r https://ntrs.nasa.gov/search.jsp?

R=19680019396 2020-02-12T16:12:29+00:00Z

NASA TECHNICAL NOTE 0-4427

LOAN COPY: RETURN TO

AFWL [WLIL-2)
KIRTLAND AFB, N MEX

A MODIFIED MULTHOPP APPROACH

FOR PREDICTING LIFTING PRESSURES
A N D CAMBER SHAPE FOR COMPOSITE
PLANFORMS I N SUBSONIC FLOW

by John E. Lamur
Langley Reseurch Center
Langley Station, Humpton, vu.

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON, D. C. JULY 1968

 TECH LIBRARY KAFB, NM

IIIlllll1llI1H
1I 11I lI llIll

A MODIFIED MULTHOPP APPROACH FOR PREDICTING

LIFTING PRESSURES AND CAMBER SHAPE FOR

COMPOSITE PLANFORMS I N SUBSONIC FLOW

By John E. Lamar

Langley Research Center

Langley Station, Hampton, Va.

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

For s a l s by the Clearinghouse for Federol Scientific and T e c h n i c a l Information

Springfield, V i r g i n i a 22151 - C F S T I price $3.00
 A MODIFIED MULTHOPP APPROACH FOR PREDICTING
LIFTING PRESSURES AND CAMBER SHAPE FOR
COMPOSITE PLANFORMS IN SUBSONIC FLOW

By John E. Lamar
Langley Research Center

SUMMARY

This report presents a modified version of Multhopp' s subsonic lifting-surface

theory which has been programed in two parts for the IBM 7094 electronic data processing
system o r Control Data 6400 computer system along with a discussion of the character of
its results. The first part is used to find both basic and additional loadings over a given
planform with known mean camber surface and the second part is used to determine the
required mean camber surface for a given planform and s e t of loadings.
F o r the loading program, various aerodynamic characteristics a r e determined on
both simple and composite planforms when the spanwise loading is symmetrical. Studies
a r e conducted to determine when these answers a r e most valid, and some results for
delta, sweptback and tapered, double delta, and variable-sweep wings a r e compared with
other theories and experiments to determine the accuracy of this method.
Application of this method is then made in predicting the aerodynamic effects of
changing the outer-panel sweep of a variable-sweep wing, of increasing the Mach number
in the subsonic regime, and of incorporating twist and camber in a wing. The first two
applications also have experimental data presented for comparison.
A study is also undertaken for the mean camber surface program to determine when
its best results a r e obtained. Subsequently, comparisons between the present and other
methods are made for a two- and three-dimensional case to aid i n the evaluation of the
present method. One application is made for a highly sweptback and tapered planform.

INTRODUCTION

In the process of wing design, one is required to determine either the load distribu-
tion from a given planform and mean camber surface o r the mean camber surface from a
prescribed planform and loading. From previous investigations a considerable amount of
information of both a theoretical and experimental nature is available to aid the designer
if the wings have delta or simple sweptback planforms. However, planforms now being
considered (which in many cases have the divergent requirements of supersonic o r hyper-
sonic cruise and subsonic loiter o r f e r r y capability) are much more complex, involving,
in many instances, cranked leading and trailing edges. Planforms of this type are classi-
fied as being of a composite arrangement.
F o r these planforms not as much general information is available, due both to their
newness and to the large number of planform variables. First-order solutions to these
problems have been found in recent years for the sonic (refs. 1 and 2) and supersonic
(refs. 3 and 4) speed regimes which a r e applicable to arbitrary wing planforms. Refer-
ences 3 and 4 have been programed for the high-speed electronic computer s o that the
designer does not have to perform the lengthy mathematical manipulations required to
obtain a solution for each wing considered.
In the subsonic speed regime, the basic theoretical method of Multhopp (ref. 5) has
received wide acceptance as one of the most accurate methods for predicting aerodynamic
loading data; consequently, it has been selected as the approach to use in solving these
two design problems for composite planforms. Some reasons for this selection a r e
(1) The chordwise pressure distributions (and other section data) as well as overall
aerodynamic characteristics a r e determined
(2) The spanwise locations of the lift-producing singularities and control points a r e
more concentrated along chordwise rows near the tip to insure an adequate representation
of the spanwise load distribution
(3) The method can as Multhopp mentioned in reference 5 be used for wings with
planform kinks at locations other than just at the plane of symmetry
A computerized solution of the loading determination problem for arbitrary wings
has been developed from the basic Multhopp method by Van Spiegel and Wouters in refer-
ence 6 and is referred to as the modified Multhopp method. Certain changes in and exten-
sions to their work have been made in the present report to increase the accuracy in
representing the actual chordwise pressure distribution and finding the influence of each
of these distributions on each point where the boundary conditions are to be met. These
modifications resulted from adding additional pressure modes (up to 10 may be used) and
by replacing the polynomial approximation technique used in the chordwise integration
with the Gaussian quadrature method.
In addition, the formulation procedure and all the modifications made to the basic
Multhopp approach a r e described as well as the techniques used in its solution for both
the load and mean camber surface determination problems. After this a study of the
resulting answers is made to determine their sensitivity to the number and location of
the points where the boundary conditions are met o r calculated. Then applications of the
programs are made to an assortment of wings, including composite wings and the results

2
are compared with other theories and, where possible, experiment. All the predictions
made from this modified Multhopp method have been obtained with the aid of the IBM 7094
electronic data processing system and Control Data 6000 series electronic computers.
A listing of the computer programs used to find the surface loadings (Langley pro-
gram A0313) and to find the mean camber surface (Langley program A0457) is pre-
sented along with a description of the input data required, sample input and output data.
and pertinent comments i n a "Supplement to NASA TN D-4427." A large portion of these
two programs is given over to the computation of the geometric representation required
by later parts of the program from the input quantities supplied. Two additional programs
(Langley programs A1590 and A1591) which are useful in obtaining geometric input data
for either program A0313 o r A0457 a r e also presented in the supplement along with brief
descriptions and sample cases of each. A request form is included at the back of this
paper.

SYMBO LS

Any convenient system of measure can be used as long as the linear dimensions of
the planform a r e based on a semispan of unity.

A aspect ratio, b2/S

a.c. aerodynamic center, in fractions of Cref, referenced to leading edge of ref-

-'Cm + -1
erence chord (positive aft), -
8CL 4

b wing span, set equal to 2

bvwbvv Multhopp's integrating coefficients

CA axial-force coefficient, Axial force

qwsref

cB,o root bending-moment coefficient, "1'

Sref 0
(CzC)basicqdq

cD,i induced drag based on the spanwise distribution of circulation

cD,ii induced drag based on axial force, CYCL - CA

Wing lift
CL lift coefficient,
qwsref

3
 overall lift-curve slope

Wing pitching moment %ef

Cm pitching- moment coefficient , about -
q 2,ef C r ef 4

overall pitching-moment- curve slope about C r ef

-
4

incremental pressure coefficient, Pupper - Plower - -AP

AcP qco q,

CS leading-edge-suction coefficient for zero degrees of leading-edge sweep,

drl

C streamwise chord at y , ~

c2dy
- mean geometric chord,
-b/2
C ~

IbI2
-b/2
c dy

streamwise half-chord at n,y

average chord, S/b

Local chord load

local lift coefficient,
4coc

local pitching- moment- curve slope about local leading edge

local slope of normal-force coefficient

root chord

reference chord, may be mean geometric chord of total wing except c

when the planform has an inboard trailing-edge chord-extension o r may be
mean geometric chord of only wing outer panel extended to plane of symme-
t r y in streamwise tip position

x-location of midchord at q
fj
jth chordal loading function

K(@,y;a,q) subsonic kernel function

Kj (@,y;q) subsonic kernel function integrated with jth loading across the chord

influence of jth loading function at qn on control point at yv and

x = (c/2)(1 - cos qs)

tin(@ s elements of influence matrix when n # v

M Mach number

m number of span stations where the'pressure modes are defined

N number of chordal control points at each of m span stations

AP lifting pres sure, Pupper - Plower

qj (77) coefficient of jth chordal loading function for additional load

qtj (7) coefficient of jth chordal loading function for twist and/or camber load

q, dynamic pressure

S total wing a r e a

Sref reference area, may be either total wing a r e a o r wing a r e a of wing outer
panels extended to plane of symmetry in streamwise tip position

S used to locate chordal control points and ranges from 1 to N

U free-stream velocity

W perturbation velocity in z-direction

X,Y ,= rectangular Cartesian coordinates nondimensionalized with respect to b/2

where origin is in plane of symmetry at half root chord (fig. 1) (they are
associated with control points)

5
 x-location of leading edge of C

Xcref x-location of leading edge of Cref

ycp, xcP centers of pressure

a! angle of attack, deg

ai induced angle of attack, radians

local angle of attack, -

-dz = -
-w radians
a2 dx U'

P Prandtl-Glauert compressibility factor, d l - M2

Y local nondimensional lift coefficient, cz c/2b

A wing tip skew angle, deg

e angle for locating control points along span (see fig. 2(a))

9 angle used to locate pressure doublets chordwise, 0 at leading edge and a

at trailing edge

K ratio of distance of leading-edge break from plane of symmetry to b/2

A outboard leading-edge sweep angle, deg

Tip chord --
x taper ratio,
Overall root chord

ratio of distance of trailing-edge break from plane of symmetry to b/2

rectangular Cartesian coordinates nondimensionalized with respect to b/2

along X- and Y-axes, respectively, and associated with pressure doublet
locations

angle used to locate chordwise rows of control points (see fig. 2(b))

inboard leading-edge sweep angle, deg

@ inboard trailing-edge sweep angle, deg

a outboard trailing-edge sweep angle, deg

Subscripts :

CP center of pressure

d distributed

le leading edge

n,v suffixes numerating spanwise stations, v station being influenced and n

m - 1 , . . .o, . . . m - 1
station doing the influencing; n,v range from - -
2 2

0 value taken at CL = 0

P pivot

te trailing edge

X x -dir e ction

Y y-direction

1 result due to twist and/or cambered wing at zero root chord angle of attack

BASIC FORMULATION

The determination of surface loadings on both plane and warped wings and the deter-
mination of the mean camber surface can be made from a number of theories already in
use. All the theories share a common feature in that the wing is replaced by a mathemat-
ical model which produces a similar downwash field to that of the wing. However, the
methods differ mainly in the potential used to produce the representative perturbations of
the flow field around the wing. These potentials are either of the velocity or acceleration
type (ref. 7). The original Multhopp method and its present modified version use the
acceleration potential approach in conjunction with the linearized Euler equations to relate
the pressure difference across the wing to the downwash field on the wing surface. This
results in the following equation which is derived in detail in reference 6 for the incom-
pressible case:
7
 Reduction of Basic Equation t o Matrix Form
Equation (1) is difficult to solve for the local pressures A p ( 5 , ~ ) since they appear
as part of the integrand and because of the integration across the second-order singular-
ity. The t e r m w/U and its resulting mean camber surface is also difficult to deter-
mine because of this s a m e required integration. A solution to either problem can, how-
ever, be made by
(1) Replacing Ap(<,v) with as many of the 10 pressure modes of unknown or known
amplitude as a r e needed

where
5= -C(~)COS 8 + d(v) (3)

(2) Employing the Multhopp quadrature formula to approximate the spanwise inte-
gration of the pressure doublets; this approximating procedure has another feature in that
every spanwise station need not be represented but only those stations where information
is required
(3) Using a logarithmic singularity correction t e r m (appendix A) to account for the
integration across the second-order singularity (this amounts to finding the finite part)
(4) Setting up equation (1) with these approximations in a matrix equation of the form

The details of formulating the matrix of influence coefficients a r e given in appen-

dix A where they have been taken for the most part directly from reference 6 and pre-
sented here for the reader’s convenience.
The points (determined from GS and yv) which are selected either to specify the
tangential flow boundary condition or to find the downwash ratio a r e called control points
and a r e located in both the spanwise and chordwise directions by the use of semicircles
which a r e laid along the span from tip to tip and along the local chord from leading to

8
trailing edge. The spanwise positions, which result from an equiarc angular displace-
ment about the plane of symmetry, can be represented by the following equation:

yv = sin -
V77
(5)
m + l

where m is the total number of spanwise stations from tip to tip and v ranges from
-11 2
1-to ~

2
- in increments of 1. (See fig. 2(a).) These positions are those
required by the Multhopp quadrature formula for the integrand evaluation when approxi-
mating the value of the spanwise integral (ref. 5).
At each spanwise location, the control points are also located for equiarc angular
displacements defined by

where N is the total number of chordwise control points and ranges from 1 to N. (See
fig. 2(b).) Note that the control points never reach the leading o r trailing edge. From
this, the control-point locations a r e found to be determinable by the following equation:

which locates them aft of the local leading edge. (See fig. 3.)
In order to solve f o r either the loading (inverse problem) o r mean camber surface
(direct problem), equation (4)is used. The loadings a r e found by (1) inverting the matrix
- a r r a y and then (2) using these values in
in equation (4)and thus solving for the qj(rln)
4,
the equations given in appendix B to determine the local and overall aerodynamic loading
description. The mean camber surface is found by the direct matrix multiplication of
equation (4)along with the procedure described in the mean camber surface determination
section of this report.

Modification Made to the Original Multhopp Method

From a study of Multhopp's original method, several features a r e seen to need
modification in order that a computerized solution of it could be made to the best advan-
tage. Aside from the logarithmic singularity correction t e r m whose best form is dis-
cussed in appendix A they are as follows:
1. The method f o r solving for the coefficients of the loading t e r m s in the inverse
problem. In Multhopp's report, a scheme of iteration on the downwash is used, this is
improved in reference 8 with a scheme of submatrix inversion and in reference 6 with a
scheme of whole matrix inversion which is also used herein.

9
 2. The chordal integration that sums the influence of the pressure doublets on con-
t r o l points not at the same span station. In Multhopp’s report, it is developed from a
MacLaurin s e r i e s then graphed for use. Van Spiegel and Wouters (ref. 6) follow the
scheme developed by Van de Vooren (ref. 9) with which the value of the integral is deter-
mined, by (a) approximating the loading functions with polynomials, (b) evaluating them at
discrete locations, and (c) multiplying the results of part (b) with predetermined coeffi-
cients. In this report, the polynomial approximation of reference 6 is replaced with a
Gaussian 20- t o 30-point numerical integration scheme.
3. The number of chordal loading functions and, therefore, control points. In
9
Multhopp’s report they are limited to two, cot 5 which gives lift with no pitching
moment about c/4 and cot 9 - 2 sin 9 which yields no lift but a pitching moment about
2
c/4. From these, the known downwash and the matrix of influence coefficients, one is
able to solve for the coefficients of the loading t e r m s which are the local circulation and
pitching moment about local c/4. Garner and Lehrian (ref. 10) have extended to four the
number of chordal loadings (and consequently, control points) that may be used but employ
combinations of t e r m s rather than separate loading modes. Van Spiegel and Wouters also
extend the number of chordal loadings and control points t o four but separate the loadings
into cot -,
9 sin 9, sin 29, and s i n 39, whose coefficients are the strength of the local
2
pressure doublets supporting that loading contribution over the chord. In this report
they have been extended to 10 with the addition of sin 49, sin 59, sin 69, sin 79,
sin 89, and sin 99 for additional flexibility.
Multhopp’ s method with the aforementioned modifications is hereafter referred to
as the present method and has been programed to handle a wide variety of composite
planforms, with a break in the leading and trailing edges as well as the variable-sweep
wings. The program at present is capable of handling s e t s of chordwise and total span-
wise stations of 1 by 41, 2 by 41, 3 by 41, 4 by 41, 5 by 39, 6 by 31, 7 by 27, 8 by 23, 9 by
21, o r 10 by 19, with the number of chord stations given first. This is not a theoretical
limitation but is near the upper limit of machine storage presently available. Also, with
the s e t s of stations listed above, “almost
.
1 .
any planform of low to moderate aspect ratio can
be handled.
The distribution of lift obtained by the present method is dependent on the subsonic
Mach number (see appendix A) at which it is computed. In this report, the Prandtl-
Glauert relations (ref. 11) have been used to account for this change but a r e valid only as
long as ACp varies linearly with a.

10
 LDAD DETERMINATION

Because of the nature of the numerical solution, different combinations of N and

m for the same planform can result in different answers for the aerodynamic character-
istics. Hence, a study was undertaken to t r y and optimize their combination so that rea-
sonably accurate results are obtained. The results of this study are presented in the next
section. Then comparisons of other theoretical methods and experiment with the present
method a r e made to gain confidence in the answers obtained with it. And lastly, illustra-
tive examples a r e given to show some applications for which the present method can be
used.

Optimization of N and m
For any given planform and pair of N and m, a set of chordal loading coefficients
J n) a r e obtained from the solution of the inverse problem for the flat plate and a r e
q.(q
used to determine the loading characteristics of the planform. However, since a different
set of q.(q ), and therefore loadings, will be found for each pair of N and m, some
J n
criterion is needed by which the most appropriate s e t of qj(qn) can be selected. One
criterion, suggested by Multhopp (ref. 5), is to examine the chordwise load distribution by
paying particular attention to the leading-edge suction force developed. The contention
being that when the overall leading-edge suction force, which is assumed equal to the
axial force and given,in coefficient form for an unswept planform by

is such as to cause the quantity

to be equal to the induced drag coefficient given by the equation

then the cotangent loading (eq. (2)) can be considered valid. This contention is based on
the conclusions of a study made by Munk in reference 1 2 where he found that the term
cD,i w a s independent of the chordwise loading distribution. In the discussion that fol-
lows, the result of equation (9) is called the suction value and the result of equation (10)
is called the circulation value.

11
 Upon comparing graphically the two coefficients just discussed for many combina-
tions of N and m f o r two rectangular planforms of different aspect ratio, as is done
in figures 4(a) and 5(a), it is noted that whereas the circulation value is essentially invari-
ant with N and m, and is plotted as a straight line without symbols, the suction value
shows a marked dependence upon N and m. If values f o r the aerodynamic characteris-
tics at each pair of N and m where the two solutions are first equal or have a mini-
mum difference are plotted as solid symbols, it is seen that the aerodynamic center and
CL, values for the A = 2 planform (figs. 4(b) and 4(c)) do not vary appreciably from
results obtained when N = 8 at the higher values of m. However, in the aerodynamic
center and CL, graphs (figs. 5(b) and 5(c)) for the A = 7 planform, it appears that the
values predicted do not necessarily agree with those at the higher N and m. One rea-
son suggested for the disagreement is that not enough values of m were available to be
used at the higher N values (storage limitation) s o that the resulting answers found by
convergence could have been established. Note that the aerodynamic center is the more
sensitive of the two; hence, a major part of the discussion that follows is concerned with
it.
From a study of figures 5(b) and 5(c), the curves for N = 2 and 4 seem to show
that a particular relationship must exist between N and m in order for the answers to
reach a level of convergence. Whatever the relationship is, it leads to results which a r e
close t o those given by the solid symbols. From the N = 4 curve in figure 5(b), it
seems that a good result from the point of view of convergence would occur at m = 37, a
solid symbol. Now if this relationship between m and N is valid, then its multiplica-
tion with lower N values to determine new m numbers should lead to results which
a r e as good as those found from the original N and m values.
An application of this procedure would be that of multiplying the ratio of (37/4) by
an N value of 2, thereby yielding an m number of approximately 19. When the curves
of aerodynamic center and CL, at N = 2 and m = 19 a r e examined in figures 5(b)
and 5(c), this combination of N and m is seen to give results which approximate well
those obtained for N = 4 and m = 37. This occurs even though N = 2 and m = 19 is
not a solid symbol which indicates that even though the ratio was obtained from a combi-
nation of N and m that was a solid symbol, applications of it at lower values of N
do not necessarily result in combinations of N and m which a r e other solid symbols.
It should be noted that had the N = 2 and m = 13 solid symbol been selected as the
numbers to be used in determining the ratio of m to N, the m value which would have
resulted from the multiplication of the ratio with N = 4 would be 26. This combination
of N and m does not give answers which approximate as well the results found from
convergence.

12
 Hence, s e t s of N and m can be calculated that will result in answers for aero-
dynamic center and CL, which approximate well those found at a solid symbol provided
they have been obtained from a solid symbol which has a larger N value than for the
combination sought. In addition, from the results of both figures 4 and 5, the relationship
between N and m is seen to indicate that an aspect ratio dependency also exists,
requiring the use of more m for a given N at the higher aspect ratios. This is dis-
cussed in more detail later.
Examining again the variation of the aerodynamic center with N and m for the
A = 2 planform in figure 4(b), it is noted that a part of the aerodynamic-center movement
can be correlated with the suction force coefficient which is seen implicitly in figure 4(a).
(
In general, a s the suction increases a decrease in CD,ii/CL2) a corresponding forward
movement in the aerodynamic center is observed. This simple relationship implies that
if any two s e t s of N and m give r i s e to the same suction coefficient, then their aerody-
namic centers will also be the same. However, this is not generally true. Look, for
example, at a value of CD,ii/CL2 of 0.20 and note all of the different combinations of
N and m that will give it and then the corresponding aerodynamic centers. Some dif-
ferences a r e seen to exist between the aerodynamic centers predicted by the different
combinations of N and m.
There a r e several possible reasons f o r this difference. The first one is that the
relationship assumed between the leading-edge suction and the aerodynamic center is not
a completely valid one since it depends not only on the qo(qn)cot ‘z t e r m s (which give
r i s e to the leading-edge suction) but also the q2(qn)sin 2s terms.
The second reason is that since in the solution for the qj(qn) set the number of
control points is required to be equal to the number of unknown coefficients, the flat plate
representation of the planform may not be made at enough control points. This could be
remedied by an overdetermined solution (Le., one where there a r e more control points
and number of equations than unknown coefficients).
The third reason follows from the second in that, with the boundary condition only
specified at a maximum of 10 points chordwise, the variation of w/U that results by
matrix multiplication using the qj(qn) set determined from the specified boundary con-
ditions does not assume the boundary condition value except at the control points. Else-
where it is either higher o r lower and because of the logarithmic singularity correction
t e r m (eq. (A9)) it is singular at the leading and trailing edges. This variation in w/U
when acted upon by the chord loading distribution should also give rise to an axial force
heretofore unaccounted for. It is called the distributed axial force and is computed by

13
 llllll I l l 1

Numerical results have not been obtained f o r CA,d since every point along the chord at
each spanwise position must contain the influence of every loading function at every other
spanwise location in order that the quantity w/U can be determined. This implies
evaluating integrals similar to those used in finding qj(qn) but at an extremely large
number of locations. In addition, the singularities at the edges of the chord would be dif-
ficult to evaluate without a coordinate transformation. (Note that if the mathematical pro-
file were truly flat, then all the axial force generated would be concentrated at the leading
edge in the form of a suction force, since w/U would be everywhere equal to a.) Thus,
only one solution f o r the axial-force coefficient has been obtained and is derived solely
from the leading- edge suction.
From figures 4 and 5, it appears (as one would probably expect) that for the most
part at the higher N and m values the answers for CL, and aerodynamic center a r e
in what appear to be converging situations. However, it has also been found that one need
not always choose the maximum N and m to achieve these answers but may by proper
insight and judicious selection obtain equally as valid answers at lower values of N and
m. Thus f o r planforms which have zero sweep on the leading edge, it is suggested that
pairs of N and m be selected for the particular planform at the desired Mach number
in a manner so as to emphasize its largest dimension; that is, planforms having large
values of PA would require a pair of N and m where m was optimized after a
suitable value for N was established, and planforms having low values of PA would
require that a large value for N be used with a corresponding reduction in m.
Selecting pairs of N and m by this procedure should give answers that will be in o r
be very close to those in the region of converging results. This can be accomplished
approximately by using the following expression:

m = (4 to 5)PAF)

A similar type of analysis which leads to equation (12) should be made for wings with
sweepback using the leading-edge-suction equation recently developed by Garner in refer-
ence 13.
In an attempt to s e e if the suggestions given would be applicable to arbitrary plan-
forms, aerodynamic-center and CL, graphs were prepared for two double delta plan-
forms which are shown in figures 6 and 7. It appears that even with taking N = 8 the
formula does not give quite enough m (only 20) to predict the aerodynamic center at the
higher m values (figs. 6(a) and 7(a)). Consequently, f o r low-aspect-ratio wings with
sweepback, an N value of from 6 to 10 should be used with as many m values as the
machine storage limitation will allow.
In addition to the overall aerodynamic characteristics which have already been dis-
cussed, the influence of different combinations of N and m has been examined with

14
regard to their influence on the span load distribution. (See figs. 8 to 11.) From this
examination, the reason becomes apparent why the induced drag parameter, cD,i/CLZ,
was stated to be essentially invariant with different combinations of N and m. The
reason is that the span loadings themselves appear for the most part to be independent of
N and m.

Comparisons and Illustrative Calculations

The following section contains a number of comparisons with other theoretical
methods and experimental data which have been used to verify the results obtained with
the present method. After the verifications have been made, illustrative calculations a r e
made to show the present methods' usefulness in computing sweep, Mach number, twist,
and camber effects. Additional illustrative calculations of the present method have been
made and a r e given in reference 14.
Comparisons with other theoretical methods.- In order to insure that the computa-
tional procedures presently employed give equivalent results to those obtained by Multhopp
(ref. 5) and Van Spiegel and Wouters (ref. 6), table I has been prepared to give a tabular
comparison of CL,, aerodynamic center, and cD,i/cL2. The indications from table I
a r e that all three methods give results for CL, and Co,i/CL2 within 2 percent of
each other and the aerodynamic centers a r e within 0.3 percent of the reference chord of
each other when at a constant value of N. When the cD,i/cL2 values a r e compared
with l / r A (0.07457 for this case) they are found not to differ from it by more than
2.2 percent, indicating that the computed span loading distributions a r e nearly elliptical.
With the present method established as giving reliable results when compared with
the basic and modified Multhopp procedures (refs. 5 and 6) and with the studies previously
conducted, which indicated that an extension in the number of chordwise pressure modes
w a s needed for certain planforms, comparisons of predicted results obtained with the
present and other methods a r e made. The following planforms a r e offered as examples
of this :

(1) The aerodynamic centers and CL, values for the four simple planforms shown
in figure 1 2 are compared in tables II and III, respectively, with results obtained by
DeYoung and Harper i n reference 15 by using a Weissinger seven-point solution. (Values
of m of 9 o r 11 were chosen since N = 4, A = 2, and p = 1.)
From table 11 the aerodynamic centers predicted by the present method a r e seen to
lie from 1.7 to 4.2 percent of the reference chord ahead of those given by reference 15
and from table IIt the values of CL, a r e seen to vary from 1.62 to 5.35 percent below
those computed by using the present method. These differences are attributed to the
more accurate accounting of the induced camber effects by the present method.

15
 TABLE I.- COMPARISONS O F RESULTS OBTAINED FOR THIS PLANFORM
WITH THE PRESENT AND THE MULTHOPP METHODS
O F REFERENCES 5 AND 6

A = 4.000
.700

cLa a.c.
~

0.0564 0.270
Present .0569 .271 .08129
0.0571 0.279 0.08069
Ref. 6 .0580 .278 .08093
.0580 .277 .08087
0 .O 5836 0.262 0.08090
Present .05834 .263 .08086
~~

TABLE II.- AERODYNAMIC CENTERS FOR TABLE m.- LIFT-CURVE SLOPE FOR
A = 2 PLANFORMS O F FIGURE 12(a) A = 2 PLANFORMS O F FIGURE 12(b)
WHEN N = 4 WHEN N = 4

) 1 Aerodynamic center for - 7-

Lift-curve slope for -
-

x
A = Oo,
= 1.0 A = 0.25
m = 11 m = 11

0.191 0.156-
I1 0::2:
k = 45O, A = 45O,
i = 1.0, A = 0.25
n = 9 m=9
..
x = 1.0
m = 11
A zoo,
m=ll

0.0427
A =45O A = 45O,
1 = 0.25, X = 1.0, X =0.25
m=9

0.0398
m=9

0.0432
.245 .192 .0415 .0376 .0425

16

. .... m I ...
 TABLE 1V.- DOUBLE-DELTA NONDIMENSIONAL LIFT COMPARISON

A -
cLa
K Aspect ratio x cos(n,/2)ef* cos(*c/2)eff
A
(Present)
~~
(4
0 1.88810 0 0.7559 2.4979 0.02086 0.02108
.25 1.41500 82.63 .52 12 2.7148 .02234 .02040
.50 1.24920 78.07 .4800 2.6025 .02253
.75 1.11830 7 5.OO .5939 1.8830 .02309
1.oo 3.90020 40.00 .9221 4.2297 .01545
a N = 8; m = 23.

Comparisons of the span loadings a r e also made between the present method and
reference 15 for two of the planforms in figure 12 and a r e presented in figures 8 and 9
for various combinations of N and m. The span loading distributions computed by the
two methods a r e in good agreement for most sets of N and m.
(2) A series of double-delta planforms, whose general planform parameters are
given in table IV, have been used to obtain values of CL A which are also presented in
a/
table IV. They are compared with the values of C&/A predicted by Spencer in refer-
ence 16. From table IV, the CL A predicted values show that a maximum difference
a/
of 6 percent or -9 percent exists between the results of the two methods for the planforms
examined. The difference is attributed to the method used in reference 16 because it does
not consider the distribution in lifting pressure over the surface as the present method
does but is a refinement of a mathematical expression relating A and sweep to the value
of C L ~for wings which have broken leading edges.
(3) The three predicted span loading curves, given in figure 13 for a sweptback and
tapered planform, come from references 15 and 17 and the present method. The span

17
_..._- ... ......

loadings are seen to differ, in t e r m s of ycp, within 0.9 percent of the semispan from each
other. The CL, results of the present method and Campbell's method (ref. 17) agree
quite well and are about 5 percent greater than the result of DeYoung and Harper (ref. 15).
For the aerodynamic center, the results of the present method and Campbell's method
(ref. 17) are 3.2 and 2.1 percent, respectively, of the reference chord ahead of that pre-
sented by DeYoung and Harper (ref. 15).
(4) The two span loading curves (obtained from ref. 8 and the present method) shown
in figure 14 for the A = 3.436 double-delta planform have both been computed using
modifications of the Multhopp approach, The curve of reference 8 is based on the two
chordal loading function combinations used by Multhopp with a process of submatrix inver-
sion of the matrix of influence coefficients to a r r i v e at the local nondimensional lift and
pitching moment. The results of CL, and cD,i/cL2 are within 2 and 1 percent,
respectively, of each other, whereas the aerodynamic centers differ by about 3.7 percent
of the reference chord. The reason for this larger discrepancy in aerodynamic center
than found for CL, o r cD,i/cL2 can be traced primarily to the different local
pitching-moment t e r m s computed by each procedure. The reason for the differences in
local pitching moments is not clear.
(5) Three span loadings a r e presented for the double-delta planform of figure 15 and
a r e computed from the following methods: Campbell's (ref. 17), subsonic kernel function
(ref. 18), and present. Since the subsonic kernel function and the present method a r e basi-
cally identical (in the reduced frequency case) except for the handling of the integration
across the spanwise singularity, it is not surprising that the span loadings should be quite
similar. Consequently, the spanwise centers of pressure ycp differ only by about 2 per-
cent semispan. Further, although the values of CL, are within 1 percent, reference 18
gives a more stable pitching moment resulting in a more rearward aerodynamic center.
Reference 17 presents a span loading which has more of its loading outboard, there-
fore, a larger value of ycp. Also, its CL, is lower and its aerodynamic center is
ahead of the others.
In figure 15(b) the same trends in the spanwise variation of the centers of pressure
are seen to occur in each of the two lifting surface theories over the first 75 percent of
the semispan. However, for q > 89 percent of the semispan, the subsonic kernel func-
tion indicates that the chordwise center of pressure (xcp) lies ahead of the wing. The
local center of pressure for Campbell's method is the centroid of the two-dimensional
loading and occurs at the local 4 4 .
The difficulty experienced by the subsonic kernel function approach can be attributed
to an improper location of collocation points used in that solution. The control points
were at $ = 0.2, 0.4, 0.6, 0.8; y = 0.15, 0.35, 0.55, 0.75, and the semispan of the region
b/2
which contains the control point was set equal to 0.10. However, since there is not an
18
 optimum location procedure specified with this method, it is unclear what number and
location of collocation points would yield the best results. Perhaps more realistic results
could have been achieved by locating points farther outboard than 0.75. A cosine distri-
bution would have made an interesting comparison, but the collocation points cannot be
located very near the tip.
A simple case w a s attempted using only the outboard panel extended to the root, with
(
the collocation points located the same as previously mentioned in t e r m s of x/c and
)'b/2 and with a region around the control point of semispan equal to 0.10. In this case
reasonable results were found which indicate that the planform shape should be of pri-
mary importance in selecting these points.
Concerning the distribution of loading along the chords (fig. 15(b)), it is seen that,
in general, both lifting surface theories have the same overall shape and differ mostly in
magnitude until the subsonic kernel function becomes influenced by the locations of the
collocation points, that is, when xcp is ahead of the planform. Comparing the present
method with that of the two-dimensional distribution, the effects of induced camber can be
seen concentrated mostly inboard of the leading-edge break. The strength of the two-
dimensional loading of any spanwise station was chosen to be that predicted by Campbell's
lifting-line solution. This was accomplished by requiring that the total circulation
strength be equal to the integral of the two-dimensional distributed loading times a suita-
ble multiplier.
Comparisons with experimental data.- Since experimental data a r e generally lacking
for the planforms previously examined, the ability of the present method to predict the
aerodynamic characteristics of these wings cannot be appraised. Therefore, the following
discussion will relate to three wings for which experimental data a r e available and com-
parisons with theoretical results will be made,
(1) The two predicted span loadings presented in figure lS(a) for the A = 2 delta
planform agree closely and estimate reasonably the experimental values found in refer-
ence 19. Also, the CL values predicted by the two theoretical methods, present and
vortex lattice (ref. 20), are in good agreement with the experimental result computed from
the CL, of the force data, since they differ by only about 5 percent. (From the variation
of CL with a determined from the integrated pressure measurements, the CL value
at a = 4.30 is lower than that found by the force tests. The difference is attributed to
the lack of enough pressure orifices near the leading edge resulting in a poor integration.)
A comparison of the three aerodynamic centers shows that the value given by the
present method is ahead of both the experimental location and the one found by refer-
ence 20. (In general, for the lower aspect ratio wings, solutions obtained using higher
values of N and m result in a more aft location of the aerodynamic center. (See

19
figs. 6(a) and 7(a).) From an examination of the local loadings (figs. 16(b) and 16(c)), it
is seen that although the experimental local centers of pressure are better predicted by
reference 20, over most of the chord (x/c of 0.1 to 0.7), the present method gives better
agreement with the experimental pressure distribution. The disagreement between the
large pressures predicted near the leading-edge singularity and those measured may have
resulted from the thickness effects present in the experimental data. The singularity in
pressure at the leading edge also causes the local centers of pressure, predicted by the
present method, to be ahead of the experimental value.
(2) The present method predicts quite closely the experimental span loading of the
A = 8.02 sweptback and tapered wing of figure 17 as found in reference 21. This is seen
not only by the distribution but i n the ycp location. Also, the agreement between the
values of CL and aerodynamic center is good. From figures 17(b) and 17(c), it can be
seen that whereas the local chordal centers of p r e s s u r e a r e accurately predicted, the
pressures a r e , in general, underpredicted over the chord from x/c of 0.1 to 0.6 and
overpredicted from 0.6 to 1.0. Because of the singularity at the leading edge, the theo-
retical pressure coefficients in this vicinity a r e in excess of experimental values.
(3) In figure 18, the theoretical and unpublished1 experimental loading and moment
distribution a r e given for a variable-sweep wing in one sweep position. The predicted
overall CL at Q! = 3.14' is over 10 percent higher than that found experimentally from
force data and this difference is mentioned later. The predicted aerodynamic center is
7 percent of the reference chord ahead of the experimental value; however, this is not a
large discrepancy when one considers that the reference chord in this case is based on
the outer panel extended to the root. (See fig. 18.)
The experimental values of span loading which a r e shown in figure 18(a) agree well
with those predicted by theory; however, there is a slight overprediction by the theory
over the outermost part of the outer panel which can also be seen in the Cn, graph pre-
sented in figure 18(b). This theoretical overprediction of local Cn, leads to the overall
CL being higher than the experimental value. The effect of the leading-edge shed vortex
and different amounts of induced angle of attack generated by the theoretical and experi-
mental loading distribution undoubtedly account for some of the disagreement seen in
figures 18(a) and (b). Good agreement is found for the Cma! data obtained from theory
and experiment. The xcP graph is just the result of the Cn, and Cm, curves and
hence shows the local center of pressure slightly off of that predicted by theory.
At the four different spanwise stations shown in figure 18(c), the chordal loadings
a r e compared and the agreement is found to be reasonable at most locations.
Illustrative calculations. - Effect of variable sweep: Changes in the aerodynamic
characteristics due to increasing o r decreasing the outer panel wing sweep (from the
'Tests performed in the Langley high-speed 7- by 10-foot tunnel.
20
streamwise tip position) can also be predicted with the present method. For example, the
planform in figure 18 has its sweep varied from 15O to 40° at a Mach number of 0.23 and
the results, shown in figure 19, are based on the area and mean geometric chord of the
outer panel extended to the root when A = 30’. Those results presented are from the
present method and unpublished experimental data measured on a semispan pressure
wing at the Langley Research Center.
The agreement obtained between theoretical and experimental results is only fair in
that the theory shows the proper trends and similar increments, but not absolute levels.
Experimental values for CD,i/CL’ were not available from force data. As reported
in references 22 and 23, the variations presented in figure 19 are typical of variable-
sweep wings.
Effect of Mach number: The effect of increasing the subsonic Mach number on the
wing-alone aerodynamic characteristics is accounted for by use of the Prandtl-Glauert
compressibility rule as mentioned earlier. (See ref. 11.) In general, the results obtained
appear to be questionable above a Mach number 0.8 because of the inapplicability of the
-.
Prandtl-Glauert transformation as M 1 (ref. 11).
Some results obtained by applying this rule to a highly sweptback and tapered wing
a r e presented in figure 20 along with those determined from Mach numbers 1.2 to 2.8 by
use of the supersonic lifting-surface theory of reference 4. Experimental values for the
wing-body combination a r e presented for comparison.
In general, the theoretical methods are able to predict reasonably well the experi-
mental flat wing overall aerodynamic characteristics (which for subsonic speeds have not
been published but a r e found in ref. 24 for supersonic speeds) even though the experimen-
tal data contained the influence of a body. The one exception is the induced drag param-
eter which does differ considerably. However, t h i s is not unexpected since the real
leading edge is sharp and the leading-edge suction is considerably less than the 100 per-
cent assumed by the theory. Further, no induced drag, called vortex drag at supersonic
speeds, is given supersonically since it w a s not available separate from the wave drag.
In the transonic region (dashed lines) the results a r e obtained by fairing from sub-
sonic to supersonic speeds.
Warped wing: The determination of the aerodynamic characteristics for a warped
wing can also be found by using the present method when the local surface slope distribu-
tion is specified. In addition to the flat plate loading characteristics, which are always
found, the basic loading features are also computed by the use of the appropriate equations
given in appendix B. Also determined are the zero-lift angle of attack ao, pitching-
moment coefficient Cm,o, and root bending-moment coefficient C B , ~ . Illustrative
results of this procedure a r e presented in figure 21 for a double-delta planform with 5O

21
of linear twist and 2-percent circular a r c camber. (Other details of this planform and
loading distribution are to be found in the "Supplement to NASA TN D-4427" since it is
one of the sample cases given.)

MEAN CAMBER SURFACE DETERMINATION

The problem of determining the mean camber surface required to support a pre-
scribed loading on a given planform is easier to solve from the basic formulation than
finding the loading associated with a prescribed surface shape for the same planform.
This can be seen from the fundamental equation in compressible subsonic flow which is
equation (1) and is given as follows:

from which the mean camber surface is found by use of the following equation:

The determination of the local Ap(5,q) from equation (1) has been shown to involve
solving the equations inversely for a part of its integrand; whereas, to find w/U at any
point, the individual specified loading must be simply multiplied with the appropriate ker-
nel function value and summed over the surface. Thus, if the more difficult problem is
solved o r programed first (as i n the present study), then the other only requires, in addi-
tion to those changes mentioned, the deletion of the matrix inversion and the addition of
the w/U integration to find z/c.
Since values of w/U a r e only computed at the control points, some representation
at other chordwise locations is needed in order that the integration of equation (13) can
be carried out. A least-square polynomial curve f i t of the w/U values at the control
points was chosen to represent the variation of w/U along the chord between the first
and last control points. The polynomial curve f i t is specified to have a degree of one less
than the number of chordal control points to insure that a good f i t can be obtained. Out-
side the range of the control points, a linear extrapolation of the polynomial curve f i t is
used to the nearest chordwise edge.

22
 Optimization of N and m
As in the load determination section, a study was undertaken here to find the rela-
tionship that N and m should have in order that the best mean camber lines could be
obtained. It was intuitively felt that these mean camber lines would occur when the most
control points were employed. For the A = 2 delta planform shown in figure 22, four
combinations of N and m w e r e selected and in each case the same surface loading
distribution w a s used. It was composed of only a sin 9 chordwise loading which varied
elliptically from tip to tip and was adjusted to give an overall lift coefficient of unity.
All other pressure mode coefficients were set equal to zero.
Figure 22(a) shows that the computed downwash w/U values at a constant m with
varying N do not lead to a much changed polynomial curve f i t , whereas at a constant
value of N varying m does. The curves appearing in this figure a r e integrated and
appear in figure 22(b). As would be expected, s e t s of N and m which have essentially
the same curve f i t lead to the same mean camber line. Thus, the mean camber lines for
the sets N = 4 and N = 8 at m = 11 a r e the same. This is also true for the sets of
N = 4 and N = 8 at m = 23.
Thus it appears that for the given surface loading, increasing the number of span-
wise stations at which the desired chord loadings a r e defined is more important in
obtaining the best mean camber lines than increasing the number of chordal pressure
modes involved (much above those needed to define the shape) at each spanwise station.
This is a similar conclusion to that reached in reference 25. (Note that the prescribed
loading does not contain any leading-edge suction and hence it is not known what the
effects of including the suction term would have on the camber for optimum combinations
of N and m.)
With the relationship that N and m should have for certain loadings established,
some comparisons with other theoretical methods are made and follow in the next section.

Comparisons
The accuracy of the present method may be appraised by comparing in figure 23
the downwash distribution at the plane of symmetry of an A = 10 rectangular wing with
that of a two-dimensional wing with the same sin 9 chord loading at each span station.
Over most of the chord the three-dimensional downwash is slightly lower than that
of the two dimensional by a constant amount. If a much higher aspect ratio wing had been
used in the computation, this difference would become smaller.
At the edges, where x/c - 0 and 1, the three-dimensional downwash is seen to
tend toward plus and minus infinity, respectively. This is caused by the logarithmic

23
 -
singularity correction term, which results from the second-order singularity
(Y
in the integrand, containing a t e r m which increases without bound at all edges. (See
!,)"
eq. (A9).)
From figure 23, it can also be noted that by using only a sin 8 loading increasing
the number of chordal control points does not change the downwash distribution at o r in
the vicinity of the chordal control points already computed but does give a more detailed
representation of the distribution.
As another check, the mean camber lines obtained with the present method for the
sweptback wing shown in figure 24 a r e presented there2 for comparison with those pre-
dicted by the mean camber surface solution of Katzoff, Faison, and DuBose (ref. 26).
Three differences a r e apparent in the procedures used and are
1. The present method rounds any kinks that occur in the wing planform (i.e., at the
root) which results in a closer representation of the isobars on the planform (ref. 5) and
hence gives a better downwash distribution than the method used in reference 25 (ref. 27).

2. The uniform chord loadings ACp used in reference 26 a r e approximated herein

by a Fourier s e r i e s of four odd sine t e r m s - sin 8, sin 39, sin 59, and sin 79. This
s e r i e s o r others like it composed of cot 9 and even sine functions give greater flexibil-
ity in approximating arbitrary pressure loadings. The amplitudes of the chord loading
were uniform over the entire a r e a and produced a lift coefficient of 1.0.
3. The present method uses the least-square polynomial curve f i t and constant slope
scheme to determine the downwash values all along the chord as discussed earlier. This
procedure differs from that used in reference 26 where the downwash distribution com-
puted at selected internal constant chord lines is completed by fairing except near the
edges of the chord. There the downwash is assumed to be equal to that required to pro-
duce a constant pressure all along the resulting mean camber lines.
The curves which are discussed in the previous paragraph are then integrated to
find the mean camber height z/c above the x-y plane. Despite the differences dis-
cussed between these two methods, figure 24 shows that the agreement between them is
good except near the plane of symmetry. This is explained in part by the first differ-
ence listed. (See also ref. 27.)
Application.- An application of the present method to the highly sweptback and
tapered planform shown in figure 253 is made where the lift coefficient for this planform
~~ - ~ -
2The results of reference 26 were found by integrating from the leading edge r e a r -
ward and hence were adjusted in figure 24 to the same zero level for comparison.
3Tabular results for this planform and loading distribution are to be found in the
"Supplement t o NASA TN D-4427" since it is used as the sample case.

24
w a s taken as unity at a Mach number of 0.30. The chordal loading function coefficients
w e r e specified so that there was (1) a zero cot 9 distribution, (2) an elliptical distribu-
2
tion of sin 9 across the span, (3) a sin 29 distribution such that the local xcp varied
parabolically from 50 percent of the local chord at the plane of symmetry to 20 percent at
the break and from the break linearly back to 50 percent at the tip, and (4) a sin 39 dis-
tribution which caused the slope of the chordal loadings to be zero at the trailing edge.
Figure 25 shows that the mean camber lines have a rearward movement of the x/c
location of maximum camber and a twist which increases toward the tip. The mean cam-
ber line at q = 0.997 is typical of stations near the tip and should be examined carefully
before a direct application is made since the downwash distribution which produced it may
be overly influenced by the singular nature of the downwash at the tip.

CONCLUSIONS

The results obtained with a modified version of Multhopp's lifting-surface theory

programed for the IBM 7094 electronic data processing system and Control Data 6400
computer system to find the surface loadings and mean camber surfaces have been
studied and the following conclusions a r e drawn:
For the surface loading program:
1. It has been found that the loading program may be applied with reasonable accu-
racy to conventional and composite planforms to find their flat plate characteristics and to
wings with twist and/or camber to determine the basic loading distribution as well as the
force and moment coefficients at zero lift.
2. In predicting the surface loading, the number and spacing of control points is more
important when seeking an accurate overall aerodynamic center than when a solution for
the lift-curve slope, span loading, o r spanwise location of center of pressure is being
sought. It has been found that at least four, and preferably more, chordal control points
should be used in obtaining aerodynamic centers. This number should depend upon the
PA of the planform being considered, since the more control points used (100 available)
at each spanwise station necessitates a decrease in the number of spanwise stations. This
may lead to a problem when values of PA a r e large. However, in general, the number
of chordal control points should be kept between 6 and 10 to insure that the resulting val-
ues of aerodynamic center lie within the region of converging results.
3. Reasonable agreement is generally obtained with both experiment and other theo-
r i e s concerning local and overall loading characteristics.
4. The effect of Mach number can be determined and the resulting agreement with
experiment is found to be reasonable.

25
 For the mean camber surface program:
1. Good agreement is achieved between the downwash required by two-dimensional
theory for a sin 8 loading and that predicted by a three-dimensional uniform sin 8
loading at the plane of symmetry of an aspect-ratio-10 rectangular wing.
2. The agreement with results from other mean camber predictions is found to be
fairly good and the present method has the capability of approximating most chord loading
shapes by the use of a Fourier sine series.
3. For an acceptable prediction of mean camber surfaces which includes no leading-
edge suction, the number of chordwise stations should be nearly equal and never less than
the number of pressure loading t e r m s required to properly define the desired shape and
the number of spanwise stations should be as many as can be accommodated by the
machine for that number of chord stations.
4. The resultant mean camber surface predicted for a composite planform whose
loading includes no leading-edge suction was found to have a reasonable twist and camber
distribution inboard of the tip. In the vicinity of the tip, the mean camber surface does
not appear to be as reasonable due to the singular nature of the downwash at the tip.

Langley Research Center,

National Aeronautics and Space Administration,
Langley Station, Hampton, Va., January 19, 1968,
126-13-01-50-23.

26
 APPENDIX A

FORMULATING THE INFLUENCE COEFFICIENT MAT=

The formulation of the influence coefficients is accomplished by integra-

ting (or approximating) the chordal influence that the pressure loading functions exert on
each control point from every other spanwise station. (See ref. 6.) Then, for symmetri-
cal spanwise loadings, the chprdal influence at q-n, LJv-n(@s),is added to that of qn,
Ltn(@s).,to give the total ch(@~),
and for antisymmetrical loadings the influence at
q-n, LJv-n(@s),is subtracted from that at qn, L&(@s). The value at the plane of sym-
j
metry, is just equal to L ~ A ) ( @when
~ ) the loading is symmetrical and zero
when antisymmetrical.
Now,for v # n,

with
v=-- m - 1
2 ’
- -m - 3,. . . , o , . . ., --
m-3 m - 1
2 ’ 2

where
f-
I

and the subsonic kernel function is given by

27

I
 APPENDIX A

where

X = -C(Y)COS @ + d(y)
5 = - C ( ~ ) C O S 8 + d(q)

Now,for n = v,

LiV(GS) = bvVKj (@s,yv;qv)+ Logarithmic singularity correction

where

The logarithmic singularity correction t e r m (LSC) arises because of the spanwise inte-
gration across the second-order singularity 1
. Its derivation is discussed by
(Y - d 2
Multhopp in reference 5, Mangler and Spencer -in reference 28, and Van Spiegel and
Wouters in reference 6 . Its best form is

-
n=--m- 1
2
I
(In - VI odd) (A9)

where the prime on the summation sign means that the t e r m when n = v is eliminated in
the summation process and

28
29
 APPENDIX B

EQUATIONS USED IN THE COMPUTATIONS

The following equations are used to determine the total local characteristics at the
design lift coefficient as well as the overall characteristics for wings with twist and/or
camber. Note that the qtj(qn) t e r m s represent the pressure mode coefficients asso-
ciated with the twist and/or camber boundary conditions, and the qj(qn) t e r m s represent
the pressure mode coefficients associated with the flat plate boundary conditions. The
J n) t e r m s a r e arrived at in a manner similar to that of the qj(qn) terms, as given
qt-(r]
by the matrix inversion of equation (4), except that the -4w('s'yv' values a r e now the
U
local angle of attack for the warped wing at the flat plate control points when the root
chord is at zero angle of attack.

Local Aerodynamic Characteristics

(a) P r e s s u r e coefficient
f-

i
+Lj=1

(b) Chord loading

L,l -

(c) Nondimensional lift coefficient (called circulation in the "Supplement to NASA

TN D-4427")

30

I
 APPENDIX B

(d) Pitching moment about local leading edge

( e ) Location of center-of-pressure loading from Y-axis

(f) Center-of -pressure loading i n fractions of local chord

(g) Span load coefficient

Overall Aerodynamic Characteristics

(a) Flat plate lift
-
m- 1
2

31

I
 APPENDIX B
(b) Flat plate pitching moment about the total wing E/4

(c) Flat plate pitching moment about the reference wing cref/4 =

(Flat plate pitching moment)total wing - Lift

(d) Flat plate rolling moment about the X-axis

(e) Flat plate location of the aerodynamic center

(Flat plate pitching moment)referenCe wing %ef

-
_ -
Lift + Xcref + 7
(f) Flat plate aerodynamic center in fractions of reference chord, measured from
the Xcref
(Flat plate pitching mOment)reference wing 1
a.c. = - +z
Lift(cr ef>

( g ) Angle of attack at zero lift

CL,1
a0 = --CLCY

where CL,1 is determined similarly to flat plate CL

32
 APPENDIX B

(h) Pitching-moment coefficient at zero lift

where Cm,l is determined similarly to flat plate Cm

(i) Root bending-moment coefficient at zero lift

33

I
 REFERENCES

1. Crigler, John L.: A Method f o r Calculating Aerodynamic Loadings on Thin Wings at a

Mach Number of 1. NASA TN D-96, 1959.
2. Smith, J. H. B.: Calculation of the Shape of a Thin Slender Wing for a Given Load Dis-
tribution and Planform. C.P. No. 385, Brit. A.R.C., 1958.
3. Carlson, Harry W.; and Middleton, Wilbur D.: A Numerical Method for the Design of
Camber Surfaces of Supersonic Wings With Arbitrary Planforms. NASA TN D-2341,
1964.
4. Middleton, Wilbur D.; and Carlson, Harry W.: A Numerical Method for Calculating the
Flat- Plate P r e s s u r e Distributions on Supersonic Wings of Arbitrary Planform.
NASA TN D-2570, 1965.
5. Multhopp, H.: Methods f o r Calculating the Lift Distribution of Wings (Subsonic Lifting-
Surface Theory). R. & M. No. 2884, Brit. A.R.C., Jan. 1950.
6. Van Spiegel, E.; and Wouters, J. G.: Modification of Multhopp's Lifting Surface Theory
With a View t o Automatic Computation. NLR-TN W.2, Nat. Lucht- Ruimtevaartlab.
(Amsterdam), June 1962.
7. Milne-Thomson, L. M.: Theoretical Aerodynamics. Second ed., MacMillan and Co.,
Ltd., 1952, p. 225.
8. Nagaraj, V. T.: Theoretical Calculation of the Aerodynamic Characteristics of a Dou-
ble Delta Wing (by Lifting Surface Theory). M.E. Thesis, Indian Inst. Sci., 1961.
9. Van de Vooren, A. I.: An Approach to Lifting Surface Theory. Rep. F. 129, Nat.
Luchtvaartlab. (Amsterdam), June 1953.
10. Garner, H. C.; and Lehrian, Doris E.: Non-Linear Theory of Steady Forces on Wings
With Leading-Edge Flow Separation. NPL Aero Rep. 1059, Brit. A.R.C., Feb. 15,
1963.
11. Kuethe, A. M.; and Schetzer, J. D.: Foundations of Aerodynamics. Second ed., John
Wiley & Sons, Inc., c.1959, pp. 200-203.
12. Munk, Max M.: The Minimum Induced Drag of Aerofoils. NACA Rep. 121, 1921.
13. Garner, H. C.: Some Remarks on Vortex Drag and Its Spanwise Distribution in Incom-
pressible Flow. NPL Aero Note 1048, Brit. A.R.C., Nov. 4, 1966.
14. Lamar, John E.; and Alford, William J., Jr.: Aerodynamic-Center Considerations of
Wings and Wing-Body Combinations. NASA TN D-3581, 1966.

34
15. DeYoung, John; and Harper, Charles W.: Theoretical Symmetric Span Loading at
Subsonic Speeds f o r Wings Having Arbitrary Plan Form. NACA Rep. 921, 1948.
16. Spencer, Bernard, Jr.: A Simplified Method for Estimating Subsonic Lift-Curve Slope
at Low Angles of Attack for Irregular Planform Wings. NASA TM X-525, 1961.
17. Campbell, George S.: A Finite-Step Method for the Calculation of Span Loadings of
Unusual Plan Forms. NACA RM L50L13, 1951.
18. Watkins, Charles E.; Woolston, Donald S.; and Cunningham, Herbert J.: A Systematic
Kernel Function Procedure f o r Determining Aerodynamic Forces on Oscillating o r
Steady Finite Wings at Subsonic Speeds. NASA TR R-48, 1959.
19. Wick, Bradford H.: Chordwise and Spanwise Loadings Measured at Low Speed on a
Triangular Wing Having an Aspect Ratio of Two and an NACA 0012 Airfoil Section.
NACA TN 1650, 1948.
20. Dulmovits, John: A Lifting Surface Method f o r Calculating Load Distributors and the
Aerodynamic Influence Coefficient Matrix for Wings in Subsonic Flow. Rep.
No. ADR 01-02-64.1, Grumman Aircraft Eng. Corp., Aug. 1964.
21. Graham, Robert R.: Low-Speed Characteristics of a 45O Sweptback Wing of Aspect
Ratio 8 From P r e s s u r e Distributions and Force Tests at Reynolds Numbers From
1,500,000 to 4,800,000. NACA RM L51H13, 1951.
22. Baals, Donald D.; and Polhamus, Edward C.: Variable Sweep Aircraft. Astronaut.
Aerosp. Eng., vol. 1, no. 5, June 1963, pp. 12-19.
23. Alford, William J., Jr.; Luoma, Arvo A.; and Henderson, William P.: Wind-Tunnel
Studies at Subsonic and Transonic Speeds of a Multiple-Mission Variable-Wing-
Sweep Airplane Configuration. NASA TM X-206, 1959.
24. Morris, Ode11 A.; and Fournier, Roger H.: Aerodynamic Characteristics at Mach
Numbers 2.30, 2.60, and 2.96 of a Supersonic Transport Model Having a Fixed,
Warped Wing. NASA TM X-1115, 1965.
25. Garner, H. C. : Accuracy of Downwash Evaluation by Multhopp's Lifting-Surface
Theory. NPL Aero Rep. 1110, Brit. A.R.C., July 1964.
26. Katzoff, S.; Faison, M. Frances; and DuBose, Hugh C.: Determination of Mean Camber
Surfaces f o r Wings Having Uniform Chordwise Loading and Arbitrary Spanwise
Loading i n Subsonic Flow. NACA Rep. 1176, 1954. (Supersedes NACA TN 2908.)
27. Thwaites, Bryan, ed. : Incompressible Aerodynamics. Clarendon Press (Oxford),
1960, pp. 321 and 363.
28. Mangler, K. W.; and Spencer, B. F. R.: Some Remarks on Multhopp's Subsonic
Lifting-Surface Theory. R. & M. No. 2926, Brit. A.R.C., 1956.

35
 (a) Planform variables.

(bl Reference wing geometry.

Figure 1.- General planform w h i c h can be used in these programs along w i t h its coordinate system and reference wing geometry.

36
I

Y=-1 Y=O Y=l

yv
8=7T 8:-27T 4 8=0
Left tip P l a n e o f symmetry Righi t i p

(a) Spanwise. Bv =
2 - m t l ’ where v ranges from - m-l
2 to m-l
2 .

x/c =o X/,=l.O
+ =O += T

L eading Trailing
edge edge

(b) Chordwise. QS =-2sn where s ranges from 1 to N.

2N t 1’
Figure 2.- Control-point locations.

37
 S

Figure 3.- Detailed chordwise location of control points. $= 4 (1 - cos QS),

38
U

Number o f span stations,m

(b) Aerodynamic center.

Figure 4.- Continued.

 /v
0 2
0 4
0 6
A 8

0 4 8 12 /6 20 24 28 32 36 40
Number o f span siaiions,m

(c) CLa.

Figure 4.- Concluded.

 -

Number of spanwise slotions, h

(a) Induced drag parameter.
Figure 5.- Effect of varying number and locations of control Points and chordal loading functions on aerodynamic characteristics of A =7 rectangular planform at M = 0.
 N
0 2

a.c.,
pen
E

Number o f spanwise stations,m

(b) Aerodynamic center.

Figure 5.- Continued.

U.C.,
pen A = 1.700
E

2.480
I

1
Number o f span stations, m

(a) Aerodynamic center.

Figure 6.- Effect of varying number and locations of control points and chordal loading functions on aerodynamic characteristics of A = 1.7 double-delta planform at M = 0.
 0 4 8 /2 /6 20 24 28 32 36 40
Number o f spun siaiions, m

(b) CL,,.

Figure 6.- Concluded.

46
-

a.c.1
-
perce
n

(a) Aerodynamic center.

Figure 7.- Effect of varying number and locations of control points and chordal loading functions on aerodynamic characteristics of A = 1.97 double-delta planform at M = 0.
 N
0 2
n 4
0 6
A 8

U 4 8 12 16 20 24 28 32 36 40
Number of span stations., m

(b) C L ~ .

Figure 7.- Concluded,

 YCP
,41154
.42420
.42654
,42688
.428

0 ./ .2 .3 .5 .6 .7 .8 .9
Q

(a) Ratio of m to N is approximately constant.

Figure 8.- Span loading coefficient variation with different combinations of N and m for A = 2 rectangular planform at M = 0.

49
 (b) Constant N.
Figure 8.- Continued.

50
 (c) Constant m.

Figure 8.- Concluded.

51
.I. 111.111 ,1111 ..,I. ,,,., I I,, I, I,,,, , 1 1 1 1 1 I..._.. ..--.-.....

i1 G,l/CL2
./59/ 8

'J1
.5 ,04312 :00049
,04324 .00005 ./59/ 7
0 6 .04295 -.00026 ./59/7
.04282 700044 ./59/7
.4
A
42
8
R e f-1 5 ,0425 - -

"0 ./ .2 .3 .5 .6 .7 .8 .9 LO
Q

Figure 9.- Span loading coefficient variation w i t h different values of N for A = 2 sweptback and tapered planform at M = 0 when m = 9.

52
 Cz.c

Figure 10.- Span loading coefficient variation with different combinations of N and m for A = 1.7 double-delta planform at M = 0.

53

I
 'I

.2
-.00306 .32336 .39533 ./6/65
.04090 -.00247 .3/047 .4/220 .I6162
.04062 -.00268 .3/592 .4/2// .I6162
./ 04074 -.00293 .32/96 .420/7 .I6162
.04067 - .00304 .32475 .42098 ./6/62

0 "!'l1lllIlIII 'I1 I / ' / I /I ,I/ /I I I I II It /I I I I I $1 I I I l l I I I I I I I I I It 4Ti

0 / 2 3 4 5 6 7 8 9 LO
Q

Figure 11.- Span loading coefficient variation with different combinations of N and m for A = 1.97 double-delta planform at M = 0.

54
 A a2.000 M=O
I

b 11.31.

a.c.,
percent
E

U
0 4 8 12 /6 20 24 28 32 36 40
Number o f span s t a t i o n s , m

(a) Aerodynamic center.

Figure 12.- Effect of varying number and locations of control points on aerodynamic characteristics for four A = 2 planforms at
M = 0 when N = 4.

55
 A *2.000 M=O

L \ I I f h *

Number o f span stat ions, m

(b) C L ~

Figure 12.- Concluded.

 Method
N=4,m=3/ .06/5 .252 .460
-- Ref /7 .06/3 263 .463
Ref 15 .0583 284.

A =6
M=O

0 ./ .2 .3 .4 .5 .6 .7 .8 .9 LO
I

Figure 13.- Span loading distributions predicted by three different theoretical methods for A = 6 sweptback and tapered planform at M = 0.

57
 iI ”
I 1 1 N = 2,m=/5
cma
.05722 -.00525 .34184
0; ~ c pl-cQ/cLz
.420 .09270
--N = 2, m = /5,r e f 8 .05792 -.003/8 .30500 .427 .09259

0 ./ .2 .3 4 .5 .6 .I .8 .9 LO
7

Figure 14.- Span loading distributions predicted by two different theoretical methods for A = 3.436 double-delta planform at M = 0.

58
 (a) Spanwise.

Figure 15.- Loading distributions predicted by three different theoretical methods for A = 1.415 double-delta planform at M = 0.

59
 XCP,
percenf
loco/ chrd

4 .5 .6 .7 .8 .9 LO
17

0 ./ .2 7
.J 4 .5 .6 .7 .8 9
X/C

(b) Local spanwise centers of pressure and chordwise load distribution.

Figure 15.- Concluded.

60
 Q = 43O

,408
,396 ,414

‘I
CLCav

-0 ./ .2 .3 4 .5 .6 .7 .8 .9 10
Q

(a) Spanwise.

Figure 16.- Loading distributions predicted by two different theoretical methods and compared with experiment for A = 2 delta planform
at M = 0.13.

61
 N=4,m=//
-- Ref 20
0 Experimenf, ref/9

XCP,
percent
/oca/ chord

-0 2 .3 .5 .6 .7 .8 .9 IIO
Q

-.6

-4
4
-. 2

./ .2 .3 .4 .5 .6 .7 .8 .9 LO
X/C

(b) Local spanwise centers of pressure and chordwise load distributions.

Figure 16.- Continued.

62
 (b) Concluded.

Figure 16.- Concluded.

63
 (I =4P
Method
N = 4,m= 37
Experiment,
ref:2/

0 ./ .2 .3 .4 .5 .6 .7 .8 .9
P

(a) Spanwise.

Figure 17.- Loading distributions predicted by present method and compared with experiment for A = 8.02 sweptback and tapered planform
at M = 0.19.

64

I l l I Ill I II I II I
 xcp,
percent
/oca/ chrd

.I .2 .3 4 .5 .6 .7 .8 .9
P

(b) Local spanwise centers of pressure and chordwise load distributions.

Figure 17.- Continued.

65
 At= 4 , m = 3 7
Q Experimenf ,ref. 21

,/IIii
, p=.55
I
I

I l

T
I
I
11

I
I
'I
p =.75

'II'1
,I
I
I
TI
I

"/ p=.96

0 ./ .2 .3 .5 .6 7 .8 .9

(b) Concluded.

Figure 17.- Concluded.

66
 0=3./4O

I Method
,295

A =4.303

M= 0.23

-0 ./ .2 .3 4 .5 .6 .7 .8 .9
7

(a) Spanwise.

Figure 18.- Loading and moment distributions predicted by present method and compared w i t h experiment for A = 4.303 variable-sweep
planform at M = 0.23.

67

I
 'rI
I

1
I

-N = 4 , m = 3 3
Qo Experiment
unpublished

local chord

0 .l .2 .3 .4 .7 1.0

(b) Local slope of normal-force and pitching-moment coefficients and centers of pressure.

Figure 18.- Continued.

68
-N = 4 , m = 3 3
0 Experiment',
unpublished

.5 8 .7 .8 .9 /.O

(c) Local chordwise loadings.

Figure 18.- Concluded.

69
I

Figure 19.- Effect of leading-edge outboard sweep on some longitudinal aerodynamic characteristics of variable-sweep wing (N = 4, m = 37, 33, and 29 for A = 15O, 300, and
40°, respectively) at M = 0.23.
 A I 1.697

2.400
\

..."...._".........".
_~__
-______

__ __

~
- ~
_ ___
~~~
_ ~

__
.-

unpublished subsonic dofa

____
Wing- alone theory (subsonic ond supersonic)

______
--- Transonic fairing

-________
_--_ ~

0 .8 I!2 I!6 20 24 28 0 4 .8 l.2 l.6 20 24 2B

M M

Figure 20.- Effect of Mach number on some longitudinal aerodynamic characteristics of highly sweptback and tapered wing (subsonic theory, N = 8, m = 23).
 ---I Tw/sf
Twist

3543
~

A =1(49/
M=O

? O HI Meun cumber
-
Hei9ht02 -
I-

C ’O
0 .2 .4 .6 .8 1.0
I

Figure 21.- Warped A = 1.491 double-delta planform at M = 0 and computer results.

72
 (a) Computed w/U and least-square polynomial curve fit.

Figure 22.- Effect of varying number and locations of control points on downwash distribution and shape of mean camber lines for A = 2
delta planform at M = 0.

73
 I 1 I1 I 1111.1.1" . 111 I I I. m1.1m.m. ..I .... . .II..I.I ,

m
//

"0 .2 4 .6 .8 LO
0 .2 I .6 .8 /O
VC

(b) Mean camber lines.

Figure 22.- Concluded.

74
 VC

Figure 23.- Comparison of two- and three-dimensional downwash predictions from uniform sin 0 chord loading at tl = 0.000.
 N = 8,m = 2 9

same
span1

.3 4 .5 .6 .7 .8 .9 1.0
X/C

Figure 24.- Chord loading distributions and comparison of mean camber shapes resulting from present and another method for A = 8
sweptback and tapered planform at M = 0.9 for a uniform area loading at CL = 1.0.

76
 r-
,345
A = 8.000
M = 0.90

40.65'

-1.000

N = 8 ,m=29
--Ref 26

./
.. i
9c 0 1
-. /
I
.f 1
I
'%c @
I
r
7
-.I

1
I
'I I-
I-
9c L
T
I
-.I
I
I
I
'I
I
1
I I
vc A
1
I T
i i1
I

I
I
I
5 $
II I
.2 .3 .4 .5 .6 .7 .8 .9 LO
X/C

Figure 24.- Concluded.

77
 A = 1.705
M = 0.30

1 I
I I
3
-

I
T ?
!
T
T I
1I 3I
I I
/
T
I
-+ I
-r
!
i i
I I
I

+ 1

i I
!
-7-

!
I
f
~

I
I
!
7I
!
i I
I
i
II I
I I
0 ./ ." 4 .5 .6 .7 .8 .9
X/C

Figure 25.- Predicted mean camber shapes for A = 1.705 highly sweptback and tapered planform at M = 0.30 when N = 4 and m = 39.

78 NASA-Langley. 1968 -1 L-4760

NATIONAL
AERONAUTICS
AND SPACE ADMINISTRATION POSTAGE AND FEES PAID
WASHINGTON,
D. C. 20546 NATIONAL AERONAUTICS A?
SPACE ADMINISTRATION
OFFICIAL BUSINESS FIRST CLASS MAIL

' T h e aeronautical and space activities of the United States shall be

.
conducted so as to contribute . . to the expansion of human knowl-
edge of phenomena in the atmosphere and space. T h e Administration
.. shalZ $rovide fa the widest practicable and appropriate dissemination
of information concerning its activities and the results thereof."
-NATIONAL AERONAUTICS
AND SPACE ACT OF 1958
%

NASA SCIENTIFIC AND TECHNICAL PUBLICATIONS

TECHNICAL REPORTS: Scientific and TECHNICAL TRANSLATIONS: Information
technical information considered important, published in a foreign language considered
complete, and a lasting contribution to existing to merit NASA distribution in English.
knowledge.
SPECIAL PUBLICATIONS: Information
TECHNICAL'NOTES: Information less broad derived from or of value to NASA activities.
in scope buthevertheless of importance as a Publications include conference proceedings,
contribution to existing knowledge. monographs, data compilations, handbooks,
sourcebooks, and special bibliographies.
TECHNICAL MEMORANDUMS :
Information receiving limited distribution TECHNOLOGY UTILIZATION
because of preliminary data, security classifica- PUBLICATIONS: Information on technology
tion, or other reasons. used by NASA that may be of particular
interest in commercial and other non-aerospace
CONTRACTOR REPORTS: Scientific and npplicatiors. Publications include Tech Briefs,
technical information generated under a NASA Technology Utilization Reports and Notes,
contract or grant and considered an important and Technology Surveys.
contribution to existing knowledge.

Details on the availability of these publications may be obtdned from:

SCIENTIFIC AND TECHNICAL INFORMATION DIVISION

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
Washington, D.C. PO546