achievable workspace, see Figure 9.
The optimized frozen design is displayed in Figure 10a, b and a picture taken from
the new Model Positioning Mechanism (MPM) in the open test section is shown in
Figure 11. The mechanism is located above the test section allowing also tests with
ground effects using e.g. rolling road devices. The dynamic characteristics of the
complete test set-up was checked by dedicated vibration tests. This included also the
effects from the building structure and the effect of higher harmonic control for
the platform actuators. It turned out that the lowest eigenfrequency at the end of the
sting (Tool Center Point TCP) was above 20 Hz, matching the predicted value.
This excellent dynamic behavior is attributed to the application of the linear direct
drive technology. This avoids a conventional ball screw drive with its elasticity in the
drive chain and allows accelerations on the actuators up to 2.5 g. The accuracy of the
overall system in pivoting angles is less than 0.005°, quite sufficient for wind tunnel
applications.
Modern Wind Tunnel Techniques for Unsteady Testing 71
Fig. 8a. Example of Hexapod application in DNW-NWB Fig. 8b. Working principle
of a Stewart platform
72 A. Bergmann
Fig. 10b. Test bed set-up of MPM in the closed test section of DNW-NWB
Fig. 11. The new Model Positioning Mechanism (MPM) in the open test section of DNW-
NWB [30]
Modern Wind Tunnel Techniques for Unsteady Testing 75
six actuators an independent seventh axis has been placed on the Stewart platform,
which can be used to excite the pitch or roll oscillation via a suitable mechanism in-
side the model that transforms the translatory motion of the additional actuator into
the required rotatory motions, cp. Figure 3b. Larger amplitudes and higher frequen-
cies are possible this way because it saves the Hexapod’s upper frame from moving
along a circular path which is also possible but which creates large undesired inertia
effects.
The major characteristic of the test rig is its high dynamic capability combined
with high and nearly constant stiffness over the whole workspace which spans
1100 mm in flow direction, 300 mm in lateral direction and 500 mm in heave
direction. To meet the demands of large amplitude and high-rate arbitrary motion
the MPM has the following advantages compared to a conventional serial axes
arrangement:
• Higher dynamics despite identical input power because lower weights are being
moved
• Higher accuracy because errors in parallel kinematics exert less effect
• Lower prices due to simpler construction and identical components for each axis
• Lower demands on tolerances for production and assembly because geometric
transformation takes the place of axial alignment
The MPM complements the today’s dynamic capabilities of the German-Dutch Wind
Tunnels. A first successful test campaign with an X31 fighter model, equipped with
remotely controlled rudders, slats and flaps was conducted [30], and demonstrates the
system ability even for simulating complex maneuvers. The test rig is complemented
by the necessary know-how of model manufacturing, fast data acquisition and
conditioning, measurement techniques like telemetric systems and so on. In addition
several dynamic tests of various transport aircraft configurations were performed in
the recent years. Most of the tests are computationally complemented by DLR as
described in [31].
5 Conclusion
A survey of the existing test rigs for dynamic testing capabilities in German wind
tunnels was given. With all the gained experience from the past activities and with
the recently developed Model Positioning Mechanism MPM even the capability of
simulating complete maneuvers in a wind tunnel now exists. In the last decade
almost ten different light-weight wind tunnel models have been built and tested.
Possibilities of CFD validation with integrated flight mechanical simulation are
thereby realized. This ability is an important step ahead in the development of a
multi-disciplinary simulation environment. Further work will be done on elastic
models to assess the influence of twist and bending of the wing on the dynamic
derivatives.
76 A. Bergmann
References
[1] Orlik-Rückemann, K.: Methods of Measurement of Aircraft Dynamic Stability Deriva-
tives, National Research Laboratories NRC, Canada, LR-254 (1959)
[2] Orlik-Rückemann, K.: Review of Techniques for Determination of Dynamic Stability Pa-
rameters in Wind Tunnels., AGARD-LS-114, pp. 3-1–3-28 (1981)
[3] Bergmann, A., et al.: Experimental and Numerical Research on the Aerodynamics of Un-
steady Moving Aircraft. In: Progress in Aerospace Sciences, vol. 44. Elsevier, Amster-
dam (in press, 2008)
[4] Loeser, T.: Experimental Research of Dynamic Derivatives of Unsteady Moved Aircraft.
In: VKI Lecture Series on Experimental Determination of Dynamic Stability Parameters
(2008)
[5] Loeser, T., Bergmann, A.: Development of the Dynamic Wind Tunnel Capabilities at
DNW-NWB. In: Proceedings of the 41st AIAA Meeting, AIAA 2003-0453, Reno (2003)
[6] Hafer, X.: Wind Tunnel Testrig of Dynamic Derivatives in W-Germany. AGARD-CP-
235, pp. 5-1–5-22 (1978)
[7] Müller, A., Klinke, H.: Ein Beitrag zur Messung der Rolldämpfung im Windkanal,
DGLR-Jahrbuch (1962)
[8] Thompson, J.S., Fail, R.A.: Oscillatory Derivative Measurements on Sting-Mounted
Wind Tunnel Models: Method of Test and Results for Pitch and Yaw on a Cambered
Ogee Wing at Mach Numbers up to 2.6., ARC R+M No. 3355, UK (1964)
[9] Leedham, H.C.: A Digital Amplitude and Phase Measure System., RAE TR-64070, UK
(1964)
[10] Thompson, J.S., Fail, R.A.: Oscillatory Derivative Measurements on Sting-Mounted
Wind Tunnel Models at R.A.E. Bedford., RAE-TR-66197 (1966)
[11] Trienes, H., Thomas, F.: Über die experimentelle Bestimmung instationärer Luftkraft-
beiwerte in englischen Forschungsanstalten“, DFL-Bericht Nr. 0117 (1961)
[12] Schmidt, E.: Die DFL-Thompsonwaage zur Messung von instationären aerodynamischen
Derivativa bei mäßigen Geschwindigkeiten, DFVLR-Bericht Nr. 0552 (1969)
[13] von der Decken, J., Schmidt, E., Schulze, B.: On the Test Procedures of the Derivative
Balances used in West Germany, AGARD-CP-235, pp. 6-1–6-17 (1978)
[14] Schmidt, E., Wagener, J.: Konstruktive Auslegung der Transkanal-Derivativwaage
(TRAD), Festschrift zum 60. Geburtstag von Prof. Försching, Inst. f. Aeroelastik, DLR-
GÖ, pp. 105–126 (1990)
[15] Schmidt, E.: Standard Dynamics Model Experiments with the DFVLR/AVA Transonic
Derivative Balance, AGARD-CP-386, pp. 21-1–21-16 (1985)
[16] Schmidt, E.: Die AVA Derivativwaage, DLR-Mitt. 74-32, 1974, and ESA TT-197 (1975)
[17] Scherer, M., Lozez, J.: Progrès réalisés dans les techniques de réserve des Dérivatives
aérodynamiques en soufflerie méthode d’oscillations forces. In: AGARD-CP-17, pp.
409–436 (1966)
[18] Schlottmann, F.: Stationäre und instationäre Rollmomentenderivative schlanker Flügel in
Rollbewegung. In: ZFW, pp. 331–344 (1974)
[19] Schmidt, E., Sauerland, K.-H.: Derivativawaage für den Transkanal Göttingen (TRAD)
Teil 1: Beschreibung der Meßeinrichtung, DVFLR IB 232-81J10 (1982)
[20] Schmidt, E., Sauerland, K.-H.: Derivativawaage für den Transkanal Göttingen (TRAD)
Teil 3: Windkanalmessungen am MBB-Pilotmodell bei Nickschwingung, DFVLR IB
232-82J10 (1982)
Modern Wind Tunnel Techniques for Unsteady Testing 77
[21] Niezgodka, F.-J.: Beschreibung eines Systems zur Messung von Nickdämpfungsbei-
werten nach der Methode der freien Oszillationen, DFVLR-Mitt. vom 29.8.1980, Köln
(1980)
[22] Cooperative Programme on Dynamic Wind Tunnel Experiments for Manoevring Aircraft,
AGARD-AR-305, Report of WG 11 (1996)
[23] Verification and Validation Data for Computational Unsteady Aerodynamics, RTO TR-
26, Report of RTO WG-003, AGARD WG-22 (2000)
[24] Greenwell, D.I.: Analysis of a Six-Degree-of-Freedom Dynamic Wind Tunnel Test Rig
Mechanism Based on the Stewart Platform, AS / HWA / TR 961 02/1, DERA, Farnbor-
ough (1996)
[25] Gough, V.E., Whitehall, S.G.: Universal Type Test Machine. In: Proceedings of the FI-
SITA 9th Int. Tech. Congress, pp. 117–137 (1962)
[26] Stewart, D.: A Platform with Six Degrees of Freedom. In: Proceedings of the Inst. Mech.
Engrs., vol. 180(1.15), pp. 371–385 (1965)
[27] Greenwell, D.I., Entwistle, P.M., Richards, D.E.: A Novel Six-Degree-of-Freedom Large-
Amplitude Motion Rig Mechanism for Dynamic Wind Tunnel Testing. In: Proceedings of
the Conference on Wind Tunnels and Wind Tunnel Techniques, Cambridge, UK, pp.
29.1–29.14 (1997)
[28] Farcy, D., Llibre, M., Carton, P., Lambert, C.: SACSO: Wire-Driven Parallel Set-Up for
Dynamic Tests in Wind Tunnel – Review of Principles and Advantages for Identification
of Aerodynamic Models for Flight Mechanics. In: 8th ONERA-DLR Aerospace Sympo-
sium, ODAS, Göttingen (2007)
[29] Wiegand, A., Weikert, S.: Vorrichtung zur räumlichen gesteuerten Bewegung eines Kör-
pers in drei bis sechs Freiheitsgraden, Internationale Patentanmeldung WO 97/22436
(1996)
[30] Rein, M., Höhler, G., Schütte, A., Bergmann, A., Löser, T.: Ground-Based Simulation of
Complex Maneuvers of a Delta-Wing Aircraft. AIAA Journal of Aircraft 45(1), 286–291
(2008)
[31] Hübner, A.: Numerical Research of Dynamic Derivatives of Unsteady Moved Aircraft.
In: VKI Lecture Series on Experimental Determination of Dynamic Stability Parameters
(2008)
Principle and Perspective of Optical Multipoint
Methods for Aerodynamic Investigations
C.J. Kähler*
Summary
In this article, the most powerful and reliable optical multipoint measurement tech-
niques applied today in fluid mechanics for estimating velocity fields, density gradi-
ents, model deformations, temperature, and pressure distributions are outlined and
advantages and limitations of the techniques are disscussed. The analysis implies that
the methods are well qualified for many scientific and engineering investigations.
However, with increasing performance of the measurement and simulation tech-
niques, the man increases the complexity of the scientific questions and expands the
range of examination. Thus, the technological progress and the scientific effort must
continue to keep up with future needs.
1 Introduction
Complex flow problems occuring in science and technology require powerful experi-
mental multipoint techniques to assess the three-dimensional, instationary effects and
to validate numerical flow simulations. By taking the enormous costs of the operation
of modern wind tunnels into account, the techniques must also be reliable and effi-
cient. The custumer, on the other hand, needs an online result of the measurement in
order to select the parameter for the next experiment or wind tunnel run. As only the
optical multipoint methods fulfill these requirements, they are applied in many uni-
versity laboratories and research organizations today. The success of these methods
relies mainly on two developments:
1. digital cameras, which capture the desired signal at millions of points simultane-
ously and
2. computer hardware, which stores, processes and visualizes the recorded or
evaluated images, such that the results become accessible to the human intellect.
In fluid mechanics the following techniques are typical examples: particle image
velocimetry (PIV), particle tracking velocimetry (PTV), molecular tagging veloci-
metry (MTV), Doppler global velocimetry (DGV), pressure sensitive paint (PSP),
infrared (IR) thermography, image pattern correlation technique (IPCT), projection
*
Now director of the Institute of Fluid Mechanics and Aerodynamics of Bundeswehr University, Munich,
http://www.unibw.de/lrt7/
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 78–104.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Principle and Perspective of Optical Multipoint Methods 79
Fig. 1. Dependency of the signal S (normalized grey values) on the number of photons given by
the intensity at the sensor Esensor multiplied with the illumination time ∆t (left). Dependency of
the signal to noise ratio SNR on Esensor ∆t (right) after [2].
inside the flow. This is done indirectly via the displacement of moving particle
groups, see Figure 2 and [3]. For this purpose the flow region under consideration is
homogeneously seeded with appropriate tracer-particles such that their injection and
presence does not affect the flow or the fluid’s properties. The concentration of the
particles must be well adjusted with regard to the finest flow structures in order to
sample the flow properly. The deviation of the particle velocity uP from the real flow
motion u must be negligible compared to the uncertainty of the imaging and recording
system and to the uncertainty due to the evaluation procedure. Therefore, olive-oil
particle with a mean diameter of around 1 µm are frequently applied for applications
in air [5]. Figure 3 shows the measured particle size distribution which depends on the
Fig. 2. Schematic of particle image velocimetry after [4]. A desired plane inside the flow is
illuminated twice by a thin laser light-sheet and the scattered light emerging from the homoge-
neously distributed particles in the direction of the imaging optics is recorded.
Fig. 3. Volumetric particle size distribution for various hole diameters (d = 0.5, 1.0, 2.0 µm),
pressure (p = 1 bar), number of holes per nozzle (n = 4 and 12) after [6]. The result with the
impactor (Imp), which rejects the large particles, illustrates the ideal distribution.
Principle and Perspective of Optical Multipoint Methods 81
Fig. 4. Light intensity scattered by spherical particles of different size in air (left: dP=1µm,
right: dP=10µm), illuminated from left with a plane monochromatic wave front. The complex
spatial intensity distribution, with a maximum in forward direction, results from the interfer-
ence between the reflected, refracted and diffracted wave front.
of the technique itself. In terms of accuracy, for example, the particles should be suf-
ficiently small and their density should exactly match the density of the surrounding
fluid. Unfortunately, this is not often feasible for a desired field of view or a given
laser power, light sheet thickness, transparency of the fluid, imaging optics and sensi-
tivity of the digital camera, as the scattering intensity decreases rapidly with decreas-
ing particle diameter. Decreasing the light-sheet width or thickness may partially help
but the size of the largest resolvable scales will decrease as well and the three dimen-
sionality of the flow may cause further problems as the number of paired particle
images decreases. To use a powerful laser seems to be the appropriate solution but
beside the costs, strong reflections from model surfaces and undesirable disturbances
of the flow due to acoustic excitation, and thermal response of the flow have to be
taken into account. An intensified camera could be applied as well but a reduced spa-
tial resolution and an increased noise level must be accepted.
2.2 Evaluation
In order to extract the displacement information from the two single exposure grey-
level patterns acquired at t and t', both images are usually evaluated by means of sta-
tistical evaluation techniques for two reasons. First, individual particle-image pairs
cannot be reliably identified due to the high seeding concentration [3]. Second, the
statistical evaluation is less sensitive to noise and discretization effects of the cameras.
In [8–10] it is shown that the particle-images must be exactly identical in size, shape
and intensity. The particle images must also be homogeneously distributed and the
structure of the pattern must be invariant under spatial transformations to ensure that
the particle-image displacement is directly related to the location of the signal
peak RD, which is calculated by cross-correlating the two images locally by means of
interrogation windows as indicated in Figure 5.
RD
t
t'
Fig. 5. Evaluation of the images acquired at t and t'. The images are sampled with small inter-
rogation windows to estimate the signal peak RD locally by cross-correlation, after [3].
When these constraints are ignored, the correlation approach is still possible but
the measurement uncertainty increases. To obtain a good signal, the following relation
has to be taken into account while setting up the experiment:
RD ~ N I FI FO F∆ (2)
where NI represents the so called image density which is actually the number of particle-
image pairs in the interrogation window which contribute to the signal strength. FI and
FO denote the in-plane and out-of-plane loss-of-particle-image-pairs as a result of the
Principle and Perspective of Optical Multipoint Methods 83
finite size of the measurement volume. F∆ accounts for the loss of correlation due to
gradients within the measurement volume. The amplitude of the cross-correlation peak
increases with increasing image density and decreases with increasing displacement, no
matter in which direction or if the displacement of the particle-images is not constant. In
order to optimize the performance of the experiment, the number of paired particle-
images is of primary importance as only these add to the signal strength. Additionally,
the number of correct pairings with respect to incorrect combinations has to be optimized
to increase the signal-to-noise ratio and thus the detectability of the signal. This can be
achieved either by changing the seeding concentration, the magnification of the imaging
system or simply by increasing the light-sheet thickness or the size of the interrogation
window. In-plane loss-of-pairs, caused by particles entering or leaving the measurement
volume during the illuminations, can be reduced by decreasing ∆t or can be eliminated
when the search window contains the corresponding particle-image pattern selected with
the template. This can be achieved by using window-shifting in connection with multi-
pass interrogation techniques, see [11] and [12] for details. Out-of-plane loss-of-pairs can
be compensated by shifting the light-sheet in the direction of the mean particle displace-
ment, as proposed in [9], or by using multiple-light-sheet arrangements as applied in [13].
The effect of in-plane gradients can be reduced by decreasing the magnification of the
imaging system, and the time separation between the two illuminations or, alternatively,
by using properly-shaped interrogation-windows whose linear dimension is reduced in
the direction of the gradient or window-deformation techniques, see [14, 15] for details.
By using sophisticated image analysis methods, the particle-image displacement can be
determined with an accuracy better than 0.05 pixel if high-quality CCD cameras are
applied in conjunction with high-quality double-pulse lasers [16–18]. For time-resolved
PIV image acquisition the accuracy is slightly reduced, as discussed in [17], because of
the larger pixels, the higher noise, the lower sensitivity of CMOS cameras according to
[2], and the decreasing pulse-energy with increasing repetition rate of the laser. However,
when the motion of the particles is recorded, more sophisticated image analysis methods
can be applied, which take the temporal variation into account according to [19]. In this
case the relation above must be extended to
in order to take the effects due to acceleration FA and path curvature FC of the seeding
particles into account. Additionally, it is useful to separate F∆ into in-plane F∆I and
out-of-plane F∆O gradient effects, as only the first one can be reduced by using the
image deformation methods mentioned above.
To illustrate the potential of the PIV technique the laminar separation bubble (LSB)
on the suction side of a l = 200mm long SD7003 airfoil was investigated at Re =
66000, see [20]. As the bubble is only a few millimeters high (hLSB) but very long,
multiple cameras were applied side by side to obtain the desired spatial resolution and
field of view. Figure 6 shows that the mean velocity distribution at the location of the
bubble (top) and the distribution of the Reynolds shear-stress component (bottom).
The results indicate that the relevant information can be accurately resolved with the
84 C.J. Kähler
Fig. 6. Mean velocity distribution (top) and distribution of the Reynolds shear-stress component
(bottom) in the separated region of an airfoil after [20]
conventional PIV technique although the velocity in the bubble and in the outer flow
are quite different.
To visualize vortices generated in the shear-layer above the bubble, the λ2 criteria was
calculated from a time-resolved PIV measurement, to detect the position x / l of the vor-
tices and their distance y / l relative to the airfoil surface. Figure 7 illustrates the vortices
which are convecting downstream with increasing measurement time tnorm= t U / hLSB.
The color implies that they move away from the model surface with increasing x / l.
Fig. 7. Temporal visualisation of regular vortices generated in the shear-layer and y/l location [20]
Principle and Perspective of Optical Multipoint Methods 85
Fig. 8. Frequency and amplitude of the vortex shedding and bubble flapping [20]
Figure 8 finally shows, that the time-resolved PIV method is also capable of esti-
mating the dominant frequencies and amplitudes of the flow vortex shedding (color
coded by Strouhal number St =f hLSB / U) in the shear-layer and re-attachment zone of
the bubble.
The conventional PIV technique described in the previous section yields reliable re-
sults as long as the flow under investigation is mainly two-dimensional and parallel to
the light-sheet. In case of 3D flows with a strong velocity component normal to the
light-sheet, the out-of-plane velocity component remains unknown and the in-plane
components are biased due to the perspective error [3, 4]. To solve this problem the
stereoscopic imaging configuration was developed [21, 22]. Using the stereoscopic
recording technique, the images of tracer particles are recorded simultaneously from
two different viewing directions, and the correct displacement (without perspective
error) of the particle ensembles are reconstructed by using the stereoscopic-equations
[3]. The accuracy of the velocity measurement depends mainly on the opening angle
between the two observation directions [23]. For conventional PIV (single light-sheet
single camera configuration) the variation of the magnification factor over the field of
view is usually negligible because the image-plane and the main plane of the imaging
system are parallel to the light-sheet according to Figure 2. As this is usually not the
case for the stereoscopic configuration, a local calibration is required to take the de-
formation of the recorded images into account [21, 22, 24, 25]. Additionally it is very
important that the location of the calibration target is precisely located at the light-
sheet location to avoid the so called offset-error, which can be easily larger than
any other error in PIV [26, 27]. Nevertheless, this technique is routinely applied in
university and industrial wind tunnels.
86 C.J. Kähler
Particle Image Velocimetry (PIV) has become a widely used technique whenever the
spatial distribution of the velocity vector together with its in-plane derivatives helps to
understand the physics of the flow. Unfortunately, the distribution of the velocity
within one single plane, captured at one instant in time, often does not yield enough
information to answer fluid-mechanical questions. Therefore, a double stereoscopic
particle image velocimetry based system was developed which is called multi-plane
stereo PIV or double-stereo PIV in the literature. It allows the determination of all
three velocity components in spatially separated planes simultaneously, (in order to
determine the three component vorticity vector field for example) or separated in time
(for space-time investigations or acceleration measurements); see [3, 4, 13, 28, 29] for
details. For this purpose, the flow field is illuminated with orthogonally linearly polar-
ized light delivered by four frequency doubled Nd:YAG lasers. The light scattered by
the tracer particles is separated by means of a polarizing beam splitter pair onto the
sensors of four progressive scan CCD cameras in stereoscopic configuration. The
multi-plane stereo PIV system is very reliable, robust and well suited for all kinds of
applications. Furthermore it is based on the conventional PIV equipment and the fa-
miliar evaluation procedure so that available PIV systems can easily be expanded.
The advantage of this measurement system with respect to other imaging techniques
lies in its ability to determine a variety of fundamentally important fluid-mechanical
quantities with high accuracy (no perspective error) simply by changing the time
sequence or light sheet position. It is also possible to further increase the measure-
ment precision by combining this method with the multi-frame evaluation approach
discussed in [19].
For the analysis of 3D flow fields the tomographic PIV method was recently devel-
oped [3, 38]. By using this recording technique the illumination of the tracer particles
within a volume is recorded by means of four cameras from different observation
directions. In this case the positions of the particles can be reconstructed by using
tomographic reconstruction algorithms [38]. Then the evaluation of the 3D particle
image cloud is performed using a 3D cross-correlation approach with the window-
shifting and image-deformation analysis outlined in section 2.2. The advantage of this
approach is that the three-component velocity field in a flat volume can be measured,
as displayed in Figure 10. The experiment was performed in a water tunnel and the
mean flow above a finite cylinder head is shown [39].
Today the main limitation results from the fact that the illumination of small parti-
cles within a volume requires high power lasers. This is in particular because the
aperture of the lenses must be closed to ensure that out-of-focus particles can be
avoided. Additionally, reflections due to illuminated models cause problems which
must be solved to expand the range of applications. Finally, it is important to keep in
88 C.J. Kähler
Fig. 10. Tomographic PIV measurement of the flow above a finite cylinder in different
representations, after [39]
mind that the information of the whole volume is stored on the cameras by using this
method. This implies that the spatial resolution is reduced, compared to conventional
PIV, as the total number of vectors which can be calculated from a given sensor is
nearly constant.
Many aerodynamic flows reveal compressible effects. Since the schlieren or interfer-
ometric methods are difficult to align and operate in large-scale wind tunnels, the
background oriented schlieren method (BOS) was developed [40]. The principle of
the technique is very similar to the aforementioned PIV method. The basic principle
can be explained on the basis of Fermat's principle of least time, which states that out
of all possible paths between any pair of points, light takes the path which requires the
shortest time, or
B
δ ∫ n ds = 0 (5)
A
The integral denotes the optical path length and n is the refractive index which is
related to the dielectric constant ε by the Maxwell relation n = ε1/2. Thus, a variation
of n alters the path of a wave towards increasing values of n, provided the spatial
gradient of n is not exactly equal to the propagation direction of the wave. The origin
of n is based on the polarisability of the atoms and molecules because any forced
vibration of an atom or molecule, caused by the incident radiation, is associated with
the radiation of spherical waves. The superposition of both waves results in a new
propagation direction with an apparently different speed. Based on this concept the
following equation can be derived
Nk e2
n = 1+ ∑ (6)
k 2ε 0 m(ω k2 − ω 2 + iγ k ω )
Principle and Perspective of Optical Multipoint Methods 89
were Nk is the number of bounded electrons per unit volume whose natural frequency
is ωk and whose damping factor is γk. The symbol e is the charge, m is the mass of an
electron and ω denotes the frequency of the incident radiation. This relation implies
the dependency of n on the density N and thus on the macroscopic pressure p and the
temperature T as well (the dependence on the frequency difference ∆ω can be ne-
glected here). When a plane dot-pattern is recorded twice (once with a constant n
between the image and the object plane and once with n ≠ const) a spatial correlation
between both images results in a displacement field. Unfortunately, this displacement
field is not unique in this imaging configuration. This becomes evident when the
regions of varying n are simple optical elements located between the dot-pattern and
the camera. In this case, it is impossible to differentiate if the imaging is performed
with a single lens or a complex optical system with the same imaging characteristics,
or if the diffraction is based on a small optical element close to the camera or a large
one close to the dot-pattern.
4.2 Evaluation
The Background Oriented Schlieren technique is fully based on the conventional PIV
recording and evaluation methods [40–42]. However, in contrast to PIV, no loss-of-
pairs can take place because the number of dots is conserved. Furthermore, the dot
density and the shape and intensity of the dots can be optimized easily for all imaging
conditions. This increases the accuracy of the evaluation from 0.05 pixel to better than
0.01 pixel with state-of-the-art evaluation schemes [16–18]. Only noise due to random
correlations and non-uniform dot displacements within an interrogation window may
lower the performance of the evaluation or truncated dots. However, due to the small
displacements which can be usually realized with this technique, these effects are of
minor significance.
4.3 Application
To illustrate the capabilities of the technique, a well defined density gradient was
generated by focusing a single laser pulse such that a plasma is generated locally as
indicated in Figure 11. Due to the strong heating, a shock-wave is generated immedi-
ately after generating the plasma, which becomes a sound wave after a certain time,
see [43, 44]. The dot-pattern applied for the illustration consists of black circles, ap-
proximately 0.1 mm in diameter, which are randomly distributed on a white back-
ground. The illumination of the pattern was achieved with a photographic white-light
flash-lamp (Metz 40 MZ-2). The registration of the dot-pattern was performed with
the PCO camera in single frame mode, placed 720 mm apart from the dot-pattern and
460 mm away from the focus point of the laser. The field of view was 69 × 51 mm2
and the integration time was set to 500 ns to assure that the displacement uncertainty
of the shock-wave is below the spatial resolution of the interrogation window. The
sequence of images displayed in Figure 12 shows the magnitude of the local
displacement, measured 15.0, 57.5 and 95.0 µs after the generation of the plasma.
The propagation of the spherical wave from the image border, were the plasma is
generated, is clearly visible.
90 C.J. Kähler
Fig. 11. Principal BOS set-up and recording geometry for calibrating the technique for quanti-
tative investigations, see [43]. The dot density is much higher for real applications.
Fig. 12. Propagation of an acoustical wave in a half space, measured 15.0, 57.5 and 95.0 µs
after the generation of the plasma with a focused laser-pulse. The stationary peak is an artefact
of the camera (strong signal writes directly into the memory of the camera as the non-light-
sensitive area of the sensor cannot protect the storage from strong illuminations).
Though the scales are just projections of the disturbance into the object plane, the
propagation velocity of the acoustic wave cannot be determined from this measure-
ment. However, when the image or object plane is translated along the optical axis, as
indicated in Figure 11, and another measurement of the same process is taken, the
technique can be calibrated for quantitative purposes. This approach can be regarded
as a one-axis tomographical method. Since ∆y=2(x+∆x+l) tan(α) and sin(α) = r / (x +
∆x), the following expression can be calculated by using the identity sin(α) = tan(α)
(1– sin2(α))-1/2: ∆y = (2r (x + ∆x + l))/({(x + ∆x)2 – r2}-1/2), see Figure 11 (right).
This equation implies that r, x and l can be determined experimentally when the
same process is observed from three non equal locations along the optical axis. How-
ever, as the location of the focal point is fixed in the present investigation, and thus l
is fixed as well, only two independent measurements were required to determine r
and x. After repeating this procedure for a slightly different ∆t, the speed of sound
was estimated to 349 m/s which indicate the potential of this method [43]. However,
it should be stated that due to the evaluation approach the spatial resolution of this
technique is at least one order of magnitude lower than classical schlieren methods,
Principle and Perspective of Optical Multipoint Methods 91
see [44] for comparison. This drawback can be compensated by using background
images with continuous grey-scale variation in combination with optical flow
evaluation methods [45, 46]. Another obvious drawback is the fact that the quantita-
tive analysis of the results is very difficult when the density variation is not as simple
as in the present experiment. Therefore tomographic reconstruction algorithms are
applied [46–48].
When the geometry of the model in the field of view is known or can be determined,
an approach based on the Hough transformation can be applied as well. The Hough
transformation is an algorithm for finding lines [50]. For the determination of the
airfoil position, the reflection of the light-sheet on the surface is used. Therefore the
curved model reflection has first to be transformed digitally into a line. This can be
done by selecting control points in the image. Since the curve holds its shape and only
changes its position from image to image, the reference curve has to be created once.
The Hough transformation is based upon the Hessian normal form representation of a
line r = x cos(φ) + y sin(φ) where the two parameters used to define the line are the
radius r and the angle φ (see Figure 13). By using a set of lines running through the
n = (cos(ϕ),sin(ϕ))
. line
r
same point in space a set of parameter pairs r, φ is obtained. When only points in
space which belong, presumably, to the searched line are used to calculate the pa-
rameters, one pair of parameters will appear the most. This is the searched pair of
parameters for the line.
For applying the Hough transformation, a criterion has to be applied which selects
only points belonging to the curve. For the used line of reflection this can be a simple
threshold filter. Using the equation given above with the found parameters at a specific
position x gives the position in the transformed space and, after transforming it back, it
gives the correct position. The potential of this approach was demonstrated in [49].
When the three dimensional deformation of a model has to be determined, the image
pattern correlation technique (IPCT) must be applied [51]. Using this technique, the
surface of the model is covered with a random dot-pattern, similar to the one applied
for the recording of BOS images in section 4. When the random dot-pattern is re-
corded from two different viewing directions according to section 3.1 the displace-
ment field between a reference image (acquired without aerodynamic loads) and the
image with aerodynamic loads can be determined for each viewing direction by using
the spatial correlation algorithms described in section 2.2. After calibrating the sys-
tem, this set of images can be used to calculate the local deformation and bending of
the model [51, 52]. This method was applied to estimate the deformation of a flexible,
flapping airfoil, displayed in Figure 14 with the setup shown in Figure 15.
The two-dimensional flapping motion of the airfoil – a superposition of a heaving
and a pitching motion – was performed with the appartus outlined in [53]. A time-
shifted trigger signal from this apparatus is captured to perform phase locked imaging
of the stereoscopic camera arrangement.
After the acquisition and evaluation of 1000 image pairs at each phase angle (im-
age with no wind and no flapping correlated with image with wind and flapping mo-
tion), the resulting set of deformation vector fields were averaged to esitmate the
deformation along one chordline in the middle of the airfoil for each phase angle as
shown in Figure 14 right.
Fig. 14. Left: Flexible airfoil for flapping wing propulsion [53]. Right: Deformation of the
flapping airfoil in the middle of the downstroke along the chord line in the middle of the airfoil.
The error bars in the chart indicate the RMS value of the local deformation.
Principle and Perspective of Optical Multipoint Methods 93
Fig. 16. Principle arrangement of the PPCT measurement technique after [54]
The drawback of the IPCT is the required dot-pattern which has to be painted on
the model. In the case of the projection pattern correlation techniue (PPCT), a power-
ful laser projector is applied to project a random pattern on the model [54]. The pro-
jection must be performed at a small angle, such that small surface deformations
induce large lateral shifts of the projection pattern. The pattern is recorded by means
of a CCD single camera and successively recorded images are cross-correlated to
determine the pattern shift from which the deformation velocity can be calculated.
This is done by taking the time-seperation between the recordings and the imaging
and projection geometry into account. To estimate the motion of the model in the
direction normal to the surface quantitatively, a calibration is required. Therfore the
relation between the pixel coordinates and the image coordinates must be determined
94 C.J. Kähler
precisely using a calibration target which is translated in the range where the model is
expected, while the experiment is running. Further details about the calibration and
error analysis can be found in [54].
In some cases it may also be possible to use the natural texture of the model to de-
termine the deformation. However, as this pattern is usually very different from the
dot-pattern, the accuracy may be reduced as the image analysis methods are optimized
for dot-patterns. However, by using evaluation methods based on optical flow [45],
this is not necessary anymore. Another advantage of this method is the resolution
which is better than the spatial-correlation based methods by one order of magnitude.
6 Infrared Thermography
The measurement of the surface heating is very important for estimating the thermal
load in the hypersonic flow regime, for instance. However, also at small Mach num-
bers, the heat transfer is frequently estimated, because it is possible to determine the
location of the transition from the laminar to the turbulent flow state, due to the dif-
ferent heat transfer rates of laminar and turbulent flows. Today infrared cameras are
usually used because they are easy to handle and instantaneous temperature measure-
ments at thousand of points are possible with a sampling rate in the kHz range. De-
pending on the sensor technology, these cameras work in different wavelength
regimes such as the short wavelength regime between 0.4 and 2–3 µm, the intermedi-
ate between 2–3 to 5 µm and the long regime between 8 and 12 µm. In fluid mechan-
ics the range between 2 and 5 µm is usually recorded by using high-resolution indium
antimonite (InSb) or mercury cadmium telluride sensors. These sensors have excellent
signal to noise ratio and are available with 14 bit dynamic range. Unfortunately, these
cameras are very expensive and up to now they have always a number of non-active
pixel elements due to the complicated manufacturing process.
By using IR thermography, the radiation collected by the sensor is a superposition
of the radiation from the model surface IT,obj, which is attenuated by the atmospheric
transmittance τ, the radiation from the surroundings IT,amb (such as wind tunnel walls),
the camera lens and camera itself, and the radiation reflected from the model IT,atm,
which is attenuated by the atmosphere in the same way as IT,obj. The emission of the
model surface depends on its emissivity ε, as the model is never a perfect black body.
By neglecting the radiation from the atmosphere, the measured radiation is given by:
6.2 Evaluation
As the dependencies are quite complex, the infrared camera must be properly cali-
brated. At first, a non-uniformity correction matrix for the pixels of the sensor must
Principle and Perspective of Optical Multipoint Methods 95
be determined by taking images of two black plates with different uniform tempera-
tures. Correction values for each pixel are then calculated accounting for the offset
and slope deviation with respect to the mean of all pixels. The absolute sensitivity is
determined using a black radiator which is basically a cavity in a constant temperature
body. Therefore the black radiator must be placed in the test section and the tempera-
ture must be varied in order to estimate the calibration curve over the range expected
during the measurements.
6.3 Application
Fig. 17. Thermal decay of the surface after heating with a laser-shot after [43]
96 C.J. Kähler
Fig. 18. Thermal load on a 240 mm long model at Ma∞=5.91 and Re∞=4·106
on the model quantitatively. The surface heat flux is calculated from the transient
temperature data using a modified Cook and Felderman method which was regular-
ized by an iterative method in order to reduce the sensitivity to the measurement noise
[56]. The effect of the small round disturbances on the centerline which is caused by
classical temperature sensors is nicely visible.
Another example is the heat transfer on a cone in hypersonic flows. Figure 19 (left)
illustrates the top view on the calibration grid attached to the cone (top image) and the
lower left image shows the same image after the de-warping. Figure 19 (right) shows
Fig. 19. Left: Calibration image of a 3D cone model (top) and after the de-warping of the image
(bottom). Right: Measured heat transfer by using an Indigo Phoenix high-speed IR camera with
InSb sensor.
Principle and Perspective of Optical Multipoint Methods 97
Fig. 20. Heat transfer distribution on a 3D cone after calibrating the three individual measure-
ments (with permission of M. Estorf)
although the slope, and thus the resolution, is slightly smaller. Unfortunately the exci-
tation wavelength is in the UV range (340 nm), so that care must be taken while oper-
ating the technique. For illumination, a Nd:YAG laser with λ=355 nm is well suited
but also low cost diodes or flash-lamps are available which match the excitation fre-
quency even better. Nevertheless, the intensity of the light must be stablized for accu-
rate measurements. As the emitted wavelength is blue, conventional CCD cameras,
with a spectral filter to separate the undesired frequencies due to the illumination from
the signal, can be applied for the registration of the emitted light, but their linearity
and sensitivity should be excellent to obtain accurate results. Unfortunately the mod-
els cannot be painted directly with Pyren. Thus an appropriate binder, which stores
high concentrations of Pyren, is required to obtain a good signal in a short time. The
binder’s permeability for oxygen should be large, especially for dynamic investiga-
tions, so that the response time is short. In order to ensure that the signal strength is
identical for all model positions, it is very important that the thickness of the paint is
constant all over the model and also that the illumination is homogeneous. As this
is hard to achieve in practice a calibration of the measurement system is required to
compensate all imperfections.
Intensity
Fig. 21. Spectral distribution of the fluorescent light in dependency on the pressure (left) and
pressure dependency of the normalized intensity (right) after [60]
7.2 Evaluation
One of the major drawbacks of the luminescent molecules is the strong dependency of
the fluorescent signal on the temperature. As most of the wind tunnels are heated up
while the experiment is running, the temperature effect has to be calibrated as well in
a calibration chamber which allows for the variation of the pressure and temperature
independently. Alternatively, a two component paint can be applied whose compo-
nents behave differently under pressure or temperature variations [61]. However, to
compensate offset effects as well, the model should also have a few classical pressure
and temperature probes, to calibrate the PSP signal or the IR thermography technique
outlined in section 6, can be applied. As it is required to acquire the intensity over
some time in order to obtain a good signal quality, the model and the equipment
should not vibrate, especially when many independent images are recorded for calcu-
lating one single low noise pressure field. This requires that the model is equipped
with some artificial markers so that the deformation or movements of the model can
Principle and Perspective of Optical Multipoint Methods 99
be detected and compensated by using the digital image analysis approaches outlined
in section 5. Furthermore, care must be taken with the humidity conditions, the con-
tamination of the model due to seeding particles used for PIV (or dirt), and the dete-
rioration of the paint. As the signal is relatively weak it is also important that the wind
tunnel windows have an appropriate coating so that losses due to reflection or ghost
images, caused by multiple reflections inside the glass window, do not lower the
precision of the measurements.
7.3 Application
The measurement principle was first sucessfully applied in 1980 to visulize an oxygen
jet [62]. With increasing performance of the method it has been applied in many dif-
ferent areas of fluid mechanics [57, 63–68]. Figure 22 shows, for example, the com-
plex pressure distribution on a 3D wind tunnel model [1, 61]. This data set makes it
possible to estimate the aerodynamic forces and moments of the whole model, as
would be possible with classical balance techniques. Nevertheless, the load of single
components, like the airfoils, flaps or tailplanes can be estimated without integrating
additional balances, provided the contribution of the shear-stress force, which cannot
be estimated with the PSP technique, is small. Additionally, many fluid mechanics
details can be examined besides the integral values. Thus, it is possible to localize the
areas where flow separation takes place or where cP becomes small as a function of
the angle of attack or flap settings for instance.
Fig. 22. Time-averaged pressure distribution measured on the 3D model surface in a transonic
wind-tunnel to calculate the force and moments; after [1] and [61]
One of the major challenges in the future is the development of paints with higher
sensitivities. This is required for investigations at small Reynolds numbers or low
pressures. Additionally, the diffusion of oxygen should be improved to enhance the
response time of the technique. This will be in particularly important for instationary
100 C.J. Kähler
measurements in the future. The use of anodized aluminum models were the lumines-
cent molecules are direcly attached to the surface may be promising in the future [69,
70]. Finally the preparation of the models and the operation of the technique should
be improved so that more universities and small companies can benefit from it. This
would push the technique in new directions.
8 Conclusion
References
[1] Tropea, C., Yarin, A.L., Koss, J.F. (eds.): Handbook of Experimental Fluid Mechanics.
Springer, Heidelberg (2007)
[2] Hain, R., Kähler, C.J., Tropea, C.: Comparison of CCD, CMOS and intensified cameras.
Exp. Fluids 42, 403–411 (2006)
[3] Raffel, M., Willert, C.E., Wereley, S.T., Kompenhans, J.: Particle Image Velocimetry – A
practical Guide, 2nd edn. Springer, Heidelberg (2007)
[4] Kähler, C.J.: The significance of coherent flow structures for the turbulent mixing in
wall-bounded flows, DLR research report DLR-FB-2004-24 (2004),
http://webdoc.sub.gwdg.de/diss/2004/kaehler/kaehler.pdf
[5] Melling, A.: Tracer particles and seeding for particle image velocimetry. Meas. Sci.
Tech. 8, 261–304 (1997)
[6] Kähler, C.J., Sammler, B., Kompenhans, J.: Generation and control of particle size distri-
butions for optical velocity measurement techniques in fluid mechanics. Exp. Fluids 33,
736–742 (2002)
[7] Born, M., Wolf, E.: Principles of Optics, 7th edn. Cambridge U. Press, Cambridge (1999)
[8] Adrian, R.J.: Statistical properties of particle image velocimetry measurements in turbu-
lent flow. In: Laser Anemometry in Fluid Mechanics III, pp. 115–129. Springer, Heidel-
berg (1988)
[9] Keane, R.D., Adrian, R.J.: Theory of cross-correlation analysis of PIV images. Appl. Sci.
Res. 49, 191–215 (1992)
[10] Westerweel, J.: Fundamentals of digital particle image velocimetry. Meas. Sci. Tech. 8,
1379–1392 (1997)
[11] Willert, C.E., Gharib, M.: Digital particle image velocimetry. Exp. Fluids 10, 181–193
(1991)
[12] Westerweel, J., Dabiri, D., Gharib, M.: The effect of a discrete window offset on the ac-
curacy of cross-correlation analysis of digital PIV recordings. Exp. Fluids 23, 20–28
(1997)
[13] Kähler, C.J., Kompenhans, J.: Fundamentals of Multiple Plane Stereo PIV. Exp. Fluids
Supplement 29, S70–S77 (2000)
[14] Huang, H.T., Fiedler, H.E., Wang, J.J.: Limitation and improvement of PIV. Exp. Flu-
ids 15, 168–174, 263–273 (1993)
[15] Scarano, F.: Iterative image deformation methods in PIV. Meas. Sci. Technol. 13, R1–
R19 (2002)
[16] Stanislas, M., Okamoto, K., Kähler, C.J.: Main results of the first international PIV Chal-
lenge. Meas. Sci. Technol. 14, R63–R89 (2003)
[17] Stanislas, M., Okamoto, K., Kähler, C.J., Westerweel, J.: Main results of the second in-
ternational PIV Challenge. Exp. Fluids 39, 170–191 (2005)
[18] Stanislas, M., Okamoto, K., Kähler, C.J., Westerweel, J., Scarano, F.: Main results of the
third international PIV Challenge. Exp. Fluids (2008) (online)
[19] Hain, R., Kähler, C.J.: Fundamentals of Multiframe Particle Image Velocimetry (PIV).
Exp. Fluids 42, 575–587 (2007)
[20] Hain, R., Kähler, C.J., Radespiel, R.: Dynamics of laminar separation bubbles at low
Reynolds-number airfoils. J. Fluid Mech. (submitted, 2008)
[21] Willert, C.E.: Stereoscopic digital particle image velocimetry for application in wind tun-
nel flows. Meas. Sci. Tech. 8, 1465–1479 (1997)
[22] Prasad, A.K.: Stereoscoptc particle image velocimetry. Exp. Fluids 29, 103–116 (2000)
102 C.J. Kähler
[23] Lawson, N.J., Wu, J.: Three dimensional particle image velocimetry: error analysis of
stereoscopic particle techniques. Meas. Sci. Tech. 8, 894–900 (1997)
[24] Soloff, S.M., Adrian, R.J., Ziu, Z.-C.: Distortion compensation for generalized of stereo-
scopic particle image velocimetry. Meas. Sci. Tech. 8, 1441–1454 (1997)
[25] Willert, C.: Assessment of camera models for use in planar velocimetry calibration. Exp.
Fluids 41, 135–143 (2006)
[26] Kähler, C.J., Adrian, R.J., Willert, C.E.: Turbulent boundary layer investigations with
conventional and stereoscopic particle image velocimetry. In: 9th International Sympo-
sium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, June
13–16, paper 11-1 (1998)
[27] Wieneke, B.: Volume Self-Calibration for Stereo-PIV and Tomographic-PIV. In: 7th In-
ternational Symposium on Particle Image Velocimetry, Rome, Italy, September 11–14
(2007)
[28] Kähler, C.J.: Multiplane stereo PIV – Recording and evaluation methods. In: Proc. EU-
ROMECH 411, Rouen, France, May 29–31 (2000)
[29] Kähler, C.J.: Investigation of the spatio-temporal flow structure in the buffer region of a
turbulent boundary layer by means of multiplane stereo PIV. Exp. Fluids 36, 114–130
(2004)
[30] Urushihara, T., Meinhart, C.D., Adrian, R.J.: Investigation of the logarithmic layer in
pipe flow using particle image velocimetry. In: Near-Wall Turbulent Flows, pp. 336–433.
Elsevier, Amsterdam (1993)
[31] Dieterle, L., Weichert, R.: Particle Image Velocimetry applied to microstructures in tur-
bulent flows. In: Rodi, Bergels (eds.) Eng. Turb. Mod. and Exp., vol. 3, pp. 391–400. El-
sevier, Amsterdam (1996)
[32] Lindken, R., Di Silvestro, F., Westerweel, J., Nieuwstadt, F.T.M.: Turbulence measure-
ments with ?-PIV in large-scale pipe flows. In: Proc. of the 11th Int. Symp. on Appl. of
Laser Techn. to Fluid Mech., Lisbon, Portugal, July 8–11, paper 12-1 (2002)
[33] Kähler, C.J., Scholz, U., Ortmanns, J.: Wall-shear-stress and near-wall turbulence meas-
urements up to single pixel resolution by means of long-distance micro-PIV. Exp. Flu-
ids 41, 327–341 (2006)
[34] Kähler, C.J., Scholz, U.: Transonic jet analysis using long-distance Micro-PIV. In: 12th
International Symposium on Flow Visualization – ISFV 12, Göttingen, Germany, Sep-
tember 10–14 (2006)
[35] Meinhart, C.D., Wereley, S.T., Santiago, J.G.: PIV algorithm for estimating timeaveraged
velocity fields. J. Fluids Eng. 122, 285–289 (1999)
[36] Westerweel, J., Geelhoed, P., Lindken, R.: Single-pixel resolution ensemble correlation
for micro-PIV applications. Exp. Fluids 37, 375–384 (2004)
[37] Billy, F., David, L., Pineau, G.: Single pixel resolution correlation applied to unsteady
flow measurements. Meas. Sci. Technol. 15, 1039–1045 (2004)
[38] Elsinga, G.E., Scarano, F., Wieneke, B., van Oudheusden, B.W.: Tomographic particle
image velocimetry. Exp. Fluids 41, 933–947 (2006)
[39] Hain, R., Kähler, C.J., Michaelis, D.: Tomographic and time resolved PIV measurements
on a finite cylinder mounted on a flat plate. Exp. Fluids 45, 715–724 (2008)
[40] Meier, G.E.A.: Hintergrundschlierenverfahren. Deutsche Patentanmeldung, DE 199 42
856 A1 (1999)
[41] Meier, G.E.A.: Computerized background-oriented schlieren. Exp. Fluids 3, 181 (2002)
[42] Richard, H., Raffel, M.: Principle and applications of the background oriented schlieren
(BOS) method. Meas. Sci. Technol. 12, 1576–1585 (2001)
Principle and Perspective of Optical Multipoint Methods 103
[43] Kähler, C.J., Scholz, U.: Investigation of laser-induced flow structures with timeresolved
PIV, BOS and IR technology. In: 5th International Symposium on Particle Image Ve-
locimety, Busan, Korea, September 22–24 (2003)
[44] Kähler, C.J., Dreyer, M.: Dynamic 3D stereoscopic PIV and schlieren investigation of
turbulent flow structures generated by laser induced plasma. In: 12th International Sym-
posium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, July
12–15 (2004)
[45] Ruhnau, P., Kohlberger, T., Nobach, H., Schnorr, C.: Variational optical flow estimation
for particle image velocimetry. Exp. Fluids 38, 21–32 (2005)
[46] Atcheson, B., Ihrke, I., Bradley, D., Heidrich, W., Magnor, M., Seidel, H.-P.: Imaging
and 3D tomographic reconstruction of time-varying, inhomogeneous refractive index
fields. In: Int. Conf. on Computer Graphics and Interactive Techniques, San Diego, Cali-
fornia, USA, Article No. 32 (2007)
[47] Venkatakrishnan, L., Meier, G.E.A.: Density measurements using the background ori-
ented schlieren technique. Exp. Fluids 37, 237–247 (2004)
[48] Goldhahn, E., Seume, J.: The background oriented schlieren technique: sensitivity, accu-
racy, resolution and application to a three-dimensional density field. Exp. Fluids 43, 241–
249 (2007)
[49] Scholz, U., Kähler, C.J.: PIV analysis of flapping wing aerodynamics. In: 13th Interna-
tional Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Por-
tugal, June 26-29 (2006)
[50] Ehrenfried, K.: Processing calibration-grid images using Hough transformation. Meas.
Sci. Tech. 13, 975–983 (2002)
[51] Hurst, D., Frahnert, H., Schinkel, R., Quix, H., Sant, Y.L.: Benchmark testing of the
model deformation measurement systems developed within the European windtunnel as-
sociation (EWA). In: Conference Proceedings, 1st CEAS European Air and Space Con-
ference, Berlin, September 10–13 (2007)
[52] Michaelis, D., Frahnert, H., Stasicki, B.: Accuracy of combined 3d surface deformation
measurement and 3D position tracking in a wind tunnel. In: Proc. of 12th International
Conference on Experimental Mechanics (ICEM 12), Politecnico di Bari (Italy), August
29 to September 2 (2004)
[53] Bansmer, S., Kähler, C., Radespiel, R., Unger, R., Haupt, M., Horst, P.: Flow field meas-
urements on a flexible airfoil for flapping wing propulsion. In: 13th International Sympo-
sium on Flow Visualization, Nice, France, July 1–4 (2008)
[54] Konrath, R., Klinge, F., Schröder, A., Kompenhans, J., Füllekrug, U.: The projected pat-
tern correlation technique for vibration measurements. In: Proc. of 6th Int. Conf. on Vi-
bration Measurements by Laser Techniques: Advances and Applications, A.I.VE.LA.,
Ancona, June 22–25 (2004)
[55] Estorf, M., Wolf, T., Radespiel, R.: Experimental and numerical investigations on the op-
eration of the Hypersonic Ludwieg Tube Braunschweig. In: 5th European Symposium on
Aerothermodynamics for Space Vehicles (2004)
[56] Estorf, M.: Image based heating rate calculation from thermographic data considering lat-
eral heat conduction. Heat and Mass Transfer 49, 2545–2556 (2006)
[57] Liu, T., Sullivan, J.P.: Pressure and Temperature Sensitive Paints. Experimental Fluid
Mechanics. Springer, Heidelberg (2004)
[58] Kautsky, H., Hirsch, H.: Detection of minutest amounts of oxygen by extinction of phos-
phorescence. Zeitschrift für Anorg, 216–222 (1935)
[59] Stern, V.O., Volmer, M.: Uber die Abklingzeit der Fluoreszenz. Phys. Z. 20, 183 (1919)
104 C.J. Kähler
[60] Engler, R., Klein, C., Trings, O.: Pressure sensitive paint systems for pressure distribution
measurements in wind tunnels and turbumachines. Meas. Sci. Tech. 11, 1077–1085
(2000)
[61] Klein, C., Engler, R., Henne, U., Sachs, W.: Application of pressure sensitive paint sys-
tems for determination of pressure field and calculation of forces and moments of models
in a wind tunnel. Exp. Fluids 39, 1114–1432 (2005)
[62] Peterson, J.I., Fitzgerald, V.F.: New Technique of Surface Flow Visualization Based on
Oxygen Quenching of Fluorescence. Review of Scientific Instruments 51(5), 133–136
(1980)
[63] Morris, M.J., Donovan, J.K., Kegelmann, J.T., Schwab, S.D., Levy, R.L., Crites, R.C.:
Aerodynamic Applications of Pressure Sensitive Paint. AIAA Journal 31(3), 419–421
(1993)
[64] McLachlan, B.G., Bell, J.H.: Pressure Sensitive Paint in Aerodynamic Testing. Exp.
Thermal and Fluid Science 10(5), 470–483 (1995)
[65] Sellers, M.E.: Application of pressure sensitive paint for determining aerodynamic loads
on a scale of a F-16C. In: 21st AIAA Aerodynamic Measurement Technology and
Ground Testing Conf., paper AIAA, pp. 2000–2582 (2000)
[66] McLachlan, B.G., et al.: Pressure-Sensitive Paint Measurements on a Supersonic High-
Sweep Oblique Wing Model. Journal of Aircraft 32(2), 217–225 (1995)
[67] Liu, T., Torgerson, S., Sullivan, J.P., Johnston, R., Fleeter, S.: Rotor blade pressure
measurements in a high speed axial compressor using presure and temperature sensitive
paint. AIAA Paper No 97-0162 (1997)
[68] Asai, K., Kanda, H., Sullivan, J.P.: Boundary layer transition detection in a cryogenic
wind tunnel using luminescent paint. J. Aircraft 34, 34–42 (1997)
[69] Kameda, M., Tabei, T., Nakakita, K., Sakaue, H., Asai, K.: Image measurements of un-
steady pressure fluctuations by a pressure-sensitive coating on porous anodized alumi-
num. Meas. Sci. Technol. 16, 2517–2524 (2005)
[70] Sakamura, Y., Matsumoto, M., Suzuki, T.: High frame-rate imaging of surface pressure-
distribution using a porous pressure-sensitive paint. Meas. Sci. Technol. 16, 759–765
(2005)
[71] Kompenhans, J.: Experimental methods for multi-diagnostics of flow fields in wind tun-
nels. J. of Visualization 10(1), 25–28 (2007)
The Second International Vortex Flow
Experiment (VFE-2)
D. Hummel
Summary
In order to understand the vortical flow around delta wings especially with rounded
leading edges and to collect new flow field data for comparison with numerical
results, the Second International Vortex Flow Experiment (VFE-2) has been carried
out in 2003 to 2008 within the framework of an RTO Task Group. The tested
configuration was a flat plate 65° swept delta wing with interchangeable sharp and
rounded leading edges. Five different models were tested in various wind tunnels
worldwide accompanied by related numerical investigations. The program of work
and some of the main experimental and numerical results according to the status in
autumn 2007 are presented in this paper, and an outlook concerning future
investigations on this configuration is given.
1 Introduction
The First International Vortex Flow Experiment (VFE-1) [1, 2] had been carried out
in 1984 – 1986 on a cropped 65° delta wing-fuselage combination in order to validate
the various Euler codes. It turned out that these were not well suited for the
calculation of the pressure distribution on slender sharp edged wings, since the
secondary separation is not modelled at all.
In the last fifteen years considerable progress has been achieved in the numerical
calculation of vortical flows by taking into account viscous effects through solutions
of the RANS equations. This means that Reynolds number effects are now included
and secondary vortices turn out. However, the pressure distribution on the upper
surface of the wing is very sensitive to the formation of the viscous regions of the
flow field, especially the boundary layers and the secondary vortices [3]. For
turbulent flows in solutions of the RANS equations a turbulence model is necessary,
which has to cover both the attached boundary layers and the secondary vortex area
properly. In order to validate the results of Navier-Stokes calculations new and more
detailed experimental data are necessary, and therefore a Second International Vortex
Flow Experiment (VFE-2) has been proposed in [4]. Since up to now most
investigations of this kind were carried out for sharp edged wings, VFE-2 has been
designed to concentrate mainly on wings with rounded leading edges.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 105–129.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
106 D. Hummel
2 Test Configuration
The configuration for VFE-2 has been chosen in such a way that all flow regimes (i.e
attached flow, separated vortical flow without and with vortex breakdown and
separated deadwater-type flow) are covered properly, and this leads to a delta wing
with a leading edge sweep of 65°. Concerning the thickness distribution a flat plate
inner portion in combination with interchangeable leading edges was desired, and
these requirements were fulfilled by the NASA configuration [5], which is shown in
Figures 1 and 2. Sets of one sharp and three rounded leading edges were available.
cR = 25.734 in
b/2 = 12 in
Λ = 65°
d = 3.30 in
t/cR = 0.0340
b/2cR = 0.4663
d/b = 0.1375
Streamwise leading-edge radii:
rLE/ c = 0, 0.0005, 0.0015, 0.0030
Fig. 1. VFE-2 configuration: NASA NTF delta wing Λ = 65°, A = 1.85 according to [5]
Fig. 2. 65° delta wing (A = 1.85) at cryogenic tests in the National Transonic Facility (NTF),
NASA Langley Research Centre [5] (By courtesy of J.M. Luckring)
The Second International Vortex Flow Experiment (VFE-2) 107
The geometry of the wing as well as the shape of the sting is given by analytical
expressions described in all details in [5]. New wind tunnel models could be built
quite easily and overall aerodynamic coefficients and pressure distributions in certain
cross sections were already available for a large variety of Reynolds numbers and
Mach numbers [5–9].
For delta wings with sharp leading edges and fixed primary separation a large num-
ber of experimental investigations are available in the literature. Therefore it was
decided to use the case of sharp leading edges within VFE-2 as reference only and
to direct the main emphasis for new experiments towards studies of the vortex for-
mation on the configuration with rounded leading edges. In the first place drag data
should be added to the existing balance measurements [5] in order to provide the
full three-component data set. Since no investigations on the flow field around this
configuration existed at all, the main general aim for VFE-2 was to provide flow
field data for comparison with numerical results. The objectives for new experi-
ments were:
• Investigations on the laminar/turbulent transition on delta wings.
• Detailed pressure distribution measurements, especially in the region of the onset
of flow separations for configurations with rounded leading edges.
• Flow field measurements in the primary and secondary vortices as well as within
the boundary layers including the distributions of the components of velocity and
vorticity, of turbulent energy and eddy viscosity.
• Investigations on the vortex breakdown flow field for delta wings with sharp and
rounded leading edges including the surface pressure fluctuations caused by the
spiral mode of vortex breakdown.
From the very beginning of VFE-2 the experimental investigations have been
accompanied by CFD calculations for the relatively simple delta wing configu-ration
including the sting. The objectives for these investigations were:
• Validation and improvement of the existing CFD codes by means of
comparisons with new experimental results.
• Code to code comparisons through calculations on common unstructured and
structured grids.
• Application of different turbulence models in RANS calculations and
comparison with new experimental flow field data.
• Assistance related to the set up, the performance and the evaluation of the new
wind tunnel experiments.
• Synergistic effects related to calculations for a complete aircraft through the
possibility for test runs on a simple configuration prior to expensive calculations
for the detailed aircraft geometry.
108 D. Hummel
Table 1. Wind Tunnel Models used in VFE-2. [Notations for leading edge shapes: (S) Sharp
edged, (RS) Rounded: Small radius, (RM) Rounded: Medium radius, (RL) Rounded: Large
radius]
Results of the PSP investigations on model 1 obtained by the PSP team (R. Engler,
Ch. Klein, R. Konrath) of DLR Goettingen, Germany, according to [3, 12, 13] are
110 D. Hummel
~c
p
−2.8
Fig. 3. Experimental pressure distribution on the 65° delta wing (A = 1.85) with rounded
leading edges (medium radius) for M = 0.4, Rmac = 3·106, α = 13°. PSI and PSP results from
DLR Goettingen (cp pressure coefficients without offset correction).
shown in Figure 3. For the configuration with medium radius rounded leading edges
at α = 13° the flow is attached in the front part. In the rear part, however, a vortical
flow is established, which is associated with two suction areas on the wing. The
strong outer one is easily identified to be due to a leading edge vortex, but at first
glance the weaker inner one seemes to be associated with a bubble-type flow
separation in the front part of the wing.
Numerical calculations have been carried out at EADS Munich (W. Fritz),
Germany, for the same case using the code FLOWER and the k-ω turbulence model.
The result according to Figure 4 showed two co-rotating vortices on each side of the
configuration with corresponding suction peaks. In spite of some differences
regarding the position of the onset of vortical flow, the two-vortex configuration turns
out very well. Surprisingly the inner vortex was as large as the outer one as shown in
Figure 5. The outer vortex is fed by vorticity up to the trailing edge, whereas the inner
vortex decays downstream more and more. The numerical results came up just after
the PSP measurements at DLR Goettingen [13] and they have been used as a
guideline for the optical setup for the subsequent investigations of the PIV team (A.
Schroeder, J. Kompenhans, R. Konrath) of DLR Goettingen, Germany, [14, 15]. The
results according to Figure 6 show excellent agreement with the numerical findings,
[16, 17].
These combined experimental and numerical results lead to a certain understanding
of the flow around the medium radius rounded leading edge VFE-2 configuration at α
= 13°: The first flow separation takes place in the front part near the apex, where the
wing is relatively thick. At its very beginning the separated flow region is located
close to the body surface, but further downstream a concentrated inner primary vortex
is formed quite rapidly. Along the leading edge the curvature radius of the wing
The Second International Vortex Flow Experiment (VFE-2) 111
Fig. 4. Numerical pressure distribution on the 65° delta wing (A = 1.85) with rounded leading
edges (medium radius) for M = 0.4, Rmac = 3·106, α = 13°. Results from EADS Munich by
means of the FLOWER code and the k-ω turbulence model.
Fig. 5. Numerical pressure distribution and vortex pattern, described by total pressure loss
contours, in the cross section at x/c = 0.75 for the the 65° delta wing (A = 1.85) with rounded
leading edges (medium radius) for M = 0.4, Rmac = 3·106, α = 13°. Results from EADS Munich
by means of the FLOWER code and the k-ω turbulence model.
112 D. Hummel
Fig. 6. Pressure (surface color), velocity (vectors), and vorticity (vector color) distributions
above the 65° delta wing (A=1.85) with rounded leading edges (medium radius) for M = 0.4,
Rmac = 3·106, α = 13°. Comparison of the PSP and PIV measurements at DLR Goettingen with
the numerical solution of EADS Munich.
remains constant, whereas the local half span increases downstream. This means in
other words that the leading edge becomes sharper towards the trailing edge of the
wing. Correspondingly the suction at the leading edge increases downstream, and
finally new flow separations take place in the rear part of the wing, forming an outer
primary vortex there. In the region of the onset of this outer primary vortex strong
interference with the already existing inner primary vortex takes place. The dominant
part of vorticity, shed from the leading edge, is now fed into the outer primary vortex,
whereas the feeding of the inner vortex with vorticity is reduced. Therefore the inner
primary vortex decays downstream due to dominating viscous effects. Unfortunately
within VFE-2 these details could not be investigated experimentally because of geo-
metric constraints of the PIV setup in the wind tunnel. Therefore these details will be
further studied through an analysis of numerical solutions.
The flow pattern described so far depends on the angle of attack. From previous stud-
ies [9] the progression of leading-edge flow separation of the outer primary vortex
with angle of attack was established as (i) a low angle-of-attack range where the flow
was attached, (ii) an intermediate angle-of-attack range where the primary separation
progressed up the leading edge, from trailing edge to apex, with increasing angle of
attack, and (iii) a high angle-of-attack range where essentially the entire leading edge
exhibited primary vortex separation.
A series of pressure distributions for M = 0.4 and Rmac = 3 million and for various
angles of attack according to the PSP measurements [13, 16] at DLR Goettingen is
shown in Figure 7. Up to α = 11.2° only the inner primary vortex exists, and the cor-
responding suction on the wing surface reaches considerable values near the trailing
The Second International Vortex Flow Experiment (VFE-2) 113
Fig. 7. Pressure distributions on the VFE-2 configuration with rounded leading edges (medium
radius) for M = 0.4, Rmac = 3·106 at various angles of attack from the PSP experiments at DLR
Göttingen [13, 16]
edge. With further increasing angle of attack the outer primary vortex is formed in the
rear part of the configuration, and already at α = 12.2° its onset has reached a position
at x/cR = 0.6. The strength of the inner primary vortex increases up to the region of the
onset of the outer primary vortex, but then decreases suddenly downstream towards
the trailing edge. This is due to the fact that the vorticity shed from the leading edge is
now fed into the outer primary vortex, and this leads to the considerable reduction of
the strength of the inner primary vortex. Another effect can also be recognized from
Figure 7: In that region where an outer primary vortex already exists, the weakened
inner primary vortex moves inboard. For further increasing angles of attack the onset
of the outer primary vortex moves upstream, see α = 13.3° and α = 15.3°. At α =
20.5° the outer primary vortex covers almost the whole leading edge, but some weak
remains of an inner primary vortex can still be detected.
114 D. Hummel
The vortex formation strongly depends on the Reynolds number as shown in Figure 8
and Figure 9. With decreasing Reynolds number the onset of the outer primary vortex
moves upstream and its strength increases, whereas the inner primary vortex is weak-
ened and its position moves distinctly inboard.
Fig. 8. Pressure distributions on the VFE-2 configuration with rounded leading edges (medium
radius) for M = 0.4, α = 13° at different Reynolds numbers from the PSP experiments at DLR
Göttingen [13, 16]
Fig. 9. Pressure distributions and flow fields on the VFE-2 configuration with rounded leading
edges (medium radius) for M = 0.4, α = 13° at different Reynolds numbers from the PSP
experiments at DLR Göttingen [16]
The Second International Vortex Flow Experiment (VFE-2) 115
In contrasting a subsonic and a transonic Mach number condition, the principal vortex
topology remains unchanged, but some compressibility effects can be recognized. In
transonic flow the formation of the outer primary vortex starts earlier [13]. Figure 10
shows the pressure distributions at α = 13° and Rmac = 2·106 for two different Mach
numbers. At the higher Mach number the onset of the outer primary vortex has moved
slightly upstream and its axis is distinctly shifted inboard. For M = 0.8 an inner pri-
mary vortex has no longer been found. If it exists at all in transonic flow, this vortex
must be very weak. The corresponding effect of Mach number on the flow field may
be taken from Figure 11. The outer primary vortex is shifted inboard and its flow field
is enlarged.
The 3D vortical flow field has been analyzed at TU Munich by means of hot-wire
measurements, in which large portions of the boundary layers were covered [20]. A
typical result for incompressible free stream conditions (M = 0.07, Rmac = 1·106) is
Fig. 10. Pressure distributions on the VFE-2 configuration with rounded leading edges
(medium radius) at Rmac = 2·106 and α = 13° for different Mach numbers from the PSP
experiments at DLR Göttingen [13]. The black line indicates the sonic pressure coefficient.
116 D. Hummel
M = 0.4 M = 0.8
Fig. 11. Flow field on the VFE-2 configuration with rounded leading edges (medium radius) at
Rmac = 2·106 and α = 13° for different Mach numbers from the PIV experiments at DLR
Göttingen [14]. Velocity vectors and axial vorticity in the cross section at x/c = 0.8.
shown in Figure 12. The inner measurement station at η = 2y/bloc = 0.4 is located
inboard of the outer primary vortex attachment line and it is thus related to the inner
primary vortex. At this station the boundary layer is very thick (2δ/bloc = 0.025) as
indicated by the time-averaged longitudinal velocity component u / U ∞ . Outside of
the boundary layer the fluctuations of almost all velocity components increase to-
wards the wing with the only exception of a slight decrease in w I 2 U ∞ , and within
asurements
Fluctuation intensity level
Re = 1.0 x 106
α = 13°
Fluctuation intensity level
Fig. 12. Results of boundary layer measurements on the VFE-2 configuration with rounded
leading edges (medium radius) for M = 0.07, Rmac = 1·106 and α = 13° at two stations (top η =
0.4, bottom η = 0.6) in the section x/cr = 0.6 according to HWA at TU Munich [37]
The Second International Vortex Flow Experiment (VFE-2) 117
surface
Surface pressure
pressu ons
fluctuations
re fluctuati
Re = 2.0 x 106
α = 13°
Fig. 13. Pressure fluctuations on the VFE-2 configuration with rounded leading edges (medium
radius) for M = 0.14, Rmac = 2·106 and α = 13° in four sections according to measurements at
TU Munich [37]
the boundary layer all fluctuations increase rapidly towards the wall. The same ap-
plies for the outer measurement station at η = 2y/bloc = 0.6, but the boundary layer is
much thinner there (2δ/bloc = 0.008). The outer measurement station is located out-
board of the outer primary vortex attachment line. Thus an accelerated outboard flow
underneath the outer primary vortex is present, which leads to the strong decrease of
the boundary layer thickness. Simultaneously measured pressure fluctuations on the
wing surface according to Figure 13 indicate, that the flow in the region of the outer
primary vortex is distinctly turbulent. At the inner measurement station, however, the
boundary layer status remains unclear: At the wall almost no pressure fluctuations
were found, whereas in the boundary layer some fluctuations are present.
Fig. 14. Pressure distributions on the VFE-2 configuration with sharp and rounded leading
edges (medium radius) at Rmac = 2·106 and α = 18° for two Mach numbers according to the PSP
experiments at DLR Göttingen [13]. The black line indicates the sonic pressure coefficient.
for both Mach numbers indicate that the primary vortex on the configuration with
blunt leading edge is located slightly more outboard than for the configuration with
sharp leading edge, and some differences concerning the suction magnitude can also
be recognized, but the overall behavior of the flow is very similar.
Fig. 15. Flow field around the VFE-2 configuration with sharp leading edges at M ≈ 0.1, Rmac
= 1·106, α = 18°. PIV results for the time-averaged flow field (left) and HWA results for the
axial velocity fluctuations at x/c = 0.6 (right) according to measurements at TU Munich [20].
The Second International Vortex Flow Experiment (VFE-2) 119
The vortical flow field with a well developed single primary vortex on each side of
the configuration has been investigated at TU Munich [18–20] for sharp and blunt
leading edges by means of PIV at α = 18°. Results for the time-averaged flow field
around the sharp edged configuration are shown in the left-hand part of Figure 15.
Within VFE-2 the case α = 18° will be used to check the various turbulence models
for their proper description of a vortical flow. For this purpose HWA has also been
applied at TU Munich [19, 20]. Concerning the time averaged velocity components
the results according to Figure 15 have been confirmed. In addition, however, the
complete field of the velocity fluctuations u', v', w' is now available. As an example
the right-hand part of Figure 15 shows the measured fluctuations of the velocity com-
ponent urms/U∞ in the cross section plane at x/c = 0.6. Again there are only minor
differences between the measurements for sharp and for medium radius rounded lead-
ing edges, but from the full field of the fluctuations u', v', w' the distributions of the
turbulent kinetic energy and the eddy viscosity can now be determined. Correspond-
ing evaluations of the experiments are presently under way, and the results will be
used for comparisons with numerical simulations of the vortical flow field by means
of various turbulence models.
Fig. 16. Flow field around the VFE-2 configuration with sharp leading edges at M ≈ 0.1, Rmac
= 1·106, α = 18°. Power Spectrum Density (PSD)u for the u-component of the velocity as
function of the reduced frequency k at x/c = 0.6, η = 0.75, ζ ≈ 0.2 according to HWA at TU
Munich [20]. (∆ denotes contour interval)
The unsteadiness of the flow can also be evaluated from these experiments by de-
termining the power spectrum density distribution of any fluctuating flow quantity as
function of the frequency at any measured position of the flow field. An example of
this kind is shown in Figure 16 for the sharp leading edge configuration. In the left-
hand graph the distribution of the fluctuations of the u-component of the velocity
urms/U∞ according to Figure 15 (right) is given, and that location near the vortex centre
is indicated, for which the power spectrum density distribution for the velocity com-
ponent u/U∞ is shown on the right-hand side. In this turbulent flow near the vortex
centre a large variety of frequencies is present with distinctly high values at low
frequencies.
120 D. Hummel
Fig. 17. Flow field around the VFE-2 configuration with rounded leading edges (medium ra-
dius) at M = 0.14, Rmac = 2·106, α = 23°. Surface pressure fluctuations (cp)rms and spectrum
density (SD)cp for certain stations in four cross sections according to measurements at TU
Munich [20].
For the VFE-2 configuration at α = 23° vortex breakdown in the primary vortex
has already progressed over the wing for sharp and for medium radius rounded lead-
ing edges. For both configurations the unsteady vortical flow field has been measured
The Second International Vortex Flow Experiment (VFE-2) 121
urms/U
Fig. 18. Flow field around the VFE-2 configuration with rounded leading edges (medium ra-
dius) at M ≈ 0.1, Rmac = 1·106, α = 23°. PIV results for the time-averaged flow field (left) and
HWA results for the axial velocity fluctuations at x/c = 0.8 (right) according to experiments at
TU Munich [20].
Fig. 19. Pressure distributions on the VFE-2 configuration with sharp and rounded leading
edges (medium radius) at Rmac = 2·106 and α = 25° for two Mach numbers according to the PSP
experiments at DLR Göttingen [13]. The black line indicates the sonic pressure coefficient.
and Fig 18, although the precise breakdown location is in general difficult to discern
from subcritical surface pressure information.
At constant angle of attack and with the increase to a transonic Mach number,
however, a distinct change of the pressure distribution on the wing can be taken from
the results for M = 0.8 (right-hand part). At this transonic Mach number a local super-
sonic zone with a terminating shock wave is formed in the vicinity of the sting mount.
The adverse pressure jump in the shock wave causes additional unsteadiness in the
vortical flow field, and most likely shifts vortex breakdown upstream.
Table 2. Summary of calculated VFE-2 CFD cases [Notations: (SLE) Sharp LE, (RLE)
Medium radius rounded LE, ● done, ○ at work (autumn 2007)]
Uni.
Case α Rmac KTH EADS TAI USAF NLR DLR
LE M Glasgow
No. (°) (Mio.) SWE GER TUR A USA NLD GER
GBR
1 13 RLE 0.2 1 ● ● ○
2 13 RLE 0.4 2 ● ● ○ ○ ○
3 13 RLE 0.4 3 ● ● ● ○ ●
4 13 RLE 0.4 6 ● ● ● ○ ●
5 13 SLE 0.4 6 ● ● ○ ●
6 18 RLE 0.4 2 ● ● ○ ○ ○
7 18 RLE 0.4 6 ● ● ● ○ ●
8 18 SLE 0.4 6 ● ● ○ ○ ● ●
9 23 RLE 0.4 2 ● ● ○ ○ ○
10 23 RLE 0.4 6 ● ● ● ○ ● ○
11 23 SLE 0.4 6 ● ● ● ○ ●
12 23 SLE 0.85 6 ● ● ○ ● ●
In the solutions of the RANS equations a large variety of existing turbulence mod-
els has been applied. For (medium radius) rounded leading edges the flow field at
lower angles of attack (α = 13°) with two vortices on each side of the configuration
has been considered. The ordinary vortex formation with one vortex on each side of
the configuration, accompanied by a secondary vortex, (α = 18°) has also been studied
numerically for sharp and rounded leading edges, and for very high angles of attack
(α = 23°) the unsteady behaviour of the vortical flow (vortex breakdown) has also
been treated. The main flow cases to be calculated within VFE-2 are listed in Table 2.
Finally a semi empirical/numerical method to predict the onset of vortical flow
over the upper surface of a delta wing with rounded leading edges with different radii
has been developed.
8.1 Numerical Results for Fully Developed Vortical Flows without Vortex
Breakdown (α = 18°)
For the calculation of the flow around a delta wing the fully developed vortical flow
without vortex breakdown is the simplest case. For this reason α = 18° has been cho-
sen for the VFE-2 configuration to be the standard case for sharp and medium radius
rounded leading edges. All numerical codes applied on structured and unstructured
grids were able to calculate this flow, and various turbulence models have been ap-
plied in these calculations. Figure 20 shows the EADS solution for the medium radius
rounded leading edge. Since the experimental results from HWA at TU Munich came
up only very recently, corresponding numerical studies with different turbulence
models have still to be carried out in order to find the best turbulence model to be
applied in calculations of vortical flows. Unfortunately the HWA investigations have
been carried out for incompressible flow at very low Reynolds numbers, and this
leads to serious convergence problems in the numerical calculations. Therefore fur-
ther studies of this topic are necessary, and they will be carried out in the forthcoming
phase of VFE-2.
124 D. Hummel
Fig. 20. Numerical solution for the vortical flow around the VFE-2 configuration with medium
radius rounded leading edge for M = 0.4, Rmac = 3·106, α = 18° according to EADS Munich,
Germany, calculated on a structured grid using the k-ω turbulence model. Distribution of the
pressure coefficient cp on the wing surface (left) and iso-surface of 1% total pressure loss in the
flow field (right).
8.2 Numerical Results for Fully Developed Vortical Flows with Vortex
Breakdown (α = 23°)
At very large angles of attack the spiral mode of vortex breakdown takes place within
the vortices over the wing. Due to the unsteadiness of the flow field time-accurate
CFD codes have to be applied. In transonic flow, in the vicinity of the sting mount of
the VFE-2 configuration terminating shocks occur in the flow field, which lead to a
considerable upstream shift of the vortex breakdown onset, see Figure 19. The time-
averaged EADS solution for this kind of flow is shown in Figure 21. From the very
beginning of VFE-2 this phenomenon has been studied numerically, in the first place
on structured grids [24], but later other members of the VFE-2 team joined with solu-
tions on unstructured grids [25, 26]. The experimentally detected effects of the shock
waves on the vortex breakdown location in the flow field could be predicted nicely by
the numerical methods, and these results were summarized in a thesis at University of
Glasgow [27].
8.3 Numerical Results for the Partly Developed Vortical Flow Around
the Configuration with Blunt Leading Edge (α = 13°)
The numerical simulation for this kind of flow turned out to be the most difficult one.
The first successful treatment of this problem within VFE-2 on a structured grid has
been shown in Figures 4 to 6, and according to the EADS solution this flow is again
shown in Figure 22.
In the meantime also solutions for co-rotating vortices on unstructured grids are
available as shown in Figure 23 for a very high Reynolds number. In this kind of
solutions serious convergence problems have to be overcome, especially for low
The Second International Vortex Flow Experiment (VFE-2) 125
Fig. 21. Numerical solution for the vortical flow around the VFE-2 configuration with medium
radius rounded leading edge for M = 0.85, Rmac = 3·106, α = 23° according to EADS Munich, Ger-
many, calculated on a structured grid using the k-ω turbulence model. Distribution of the pressure
coefficient cp on the wing surface (left) and streamline visualization of the flow field (right).
Fig. 22. Numerical solution for the vortical flow around the VFE-2 configuration with medium
radius rounded leading edge for M = 0.4, Rmac = 3·106, α = 13° according to EADS Munich,
Germany, calculated on a structured grid using the k-ω turbulence model. Distribution of the
pressure coefficient cp on the wing surface (left) and iso-surface of 1% total pressure loss in the
flow field (right).
Reynolds numbers and low Mach numbers, and therefore improvements on this sub-
ject are necessary.
Due to the lack of experimental information about the onset of the inner primary
vortex and the interference between the two vortices in the onset region of the outer
primary vortex, reliable numerical results will play an important role in the under-
standing of the flow physics. Figure 24 shows latest [autumn 2007] calculated results
on a structured grid by EADS Munich (left) and on an unstructured grid by DLR
Braunschweig (right). It is expected that the detailed analysis of this kind of numeri-
cal solutions will lead to the proper understanding of the flow structure.
126 D. Hummel
0.90
0.85
0.80
0.75
0.70
0.65
0.60
x/cr
flow
Fig. 23. Numerical pressure distribution and vortex pattern described by axial vorticity contours
in various cross sections x/cr = const. for the VFE-2 configuration with medium radius rounded
leading edges for M = 0.4, Rmac = 60·106, α = 13°. Results from KTH Stockholm obtained on an
unstructured grid [28].
Fig. 24. Numerical solutions for the flow around the VFE-2 configuration with medium radius
rounded leading edges at M = 0.4, Rmac = 3·106, α = 13.3°. EADS solution [41] on a structured
grid with pressure distribution and streamlines in the flow field (left) and DLR solution [46] on
an unstructured grid with pressure distribution and upper surface friction lines (right).
9 Outlook
Up to now the Second International Vortex Flow Experiment (VFE-2) has been car-
ried out within the framework of an RTO Task Group, in which specialists in CFD
and experimental techniques worked closely together. The status of VFE-2 has been
reported from time to time [3, 31, 32]. At the end of the RTO phase of VFE-2, main
results of this cooperation will be presented in two sessions at the 46th AIAA Aero-
space Sciences Meeting 2008, [33–47], and all results will be included in a Final
Scientific Report to be published by RTO in 2008.
The Second International Vortex Flow Experiment (VFE-2) will now enter a new,
open phase: Many existing experimental and numerical results will be improved in
the future, some inadequately treated problems should be considered anew, and even
new tasks can be dealt with. In order to encourage future investigations the Final
Scientific Report will provide some material: The VFE-2 configuration is very simple
The Second International Vortex Flow Experiment (VFE-2) 127
and described in analytical form. Therefore new wind tunnel models can be built quite
easily. The Final Scientific Report will contain some experimental data for use in
comparisons with future calculations, and in addition a structured and an unstructured
starting grid will be included in order to promote new calculations.
The scientific community is invited to join this program for future work.
References
[1] Drougge, G.: The international vortex flow experiment for computer code validation. In:
ICAS-Proceedings 1988, vol. 1, pp. XXXV–XLI (1988)
[2] Elsenaar, A., Hjelmberg, L., Bütefisch, K.-A., Bannink, W.J.: The international vortex
flow experiment. AGARD-CP 437, vol.1, pp. 9-1–9-23 (1988)
[3] Hummel, D.: Effects of boundary layer formation on the vortical flow above slender delta
wings. In: RTO AVT Symposium on Enhancement of NATO Military Flight Vehicle
Performance by Management of Interacting Boundary Layer Transition and Separation,
Prague, Czech Republic, October 4–8, 2004, Meeting Proceedings RTO-MP-AVT-111,
30-1–30-22 (2004)
[4] Hummel, D. and Redeker, G.: A new vortex flow experiment for computer code valida-
tion. In: RTO AVT Symposium on Vortex Flow and High Angle of Attack Aerodynam-
ics, Loen, Norway, May 7–11, 2001, Meeting Proceedings RTO-MP-069 (I), SYA 8-1–8-
31 (2003)
[5] Chu, J., Luckring, J.M.: Experimental surface pressure data obtained on 65o delta wing
across Reynolds number and Mach number ranges. NASA TM 4645 (1996)
[6] Luckring, J.M.: Reynolds number and leading-edge bluntness effects on a 65o delta wing.
AIAA Paper 2002-0419 (2002)
[7] Luckring, J.M.: Transonic Reynolds number and leading-edge bluntness effects on a 65o
delta wing. AIAA Paper 2003-0753 (2003)
[8] Luckring, J.M.: Compressibility and leading-edge bluntness effects for a 65o delta wing.
AIAA Paper 2004-0765 (2004)
[9] Luckring, J.M.: Reynolds number, compressibility, and leading-edge bluntness effects on
delta-wing aerodynamics. ICAS Proceedings, Paper 4.1.4 (2004)
[10] Lamar, J.E., Cronin, C.K., Scott, L.E.: A review of steps taken to create an international
virtual laboratory at NASA Langley for aerodynamic prediction and comparison. Pro-
gress in Aerospace Sciences 40, 163–172 (2004)
[11] Lamar, J.E., Cronin, C.K., Scott, L.E.: Virtual laboratory enabling collaborative research
in applied vehicle technologies. In: RTO AVT Symposium on Flow Induced Unsteady
Loads and the Impact on Military Applications, Budapest, Hungary, April 25–28, 2005,
Meeting Proceedings RTO-MP-AVT-123 (2005)
[12] Klein, C., Sachs, W., Henne, U., Engler, R.H., Wiedemann, A., Konrath, R.: Develop-
ment of PSP technique for application on the VFE-2 65o delta wing configuration in sub-
and transonic flow. AIAA Paper 2006-0059 (2006)
[13] Konrath, R., Klein, C., Engler, R.H., Otter, D.: Analysis of PSP results obtained for the
VFE-2 65o delta wing configuration at subsonic and transonic speed. AIAA Paper 2006-
0060 (2006)
[14] Konrath, R., Schroeder, A., Kompenhans, J.: Analysis of PIV results obtained for the
VFE-2 65o delta wing configuration at sub- and transonic speeds. AIAA Paper 2006-
3003 (2006)
128 D. Hummel
[15] Schroeder, A., Agocs, J., Frahnert, H., Otter, D. Mattner, H., Kompenhans, J. and Kon-
rath, R.: Application of Stereo PIV to the VFE-2 65o delta wing configuration at sub- and
transonic speeds. AIAA Paper 2006-3486 (2006)
[16] Konrath, R., Klein, Ch., Schroeder, A., Kompenhans, J.: Combined application of Pres-
sure Sensitive Paint and Particle Image Velocimetry to the flow above a delta wing. In:
12th International Symposium on Flow Visualization, September 10–14, 2006, DLR
Goettingen, Germany (2006)
[17] Sheplak, M.: The year in review: Aerodynamic measurement technology. Aerospace
America (December 2006)
[18] Furman, A., Breitsamter, Ch.: Delta wing steady pressure investigations for sharp and
rounded leading edges. Notes on Numerical Fluid Mechanics and Multidisciplinary De-
sign (NNFM), vol. 92, 77–84 (2006)
[19] Furman, A., Breitsamter, Ch.: Investigation of flow phenomena on generic delta wing. In:
ICAS Proceedings 2006 (2006)
[20] Furman, A., Breitsamter, Ch.: Stereo-PIV and Hot-Wire investigations on delta wing. In:
1st European Air and Space Conference on Council of European Aerospace Societies
(CEAS), Berlin, September 10–13 (2007)
[21] Visbal, M.R.: Computed unsteady structure of spiral vortex breakdown on delta wings.
AIAA Paper 96-2074 (1996)
[22] Müller, J., Hummel, D.: Numerical analysis of the unsteady flow above a slender delta
wing at large angles of attack. In: ICAS Proceedings Harrogate 2000, Paper ICA 0252
(2000)
[23] Pirzadeh, S.Z.: Vortical flow prediction using an adaptive unstructured grid method. In:
RTO AVT Symposium on Vortex Flow and High Angle of Attack Aerodynamics, Loen,
Norway, May 7–11, 2001, Meeting Proceedings RTO-MP-069 (I), SYA 13-1 to 13-36
(2003)
[24] Schiavetta, L.A., Boelens, O.J., Fritz, W.: Analysis of transonic flow on a slender delta
wing using CFD. AIAA Paper 2006-3171 (2006)
[25] Schiavetta, L.A., Badcock, K., Cummings, R.: Comparison of DES and URANS for un-
steady vortical flows over delta wings. AIAA Paper 2007-1085 (2007)
[26] Schiavetta, L.A., Boelens, O.J., Fritz, W., Cummings, R.M.: Sting effects on transonic
delta wing experiments. In: 3rd International Symposium on Integrating CFD and Ex-
periments in Aerodynamics, June 20–21, 2007, US Air ForceAcademy, CO, USA (2007)
[27] Schiavetta, L.: Evaluation of URANS and DES predictions of vortical flows over delta
wings. PhD thesis, Faculty of Engineering, Department of Aeronautical Engineering,
University of Glasgow (2007)
[28] Crippa, S., Rizzi, A.: Numerical investigation of Reynolds Number effects on a blunt
leading-edge delta wing. AIAA Paper 2006-3001 (2006)
[29] Crippa, S., Rizzi, A.: Initial steady/unsteady CFD analysis of vortex flow over the VFE-2
delta wing. In: ICAS Proceedings 2006 (2006)
[30] Crippa, S., Rizzi, A.: Reynolds number effects on blunt leading edge delta wings. In:
Council of European Aerospace Societies (CEAS), 1st European Air and Space Confer-
ence, Berlin, September 10–13 (2007)
[31] Hummel, D.: The Second International Vortex Flow Experiment (VFE-2): Objectives and
first results. J. Aerospace Engineering 220(6), 559–568 (2006)
[32] Hummel, D.: The Second International Vortex Flow Experiment (VFE-2): Objectives and
present status. AIAA Paper 2007-4446 (2007)
[33] Hummel, D.: Review of the Second International Vortex Flow Experiment (VFE-2).
AIAA Paper 2008-0377 (2008)
The Second International Vortex Flow Experiment (VFE-2) 129
[34] Luckring, J.M.: Initial experiments and analysis of vortex flow on blunt edged delta
wings. AIAA Paper 2008-0378 (2008)
[35] Konrath, R., Klein, Ch., Schroeder, A.: PSP and PIV investigations on the VFE-2 con-
figuration in sub- and transonic flow. AIAA Paper 2008-0379 (2008)
[36] LeRoy, J.F., Rodriguez, O., Kurun, S.: Experimental and CFD contribution to delta wing
vortical flow understanding. AIAA Paper 2008-0380 (2008)
[37] Furman, A., Breitsamter, Ch.: Turbulent and unsteady flow characteristics of delta wing
vortex systems. AIAA Paper 2008-0381 (2008)
[38] Coton, F., Mat, S., Galbraith, R.: Low speed wind tunnel characterization of the VFE-2
wing. AIAA Paper 2008-0382 (2008)
[39] Luckring, J.M., Hummel, D.: What was learned from the new VFE-2 experiments? AIAA
Paper 2008-0383 (2008)
[40] Nangia, R.K.: Semi-empirical prediction of vortex onset and progression on 65° delta
wings (RTO-AVT-113, VFE-2 facet). AIAA Paper 2008-0384 (2008)
[41] Fritz, W.: Numerical simulation of the peculiar subsonic flow-field about the VFE-2 delta
wing with rounded leading edge. AIAA Paper 2008-0393 (2008)
[42] Gürdamar, E., Ortakaya, Y., Kaya, S., Korkem, B.: Some factors influencing the vortical
flow structures on delta wings. AIAA Paper 2008-0394 (2008)
[43] Schiavetta, L.A., Boelens, O.J., Crippa, S., Cummings, R.M., Fritz, W., Badcock, K.J.:
Shock effects on delta wing vortex breakdown. AIAA Paper 2008-0395 (2008)
[44] Cummings, R.M., Schütte, A.: Detached-eddy simulation of the vortical flowfield about
the VFE-2 delta wing. AIAA Paper 2008-0396 (2008)
[45] Crippa, S., Rizzi, A.: Steady, subsonic CFD analysis of the VFE-2 configuration and
comparison to wind tunnel data. AIAA Paper 2008-0397 (2008)
[46] Schütte, A., Lüdeke, H.: Numerical investigations on the VFE 2 65-degree rounded lead-
ing edge delta wing using the unstructured DLR-TAU-Code. AIAA Paper 2008-0398
(2008)
[47] Fritz, W., Cummings, R.M.: What was learned from the numerical simulations for the
VFE-2? AIAA Paper 2008-0399 (2008)
Aerodynamic Wing Design
for Transport Aircraft – Today
Summary
Increasing fuel costs and environmental burden in particular with respect to the CO2
emissions require a substantial improvement of efficiency of transport aircraft. In the
past a multiplicity of improvements in aircraft development has been carried out. In a
historical overview the most important ones are briefly described. Different aerody-
namic technologies related to the clean wing are discussed in detail: swept wing, tran-
sonic adaptation, laminar technology, forward sweep, elasticity. Today some of these
technologies are fully exploited; others are well known but not yet applied. The paper
describes the current status of these technologies and identifies further potential.
1 Introduction
The Vision 2020 of the ACARE group asks for a reduction of fuel consumption by
50% until 2020 relative to the consumption in 2000. Seven years of the 20 are already
gone, but the realisation of an aircraft with advanced efficiency as requested is not in
sight. Several disciplines have to contribute to achieve the ACARE goal. In particular
three main shares are expected: The total drag of the aircraft has to be reduced in
the order of 15% to 20%, the weight of the aircraft has to be reduced by about 5% and
the efficiency of the engines has to be improved by 20% to 25%. The reduction of the
total drag has to occur largely by wing and empennage drag reduction. In the follow-
ing those technologies which in the past allowed big steps of improving efficiency
and others which probably could contribute to future improvements will be discussed.
The design of wings of transport aircraft for transonic speeds is affected by a multitude
of technologies being developed over the past ten decades. Some of the most important
will be discussed here briefly. In the last decade of the 19th century Lilienthal [1] ex-
perimentally found that cambered thin wings can produce higher lift than thin flat
plates. Lilienthal measured the lift of different test wings mounted on the tip of a pow-
ered rotating lever. All his different flight apparatuses were equipped with a thin cam-
bered wing. Sufficient structural strength was achieved by wire bracing the wing.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 130–144.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Aerodynamic Wing Design for Transport Aircraft – Today 131
Hugo Junkers [2] was the first who introduced the thick cantilevered wing with big
advantages for both aerodynamic and structural properties. From an aerodynamic
point of view the thick wing allows a larger range of angles of attack without sepa-
rated flow and associated moment variations. The load carrying structure could now
be placed inside the wing leading to a strong reduction of aerodynamic drag.
Some years later in 1919 Handley Page flew the first wing with a leading edge slat
in order to improve the low speed behaviour at take off and landing. G. Lachmann
first developed this technique in 1917 [3] and filed a patent in 1918 which was re-
jected by the German patent office. Another term of about 8 years had passed when
Fowler realized his famous extendable trailing edge flap with a gap between main
wing and flap. By means of these two devices the maximum lift coefficient could be
more than doubled enabling the design of a smaller wing with less drag in cruise
flight but unchanged take off and landing speeds.
A discovery shaping all modern high speed wings has been published 1935 by
Busemann [4]. For transonic and supersonic flow velocity he found that swept wings
allow higher flight Mach number than unswept wings of the same thickness ratio.
1939 Ludwieg [5] confirmed the approach of Busemann. The first transport aircraft
with a swept wing the Boeing XB47 flew in 1947 [6].
An important rule for supersonic design of aircraft but also for transonic speed is
well known as “area ruling”. In 1944 Frenzl [7] obtained a patent for speeds near
M=1 and 1955 Whitcomb [8] formulated his general area rule for transonic and super-
sonic speeds. For a smooth distribution of the cut surface of an aircraft configuration
with the Mach cone applied in flow direction the wave drag can be reduced.
A further important step on the way to transonic flight speed was the development
of airfoil shapes particularly suited for high speed. They are characterized by a large
local zone of supersonic flow speed followed by a slight shock nearly without wave
drag. First attempts are published in 1940 by Kawalki [9]. In the 60’s Whitcomb de-
signed his so-called supercritical airfoils. Improvements by Korn, Eberle, Redeker
and others lead to the well known modern transonic airfoils.
A drag reduction technique which to date could not be realized for transport air-
craft is the laminar flow technology. First test results from 1938 were published by
Doetsch [10] in 1940. At the same time the development of the famous NACA 6
series of NLF airfoils was started. Already in 1946 Pfenninger [11] published first test
results of airfoils with laminar boundary layer suction enabling laminar flow over the
whole airfoil surface. These early trials were performed at low and medium Reynolds
numbers between 1 and 10 Millions. Later in wind tunnel and flight tests Reynolds
numbers up to 30 Million were achieved. It could be demonstrated that the skin fric-
tion drag and the friction induced pressure drag can be reduced by up to 50%. But the
requirements concerning surface smoothness and cleanness are extremely high und
thus this promising technology has not yet been applied on transport aircraft.
Fig. 2. Swept wing test results by Göthert (1941) in the DVL high speed wind tunnel [5]
Aerodynamic Wing Design for Transport Aircraft – Today 133
The principle of the swept wing was first published by Busemann in 1935 [4].
Figure 1 shows the governing effect. With increasing sweep angle only the compo-
nent of the Mach number perpendicular to the wings leading edge is of importance.
The component in span direction has no influence on the pressure distribution. The
comparison of an unswept rectangular wing with a wing swept by 35° having the
same span and airfoil section both being tested in 1941 by B. Göthert in the 2.7 m
DVL wind tunnel, clearly showed the advantage of the swept wing concept, see
Figure 2. The strong drag rise indicating rapidly increasing wave drag is shifted to
higher Mach values by about 10%. The strong effect of the sweep angle on the pres-
sure distribution is displayed in Figure 3 showing results computed with an Euler
method for the well known Korn airfoil [12]. This airfoil is designed for M∞=0.815,
cl=0.644 and a sweep angle of 23°. Reducing the sweep angle at constant Mach num-
ber and lift coefficient results in a strongly increasing shock moving downstream. The
appearance of such a growing shock is combined with a strongly increasing wave
drag indicated by the blue symbols in the right part of Figure 3. The red and the black
symbols indicate Mach numbers and lift coefficients based on the flow velocity com-
ponent perpendicular to the leading edge.
Unfortunately the swept wing also has some disadvantages. The sweep of the wing
induces principally a flow component in span direction and the pressure gradient
perpendicular to the oncoming flow twists the velocity vectors inside the boundary
layer. Both effects result in a slightly higher momentum loss of the boundary layer
and thus in a reduced maximum lift coefficient. This is reflected in Figure 4. The
maximum lift coefficients of the infinitely swept wing are 8% lower than those trans-
formed from 2D calculations. Another disadvantage is shown in Figure 5. The sweep
effect leads to a reduced inboard lift and an increased outboard lift. Thus the induced
drag is increased and flow separation starts outboard followed by a dangerous pitch
up moment. For cruise condition this behaviour can be compensated by twisting of
the wing. But even then at higher lift coefficients the same behaviour starts again and
has to be compensated by the layout of the high lift system.
Fig. 3. Pressure distributions and aerodynamic coefficients of an infinitely swept wing with
Korn airfoil, designed for 23° sweep angle and M=0.815 calculated for different sweep angles
(FLOWer calculation, Euler)
134 K.H. Horstmann and T. Streit
2.5D swept
2D Re
2D Re’
cases
Fig. 4. Influence of sweep angle on maxi- Fig. 5. Influence of sweep angle on distribu-
mum lift coefficient tion of circulation along span
Another important item is the development of high speed airfoils. Already in 1940
Kawalki described some basic attributes and design aspects of modern transonic air-
foils, shown in Figure 6. By means of the Theodorsen method, Kawalki investigated
leading edge radius, thickness and camber distribution in order to design an airfoil
with minimum local flow speed at given lift coefficient. In 1941 Göthert performed
systematic airfoil investigations based on Kawalkis findings, see [13]. The further
development of transonic airfoils is linked with names Whitcomb, Korn, Eberle, Re-
deker [14–17] and others. Some well known types of pressure distributions of super-
sonic airfoils are shown in Figure 7, see Körner [18].
Important criteria for the assessment of transonic airfoils are the amount of lift in
the local supersonic region and the behaviour of drag and moment coefficients at off-
design conditions with respect to angle of attack and Mach number. The Peaky Airfoil
with its strong suction peak at the leading edge allows a shock free recompression
from supersonic to subsonic speed, but the lift contribution of the supersonic range is
small. The highest value of supersonic lift can be achieved with the Whitcomb type of
pressure distribution, but the off-design behaviour is marked by high shock induced
drag and strong flow separation. A good compromise between the different goals is
marked as slopy roof top. It combines an acceptable amount of supersonic lift almost
without any wave drag and an excellent off-design performance. Most of the modern
transonic airfoils are based on this type of pressure distribution. In Figure 8 the result
of an airfoil optimisation process by use of the adjoint method of the TAU code is
presented. The pressure distribution of the original airfoil marked by the dashed line
is characterized by a shock at 50% of chord on the upper side. The objective was to
Aerodynamic Wing Design for Transport Aircraft – Today 135
Fig. 6. Investigations on optimum velocity distributions for high speed airfoils according to
Kawalki, 1940 [14]
Fig. 7. Pressure distributions of different types of transsonc airfoils, according to Körner [18]
reduce the drag at constant lift coefficient and constant thickness ratio maintaining the
flow conditions (M∞=0.73, cl=0.8, Re=6.5·106). The shape modification was imple-
mented by a parametric description of the camber line with 20 parameters. For each
optimisation step two adjoint calculations (for cl and cd) have to be performed giving the
gradients of all 20 parameters. Already after seven steps the drag converged on the final
level. The result of the optimisation given by the solid line shows an excellent shock
136 K.H. Horstmann and T. Streit
cD
[cnts]
cL
cp
x/c
free pressure distribution with a drag reduction of about 39% compared to the original
configuration. The type of pressure distribution is close to the slopy roof top one.
A completely different type of pressure distribution in the local supersonic range is
essential for a laminar airfoil for transport aircraft. The high Reynolds number of about
20 Million requires a weak decrease of pressure in the range of the laminar boundary
layer in order to stabilize the boundary layer against increasing Tollmien-Schlichting
instability. A typical pressure distribution is shown in Figure 9, see [19]. The contribu-
tion to lift of the local supersonic speed zone of the laminar airfoil is smaller than that
of the slopy roof top type assuming the same pressure rise at the shock.
The design of a laminar transonic wing is also strongly affected by the local super-
sonic flow field on the upper side of the wing, but it is clearly dominated by the ef-
forts to keep the different types of instabilities on upper and lower side of the wing as
low as possible within a range of lift coefficients, Mach numbers and Reynolds num-
bers. On an unswept wing instabilities of the laminar boundary layer mainly appear in
flow direction, well known as Tollmien-Schlichting instabilities (TSI) [20]. Wave
packages of TS waves are induced which grow in flow direction and finally lead to
transition into turbulent flow. On a swept wing different types of instabilities can
Aerodynamic Wing Design for Transport Aircraft – Today 137
appear: The velocity profile of the laminar boundary layer is twisted, see Figure 10.
The component in flow direction generates TS waves as described above whereas the
component across the main flow direction can also become unstable (cross flow
instability, CFI) and can generate longitudinal vortices, which also may induce transi-
tion, see [21]. Finally the flow along the attachment line of the swept wing can become
unstable followed by transition (ALT). Turbulent boundary layer at the attachment line
infects the whole wing downstream and thus has to be avoided in any case.
For the design of a transonic laminar wing more limitations have to be taken into
account than for a turbulent wing. Besides avoiding transonic wave drag the stability
of the laminar flow governs the design, see Figure 11. TSI is less stable, the lower the
pressure decrease is in flow direction, and CFI is less stable, the higher the pressure
decrease is. The upper surface of the wing for example becomes turbulent by TSI at
lift coefficients above the design range and turbulent by CFI at lift coefficients below
the design range. Similar but vice versa transition occur on the lower side. Each wing
side has its own laminar drag bucket and of course they have to cover the same range
of lift coefficients. The stability of the laminar boundary layer generally depends on
the Reynolds number: Increasing Reynolds number generates more unstable boundary
layer. The cross flow instability additionally depends on the effective sweep angle of
the wing leading edge, see [22]. At a Reynolds number of 20 Million a maximum
sweep angle of about 20° may serve as guiding value.
Fig. 12. Plan view of TELFONA Pathfinder Fig. 13. Section pressure distribution of Path-
model for testing of NLF potential in ETW finder wing near design conditions
Aerodynamic Wing Design for Transport Aircraft – Today 139
Fig. 14. Paralel isobar design of the TEL- Fig. 15. Section pressure distribution de-
FONA Pathfinder wing in presence of belly signed for TELFONA Performance wing
fairing and fuselage
The opinions about a forward swept wing are always dominated by a well known
disadvantage. The bending of a forward swept wing e.g. due to gust loads leads geo-
metrically to an additional local angle of attack in particular in the outer wing region
which increases the gust load: The larger the sweep angle and the stronger the bend-
ing the higher the additional local angle of attack and thus the additional load.
For the application of the laminar flow technology forward sweep has a big advan-
tage. The leading edge sweep angle of a natural laminar flow wing for a Reynolds
number of about 20 Million is due to cross flow dominated transition restricted to
values below 20°. For the transonic behaviour of the wing the local sweep angle at the
shock location is the significant parameter. A higher local sweep angle allows for
higher local lift, higher airfoil thickness ratio or higher flight Mach number. On lami-
nar wings the shock appears at the end of the laminar range at about 50% of chord
length. For a typical backward swept transport aircraft wing with an aspect ratio of 9
and a taper ratio of 0.25, the sweep angle of the 50% line is about 13° and the leading
edge sweep angle has a value of 20°. The forward swept wing with a leading edge
sweep angle of –20° in contrast features a sweep angle of the 50% line of 26°. This is a
remarkable advantage for the transonic behaviour of a natural laminar flow wing. It
demands including the forward swept wing into the studies of laminar flow wing design.
Another interesting aspect of the forward swept wing is the change of the load dis-
tribution compared with the backward swept wing. Figure 17 shows simple lifting line
results of three tapered wings of different sweep angles. The local lift coefficient of
the unswept wing is due to the small - but typical - taper ratio higher in the outboard
region than in the inboard region. This tendency increases with increasing backward
sweep angle and requires at last a nose down twist of about 4° to 5° in order to get an
acceptable load distribution. On the forward swept wing in contrast the lift coeffi-
cients are evenly distributed along the span and a wing twist is not required. Figure 18
gives an illustration of this behaviour. The trailing vortices of an unswept wing with
elliptical load distribution induce a constant downwash along the span. On a back-
ward swept wing the lifting vortices induce an additional downwash in the centre
140 K.H. Horstmann and T. Streit
CL
CL
CL
Taper Ratio
Fig. 16. Drag polars of airfoils for different Fig. 17. Distribution of lift coefficients of
sweep angles at laminar and fully turbulent untwisted wings at different sweep angles
boudary layer (TELFONA)
Fig. 18. Upwash and downwash distribution at wings of different sweep angles
region of the wing. The trailing vortices from the tip geometrically have a smaller
influence on the wing and induce a relative upwash compared to the unswept wing.
On a forward swept wing in contrast the lifting vortices induce a relative upwash near
the midsection and the stronger influence of the trailing vortices from the tip region
Aerodynamic Wing Design for Transport Aircraft – Today 141
Fig. 19. Lift increase along span of BSW- Fig. 20. Surface pressure distribution of a
and FSW wing calculated by lifting line FSW, designed for fully turbulent boundary
method layer
induces an additional downwash in the outboard region of the wing. The size of this
effect is presented in Figure 19 showing the lift increase along the span of a typical
backward swept transport aircraft wing and a corresponding forward swept wing. The
lift increase in the outbord region of the BSW is about 20 to 30% higher than for the
FSW. This aerodynamical advantage of a forward swept wing may even compensate
the additional angle of attack geometrically caused by bending.
First investigations on a forward swept wing designed for turbulent flow, see
Figure 20, show that the reduced downwash near midsection may induce high suction
peaks near wing root combined with a stronger shock and the danger of flow separa-
tion at the wing fuselage intersection.
The large benefits of the forward swept wing due to laminar flow technology have
to be assessed in detail against known drawbacks and the validity of technological
prejudices has to be checked carefully.
A technology which also could increase the efficiency of today’s aircraft is the tar-
geted use of the elastic deformation induced by gust or maneuver loads. The wing
142 K.H. Horstmann and T. Streit
Fig. 22. Isobar distribution on a transport Fig. 23. Influence of wing deformation on
aircraft wing at a load factor of n=2.5 relative values of wing root bending moment
and drag coefficient
4 Perspectives
The above described technologies being used for the design of all modern transonic
aircraft are the swept wing and the supercritical airfoil/wing design. In combination
with modern design and analysis methods developed during the last 50 years they
contribute considerably to the highly efficient wings of today’s aircraft.
Aerodynamic Wing Design for Transport Aircraft – Today 143
References
[1] Lilienthal, O.: Der Vogelflug als Grundlage der Fliegekunst, 1st edn. (1889); (reprinted
2003) ISBN 3-9809023-8-2
[2] Schmitt, G.: Junkers und seine Flugzeuge, 2. Auflage Transpress VEB Verlag für Verke-
hrswesen, Berlin (1986) ISBN 3-344-00065-9
[3] Lachmann, G.: Experiments with Slotted Wings, NACA TN71, 1921, translated from
Zeitschrift für Flugtechnik und Motorluftschiffahrt (1921)
[4] Busemann, A.: Aerodynamischer Auftrieb bei Überschallgeschwindigkeit, Luftfahrtfor-
schung Bd.12 (1935)
[5] Ludwieg, H.: Verringerung des Widerstandes von Tragflügeln bei hohen Geschwindig-
keiten durch Pfeilform, AVA-Bericht 39/H18 (1939)
[6] Meier, H.U. (ed.): Die Pfeilflügelentwicklung in Deutschland bis 1945. Bernard & Graefe
Verlag, Bonn (2006)
[7] Frenzl, O., Hertel, H., Hempel, W.: Widerstandsarme Gestaltung von Hochgeschwindig-
keitsflugzeugen, auch von solchen mit außerhalb des Flugzeugumrisses liegenden Ver-
drängungskörpern, German Patent Office, Patent 932410 (1944)
[8] Whitcomb, R.T.: A study of the Zero-Lift Drag-Rise Characteristics for Wing-Body
Combinations near the Speed of Sound, NACA RM L 52 H 08 (1952)
[9] Kawalki, K.H.: Profile geringster Übergeschwindigkeiten bei vorgegebenen Auftriebs-
beiwerten. LGL-Bericht Nr. 156, pp. 163–169 (1942)
[10] Doetsch, H.: Untersuchungen an einigen Profilen mit geringem Widerstand im Bereich
kleiner ca-Werte, Jb, dtsch. Luftfahrtforschung, Bd. I, pp. 54–57 (1940)
144 K.H. Horstmann and T. Streit
Summary
A review is presented of some previous papers on two control techniques with applica-
tion to axial-flow compressors. The first technique is a passive technique using stator
blades with a forward swept leading edge as a means to prevent an early breakdown of
the stator hub flow in a highly loaded single-stage axial-flow low-speed compressor.
The second technique is a predetermined technique utilizing rotor tip injection to sup-
press spike-type stall inception in three single-stage axial-flow low-speed compressors
of different design. The first technique is shown to work highly successful, especially
at off-design conditions. The second technique is shown to be highly effective in rais-
ing the stall limit line. On the other hand, the associated efficiency drop can be signifi-
cant, so that the injection technique may not be recommended for steady state
operation. Injection should only be employed to enhance compressor stability during
transient operation that means for a short time period only.
1 Introduction
Flow control is used to effect beneficial changes in a given flow. The corresponding
control techniques may be passive, requiring no auxiliary power and no control loops,
or active, requiring energy expenditure and control loops. Active control is further split
into predetermined or reactive. Predetermined control means control with energy input
but without regard of the particular state of the flow. The loop is an open one. Reactive
control means control with continuously adjusted energy input guided by some suitable
measurements. The corresponding control loop may be open or closed [1].
Two of the aforementioned control techniques were used in the work described in
[2–4] on which the present paper is based: the first is a passive and the second a pre-
determined active technique. A passive technique was applied to prevent an early
breakdown of the stator hub flow in a highly loaded single-stage axial-flow low-speed
compressor [2, 3] and a predetermined active technique was implemented in a test rig
to suppress spike-type stall inception [4]. In the first case (chapter 2) two different
stators, A and K, were investigated. Stator A is a conventional design with blades of
rectangular planform while stator K is an unconventional design with blades of spe-
cial planform characterized by a forward swept leading edge with increasing sweep
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 145–167.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
146 U. Stark and H. Saathoff
angle towards hub and casing. In the second case (chapter 3) three single-stage axial-
flow low-speed compressors of different design were tested with and without rotor tip
injection.
The intention of the present paper is to summarize the more important results of the
work described in [2–4] and to highlight the possibility to passively or actively con-
trol the flow in turbomachines.
Performance tests and flow field investigations were carried out on a highly loaded
single-stage axial-flow low-speed compressor, Figure 1, at the Pfleiderer-Institute,
Technical University of Braunschweig. The compressor has an outer diameter
Do = 0.4 m and a hub to tip ration Di/Do = 0.55. The rotor comprises 16 blades of
standard (A10) NACA 65-series blade sections and the aerodynamic design is such as
to give a free vortex flow for the design parameters φ = 0.45 and ψ = 0.6. The rotor
tip clearance to chord ratio is s/l = 0.0067 with a clearance s = 0.4 mm and a chord
length l = 60 mm.
The original stator A is a conventional design with constant chord blades of stan-
dard (A10) NACA 65-series blade sections. The chord length is l = 90 mm and the
maximum thickness d = 5.4 mm at all radii. The blade height is h = 90 mm and the
AR = 1.0
Rotor Stator A
A10 A10 Do
r NACA 65 NACA 65 Di
θ z
A K
Fig. 2. Stator blades A (left frame) and K (right frame)
blade aspect ratio AR = 1.0, see Figure 2. The stator hub turning angle ∆β ≈ 60° and
the diffusion factor D ≈ 0.65 are high and the de Haller number DH ≈ 0.55 is low.
With these design point loadings at about 5% span the original stator A develops
increasing hub endwall and corner separation with decreasing flow rate. The corre-
sponding endwall blockage and losses were found to be the main reason for an inade-
quate off-design performance. Therefore a second stator K, see Figure 2, was
designed and tested with the objective to significantly improve the off-design per-
formance keeping the design point turning angle ∆β, diffusion factor D and de Haller
number DH constant at hub and casing, respectively. To that end all blade sections of
the stator were completely redesigned based on a measured inlet angle distribution.
The same distribution has already been used for the design of stator A.
The second stator K is an unconventional design using circular arc mean lines of
constant curvature at all radii resulting in an increasing chord length towards the hub
and, to much lesser extent, towards the casing. The circular arc mean lines were com-
bined with NACA 65-series thickness distributions of maximum thickness
d = 5.4 mm and the resulting sections were radially stacked in such a way as to give
an unswept (straight) trailing edge and a forward swept leading edge with increasing
sweep angle towards the hub (φH = +30°) and casing (φC = +10°). At the same time
the number of blades has been reduced from 19 to 15 (−20%) in order to keep the
diffusion factor similar for stators A and K at hub and casing. The usage of circular
arc mean lines instead of standard A10 mean lines leads to a basic reduction in lead-
ing edge loading of the blades, which is reinforced by the planform (sweep) produced
loading reduction [5, 6]. For stator blades of this type a three-dimensional relief of the
endwall flow could be expected together with a reduction in cross-passage pressure
gradient and flow, boundary layer separation and losses and altogether with an im-
proved off-design performance.
The stage was run at a constant speed (4500 rpm) and the air flow rate adjusted by
means of a throttle and an auxiliary fan until a prescribed flow coefficient φ was ob-
tained. The velocities encountered in the stage were such that the flow may be regarded
as incompressible. The Reynolds number based on chord and relative inlet velocity at
the rotor tip is Re = 4.1·105 for the design flow conditions. The rotor/stator distance is
a = 80 mm from rotor trailing edge to stator leading edge at mid-span height. This is
comparatively large but leads to a reduced unsteadiness of the stator flow.
148 U. Stark and H. Saathoff
Sweep and dihedral have been used for some time to achieve improvements in ef-
ficiency and stall margin. Most of the published work has focused on rotor blades,
subsonic and transonic rotor blades [7–20] with much less information being available
for subsonic stator blades [12, 13] and [21–24]. Special features of the present sub-
sonic stator investigations are i) the extremely high loading of the stator hub and
casing in terms of turning angle, diffusion factor and de Haller number and ii) a con-
siderable reduction of the blade count in spite of the extreme loading.
Fig. 3. Comp. grid, top frame: blade to blade view, bottom frame: meridional view
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 149
The numerical calculations of the present paper were performed using the CFD package
Fine/Turbo of NUMECA Int. S. A.. The code solves the full Reynolds-Averaged Navier-
Stokes equations (RANS) on structured multi-block grids with mixing planes between
adjacent blade rows. An explicit time marching, implicit residual smoothing 4-step
Runge-Kutta procedure, a cell-centered second-order finite-volume discretization and the
algebraic turbulence model of Baldwin-Lomax without wall function were used. In addi-
tion, local time stepping and multi-grid capability were applied to enhance convergence.
For both blade rows standard elliptical H-type grids were used to mesh the numeri-
cal domain as shown in Figure 3, giving approximately 685000 points. The rotor tip
clearance was resolved separately with a so-called butterfly grid block of about 30000
points, making 715000 point in total for a complete stage with y+ values smaller than 5.
Calculated radial total pressure (pt = 101.3 kPa) and total temperature (Tt = 288 K)
distributions together with zero inlet swirl were imposed at the compressor inlet
(80 mm in front of the rotor). Off-design operating points were calculated by increas-
ing the static back pressure and the last numerically stable operating point is consid-
ered to be close to stall.
0.80 0.90
ψ η
0.75 0.85
CFD A
CFD K
0.70 Exp. A 0.80
Exp. K
0.65 0.75
0.60 0.70
0.55 0.65
0.35 0.40 0.45 0.50 ϕ 0.55
does not show much influence on the design point performance of the compressor, but
has an influence on the off-design performance, which is significantly improved. The
compressor is known to be a rotor tip-stalling machine so that a change in stalling
flow range has not been expected. The agreement between the experimental and nu-
merical results is good at φ = 0.45, with a small discrepancy gradually increasing as φ
decreases.
Mid-Passage Pressure Distributions on the Hub
Surface pressure distributions may be used to analyse the flow next to the surface
under consideration. Figure 5, for example, shows at flow rates φ = 0.45 (design flow
rate) and φ = 0.37 (part load) a leading edge to leading edge comparison of measured
and calculated hub mid-passage static pressure distributions cp(z) for stator A and K
between inlet and outlet. The pressure coefficient is defined as
cp(z) = (p(z) − p2)/(ρ/2)UT2 and is plotted against the axial direction z.
As shown in Figure 5, the deceleration of the flow starts directly at the beginning
of the passage for stator A, which is a camberline (A10) effect and at some point
further downstream the passage for stator K, which is a planform effect. For stator K
the pressure distributions at φ = 0.45 and φ = 0.37 are quite similar in shape indicating
a hub boundary layer development without dramatic changes. The corresponding
pressure distributions for stator A are completely different and indicate severe hub
boundary layer separation at φ = 0.37 by a long drawn-out pressure distribution. In
spite of thick and partly 3D-separated endwall boundary layers the agreement be-
tween experimental and numerical results is surprisingly good.
LE A/K TE A K LE A/K TE A K
0.5 0.5
cp CFD A cp CFD A
CFD K CFD K
0.4 Exp. A 0.4 Exp. A
Exp. K Exp. K
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
-0.1 -0.1
-0.2 -0.2
-20 0 20 40 60 80 100 120 140 160 180 -20 0 20 40 60 80 100 120 140 160 180
z [mm] z [mm]
Fig. 5. Measured and calculated hub pressure distributions along the mid passage streamline,
left frame: design flow rate φ = 0.45, right frame: near stall flow rate φ = 0.37
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 151
0.5 0.5
cp cp
0.3 0.3
0.1 0.1
-0.1 -0.1
-0.3 -0.3
CFD A CFD K
Exp. A Exp. K
-0.5 -0.5
-0.7 -0.7
20 40 60 80 100 120 140 20 40 60 80 100 120 140
z [mm] z [mm]
0.5 0.5
cp cp
0.3 0.3
0.1 0.1
-0.1 -0.1
-0.3 -0.3
CFD A CFD K
Exp. A Exp. K
-0.5 -0.5
-0.7 -0.7
20 40 60 80 100 120 140 20 40 60 80 100 120 140
z [mm] z [mm]
0.5 0.5
cp cp
0.3 0.3
0.1 0.1
-0.1 -0.1
-0.3 -0.3
CFD A CFD K
Exp. A Exp. K
-0.5 -0.5
-0.7 -0.7
20 40 60 80 100 120 140 20 40 60 80 100 120 140
z [mm] z [mm]
Fig. 6. Measured and calculated blade pressure distributions at the hub, flow rates φ = 0.45,
0.41 and 0.37 (from top to bottom), left frames: stator A, right frames: stator K
152 U. Stark and H. Saathoff
experimental distributions were measured on the hub using pressure taps close to the
blades and the numerical distributions were generated using again the CFD package
Fine/Turbo of NUMECA Int. S. A.
The first row in Figure 6 is for stator A and K at φ = 0.45. Experimental and nu-
merical distributions show good agreement, for stator A as well as for stator K. As
mentioned earlier in the paper, stator K blades were expected to show reduced leading
edge loading and, as a consequence, to initiate less secondary flow and losses. This
reduction has been achieved and may be seen in Figure 6 comparing stator A and K
pressure distributions at φ = 0.45.
Fig. 7. Stator A hub oil flow visualisation, design flow rate φ = 0.45
Fig. 8. Stator K hub oil flow visualisation, design flow rate φ = 0.45
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 153
Fig. 9. Stator A hub oil flow visualisation, near stall flow rate φ = 0.37
Fig. 10. Stator K hub oil flow visualisation, near stall flow rate φ = 0.37
154 U. Stark and H. Saathoff
Oil flow pictures are photographs of especially prepared surfaces showing the surface
pattern of the shear stress lines. The directions of these lines are nearly identical with
the flow directions of the fluid particles next to the surface. Following the criteria de-
scribed in [25] the characteristic features of the surface flow may be identified, as for
example: i) three-dimensional separation lines (SL, convergence of the shear stress
lines), ii) three-dimensional (re-)attachment lines (AL, divergence of the shear stress
lines) and iii) so-called singular points (saddle and nodal points, foci). Based on these
features, a first idea of the flow field close to the wall may be developed.
An analysis of the hub endwall flow starts best with Figures 7 and 8 for φ = 0.45.
By comparison it may be seen that both configurations develop a classic secondary
flow superimposed on a primary flow. At about the end of the entrance region, the
endwall boundary layer is separated in stator A, as well as in configuration K, with
the important difference, however, that in A a vortex – with a spiral point at the hub –
is established, which is not like that in K. The vortex in A has a strong influence on
the downstream flow, which is, due to the vortex induced velocities, partly reversed
between the cross-passage attachment line and the spiral point. In K, without a vortex
in the passage, the flow is axial forward throughout the passage. The deterioration of
the hub endwall flow with decreasing flow rate is much faster in stator A than in sta-
tor K. This has already been mentioned and is now confirmed for the special flow rate
φ = 0.37, see Figures 9 and 10. Figure 9 shows axial backward flow in stator A
throughout the passage between outlet and inlet while Figure 10 for stator K indicates
backflow in a small region only, similar to stator A at φ = 0.45 in Figure 7.
These and the other results closely correspond to the expectations that have been
expressed at the beginning of the paper. At high stator hub loading, when a complete
breakdown of the hub endwall flow is likely to occur, the introduction of forward
swept stator blades turns out to be an effective means of passive control of the hub
endwall flow.
The stages were run at a constant speed of n = 4500 rpm and the air flow rate was
adjusted by means of a throttle and an auxiliary fan until the upstream axial velocity
to produce the desired flow coefficient φ was obtained. The velocities encountered in
the test facility are such that the flow may be regarded as being incompressible. Over-
all performance characteristics were evaluated i) in terms of static-to-static pressure
rise (ψ(φ)) and ii) in terms of total-to-static pressure rise (ψts(φ)).
Fig. 13. Radial probe arrangement upstream and downstream from the rotor
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 157
the transient behaviour of the inception process as well as its motion relative to the
rotor right from the beginning, two cranked total pressure probes (probes 1 & 2)
were used [32], each equipped with a high frequency miniature pressure transducer
(Kulite XCS-062-5D). Measurements of the unsteady total pressure were recorded at
∆z = −15 mm upstream of the rotor for several radial positions r’ between casing and
mid-span as well as probe angles ε (defined in Figure 14). In addition to the up-
stream total pressure probes (1 & 2) two cylindrical pressure probes (3 & 4) with
lateral downward facing measuring holes (equipped with the same pressure trans-
ducers as mentioned before) were used to acquire unsteady pressure data just down-
stream of the rotor, see Figures 13–15. With reference to the 2D pressure distribution
of a cylinder and in consideration of the results of a special calibration, the axes of
the lateral measuring holes were aligned with respect to the rotor axis in such a way
(ε = 45°) that axially backward directed flow was indicated by a change of sign of
the pressure signal. The radial position of the four pressure probes was r’ = 2 mm
below the casing when spike-type stall inception was investigated. Both types of
measurements were carried out with a sampling frequency of 50 kHz, corresponding
to about 42 samples per blade passage. The available data storage was divided up in
equal shares to save the pre- and post-stall history using the stall event itself as an
external trigger.
As an example, taken from [4], Figure 16 shows the time history of the unsteady
pressure as measured by probe number 1 and 3 during spike-type stall inception for
stage 2, set at 1.5 percent tip clearance. During stable compressor operation the pres-
sure distribution in front of the rotor (trace of probe 1 in Figure 16, top frame) is
formed by the upstream effect of the rotor blades. For each rotor revolution 16 blades
can be identified passing the probe. A first disturbance of the pressure distribution is
shaped like a spike and represents a distinct pressure and suction peak of small
circumferential extension due to unsteady spillage of tip clearance flow ahead of the
Fig. 16. Unsteady pressure distributions up- and downstream of the rotor, stage 2, tip clearance
to chord ratio of s/l = 0.015
160 U. Stark and H. Saathoff
rotor [30]. The initial spike then grows in circumferential as well as in radial direction
[4] until, after less than 10 rotor revolutions, a new steady state is reached. Thus it
may be concluded that in the present example there is one local flow disturbance that
grows into a fully developed stall cell. As mentioned before, there is a second condi-
tion for spike-type stall inception that must be fulfilled simultaneously with the first
condition. This second condition is the appearance of locally confined, unsteady back-
flow at the rotor outlet close to the casing wall. The corresponding pressure distribu-
tion (trace of probe 3 in Figure 16, top frame) shows axially backward directed flow
exactly at the position of the first spike in the trace of probe 1. To the best of the au-
thors’ knowledge this is the first time that the numerical results described in [31] were
experimentally verified.
A close-up of the traces is shown in the middle frame of Figure 16 demonstrating
more clearly that the first spike in trace 1 directly links up with a backflow indicating
spike in trace 3. Spikes of this type occur several times, irregularly distributed around
the circumference before the first spike appears in trace 1.
The bottom frame of Figure 16 represents a still closer view of the pressure distri-
butions together with a number of rotor blades. The constellation of the rotor blades
and the pressure distributions indicates that backflow occurs near the trailing edge and
spill forward near the leading edge of the same rotor blade.
Simulated Rotor Tip Flow
For a better understanding of the results presented so far, the rotor relative flow for
stage 1 was simulated in a rotor tip equivalent 2:1 linear cascade model with a clear-
ance (3 percent of the blade chord length) on one side [26]. Visualised by an oil
technique, Figure 17 from [26] illustrates spill forward for the rotor equivalent com-
pressor cascade at β1 = 64°, which closely corresponds to the stall inception point of
the machine. Starting at the roll-up point, which is close to the leading edge of the
blade, a well defined clearance vortex – with S-shaped shear stress lines on the wall –
begins to develop and may be traced up to the 65 percent chord position, where i) the
S-shape of the shear stress lines suddenly disappears and ii) the cross-section of the
vortex rapidly expands – both of which are indications for an abrupt degeneration of
the clearance vortex. In addition, the oil flow picture shows tip clearance fluid cross-
ing the blade passage diagonally, then contracting and finally spilling forward ahead
of the next blade and from there into the adjacent passage along the row. Together
with the clearance vortex the spilled clearance flow acts as a flow obstruction causing
the incoming endwall boundary layer to separate along a circumferential separation
line close to the leading edge plane. The topological essentials of the endwall flow
were summarized in a sketch (Figure 17, right frame). Moreover, estimated velocity
triangles were marked at positions of axial backflow just upstream and downstream of
the blading, respectively. These velocity triangles explain the orientation of the probes
in Figure 14 and link the main flow features of the simulated rotor relative flow to the
initial disturbances found in the unsteady pressure distributions discussed in the last
paragraph.
Overall Compressor Performance with Tip Injection
From the results presented so far it is known that two flow events must be simultane-
ously present for spike-type stall inception to occur: i) spillage of the clearance fluid
ahead of the next blade and from there into the adjacent passage and ii) backflow of
the tip clearance fluid from adjacent blade passages at the trailing edge plane. Based
on this knowledge an air injection technique has been designed with the intention to
enhance the momentum of the incoming flow. The characteristics for all three stages,
set at 3 percent rotor tip clearance, were measured for various injection flow rates V&N
using 8 and 16 nozzles, respectively, equally spaced around the annulus at
∆z = −3 mm upstream of the rotor tip, see Figure 18. The nozzle flow was controlled
by a flow meter, a pressure gauge and a throttle valve.
Figure 19 presents the measured stage pressure rise versus flow rate characteristics
with and without tip injection. The characteristics show a significant shift of the last
stable operating point towards lower flow and higher pressure coefficients. In addition
Figure 19 shows the injection technique to be more effective with 16 nozzles than
with 8 nozzles.
Further experiments with air injection demonstrated the capability of the stages to
recover from a stalled condition. Starting with a stable operating point on the secon-
dary characteristic, the injected air was continuously increased until the pressure rise
and mass flow suddenly jumped to significantly higher values on the extended pri-
mary characteristic, as indicated by arrows in Figure 20, left frame. Hence it may be
concluded that during fully developed rotating stall unsteady spillage of the rotor tip
clearance flow still exists, see [33].
Various radial positions r’N of the nozzles and injection angles ε were investigated
to determine the sensitivity of the compressor flow to the nozzle arrangement. As a
result of these measurements it was found that the performance of all three stages
162 U. Stark and H. Saathoff
Fig. 19. Stage characteristics with tip injection, tip clearance to chord ratio of s/l = 0.03
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 163
Fig. 20. Stall recovery (left frame) and influence of injection angle (right frame),results for
stage 2 with tip clearance to chord ratio of s/l = 0.03
decreases with increasing r’N. This result supports the previous findings that spike-
type stall inception is due to unsteady spill forward of the rotor tip clearance flow.
Figure 20, right frame, illustrates the effect of the injection angle ε on the stall flow
rate for stage 2 with 16 nozzles and the maximum injection flow rate V & N = 0.019
3
m /s. For positive injection angles ε (jet direction against rotor rotation) the stall flow
rate φS increases continuously and is for the highest ε values even higher than the stall
flow rate without injection.
The attainable shift of the last stable operating point towards lower flow and higher
pressure coefficients increased with increasing injection flow rate. The corresponding
jet power added to the flow has to be taken into account calculating the stage effi-
ciency. For the high injection flow rates investigated in the present study a significant
drop of the stage efficiency was determined [34]. Low injection flow rates on the
other hand did not decrease the efficiency while already producing a noticeably in-
creased operating range of the compressor stages.
blades shows a much better off-design performance than the corresponding datum
stage with conventional (unswept) stator blades.
For three single-stage low-speed compressors, the introduction of a rotor tip injec-
tion technique was found to be highly effective in shifting the last stable operating
point towards lower flow and higher pressure coefficients on the respective stage
characteristic. The measurements were carried out for various injection flow rates
with 8 as well as 16 nozzles.
The corresponding stage efficiencies were not presented in the paper, but may be
found in [34], where they show a reduction of up to 14 percent for the tests with injec-
tion depending on stage design, tip clearance, injection flow rate and nozzle number.
With these results, the injection technique may not be recommended for steady state
operation. Injection should only be used to enhance compressor stability during tran-
sient operation, for which the negative performance impacts become negligible [35].
Acknowledgement
The authors would like to acknowledge the following colleagues for their contribu-
tions to the original papers [2], [3] and [4] on which the present review paper is based:
Dr.-Ing. S. Baumgarten (KSB, Frankenthal), Dipl.-Ing. C. Clemen (Rolls-Royce
Deutschland, Dahlewitz), Dr.-Ing. A. Deppe (Lufthansa Technik, Hamburg), Dr.-Ing.
J. Friedrichs (MTU Maintenance, Hannover), Prof. Dr.-Ing. G. Kosyna and Dr.-Ing.
D. Wulff (Pfleiderer-Institut für Strömungsmaschinen, TU Braunschweig).
Nomenclature
Nomenclature Chapter 2
a distance rotor/stator β flow angle
AR aspect ratio ∆β flow turning
cp pressure coefficient η efficiency
d max. thickness θ coordinate
D diameter λ stagger angle
D diffusion factor ρ density
DH de Haller number φ sweep angle
h blade height φ flow coefficient
l blade chord ψ pressure coefficient
p static pressure Abbreviations
pt total pressure CFD computational fluid dynamics
r coordinate Exp. experiment
Re Reynolds number LE leading edge
s clearance TE trailing edge
t space AL attachment line
Tt total temperature SL separation line
U rotor velocity Indices
W flow velocity 1, 2, 3 axial stations
y+ non-dimensional wall distance i, o inner, outer
z coordinate C, T, H casing, tip, hub
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 165
Nomenclature Chapter 3
D diameter β1 inlet flow angle
l blade chord θ coordinate
n rotational speed ε probe or nozzle angle
r coordinate φ flow coefficient
r’ radial distance from casing ψ pressure coefficient
s clearance Indices
injection flow rate i, o inner, outer
z coordinate ref reference
S stall
ts total-to-static
References
[1] Gad-El-Hak, M.: Flow Control: Passive, Active and Reactive Flow Management. Cam-
bridge University Press, Cambridge (2000)
[2] Friedrichs, J., Baumgarten, S., Kosyna, G., Stark, U.: Effect of Stator Designon Stator
Boundary Layer Flow in a Highly Loaded Single-Stage Axial-Flow Low-Speed Com-
pressor. ASME Journal of Turbomachinery 123, 483–489 (2001)
[3] Clemen, C., Stark, U., Friedrichs, J., Baumgarten, S., Kosyna, G.: Effect of Mean Line
Shape, Sweep and Dihedral on Stator Performance of a Highly LoadedSingle-Stage Ax-
ial-Flow Low-Speed Compressor. In: Proceedings of the 6th European Conference on
Turbomachinery, Fluid Dynamics and Thermodynamics, pp. 34–45 (2005)
[4] Deppe, A., Saathoff, H., Stark, U.: Spike-Type Stall Inception in Axial-Flow Compres-
sors. In: Proceedings of the 6th European Conference on Turbomachinery, Fluid Dynam-
ics and Thermodynamics, pp. 178–188 (2005)
[5] Multhopp, H.: Methods for Calculating the Lift Distribution of Wings (Subsonic Lifting-
Surface Theory), A.R.C., R.M., No. 2884 (1950)
[6] Küchemann, D.: A Simple Method for Calculating the Span and Chordwise Loading on
Straight and Swept Wings of Any Given Aspect Ratio at Subsonic Speeds., A.R.C., R.M.,
No. 2935 (1952)
[7] Mohammed, K.P., Prithvi Raj, D.: Investigations on Axial Flow Fan Impellers with For-
ward Swept Blades. ASME Journal of Fluids Engineering 99, 543–547 (1977)
[8] Yamaguchi, N., Tominaga, T., Hattori, S., Mitsuhashi, T.: Secondary-Loss Reduction by
Forward-Skewing of Axial Compressor Rotor Blading. In: Yokohama International Gas
Turbine Conference (1991)
[9] Inoue, M., Kuroumaru, M., Furukawa, M., Kinoue, Y., Tanino, T., Maeda, S., Okuno, K.:
Controlled-Endwall-Flow Blading for Multistage Axial Compressor Rotor. ASME Paper
97-GT-248 (1997)
[10] Wadia, A.R., Szucs, P.N., Crall, D.W.: Inner Workings of Aerodynamic Sweep. ASME
Journal of Turbomachinery 120, 671–682 (1998)
[11] Hah, C., Puterbaugh, S.L., Wadia, A.R.: Control of Shock Structure and Secondary Flow
Field Inside Transonic Compressor Rotors Through Aerodynamic Sweep. ASME Paper
98-GT-561 (1998)
[12] Lyes, P.A., Ginder, R.B.: Low-Speed Compressor Tests of Swept and Bowed Blade De-
signs. ISABE Paper 99-7048 (1999)
166 U. Stark and H. Saathoff
[13] Gallimore, S.J., Bolger, J.J., Cumpsty, N.A., Taylor, M.J., Wright, P., Place, M.M.: The
Use of Sweep and Dihedral in Multistage Axial Flow Compressor Blading. Part I: Uni-
versity Research and Methods Development. Part II: Low and High Speed Designs and
Test Verification. ASME Journal of Turbomachinery 124, 521–541 (2002)
[14] Denton, J.D., Xu, L.: The Effects of Lean and Sweep on Transonic Fan Performance.
ASME Paper GT-2002-30327 (2002)
[15] Passrucker, H., Engber, M., Kablitz, S., Hennecke, D.K.: The Effect of Forward Sweep in
a Transonic Compressor Rotor. In: Proceedings of the 5th European Conference on Tur-
bomachinery, Fluid Dynamics and Thermodynamics, pp. 141–150 (2003)
[16] Scott McNulty, G., Decker, J.J., Beacher, B.F., Khalid, A.S.: The Impact of Forward
Swept Rotors on Tip Clearance Flows in Subsonic Axial Compressors. ASME Journal of
Turbomachinery 126, 445–454 (2004)
[17] Goller, M., Gümmer, V., Clemen, C., Swoboda, M.: Enhancement of Highly- Loaded Ax-
ial Compressor Stage Performance Using Rotor Blade Tip Tailoring. Part I – Numerical
Design Studies. In: Proceedings of the 6th European Conference on Turbomachinery,
Fluid Dynamics and Thermodynamics, pp. 88–99 (2005)
[18] Rohkamm, H., Kosyna, G., Saathoff, H., Stark, U.: Enhancement of Highly- Loaded Ax-
ial Compressor Stage Performance Using Rotor Blade Tip Tailoring. Part II – Experimen-
tal Results. In: Proceedings of the 6th European Conference on Turbomachinery, Fluid
Dynamics and Thermodynamics, pp. 100–110 (2005)
[19] Bergner, J., Kablitz, S., Hennecke, D.K., Passrucker, H., Steinhardt, E.: Influence of
Sweep on the 3D Shock Structure in an Axial Transonic Compressor, ASME Paper 2005-
GT-68835 (2005)
[20] Braun, M., Seume, J.R.: Forward Sweep in a Four-Stage High-Speed Axial Compressor,
ASME Paper 2006-GT-90218 (2006)
[21] Tweedt, D.W., Okiishi, T.H., Hathaway, M.D.: Stator Endwall Leading-Edge Sweep and
Hub Shroud Influence on Compressor Performance. ASME Journal of Turbomachin-
ery 108, 224–232 (1986)
[22] Weingold, H.D., Neubert, R.J., Behlke, R.F., Potter, G.E.: Bowed Stators: An Example of
CFD Applied to Improve Multistage Compressor Efficiency. ASME Journal of Turbo-
machinery 114, 161–168 (1992)
[23] Gümmer, V., Wenger, U., Kau, H.P.: Using sweep and Dihedral to Control Three-
Dimensional Flow in Transonic Stators of Axial Compressors, ASME Paper 2000–GT-
491 (2000)
[24] Fischer, A., Riess, W., Seume, J.R.: Performance of Strongly Bowed Stators in a Four-
Stage High-Speed Compressor. ASME Journal of Turbomachinery 126, 333–338 (2004)
[25] Tobak, M.T., Peake, D.J.: Topology of Three-Dimensional Separated Flows. Annual Re-
view of Fluid Mechanics 14, 61–85 (1982)
[26] Saathoff, H.: Rotor-Spaltströmungen in Axialverdichtern, ZLRForschungsbericht 2001-
05 (Dissertation, TU Braunschweig) (2001)
[27] Rohdenburg, M.: Kennfeldberechnungsverfahren für Axialventilatoren mit Berücksichti-
gung der Naben- und Gehäusegrenzschichten, Mitteilungen des Pfleiderer- Instituts für
Strömungsmaschinen, Verlag und Bildarchiv W.H. Faragallah, Heft 8 (Dissertation, TU
Braunschweig) (2003)
[28] Camp, T.R., Day, I.J.: A Study of Spike and Modal Stall Phenomena in a Low-Speed Ax-
ial Compressor. ASME Journal of Turbomachinery 120, 393–401 (1998)
[29] McDougall, N.M.: Stall Inception in Axial Compressors, Ph. D. Dissertation, University
of Cambridge (1988)
Passive and Active Methods to Enhance Axial-Flow Compressor Aerodynamics 167
[30] Dobat, A., Saathoff, H., Wulff, D.: Experimentelle Untersuchungen zur Entstehung von
Rotating Stall in Axialventilatoren, VDI-Berichte Nr. 1591, pp. 345–360 (2001)
[31] Vo, H.D.: Role of Tip Clearance Flow on Axial Compressor Stability, Ph. D. Disserta-
tion, Massachusetts Institute of Technology (2001)
[32] Brodersen, S., Wulff, D.: Measurements of the Pressure and Velocity Distribution in
Low-Speed Turbomachinery by Means of High-Frequency Pressure Tansducers. ASME
Journal of Turbomachinery 114, 100–107 (1992)
[33] Werle, H., Gallon, M.: Maquette Hydraulique de Turbomachine. La Recherche Aerospa-
tiale 5, 267–288 (1977)
[34] Deppe, A.: Experimentelle Untersuchungen zur Auslösung von Rotating Stall in Axial-
verdichtern, ZLR-Forschungsbericht 2008-01 (Dissertation, TU Braunschweig) (2008)
[35] Horn, W., Schmidt, K.-J., Staudacher, S.: Effects of Compressor Tip Injection on Aircraft
Engine Performance and Stability, ASME Paper GT2007-27574
Flow Analysis of Augmented High-Lift Systems
Summary
Recent progress in numerical flow simulation methods and flow measurement tech-
niques opens up new potentials for designing and optimizing augmented high-lift
systems. This research is directed towards substantially higher lift coefficients,
needed for future generations of quiet transport aircraft with short take-off and land-
ing capabilities. Based upon a review of several decades of powered lift research the
present investigations focus on advanced circulation control concepts and their inte-
gration with modern transonic wing sections. The gains of blowing moderate amounts
of compressed air over carefully selected Coanda surfaces are studied. Numerical
flow field analysis is used to identify performance sensitivities and explore the design
parameter space. Directions of future research needed to mature this technology for
transport aircraft applications are identified and discussed.
1 Introduction
The importance of high lift coefficients for the flight performance of transport aircraft
is well known. The state of the art in designing mechanical high lift systems allows
sizing the clean wing for optimum cruise performance given the operational require-
ments and overall configuration design constraints. Remarkable progress has been
achieved with these conventional high lift systems in terms of decreasing mechanical
complexity, weight and enhanced lift potentials. Typical maximum lift coefficients
for a three-element slotted wing section in landing configuration are now around 4.
They yield overall aircraft lift coefficients for a modern swept wing transport of
around 2.6 which is about twice the number obtained for the clean configuration.
A Vision of future air transport, however, has to address the mobility requirements
of highly developed societies. A new class of innovative transport systems is foreseen
for operation in densely populated areas to directly link city centers while providing
maximum flexibility with respect to changing mobility demands. The technological
challenges for the realization of such a system reach beyond those usually occurring
in the design of today’s aircraft. Noise emissions have to be brought down to a level
compatible with urban community noise to enable day and night operations. To estab-
lish airports close to city centers, runway lengths have to be reduced to a minimum,
thus Short Take-Off and Landing (STOL) performance of the aircraft becomes man-
datory. As a requirement to be met under all circumstances, the operation has to com-
ply with or even surpass today’s safety standards.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, pp. 168–189.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Flow Analysis of Augmented High-Lift Systems 169
The STOL capabilities as a prerequisite for minimum runway lengths can only be
achieved by pursuing innovative high-lift concepts even if extreme lightweight air-
frame materials and structures are used. Figure 1 displays the effects of wing loading,
engine thrust installed and maximum lift coefficients of the start configuration of a
modern transport aircraft on the start length. Assumed data for a typical 150 seat air-
craft are highlighted as a reference. Reducing the wing loading deteriorates flight
performance, as aircraft weight, lift and drag largely increase. Installing larger engine
thrust would mainly increase A/C cost and hence operating cost. In computing varia-
tions of the maximum allowed lift coefficient reasonable estimates of induced drag
have been defined along with a suited function of effective airfoil drag versus maxi-
mum lift that represents state-of-the-art blown flap systems as described in Chapter 4.
Moreover, current safety factors that keep the take-off speed well above the stalling
speed are also taken into account. Note that the rather low reference maximum lift
coefficient is dictated by current one-engine climb rate requirements. It becomes
obvious that substantial reductions of runway lengths are only feasible by both,
increasing the maximum lift coefficients by significant factors and moderately in-
creasing installed thrust. This calls for revolutionary concepts that most likely use
synergisms between lift and propulsion systems.
2000 12,5
1600 10
Start Length / m
1200 7,5
L/D
800 5
Fig. 1. Effects of wing loading, thrust-to-weight ratio and maximum lift coefficient on start
length (rolling distance plus transition length). Hollow symbols denote reference aircraft.
Advantages of using slotted wing shapes and blowing to improve lift are known
since the late 1920s [1–3]. Numerous powered lift concepts have been explored since,
over the past seventy years, to obtain wing circulation above values possible for non-
powered wings. For many years Hermann Schlichting supported research on powered
lift concepts and reported about the results [4], which were strongly stimulated by the
advent of jet engines after the Second World War. Figure 2 shows several concepts
that were investigated in detail and even flight tested, according to [5]. While classical
boundary layer control by blowing and suction, Figure 2a, aims to achieve the lift
coefficients theoretically possible for unpowered wings, several powered lift concepts
aim for lift coefficients beyond that limit, thereby employing super circulation.
It was found that the lift increase yielded by jet flaps, Figure 2b, where blown air is
typically directed downwards by about 30 deg, is rather limited because the jet exits
170 R. Radespiel, K.-C. Pfingsten, and C. Jensch
from the airfoil surface at a fixed angle. Three concepts of Figure 2c use the approach
of externally blown flaps by the propulsion exhaust. These approaches share the po-
tential of medium powered lift coefficients achievable for a given engine thrust-to-
weight ratio [6]. The externally upper surface blown flap is quite attractive since the
jet engine exhaust can be re-distributed along a larger part of the wing span, and the
noise of upper surface blowing is shielded by the wing. However, the external blow-
ing of highly curved flaps comes with mechanical complexity [7]. Thrust losses due
to engine air ducting are encountered and wave drag increases drastically at high
speed flight [8]. This solution is therefore restricted to specific configurations domi-
nated by STOL requirements.
Much higher powered lift augmentation for a given thrust-to-weight ratio may be
achieved by using internally flow blown flaps or flows over Coanda surfaces close to
the trailing edge known as circulation control. The functional principle of circulation
control is blowing from a slot ahead of the curved Coanda surface in order to move the
separation point in the downstream direction, possibly around the rounded trailing edge
onto the bottom surface. Thus the circulation of the airfoil can be adjusted by varying
the amount of blowing. Circulation control has several interesting features: a) High
force augmentation per input blowing momentum, b) Small flap sizes or even no mov-
ing aerodynamic surfaces required, c) Large lift at low angles of attack, d) high poten-
tials for generating aircraft control moments. It is felt that the two approaches of
Flow Analysis of Augmented High-Lift Systems 171
Figure 2d and their combinations have good potentials for powered lift of future trans-
port aircraft. Existing knowledge on these systems will therefore be reviewed in the
following chapter, where the focus is on data useful for transport aircraft applications.
The jet remains attached to the surface because of its entrainment capabilities
(Coanda effect) and low pressures occur at the curved surface due to the balance of
centrifugal and pressure forces across the wall jet. Most important flow parameter is
the blowing momentum coefficient
where m & is the blown mass flow rate and Vj denotes the jet velocity at the slot posi-
tion. The change of lift with momentum coefficient in Figure 3b indicates that blow-
ing with small momentum coefficients to about 0.1 functions as boundary layer
control and lift augmentations up to 80 times the value of cµ have been reported. The
rear stagnation point moves beyond the trailing edge on the lower surface at larger
momentum coefficients and super circulation is obtained, however with decreasing
lift augmentation rates. Figure 4 displays typical results achieved for single slotted
symmetric and cambered elliptic airfoils as concluded by Englar [12] who could also
demonstrate good lift-over-drag performance with blowing. Englar’s early work also
identified target regions for the geometrical parameters of effective Coanda blowing,
which are the slot height h, and the radius of surface curvature r, versus chord length
C [13], according to Figure 5. In general rather small slot heights improve lift at con-
stant cµ because larger jet velocities enhance flow entrainment.
172 R. Radespiel, K.-C. Pfingsten, and C. Jensch
t/c = 0.20
6
f/c = 0.05
l clC
t/c = 0.15
3 f/c = 0
0 .120
Present DR flap
0 .100
r/C=2% r/C=5%
0 .075
Thomas [14]
h/r
0 .050
0 .025
Fig. 5. Effective Coanda performance for different radius and slot heights (dashed area accord-
ing to Englar [13])
Internally blown flaps may be viewed as a special case of circulation control since
they use the Coanda effect at their curved leading edge if large flow turning angles are
to be accomplished. However, the turning for a given flap setting is limited and then
the resulting lift increase with blown momentum above the value needed for avoiding
flow separation is rather low, similarly as for the jet flap. Detailed early investigations
into the potentials of blown 26% flaps were undertaken by Thomas [14], where flap
angles and momentum parameters were varied at Reynolds numbers up to 0.8·106.
Thomas reported an increased maximum lift coefficient of ∆cl,max = 1.8 for cµ = 0.075
at flap deflections of 60 deg versus the non-blown case without deflection, and the
gain factor by blowing was around 13. Note that the maximum lift increases reported
in the assessments of this review are generally obtained versus the reference configu-
ration without deflection. The reported gain factor is the ratio of ∆cl,max/cµ for fixed
flap deflection. Inspection of Figure 5 indicates that one reason for the limited gain
Flow Analysis of Augmented High-Lift Systems 173
a) Movable plenum solution with track [16] b) Fixed plenum solution with 2 moving parts
factor was the relatively large slot height, furthermore the airfoil Reynolds number
was quite low. Thomas also derived an engineering flow model that can be used to
predict the momentum coefficient needed to avoid flow separation on blown flaps.
Additional gains in maximum lift by leading edge blowing were later reported by
Gersten and Löhr [15], for similar experimental setups, however the gain factor
achieved with this type of leading edge protection was even lower. Other results for a
blown 23% high lift flap by Englar [19] indicate larger gains possible, similarly to the
data reported in [4].
The elliptical circulation control airfoils share the problem that drag is high at
lower lift coefficients (cruise) where powered lift is not needed. Early solution ap-
proaches to this problem during the 1970s used mechanical devices that would con-
vert the wing’s sharp trailing edge into a circulation control wing. Figure 6a displays a
variant where the pressurized plenum is moved to the trailing edge using a track sys-
tem and a thin flap is used for guiding the flow along the lower airfoil surface with
circulation control. This concept resulted in relatively large lift coefficients for short
flap lengths. Compared to the blown flap of above this is a promising approach for
high turning/high cµ requirements, however it is mechanically complex and probably
not useful for large aspect ratio wings. A mechanically simpler variant was also de-
veloped as shown in Figure 6b, using a single hinge that pays a penalty of losing wing
area in the high-lift mode.
An alternate solution to the problem is displayed in Figure 7a where the bluntness
of the single-slot Coanda surface at the trailing edge is reduced to 2% to achieve
lower cruise drag for a 17% supercritical airfoil. Maximum lift increase by 2.4 is
reported for cµ = 0.1 [17], [19], with a gain factor of 24. However, the drag penalty in
cruise seems to be around 2·10-3 which can be possibly alleviated by allowing for
small blowing rates during cruise. This concept was further explored for general avia-
tion airfoils by Jones [18] who modified the 17% GA(W)-1 airfoil section by adding
different Coanda surfaces with two slots and a total of 4% bluntness at the trailing
edge, see Figure 7b. Jones reported an increase of maximum lift by 2.0 for cµ = 0.075
174 R. Radespiel, K.-C. Pfingsten, and C. Jensch
d/c = 0.04
a) Single slot trailing edge for super critical airfoil b) Double-slot Coanda surface on
GA(W)-1 airfoil for cruise drag
alleviation
Fig. 7. Fixed blunt trailing edge solutions with reduced size for lower cruise drag
which translates into a gain factor of 26. He used weak blowing through both slots to
reduce equivalent drag values at cruise conditions by a factor of 2.5. Pulsed blowing
was successfully tested in order to reduce mass flow requirements.
Another solution to the drag problem was given by Englar [19] who introduced the
dual radius flap, according to Figure 8. Here, a flap of less than 10% chord length pivots
about a hinge at the lower surface thereby opening a rather small radius of surface cur-
vature. The upper surface of the flap has a somewhat larger curvature and is not too far
away from the original reference airfoil. Thus the Coanda effect can be exploited up to
high flow turning angles if needed. Maximum lift is increased by 2.5 for cµ = 0.075 by
this device with a gain factor of 22, according to [12]. This concept was also used in
recent work at NASA Langley [20], where a 7% dual radius flap was integrated with the
GA(W)-1 airfoil. This work focuses on experimental techniques and flow field analysis
and it determines requirements for detailed flow field measurements.
Further exploration went into the use of circulation control for control purposes.
Testing of dual slotted cambered airfoils showed that the individual slots yielded the
desired control authority in both upper and lower directions and hence the range of
control doubled [21]. Later it was found that two slots can help to prevent excessive
turning of the primary wall jet at larger values of cµ that would otherwise cause large
lift losses [22].
In conclusion it is felt that the aerodynamic design of future STOL transport air-
craft configurations requires more detailed work in the areas of
• integrating powered lift approaches with high gain factor potential into modern
transonic airfoil section at high Reynolds numbers.
• detailed further exploration of powered lift concept with rather low cµ, which is
desired if bleed air from the bypass of modern jet engines is used.
• establishment of detailed experimental aerodynamic data sets that can be used to
validate numerical flow simulation approaches.
Flow Analysis of Augmented High-Lift Systems 175
Since the 2004 Workshop a further experimental set of data that includes detailed
flow field measurements by Novak [24] has become the target of validation activities
and some recent results will be displayed below. The model with a round trailing edge
for the wind tunnel experiments had a chord length of c = 0.38 m and a span of
b = 0.61 m. The height of the slot used for blowing was h/c = 0.002. Momentum coef-
ficients of cµ = 0.03 and cµ = 0.1 were investigated. The windtunnel experiments were
accomplished for Re = 106, Ma = 0.12 and the geometric angle of attack was zero. In
addition to the surface pressure distribution Novak measured velocity profiles on the
Coanda surface by using Laser Doppler Velocimetry. The measured surface pressure
distributions were used to compute the integrated force coefficients in tangential and
normal directions, with respect to the chord of the model. Due to the interaction of the
sidewall boundary layer with the model, vortical structures were generated. These
three-dimensional structures induced a strong downwash along the span, which re-
duced the effective angle of attack.
The flow simulations of the present paper are performed by solving the Reynolds-
averaged Navier-Stokes equations using the DLR hybrid unstructured flow solver TAU
176 R. Radespiel, K.-C. Pfingsten, and C. Jensch
[25] which is based on a finite volume scheme for spatial discretization. The code proc-
esses meshes with different types of cells and combines the advantages of structured
grids to resolve boundary layers with the flexible grid generation of unstructured grids.
To accelerate the convergence to steady state, local time stepping, residual smoothing
and a multi-grid technique based on agglomeration of the dual-grid volumes are avail-
able. All computations are undertaken assuming the boundary layer to be fully turbulent.
The Spalart-Allmaras (SA) turbulence model [26] is used for the present computa-
tions. In the standard model the eddy viscosity, vt, is a function of v which is deter-
mined from a transport equation:
∂ ( ρν~ ) ∂ ( ρu iν~ )
+ = P + Ddiff Ddiss . (2)
∂t ∂xi
The terms on the right hand side represent production, gradient diffusion and the near-
wall dissipation of the turbulent kinematic viscosity. In the extended Spalart-Allmaras
model for rotation and/or curvature effects (SARC) the production term is multiplied
by the rotation function fr1:
~
P = cb1 ρS ν~f r1 . (3)
The function fr1 takes the effects of strain-rate/rotation, the Lagrangian derivatives of
the strain tensor axis angle and the system rotation into account [27]. A simplified
SARC model called SSARC based on unpublished work of Spalart [28] is also im-
plemented in the TAU code. The simplifications therein are based on the assumption
of locally vanishing streamwise derivatives of the velocity vector.
The comparisons given below include flow field data, force coefficients and the sepa-
ration point on the Coanda surface. The position of the separation is indicated by the
angle θsep which is counted from the beginning of the circular trailing edge on the
upper surface.
Figure 9 displays the mesh for the numerical simulation around the Novak profile
of ref. [24].
For all computations the total number of points is about 105. The structured area
close to the surface consists of 45 cells. In the vicinity of the jet slot the grid for the
cµ = 0.03 cµ = 0.1
0.1 0.1
z/c
z/c
0.0 0.0
-0.1 -0.1
0.8 0.9 1.0 1.1 0.8 0.9 1.0 1.1
x/c x/c
Figure 10 shows the measured flow field around the wind tunnel model for two dif-
ferent momentum coefficients. Separation from the Coanda surface occurs in the ex-
periments at θsep ≈ 85 deg for cµ = 0.03 and at θsep ≈ 115 deg for cµ = 0.1. These results
are to be compared to numerical results of the SA and SARC models in Figure 11.
a) SA turbulence model
The SA model results in too large separation angles of 91 deg and 134 deg, respec-
tively, whereas the SARC model gives 84 deg and 114 deg which is much closer to
the experiments. Quite similar results are reported by Swanson [29]. Note further that
Fig. 12. Comparison of curvature effects on the computed turbulent viscosities around the
trailing edge
a) θ = 0° b) θ = 80°
c) θ = 110° d) θ = 120°
Fig. 13. Near-wall velocity profiles along the Coanda surface for cµ = 0.10 (angle θ in circum-
ferential direction counts from slot location)
Flow Analysis of Augmented High-Lift Systems 179
the Spalart-Shur curvature correction does increase turbulent viscosities in the free
shear layer of the wall jet, whereas viscosities are reduced very close to the wall,
according to Figure 12. This combination results in earlier separation, compared to
the standard model. Note that the simplified turbulence model SSARC yields results
very close to SARC.
Quantitative details about the ability of the simulation method to predict the near-
wall flow behavior is assessed by plotting mean velocity profiles. Figure 13 gives a
sample result for cµ = 0.1.
It is seen that the turbulence model with curvature correction predicts the develop-
ment of the wall-jet reasonably well, whereas the standard model computes too little
mixing of the jet shear layer. Improved simulation accuracy with curvature effects
included is also appearent from the pressure distribution, Figure 14, where the too
Fig. 15. Development of normal force coefficient with blowing for the Novak configuration in
the windtunnel using corrected angles of attack
180 R. Radespiel, K.-C. Pfingsten, and C. Jensch
large turning angle of the flow simulation with the standard SA model results in an
overprediction of suction along the whole upper surface. With the SARC model there
is still significant overprediction of suction along the Coanda surface which indicates
that there are some remaining differences in the force coefficients as well. These
differences are finally seen in Figure 15 where the resulting normal force coefficients
are displayed including the results of Swanson [29]. The gains of normal force with
blowing is reasonably well predicted taking into account the relatively large wind
tunnel interference effects to the nominally 2D flow over the airfoil.
Ma∞ Re
Start 0.21 29·106
Cruise 0.72 36·106
Fig. 16. Reference transonic airfoil (not to scale) and flow parameters used for investigations of
Chapter 4
Flow Analysis of Augmented High-Lift Systems 181
4.1 Geometry
Following the above design requirements the generation of airfoil variants with circu-
lation control that keep the characteristics of the basic profile are now investigated.
The x-wise position of the slot is defined together with the length of the flap, since the
slot is positioned directly upstream of the movable surface. Upstream of this position
the original upper surface is used. Downstream of the vertical slot the upper surface of
the flap is reshaped.
There are two distinct design drivers for the flap geometry. On the one hand the
Coanda radius should be large to enable the Coanda effect. On the other hand the
profile should not exhibit additional drag at cruise flight, which means it needs a thin
trailing edge and the new profile should be rather close to the shape of the reference
profile, at the cruise setting of the device.
Then there are three general options of designing the upper surface of the flap, as
sketched in Figure 17. In version A the Coanda radius is hidden in the profile whilst
in cruise flight. With some geometrical refinements of the knuckle shape the circula-
tion control profile and the basic airfoil can be made virtually identical at cruise con-
ditions. If the flap is deflected for take-off and landing the Coanda radius appears
downstream of the slot.
In version B the the upper surface downstream of the flap is translated downwards
to create the slot. Therefore, in cruise the upper surface of the flap is slightly lower
and at the position of the slot is a small backward facing step. As for version A the
Coanda radius is hidden when the flap is not deflected. Note that the axial position of
the hinge line can be used to obtain continuity in surface slope with version A and B,
not shown in Figure 17.
C
Fig. 17. Geometric design options for a blown flap with circulation control
Version C describes a dual-radius flap, which was first shown by Englar [19]. Here
the upper surface of the flap is composed out of two circle segments. The first circle
segment is inside the geometry whilst in cruise. It is always chosen as large as possi-
ble, so the radius is about the local thickness of the profile at the slot position, as for
versions A and B. The second radius is used to achieve additional turning thereby
achieving higher lift. Here its radius is chosen in a way that the position of the trailing
edge is unchanged. As the second segment is from a very large circle, the new shape
of the upper surface stays close to the original contour.
182 R. Radespiel, K.-C. Pfingsten, and C. Jensch
0.1 0.1
0 0
-0.1 -0.1
Large high-lift flap Dual-radius flap
-0.2 -0.2
0.6 0.8 1 0.6 0.8 1
Fig. 18. Investigated flap geometries
Fig. 19. Typical mesh for transonic airfoil with large blown high-lift flap
A large high-lift flap with a flap length of cflap/c = 0.3 is here designed according to
version B to numerically assess the benefits of a simple blown flap device as a refer-
ence. Furthermore, a dual-radius flap (version C) is realised with a relative flap length
of cflap/c = 0.07. Both flap geometries are given in Figure 18. A typical computational
mesh with 0.7·105 points used is shown in Figure 19.
For a slot height of h/c = 0.001 and maximum slot flow total pressure, pt,jet/p∞ = 1.68,
the momentum coefficient is cµ = 0.043. The lift curves of the large high-lift flap
using this setting in Figure 20 indicate that lift coefficients of around 4.5 are feasible.
Significantly higher deflection angles than 60 deg need larger momentum coefficients
and hence larger total pressures to avoid flow separation on the flap, see section 4.3
below. It is also seen that higher deflection angles shift the angle of attack with
maximum lift to smaller values. By doubling the slot height the maximum lift is in-
creased by about 7%. However, the smaller slot height is the preferred solution, be-
cause only half the amount of bleed air is needed. Figure 21 indicates that this design
Flow Analysis of Augmented High-Lift Systems 183
Fig. 20. Lift coefficients for the large high-lift flap: Effects of flap deflection and slot hight,
cµ = 0.043
Fig. 21. Lift coefficients for large high-lift flap: Effect of reducing slot total pressure below
pt,jet/p∞ = 1.68
enjoys a safety margin of 20% slot total pressure against boundary layer separation at
a deflection angle of 40 deg. Note that the blown flap does not generate significantly
higher pitching moments than a modern high-lift airfoil with slat and Fowler flaps.
The assessment of computed drag values must take the loss of thrust due to bleed
air extraction from the engine into account. Here we assume that the blown momen-
tum coefficient is equal to the thrust loss of the engine. Hence the cl/cd,eff values
shown in Figure 22 use cd,eff = cd + cµ. It is seen that a flap deflection angle of 20 deg
is a good choice for aircraft take-off and climb whereas 60 deg should be used for
landing.
In conclusion the large high-lift flap yields increased maximum lift coefficients by
2.6 for Cµ = 0.043 with a gain factor due to blowing of about 60. This is significantly
184 R. Radespiel, K.-C. Pfingsten, and C. Jensch
Fig. 22. Effects of flap deflection angles and slot height on lift-over-effective-drag values
larger than the gain factors observed for similar blowing rates in the literature, albeit
those were obtained at lower Reynolds numbers.
A comparison of the lift potential for the large flap and the dual radius flap is
shown in Figure 23. It can be seen that the much shorter and lighter dual-radius flap
achieves maximum lift coefficients of 3.2 for cµ = 0.043. The dual radius flap obtains
increased maximum lift coefficients by 1.4 for cµ = 0.043 with a gain factor due to
blowing of about 30. It is speculated that the choice of a larger first Coanda radius
would deliver somewhat better aerodynamic performance, since the present solution
is rather far away from the optimum parameter range, according to Figure 5. How-
ever, this would require a lower hinge line position outside of the airfoil contour with
some cruise drag and weight penalties.
The relative efficiencies of the present blown designs are also highlighted by Fig-
ure 24 which compares the lift due to blowing for the present solutions and data pub-
lished previously.
Fig. 23. Comparison of lift coefficients for large high-lift flap and dual-radius flap, cµ = 0.043
Flow Analysis of Augmented High-Lift Systems 185
Detailed analysis of the airfoil stall behavior is obtained from Figures 25 and 26,
which display computed Mach number contours and streamlines at angles of attack
beyond maximum lift. It is seen that the Coanda jets still follow the flap contour
whereas low flow velocities in the wake from the main wing occur further away from
the surface (free flow separation). This phenomenon decreases the effective flow
turning thereby reducing the overall lift.
The present results for the large high-lift flap (design version B of Figure 17) are
also applicable for a flap of type A, because the high-lift characteristics of versions A
and B are almost identical. The cruise performance of a version A flap is identical to
the performance of the reference airfoil, as in cruise the geometries are virtually the
same. Thus the high-lift performance of the investigated large high-lift flap can be
Fig. 25. Stall analysis of large high-lift flap Fig. 26. Stall analysis of dual-radius flap
η = 40°, α = 6° η = 60°, α = 12°
186 R. Radespiel, K.-C. Pfingsten, and C. Jensch
achieved without degrading the cruise performance assuming a flap of version A can
be generally used.
For an assessment of the cruise behaviour of the dual-radius flap the small remain-
ing slot of the zero-deg. configuration is closed with a straight cap. While the refer-
ence airfoil has a drag coefficient, cd = 0.0073, at its design point (Ma∞ = 0.72,
cl = 0.4), drag of the dual-radius flap is cd = 0.0075. It is felt that this relatively small
increase could be further diminished by adjusting the flap deflection angle and opti-
mizing the slot cap geometry. Thus the requirement of vanishing cruise drag penalty
can be fullfilled with both high-lift designs.
The use of internally blown flaps with extended lift capability beyond the range of
classical slotted high-lift systems is finally investigated. We address two effects that
have to be well understood when moving to higher lift coefficients. The first phe-
nomenon is the occurrence of transonic stall and the second relates to subsonic lead-
ing edge stall.
The occurrence of transonic flow at the leading edge is highlighted in Figure 27 for
the large high-lift flap with deflection angles of 60 deg.
The circles on the lift curves indicate the first angle of attack with supersonic
points in the flow field. It is seen that transonic stall is obtained at the Mach number
of 0.21, whereas the airfoil exhibits subsonic stall at Ma∞ = 0.15. The subsonic stall
occurs gentler, and at somewhat larger angles of attack as compared to transonic stall.
Fig. 27. Effects of free stream Mach number and leading edge blowing on lift coefficients for
large high-lift flap with η = 60°
Flow Analysis of Augmented High-Lift Systems 187
edges is recommended in order to delay transonic stall and to open the way for achiev-
ing higher lift coefficients as compared to present conventional slotted designs.
An important aspect for future application of augmented lift in transport aircraft
concerns the required momentum coefficients. Significant smaller mass flow rates as
predicted in the present work could make the use of alternate aircraft systems for
supplying compressed air feasible. This would significantly ease the problem of flight
safety, compared to engine bleed air solutions. It is therefore suggested to investigate
actuated and dynamical Coanda blowing devices and geometrically variable flaps.
The present review also reveals that the existing data base of detailed flow field
measurements is rather limited. It is proposed to perform validation experiments of
representative geometries that include the effects of exciting the shear-layer instabili-
ties of the curved wall jet. The corresponding validation work should resolve the still
open question whether second-order RANS closures are suited for future aerodynamic
design work on augmented high-lift systems or wall-modeled LES are needed to de-
scribe the relevant features of these turbulent shear flows.
Finally, there is the need for detailed research on flows around 3D wings with effi-
cient augmented high-lift systems. This research should address the effects of wing
sweep on the stalling behavior as well as on advanced lift augmentation schemes to
obtain high gain factors due to blowing for real applications with finite span.
References
[1] Wings with Nozzle Shaped Slots. NACA Translation TM 521, July 1929, from Berichte
der Aerodynamischen Versuchsanstalt in Wien 1(1) (1928)
[2] Bamber, M.J.: Wind tunnel tests on airfoil boundary layer control using a backward-
opening slot. NACA Report 385 (1932)
[3] Hagedorn, H., Ruden, P.: Windkanaluntersuchungen an einem Junkers- Doppelflügel mit
Ausblaseschlitz am Heck des Hauptflügel. Bericht A64 der Lilienthal-Gesellschaft für
Luftfahrtforschung (1938)
[4] Schlichting, H.: Aerodynamische Probleme des Höchstauftriebes. Z. Flugwiss. 13(1), 1–
14 (1965)
[5] Nielsen, J.N., Biggers, J.C.: Recent Progress in Circultion Control Aerodynamics. AIAA
Paper 87-001 (1987)
[6] Yaros, S.F., et al.: Synergistic Airframe-Propulsion Interactions and Integrations.
NASA/TM-1998-207644 (1998)
[7] Stephens, V.C., Riddle, D.W., Martin, J.L., Innis, R.C.: Powered-Lift STOL Aircraft
Shipboard Operations – A Comparison of Simulation, Land-Based and Sea Trial Results
for the QSRA. AIAA Paper 81-2480 (1981)
[8] Pfingsten, K.-C., Radespiel, R.: Use of upper surface blowing and circulation control for
gapless high lift configurations. In: CEAS/KATnet Conference on Key Aerodynamic
Technologies, Bremen, June 20–22 (2005)
[9] Davidson, I.M.: Aerofoil Boundary Layer Control System. British Patent No. 913, 754
(1960)
[10] Kind, R.J., Maull, D.J.: An Experimental Investigation of a Low-Speed Circulation Con-
trolled Airfoil. The Aeronautical Quarterly, vol. XIX, pp.170–182 (May 1968)
Flow Analysis of Augmented High-Lift Systems 189
[11] Englar, R.J., Applegate, C.A.: Circulation Control – A Bibliography of DTNSRDC Re-
search and Selected Outside References (January 1969 through December 1983).
DTNSRDC-84/052 (September 1984)
[12] Englar, R.J.: Overview of Circulation Control Pneumatic Aerodynamics: Blown Force
and Moment Augmentation and Modification as Applied Primarily to Fixed- Wing Air-
craft. In: Joslin, D., Jones, G.S. (eds.) Applications of Circulation Control Technology,
Progress in Astronautics and Aeronautics, vol. 214, AIAA (2006)
[13] Englar, R.J., Williams, R.M.: Design of Circulation Controlled Stern Plane for Submarine
Applications. DTNSRDC, Technical Note AL-200, AD901-198 (1971)
[14] Thomas, F.: Untersuchungen über die Grenzschichtbeeinflussung durch Ausblasen zur
Erhöhung des Auftriebes. Dissertation, Institut für Strömungsmechanik, TU Braun-
schweig (1961); Gekürzte Fassung: Z. Flugwiss., vol. 10, pp. 46–65 (1962)
[15] Gersten, K., Löhr, R.: Untersuchungen über die Auftriebserhöhung eines Tragflügels bei
gleichzeitigem Ausblasen an der Hinterkantenklappe und an der Profilnase. Instituts-
bericht 62/34, Institut für Strömungsmechanik der Technischen Universität Braunschweig
(1962)
[16] Englar, R.J.: Circulation Control for High Lift and Drag Generation on STOL Aircraft.
AIAA Journal of Aircraft 12(5), pp. 457–463 (1975)
[17] Englar, R.J.: Low Speed Aerodynamic Characteristics of a Small Fixed Trailing Edge
Circulation Control Wing Configuration Fitted to a Supercritical Airfoil. DTNSRDC,
Rept. ASED-81/08 (March 1981)
[18] Jones, G.S.: Pneumatic Flap Performance for a Two-Dimensional Circulation Control
Airfoil. In: Joslin, D., Jones, G.S. (eds.) Applications of Circulation Control Technology,
Progress in Astronautics and Aeronautics, vol. 214, AIAA (2006)
[19] Englar, R.J., Huson, G.G.: Development of Advanced Circulation Control Using High-
Lift Airfoils. Journal of Aircraft, 21(7), pp. 476–483 (1984)
[20] Jones, G.S., Yao, C.-S., Allan, B.G.: Experimental Investigation of a 2D Supercritical
Circulation-Control Airfoil Using Particle Image Velocimetry. AIAA Paper 2006-3009
(2006)
[21] Abramson, J.: Characteristics of a Cambered Circulation Control Airfoil Having Both
Upper and Lower Surface Trailing Edge Slots. NSWCCD-50-TR-2004/030 (April 2004)
[22] Rogers, E.O., Donnelly, M.J.: Characteristics of a Dual-Slotted Circulations Control
Wing of Low Aspect Ratio Intended for Naval Hydrodynamic Applications. AIAA Paper
2004-1244 (2004)
[23] Joslin, R.D., Jones, G.S.: Applications of Circulation Control Technology. In: Progress in
Astronautics and Aeronautics, vol. 214. AIAA (2006)
[24] Novak, C.J., Cornelius, K.C., Roads, R.K.: Experimental Investigations of the Circular
Wall Jet on a Circulation Control Airfoil. AIAA Paper 87-0155 (1987)
[25] Gerhold, T.: Overview of the hybrid RANS code TAU. In: MEGAFLOW – Numerical
Flow Simulation for Aircraft Design. Notes on Numerical Fluid Mechanics and Multidis-
ciplinary Design, vol. 89, pp. 81–92. Springer, Heidelberg (2005)
[26] Spalart, P.R., Allmaras, S.R.: A One-Equation Turbulence Model for Aerodynamic
Flows. AIAA Paper 92-0439 (1992)
[27] Spalart, P.R., Shur, M.: On the Sensitization of Turbulence Models to Rotation and Cur-
vature. Aerospace Science and Technology 5, 297–302 (1997)
[28] Spalart, P.R.: Evaluation of Simplified Rotation/Curvature Correction. Personal Commu-
nication (2006)
[29] Swanson, R.C., Rumsey, C.L., Anders, S.G.: Progress towards Computational Method for
Circulation Control Airfoils. AIAA Paper 2005-0089 (2005)
Dankesrede der Familie Schlichting zum Abschluss des
wissenschaftlichen Kolloquiums am 28. Sept. 2007 aus
Anlass des 100. Geburtstages von Prof. Dr. Hermann
Schlichting in der Aula der TU Braunschweig
R. Schlichting
Im Namen der Familie Schlichting möchte ich dafür danken, dass aus Anlass des 100.
Geburtstages unseres Vaters, Schwiegervaters, und Großvaters heute dieses Gedächt-
nis-Kolloquium stattgefunden hat.
Drei der sieben Enkelkinder sind heute auch hier. Sabine, Meike und Sven sitzen
hier vorne nebeneinander. Danke, dass ihr es möglich gemacht habt dabei zu sein, das
ist nicht selbstverständlich! Die Kinder und Schwiegerkinder sind vollzählig anwe-
send – so ist das bei den Schlichtings!
Ein besonderer Dank geht an alle, die diese Veranstaltung so vorbildlich organis-
iert haben. Den größten Anteil zum Gelingen haben meines Wissens Sie, lieber Herr
Hummel, beigetragen – dafür ganz herzlichen Dank!
Ein weiteres Dankeschön auch allen Referenten! Ihre Beiträge haben diese Veran-
staltung zu einem inhaltsreichen Kolloquium gemacht, was ganz bestimmt auch im
Sinne unseres Vaters gewesen wäre!
Wir sind dankbar, dass das wissenschaftliche Werk unseres Vaters bis zum heuti-
gen Tage nachwirkt und dass viele Weggefährten unseres Vaters das durch ihre Teil-
nahme an diesem Kolloquium zum Ausdruck gebracht haben.
Schon lange haben Powerpoint Präsentationen die quadratischen Dias abgelöst.
Dennoch war heute noch einmal der Schlichting'sche Geist zu spüren, der wohl vor
allem in den 50er und 60er Jahren die Arbeit an Institut und Hochschule geprägt hat.
In diesem Jahr ist die Stadt Braunschweig zur Stadt der Wissenschaft gewählt
worden. Es ist eine glückliche Fügung, dass der runde Geburtstag unseres Vaters
genau in diese Zeit fällt. Dadurch bekamen die beiden Institute der TU und des DLR
die Chance sich aus diesem Anlass zu präsentieren – was meines Erachtens auch sehr
gut gelungen ist.
Gestatten sie mir zum Schluss noch einen kleinen Vergleich zwischen Carl Frie-
drich Gauß und Hermann Schlichting. Es wird immer wieder die Frage gestellt, ob
Gauß nun ein Göttinger oder ein Braunschweiger Wissenschaftler gewesen ist. Für
unseren Vater ist diese Frage – aus unserer Sicht – spätestens seit heute beantwortet.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, p. 190.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Vote of Thanks by the Family Schlichting, Terminating
the Scientific Colloquium on September 28, 2007 at the
Occasion of the 100th Anniversary of the Birthday of
Prof. Dr. Hermann Schlichting
R. Schlichting
Transcript by D. Hummel
On behalf of the family Schlichting I would like to express our sincere thanks for the
organization of this colloquium in memory of our father, father-in-law and grandfa-
ther at the occasion of his 100th birthday.
Three of the seven grandchildren are with us today: Sabine, Meike and Sven are
sitting right in front side by side. Thanks to you three for having made it possible to
attend – this is not self-evident! All children and children-in-law are present in the
audience – as it is common practice in the Schlichting family!
Special thanks may be addressed to those who have excellently organized this col-
loquium. As far as I know the most important part of the success is due to your con-
tribution, dear Mr. Hummel – hearty thanks for all this.
Further thanks may be addressed to the speakers! Your contributions made this col-
loquium an important scientific event, and I am sure this would have been to the mind
of our father.
We are grateful to notice that the work of our father is still actively present in to-
day’s science, and that many fellow scientists of our father demonstrated this fact by
attending this colloquium.
Since a long time PowerPoint presentations have replaced the quadratic Schlicht-
ing-slides. But nevertheless today once again we experienced the Schlichting spirit,
which played an important role during the 1950ies and 1960ies for the work at the
Institute and in the University.
This year Braunschweig has been selected to be the German “City of Science” and
the 100th birthday of our father was an opportunity for both the TU Institute and the
DLR Institute to present their scientific programs to the public. From my point of
view this task has been solved very well.
Finally I would like to draw a small comparison between Carl Friedrich Gauß and
Hermann Schlichting. (Both lived and acted in Göttingen and in Braunschweig). For
Gauß the question has been often put forward as to whether his scientific home was in
Göttingen or in Braunschweig. For our father this question is – from our point of view
– definitely answered since today.
R. Radespiel et al. (Eds.): Hermann Schlichting – 100 Years, NNFM 102, p. 191.
springerlink.com © Springer-Verlag Berlin Heidelberg 2009
Author Index
Available Volumes
Volume 102: Rolf Radespiel, Cord-Christian Rossow, Benjamin Winfried Brinkmann (eds.):
Hermann Schlichting – 100 Years - Scientific Colloquium Celebrating the Anniversary of His Birthday,
Braunschweig, Germany 2007. ISBN 978-3-540-95997-7
Volume 101: Egon Krause, Yurii I. Shokin, Michael Resch, Nina Shokina (eds.): Computational Science and
High Performance Computing III - The 3rd Russian-German Advanced Research Workshop, Novosibirsk,
Russia, 23–27 July 2007. ISBN 978-3-540-69008-5
Volume 100: Ernst Heinrich Hirschel, Egon Krause (eds.): 100 Volumes NNFM and 40 Years Numerical
Fluid Mechanics. ISBN XXXXXXXXXX
Volume 99: Burkhard Schulte-Werning, David Thompson, Pierre-Etienne Gautier, Carl Hanson, Brian
Hemsworth, James Nelson, Tatsuo Maeda, Paul de Vos (eds.): Noise and Vibration Mitigation for Rail Trans-
portation Systems - Proceedings of the 9th International Workshop on Railway Noise, Munich, Germany, 4–8
September 2007. ISBN 978-3-540-74892-2
Volume 98: Ali Gülhan (ed.): RESPACE – Key Technologies for Reusable Space Systems - Results of a
Virtual Institute Programme of the German Helmholtz-Association, 2003–2007. ISBN 978-3-540-77818-9
Volume 97: Shia-Hui Peng, Werner Haase (eds.): Advances in Hybrid RANS-LES Modelling - Papers con-
tributed to the 2007 Symposium of Hybrid RANS-LES Methods, Corfu, Greece, 17–18 June 2007. ISBN
978-3-540-77813-4
Volume 96: C. Tropea, S. Jakirlic, H.-J. Heinemann, R. Henke, H. Hönlinger (eds.): New Results in Numer-
ical and Experimental Fluid Mechanics VI - Contributions to the 15th STAB/DGLR Symposium Darmstadt,
Germany, 2006. ISBN 978-3-540-74458-0
Volume 95: R. King (ed.): Active Flow Control - Papers contributed to the Conference “Active Flow Control
2006”, Berlin, Germany, September 27 to 29, 2006. ISBN 978-3-540-71438-5
Volume 94: W. Haase, B. Aupoix, U. Bunge, D. Schwamborn (eds.): FLOMANIA - A European Initiative
on Flow Physics Modelling - Results of the European-Union funded project 2002 - 2004. ISBN 978-3-540-
28786-5
Volume 93: Yu. Shokin, M. Resch, N. Danaev, M. Orunkhanov, N. Shokina (eds.): Advances in High Perfor-
mance Computing and Computational Sciences - The Ith Khazakh-German Advanced Research Workshop,
Almaty, Kazakhstan, September 25 to October 1, 2005. ISBN 978-3-540-33864-2
Volume 92: H.J. Rath, C. Holze, H.-J. Heinemann, R. Henke, H. Hönlinger (eds.): New Results in Numer-
ical and Experimental Fluid Mechanics V - Contributions to the 14th STAB/DGLR Symposium Bremen,
Germany 2004. ISBN 978-3-540-33286-2
Volume 91: E. Krause, Yu. Shokin, M. Resch, N. Shokina (eds.): Computational Science and High Perfor-
mance Computing II - The 2nd Russian-German Advanced Research Workshop, Stuttgart, Germany, March
14 to 16, 2005. ISBN 978-3-540-31767-8
Volume 87: Ch. Breitsamter, B. Laschka, H.-J. Heinemann, R. Hilbig (eds.): New Results in Numerical and
Experimental Fluid Mechanics IV. ISBN 978-3-540-20258-5
Volume 86: S. Wagner, M. Kloker, U. Rist (eds.): Recent Results in Laminar-Turbulent Transition - Selected
numerical and experimental contributions from the DFG priority programme ’Transition’ in Germany. ISBN
978-3-540-40490-3
Volume 85: N.G. Barton, J. Periaux (eds.): Coupling of Fluids, Structures and Waves in Aeronautics - Pro-
ceedings of a French-Australian Workshop in Melbourne, Australia 3-6 December 2001. ISBN 978-3-540-
40222-0
Volume 83: L. Davidson, D. Cokljat, J. Fröhlich, M.A. Leschziner, C. Mellen, W. Rodi (eds.): LESFOIL:
Large Eddy Simulation of Flow around a High Lift Airfoil - Results of the Project LESFOIL supported by
the European Union 1998 - 2001. ISBN 978-3-540-00533-9
Volume 82: E.H. Hirschel (ed.): Numerical Flow Simulation III - CNRS-DFG Collaborative Research Pro-
gramme, Results 2000-2002. ISBN 978-3-540-44130-4
Volume 81: W. Haase, V. Selmin, B. Winzell (eds.): Progress in Computational Flow Structure Interaction -
Results of the Project UNSI, supported by the European Union 1998-2000. ISBN 978-3-540-43902-8
Volume 80: E. Stanewsky, J. Delery, J. Fulker, P. de Matteis (eds.): Drag Reduction by Shock and Boundary
Layer Control - Results of the Project EUROSHOCK II, supported by the European Union 1996-1999. ISBN
978-3-540-43317-0
Volume 78: M. Hafez, K. Morinishi, J. Periaux (eds.): Computational Fluid Dynamics for the 21st Century.
Proceedings of a Symposium Honoring Prof. Satofuka on the Occasion of his 60th Birthday, Kyoto, Japan,
15-17 July 2000. ISBN 978-3-540-42053-8
Volume 77: S. Wagner, U. Rist, H.-J. Heinemann, R. Hilbig (eds.): New Results in Numerical and Experi-
mental Fluid Mechanics III. Contributions to the 12th STAB/DGLR Symposium, Stuttgart, Germany 2000.
ISBN 978-3-540-42696-7
Volume 76: P. Thiede (ed.): Aerodynamic Drag Reduction Technologies. Proceedings of the CEAS/DragNet
European Drag Reduction Conference, 19-21 June 2000, Potsdam, Germany. ISBN 978-3-540-41911-2
Volume 75: E.H. Hirschel (ed.): Numerical Flow Simulation II. CNRS-DFG Collaborative Research Pro-
gramme, Results 1998-2000. ISBN 978-3-540-41608-1
Volume 66: E.H. Hirschel (ed.): Numerical Flow Simulation I. CNRS-DFG Collaborative Research Pro-
gramme. Results 1996-1998. ISBN 978-3-540-41540-4