Você está na página 1de 42

ANNUAL

REVIEWS Further
Quick links to online content

Copyright 1972. All rights reserved

WING-BODY AERODYNAMIC INTERACTION1 8031


HOLT ASHLEY

Stanford University, Stanford, California


Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

WILLIAM P. RODDEN

Consulting Engineer, Los Angeles, California


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

1. INTRODUCTION

When addressing the topic of this article, one is struck by a curious


disparity between the needs of the designer of vehicles that move through the
air or water and the subject matter that has proved most interesting to the
theoretical aerodynamicist. Clearly the former is concerned with the flow
over, and the pressures, forces, and moments experienced by, generally
shaped streamlined bodies and multiple lifting surfaces (wings, tails, con
trols) in intimate combination. Until recently he has had to place primary
reliance for such information on measurements by wind tunnel and other
experimental means. Except in particular cases, he has obtained only a
rather piecemeal and preliminary sort of guidance from the products of
rational theory, whose focus has been single, isolated, planar wings and
slender bodies of revolution.
Interference was the word chosen to describe situations where flow over
one lif ting or body element was significantly affected by the presence of an
other-as if, somehow, this interaction interfered with the purist's more
appealing efforts to analyze each aerodynamic entity standing alone. It is no
accident that 32 years intervened between the attainment of powered air
craft flight and the first appearance of even a short descriptive chapter on
,

flow over the "Airplane as a Whole (Durand 1935). Many additional years
"

rence & Flax 1954, Ferrari 1957). Most of the basic tools of interference or in
preceded the publication of two truly comprehensive survey articles (Law

teraction theory, as it exists today, were understood by th e latter date. With


regard to the complete description of fluid motion past an assemblage of
streamlined elements, notable subsequent progress has been made toward
realizing the full potentialities of this theory by means of successive gen
erations of high-speed data-processing equipment.
The limited objective of the present authors will be to review and
1 The contributions of Joseph P. Giesing of the Douglas Aircraft Co., Long Beach,

Calif., to Section 2 are gratefully acknowledged. The portion of this article contributed
by Holt Ashley was supported by the Air Force Office of Scientific Research, Office
of Aerospace Research, under Contract No. F44620-68-C-0036.

431
432 ASHLEY & RODDEN

exemplify some of the more successful methods known to them for analyzing
the near flow field and aerodynamic loading experienced by three-dimen
sional combinations of slender elements in subsonic or supersonic main
streams. There are certain topics that will be omitted altogether, and others
that have been summarized so thoroughly elsewhere that more than a litera
ture citation here seems redundant. A brief listing of these two categories in
tHis Introduction, we believe, constitutes a partial survey of significant past
contributions.
The following are excluded from the present scope:
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

Interaction with rotors and propellers.-This can be an important effect on


ships, helicopters, and high-performance subsonic aircraft. Ferrari (1957)
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

covers the propeller extensively in his Chapter 3, and most subsequent at


tempts to improve on his methods are of a semiempirical nature.

Free surface and cavitation effects in liquids. While they have obvious
-

significance both for hydrofoil and displacement ships, attention to these


phenomena would eause this survey to exceed its allotted space.

Transonic flow over wing-body combinations with shocks.-In its greatest


generality, this problem is characterized as essentially nonlinear. That is, the
methods of singularity superposition that form the present theme would
seem to be excluded ab initio. By means of the "local linearization" tech
nique, however, Landahl ( 1964) has proposed how the influence of thickness
on the lifting airloads of an isolated surface might be represented. Among
several incomplete tries at mechanizing Landahl's suggestions, one for finite
wings (Andrew & Stenton 1968) deserves mention, but wing-body combina
tions constitute unfinished business.

Far fields and the sonic boom. - Whe nthe source of disturbances is a
sonic speed relative to the surrounding
streamlined aircraft flying faster than
gas, it has long been known that the principal far-field nonlinear effect is a
displacement of wavefronts from positions calculated by wholly linearized
acoustics. The implications have been thoroughly surveyed by Hayes (1971),
who also discusses such influences as those of atmospheric nonuniformity,
winds, and turbulence. It is noted that methods reviewed in Section 3 are
well adapted to finding the supersonic near field, on which boom predictions
can be based.

Hypersonic flow.-This regime is defined by the conditions


(MO)-l =
0 (1) (1)
where M is the flight Mach number and 0 a dimensionless parameter that
measures the maximum disturbance created by an object moving through a
gas (e.g., maximum surface slope relative to the flight direction, thickness
WING-BODY AERODYNAMIC INTERACTION 433
ratio or angle of attack of wing, fineness ratio of body, etc.) . The field dif
ferential equations are essentially nonlinear, and the flow is marked by
nonisentropy and deviations from the state relations for a perfect gas.
Either of these facts would cause it to fall outside the restrictions adopted
here. A few references are given below, however, to method of characteristics
and other techniques that are potentially applicable to hypersonic wing
body problems.
* * *

Aerodynamic interaction comprising, as it does, a melange of fairly dis


Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

tinct physical situations, one is dismayed if not surprised at the disorderly


state of its literature. In an attempt to achieve some organization while keep
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

ing the reference list within bounds, the authors next propose to identify
several topics that are relevant but have already been well reviewed in earlier
books or articles.
Two-dimensional flow, biplanes, and cascades.-I t is hard to imagine a
"body" as giving rise to a purely p'lane flow. Nevertheless, this subject de
serves mention if only because its relative mathematical tractability has
stimulated such an enormous volume of research. For instance, Chapter XI I
of Thwaites (1960) sU,mm
two-dimensional biplanes and infinite arrays of lifting airfoils. Notable early
contributions by the exact method of conformal transformation are due to
Garrick (1936, 1944). The extension of nonlinear and small-perturbation
solutions for compressible fluid appears in Chapter 12 of Woods (1961). The
theory of oscillating cascades was fully developed by Siihngen & Meister
(1958) for the incompressible case. More recently, compressibility effects on
oscillating cascades and closely related problems were introduced by such
authors as Jones & Rao ( 1970) and Bland (1968). The interesting attempt to
construct nonlinear incompressible solutions for general unsteady motions
of interacting airfoils is typified by Giesing (1968a).

Three-dimensional incompressible flow over configurations with lifting


surfaces and wakes, as treated by numerical solution of the exact Green's theorem
for Laplace's equation Termed "calculation of potential flow about arbi
.-

trary bodies," this approach was fully described by Smith & Hess (1967) for
nonlifting shapes, which can be simulated with distributed sources. Djojodi
hardjo & Widnall (1969) published a valiant attempt to determine the true
evolution of the wake and loading of an isolated wing with finite thickness
and incidence. The similar treatment of wings and bodies in combination
would seem to be possible at present levels of computer capacity, but the
promise has not yet been realized.

Classical slender-body theory. Surely the richest source of approximate


-

results on interfering wings and bodies is the slender-body idealization, whose


full exposition is usually attributed to Ward (1949). The key ideas are that
434 ASHLEY & RODDEN

flow past a high-fineness-ratio body, a very low-aspect-ratio wing, or a suit


able combination of these two is characterized by a nonlifting part plus a
"crossflow" part associated with angle of attack and other motions that pro
duce net fluid momentum normal to the direction of flight. In the near field,
this crossflow simplifies to two-dimensional, incompressible motion normal
to the body axis; thus, complex-variable methods can be used to estimate
lift, pitching moment, sideforce, etc. Only the supersonic wave drag is found
to be affected by the nonlifting flow, whose near field consists of a two
dimensional component plus a correction, dependent only on the axial
coordinate x and determined by the axial distribution Sex) of the cross-sec
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

tional area of the configuration.


The most comprehensive exploitation of these concepts for force and
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

moment estimation appears in Nielsen ( 1960), especially his Chapter 5 on


interference. Chapter 2 of Ferrari ( 1957) also summarizes the application to
combinations of wings, bodies, and tails. From the extensive literature on
slender-body methods adapted to interacting flows, one other paper (Bryson
1953) is singled out. His use of the momentum theorems of fluid mechanics
to determine lateral forces and other stability derivatives is a landmark in
the evolution of this subject. Some additional references are mentioned in
Section 2.

Simplified approaches to particular interference problems.-The methods


described in Sections 2 and 3 below are aimed toward simultaneous determi
nation of complete flow fields for general systems. It is therefore judged that,
at the price of considerably greater computational effort, they subsume all
earlier, more approximate schemes for partially representing the effects of
interference. No denigration of the simplicity, convenience, or enhanced
physical understanding associated with such schemes is implied. Indeed,
some account of them is essential as introduction to the more elaborate

will be found in Section 2.


analyses that will be the final design tools of the future; further discussions

It is helpful to consider an example. An unswept, thin wing of moderate

centerline) roughly 1/2 is placed at small angle of attack ex in a subsonic main


aspect ratio and of taper ratio (chordlength at tip divided by chordlength at

stream. It is a familiar fact that the spanwise distribution of circulation or


running lift will resemble the upper half of an ellipse, placed across tire wing
span as major axis.
If now an elongated fuselage of circular cross section is centered in the
wing,2 and the combination is placed at incidence so that the fuselage axis
also inclines at ex to the flight direction, two things happen. First, the por
tions of the wing near its root are exposed to a higher effective angle of attack
because of crossflow past the fuselage (cf the slender-body model discussed
above). Similar to two-dimensional streaming around a circular cylinder, this

2 Cf the sketch in Figure 1.


WING-BODY AERODYNAMIC INTERACTION 435
crossflow "induces" a total angle of a lmost 2a j ust where the win g
intersects
the fuselage sidewall. Secon d, the fuselage itsel f is a much less efficient gen
erator of lift (the slender-body lift is zero), and its contribution arises mainly
from the difference in pressure carried over onto it from the upper and lower
wing surfaces. The net effect, as compared with an isolated wing, is a marked
increase in r unning load just outboard of the wing roots but a reduction over
the body. The total lift for a given angle (or "lift-curve slope" CLJ is found

about 10 percent, and decreased when the body is larger. The fuselage is also
to be increased when the ratio of fuselage diameter to wingspan is less than

found to experience a nose-down pitching moment, which can be attributed


Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

through slender-body methods (Ferrari 1957, Section C, 8) to the curved

A typical calculation is shown in Figure 1. These results are taken from


flow fiel d induced along its axis by the wing circulation.
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

1.0 r-------r---.---

0.07
0
hi =

hi =

0.8 1-=::-l'---+---;;:__1f__---_t_---__+---

O.B
0.6
=

Y.. =

Cl' = 0.3

0.21---- 1
..a
.=

0.2 0.4 0.6 0.8 1.0


O---------L----------------------

Tl
o

FIGURE 1. Spanwise distributions of dimensionless bound circulation per unit


a ngle of attack, at three subsonic Mach numbers, for the indicated wing-body com
bination and isolated wing. Wing is untwisted, and both wing and body axis are
inclined at a to the flight direction. (Taken from Fig. C. 9a of Ferrari 1957).
436 ASHLEY & RODDEN

Ferrari (1957). They were found by his adaption of Multhopp (1941), who
proposed that lifting-line theory for isolated wings be corrected by centering
in the wing an infinite circular (or elliptical) cylinder and essentially adding
image trailing vortices so that this cylinder is caused to be a streamline of the
flow in the "Trefftz plane" far downstream of the wing. Such numerical esti
mates can be shown to agree quite well with suitably interpreted measured
data.
Many other cases like the foregoing could be referenced. The approach
to wing-tail interference in Chapter 1 of Ferrari (1957) is one. He makes the
assumption-usually exact in supersonic flow because of the "law of for
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

bidden signals" but questionable at M < I that any influence at the wing
-

of the flow around an aft stabilizer is negligible. I t becomes only a question


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

of estimating the flow field or "downwash" at the tail due to the wing and its
wake, then calculating the tail loads as those on an isolated surface exposed
to known free-stream perturbations. When a central body is present in such
problems, it is typically idealized as an infinite cylinder whose presence can
be simulated through imaging. For instance, the supersonic wing-body
theory of Morikawa ( 1952) combines this model with Laplace transforma
tion on the x coordinate to simplify the analysis. Since it proves difficult to
handle cases where the wing plane does not contain the fuselage axis, most of
these theories deal only with this midwing position .

Also in the category of "simplified approaches" are the valuable arsenal


of methods for the design of swept wings and fuselages in combination, which
have stemmed from the pioneering work of Kiichemann (e g Kiichemann
. .,

& Weber 1953) in Great Britain. For subcritical, subsonic flight these are well
summarized in Bagley (1962) while Lock & Bridgewater ( 1967) describe the
more recent extensions to the transonic and supersonic ranges.
As Lock & Bridgewater ( 1967) point out about their subject, "a useful
and possibly superior alternative might thus be provided to some of the older

reference to physical reality. " These same words contain a valid criticism of
design methods which are based on theoretical considerations without much

most of the contents of Sections 2 and 3 below, which are really just another
evolutionary step toward the remote goal of flow-field analysis for wholly
arbitrary configurations. Current practice in the United States tends to
demonstrate that techniques such as Kiichemann's and those of Woodward
( 1968) are complementary, in the sense that the former serve to guide the
preliminary shaping of the flight vehicle whereas the latter are useful for de
termining the entire near field at later design stages. Formal configuration
optimization, for maximum supersonic lift-drag ratio and the like, is another
highly mathematical process for which the bridge to simple physical under
standing has not yet been built.

Nonlinear numerical methods for supersonic jl.ight.-Of particular impor


tance in this connection is the flow over hypersonic aircraft or entry vehicles,
which have blunted noses for thermal protection and therefore generate
WING-BODY AERODYNAMIC INTERACTION 437
mixed subsonic-supersonic flow fields. The many approximate schemes de
veloped prior to 1959 have been thoroughly reviewed in the book by Hayes &
Probstein (1959) . Typically the elliptic field up to the sonic line was first de
termined by finite-difference integration of the governing equations behind
some estimated position of the bow shock wave. At higher Mach numbers,
the "shock layer" is thin, and quasi-one-dimensional approximations like the
metho of integral relations simplified this process and yielded the shock
shape p'roduced by a given body. Method of characteristics was the early
choice for analysis of supersonic regions and still proves valuable for certain
wholly supersonic flows (see, e.g., Powers & Beeman 1971).
Unfortunately, despite its conceptual elegance and apparently efficient
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

organization of the calculations, method of characteristics has proved a dis


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

appointment for flows in three and four (i.e., with time dependence) dimen
sions. The work of Sauerwein (1964) , for example, succeeded in overcoming
many of the difficulties of geometrical conceptualization and stability of
multidimensional characteristics. Nevertheless, long computer runs re
mained, as did a slow drift with time of the computed results in cases where
there should have been a transient approach to steady flow.
As of the present writing, none of the nonlinear numerical schemes has
been applied to general wing-body combinations. The most promising would
seem to be that of Moretti and collaborators3 (e.g., Grossman, Marconi &
Moretti 1971; Moretti & Abbett 1966) . The flow around blunt noses is ren
dered hyperbolic by introducing artificial time dependence, then conducting
a finite-difference solution of the unsteady equations until the steady state is
reached. For supersonic and hypersonic regions, a straightforward down
stream " marching" computation is chosen in preference to characteristics.
The instability of many competing methods is not encountered. With cur
rently available computer capacities, it is believed that fairly complicated
interacting configurations will soon be analyzed in this way.
* * *

The foregoing lists of topics include selected literature citations which, in


turn, cover much of what has been published on wing-body interaction. By
way of bringing the subject up to date, we mention here the proceedings of
three recent conferences, which contain additional relevant material. The
first of these was a symposium held at NASA Ames Research Center ( NASA
1969) , where at least nine of the papers presented were germane to the sub
ject of this article. The second and third (AGARD 1970, 197 1) deal specifi
cally with interference and were organized by different panels of the NATO
Advisory Group for Aeronautical Research and Development (AGARD).
The latter (AGARD 197 1) contains extensive theory and examples on un
steady motion of interacting systems of lifting surfaces (three-dimensional

I Note that, in the first reference, the pattern of streamlines and fluid properties

is given over most of the length of a three-dimensional hypersonic vehicle shape.


438 ASHLEY & RODDEN
biplanes, wing-tail combinations, complete empennages, etc), which are
deemed to fall within the present scope. The paper by Rodden, Giesing &
Kalm{m also introduces body interference.
This section is closed with an account of the physical and analytical basis
of the particular theories selected for exposition below. Since the reader is
presumed to possess some familiarity with aerodynamics of a p erfect gas, he
is referred to textbooks on the subject (e.g.. Ashley & Landahl 1965) for
further detail.
The working fluid is assumed to be an inviscid perfect gas, so that in
stantaneous pressure p, density p, and absolute temperature T are connected
by the state equation
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

p = RpT (2)
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

Like the gas constant R, the specific-heat ratio 'Y is uniform throughout the
flow. The fluid particle velocity vector 0, observed in an inertial reference
frame, is assumed to be irrotational. Thus the influence of the small viscosity,
which is always present in fact, remains confined to surfaces of tangential 0
discontinuity simulating attached boundary layers and thin wakes. It is
noted that slender-body methods can approximate separated wakes by
means of free-vortex systems with "feeding sheets" (cf appropriate sections
of Nielsen 1960), but this refinement is not pursued here.
An immediate consequence of irrotationality is the existence of a velocity
potential<I>,
Q = grad <I> (3)

from all disturbance sources consists of uniform streaming at fixed speed U


Let it be assumed, for temporary convenience, that the fluid motion remote

parallel to the x direction of Cartesian coordinates. It is then customary to


define a disturbance potential cP by some such relation as
<I> = U[x + 4>] (4)
The disturbance velocity (U grad ) has components u, v, and w parallel to
the coordinate axes.
The principal further assumption that underlies most wing-body methods

(u, v, w) are small compared with U and the dimensionless deviations of the
is one requiring small perturbations: that almost everywhere in the field

5,
state variables from their free-stream values p, p, etc are correspondingly
small. Two conditions are placed on M and the disturbance parameter
(Mc'l)2 1 (5)
and
((M25) h 11) 1 +
M2 -1
(6)

Here (6) is added to ensure (for bodies, at least; cf Chapter 12 of Ashley &
WING-BODY AERODYNAMIC INTERACTION 439

equations. Under the stated circumstances, it is well known that cp or any of


Landahl 1965) the absence of transonic nonlinear terms in the differential

the velocity perturbations is governed by

(1 - M2)cpxx + cp"" + CP
2M2 M2
- CPxt - U2 CPtt = 0 (7)
U
Equation (7) enforces both conservation of mass and Newton's law of
motion. During its derivation, the state variables are eliminated through
linearization of a first integral of Newton's law known variously as Ber
noulli's or Kelvin's equation. The most useful form of the latter is one that
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

permits the pressure p to be computed algebraically from previously estab


lished kinematics of the flow. For instance, an exact exp'ression for the "pres
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

sure coefficient" Cp is as follows:

cp p - p",
-
=
1
"2P", U2

_2_ {[1 _ -y -1 (aq, + Q.Q - U2 ) "yf("Y-l I} (8)


=
-yM2 a",2 at 2 J
(Here free-stream sound sped a", is the only symbol not previously defined.)
For later purposes, it is remarked that certain truncated binomial ex
pansiops of the bracketed quantity in (8) are consistent with the small-per
turbation approximation. Thus, for steady flow near slender bodies

(9)

is appropriate, whereas for unsteady flow over thin wings the fully linearized
version holds,

Cp ---- cpt
2u 2
(10)
U U2
=

As will be seen, however, a stratagem that often yields more accurate airload
estimates in wing-body problems is the (mathematically inconsistent) as
sociation of (9) or even (8) with velocity fields determined from (7) .
In suitable combinations, three types of boundary conditions are needed
for uniqueness when solving (7). These are as follows:

Conditions at infinity.-Whatever the nature of the disturbance pro


duced by a wing-body combination, it must die out uniformly with increas
ing distance away from both the configuration and any downstream wake
that may trail behind it. When the motion is unsteady and/or the flight
supersonic, care is sometimes needed to enforce Sommerfeld's condition that
waves propagate outward at sonic speed away from their sources.
440 ASHLEY & RODDEN

Impermeability of body surfaces.-The exact requirement is that, at the


surface
F(x, y, z, t) = 0 (11)
of any immersed object, fluid particles are constrained to move according to
DF
- ==
(a
- + Q . grad
-+ ) F = 0 (12)
Dt at
Equation ( 12) obviously reduces to a vanishing normal velocity Vn at the fixed
surface whenever the flow is steady, which is a special form of Neumann con
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

The specification 0 1 of small surface slope usually permits (12) to be


dition.
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

solved explicitly for a combination of the lateral velocity components v and


W (0, the transverse radial component Vr on a body of revolution) . For in
stance, consider the upper surface of a wing that is performing unsteady
motions normal to the xy plane, to which it is nearly parallel. Then the func
tion F is given by
F = z - zu(x, y, t) = (13)
with Zu a prescribed quantity. For small disturbances,
( azu
az" 1 az,,
-,-,-- 1
) (14)
ax ay u at
whence it follows that (12) can be approximated by

-+ u-
az" az"
w(x, y, Zu, t) = (15)
at ax
for all (x, y) on the planform a,ea. A further approximation to (1St, accord
ing to which it is enforced not at Z=Zu but at z=O+, the upper side of the
planform's projection, is adopted in isolated wing theory. It is not acceptable
for body surfaces, however; the manner of its application to interaction prob
lems will be taken up later.

The Kutta condition.-For lifting surfaces it has been found that an addi
tional specification is needed to render the bound circulation unique. This
takes the form of requiring flow field quantities to behave continuously when
passing across the (sharp) " trailing edge," which is defined to be that portion
of the planform boundary whose outward normal in the xy plane makes less
than a right angle with the free-stream direction.
By way of anticipation, one may summarize the favored approach to in
teraction theory as follows. The boundary-value problem is solved by super
position of suitable singular solutions of a linear differential equation
[equation (7) or its steady counterpart]. " Nonlinearity" is introduced by
WING-BODY AERODYNAMIC INTERACTION 441
satisfying boundary condition ( 12) at or near the actual positions of the wing
and body elements. If j ustified by improved agreement with experimental
data, a nonlinear relation such as (8) may be employed for pressures and
aerodynamic loads.
There are several dimensions along which the subject matter may be
further subdivided. For steady flow the most clear-cut distinction occurs be
tween subsonic and supersonic flight, which characterize Sections 2 and 3, re
spectively. It has also been customary to separate steady and unsteady
motion of the boundaries, the latter solutions being useful in aeroelasticity

treat steady flow as a limit, for vanishingly small frequency w, of the case of
and certain dynamic response calculations. The authors prefer, however, to
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

faces Cd the conference proceedings, AGARD 1971), unsteady aerodynami


small simple harmonic oscillation. Particularly for interacting lifting sur
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

at w=O.
cists have taken the lead in developing methods that also have great utility

Such interactions as those between wing and tail or some other non
coplanar lifting system are regarded as special cases of a multi-element con

surfaces also belong in this category (d presentation by Ashley in Part I of


figuration with bodies present. It is believed that certain types of control

AGARD 1971) .
Some theories are better adapted to the determination of pressure dis
tribution and loads all over the surface of the wing-body system; resultant
forces and moments are then found by numerical integration. Others orga
nize the computation in such a way that some or all of the forces are more
conveniently obtained from momentum flux through a far-field control sur
face. These are, respectively, exemplified by the Woodward and the NASA
Langley approaches described in Section 3.
Finally, the authors discern important distinctions based on the degree
to which the chosen singularities violate local streamline continuity over the
surfaces on which they are distributed. All interaction theories must be de
signated "finite-element m.ethods," because they satisfy the flow tangency
condition only at a finite number of stations and thus d etermine the strengths
of a finite number of singularity elements. However, these singularities vary
all the way from the pressure modes used in subsonic wing theory, each of
which gives rise to a continuous distribution of upwash w over the entire
lifting-surface planform, to vortex lattices and doublet patches of constant
strength, which produce infinite w around their boundaries and in the wake.
A systematic study of the implications, for computational efficiency and ac
curacy, of various element choices has yet to be made (see, however, the at
tempt discussed by Djojodihardjo & Widnall 1969). It would have signifi
cance for the refinement of interaction theory.
2. INTERFERENCE AT SUBSONIC SPEEDS

Isolated planar surfaces.-The general formulation of the subsonic lifting


problem was given as an integral equation by Kiissner (1940), using Prandtl's
442 ASHLEY & RODDEN
acceleration potential. The acceleration (or pressure) potential eliminated
the need for consideration of flow-field characteristics, e. g., the wake, not on
the lifting surface and provided the basis for most later subsonic develop
ments. Kiissner suggested a solution to the integral equation by series
methods, and the terms in the series used in later studies assumed the form of
pressure loading functions that exhibited the properties of exact two- and
three-dimensional solutions, i.e., the square-root leading edge singularity, the
Kutta condition at the trailing edge, and the elliptical spanwise variation.
The general formulation was soon followed by various forms of practical
implementation of the series solution based on the pressure loading functions.
Among these are the variations of Falkner (1943) , M ulthopp (1950), and
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

Truckenbrodt (1953) for the case of steady flow. Practical implementation


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

of Kiissner's general for' m ulation for oscillatory motion awaited a suitable


computational form for the kernel of the pressure-downwash integral equa
tion. This was obtained by Watkins, Runyan & Woolston ( 1955). Practical
solutions to the oscillatory lifting-surface problem by what became known as
the kernel-function method were presented shortly thereafter by Stark
( 1958), Watkins, Woolston & C unningham (1959), and Laschka ( 1963b) .

Interfering nonplanar surfaces.-A logical development of lifting-surface


theory would have included the addition of control surfaces as the next step.
However, the flutter problems encountered in the development of the T-tail
configuration, followed later by flutter problems in the variable-geometry
(swing-wing) configuration (see, e.g., Topp, Rowe & Shattuck 1966), served
notice that aerodynamic coupling among interfering surfaces was a more
immediate concern.
Expressions for the kernel of the integral equation relating lifting pressure
and normalwash on multiple nonplanar interfering surfaces have been given
by Laschka (1963a), Rodemich (see Vivian & Andrew 1965), Yates ( 1966),
Landahl (1967), and Berman, Shyprykevich & Smedfj eld ( 1970). Expres
sions are given for all three velocity components, downwash, sidewash, and
backwash, by Laschka and Berman et al; however, it is noted that the ex
pressions of Berman et al contain typographical mistakes that prevent
reconciliation with Laschka's expressions. The expressions of Rodemich,
Yates and Landahl give only the normalwash perpendicular to the free
stream, i.e., downwash and sidewash. The signs in Landahl's expressions
were clarified by Harder & Rodden (1971).
The development of the nonplanar kernel function was followed immedi
ately by practical solutions using the pressure loading functions appropriate
for the intersecting fin and stabilizer of the T-tail configurations. Two similar
solutions by the kernel function method were given in 1964 by Stark ( 1964)
and Davies ( 1964) . The recognition of the flutter problems resulting from
wing-tail interference in variable-geometry configurations was immediately
followed by a number of special solutions for wing-tail interference. These
include developments by Laschka & Schmid (1967), Albano et al ( 1970) ,
WING-BODY AERODYNAMIC INTERACTION 443
Davies (Part I, AGARD 1971), Bohm & Schmid (Part II, AGARD 1971),
Dat & Akamatsu (Part I, AGARD 197 1) and Andrew (197 1) . Although the
technique employed in these procedures for treating interference is general,
the computer algorithms for obtaining solutions have been selected with
specific configurations in mind. This is a practical limitation since the pres
sure loading functions for each component of a multiple-surface configura
tion must be chosen according to its boundary conditions, e.g., the lifting
pressure must approach zero at a free edge (tip) but not at an intersection
(root) . The method of B ohm & Schmid perhaps is the most systematic of
the various solutions and is capfLble of analyzing quite general configurations.
As in any series method, the solution of the pressure-normalwash integral
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

equation by the kernel-function method contains truncation errors, and


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

questions of convergence must be answered. The exact solution for the circu
lar wing oscillating in incompressible flow (Schade & Krienes 1940) provides
a test for convergence, but no such test exists for arbitrary planforms. For
the most part, questions of convergence are answered by comparative calcu
lations on standardized configurations (such as those proposed by the Struc
tures and Materials Panel of AGARD) . This is a necessary but insufficient
condition for convergence, particularly since similar methods are being com
pared. A comparison of solutions for the generalized aerodynamic forces on
an oscillating T-tail has been reproduced by Ashley (Part I, AGARD 1971),
and some indication of the scatter of results by various methods is shown;
Andrew (1971) has since obtained new results, but the general question of
convergence still deserves further study,

Finite element methods.-The previous discussion was limited to the ker


nel-function method because it has been the outgrowth of traditional meth
ods that were devised long before the appearance of the high-speed digital
computer. The practical constraint that was placed on the early develop
ments was a minimal number of unknowns and, hence, the series solution
was chosen using pressure loading functions that, presumably, yielded rapid
convergence . Although the elegance of current kernel-function method
solutions could never have been achieved without the digital computer, the
philosophy of the approach dates from 1940. As Argyris ( 1970) has noted,
a digital computer is not a larger-scale desk calculator. With the digital com
puter available, the number of unknowns in a system of equations may be
permitted to rise considerably, and therefore assumptions of pressure loading
functions become unnecessary. Since they are not always suitable, it may
be preferable to avoid them, particularly when seeking solutions for non
planar interfering configurations.
Such was the reconsideration of the vortex-lattice method of Falkner
( 1943) for steady flow in the light of the high-speed digital computer of the
early 1960s. The use of discrete horseshoe vortices with swept bound legs
was demonstrated in a number of similar and independent developments of
the vortex-lattice method by Rubbert (1964), Dulmovits (1964), Hedman
444 ASHLEY & RODDEN
( 1965) , and Belotserkovskii (1967) . Later solutions for steady flow have
replaced the horseshoe vortex system by a quadrilateral ring vortex system
on the mean camber surface throughout the lifting surfaces except near the
trailing edges, where a horseshoe vortex is placed. This approach has been
taken by Monical (1965), Rubbert & Saaris (1968) , and Tulinius (1971) .
Rubbert and Saaris also account for thickness effects by placing a system
of source p.anels on the airfoil surface, whereas Tulinius treats thickness by
using a source-lattice system on the mean camber surface. A rather different
finite element is the constant pressure panel proposed by Woodward (1968)
in his study to unify the subsonic and supersonic analyses of wing-body
combinations. Unfortunately, Woodward's unified approach does not obtain
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

the accuracy at subsonic speeds that it achieves at supersonic speeds.


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

An extension of Falkner's method to the oscillatory case was given by


Runyan & Woolston ( 1957) in an early form of the kernel-function method.
This solution was soon reconsidered in view of the successes of the newer
vortex-lattice method. The extension of the vortex-lattice method has be
come known as the doublet-lattice method and resulted from similar and
independent developments by Stark (see Landahl & Stark 1968) , Albano &
Rodden (1969) , Petkas ( 1969), and Houbolt (1969). A number of further
developments and applications of the doublet-lattice method have been
made by Rodden, Giesing & KiiJman (e. g., Part I I, AGARD 197 1 ) . Although
the algorithm of Albano and Rodden has received wide acceptance in de
velopment of computer programs, a number of alternative formulations have
been proposed by Landahl & Stark ( 1968) and by Houbolt (1969); these are
deserving of further investigation.

Control surjaces.-As noted before, the inclusion of control surfaces in


the analysis of lifting surfaces was the next order of business until practical
design problems on interfering configurations assumed greater importance.
The analysis of multiple control surfaces creates no special problems for
finite-element methods. Convergence is always a question, of course, and the
control surface requires a close spacing of elements near the hinge line and
side edges, but there is no difficulty in principle. On the other hand, the
pressure loading functions required by the kernel-function method introduce
a new boundary-value problem into lifting-surface theory. Basic investiga
tions of this problem have been made by Landahl ( 1968), Ashley & Rowe
( 19 70), White & Landahl ( 1968), and Navarro Crespo & Cunningham (1969) .
Applications have been rather limited, although two recent investigations by
Darras & Dat (Part I I, AGARD 1971) and by Hewitt (Part II, AGARD
1971) appear promising. An earlier development by Berman et al (1968)
was based on rather arbitrarily chosen pressure loading functions and has
not been observed to exhibit reliable convergence characteristics.

Wing-body interjerence.-As mentioned in Section 1 , the early develop


ments of wing-body interference theory included slender wing-body methods
and lifting-line methods. The slender wing-body theory was initially de-
WING-BODY AERODYNAMIC INTERACTION 445
veloped by Spreiter ( 1948) and Ward (1949) and later extended to unsteady
flow by Miles (1959). The simplicity of this theory makes it highly desirable
for load prediction, but, unfortunately, it is severely restricted in its applica
tion, although a recent refinement has been made for transonic flow by
Sprciter, Stahara & Frey (NASA 1969, pp. 53-73) . A lifting-line theory was
developed by Multhopp (1941) for high-aspect-ratio wing-fuselage combina
tions. The upwash at the lifting line is determined in the Trefftz plane where
the fuselage has been mapped onto a vertical slit. The mapping affects the
position and shape of the wing, e.g., a wing not at midplane is warped into a
nonplanar shape. Refinements to this technique have been made by Weber,
Kirby & Kettle ( 1956) and others. Other similar variations of this approach
have also been considered by Kuchemann & Weber (1953) , and a recent
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

A vcry powerful tool for predicting wing-body interference is the method


progress report has been presented by Ktichemann (AGARD 1970) .
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

of images. The method of images is based on the Thompson Circle Theorem


of hydrodynamics. The basic idea is to match each singularity external to a
body with one internal to thc body. The image exists to negate the flow
through the body surface generated by the external singularity. The strength
and location of the image are directly related to the strength and location
of the external singularity and the body shape; thus, no new unknowns are
introduced. The image system is not completely effective in reducing the
body surface to a streamline since two-dimensional theory is used to develop
it. Wu & Talmadge (1961) have developed a very complicated expression
for a "residual" potential that, when added to the image system, renders the
body a streamline. Actually they developed the method for wings passing
through circular jets (assumed infinite in length) , but the techniques can be
applied to bodies. Wu & Talmadge concluded that the residual potential flow
is small compared with that generated by the external singularities and their
images. That such is the case is borne out by many theories that ignore the
residual flow.
Zlotnick & Robinson ( 1954) and Gray & Schenk (1953) have used the
unimproved image system (without the residual potential) . Zlotnick and
Robinson used a discretized Weissinger approach in which the wing is ap
proximated by a series of unswept horseshoe vortices (lying along the one
quarter chord) and their images. Gray and Schenk have used the image
approach in an iterative manner. The wing loading is first calculated exclud
ing body effects. Next an image system of this loading is placed within the
body and its effect on the wing determined. The wing loading is then recalcu
lated, using the new upwash caused by the image system.
A general lifting-surface theory for wing-body combinations, which uses
the method of images, has been developed recently by Giesing ( 1968b) . The
vortex-lattice method is used to represent the lifting surfaces. The fuselage
is represented by an image system and an axial doublet system. Each horse
shoe vortex is given an image within the body. The residual flow is accounted
for by using a piecewise constant doublet distribution.
The image methods discussed above apply to bodies of circular cross
446 ASHLEY & RODDEN
section. Borland (1966) has extended the image concept to elliptic cross
sections. Chou ( 1966) has generalized the image concept to include nacelles;
an attempt was also made to generalize the image approach so that longi
tudinal variations in cross-sectional area can be considered. Recent works by
Spangler & Mendenhall (NASA 1969, pp. 703-19) and by Giesing et al
(1971, Part I I) have extended the image technique to very complicated
configurations; e.g., Spangler and Mendenhall present a span loading for
transport aircraft with both nacelle and pylon effects. The method developed
by Giesing et al is very general and can determine the loading (chordwise
and spanwise) on an assemblage of lifting surfaces (wings, tails, pylons, con
trol surfaces, etc) and bodies (fuselage, nacelles, stores, etc) in or out of
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

ground effect for either steady or oscillatory motion. Lifting surfaces may be
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

of arbitrary planform and dihedral. Bodies may have elliptic cross sections
and an arbitrary distribution of width or radius. This method is a combina
tion of lifting-surface theory, the method of images, and slender-body theory.
The doublet-lattice method is used to model the lifting surfaces. Slender
body theory is used to model the bodies. The interference between the lifting
surfaces and bodies is obtained by the use of images and slender-body theory.
Example calculations by this approach are shown in the next section.
Body-element methods are those that use surface singularities to satisfy
the Neumann boundary conditions on the body surface. These methods may
be placed in two categories : those (a) that use lifting surface elements and
(b) that use source elements. In the first category are the methods of Wood
ward (1968), Kalman et al (1971) , and Giesing et al (1971, Part I) . The gen
eral approach of these methods is to replace the body by an axial doublet
distribution and an annular wing of constant cross section. Elements cannot
be placed on the actual body shape nor can the elements be inclined to the
flow. The kernel function of Berman et al (1970), mentioned above, was de
veloped for lifting surfaces that are inclined to the flow. The geometry of the
wake associated with this kernel function, however, is the same as that asso
ciated with the classic kernel, i.e., the wake extends to downstream infinity
from the sending point along a line parallel to the free stream. If this type of
kernel is used on an inclined body surface, then wakes emanating from
points near the body leading edge will thread through the body surface near
its trailing edge. This situation is not physically meaningful and may lead
to numerical difficulties.
Source-type elements are placed on the body surface for the methods of
the second category. A distribution of vorticity is placed either on the wing
camber line or on the surface while a distribution of source or doublet ele
ments is placed on the wing and body surfaces. In this category are the
methods of Rubbert & Saaris (1968), Labrujere et al (see AGARD 1970),
Djojodihardjo & Widnall (1969), Hess (1970), and Tulinius (1971) . The
method of Tulinius is somewhat different from the rest. Sources and vortices
are placed on the wing camber line while quadrilateral vortices are placed
on the body surfaces. These methods furnish surface pressure distributions
and not just loading, but they are restricted to steady flow. Of all of the
WING-BODY AERODYNAMIC INTERACTION 447
methods described above, only slender wing-body theory, the method of
Kalman et al (1971) , and the previously mentioned methods of Giesing et al
furnish solutions for wing-body combinations in oscillatory motion.

Theoretical and experimental correlations.- The success of any theory must


ultimately be measured by its ability to predict accurate experimental re
sults, although an intermediate criterion is that different approaches to a
particular mathematical idealization of a ph,ysical problem should yield the
same results. The kernel-function method has enjoyed a traditional Ifosition
in the solution of the aerodynamic lifting-surface problem: a series solution
to integral equations by collocation methods is a classical mathematical
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

approach, and the convergence of the usual series of pressure loading func
tions has always seemed assured in view of its selection on the basis of exact
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

two- and three-dimensional solutions, the current difficulties with the control
surface problem notwithstanding. The finite-element methods, on the other
hand, seem to possess none of the traditional values other than some approxi
mation to the calculus of infinitesimals. Indeed, the success of the finite
element methods remains to be explained, although the study by James
(1969) of the two-dimensional steady case sheds some light on the correct
convergence characteristics of the finite-element methods. Nevertheless, the
finite-element methods have compared favorably with the kernel-function
method where they are comparable, and both have been compared favorably
with experimental results. The original developments of the vortex- and
doublet-lattice methods were compared extensively with experimental and
other theoretical results in the references cited above. Later correlations
have been presented by Rodden & Liu (1969) and by Ka.lman, Rodden &
Giesing (1971) . Correlations for the constant-pressure-panel method of
Woodward have been presented by Bradley & Miller (1971).
An experimental study by Forsching, Triebstein & Wagener (Part II,
AGARD 1971) has yielded extensive pressure measurements on a swept wing
with two control surfaces oscillating in an incompressible flow. LaBarge
(1971) has made a correlation study using Petkas' version (1969) of the
doublet-lattice method. Results for two of the modes of vibration are taken
from LaBarge (1971) and are shown in Figures 2 and 3. The calculations
were carried out by dividing the span into eleven unequal-width strips; the
strip widths were chosen so as to place the centerlines of three of the strips
at the three measurement stations. Two chordwise divisions were considered
to investigate convergence. Nine and thirteen unequal chordwise divisions
were chosen, the small chord-length elements being concentrated near the
leading edge and hinge line. Comparison of the 99-element solutions with the
143-element solutions indicated convergence of the 99-element solutions, but
an improved definition of the leading-edge and hinge-line singularities results
from 143 elements.4 Figures 2 and 3 correspond to Figures 9 and 15, respec
tively, of Forsching, Triebstein & Wagener (Part I I , AGARD 1971), al-
4 Note that only the theoretical results for 143 elements have been included on
the figures.
448 ASHLEY & RODDEN
.09

.06

Z U '0 3
1- - 0.
w <l
Q
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

1..1..
I..I.. w

U
a:: 0
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

w
a::
::::>
Of
if

w
-:03 <
--- <
a:: a w :' O . O
Cl.
a i f 0 . 67
=


-: 0 6 CI o f = - 0 . 6 5
: 0.
.02
I 0 0
>
0 00 0

z
ffi
o kR
- -
-
sVj-
- -E....!!.- v

-

0 2 ----2
--4---- 60
H

e8 0---- I OO
i n g-
PERCENT CHORD
0 0 0

FIGURE 2. Comparison between doublet-lattice theory and measurements of the


pressure difference tlei> between the lower and upper surfaces of a wing with two
flaps. The inboard flap (if) and outboard flap (of) oscillate sinusoidally with the
amplitudes and phase indicated. Real and imaginary parts of tlCp relate to com
ponents of the sinusoidal loading in phase and 90 ahead in phase with angular dis
placement of the inboard flap. (See text and references for further detail. Figures 2
and 3 are adapted from LaBarge 1971.)

though Figure 3 has been corrected to show that the outboard flap was locked
during the test run and to reverse the sign of the imaginary part of the pres
sure coefficient. GeneraIIy good agreement is noted between the test and
calculated results. However, significant differences appear toward the trailing
edge of the control surfaces ; it is apparent that accurate hinge-moment calcu
lations will be obtained only when viscous interactions are accounted for.
Data for wings with control s urfaces have also been p resen ted by Tijde
man & Zwaan (Part I I , AGARD 1971). The data are for transonic Mach
numbers including the high subsonic and low supersonic regimes. Correlation
WING-BODY AERODYNAMIC INTERACTION 449
.06 of if
aw"" 0.17
aif = - 0 . 80
a of = 0 . 0
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

-:09

100
PERCENT CHORD
20 40 60 H i ng e 80

FIGURE 3 . Comparison between doublet-lattice theory and measurements of the


pressure difference I!. Cp between the lower and upper surfaces of an oscillating wing
with two flaps. Conditions similar to Figure 2 except that wing oscil1ates in pitch,
with outboard flap (of) locked and inboard (if) moving in opposition. Phase of I!.Cp
referred to wing angular displacement.

studies of these data would indicate the extent to which shock waves should
be accounted for in addition to the viscous effects already noted above to be
important in the prediction of hinge moments.
Next considered are correlations (Figures 4-7) taken from Giesing, Kal
man & Rodden (197 1 , Part I I) for wing-body combinations. Figure 4 presents
experimental data for spanwise loading obtained by Martina (1956) for a
highly swept wing on a long cylindrical fuselage. The first wing-body theory
of Giesing (1968b) , which includes the method of images, is compared with
the experimental data in Figure 4; the refined wing-body theory of Giesing
450 ASHLEY & RODDEN

o.5 .....---

0.4

0. 3
kr = O.O
o 11
),=0.34
AR=8.0

M =0.2
c = AVERAGE CHORD
-0- GIESING, KALMAN S RODDEN
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

- GIESING
o EXPERIMENT, MARTINA
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.


O
I NG

(' 1 -_
W
7 O. O .5
A--0 . --
O
--
0
.-Q O
8-- . 9---7
I
'

y /s

FIGURE 4. Comparison between measurements and two predictions of the steady
subsonic spanwise load distribution on a swept-wing-body combination at angle of
attack; Ct, C, X, and AR are sectional lift coefficient, local wing chord, taper ratio,
and aspect ratio, respectively. (See text for further detail.)

et al (1971, Part II) is also shown. Both theories show excellent agreement
with the data. Another comparison of the refined wing-body theory of Giesing
et al with experimental data for spanwise loading is shown in Figure 5 for an
aircraft with wing-mounted engine nacelles. The data lie above the calculated
results, probably because of the development of a leading-edge vortex on the
highly swept wing. A final comparison of wing-body interference in steady
flow is shown in Figure 6, which compares spanwise loading calculated by
the methods of Labrujere et al (AGARD 1970) and Giesing et al with
experimental data of Korner (1969) . The experimental data were measured
at low Reynolds number and thus show the classic reduction in lift caused by
viscous boundary-layer effects. The results of Labrujere et al lie above those
of the image method because those authors have accounted for effects of wing
thickness that are not considered in the image method. The only experi
mental lifting-pressure data known to the authors for wing-body combina
tions in oscillatory motion have been obtained by Cazemier & Bergh ( 1964)
at low Reynolds number and Mach number. A comparison of these data with
the image method of Giesing et al is presented in Figure 7 ; the agreement
is generally good.
A source of indirect correlation of oscillatory aerodynamic interference
theory is a flutter test of interfering surfaces. A number of such tests have
been performed since flutter was first observed by Topp et al ( 1966) to be a
significant problem on variable-geometry configurations. Efforts to predict
WING-BODY AERODYNAMIC INTERACTION 451
0. 1 4
o
o
0. 1 2
o

0.10

o.oa
a: I I
u
.....
o GIESING, KALMAN a RODDEN
-<
u
o EXPERIMENT, . WALKER
u
0.06
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

a = 4 DEG PLUS CON I CAL CAMBER


kr = O
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

0.04 elf ROOT C HORD

0.02

0
0 .
. I --O 2 0.
3--
0.4O 5
. 0
.6--
O. 7-0
.a
-
0

1 .0
y/ s

FIGURE 5. Comparison between theory and experiment for the steady subsonic
spanwise load distribution on an aircraft with cambered wing and wing-mounted
nacelles.

the measured flutter characteristics have been made by Sensburg & Laschka
(1970) , Albano, Perkinson & Rodden (1970, Part II), Seidel & Sensburg (Part
II, AGARD 1971) , and by Mykytow et al (1970) . The predictions have
generally been reliable and serve to confirm the validity of the aerodynamic
methods employed. However, the additional parameters present in the flutter
problem, i.e., stiffness and inertia, make specific evaluation of an aerody
namic method problematical since no evidence of sources of discrepancy,
e.g., viscous effects, is obtained. Measurement of oscillatory pressures on
interfering surfaces is a preferable but, admittedly, more difficult alternative.
Experimental measurements of oscillatory air loads (lift and moment) on
two interacting lifting surfaces in tandem have been presented by Destuynder
(Part II, AGARD 1971) . However, no correlation studies of the data are
known to the authors.
3. SUPERSONIC SPEEDS

The history of finite-element methods for interacting supersonic lifting


surfaces may be traced back to the work of Etkin (1955) and Pines, Dugundji
& Neuringer (1955) on planar wings. The latter authors divided the plan
form and the disturbed area ("diaphragm") of the xy plane off the planform
into small square elements or facets. Each facet was covered with an oscilla
tory source sheet of constant strength amplitude. A matrix of aerodynamic
452 ASHLEY & RODDEN
,. /
o GIESING, KALMAN S
RODDEN
- LABRUJERE
EXPER I M ENT
kr = 0

M = O. I I S
aw = So
A = O. A R =S.O.A= 1 .0. DIe = 1 .0
a f = So
Ole = 1 .0
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

0. 1

0
0 0.1 0.2 0.:3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
y/s
FIGURE 6. Comparison between theory and experiment for the spanwise distri
bution of sectional lift coefficient Ct on a rectangular wing-body combination.

influence coefficients was constructed, giving the upwash w at the center of


each facet due to each source element whose influence could be felt there.
I n the event that the wing motion is specified and the loading to be predicted,
the source strengths were determined from a choice of conditions: for facet
centers on the planform, boundary condition (15) was enforced ; on the dia
phragm no discontinuity in pressure was permitted between the upper and
lower surfaces. Given these source strengths, we can find pressure distribu
tion and resultant loads from the linearized relation (10) .
I t i s worth remarking that the source-element scheme can actually repre
sent two different physical situations, or their combination by additive super
position. The first involves a wing of symmetrical thickness distribution at
zero a, in which case the strengths are zero off the planform and directly
proportional to thickness slope otherwise. The second consists of a zero
thickness wing with arbitrary small a , camber, and twist. The diaphragm is
then regarded as an artificial barrier between the upper and lower fields.
By appeal to antisymmetry, the source strengths and Cps on the lower side
are then known to be equal and opposite to those on top.
WING-BODY AERODYNAM IC INTERACTION 453

Q.
U
<l
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

" "

0.5 1 .0 GIESING, KALMAN a RODDEN


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

x/c o EXPERIMENT BERGH a


1 5.0 CAZEMIER

\
PLUNGE ROLL
30.0 STATION NO. 2
STATION NO. 3
\
1 0.0 \
\
q
20.0
\
\

\
b
\
\
b...
b..
''tl...
1m

-5.m:--,----,l
-u 0.5 1.0
x/c

FIGURE 7. Comparison between theory and experiment for the pressure difference
across three spanwise wing stations oscillating in the indicated motions.

These source-sheet procedures were refined by adaptation to higher


capacity computers and by introduction of other facet geometries, notably
diamond-shaped elements ("characteristic boxes") bounded by Mach lines
and rectangular elements ("Mach boxes") with Mach-line diagonals. The
extension to interacting pairs of surfaces was suggested by Ashley ( 1962) . &
This generalization was automated, among other investigators, b y Moore &
Andrew (1965) for intersecting planar surfaces and by Andrew ( 1967) for
more complex configurations.
M ore recently Kariappa & Smith (1971) demonstrated the utility of a
triangular facet, whose displacements could be made compatible with struc-
6 See also the discussion for steady flow in Chapter 11 of Ashley & Landahl (1965).
454 ASHLEY & RODDEN
tural elements used for analyzing the wing deformations. The Woodward
(1968) area element also appears adaptable to oscillatory, as well as steady,
motion of lifting systems (see Ashjley, Part I, AGARD 1971).
It is worth noting that Green's theorem for the modified wave equation
of supersonic flow yields a doublet solution that is conceptually preferable
to the source for steady or unsteady lifting problems. This approach forms
the basis of the Watkins & Berman (1956) integral equation for the super
sonic planar wing. Harder & Rodden (1971) have generalized the kernel func
tion of this equation to cover nonplanar surfaces. A recent paper by Wood
cock (Part I , AGARD 197 1) presents results for pairs of triangular surfaces,
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

found by numerical solution of the doublet-sheet equation with diamond


shaped facets.
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

The evolution of supersonic theory for isolated bodies also contains a


thread of development that seems markedly superior to the classical slender
body formulation and that has had an influence on modern interaction
techniques. Lighthill (1948) , in the case of parallel or inclined steady flow
past axisymmetric shapes that may have slope discontinuities, pioneered the
idea of distributing ring singularities along the actual surface of the body
and thereby satisfying the tangency boundary condition at the proper loca
tion F = O. I mportant generalizations appear in a paper by Kacprzynski &
Landahl (1967) . They not only broaden Lighthill's solution to cover small
simple harmonic vibrations but also show, by the method of "parametric
differentiation," how the exact nonlinear boundary condition ( 12) can be
successfully introduced. Their results agree remarkably with known solutions
for cones and with pressure measurements on discontinuous bodies. Although
they do not discuss extensions to wing-bodies or other body cross-sectional
shapes, these are believed to constitute a promising topic for future research.
With the foregoing introductory remarks, the remainder of this section
is devoted to the details of two currently popular precedures for general

scheme has seen wider application for M> 1, which is the reason for its
supersonic configurations. As mentioned in Section 2, the Woodward (1968)

presentation here. Nevertheless, several examples at high subsonic M given


by Bradley & Miller (1971) from the computer program described by Wood
ward & Hague (1969) seem to indicate that earlier dfficulties with colloca
tion-point locations have been overcome.
Only steady flow is considered, the full extension to oscillatory motion
not yet having been mechanized. Figure 8 summarizes the manner in which
various singular solutions of Equation (7) are put together to satisfy tin 0
=

at the approximate surface locations of a collection of wings, tails, cambered


bodies of circular cross section, and cylindrical ducted bodies. Along an x
axis coinciding with the axis of each closed body, there are first distributed
line sources and doublets of running strengths f. and fD. Their disturbance
potentials are, respectively,

q, = - Re {fV(x -
J.W d
}
)2 - (M2 - 1 )r2
(16)
WING-BODY AERODYNAMIC INTERACTION 455
BODY INTERFERENCE PANELS
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

FIGURE 8. The manner in which line singularities are used to represent bodies
and the arrangement of facets for lifting surfaces and wing-body interactions, for the
theory of Woodward (1968).

and
cfm =
cos 8
R
e {f !nW [x - ]d } (17)
r v(x - )2 - (M2 - 1 ) r2
where r and 8 are polar coordinates in the cross section, 8 0 being at the =

vertical xz plane. In the supersonic theory, linearly varying source and


doublet strengths are started from the conical nose at O. A series of up to
=

50 quadratically varying elements [with f. and fD proportional to ( -i) 2 ]


are started a t a set o f points i o n the body axis. At discrete stations along
=

a parabolic-arc approximation to the body surface rB(x) , the strengths of


these elements are determined in advance of the interference calculation
from the following requirements [cf (12) ] : the sources match the condition

flr =
[
U 1 +- -
acp. arB ] (18)
ax dx
associated with the axisymmetric body disturbance, while the doublets match

Vr = U cos 8
[d zc
- - a
] +
aCPD drB
U- - (19)
dx ax dx
Equation (19) means that the doublets must cancel a radial velocity that
would otherwise exist owing to the body's angle of attack a plus the slope
dzc/dx of its (slightly) cambered axis. It is worth mentioning that many
fine points (necessarily omitted here) associated with all the boundary condi
tions, formulas for u, v, and w due to the various singularities, etc, are
elaborated by Woodward, Tinoco & Larsen (1967) . The complete computer
program appears in LaRowe & Love (1967).
Figure 8 shows the mean plane of each lifting surface as represented by
(up to 100) panels or facets, each in the form of a trapezoid with two sides
456 ASHLEY & RODDEN

parallel to the xz plane of symmetry of the vehicle and the other two inclined

typical wing adjacent to the horizontal xy plane. Its thickness distribution


arbitrarily in a manner chosen to fit the planform geometry. Consider a

2zT(x, y) creates a symmetrical flow disturbance, which is represented in ad


vance of the actual interference calculations by placing sources of strength
f. (, Tf) over each facet. A typical potential is

(20)

where the integral extends over the facet area, the "real part" again being
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

taken to eliminate forbidden signals (disturbances must, of course, also be


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

eliminated from the forward Mach cone from any singularity) . As with the
body line sources, f. is taken to be constant over facets originating on the
leading edge and varies linearly with over aft facets. From a familiar prop

azTjax over that same area.


erty of source sheets,f. in any facet is completely fixed by the thickness slope

As in the case of the lifting type of area singularities, which will be dis
cussed next, the algebra of evaluating the source disturbances is facilitated

two negative and each bounded by a streamwise line to x = 00 and a swept


by representing the facet's area as the sum of four source sheets, two positive,

back leadi ng edge extending laterally to infinity (d the figure on p. SO of


Woodward, Tinoco & Larsen 1967) .
To complete the solution of the full interaction problem, lifting elements6
of constant pressure difference I::. p (proportional to I::.u in steady flow) are
placed over each lifting-surface facet as well as over rectangular facets whose
locations are picked, as closely as possible, to ensure that normal velocities
due to interaction can be cancelled at each body. Figure 8 demonstrates,
however, that in the current version of Woodward's method these rectangu
lar facets are arranged on circular cylinders with radius somehow related to
the average radius of the corresponding body. An excellent opportunity for
increased accuracy-at the cost of lengthier compu tation-would seem to
consist in locating these elements closer to the true body surfaces.
D ucted bodies are not included in the Woodward, Tinoco & Larsen ( 1967)
program, but they are now being successfully simulated (Woodward &
Hague 1969) by means of similar hollow cylindrical arrays of rectangular
lifting elements.

(l::.p/!Pct:J U2) set equal to unity, would be described by the potential


A typical lifting element on the xy plane, with its dimensionless strength

<PAp(X, y, z)
1
4 11'"
{zf r [x-t]ddTf (2 1)
= - Re
J l(Y-Tf) 2+z2]v(x-tP- (M2 _ 1) [(Y-TfP+z2J }
6 This singularity is sometimes called a "u-doublet sheet."
WING-BODY AERODYNAM IC INTERACTION
45 7

When finding velocities u , v , and w and treating the limit of (21) as zO,
one must take finite parts of the singular integrals in the usual manner.
A control point is selected for boundary-condition purposes within the
area of each facet ; Woodward, Tinoco & Larsen (1967) indicated that a
station at 95 percent of the streamwise chord through the centroid yields
most accurate results. At this station on each rectangular body facet, the
condition Vn 0 is enforced, where Vn is the normal velocity induced there by
=

the sum of all flow singularities whose effects have not already been ac
counted for by conditions like (18) and (19) .
Similarly, at each wing station, Vn is computed due to all body-axis
sources and doublets, all body and wing lifting elements, and wing sources
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

located on lifting surfaces other than the one at which the boundary condi
tion is being applied. Typically, the resulting Vn is required to satisfy an
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

equation like
(22)

which ensures that streamlines over this wing are sloped so as to account for
its angle of attack and follow its camber surface zc(x, y) . Unlike (18) and
(19) , equation (22) is wholly linearized. Together with the conditions Vn 0 =

at body facets, the conditions in (22) therefore produce a system of linear,


simultaneous equations in the unknown values of /),.p on the entire system of
facets. These unknowns may number several hundred, but the solution
proves quite straightforward by current methods.
Once all singularity strengths have been found, the pressure distribution
over the configuration is obtained by programming (8) , (9) , or (10) . When a
nonlinear form of Bernoulli's equation is used for this purpose, note that the
aforementioned /),.p should be regarded as a doublet strength rather than the
physical lifting pressure on its particular facet. Care must be taken to transfer
body pressures from the rectangular elements of Figure 8, where the interac
tion conditions are satisfied, to the actual body surface. Resultant aerody
namic forces and moments on the configuration, including drag, can finally
be summed by a rather elaborate process, which is weB described by Wood
ward & Hague ( 1969) .
Many supersonic applications of Woodward's methods can be seen in the
references already cited, as well as in Carmichael & Woodward (1966) , Car
michael (NASA 1969, AGARD 1970) , Bradley & Miller (197 1) , and else
where. Only the direct determination of loading on a given vehicle geometry

such things as adjusting zc(x, y) of a given planform to yield minimum wave


is discussed above, but newer versions of the program are capable of doing

drag at a fixed lift (or lift and pitching moment) in the presence of one or
more interacting bodies.
Figures 9 through 13 have been selected from Woodward, Tinoco &
Larsen ( 1967) . All show measured and computed pressure distributions, by
way of emphasis that such details are a natural byproduct of this theory.
458 ASHLEY & RODDEN
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

BODY PRESSURE TAPS LOCATED


-
ALONG BODY MERIDIAN LINES
AT 9 =0AND45
ALL DIMENSIONS IN INCHES

f ---- "l
z

100 WING
84 BODY PANELS
PANELS

9=0, 30 60,90 , 1 20,150 ,180

r----III;tt tl 111 !;td;


--1;;l;t IIIIIQt;t ;j -
l l l---- ..x

FIGURE 9. A wing-body configuration, with locations of taps used for wing pressure

(Figures 9 throuh 13 adapted from Woodward, Tinoco & Larsen 1967).


measurements and facets chosen for automating the Woodward supersonic theory.
WING-BODY AERODYNAMIC INTERACTION 459

o EXPERIMENTAL
-- NIELSEN THEORY
-- PRESENT METHOD
IXw 1.92
C(S D O
RN - L5xl06
p . C,- C,
90 MERIDIAN ( CC- O t

o
o
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

O L-__-+______--+_------_+--==
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

45 MERIDIAN

1!!.
CIIw
-1

O ------4-____-+____-4______________

o
TOP MERIDIAN
-2
- _ .Q _ -
-;-
fJp
--

Cltw -1

e 0
-. 2 0 .2 .8 1.0 1.2

BODY STATION AS FRACTION OF WING CHORD . x/C

FIGURE 10. Measured and predicted pressures on the surface of the body shown
in Figure 9. RN is chord Reynolds number. The values of wing and body angles of
attack a,. and aB are shown.
460 ASHLEY & RODDEN

G> EXPERIMENTAL
NIELSEN THEORY
PRESENT METHOD -
M =1.48
-3

Qw= l.92
0:6 0"
-2 RN = L5 x106
P =C -C
p p
(0: - 0)
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

-1
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

-3
v -2.58
r

/JP
rw -2

-,l

-3
r
!-1.92

,8p
Qw
-2
(i) (i) (i)
--- --- - ...:. - - - -

! -L25
r

-1

I I I
o .2 .4 .6 .8 LO
WING CHORD FRACTION X/C

FIGURE 1 1 . Measured and predicted pressures at four stations on the upper wing
surface of the Figure 9 configuration; x and y are spanwise and chordwise coordinates,
c is wing chord, and r is body radius.
WING-BODY AERODYNAM I C INTERACTION 46 1

(!)
- - - NIELSEN THEORY
EXPERIMENTAL

, -- PRESENT METHOD OCw = 2


-3
CX B - 2
l" AN a 1.5xl06
.... P Cp-C
"
Q: = O)
-2

_ ..G)-

- 1
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

-3

45 MERIDIAN
-2

,liP
tl w -1

-3

2
TOP MERIDIAN
-

/3P
;-
- 1

o .2 .4 .6 ,8 1.0 1.2
WING CHORO FRACTION X!c

FIGURE 1 2 . Data similar to Figure 10, but with both wing


and body at the same angle of attack.
462 ASHLEY & RODDEN

102.19, Ill, 155, l


PRESSURE TAP LOCATIONS
(J = 0, 25, SO, 7S,

S WING
OD
PANELS
P NE S AA
9 B Y :L
:====::_ Y= Y=5.S
Y=2.iiii
100

t l!'YI=I:
PANELING SCHEME !!::::"
M= I.S 0 = 40
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

WIND TUNNEL DATA THEO RY


o UPPER SURFACE
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

.. LOIVER SURFACE
-0.20

-0.16
-0. 12

-o.08
-1).04 . G)
Cp
x/c
r'-
e
-
t------ x/c
e -o-Q.El e
......-

Y = 4.0
0.04
e
O.OS . :
!
0.12 I
j

G Ci)
-0.20 G Q
-O'16 Q . Q . Q
QQ QGQ '
-0. 12

-0.08

-0.04
0.2

e
0,4
Ci)

0 . 6 O.S . LO

e e e
x/c

. 0.2 0.4 0.6 e e
iii

0.8 e 1.0 xlc

e ",->..--...

FIGURE 13.
: /.--- ,

Measured and predicted pressures on the upper and lower surfaces


j
Y " 7.68

of a swept, cambered wing under the conditions illustrated.


WING-BODY AERODYNAM I C I NTERACTION 463
Figure 9 depicts a simple wing-body combination tested at M 1.48 and=

analyzed by Nielsen ( 1952) , the pattern of facets used when applying the
Woodward theory, and the locations of static-pressure holes. Comparisons
between measured and predicted differences between Cp and its value at
zero incidence are shown on Figures 10, 1 1 , and 12, under circumstances
explained in the captions. For these cases there is little to choose between
the accuracy of the present method and the simpler interference theory of
Nielsen (1952), except that Woodward is more successful in estimating body
pressures near the wing leading edge.
Figure 13 shows a cambered wing centered in a body with slight boat
tailing. The theory is again seen to estimate local Cp quite well at four span
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

wise wing stations. These results are perhaps more satisfactory than
individual point-by-point comparisons might suggest, since it is notoriously
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

harder to predict local pressure than lifting-pressure difference on such a


configuration.
The second wing-body interaction theory selected for emphasis here
really comprises a collection of techniques that have been evolved over more
than a decade at NASA Langley Research Center. Their origin is the linear
ized theory of Hayes (1947), who conceived of the disturbance produced by
a vehicle in supersonic steady flight as the effect of volume, lifting, and side
force elements. He suggested the computation of inviscid drag from the far

volume with generators parallel to the free stream (d the account in Chapters
field momentum flux through the sides and rear face of a cylindrical control

9-10 of Ashley & LandahI 196S) .


Carlson & Harris (pp. 639-58, NASA 1969) and Shrout (Article 8,
A G A R D 1970) are current sources that describe the computer program in
use at Langley. To paraphrase the former article, the analyst requiring
information on resultant supersonic airloads has a choice. He may adopt a
near-field method like Woodward's but does so at the cost of a very time
consuming numerical calculation of wing and body pressures. Alternatively,
he may select Hayes' momentum approach for all significant forces and
moments except skin-friction drag, whose determination falls outside the
present scope (see the cited literature) . Nevertheless, a separate near-field
computation is still needed to ascertain the lift and sideforce elements ap
pearing in the momentum formulas. As a way of minimizing the impact of
this dilemma, the Langley researchers combined the most convenient fea
tures of the two approaches.
Taking the inviscid supersonic drag as an example, one notes that it may
be approximated by the sum of the zero-lift wave-drag due to the volumetric
displacement of all bodies and lifting surfaces, the vortex drag due to lift
that shows up as kinetic energy of trailing vortices, and the wave drag due
to lift that appears as momentum flux across the lateral cylindrical control
surface. The first of these is most readily found from the Hayes ( 1947)
"supersonic area rule"
464 ASHLEY & RODDEN
2
Dow U
p""81T f 2.-f tf tS//(XI, 8)S//(xz, 8)ln I
0 0 0
Xl - X2
e
1 dx1dxzd8 ( 23)

Here two integrals are taken over body length l in terms of dummy variables
Xl and X2. Sex, 8) is the sectional area of the entire configuration cut by a plane
passing through the axis at station x, inclined at the Mach angle sin-1 (1/M)
to the flight direction and also normal to a plane through the axis at the angle
8 defined below ( 1 7) (d Figure 9-6 of Ashley & LandahI 1965) . The two parts
of drag due to lift are obtained by the method of Middleton & Carlson ( 1965),
which is a near-field scheme based on pressure elements resembling Wood
ward's but applied to a zero-thickness projection of the entire vehicle having
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

the twist, camber, and a of the mean surface.


by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

From the latter drag contributions, one can omit the "leading-edge suc
tion," which is harder to eliminate in the momentum formulation of the
theory and which, in the physical world, is not usually achieved at subsonic
edges of thin wings.
The assumptions of zero volume in the computation of drag due to lift
and of zero lift in the volumetric wave drag prevent consideration of mutual
interaction between volume and lift. That such interaction may occur is
evident from the nonlinear relationships between wave drag and the various
element strengths. Carlson and Harris assert, however, that it "appears to
be negligible for slender supersonic-transport configurations but . . . may
be significant for supersonic-dash vehicles." It is noted that a recent paper
by Bonner (1969) is representative of current work that accounts for this
interaction and may point the way to its future inclusion in the Langley
programs.
Figure 14 reproduces, from Carlson and Harris, typical computer-graphic
representations of a fighter with close-coupled wing and horizontal stabilizer,
as these might be employed for the two parts of the potential-drag computa
tion. Skin-friction drag is determined primarily from wetted area, Reynolds
number, and Mach number. All such quantities as lift, pitching moment,
and sideforce are evidently calculated from the zero-volume model by nu
merical summation of the forces experienced by individual loaded facets at a
particular choice of angles of attack and sideslip.
Figure 15 illustrates the manner in which the components of drag might
be expected to vary with lift for a supersonic bomber such as the B-58, with

In Figures 16 and 1 7, the quantities CD, CL, and Cm are the total drag, lift
its surfaces cambered so as to minimize drag at some cruising flight condition.

force, and nose-up pitching moment, each rendered dimensionless by the

(P0lJ/2) U2, wing planform area, and (in the case of Cm) mean wing semichord.
standard aeronautical practice of dividing by flight dynamic pressure

The particular configurations and Mach numbers are indicated on sketches.


In Figure 17, the symbol 0 represents angular deflection of the trimming
control surface, which is a canard elevator on the right-hand vehicle and a
WING-BODY AERODYNAMIC INTERACTION 465
COMPUTER DRAWINGS OF NUMERICAL MODELS

MODEL USED FOR WAVE DRAG


Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

FIGURE 14. Computer-graphic representations of the two models used for drag
prediction of a. fighter. (Figures 14 through 1 7 adapted from Carlson & Harris, NASA
1969.)

conventional all-movable stabilizer on the left. At the scales given, the mea
sured values of the various aerodynamic quantities show quite acceptable
agreement with predictions. A possible exception is seen in the drag estimates
for the fighter in Figure 16. Carlson & Harris (NASA 1969) remark that
"other comparisons of the fighter aircraft data with theoretical results indi
cate that the discrepancy is due more to the vertical displacement of airplane
components than to component thicknesses."
CONCLUDING REMARKS

I t should be evident from the foregoing all-too-brief account of interac


tion theory that it is both a complicated subject and one in which computer
automation is more nearly in a state of revolution than evolution. Within a
few years, programs should be available that will solve the linear potential
equation, with boundary conditions satisfied by placing appropriate discrete
singularity elements at a close approximation to all the true wing and body
surfaces. The following " nonlinearities" will be included : pressure-velocity
relations such as (8) and (9) ; boundary conditions that partially account for
x-velocity perturbations as in ( 1 8) and ( 19) ; wakes trailing streamwise from
the actual positions of trailing edges; and/or estimates of self-deformation
of wing wakes as they affect aft tail surfaces and the like. Such programs will
play an important role in optimizing configurations for maximum aerody-
466 ASHLEY & RODDEN

- NEAR FIELD
DRAG
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

U FT

FIGURE 15. Schematic of the drag build-up for a supersonic airplane.

S K I N-FR I CTI ON DRAG


o EXPERIMENT
. WAVE DRAG (ZERO LIFT)
THEORY
DRAG DUE TO LIFT
--

M= 2.7

CD

FIGURE 16. Measured and predicted coefficients of drag vs lift for the
indicated vehicles at two supersonic Mach numbers.
WING-BODY AERODYNAM I C INTERACTION 467
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

8t deg p TH EORY 8e deg

3
o o 0
<> 5
- 5 o
-10

FIGURE 1 7. Measured and predicted coefficients of moment and drag vs lift for the
indicated vehicles at M = 2 and for different angular positions of the longitudinal
trimming surfaces.

namic efficiency, as they have already done in the case of the US supersonic
transport.
That these realizations of theory are quite successful in estimating air
loads on streamlined vehicles at small inclinations to the direction of flight
should be apparent from the foregoing selection of examples. A certain hu
mility, however, is called for. Numerous defects of current formulations
have already been pointed out, and a few will be mentioned for emphasis
here. The entire area of transonic flow cannot be described satisfactorily,
especially when shocks lie part-way back on wing surfaces and interact with
the boundary layer ; cf literature survey by Newman & Allison (1971). Flow
separations and thick wakes elude theoretical analysis, even in the relatively
simple instance of concentrated vortices emanating from highly swept leading
edges (see Chapter 4 of Nielsen 1960; Polhamus 197 1 is a recent paper on
this same topic) . The true positions and "rolling up" of wakes-thin as well
as thick-are inadequately represented. Aerodynamic forces, particularly
the drag due to lift, cannot be as effectively calculated in unsteady motion
as in steady. Finally, the aerodynamics of vertical- and short-take-off
vehicles in slow flight holds many frustrations for the industrious analysts
and programmers to whose hands the field of wing-body interaction now
seems to have been entrusted.
468 ASHLEY & RODDEN

LITERATURE CITED
Albano, E., Perkinson, F., Rodden, W. P. Thin Wings in Subsonic Flow. New
1970. Subsonic lifting-surface theory York: Plenum

/
aerodynamics and flutter analysis of in Berman, J. H., Shyprykevich, P." Smedf
terfering wing horizontal-tail configura jeld, J. B. 1970. A subsonic nonplanar

/
tions; Part I -Subsonic oscillatory aero kernel function for surfaces inclined to

/
dynamics for wing horizontal-tail con the freestream. J. A ircr. 7 : 1 88-90
figurations; Part II-Wing tail flutter Berman, J. H., Sh yprykevich, P., Smedf

/
correlation study. USAF FDL-TR-70-59, jeld, J. B., Kelly, R. F. 1968. Unsteady
Parts I and II. aerodynamic forces for general wing con
Albano, E., Rodden, W. P. 1969. A doublet trol-surface configurations in subsonic
lattice method for calculating lift dis flow. USAF FDL-TR-67-117
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

tributions on oscillating surfaces in sub Bland, S. R. 1968. Two-dimensional oscillat

esi
sonic flows. AIAA J. 7 :2 79-85, 2192 ing airfoil in a wind tunnel in subsonic
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

Andrew, L. V. 1967. Unsteady aerodynam compressible flow. PhD th s. North


ics for advanced configurations, Part VI Carolina State Univ.
-Application of the supersonic Mach Bonner, E. 1969. Expanding role of poten
box method to T-tails, and top-mounted tial theory in supersonic design. J. A ircr.
verticals. USAF FDL-TDR-64-15Z, Part 8 : 347-53
VI Borland, C. J. 1966. Methods of calculating
Andrew, L. V. 1971. Subsonic generalized aerodynamic loads on aircraft structures,
forces on aerodynamically 'interfering Part I-Wing-body interference effects.
surfaces. USAF FDL-TR-71-55 USAF FDL-TR-66-37, Part I
Andrew, L. V., Stenton, T. E. 1968. Un Bradley, R. G., Miller, B. D. 1 9 7 1 . Applica
steady aerodynamics for advanced con tion of finite-element theory to airplane
figurations, Part VII-Velocity poten configurations. J. A irer. 8 :400-5
tials in non-uniform transonic flow over Bryson, A. E. 1953. Stability derivatives for
a thin wing. USAF FDL-TDR-64-15Z, a slender missile with application to a
Part VII wing-body-vertical tail configuration. J.
A GARD Conference Proceedings No. 71. A eronaut. Sci. 20:297-308
1 9 70. Aerodynamic interference (Pre Carmichael, R. L., Woodward, F. A. 1966.
print) An integrated approach to the analysis
A GARD Conference Proceedings No. 80. and design of wings and wing-body com
1 9 7 1 . Symposium on Unsteady Aerody binations in supersonic flow. NASA TN
namics for Aeroelastic Analyses of Inter D-3685
fering Surfaces, Parts I and I I Cazemier, P. G., Bergh, H. 1964. Messungen

I I
Argyris, J . H. 1970. The impact o f the instationarer Druckverteilungen am
digital computer on engineering sciences. Fliigelhalbmodell der VJ- 0 - C . NLR
A eronaut. J. 74 : 1 3-41, 1 1 1-2 7 Rept. F. 23Z
Ashley, H. 1962. Supersonic airloads on in Chou, D. C. 1966. Methods of calculating
terfering lifting surfaces by aerodynamic aerodynamic loads on aircraft structures,
influence coefficient theory. The Boeing Part II I-Effects of engines, stores and
Co. Rept. D-Z-ZZ67 wing-tail interference. USAF FDL-TR-
Ashley, H., Landahl, M. T. 1965. Aero 66-37, Part III
dynamics of Wings and Bodies. Reading, Davies, D. E. 1964. Generalized aero

w
Mass. : Addison-Wesley dynamic forces on a T-tail oscillating
Ashley, H., Ro e, W. S. 19 70. On the un harmonically in subsonic flow. Roy.
steady aerodynamic loading of wings A ircr. Est. Rept. No. Structures 295
with control surfaces, Z. Flugwiss. 1 8 : Djoiodihardjo, R. R., Widnall, S. E. 1969.
321-30 A numerical method for the calculation

S t
Bagley, J. A. 1962. Some Aerodynamic of nonlinear, unsteady lifting potential
Principles for the DeSign of wep Wings. flow problems. AIAA J. 7 :2001-9
In Progress in A eronautical Sciences, ed.
D . Kiichemann, Vol. 3, Chap. 1 . Elms
Dulmovits, J. 1964. A lifting surface
method for calculating load distributions
ford, NY: Pergamon and the aerodynamic influence coefficient
Belotserkovskii, S. M. 1967. The Theory of matrix for wings in subsonic flow. Grum-
WING-BODY AERODYNAMIC INTERACTION 469
man A irer. Eng. Corp . Rept. A D R 01-02- Hayes, W. D. 1947. Linearized supersonic
64. 1 flow. North Am. Av. Rep. AL-222. (Re
Durand, W. F. 1935. Airplane as a Whole printed as Princeton University AMS
General View of Mutual Interactions Rept. No. 852)
Among Constituent Systems. In A ero Hayes, W. D. 1971 . Sonic boom. Ann. Rev.
dynamic Theory, ed. W. F. Durand, Sect. Fluid Mecho 3 :269-90
P, vol. VI. Berlin : Springer-Verlag (re Hayes, W. D., Probstein, R. F. 1959.
printed 1963 by Dover Publications) Hypersonic Flow Theory. New York,
Etkin, B. 1955. Numerical integration London: Academic
methods for supersonic wings in steady Hedman, S. G. 1965. Vortex lattice method
and oscillatory motion. Inst. A erophys., for calculation of quasi steady state load
Univ. of Toronto, Rept. 36 ings on thin elastic wings. A eronaut. Res.
Falkner, V. M. 1943. The calculation of Inst. Sweden Rept. 105
aerodynamic loading on surfaces of any Hess, J. L. 19 70. CalcUlation of potential
flow about arbitrary three-dimensional
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

shape. A eron aut . Res. Coun. Rep . Mem.


No. 1910 lifting bodies-Phase II Final Rept.
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

Ferrari, C. 1957. Interaction Problems. Douglas A ircr. Co. Rep. JO 971-01


In High SPeed A erodynamics and Jet Houbolt, J. C. 1969. Some new concepts in
Propulsion, ed. H. R. Lawrence, A. F. oscillatory lifting surface theory. USAF
Donovan, Section C, Vol. VII. Princeton, FDL-TR-69-2
N. J. : Princeton Univ. Press James, R. M. 1969. On the remarkable ac
Garrick, 1. E. 1936. Potential flow about curacy of the vortex lattice discretization
arbitrary biplane wing sections. NACA in thin wing theory. Douglas A ircr. Co.,
Rept. 542 Rep. DAC-67211
Jones, W. P., Rao, B. M. 1970. Compressi
Garrick, 1. E. 1944. On the plane potential
bility effects on oscillating rotor blades in
flow past a lattice of arbi trary airfoils.
hovering flight. A IA A J. 8:321-29
NACA Rept. 788
Kacprzynski, J. J., Landahl, M. T. 1967.
Giesing, J. P. 1968a. Nonlinear interaction
Recent developments in the supersonic
of two lifting bodies in arbitrary unsteady
flow over axisymmetric bodies with con
motion. J. Basic Eng. 90:387-94
tinuous or discontinuous slope. A IAA
Giesing, J. P. 1968b. Lifting sutface theory
Paper No. 67-5
for wing-fu selage combinations. Douglas
Kalman, T. P., Rodden, W. P., Giesing,
A ircr. Co. Rep. DAC-67212
J. P. 1971. Application of the doublet
Giesing, J. P., Kalman, T. P., Rodden, lattice method to nonplanar configura
W. P. 1971. Subsonic unsteady aero tions in subsonic flow. J. A ircr. 8:406-13
dynamics for general configurations ;
Kariappa, and Smith, G. C. C. 1971. Further
Part I-Direct application of the non
developments in consistent unsteady
planar doublet-lattice method ; Part II
supersonic aerodynamic coefficients.
Application of the doublet-lattice method AIAA Paper No. 71-177
and the method of images to lifting Korner, H. 1969. Untersuchungen zur
surface/body interference. USA F FDL
Bestimmung der Druckverteilung an
TR-71-5, Parts I and II
Fliigel-Rumpf-Kombinationen, Teil I :
Gray, W. L., Schenk, K. M . 1953. A method Messergebnisse fiir Mitteldeckeranor
for calculating the subsonic steady-state dnung aus dem 1 . 3m-Windkanal.
loading on an airplane with a wing of DFVLR Rept. No. 0562
arbitrary plan form and stiffness. NA CA Kiichemann, D., Weber, J. 1953. The sub
TN 3030 sonic flow past swept wi ngs at zero lift
Grossman, F., Marconi, F., Moretti, G. without and with body. A eronaut. Res.
1971. A numerical procedure to calculate Counc. Rep. Mem. No. 2908
the inviscid flow field about a space Kiissner, H. G. 1 940. Allgemeine Trag
shuttle orbiter traveling at a supersonic/ fl1(chentheorie. Luftfahrtforschung 7 :3 70-
hypersonic velocity. Paper 6, NASA 78
Space Shuttle Techno!. ConL NA SA TM LaBarge, W. L. 1971 Correlation o f the
X-2272 oretical and experimental pressure dis
Harder, R. L., Rodden, W. P. 1971. Kernel tributions over an oscillating wing with
function for nonplanar oscillating sur two control surfaces. Lockheed Calif. Co.
faces in supersoni c flow. J. A ircr. 8:677- Rept. LR-24737
79 Landahl, M. T. 1964. Linearized theory for
470 ASHLEY & RODDEN
unsteady transonic flow. Symposium Monical, R. E. 1965. A method for repre
Transsonicum, I UTAM Symposium senting fan-wing combinations for three
A achen 1962, ed. K. Oswatitsch. Berlin: dimensional potential flow solutions. J.
Springer-Verlag A ircr. 2 :52 7-30
Landahl, M. T. 1967. Kernel function for Moore, M . , Andrew, L. V. 1965. Unsteady
nonplanar oscillating surfaces in a sub aerodynamics for advanced configura
sonic flow. AIAA J. 5 :1 045-46 tions, Part IV-Application of the super
Landahl, M. T. 1968. Pressure-loading sonic Mach box method to intersecting
functions for oscillating wings with con planar surfaces. USAF FDL-TDR-64-
trol surfaces. A IAA J. 6 :345-48 152, Part IV
Landahl, M. T., Stark, V. J. E. 1968. Moretti, G., Abbett, M. 1966. A time-de
Numerical lifting-surface theory-prob pendent computational method for blunt
lems and progress. A IAA J. 6 :2049-60 body flows. A IAA J. 4 :2 1 36-41
LaRowe, E., Love, J. E. 1967. Analysis and Morikawa, G. K. 1952. A non-planar
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

design of supersonic wing-body combina boundary problem for the wave equation.
tions, including flow properties in the Quart. A ppl. Math. 1 0 : 129-40
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

near field, Part I I-Digital computer Multhopp, H. 1941. Zur Aerodynamik des
program description. NASA CR-73107 Flugzeugrumpfes. Luftfahrtforschung 1 8 :
Laschka, B. 1963a. Das Potential und das 52-66
Geschwindigkeitsfeld der harmonisch Multhopp, H. 1950. Methods for calculat
schwingenden tragenden Flache bei Un ing the lift distribution of wings (sub
terschallanstromung, ZAMM 43:325-33 sonic lifting surface theory). A eronaut.
(see erratum in ZAMM 43:284) Res. Counc. Rep. Mem. No. 2884
Laschka, B. 1963b. Zur Theorie der har Mykytow, W. J., Noll, T. E., Huttsell, L.
monisch schwingenden tragenden Flache J., Shirk, M. H. 1 9 70. Subsonic flutter
bei Unterschallanstromung. Z. Flugwiss. characteristics of a variable sweep wing
1 1 :265-91
Laschka, B., Schmid, H. 1967. Unsteady FDL-TR-69-59
and horizontal tail combination. USAF

aerodynamic forces on coplanar lifting NASA SP-228. 1969. Analytical methods in


surfaces in subsonic flow (wing-horizontal aircraft aerodynamics
tail interference) . Jahrb. 1967 WGLR, Navarro Crespo, A., Cunningham, H. J.
2 1 1-22 1969. Development of three-dimensional
Lawrence, H. R., Flax, A. H. 1954. Wing pressure distribution functions for lifting
body interference at subsonic and sUper surfaces with trailing-edge controls based
sonic speeds-Survey and new develop on the integral equation for subsonic
ments. J. A ero. Sci. 21 :289-324, 328 flow. NA SA TN D-5419
Lighthill, M. J. 1948. SUpersonic flow past Newman, P. A., Allison, D. O. 1 9 7 1 . An
slender bodies of revolution the slope of annotated bibliography on transonic flow
whose meridian section is discontinuous. theory. NA SA TM X-Z363
Quart. J. Mech. Appl. Math. 1 :90-102 Nielsen, J. N. 1952. Quasi-cylindrical theory
Lock, R. C., Bridgewater, J. 1967. Theory of wing-body interference at supersonic
of Aerodynamic Design for Swept speeds and comparison with experiment.
Winged Aircraft at Transonic and NA CA Rep. 1252
Supersonic Speeds. In Progress in A ero Nielsen, J. N. 1960. Missile A erodynamics.
nautical Sciences, ed. D. Kiichemann, New York: McGraw-Hill
Vol. 8, Chap. 2. Elmsford, NY: Perga Petkas, J. S. 1969. Oscillatory aerodynamic
mon representation using discrete load line ele
Martina, A. P. 1956. The interference effect ment (DLLE) method. Lockheed-Georgia
of a body on the spanwise load distribu Co. Unpublished Memo
tion of two 45 sweptback wings of aspect Pines, S., Dugundji, J., Neuringer, J. 1955.
ratio 8.02 from low-speed tests. NACA Aerodynamic flutter derivatives for a
TN 3730 flexible wing with supersonic and sub
M iddleton, H. W., Carlson,H. W. 1965. A sonic edges. J. A eronaut. Sci. 2 2 : 693-700
numerical method for calculating the flat Polhamus, E. C. 1 9 7 1 . Predictions of vor
plate pressure distributions on supersonic tex-lift characteristics by a leading-edge
wings of arbitrary planform. NASA TN suction analogy. J. A ircraft 8 : 193-99
D-2570 Powers, S. A., Beeman, E. R., Jr. 1 9 7 1 .
Miles, J. W. 1959. Unsteady Supersonic Flow fields over sharp edged delta wings
Flow. Cambridge : Cambridge Univ. Press with attached shocks. NA SA CR-1738
WING-BODY AERODYNAM IC INTERACTION 471

Rodden, W . P., Liu, D . T. 1969. Correla of thick wing and pylon-fanpod-nace1le

rotor/wing configurations. J. A ircr.


tion of the vortex lattice method on aerodynamic characteristics at subcritical
speeds. North Am. Rockwell Corp. Rep.
6 :375 NA-71-447
Rubbert, P. E. 1964. Theoretical charac Vivian, H. T., Andrew, L. V. 1965. Un
teristics of arbitrary wings by a non steady aerodynamics for advanced con
planar vortex lattice method. Boeing Co. figurations, Part I-Application of the
Rep. D-6-9244 subsonic kernel to nonplanar lifting sur
Rubbert, P. E., Saaris, G. R. 1968. A gen faces. USA F FDL-TDR-64-152, Part I

method applied to V/STOL aerody


eral three-dimensional potential flow Ward, G. N. 1949. Supersonic flow past
slender pointed bodies. Quart. J. Mech.
namics. SAE Paper No. 680304 A ppl. Math. 2 : 75-97
Runyan, H. L., Woolston, D. S. 1957. Watkins, C. E., Berman, J. H. 1956. On the
Method for calculating the aerodynamic kernel function of the integral equation
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org

loading on an oscillating finite wing in relating lift and downwash distributions


subsonic and sonic flow. NASA Rept. of oscillating wings in supersonic flow.
1322 NACA Rep. 1257
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

Sauerwein, H. 1964. The calculation of two Watkins, C. E., Runyan, H. L., Woolston,
and three-dimensional inviscid unsteady D. S. 1955. On the kernel function of the
flows by the method of characteristics. integral equation relating the lift and
A FOSR 64-1055, M I T Fluid Dynam. downwash distributions of oscillating
Res. Lab. finite wings in subsonic flow. NACA
Schade, T., Krienes, K. 1940. Theorie des Rep. 1234
kreisformigen schwingenden Tragflache Watkins, C. K, Woolston, D. S., Cunning
auf potential-theoretischer Grundlage. ham, H. J. 19 5 9. A systematic kernel
Luftfahrtforschung 1 7 :387-400 function procedure for determining aero
Sensburg, 0., Laschka, B. 1970. Flutter in dynamic forces on oscillating or steady
duced by aerodynamic interference be finite wings at subsonic speeds. NASA
tween wing and tail. J. A ircr. 7 : 3 19-24 Tech. Rep. R-48
Smith, A. M. 0., Hess, J. L. 1967. Calcula Weber, ]., Kirby, D. A., Kettle, D. A. 1956.
tion of Potential Flow about Arbitrary An extension of Multhopp's method of
Bodies. In Progress in A eronautical calculating the spanwisc loading of wing
Sciences, ed. D. Kiichemann, Vol. 8,
Chap. 1. Elmsford , NY: Pergamon
fuselage combinations. A eronaut. Res.
Counc. Rep. Mem. No. 2872
S6hngen, H., Meister, E. 1958. Beitrag zur
White, R. B., Landahl, M. T. 1968. Effect
Aerodynamik eines schwingenden Git of gaps on the loading distribution of
ters. ZAMM 38 :443-65
planar lifting surfaces. A IAA J. 6 :626-31
Woods, L. C. 1961. The Theory of Subsonic
Spreiter, ]. R. 1948. Aerodynamic proper
ties of slender wing-body combinations at
Plane Flow. Cambridge : Cambridge Univ.
subsonic, transonic and supersonic
Press
speeds. NACA TN 1662
Stark, V. J. E. 1958. A method for solving Woodward, F. A. 1968. Analysis and design
the subsonic problem of the oscillating
of wing-body combinations at subsonic
finite wing with the aid of high-speed and supersonic speeds. J. A ircr. 5 :528-34
digital computers. Saab A ircr. Co. TN 41 Woodward, F. A., Hague, D. S. 1969. A
Stark, V. J. E. 1964. Aerodynamic forces on computer program for the aerodynamic
a combination of a wing and a fin oscillat analysis and design of wing-body-tail
ing in subsonic flow. Saab A ircr. Co. TN combinations at subsonic and supersonic
54 speeds, Vol. I : Theory and program
Thwaites, B., Ed. 1960. Incompressible utilization. Gen. Dynam. Ft. Worth Div.
A erodynamics. Oxford : Clarendon Press Res. Rep. ERR-FW-867
Topp, L. J., Rowe, W. S., Shattuck, A. W. Woodward, F. A., Tinoco, E. N., Larsen,
1966. Aeroelastic considerations in the de J. W. 1967. Analysis and design of super
sonic wing-body combinations, including
flow properties in the near fiel d, Part 1-
sign of variable sweep airplanes. Proc.
Fifth Int. Congr. A eronaut. Sci., London
Truckenbrodt, K 1953. Tragfiachentheorie Theory and application. NASA CR-
bei inkompressibler Str6mung. Jahrb. 73106 (Part I)
1953 WGL, 40-65 Wu, T. Y., Talmadge, R. B. 1961. A lifting
Tulinius, J. R. 1971. Theoretical prediction surface theory for wings extending
472 ASHLEY & RODDEN

through multiple jets. Vehicle Res. Corp. Zlotnick, M., Robinson, S. W. 1954. A
Rep. No. 8 simplified mathematical model for cal
Yates, E. C., Jr. 1966. A kernel-function culating aerodynamic loading and down
formulation for nonplanar lifting surfaces wash for wing-fuselage combinations
oscillating in subsonic flow. AIAA J. 4 : with wings of arbitrary plan form. NACA
1486-88 TN 3057
Annu. Rev. Fluid Mech. 1972.4:431-472. Downloaded from www.annualreviews.org
by Technische Universiteit Eindhoven on 06/12/14. For personal use only.

Você também pode gostar