(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
For permission to copy or to republish, contact the copyright owner named on the first page. For AIAAheld copyright, write to AIAA Permissions Department,
1801 Alexander Bell Drive, Suite 500, Reston, VA, 201914344.
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
AIAA 20020555
WAKEINTEGRAL DETERMINATION OF AERODYNAMIC DRAG, LIFT AND MOMENT IN THREEDIEMENSIONAL FLOWS
J. C. Wu*
Applied Aero, LLC, 
Zephyr Cove, Nevada 
C. M. Wangt 

Applied Aero, LLC, 
Zephyr Cove, Nevada 
K. W. McAlistert
Army Aeroflightdynamics Directorate, Ames Research Center, Moffett Field, California
Abstract
New wakeintegral expressions for the determination of aerodynamic load on finite wings and rotors are established using a vorticitymoment theorem. Compared to previous wakeintegral expressions based on the momentum theory, the new expressions connect the wake flow properties more directly to the aerodynamic load. They offer enhanced physical understanding of the flow mechanisms responsible for the production of aerodynamic force and moment and are simpler and more efficient to use. Windtunnel experiments are performed to validate the wakeintegral expressions for the thrust and the torque on rotors in slow climb. A threedimensional particleimage velocimetry system is used to obtain velocity values in the nearwake of a model rotor. Thrust and torque values determined using the wake data are presented and compared with balancemeasured values.
1. INTRODUCTION
A lifting body in flight always leaves behind in the fluid a footprint  the wake. For more than a century, the aerodynamicist has searched for the connection between this footprint and the aerodynamic load on the body. L. Prandtl connected the down wash induced by trailing vortices  parts of the wake  to the induced drag on the finite wing. The profound contribution of the resulting liftingline theory to theoretical aerodynamics cannot be overemphasized. The research described in the present paper is centered on the wake integral approach, which also connects the wake to the aerodynamic load. This method, however, differs from the liftingline theory in that it focuses not on the downwash induced by the wake, but on the wake itself. The wakeintegral method does not require the inviscid fluid idealization and is useful in evaluating both the inviscid drag and the viscous drag.
A wake integral in a general context is an integral over a transverse surface downstream of a lifting solid body. For the present work, the term ^{4} wake integral' is used in a more restricted context to designate a special surface integral whose integrand vanishes outside the
vortical wake region. A. Betz
2
pioneered the wake
integral concept and successfully established a wake integral expression for the steady profile drag (also
* President, Associate Fellow t Chief Aerodynamicist t Research Scientist Copyright © 2002 by J. C. Wu. Published by the American Institute of Aeronautics and Astronautics, Inc. with permission
called the parasite drag ^{3} ) in 1925. E. C.
J. C. Wu et al.
5
Maskell ^{4} and
derived a wakeintegral expression for
the induced drag in the 1970s. These wakeintegral expressions allow the separate determination of the induced drag and the profile drag on the lifting body through wake surveys over a single wake plane. Since measurements are required only in small wake regions where the vorticity is nonzero, both the profile drag and the induced drag can be determined efficiently and accurately. The advantages offered by the method in design diagnostics are obvious.
Efforts have been in progress in recent years at several universities and governmental and industrial laboratories at various points of the world to further develop the wakeintegral method. Wind tunnel studies of many aerodynamic shapes of practical importance, including car shapes, have been performed using the method. In a recent review article ^{6} on drag prediction and reduction, I. Kroo referred to many recent efforts, noting that successes have been reported along with several open issues that require further investigations.
Previous studies of the wakeintegral method are mostly concerned with steady aerodynamic drag. The
present paper reports selected results of a research
program initiated in 1996 and completed
recently ^{7} .
The aim of the program is to generalize the wake integral method for unsteady flow applications, in particular helicopter rotor applications. Under this program, new wakeintegral expressions are derived for the finite wing. New wakeintegral expressions are also derived for the thrust and the torque on the rotor in axial flight, including hover. Windtunnel experiments
are performed to validate the rotor expressions.
1
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
2. VORTICITY MOMENT AND VORTEX LOOP
The conceptual foundation of the present research
is described in detail in a recent report . Several long
standing issues of viscous aerodynamics are examined from the vorticitydynamics viewpoint in the report ^{8} . In this paper, a vorticityloop method for aerodynamic analyses is described. This method is based on the
previously developed vorticitymoment theorem ^{9} . New wakeintegral expressions are derived for the finite
wing and for the rotor in axial flight using this method.
The vorticitymoment theorem ^{9} contains several
mathematical statements involving integrals of vorticity moments. These statements are derived mathematically
rigorously from the NavierStokes equations. For the
aerodynamic force F on a solid body, the statement is:
that point. The strength of each tube (the integrated
vorticity strength co over the tube's crosssection) is the
circulation F around the
tube. Since the vorticity is
solenoidal, F is the same at all sections of the tube.
Hence the vorticity strength co is inversely proportional
to the crosssectional area of the vorticity tube.
If one
views the vorticity field
in
R _{f} (or
R^)
as
composed of a system of vorticity tubes with small
crosssectional areas, then the vorticity in each tube can
be approximated by a vortex loop F = Ft, where t is the
unit tangent vector of the loop's path C, as shown in
Figure la. The vector t points in the direction of the
vorticity vector in the tube. The term 'vortex loop' is
used in the following discussion for convenience. The
conclusions are obviously valid for the closed tubes of vorticity that the vortex loops approximate.
= ^p—f
2
dt JR.
rxcodR + p—fvdR
dt JR
S
(1)
where p
is the density of
the fluid;
R«, is
the
infinite
unlimited region composed of the solid region R _{s} and
the fluid
region
R _{f} ; r
is
a position
vector; v
is
the
velocity vector; and CO is the vorticity vector defined by co = V x v.
The last term in (1) vanishes if the solid motion is
rectilinear and does not change with time. For most
practical applications, the contribution of this term to the aerodynamic force is negligibly small even if the solid is accelerating or rotating. In such applications, the first term in (1) determines the aerodynamic force.
This term states that F is equal to Vip times the rate of
change of the first moment of vorticity in Roo. region reduces to R _{f} if the solid is not rotating.
This
A Cartesian system of coordinates (x,y,z) with the
unitvector set (i j,k) is used in the following discussion
of the vortex loop method.
If the freestream velocity,
V = Ui, is aligned to the xaxis and the span of the solid
is in the ydirection, then the lift L and the drag D on
the solid
are
the
z
and
the
x components of
F
respectively.
The vectors r, v, and CO are stated as r =
xi + yj + zk, v = ui + vj + wk, and CO = £i f
Tjj + £k.
The vorticity field,
as
the curl
of
a vector
field
(specifically, the velocity field),
is solenoidal,
i.e.,
divergence free. It has been shown ^{8} that the regions R _{f}
and R _{s} can be considered together kinematically. A
vorticity field in R _{f}
(or more generally in RJ can be
viewed as being composed of closed tubes of vorticity ^{8} whose walls are vorticity lines, i.e., lines whose tangent at each point is in the direction of the vorticity vector at
The elemental vorticity moment rXCOdR of an
elemental region dR is approximated by rxFds, or F(rxtds), where ds is an elemental segment of the loop. If the vortex loop lies in the xy plane z = z _{h} then tds =
idx + jdy
and rxtds = Zi(idy + jdx) + k(xdy  ydx).
The integration of rxcodR over the vorticity tube R _{t} is
frxcodR^Fziif
JR,
Jc
dy + Fzjfdx+rkf
Jc
Jc
(xdyydx).
The first
two integrals in this expression
are
zero.
Using Green's theorem, it can be shown that the last integral gives twice the area enclosed by C. Hence the
vorticity moment A of the vortex loop F is normal to
the plane of 
the loop 
and its magnitude is twice the 

loop's circulation F times the loopenclosed area A: 

A = 2FA = 2FAn 
_{(}_{2}_{)} 

where n is the unit vector normal to the plane of the 

loop and points in the direction of advance of 
a right 
handed screw as the loop is traveled in the direction t.
A change with time of the vorticity moment A causes a
force F _{r} on the solid which is, according to (1) and (2):
F _{r} =p(FA )
_{(}_{3}_{)}
The force F on the solid is the sum of F _{r} over all loops of the system representing the vorticity field.
If the path C is divided into two parts, C\ and C _{2} , as
shown in Figure
Ib, and the two dividing points are
connected by a line C', then one has two closed paths: a
path Cy formed by joining C' to C\ and another
path
C _{2} ' formed by joining C' to C _{2} . Consider a vortex loop FI on the path Q' and another loop F _{2} on the path C _{2} '.
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
If F! = T _{2} and t _{2} =  tj on the line
C', then F, + F _{2} = 0
and the combined strength of the vortex line on C'
is
zero. The two smaller vortex loops FI and F _{2} together
are thus equivalent to a single vortex F on the path C. This means that any vortex loop F is divisible into two
smaller loops. Successive divisions give an arbitrary
number of smaller loops that are, in aggregate,
equivalent to the loop F. A nonplanar vortex loop can
be divided into a number of small loops that are
approximately planar. Any specific vorticity distribution can be approximated by various systems of
vortex loops configured with a great deal of flexibility.
The 
vector area 
A 
can 
be 
expressed 
in 
the 
component form A = A _{x} i + A _{y} j + A _{z} k, where A _{x} , A _{y} , A _{z} 

are the projected areas 
of A 
in the 
yz, zx 
and 
xy 

planes respectively. 
Also, 
(2) 
indicates 
that 
the 

vorticity moment A of a vortex loop depends only on 

the strength of the 
loop and 
its 
size and direction. 
Hence A is independent of the shape and location of the
loop. In other words, a planar vortex loop with a fixed
enclosed area may deform in its own plane and undergo rectilinear motions without altering its vorticity moment. These facts greatly facilitate the use of the
vortexloop method in aerodynamic analyses.
3. LIFTING LINE THEORY
The
liftingline
theory models the
steady
flow
around a wing of finite span by a horseshoeshaped vortex system ^{l} . This system is composed of a lifting
line representing the circulation F(y) around the wing
and a trailing vortex sheet representing a thin wake.
With this flow model, the KuttaJoukowski theorem is
used to derive expressions for the lift L and the induced
drag Dj on the wing. The downwash, w, at the lifting
line location is viewed as a modifier of the fresstream velocity, hence also the angle of attack, thus causing the
induced drag. The expressions for L and Dj are then:
fb/2
= pUj
J—b/2
r(y)dy
(4)
(5)
The vortex loop method is used to rederive (4) and (5) as follows. Consider first the idealized case of a
wing with a constant circulation F. The vortex theorem
of Helmholtz squires that this lifting line not to end in
the fluid. The liftingline flow model is, in this case, a vortex line composed of the lifting line and two semi
infinite vortex lines, called tip vortices, trailing from the
tips of the lifting line. One thus has an openended
horseshoeshaped vortex system. This system is complete if the presence of the starting vortex is recognized ^{8} . The complete system is a rectangular vortex loop. The starting vortex connects the tip
vortices and closes the horseshoeshaped system far
downstream. As the starting vortex moves away from
the wing, the tip vortices grow. The rectangular closed
vortex loop elongates and the loop remains closed.
Let 
the lifting 
line 
lie 
on 
the 
yaxis and 
extend 
between y = b/2 and b/2, b being the span of the wing. 

If the rectangular vortex loop lies in the z = 0 plane, 

then t = j, i,  j, and i respectively 
on the lifting line, 
the tip vortex at y = b/2, the starting vortex, and the tip vortex at y = b/2. Then n = k and the area A enclosed
by the loop increases at the rate Ub. Hence, according to (3), the growth of the rectangular vortex loop causes
a lift on the wing in the amount pUbF.
With the wing circulation F(y), the strength y(y) of
the trailing vortex sheet in the liftingline flow model is
required
by the Helmholtz vortex theorem to be ^{l}
= dF/dy
_{(}_{6}_{)}
The complete flow model includes the starting
vortex
'closing'
the
trailing
downstream of the lifting line.
vortex
sheet
far
Consider a system of
rectangular vortex loops placed side by side in the z = 0 plane. The vortex loops are labeled sequentially from 1
to J. The vortex loop j has a liftingline segment on the
yaxis with the strength Fj =F(Vj) and the length 8y = b/J. Let yj = (b/2)+j8y and the jth liftingline segment
be
in
the
range y _{}} .\
<
y <
yj. There
are two
trailing
vortices belonging to the vortex loop j, one at y = y^
with t =  i and the other at y = yj with t = i.
Coexisting
at y = ^ (except the tip points j = 0 and j = J) are two
vortex lines: the vortex line Fji belonging to the vortex
loop j and the vortex line Fj+ii belonging to the loop
j+1. The combined strength of the two vortices is [F(Vj)

Fty,)].
As 8y»0,
[F( _{y}_{j} ) 
r(y _{H} )]/8y > dF/dy.
The tip vortices of the J vortex loops in the vortex loop
system become the trailing vortex sheet with the
strength given by (6). The set of J vortex loops is thus
an approximation of the liftingline vortex system.
With the liftingline flow model, the trailing vortex
sheet lies in the z = 0 plane. The jth vortex loop has the
strength Fj
and its
area Aj increases at
the rate U8y.
According to (3), this vortex loop causes a lift pUFjSy.
The total lift caused by the system of vortex loops is the
summation of this quantity over all the loops in the vortex loop system. In the limit 8y —•» 0, the summation becomes (4).
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
With the vortex sheet lying in the z = 0 plane, the liftingline flow model predicts a zero induced drag. Prandtl developed the flow model by assuming * "the
vortices move away from the wing backwards with the
rectilinear velocity V". To rederive the induced drag expression (5) using the vorticitymoment theorem, this assumption needs to be modified to include the velocity component, w, in the analysis. With vortices moving
with the flow, w causes the vortex loops to be inclined
to the z = 0 plane. The disturbance velocity caused by
the wing is small compared to the freestream velocity. Thus w«U and the angle of inclination of each vortex loop to the z = 0 plane is very small. The zcomponent
of the area Aj, (Aj) _{z} , is = A _{r} This component area
grows at the rate U5y and causes, as shown, the lift
pUFjSy. The xcomponents of the area Aj, (Aj) _{x} , is = (w/U)Aj . This component area grows at the rate Wj5y
and, according to (3), causes a drag in the amount

Wjp Fj8y. The total drag caused
by the loop system is
the summation of this amount over all loops.
In the
limit 8y—»0, one has (5).
As discussed, vortex
loops can be divided
into
smaller loops. This fact
leads to a simpler way to re
derive (4) and (5). At the time level t = T, introduce a cut at the plane x=xi >0 to divide the system of J vortex loops into two systems each containing J smaller loops:
a system S _{u} upstream of the cut (in the region x < \\) containing the lifting line and a second system S _{d} downstream of the cut (in the region x > x^ containing
the starting vortex, as subsequent time level
shown in Figure 2a.
T
+
8t,
the
system
At
S _{u}
the
has
expanded and the system S _{d} has moved downstream
with the flow. 
If the shape and the inclination of the 
vortex loops in S _{d} 
collectively remain unaltered during 
the time period 8t, then, according to the discussions in the last paragraph of Section 2, the vorticity moment of
the system of loops in S _{d} at the new time level T + 8t is
the same as that at the old time level T.
The system S _{d} 

With a steady 

time 
level 
I. 
therefore does not cause a force. At the new time level
T + St, again introduce a cut at the plane x = Xi to divide
the system S _{u} into two new systems.
flow, the new upstream system at the new time level is
identical
to
the
system
S _{u}
at
the old
Therefore the change of vorticity moment that took
place during 8t is attributable entirely to the vorticity
moment of the new downstream system, shown in
shade in Figure 2b. This system occupies the region xi
< x < \i + U6t.
The jth loop in this region is inclined to
the z = 0 plane at a small angle. This vorticity moment
of the loop is 2FjWj5y8ti2rjU8y8tk. This newly
emerged vorticity moment causes a lift pUFj8y and a
drag pWjFj8y. Summing these forces over the loops in the new downstream system and letting 8y —> 0, one has (4) and (5).
4. WAKE INTEGRALS FOR THE FINITE WING
Using
(6), one obtains
F = d(yF)/dy +
yy.
The
integration of d(yF)/dy over the span of the wing is zero
since F=0 outside the wing tips. One thus has, from (4),
L = pU
pb/2
Jb/2
yydy
(7)
Using (6), one has wF = wd(yF)/dy +ywy.
For a
symmetric wing, the term wd(yF)/dy is antisymmetric
with respect to y=0. The integration of this term over
the span of the wing is therefore zero and (5) becomes
/ 2
b/2
yw(y)y(y)dy
(8)
The strength y approximates the integrated vorticity
value across the wake layer. With the layer inclined at
a very small angle to the plane z=0, y is the integration
of £ respect to z over the wake region. 
One thus re 

expresses (7) and (8) in the wakeintegral form: 

puJ 
(9) 

=p f 
yw^dydz 
(10) 
where Wis the wake crosssection.
Equation (10) is a new wakeintegral expression for the induced drag. An wakeintegral expression for the induced drag, developed previously ^{5} on the basis the momentum theory, is in the form
(11)
where \/ is a stream function in the yz plane.
It has been shown ^{8} that the new expression (10) is equivalent to the previous expression (11). With
measured wake velocity values, corresponding vorticity
values can be computed easily. The induced drag can
then be evaluated using (10). The numerical procedure
required is simple and efficient. In contrast, the use of (11) requires the computation of the stream function
\/ by integrating the velocity values. The procedure for
the integration is relatively complex and prone to error.
The use of the new wakeintegral expression (10) is
therefore preferred over the previous expression (11).
The equivalence of (10) and (11) endorses the use of the vortexloop method in aerodynamic analysis.
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
A wakeintegral expression for the profile drag ^{8} on
the finite wing can be established using the vorticity
moment theorem. The wakeintegral expressions link
the footprints of the wing to the aerodynamic force on
the wing. This linkage is discussed in Section 6 in
connection with wakeintegral results for the rotor.
5. WAKE INTEGRALS FOR THE ROTOR
A cylindrical coordinate system (r, 0, z) with the
unitvector set (e _{r} , e _{e} , e _{z} ) is used in the present study of
the hovering rotor problem. The vectors r, v, and CO are
stated as r = e _{r} r + e _{z} z, v = e _{r} v _{r} + e _{e} v _{e} + e _{z} v _{z} , and (0 =
e _{r} 0)r + ee (Oe + e _{z} civ
A stationary reference frame at
rest relative to the fluid far from the rotor is used. The
rotor disk is placed in the z = 0 plane. The rotor rotates
about the zaxis with the angular velocity Q.
The circulation F(r) around the blade depends on
the span location. According to the Helmholtz vortex
theorem, the blade must leave behind wake vorticity as
it advances azimuthally. For a thin wake, the vorticity
content of the wake can be approximated by a vortex
sheet with the strength y = dF/dr. (The negative sign in
(6) is absent with the ordering of the unitvector set e _{r} ,
e _{e} , e _{z} .) For a rotor in hover or in climb, the velocity v _{z}
transports the wake vorticity continually in the axial
direction. A helical wake is therefore present under the
rotor disk.
The blade circulation is connected through
this helical wake, which is in turn connected to the
starting vorticity at the far end of the helical wake.
Consider a system of J helical vortex 
loops. Let 
the jth loop contain a liftingline segment of strength Fj 

= F(VJ). Let this segment be on the raxis in the z = 0 

plane and occupy the radial range TJ.\ 
< r < TJ, with TJ = 
rj_! + 5r, 6r = R/], R being the rotor disk radius. This
lifting line segment is trailed by two helical tip vortices.
These tip vortices are connected far downstream by a
starting vortex segment. This system of vortex loops
approximates the lifting line of the blade, the helical
vortex sheet, and the starting vortex.
At the time level T, designate the blade position by
6=0 . Introduce a cut at the plane 6 = 81 < 0 to divide 

the system of 
J vortex 
loops into two systems each 
containing J smaller loops: a system S _{u} in the region 6 >
61, containing the blade, and a second system S _{d} in the
region 0< 61, containing the starting vortex.
subsequent
time
l^^el
T
+
8t,
the
system
At the
S _{u}
has
expanded. Introduce a new cut at the plane 0 =0i+ QSt
to divide the system S _{u} into two new systems each
containing J smaller loops. Following the discussions
of Section 4, the newly emerged vorticity moment in
the pieshaped region
0 _{t}
<
0
<
0! + QSt
causes the
aerodynamic force on the rotor.
The jth vortex loop in the pieshaped region
encloses the
area
rjQ 8r8te _{z}  (v _{z} )j8r8te _{e} .
Therefore,
according to (2) and (3), there is a thrust prjQFj8r and
an induced drag
p(v _{z} )jF8r caused by the jth loop. The
thrust on the blade is the sum of prjQFjSr over j. In the
limit as Sr »0, one obtains an expression for the thrust
on one blade. For an nbladed rotor, the thrust T is
given by
C ^{H}
rFdr
T = npQ
Jo
(12)
Using
the
relation
y
=
dF/dr,
one
obtains
rF
=
V ^{/} id(r ^{2} F)/dr Vx ^{2} y. Since F = 0 immediately outside the blade tip, the integration of Vkl(r ^{2} r)/dr over the span of
the blade is zero. One therefore obtains from (12)
C
_{n}_{p}_{Q} _{J}_{o} r^ydr
_{(}_{1}_{3}_{)}
The strength
y
is
the
integrated
vorticity
value
across the wake. With v _{z} « rQ, y is the integrated (Oe
value across the wake with respect to z.
Equation (13)
is expressed as a wakeintegral expression:
(14)
Wake integral expressions have been obtained for
the induced drag and the profile
drag ^{7} on
the rotor.
More important are wakeintegral expressions for the
torque Qi due to the induced drag and the torque Q _{p} due
to the profile drag. These expressions have been
derived using the vorticitymoment theorem ^{7} :
r v co^drdz 
(15) 

I 
Z 
D 

Q _{D} =np 
r zco _{r} drdz 
(16) 

The GO,, vorticity 
layer 
is 
left behind by boundary 

layers surrounding the blade. 
For 
the thin 
wake, the 
computation of (Or can be avoided by placing o\ =
d\o/dz into (16). By noting that z(3ve/3z) = 3(zv _{e} )/3z
v _{e} , and that the integration of 3(zvq)/dz over the layer is 0, one obtains an expression in terms of v _{e} :
Qo=nf
'iv
r
v^drdz
e
(17)
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
6. EXPERIMENTS AND RESULTS
Rotor tests were performed in the U.S. Army Aeroflightdynamics Directorate (AFDD) 7' by 10' Wind Tunnel at the NASA Ames Research Center. A twobladed model rotor was mounted in the settling chamber of the wind tunnel. The axis of the model rotor was aligned with the tunnel flow direction. The wake of the model rotor passed through the contraction section and the test section of the tunnel. In previous rotor tests in the AFDD 7' x 10' wind tunnel, F. Caradonna et al. ^{M} demonstrated the advantages of simulating rotor climb flows using this test configuration. Descriptions ^{n} of the test configuration, the physical layout, the rotor, and the instrumentation of these previous tests are for the most part applicable to the present tests. Modifications and additions were made to obtain particle images in the wake of the model rotor and to address the issue of rotordriven flow returning to the settling chamber.
The flow circuit of the wind tunnel is shown in
Figure 3. A flow seeder was used to introduce particles
into the flow for particle imaging.
A threedimensional
particleimage velocimetry (PIV) system was used to obtain particle images in the near wake of the model
rotor. Major components of the PIV system are two 2,000 x 2,000 pixel digital cameras, lens sets, remote focus system, highspeed interface and digital links, control cables, computers for acquiring and storing particle images, and laser light source and mirror systems. Figure 4 shows the camera and light sheet configuration used.
The AFDD 7' x 10' wind tunnel is a closedcircuit tunnel. The crosssection of the settling chamber is 30' x 31'. The crosssection of the test section is 7' x 10'. The model rotor has a nominal diameter of 7' and a true diameter of 6.283'. The wake of the model rotor was expected to flow through the test section with minimal interaction with testsection walls. With the tunnel drivefan off, the model rotor acted as a substitute drive fan and created a flow through the tunnel's flow circuit. Thus, with the tunnel drivefan off, a climb condition rather than a true hover condition was expected to exist in the settling chamber.
To evaluate the strength of the rotordriven flow, a curtain was installed at the air exchanger section of the tunnel to block the rotordriven flow from returning to the settling chamber. Fresh air was admitted to the settling chamber through openings downstream of the curtain, as shown in Figure 3. Prior to acquiring wake data using the PIV system, tests were run both with the curtain in place and with it removed. During these tests, the tunnel drivefan was off and the rotor operated
at 870 rpm. Tests were run with the collective pitch angles of the rotor blade set at 1°, 3°, 5°, 7°, 9° and 11°. Thrust and torque were measured using the balance mounted on the rotor's drive shaft. The measurements showed that the thrust and the torque on the model rotor were not significantly affected by the blockage of the rotordriven flow.
In these test runs, flow velocities were measured in the test section using vane and thermoanemometers. Total volume flow rates through the test section were determined from the measured test section velocities. It was found that the flow through the test section was reduced between 14% and 21% by the blockage of the rotor driven flow. Balancemeasured thrust values were used to estimate the flow through the rotor disk using the axial momentum theory. It was found that flow through the test section was between 2.5 and 2.9 times the estimated flow through the rotor disk, indicating that a sizeable portion of the flow through the test section did not go through the rotor disk.
Measured velocity contours in the test section indicated that the rotor wake was diffused by the time it entered the test section. It is postulated that the wake, together with the fluid it entrained on its way to the test section, accelerated slightly in the contraction section. The acceleration lowered the static pressure in the test section slightly. This lowered test section pressure created a flow external to the 'slipstream' of the wake. The flow through the test section is therefore composed of the rotor wake and a flow external to the rotor wake.
The momentum flux at the test section was estimated using measured average velocities and found to be greater than the thrust on the rotor. This excess of momentum flux supported the view that a pressure difference existed between the settling chamber and the test section. The average settling chamber flow velocity was very low. (With the curtain installed, this velocity was 1.55 fps for the 11° case, of which 0.54 fps was due to the estimated flow of the rotor wake.) Only a minuscule pressure difference would produce the measured amount of flow through the test section. Experimental verification of this minute pressure difference is therefore difficult.
Wakeintegral expressions presented in Section 5 are applicable to rotors in axial flight, with hover as a special case. The question as to what specific flow rate corresponds to the true hover state is not essential to the present work. Future test runs with the test section access doors open to equalize the static pressure in the test section with ambient air are desirable. With the access doors open, the measured flow rate through the test section can be used to establish the true hover state.
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
The curtain at the air exchanger section of the
tunnel was installed to block the return of the rotor
driven flow during PIV test runs. The tunnel drive fan
was off. Two pulsed laserlight sheets were introduced
in a plane parallel to the rotor's axis, as shown in Figure
4. As the rotor rotated, its blades passed through this
light sheet repeatedly. The rotor was operated at 870
rpm to match the maximum pulse rate of the laser. The
bladetip speed was about 286 fjps and compressibility
effects were not important. The light sheet was aligned
to the trailing edge of the blade at the instant the blade
advanced passed the sheet.
This instant of time was
used as a reference time level. A set of 25 images was
acquired at the same blade azimuth during a series of
blade revolutions. The images were combined to
produce timeaveraged
velocity
fields
at
specific
relative positions between the blade and the wake
survey plane.
Figure 5 shows the geometry of the rotor blade and
the position of the light sheet relative to the blade at the
time levels particle images were acquired. The light
sheet was stationary while the blade advanced during
the tests. The distance between the blade tip and the
light sheet designates the relative position of the
particle images (wakesurvey plane) and the blade. For
example, a 2.0c (twochord) wakesurvey plane
designates particle images acquired at the instant the
blade tip advanced two chords from the plane.
Particle images were acquired for two collective
pitch angles, 5° and 11°, in three contiguous rectangular
data patches along the blade span. Each patch covered
approximately 6" of span and 10" of axial distance.
The three patches together covered about 17.6" of span
extending between 20.9" from the rotor axis to 0.8"
outboard of the blade tip. In the axial direction, the
boundaries of each zone were about 3" upstream and 7"
downstream of the rotor disk.
Figures 6 and 7 show contours of the velocity v _{e}
and the vorticity Cfle at the wake survey planes 2.0c for 

the 5° and the 11° cases. 
The v _{e} velocity deficit layer 
represents a layer of ov 
This layer is the footprint of 
the two boundary layers on the blade surface. This 
layer is composed of two sublayers, one from the
upper boundary layer and the other from the lower
boundary layer. The (Or contents of the two sublayers
have different signs.
vorticity
o\ in the two
The positive and negative
layers are
connected by ci>z to
form closed vorticity loops in the 9plane. The
presence of the vorticity co _{z} , though not shown, can be
inferred from the presence of the 0^ sublayers. As time
progresses, vorticity loops emerge in successive 0
planes.
With
downwash,
a
helical
wake
layer
composed of vorticity loops that lay in planes normal to
the helical layer is formed. In order to determine the
profile drag, the substructure of the helical o\ layer
must be recognized. If the a^ layer is approximated as
a vortex sheet, in other words, a layer of zero thickness,
then the profile drag cannot be detected. This is
because the approximation makes vorticity moment zo\
zero and therefore (16) gives a zero profile drag.
The
approximation masks the deficit of v _{e} in the wake and
thus (17) gives a zero profile drag.
The (Oe vorticity in the wake is the footprint of the
circulation change along the span of the blade. This
footprint is linked by (14) and (15) 
to the thrust and the 

induced torque on the rotor. Figures 6 and 7 show that 

the coe wake associated with each blade is composed of 

a strong tip vortex, i.e., a helical 
tube of intense co& 

trailing the blade tip, and 
a weak helical layer of 
co& 

inboard of the tip vortex. 
The blades of the present 

tests are twisted. The observed (0& distribution 

indicates that the circulation around the blade changes 

slowly along the span and drops abruptly to zero 

outside the tip. 

The 
sign of 
co^ in the inboard 
layer is opposite to 

that 
in 
the 
tip vortex. 
The 
(Oe layer can be 
approximated by a vortex sheet, without losing pivotal
information about either the thrust or the induced torque
on the rotor. This is because this inboard co^ layer,
unlike the co^ layer, is not composed of sublayers
containing vorticity of different signs. This OG& layer is
a part of the vortex loops lying in the helical wake
sheet, not normal to the sheet. The presence of a hub
vortex and a starting vortex is inferred by the presence
of the helical layer of
cog.
The hub vortex and the
starting vortex, together with the circulation around the
blade, complete the vorticity loops containing the
vorticity 009. The hub vortex and the starting vortex are
both outside the three data patches of the present tests.
The presence of tip vortices is evident in Figures 6
and 7. Two traces of tip vortices appear in Figure 7 for
the 11° case. The one very close to the rotor disk is
associated with the blade that most recently passed
through the wakesurvey plane. For convenience, this
blade is called the first blade. The second trace is
associated with the second blade, which is about 180°
from the survey plane at the instant particle images are
taken. A third trace of a tip vortex is observed in
Figure 6 for the 5° case. This third trace is the footprint
of the first blade during its previous passage through
the wakesurvey plane.
The 
layers of 
C0r and 
0)9 leave the blade together 

and 
they 
are transported 
in 
the 
fluid 
by identical 
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
physical processes of convection and diffusion. The
two layers therefore occupy the same physical space.
The structures of the rotor wake described above
are evident at all wakesurvey planes. The thinness of
the inboard vorticity layers in Figures 6 and 7 indicates
that for both the 5° and the 11° cases, any significant
flow separation, if present, is restricted to the root
portion of the blade not covered by the data patches.
As the tip vortex moves axially, it also moves
inboard. Figure 6 shows that, for the 5° case, the tip
vortices move axially at a speed substantially slower
than that of the inboard wake layer. The movements of
the second blade's tip vortex bring it to the path of the
first blade's inboard wake layer. A strong interaction
between the inboard wake layer of the first blade and
the tip vortex of the second blade then occurs. For the
11° case, the axial speed
of the
tip vortex is greater.
The strong interaction between the inboard vorticity
layer of the first blade and the tip vortex of the second
blade is not observed in Figure 6.
Spurious vorticity along the boundaries connecting
the three data patches is observed in Figures 6 and 7.
This spurious vorticity is attributable to an inexact
matching of the three data patches in the tests. For
wakeintegral analyses, this spurious vorticity is filtered
and disregarded. Figure 7 also shows widespread
traces of background noises. The noises are weak and
do not have significant effects on wakeintegral results.
Wake data at the 0.5c wakesurvey plane contain
excessive spurious values. The quality of these data is
not sufficiently high for meaningful aerodynamic
analyses. For the 11° case, wake data for the innermost
data patch are either missing or not of sufficiently high
quality at the 1.0c, 4.0c and 5.0c wakesurvey planes.
The qualities of all other acquired wake data are
comparable to those shown in Figures 6 and 7.
Because of the strong interaction between the
vorticity layers left behind by the first blade and the tip
vortex left behind by the second blade, the wake data
for the 5° case are not suitable for the evaluation of the
profile torque. Profile torque values are determined
using (17) and wake data for the 11° case. As noted,
the three data patches cover only the outboard r > 20.9
portion of the wake. In evaluating the profile torque,
the contribution of the missing inboard wake data is
estimated by assuming the inboard o\ layer does not
change with the span in the root portion of the wake.
Based on this assumption, the missing c\ layers in the
wakesurvey planes 2.0c and 3.0c are estimated to
contribute 17% of the total profile torque. For the
wakesurvey planes 1.0c, 4.0c and 5.0c, the missing
8
(Or layer in the root portion of the blade, including those
in the innermost data patch, is estimated to contribute
36% of the total profile drag.
Wake data at the 2.0c and the 3.0c wakesurvey
planes for the 11° case show that the o\ content in the
wake layer does not change rapidly in the two inboard
data patches. The estimated contributions of the
missing inboard data do, however, introduce
uncertainties in the evaluation of the profile torque.
This uncertainty is due in part to the physical presence
of the root structure of the model rotor. Also, with the
twisted blade, it is possible that flow separates over a
root portion of the blade, especially in the 11° case.
The vorticity 0)9 in the inboard layer is found to be
very
weak.
For
example,
for
the
11° case,
the
magnitude of the integrated co& value in the inboard
layer is determined to be 1.4% of that in the tip vortex
at the 2.0c wakesurvey plane. As (15) and (14) show,
the contributions of (Oeto the induced torque and the
thrust are weighted by the factor r ^{2} . The missing data in
the root portion of the blade span is therefore
unimportant in the evaluation of the induced torque and
the thrust using wakeintegrals. Since the tip vortex is
located in the outermost data patch, the induced torque
and the thrust on the rotor can be accurately determined
using only wake data in this outermost data patch.
Induced torque values, determined using (15), are
shown in Figure 8 for the 11° case. Total torque values
are obtained by adding the values of profile torque,
determined using (17), to the induced drag values. The
very good agreement between the balancemeasured
value and the total torque values determined using wake
data at survey planes 1.0c and 2.0c is unforeseen
since, as discussed, the missing inboard wakedata
introduce uncertainties in computing the profile torque.
Figure 9 shows the thrust on the rotor determined
using (14). The agreements between the wakeintegral
results and the balancemeasured thrust at all wake
survey planes for both the 5° and the 11° cases are
reasonably good and encouraging.
Wakeintegral expressions are derived in Section 5
by analyzing the rate of emergence of new vorticity
moment in the wake. It is therefore preferable to use
wakesurvey planes close to the blade. As discussed,
the o>r layer is composed of two sublayers containing
OT with opposite signs.
disperses the
vorticity
As the wake ages, diffusion
and partially
annihilates the
positive Or and the negative 0)r in the two sublayers.
The wakeintegral expression (16), or equivalently (17),
therefore provides more accurate profile torque values
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
at wakesurvey planes closer to the blade.
Diffusion
effects are less important in the determination of the
thrust and the induced torque since (0& resides
wholly in tip vortices.
nearly
7.
CONCLUSIONS
The wakeintegral method connects the footprints
left behind by a solid body in flight to the aerodynamic
force
and
moment
on
connection, the task of
the
body.
Through
this
solving a threedimensional
aerodynamic
flow
problem
is
reduced to
one
of
evaluating the footprints in a twodimensional planar
area. Information about these footprints can be
acquired either experimentally or computationally. By
reducing the dimensionality of the information required
to determine the aerodynamic load from three to two,
the method offers major advantages in all three
branches of aerodynamics  theoretical, experimental
and computational. The method is efficient since the
required footprint information is restricted to the small
vortical wake region of the flow.
The central theoretical task of the wakeintegral
method is the establishment of wakeintegral
expressions. In the present research, a vorticityloop
method was developed and used to derive new wake
integral expressions for the finite wing problem.
Compared to previous wakeintegral expressions for the
induced and the profile drags, the new wakeintegral
expressions are remarkably simpler and more efficient.
New wakeintegral expressions are also derived, using the vorticityloop method, for the thrust, the
induced torque and the profile torque on the rotor.
These expressions connect the footprints of the rotor
blade to the aerodynamic load on the rotor. The
azimuthal component of the wake vorticity is connected
to
the thrust
and
the
induced
torque.
The
radial
component of the wake vorticity is connected to the
of the wakeintegral method. The power of three dimensional particleimage velocimetry in experimental
aerodynamics has also been demonstrated. In addition
to providing quantitative wake data, particle imaging
has brought into focus wake features often disregarded
in the past. These wake features are relatively
inconspicuous, but important to viscous and unsteady
aerodynamic analyses.
Efforts of the present program have laid the
foundation for continued efforts to construct a practical aerodynamic design tool using the wakeintegral method.
Acknowledgements
The contribution of the windtunnel taskteam for
the
present
research
is
gratefully
acknowledged.
Members of this team include Anita I. Abrego, Brian H. Chan, Steven Chan, Lauura Galvas, Joel T. Gunter, Elizabeth M. Hendley, Jon L. Lautenschlager and David W. Pfluger. Samuel S. Huang served as the on
site engineer of Applied Aero throughout the planning and execution phases of the wind tunnel tests. Dr. Luiz
Lourenco designed the particle image velocimetry system and provided related technical support, including the processing of particle images. Dr. Chee
Tung's
support
and
timely
advice
throughout
this
research program is also gratefully acknowledged.
References
1. Prandtl, L. "Applications of Modern Hydrodynamics to Aerodynamics", Report No.l 16, National Advisory
Committee on Aeronautics, 1921
2. Betz, A. "Ein Verfahren zur Direkten Ermittlung des Profilwiderstandes", Zeitschrift fur Flugtechnik und Motorluftschiffahrt, Vol. 3, 1925
profile torque. The axial component of the wake 


vorticity does not need to be known explicitly. 
Its 
Cambridge 1997 

presence 
in 
the 
wake 
and 
its 
contribution 
to 
the 

aerodynamic load are 
inferred 
from 
those 
of 
the 

C. 
"Progress Towards a Method of 
azimuthal and
radial components of the wake vorticity.
With the new wakeintegral expressions, the use of wake data very close to the trailing edge of the lifting
body
is
preferred.
This fact
offers
an
important
advantage to the use of CFD in wakeintegral analyses.
Numerical methods capable of accurately simulating
the near wake are useful, even i* ihe far wake cannot be
accurately simulated because of numerical diffusion.
Experiments performed in the present research
have validated the practicality and the major advantages
Measurement of the Components of the Drag of a
Wing of Finite Span", Technical Report 72232, Royal
Aircraft Establishment, 1973
5.
Wu, J.
C.,
Hackett, J.
E.,
and Lilley, D.
E.
"A
General Wake Integral Approach for Drag
Determination in ThreeDimensional Flows", AIAA
Paper No. 790279, 1979
6. Kroo, I. "Drag due to Lift: Concepts for Prediction and Reduction", Annual Review of Fluid Mechanics, Vol. 33, 2001
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
7.
Wu,
J.
C.
and
Wang,
C.
M.
"Separate
Determination of Coexisting Components of
Aerodynamic Drag on Rotors", USAAMCOM TR
01D10, Aviation Applied Technology Directorate, U.S. Army Aviation and Troop Command, 2001
8. Wu, J. C. "Theoretical Aerodynamics based on Vorticity Dynamics", Applied Aero LLC, Report 2001A1231,2001
9. Wu, J. C. "Theory for Aerodynamic Force and Moment in Viscous Flows", AIAA Journal, Vol. 19, 1981
10.
Caradonna, F., Henley, E., Silva, M., Huang, S., Komerath, N., Reddy, U., Mahalingam, R., Funk,
R., Wong, O., Ames, R., Darden, L., Villareal, L.,
and
Gregory, J., "An
Experimental Study of a
Rotor in Axial Flight" Proceedings, Specialists'
Meeting for Rotorcraft Aeroacoustics and
Aerodynamics, American Helicopter Society,
1997.
Air Exchanger Section
J
Settling Chamber
Figure 3. Wind tunnel flow circuit.
Closed tube
of varticitv
Figure la. Vortex loop approximation of vorticity tube.
Figure Ib. Division of vortex loop into smaller loops.
Figure 2a. Liftingline vortexloop systems at time t = T.
1
KUSt
New upstream system
Transported Sd
Expanded S _{u} Figure 2b. Liftingline vortexloop systems at time t = T + 5t.
10
Figure 4. Particle imaging system layout.
75.40
4.03
7.55 degree linear twist
Figure 5. Blade geometry and wakesurvey planes.
American Institute of Aeronautics and Astronautics
(c)2002 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.
 velocity  I: _______© u 6.00 7.36 8.71 10.07 11.43 12.79 ^{1}^{4}  14 15^0 16>86 18^1 19^7 20.93 22.29 23.64 25.00 , , . i i / . 1 1 1 1 1 1  vorticity • t VORX 500.00 463.57 : i C*=> o <=> 0 o ' m 0 => ° c •*• o « £K^OC^CK ^~ir _{^}$\ _{e} ^^~^~^S^' < id S • • f ) • , , , 1 , ( ,,,!, r(ln) , , 1 1 ! . r Figure 6. Streamwise velocity and vorticity contours. 2chord wake survey plane 5degree collectivepitch angle velocity : vorticity 30 tin) t r<ln) SI !~! ^sw* tea B S P B ^{K}^{s}^{>}^{i} • • >W 427.14 M, 390.71 r*n 354^9 rTJ 317.86 281 H ^{2}^{4}^{5} °° _{l}_{i} 208 57  n 43  S 172.14 l^ 135.71 9929 62.86 26.43 10.00 VORX • 500.00 H 463.57 m H 427.14 390.71 H 354.29 317.86 281.43 245.00 208.57 172.14 135.71 99.29 62.86 26.43  10.00 Figure 7. Streamwise velocity and vorticity contours. 2chord wake survey plane 11degree collectivepitch angle 

BatenceMeasuredJotal 

a 
WakeIntegral, Induced 

• 
WakeIntegral, Total 

i 234 
5 

Distance between Blade and Wake Plane(chord) 

Figure 8. Rotor torque, 11degree collective pitch. 

5degree collective 

BalanceMeasured 

• 
WakeIntegral 

Distance between Blade and Wake Plane(chord) 

t 11 degree collective 

 

   BalanceMeasured WakeIntegral • Distance between Blade and Wake>lane(chord) 

Figure 9. Rotor thrust, 5 and 11degree collective pitch. 
11
American Institute of Aeronautics and Astronautics