3217
3217
3217
M I N I S T R Y . OF A V I A T I O N
L O N D O N : H ER M A J E S T Y ' S S T A T I O N E R Y O F F I C E
1961
- PRIC~.: lls. 6d. N~.T
Formulae for Calculating the Camber Surfaces of
Thin Swept-Back Wings of Arbitrary Plan-form
with Subsonic Leading Edges, and
Specified Load Distribution
B y G. M. ROPER, M.A., P h . D .
Summary. Formulae for calculating the-gradients and ordinates of the camber surfaces of swept-back
wings of arbitrary plan-form with subsonic leading edges, and specified load distribution, are given, including
those which have been programmed and used for DEUCE calculations for some swept-back and M-wings with
curved leading edges.
Some methods for the numerical calculation of singular integrals are given.
For polygonal wings with simple load distributions, the equation of the camber surface is given in closed
form. This is useful for obtaining approximate results for more general plan-forms.
1. Introduction. This paper is a brief account of a method for calculating the gradients and
ordinates of thin swept-back wings of arbitrary plan-form, with subsonic leading edges. No claim
is made that anything is new except perhaps the suggested treatment of singularities for numerical
integration, and some of the formulae and integrations produced.
Integrals, based on linearised supersonic theory, for calculating the upwash or the streamwise
gradients of a wing surface for a given load distribution (the direct problem--Ref. 3) have been known
for some time. But until fairly recently, when large scale calculations with the aid of electronic
computers became possible, little use had been made of these integrals for wing design. The need
of a method for designing camber surfaces to produce specified load distributions has arisen in
connection with a research programme on problems associated with flight at low supersonic
speeds (Refs. 5, 6).
In this paper will be found:
(1) the basic integrals and formulae for calculating t i e streamwise gradients and ordinates of a
camber surface to support a given load distribution, the wing plan-form being arbitrary,
with subsonic leading edges and any trailing edge;
2. Calculation of the Shape of the Camber Surface of a Wing of Given Plan-form for a Given Load
Distribution. Considering incidence effects only, and using the linear small perturbation theory
of supersonic flow, Volterra (Ref. 1) has shown that the acceleration potential f~ can be expressed
in the form
1 a f f Af2(x-~)(z-C)dS
~ ( x , y , z ) - 2~ax T [ ( y - ~ ) ~ + ( z - ~ ) ~ ] { ( x - ~ ) ~ - ~ [ ( y - ~ ) ~ + ( z - )]} 11 , (1)
where f~ = V ~x' V is the free stream velocity, parallel to the x-axis, x, y, z are a right-handed
system of rectangular Cartesian co-ordinates, z being measured upwards, and ¢ is the perturbation
velocity potential. M is the free stream Mach number and/?~ = M 2 - 1. Af~ is the jump in the
value of f~ across the surface of the wing, and integration is over the part, ~, of the wing surface
for which ( x - ~)~ - fl~ [ ( y - ~)~ + ( z - ~)~] > 0 and x - ~ > O.
2
Af2 is given in terms of the loading coefficient, ACe, by the linearised Bernoulli equation:
The streamwise gradient of the camber surface is given by: (according to the linear theory)
8z~_ 1 8
y, (3)
Ox V 8z V 2 8z ~o
Taking the wing to lie approximately in the plane ~ = 0, the streamwise gradient at point (x, y, z)
of the surface is given by
8z_ l 8 f A C e z ( x - ~) d(&7
8x 4~ 8z f r [(Y-- ~7)z + Z~] [(X-- ~)2 _ / ~ 2 { ( y _ ~7)2 + z2}]112 "
(4)
P P
where j denotes the 'generalised principal part of the integral'. (see Appendix II.)
~,(~, 7) dv
where [- denotes the 'finite part of the integral', and ~u(~, ~7) is the velocity potential on the upper
surface of the wing. (See Appendix II.)
In general, the gradient must be evaluated numerically from (6) or (7). The integrand becomes
singular along the Mach lines through the point (x, y) and, in Equations (6), also along the line
~/ = y. Also, for some forms of ACe, the value of Oz/ax given by Equations (6) or (7) may become
infinite along the leading or trailing edges, or along certain lines ~7 = constant. Methods of dealing
with these singularities are discussed in Section 5.
For some forms of ACe, it may be more convenient to use characteristic co-ordinates (r, s).
Taking the Mach lines through the origin as axes of co-ordinates, the transformation formulae are:
x = -tiM(r + s), M
," = ~ ( x - flY),
1 M
y = ~ ( s - r), s = yfi (x + fly), (8)
(82476) A*
and the formula for the streamwise gradient is
P P
a~_ M ( (¢~(q,*l)drld,1
(lo)
ax 4~ v J~ J{(r - ,1) (s - $1)}~~'
(which can also be put into other forms).
T h e ordinates z of the camber surface are found by numerical integration from
~(~) = f(~);
~z(r/) = x - / 3 [y - , [ ~<g(w), x - f(~?) >/9 lY - ~7 [
or ~2(~/) = g(~?) ~< x - 19 [ y - ~7 1;
~1 = ~ = 0 for x -f(~?) </3 ] y - ~ 1, (16)
where ~: = f(~l), ~ = g(~?) are the equations of the leading and trailing edge s respectively. ~h, % are
given, by the equations
x - f ( . l ) = /~(Y- Vl),
x - f(~/e) = / ~ ( % - y ) . (.17)
4
"ql, % are the extreme limits of integration for ~/. Intermediate limits (due to the bounding of the
region of integration by other leading or trailing edges) are given by similar equations with the
appropriate functions replacing f071), f0%).
Reduction formulae for the calculation of the integrals X~, 1 are given in Appendix I.
(2) If
ACv = Go(~:) + £ [rl+G,~(~)], (18)
(19)
where
P
x-~
O?"
~1(~) = Y 5 >~f-l(~:);
~<x
~(8) = f - l ( ~ ) .< y + _ _
5
OY ~]2(~) = Y + ~<f-*(~);
where ~/ = f S ( ~ ) is the equation of the leading edge. (For a yawed wing ~h or % might be a point
on a trailing edge.)
Reduction formulae for the calculation of the integrals Y~, 2 are given in Appendix I.
(3) Uniform (or variable) chord loading and load varying linearly along the chord:
= A B{~ - f(r/)}
g(r/) - f ( r l ) ' (22)
where ~ = f0?), ~ = gO7) are the equations of the leading and trailing edges respectively and A, B
are constants (or functions of ~7).
This is a special case of (1) with
that is by the points of intersection of the fore Mach lines through the point (x, y) with the leading
and trailing edges respectively. The integrands in Equation (26) are of the form
P(x' ~){(x-~)~
( y _ -~)~/3Z(Y-~'l)~}l/2 + ~ H(~) cosh-~ l /3 x~ -- -~ l7 I '
(30)
where ~ = f(~7) or g(7), and P(x, 7) is a function of x and 7, and thus become singular along the
line 7 = Y in the region of integration.
Methods for dealing with these singularities and those which occur at the leading and trailing
edges are given in Section 5.
3. Formulae .for Calculating the Shape of the Camber Surface of Two Particular Swept-back Wings
with Uniform (or Variable) Chord Loading, and Load Varying Linearly along the Chord. Formulae
for calculating the shape of the camber surface of two particular swept-back wings (see Refs. 5, 6),
with load varying linearly along the chord, are given.
For each w i n g : -
The equation of the leading edge is given by ~ = f(7), and the equation of the trailing edge by
T h e loading coefficient is
B(~ - f(w)} (31)
ACp=A g(~) - f 0 7 ) '
where A, B are constants.
(The formulae given below also apply if A, B are functions of %)
All lengths are measured in semi-span lengths.
Formulae for calculating the gradient, 3z/3x, for two swept-back wings, are given below. T h e
ordinate, z, is calculated from Equation (11) or (12).
(a) Swept-back wing with partly curved subsonic leading edges and straight subsonic trailing
edges (Fig. 1):
T h e plan-form of the wing is given by
o ~<. I~1 ~< ½, f07) = k 1~7l; }
~<~ Iwl<-a,f(w)=kl~I+c[1-{2(1-lwl)}~¢V;, g(7) = c + k l T I . (32)
The length of the root chord is c and the leading edge sweep-back angle is tan-lk.
T h e streamwise gradient of the camber surface at the point (x, y, z) is given by (y > 0)
P
and '
f ~2F(x, Y, u O d ~ = (G~,~)I, ifx /3y~< C; (33)
Region 0~[~
½<y<l
0<y<½
4. Approximate Formulae for the Calculation of the Ordinates of the Camber Surfaces of Wings with
Subsonic .Leading Edges and Subsonic or Supersonic Trailing Edges, with Uniform Chord Loading.
An approximate method for the calculation of the ordinates of the camber surface of a wing of any
plan-form with subsonic leading edges and any trailing edge is suggested, whereby curved leading
and trailing edges are approximated by polygons. It is also useful in some cases to use an approximate
formula for the prescribed load distribution, so that formulae for the gradients and ordinates at
points on the wing can be obtained in closed form.
The formulae can be used for calculations on a desk machine or for Interpretive Scheme calcula-
tions on a D E U C E computer. This method was used for some preliminary calculations for both the
swept-back wing and the M-wing mentioned in Section 3; these served as a partial check on the
D E U C E calculations for which the integrals given in that section were programmed.
The resulting approximate formulae for any wing with prescribed uniform chord loading are
given below. Similar formulae could be derived for other loadings.
Using the notation of Section 3, the loading coefficient is
[Alternatively, circumscribing polygons could be taken, the points of tangency being (say) at
y = ½(y~+~+yr).]
Writing
1
= 3(y-y,,)/(*-=,,), u~ = c°sh-1/l'~--~-' (44)
l+Pr 1 -t- Qr
p,,= ½log l - P , . ' q,. = ½ log [ ~ Q , . , (45)
the formula for the ordinate z at the point (x, y) of the plan-form can be written in the form
= +
fi [ B (x-xr)2R2(t%) 1 , (46)
with xr = f(y~) in the first sum [L], and x r = g(y~) in the second sum [ T]. Rl(tg,.), Rz(~9~) are functions
of t% given below. The summation of each sum is for values of r for which both x,. - fiy~ < x - [3y
and x~ + fiy~ < x + fly.
For points (x, y) on the wing plan-form for which x +_ fly <~g(yr) +_ fly,., or for a wing with all
supersonic trailing edges, the second sum [T] does not appear.
The functions R1, R~ are given by
RI(#,) = (a) + (b) (47)
(1 -va,2) a!2
)r--lUr o ' 2
(d) = ( 3 - 2a,_~,) V'(1 -~,Y) - - - 2 - {(217-*- 1)t~,
- 41r_~,. + 2} + 2 ~/(~,._~-
~- 1) . (1 - ~,_~e,)2pr
Y = 1
~(Yr -l- Y~+l), and ~-
~(y+,+I-yD
2(x- xD
Also the approximation to the local chord might be taken as
If point (x, y) lies on a leading-edge or trailing-edge segment of slope + k;, va~.= ___ l/x,,
ur = cosh-15,., and the last terms in (b) and (e), [(1 - Z~)q~, (1 - ~r)2q,.], tend to zero.
(Point (xr, y,.) on an edge would be taken as a point on segment of slope hr_~. )
10
If Ar_l.= - 2~ and cr_ 1 = c~
For the swept-back wing (Figs. 1, 2) for which g(~) = c + k ]7 I and, for 0 < ~ ~< ½,f(~) = h ]~ ],
for all points (x, y) on the wing such that x + py 4 ½(k+p), (i.e., all points upstream of the after
Mach lines from leading-edge points y = + ½) the exact solution of the linearised equation is
obtained from (46) with (50)-(53) by putting r = 0, x 0 = Y0 = 0, Yl = ½, k,. = k = - k ~ _ l in
sum ILl when x + fly < c, and in both sums [L], [7"] when x + fly > c.
It is not suggested that it would be better to use approximate formulae than to use the integral
forms, except perhaps for parts of the wing where the approximate formulae become exact. T h e
above formulae were used to obtain some preliminary results fairly quickly before a D E U C E
programme had been made. T h e y were also found useful later for checking.
5. The Numerical Evaluation of some Singular Integrals. Analytical methods for evaluating the
'finite part' of an integral, the 'Cauchy principal part' and the 'generalised principal part' of an
integral are well known (see Refs. 3, 4 for example) and details will not be given here. For purposes
of reference, definitions and relevant formulae are given in Appendix II.
For the integrals in Equation (7) in Section 2, one of the usual methods for evaluating the finite
part of an integral can be used (see Appendix II). Equations (6) are derived from (7) by an integration
by parts and evaluation of the finite part.
Integrals discussed here are those which arise from Equations (6):
(1) when integration with respect to ~ is performed analytically;
(2) when integration with respect to ~/is performed analytically;
(3) when the double integral is evaluated numerically;
(4) when po!nt (x, y) is on a leading or trailing edge.
11
(1) I f ACp(~, 7) is of such a f o r m that a first integration w i t h respect to ~ can be p e r f o r m e d
analytically, in general, integrals of the f o r m
P
~'~ P(~, 7) { ( ~ - F(7)) ~ - ~ ( Y - 7)~F ~
or
f ,11 (y _ 7) ~ d7
have to be evaluated numerically. ~ = F ( 7 ) is the equation of a leading or trailing edge, and the
functions P(x, 7), F(7) are continuous in the range 7t ~< ~? ~ 72.
F o r the numerical evaluation of these integrals w h e n 7~ < Y < 72, w r i t e
P P
?f°r; ~ll
=
~ql
+
d b
+
a
,
w h e r e a, b are suitably chosen, and 71 < a < y < b < 72. (It is usually not convenient to take a = 71
or b = 7~, one reason being that the integrand m a y diverge for a different reason near 7 = 71 or 72.)
T h e integrands in the first and second integrals are finite and cause no difficulty. T h e third integral
can be put in the f o r m
P
d7
1 1
b-y
P(x, b) . R(~, y, b) - - - - -
y-a
P(x, a) . R(~, y, ~)
b-y fo P(x,
+ [P'(x, y) ( x - F(y)) - P(x, y). F'(y)] log - -
y-a - 3~ n) d7
. R(~, y, 7)
where R(x, y, 7) -- { ( x - F ( 7 ) ) ~ - fi2(y_ 7)e}v~, and the dash indicates differentiation with respect
to 7 or y.
T h e integrands of the three integrals in (54) are finite at 7 = Y provided that P(x, ~7) and P'(x, 7)
exist and are single-valued at 7 = Y. A formula similar to (54) has been used in a D E U C E p r o g r a m m e .
Another f o r m (useful if the expansion converges sufficiently rapidly near 7 = Y) can be obtained
b y expanding R(x, y, 7) as a p o w e r series in ( y - ,/). T h u s
P
(55)
- . ( y - 7) ~ ~ ( x - F(7)) ~ - g ( x - F(7)) ' - "
12
The first integral
P P
,~b P(x, ~ ) ( x - F ( ~ ) ) b p1(x ' 7) .
3o (y- v)~ f
is equal to
Pl(x, b) Pl(x, a) b- y ( ~ P~'(x, ~1) - PI'( x, Y)
- - + P~'(x, y ) log &7. (56)
b-y y-a y a .I Y ~7
The integrands in (56) and the remaining terms of the series in (55) are finite or zero at ~7 = Y-
The numerical evaluation of the integral
P
b cosh -1 x - F07 )
f
is fairly simple (provided that ACp and Fi~ ) are suitably chosen). One method is to write the
integral in the form
P
f x Q(x, y, ~) d~
o 3--?
occur, the function Q(x, y, ~) being continuous in the range 0 ~< ~ ~< x, and generally (but not
always) of such a form that the integral converges near ~ = x. The treatment of this integral depends
on the particular forms of ACp(~:, ~1) and F(~?).
(3) If the double integral in Equation (6) is evaluated n u m e r i c a l l y : -
The double integral to be evaluated is
P P
P P {(~-
13
(82476) B
~1 = f ( v ) and ~ = x - fl [ y - ~2 [ or g(v), and the first integral is of the form
P
P(x, Y, ~1) . R(x, Y, ~)
f~ dn, w < Y < ~2,
~ ( Y - ~)~
the same form as given in (54) and (55).
The integrand of the second integral in (58) is finite in the region of integration except near
r] = y, (ACp(~, .q) being suitably chosen). For integration over a strip a <~~7 <<.b, the integral
could be written in the form
ly-vl
I{(x-~)2 - [32(Y-~7)2}aI2 ~i ACp( ~, ~) - (x-~) ~ A Cp(~, ~7)1
(y _ ~)~ d~ &
where ~:1 = f(a) or f ( b ) (whichever is least), and in the second integral a(~), b(~) vary over part of
the range ~1 ~< ~'~ x for which the integral is to be evaluated. Before numerical evaluation, the
second integral in (59) should be put in a form similar to that given in (54).
(4) W h e n the point (x, y) is on a leading or trailing edge (Equation: ~ = F(~))), x = F(y) and
~1 or % is equal to y; ~1, ~2 are given by
x - F(nr) = 5 I s - I r= 1,2.
f ~2P(x, ~)
~Jl
cosh -1 x - F(~7) d~/
-~-~
remain finite provided that P(x, ~) is finite in the region of integration. But integrals of the form
P
,,2 P(x, ~) {(x - F(~/)) ~ - fi2(y _ 71)2}112
f
become infinite unless ACt,(~, ~?) is so chosen that
exists.
If this limit does not exist, the integral has a logarithmic singularity when ~7, = y, although, in
general, an analytical integration of the singular expression with respect to x is possible (and finite)
at x = F(y). This causes some difficulty when numerical methods are used, especially with respect
to the camber shape near the wing tips.
Extrapolation near leading and trailing edges has been used in some existing D E U C E programmes.
This is not very satisfactory. A programme for calculating the shape of the surface at the wing tips,
where the distances between leading and trailing edges are small is then not possible. It may not
even be possible to obtain the shape near the tip' by extrapolation, since the ordinates extrapolated
along different lines through the tip may not converge to a unique point there.
14
Some research on methods of programming for such singularities could b e done. One method of
dealing with the difficulty is suggested below. (The method is given for an integration with respect
to ~7, but a similar method could be used for numerical calculation of the double integral.)
If the point (x, y) is on an edge ~ = F(7), so that the limit % = y, writing
(for small local chord, e.g., near the wing tips, a could be taken equal to 71), the contribution to
the ordinate z at point (x, y) from an integral of the form
P
p(x, 7). R(x, y, 7)
f
,,2
d7
could be written
m z = ( m Z ) l -~- ( A z ) ~ ,
where
(AZ) 1 = P(x, 7) { ( x - F(7)) 2 - fi=(y - 7)e}1'=
;U
(60)
and
(A~)~ = _ {(F'(y))~ -~)~J~ limit
x---> F ( y )
*o P(~', y) Q'. y-
x
~/&'. (61)
T h e two integrals in (60) should cause no difficulty, and (61) can be put in the form:
It must be remembered that in the integrand, % - 72(x', y), a function of x', y, given by
The integrand in (62) converges at x' = F(y) and it should be possible to programme (62) for
D E U C E , or other electronic computers , and thus avoid the need for extrapolation near the edges
and a certain amount of guessing at the wing tips.
15
(82476) B2
The formula for the total ordinate z at a point (x, y) on an edge would be of the form
z = zo + dx + Z (Az), (64)
x0
where Ozl/ax is the contribution to the gradient from other integrals which can be evaluated at the
edge, and x0, z 0 are arbitrary constants or functions of y.
A similar formula could be derived if ~/1 = Y.
6. Conclusion. A method has been given for calculating the shapes of camber surfaces of
swept-back wings of arbitrary plan-form, with subsonic leading edges and specified load distribution,
with particular reference to some of the difficulties encountered in the numerical integrations. The
method is primarily intended for programming for calculations on an automatic digital computer.
It is obviously not possible to produce perfectly general formulae or programmes (beyond the basic
integrals from which all such formulae would be derived) to suit all plan-forms and all load
distributions. The particular methods used must depend partly on the type of load distribution
chosen and (to a lesser degree) on the plan-form of the wing.
16
NOTATION
(ao,0r
f ~bF(x, y, ~7, ur)&7, r = 1 or 2
P,
I+P,.
P, = } log
1+ Q,
q, - ½ log
M
It - 213 (x-fly) (a characteristic co-ordinate)
It = cosh-1
V Free-stream velocity
v,
_ 1 [~,~- 1~,~
A, t ~ ]
I+V,
~, = ½ log
Wr
{;~,.~- 1~,~
l+W,
w~ = ~ log
cos-15(y- 7)
x-~
fi(Y - y D / ( x - ~ )
lk, l//~
Chordwise co-ordinate--of. (l), etc. (variable of integration)
Normal co-ordinate--of. (1), etc. (variable of integration)
Spanwise co-ordinate--of. (1), etc. (variable of integration)
Region of wing surface for which
, ( x - ~)~ - ~ [ ( y - ~)~ + ( ~ - ~)~]/> 0, x - ¢ > 0
Velocity potential
f~ Acceleration potential
19
REFERENCES
No. Author Title, etc.
1 M . A . Heaslet, H. Lomax and Volterra's solution of the wave equation as applied to three-
A. L. Jones dimensional supersonic airfoil problems.
N.A.C.A. Report No. 889. September, 1947.
2 M . A . Heaslet and H. Lomax.. The use of source-sink and doublet distributions extended to the
solution of boundary values problems in supersonic flow.
N.A.C.A. Report No. 900. 1948.
H. Lomax, M. A. Heaslet and Integrals and integral equations in linearized wing theory.
F. B. Fuller N.A.C.A. Report No. 1054.
Supersedes N.A.C.A. Tech. Note 2252. 1951.
4 M.A. Heaslet and H. Lomax.. High speed aerodynamics and jet propulsion. Vol. VI--General
theory of high speed aerodynamics.
(Ed. Sears), Section D, Chapter 3.
Oxford University Press. 1955.
6 J.A. Bagley and J. A. Beasley .. The shapes and lift-dependent drags of some swept-back wings
designed for M 0 = 1.2.
C.P. 512. June, 1959.
7 J. Weber . . . . . . The shape of the centre part of a sweptback wing with a required
load distribution.
R. & M. 3098. May, 1957.
8 Prof. I. C. Cooke .... The centre section shape of swept tapered wings with linear chord,
wise load distribution.
C.P. 470. September, 1958.
20
APPENDIX I
Evaluation of the integrals:
(i)
P
(ii)
' J~l(~) ( Y - n)~ { ( x - ~)~ - 5 ~ ( y - ~)~}~
(i) Writing
For r = O:
nX,~,o = (2n - 1 ) x X , ~ _ ~ , o - ( n - 1) {x2 - f i 2 ( y _ ~)2} X~-2, o
- x o,o + I, (68)
where
x-~
u = cosh -1 fi(-~-~)l "
Also
= /3
[ 1"
tan 0 , (72)
=- y Y o , o - ~
1 (x-~) [l"
sinO ~ol , (75)
11 /3(y-,)l.~ 1
Y00, = /~ COS-1 X-- ~ -]'ql ~ fl (02-- 01)' (76)
where
0 = cos-~ 5(y- 7)
22
APPENDIX II ,
succeeding operations. In using the symbol I f all singularities for which the order of integration
Iv
is irreversible are excluded from the area of integration and are treated separately. This symbol is
not used in this report.
By definition:
a (Y _ ~ + 1 1 2 g ~ -- (2n) ! ~Y ,o a (Y -- 7) 1/2
n~>l.
A sufficient condition for convergence is that the function A(~) is continuous at ~ = y and
integrable elsewhere in the region of integration. ..
In particular,
-2
n/> 0. (79)
(2n - 1)(y - a)'~-Y~
Also
f l i . A(y,
~ 1 , 7) ~d,7 - ( - (2n),
1)~2=~n ! (Uy)
a '~ j~(y--~
(~' A(y, 7)d~, (80)
provided that
limit ( y - 7)1I~ A(y, 7) = O.
r/--+y
Writing
23
it can be shown that
Similarly
dr/= dr/
o ( s - r/P+~= Jo ~-r/~=
+ 52" [-(- 1)"-~'+* 2 1 ! (82)
,.=o L ( n - r ) ! (2r-1) A('~-r)(Y'Y)(y-a) r-lt2J '
where
C(y;y, r/) =
-
52 i-(-
L~.1), A~(y, y) (y- r/),]
and A(')(y, y) denotes
E(~ A(y, r/)
1 .
In both (81) and (82) the integrand on the right-hand side is finite at r / = y.
•
It is also useful to remember that if the indefinite f f ( Y , r/) .ar/ - F(y, r/) exists, and
integral j(y_~)741,2
if a < y < b and f(y, r/) is real in the interval a <~ r~ <~ b, then
The methods shown above can also be used to evaluate an integral of the form
( A(~, y, ~, r/)d~dr/
f
where the region of integration is defined by ( x - ~)~ > fi=(y - r/)=.
By definition:
P
f[ A(~) d r l - 1 {~]n+* b
(r/--~+* n ! kay] f ~ A(r/) log Is - r/] dr/
1 ( 3 ) ~ fo A(W),
= 7., ~5 Y ° ~ - y "
(84)
a<y<b, n>0.
24
W h e n n = O, the above integral becomes the Cauchy principal part
fb nA(*1)
Y° - y d*1 = a
ay fba A(n) J+og I n - y l d n . (85)
For convergence it is sufficient to assume that A(*1) and its first n derivatives exist and are single-
valued at ~ = y, and that elsewhere, A(*1) is continuous or with integrable singularities.
In particular
P
--- 1
I/+ I-(bYy)'+ (aYy)¢+1 n/> ]. (86)
Also
p
provided that A(y, .1) and its first n derivatives with respect to y and .1 exist at .1 = y.
Writing
P p
(88)
-,.~=1+ [.~Z~
r+++-+>+
+r (b-Y) + ' +I
(a-Y) +
Similarly
P
A('~)(y, y) b- y
f+° (.1-y)++~'+.1
A(y, ,+) ~ = F ° A(y, ~.1)-
- y ~(s;
~ y, .1) d*1 + - -n! log - -
y - a
+ FA(~-0(Y-'.1) 1(b-y)
- ,.=~E k (n-r)! r
1 + °
In both (88) and (89) the integrand on the right-hand side is finite at ~/ = y.
25
o
fJ
/"
/ /
%%,
ta~-'k
,%,%,%'N %~
/
J %,%,
J
7/./7 y
/ 7
f
J
J
'~.. /3@= k/~,~,='/*)
-%
J
7
f
f
/ \ %,% /J"
\
%%
Z "~. /
/ / %./"
//. 7%....
J *
IN "-..\ %,%,%.
=C
×'5"\'-% NN~ \ .-" I.~'t(x~+l;%.f)
kr.+)
x-Xt~:
Co&),
• °l At~.. ~ ..q,~
('~k,,o ,~-,.. 6
F"
*,5
FIG. 4. M-wing (b), showing the different regions.
27
(82476) 'J[L 66]999 K.5 9/61 Hw.
Publications of the
Aeronautical Research Council
Special Volumes
Vol. I. Aero and Hydrodynamics, Aerofoil*, Controls, Flutter, Kites, Parachutes, Performance, Propulsion,
Stability. x26s. (post 2s. 6d.)
Vol. II. Aero and Hydrodynamics, Aerofoils, Airscrews, Controls, Flutter, Materials, Miscellaneous, Parachutes,
Propulsion, Stability, Structures. x47s. (post 2s. 6d.)
Vol. III. Aero and Hydrodynamics, Aerofoils, Airscrews, Controls, Flutter, Kites, Miscellaneous, Parachutes,
Propulsion, Seaplanes, Stability, Structures, Test Equipment. x895. (post 3s. 3d.)
Index to all Reports and Memoranda published in the Annual Technical Reports
I9o9-x947 R. & M. 260o 6s. (post 2d.)