C.P.

No 72
(8549)
A R.C Technical Report

MINISTRY OP SUPPLY

-AERONAUTICAL RESEARCH COUNCIL

- CURRENT PAPERS

Approximate Two-dimensional Aerofoil Theory.

Part v. The Positions of Maximum Velocity

and Theoretical CL Ranges

S. GoldsteIn. F R.S

LONDON HER MAJESTY’S STATIONERY OFFICE

1952
Price 5s. 6d net.
 NAnOt$AL AERONAUTICAL ESTABLISH-
LUBRWARY

Approhmte' 'hm-Disensiond. Aerofoil Theory.

The.Bsitions of Nxcimm Velocity and Theoretical CL-Ranges )I
-By-
5. Goldstein, fid4.S.

1. We may find the positions of maximum velocity on the upper surf'ace

with suf'fzoient preclszon for practlc.1 purposes by the following rules.
When ( CL - s optI , 1s not large, so;;e the equation
(
>-CL= 2eL cotO + sin*O[gA (V) t gi (d)]
I II 1
t- I + -- CL opf. (1 + cc=@!) } /[$ + E.Z.] . . . .. (1)
2 i a0 21t 1
by plotkng the right-hand side agamst 0 snd readmg off the values of E
.for which it 1s equal to specified values of CL; or, if a0 = 2ri, solve
the equation

CL - CL opt
------- =
*CL -------
, ;l,r!? + [1 - cos ij [ gp) + gp 1-J
2rr
. . .. . (2)

similarly by plotting the right-hsnnd side and rending off the values of 0
for which it is equal to (CL - CL .pt)/2H. If the right-hand side stays
practically constant over a considerable range of values of 8 , then for the
corresponding value of CL we ha+ a 'flat' maximum, which we do not attempt
to locate kqth any precxsion. As CL increases, after a certain stage @
becomes smaller. ihen A, defined by

I..... (3)

is large compared with ' 2oL, al-113 $ large compared with

i3 is given simply by
 -2n

If, however, we proceed to very large values of CL - CL opt and very dl

values of 0 ) we must change the definition of A to

. .... (6)

for igntrc lines without singularities at & = '0 (gi(0) = 0), 8 is then
given by

\ 0 = 2PI/A - J (2p~)yt(o); ,......... (7)

for centre lines with singularities at Q= 0 (gi(0) f 0) 8is given by
substituting 2PT/h for 0 in the right-hand side of the equation

v2(d) - A’+‘(@)~‘(O)
0 = ------------------.-, I.... (8)
I+ yJ (8) ‘P(O)
snd then proceeding by successive approximation if necessary. In thcsc last
casts, nowever, mess ~ounaary layer suczlon 1s employed, restricted boundary
layer separation will probably modify the theoretical pressure distribution very
considerably oven if the aerofoil is not stalled, so the results will not have
much practlcsl significance at present.

The equations above also apply to the lower surface if we remember that
8 is negative, gsr *Vsr gl.9 Yvb cvcn functions of B, snd g;, y',
'Y 0 odd functions of 0 . It is, however, convenient always to Cons=.l cr "if
t ese
functzons in the range 0~ 8 g t7' and WC may da so if we make the follol
changes in the equations. In (I) change the Sims of 2pL cot 0 .and sin 3 ogi(6)
In (2) change the signs of CL - CL opt and of g!(Q)). C2mnge the sign of
CL - CL opt in the definition of h in (3): and the sign of gi in (4). Then
is unoltercd. Change tho sign of the right-hand side of equation (6),
%ning h; then (7) and (8) <sr?reunaltered, but, whereas on the upper surface,

-Y = Ys(f ) +yo(E), -qJ' = v;(O) + yp), ..... (9)

for the lower surface we must take

il = Y,(O) -y&(S), y" = $p) - y;(B). . . . .. (IO)

Those npproxirmte methods have been tested by Mr. E. J. Richard.&, who has
applied them to N.A.C.A. I6 series and Clark Y aerofoils, with satisfactory results
for practical purposes,

2. A discussion is given of possibli definitions of CL-ranges for

low-drag aerofoils. The "thcoretlcsl" CL-range is defined as the range of values
of CL for which the velocity continually increases, on both the upper and lower
surface, from the stagnation point to the &signed position of maximum velocity at
the design CL. (This definition a@ies strictly only when the slopes of the
graphs of gs +, gi are disoontinuous at the design position of maximum velocity,
which is the case for low-drng nerofoils as now &signed; if the graphs of gs !: gi
are rounded off, WC should require the velocity to increase only up to thebeginning (
the rounaiup-off).

To obtain a CL-range of <any practiwlly significant size, g;(6), wh.Gzh

is small coward \mth 1, must be positive =s.ndlarge compared w.th 2@1,, except
perhnps for smn'l values of 8. For a0 = 2n(, best results NTC obtzincd by
Iding g!(B) = 0, and for practical values of ao, this theorem remnins
practicdy correct.
If/
 -3-

If, at the relcvsnt value of 0, g&(o) + g:(c) are large enough con~ared
with 2eL for

to have minims for small vnluos of 0, then the Ci-rango is given, quite
w-W, by

. . .. . (11)

except that, for highly cnmbered thin aerofoils it seems worth while, having
found thL positions of the maximum snd minimum of the expressions above, to
substltuto the vducs of. !j2 at those positions for 2pL in order to find the
values. 'ybr,c generally, end with no very.grent accuraoy,'if a0 = 2'Tf me have
thceonditidnstk~ t;~(CL-CLopt)~tMtexccedthc~~vslvcsof

cot E
2pL ;l-,,o t (1 - DOSE)(gA +, g;, ..... (12)

rcspeotively; for a0 f 277, WChave the m3rc elaborate conditions (15) of

the text.
'
If gi b 0, the middle of the CL-range is displaced from
CL = CL opt*
For the keof-to-o' ‘ncrofoils of RB6 and. 7 of Ref.5 (c$/dx, = s = oonstsLnt
for 0~s x&X), with c&xc-lines dcnigncd for constant .smroxirdo landing for
O<x<X(J&/dX = 0 for 06x&X), t' hc: CL-range is given by ,

( )
-
a0
1
+ --
1
2?f

WC suppose the positlon of dmum vclooity,

I CL -.C, Apt 6 1.4756

X, fixed, rind also hj(2pT)

(2pL)z

fixed.
&. ..... (13)

[ For a cuspcd aerofoil J (2 T) = 0; it till. in so case be small; O.CY+

would be n large value.] I in addition the mtimwn velocity is given (theoretical
criticd.LBch nunbcr given), or the thieknoss for a given value of x, or the '
maximum thickness, then there is a value of s lrhich mskds the CL-rengc a
maxim- (For the oases when the maximum velocity is given, or the thickness for
a given x, see the cxmples in the text.) In particular, when the m;udmum
thiclcnoscrchord ratio, t, is given, the value of s mny be found from the
fonmiLacl
s = 1.25 t - o.25J(2pT) for x = 0.4 -
= 0.9818 t - 0i2fd.+J(2pT) for X = 0.5 I . . (14).

= 0.8095 t - O.l72,/(2p~) for 'X = 0.6

With/
 -4-
IWith normal values of t snd J((2PT), the values of s so found are probably
large enough for a stage to have been reached when any practicable irlorease in
s would not, in any case, mahe sny practical difference to the tolerance for
imviness of the surface. (An increase of the velocity gradient by a large factor
would probably mske a difference, which is why we expect the tolerance to be
greater near the nose than elsewhere)
Oorrospcnding to the above formulae for s, we have the following formulae
for the parameters a, b, 0 needed to find tho ordinates of the fairing5.
For X = 0.4,
a I 1.0125t- 0.3 ,'(2(3T), b = 1.5125 t -
-0 = 0.9085 t -
for X -ii 0.5,
a = 009453t- 0.208 ,,/@{T), b = 1.4362 t -
-0 zz 1.0319 t -
for x = O-6,

a z 0.8908 t - 0.156 ti;(2pT), b = 1.3765 t -

-c = 1.2121 t -

Tables of the m&mum CL-ranges of these aemfoils are given for thicknesses
between 8 end 22 per cent, for A(2pT) = 0 aud 0.02. For hj(2pT) = 0,
the CL-range is m%.rly proportional to t7/4.

1. Introduction
To calculate the theoretical critic&l compressibility speed for a given
acrofoil section af a given CL we first compute the grcatost velocity on tho
aorofoil contour. This greatest velocity may be found. by graphical or numerical
methods from a graph or table of the velocity distribution, but when we require the
answers for a nunitxr of aorofoils over a range of CL-values, such a method is,
long and laborious, and has been found in practice in sort cases to be prohibitively
long. Some simplification is, therefore, nooessnry. It appears that it is j
possible to find s&ple formulae for the positions of the volo&ity maxima; sqd
formulae, though r&or crude, scem to be sufficiently accurate for practical )
pu_rEssos. The act&i maximum values of the velocity ore then easily computeq,
by Appro-tion III, for the v?Jucs of , x and ,o so found? end for those values
only* If there ore two maxima for any CL, both must be computed and the l,yger
chosen (unless wo luvw bcforchnnd which will bc the larger).
To oonptko tho mnxLmun vnlucs of the velocity no great accuracy is
necess‘ary an oaloulating their positions, since a smell error in the position s
produces a second-order error in the value. As \cL( increases, however, a peak
in the velocity graph develops near the loading edge of the aerofoil - on the
upper surface for CL positive and on the lowor surface for CL negative - and
the absolute error in the calculated vnluc of &' at the nm.xbum must be small if
we <sre to avoid the possibility of fairly large errors in the calculated maxirmr
VgLuo. (T'nepcrccnta~ crroc need not be very smQ1, since G itself is &&.l).
In other words, spocMl. attention must be paid to the nose of the aorofoil
small).
(b\
The/
-- -
*x is the distsncc, perallcl to tho chord, of a point on the aorofoil surface,
measured as a fraction of the length of the chord, ‘and x = $(I - cos @)j
O,< e < Tf on the upper surface, 0 2 0 >-77' on the lowor surface.
 -5-
The positions ofthe maxima being calculated from an approxlmateformula,
$? da/d@ (where q is the velocity) remains small over a large range of 8
in the neighbourhood of the maximum, our determination of its position will be
subject to quite considerable error. This error, however, will not be wortsnt
in the caloulation of the anximum v&he of q, since the graph of. q till vary
slowly over the whole range of f: in question.

The determination of the position of the maximum velocity may also be

considered to give some information on the probable position of transition-to
turbulence in the boundary layer. Care will be needed, however, in using this
information. Other factors in addition to the velocity distribution (waviness
and roughness of the surface, turbulence in the air, Reynolds number) affect the
position of transition, and the effect of the velocity distribution, end its
interaction with the otiier factors, do not depend solely on the position of the
maximum velocity. We do know that if the velocity fslls off steadily snd not
too slowly after the maximum, then in practice, at high Reynolds nunicers,
transition will not be delayed to any appreciable extent beyond the maximum,
Whether transition will occur before the maximum will dcpend on the velocity
gradient, the state of the surface, the twbuloncc in the air and the Reynolds
nunberT In certain circumstances, also, transition may occur well after the
velocity maximum - for example, with a good surface and low turbulence in the air,
if the msxkmnn is followed by a small fall in velocity and the velocity begins t
rise again,(Big.l), or, at Reynolds numbers which arc not too large (106 to 2.10 8 ),
if the msximum is very "flat". The former state of affairs (Pig.1) alrplics near
the nose of a good many aerofoils at certain values of CL. To sum up we may say
that, for the purposes of discussing tho probable position of transition, a rough

calculation of the positions of msximum velocity may be of some restricted use for
a preliminary vsorting-out", but much more will be necessary for sn acrofoil which
it is proposed to study,in any detail, so only very.rough c&xitat~ons of the
/ positions of the me&ma will bc needed.; inparticular, if q is varying very
slowly, then that is itself probably all wc need to know - the exact position of
the meximum in a "flat" portion of the graph of q is not of sny practical
interest.

Closely oonncotod with the questaon of the positions of the velocity

maxima is the discussion of the "theoretical" ~-ranges for low-drag aorofoils
according to a definition we have used for some time now, the "theoretical"
CL-rsnge being dofined as the ramp of values of CL for which the velocity
continually increases, on both surfaces, from the stagnation point to the designed
position of maximum velocity at the design CL?* A slight oxtension of the

analysis/
--se-- F--.- e-m
%or some further.rcmerks on this subject, see B6.

9~3 low-drag aorofoils are design4 at present, the slopes of the graphs of the
velocity on Approximation I (i.e. the slopes of the graphs of gs 2 gi) are
discontinuous at the design position of .maximum velocity, and the definition
given applies strictly only to such 0asC.s~ If the graphs of gs 2 gi are
round& off, WC shouJA require the velocity to increase only up to tho beginning
of the rounding off.
 -6-

analysis for finding the positions of the velocity maxima enables us to discuss
these "theoretical" CL-ranges, but before we proceed to the analysis, a short
discussion of definitions of CD-range may not be out of place. The general
notion is that the CD-rsnge is the range of vslues of CD for which the aerofeil
drag stays low and praotically constant. Such a statement is, however, too
vague to serve as a definition until the ciroumstanoes are specified in whioh
the drag is to be determined. One such definition is the range of values of
C$ for which the drag stays low and constant when measured. on models made as
carefully as possible and tested in a 1owLturbulence wind-tunnel at a given
Reynolds number, say 30 x 106,~ This %ind-tunnel" definition would certainly
be useful if a suitable tunnel and model-making facilities were available. The
third definition - the "praoticsl" definition - is the lnost probable range under
practioal conditions of msnufacture, flight and maintononoe; values according to
this definition are the values we should all like to be able to give but none of
us can. It is probable that In future all three definitions will be used.. Our
"theoreticsit" definition will be of use r0r.a preliminary "sorting-out". Its
main disadvantage is that it tskes no account of the nature of the velocity
curve after the maximum; if, for example, there is only a small Gill in velooity
after the rmximum (Pig.l), it does not take into scoount the magnitude of the
velocity gradient thereafter, which may be fairly lsrgc in some oases and
practically zero in others. As soon as really good surfaces become practicable,
and a small velocity followed by a ris o does not neoessnrily lead to transition,
I this point till have to be borne in mind. b%oenwhile, it is doubtful if, at
* present, it would be possible in practice to delay transition in this way, SO
aerofoils may porhaps be expected to be in the same "order of merit" as regards
CL-rangeswhether srrsnged according to the "themretical" or "practical" definition.
In fact, with present surfaces, the "theoretical" definition may be nonrer to the
d "practical." one than the "wind-tunnel" definition would be?"
2. The Approximate Calculation of the Positions of Maximum Velocity on the
Upper Sudace
We wish to use the simplest possible method to calculate the position
(or positions) of maz&num velocoty. The crude, linear Approximation I cannot
be used; q/U is infinite at 6 = 0 except for CL = CD oat
to Approximation I. We therefore use Approximation II, accordang
on the upper surface,

9 1 + $qFj 1 cost
----__---- (1 + g, + ii) sin6 + CD - + m-w-
U = (Y2 + siIPi$ i _ 271 a0 )

11 1
-- em+- 27~ c, opt (1 + OJs b) 9
2 c a0 ) \ so/
?Xis "wind-tunnel" definition may be that adopted by Amoriosn workers on the
subJect, though, if so, It is not ,clear which Re olds number, if any, they
adopt as stsndsrd. Jacobs, Abbott and Davidson F , however, write of the
C -range as the range "over whioh the pressure distribution rcmclins favorabla",
wk,ich suggests that they adopt the definition of what we c‘all the "theoretical"
CD-range, though no exaot definition is given of when a pressure distribution
is favourable. It IS of interest that J&s. Moore, in working out2 the velocity
distribution on N.A.C.A.66, 2-015, cals~ worked out the theoretical C&e,
.snd found it to be a.166 (l?ig.2), in place of tho LO.2 ind~atea ir the title
of the aerofoil. This acrofoil is one of an oldor scrux5; it has now been
replaced by N.A.C.A.662 - 015, and it would be of interest to' repeat the
calculations on tho now aerofoil.

**A fourth possible definition, namely the rango of values of CD for which
transition stays at or behind the designed position of the maximum velocity,
has not been included, since it is still not possible to calculate the position
of transition, and the drag is easier to measure.
 -7-
so

a 9 I+@.
20
-a --- Sin'J (l+J2 + sin2 B)(gA + gi)
de
0u = 7,V2+sin2Q)3z
i

+ by2 003 5 -'rmq' sinf?)(l + &, + gi) - cL[sb6( ;6"'+$)

, (cd 1
+u;lq.--.-+~ .
I

II 1
I
+- -w + 7' CL opt L sin (3 (I + :y* + oosi,) + i~lp (1 + ooso,]
2 i 2H I
aO )
.. . . . (I)
where the dash denotes differentiation with respect to 0. This expression
is too complioat&d to be of practical use; it must bo simplified by approximation.
Ike t for wry large valves of CL ~0 pry say that the terms of the first order
FI

those of the second order are

and those of the third order are

, .
CL
-II*@ + g:) sinQ + (-VT cosQ - ?fv* sm3)(gs + gi) - -- y* sin Q
a0
1 1 ,I
+- -+- CL opt u*
I
sin J - CT, ( -'OS' + - ' j VW
2 a, 2/H -\a0 277r//"
,( ' )

11 1
+- --+--- CL opt (1 + cosQ)yy'.
2 a0 2ti t
( )

The terms of the first order are all snnl.1 whon 6) is smsll, whereas those of the
scoond and third orders are rat, so it is inmediately clear, as explained in the
introduction, that YB may not simply neglect the,terms of the second and third
orders If inoorrcot results nrc to be avoided for small values of 0. oil the
other hand, as we LSiso explained in the introduction, we abandon tho requirement of
even fair accuracy m the case of a very 'flat' msximum, nn8. thon tho neglect of
the firgt four terms of the third order wdl.d appear to be always justifL?ble.
For 7p '(g' + g') sin 8 still alwsys be small either compared. with sin3 @(gL + g!)
or OO~XKICI~~~~~ ~2 00s 6' - lf li!' sin@; gs + gi will~be smK!A COM@W~
. Mth 1; certaitdy ma 1 compared with 1. It 1s eqwdly clear that,
at any rate for smll 8, the first term of the second order may not bc
neglected;. &en @= 0 this term is simply 2@L, whcrc PL is the radius of
curvnturo of the aerofoil section et the leading cd.@. To hat extent the influence

of/
 - 8-
of the second term of the second order, and of the last two terms of the third
order, may he neglected, is a more difficult question to decide; it seems
probable that the formr, since it is of the second order and is, moreover, small
when 8 is small, may always be neglected, snd that the last two terms of the
third order may be neglected except when CL is very large snd 8 very saall.
We shsll presently give some numerical examples and shall sea that-the above state-
ments Bre correct.
The second snd third order terms are of mst importance when e is smsll,
so io osrry the analysis further for szitl (j neglecting completely the first
four term of the third order. In the tem'of th first az.3 second orders ws
write 1 r^or 00s 8 and G for sinFj; for W2 we writ0

-q2Fo + 2 ~w3)u:‘(O),

and for l+Jl/J' we mite y (O)i/!l(O). In other words we expand tho first snd
second order torms in powers of ti , and keep only the terms~ in 80 snd 01.
In the last two third order tern wc keep only the term in (70. Also, on the
"pper surface,

vJo(o) = 0, $$o = i/W&

Hence as/a6 = 0 on the upper surface for small G when, approximately,
2?L + G'&!pL)Y*(+ - A[6 + J (2eL) V(o)] = 0
where
A = (;+&I) (CL"CLqt)i . . .. . (2)

i.e.
2PL - 4J(2PL)Y'(C)
0 = -I_------* . . . .. (3)
h - J (2p JJ q ’ (0)

In order that a" should be small, the denominator must be large compared
with the nmmrator, i.e.

~-,/(2~~)-‘4”03) 3 ~CL- ‘hJ WL)Y'(O),.

i.e.

h [ll +n/(2~$p&0~ 2 2PL +~Pe3yW,

i.e., sinoe J @PiI V’(O) is small compared with I,

+ 2PL +J(2,?#P(0).

Home oertainly

h $Y,/fqq y'(o),
the sooond term in the denominator may be negloctod, mnd approximately

' 2pL
- J (q IJ $J ' (0) ..... (4.)
8 = -7'
,- Thus/
 -Y-
Thus of the terms of the second order in I I in (I), the second could have been
:I
neglected and the first replaced by its value at 0=0.
Before WC moceed to numerical illustrations I;% must remark that the
last stcl> in our NIR~~SXS is lnvnlid if there is n singularity at B= Oin
the equation of the centre line; such smngula+.tlos occur in certain modcn ccntrc
lines, for khlch &i(O) f 0, the approximate losdinp, bang taken as constmt
from x I 0 either over the who16 chord or over some fract-ion of it from t'ne
leading e&c. As n result both dyo/dx rind the npproximntc vGKluc of
at8 @ = 0, and
may bc, nor
In place of (2) wo have then, to begin lrith,

------_---------------~, . . . . . (5)
h+qJ(Q)Vl'(W
If f3= 0 is the relewnt (small) root of this equation, yc(@o) w-ill
be small com&rcd with ‘\ys(80), as w shrill xc later in numoricnl ex~amplcs,
cm-d lf' s(Qo), rtillbc nearly equal to ys(0), i.e. to d((2pL), so y(oo)
will be of the s‘amo order of magnitude as n/ (2pL), We shall also shw later, by
nuwrical exnmples, that, although V'(s) is 1ognrithmicKLly infinite at
8 = 0, yet, for q.tc sm-lll vnlucs of Qo9 y/'(Q,) for centre lines I-6th
singulnrltics is of about the sync order of m?gnitud? ns for centre lines
with no singul&,tics. It follow th?t, j'ust as wo could!$k+/(2&) '/'l(O)
&thc donormnntor in (3), so VC may neglect v(Q)v/'(@) 3n the denonunntor in
,yl 5(81;cz"p" ;~~i~$~&~, Howick, thz order of ;;?;;l
SO it 1s dl,flcult to forotcl horr lnrec a
pcrcontago~orrkr ~211 bo lnvolvod'ln substituting 2p in the
nmorator of (5). If we do make this substitution, (4) rcducos to

..... (6)

This equation may be solved by succossivo approuimation; for the first approximation
TIC put Q 5 2pJX in I$'(@), so that , ,

... .. (7)

for the second aiT?roximat]on wo nubstltuto from (7) into \y*(B) in (6), ema
so on. The first 6?proximation (7) ~6.11 usually bc sufflcicntly ncctrnto; ln
this fain the cquatlon 3s equally applicable If thcrc is no singularity in
Q= 0, suxc in such CIZSCS, in fact, the difference bekocn (4) cud
is ncglz.giblo. ~reovcr, on the upper surface,
lfJy'(@) = yJ;co, +4@3),
ma, uliLcss as n conscquoncc of singulnritks
in the cquatlo %::;hDi:c:::;;t%:;t $$),"=~~(,~d?~~~~~) ~511 be
negligibly small, so vo my rcplncc such
?].scontinuitlcs occur on N.A.C.A Y'
00 nerofoils. On. N.R.~.h.0012, for'exnmple, the
qqroxirmto vnluc of Ijy' at b= 0 chnngos discontinuously from +0.0378 to
-0.0378 ns VC pass from the lcwcr to the u or surfnco. For such norofoils, J$'
should not bc roplncod by yh m (4), (?cnd (7)?
Smcc/
--- ----____ .----------~~L__--_-___,_-____----_------------------------------------ -----
*On N.A.C.h.0012, the nccurntc vnluo of VI' J.S zero ?.t 8= 0, but rises wry
rqndly to npproxirmto agrccncnt with the values c.dm.iLntcd on the approximate
theory.
VC should rwnt?on that the slngukulty in the cquntion of the ncrofoll contour
which produces ~1 dascontmnulty in vi(@) at @ = 0, also mnkcs g logLarlth-
mically ini'initc at @ = 0. But for N.h.C.I",. 0012, ' 545 ~rhon
Q = O.OJ+ rind 0.3574 when d= 0.02, compared wzth 0.?~331~t0fl$ 2 -14 H.
 - 10 -

Since, for small values of 6 , -q'(G) is larger 2nd y(O) varies

more rapidly for N.A.C.A. fomfigure acrofoils than for any other common symetri-
onl types, for ow numcrioal cxsmplcs WC shall sqposo the aerofoil fairing to have
the shape of N.A.C.A.0012; ad we shsll consider tVio centre lines, one mth, ‘and
onz;tho10, a singularity at Q= 0. For the former we take the pnrnbolic

YC = 0.08 x(1 - x),

whioh corresponds to 2 per cent camber and a CL opt roughly 0.25; for the latter
the centre-line for constant approximate loading,
I ‘ +
yo = - -- xlnx+(l-x)ln(l-x),
16n t J
$~,~cz3ponds to a CL opt of 0.25 and a camber of roughly 1.4 per cent. For

yy, = o.a!+ SinQ, y; P 0.04. cost:;

for the latter

Then J(2pL) = 0.1781. Let us take x = 1/6 to start with, so that,

for a0 = 2d, CL - CL opt = r(/6. !l%en 2PL/A = 0.1903. For this
value of 8,s q&G;) e 0.1715, Y'(F) = -0.0324, end, according to the
formulae above, y. = 0.0076, 0.0393 for the first ocntre line and
'to = 0.0108, -11; = 0 . 0373 f?tL seoond . Hence, on the wer surface,
for the first centre line
y2 = 0.03208, ?f'/A = 0.1925,
Y = 0.1791,
y’ q o.ogY,

Y’y = 0.0012,

and for the second

1 = 0.1823, 'i" = 0.03323, p/x = 0.1994, y" = 0.0049,
'fly' = o.oooy.
be neglected in the denominator of (5),
~~~o~~~e~~o~~~.~yt~~I ccond terms in (4.), (6) snd (7) my be
neglected. We shall, hokever, wish to make the ssme alJproximations for the
lower as for the u-r surface; for the lower surface,
y = 0.1639, y2 = 0.02686, 1+2/x = 0.1612, jy~'l = o.o717,~y/yi~( = 0.0118

for/

+In is used for log,.

 - 11 -
for the first centredz~e, ad

$J = 0.1607, u.2 = o.ozy32, ,q‘2/A = 0.1549, II+,'1 = 0.0697, IyJy'\ = 0.0112,

for the second. Here both lv‘ly' and the difference of y2 from 2p~ may not
be completely negligible. We shall See, however, that the sign of 7y‘\y' is to
be taken as negetive, and the effect of‘Woluding the q?+Jw' terms in (5) is
therefore opposite in sign from the effect of tsking the above values of y2
In fact, if we calculate a second approximation to 0 from
Xz"Kf, 2PSL=* 0.1060 for the first centre line and 9 = 0.1781 for the
;* even In the latter case the percentage difference from the first
approximation 0.1903 (about 63 per cent) is probably tolerable for th'e purposes
we have in. mind. That the effects tend to cancel is not fortuitous; the
difference between yq G) red 2pL is cleverly related to the sign and
magnitude of -V wt.

As a second numerical example we consider the same aerofoil, but double

the value of h, so that the value of G is halved. With hti=1sl/3*
CL - CL opt = rf/3 for a0 = ?I. The first approxlmat,ion to *
2QJh = 0.0952. For this value of F , Vs.(@) 7 0.1747,
y i(o) = -0.0344; for the first centre lint VI, = 0.0038, ‘ly; = 0.0398,
and for the second On the upper surface
7v o = 0.0068, y:, = o.osoa?

-q = 0.1785, y2 = o.ojla6, ?+2/h = 000956, -qJ' = 0.0054, yl+J' = 0.000?6

for the first centre line, and

'/J = 0.1815, y2 = 0.03294, -p/A = 0.0988, l/J' = 0.0164, u/y' = 0.00298

for the second.; on the lower surface
'1' = o.j709, ltJ2 = 0.02921, v2/1, = 0.0876, 1-p 1= 0.0742, pjqJ') = 0.0~266

for the first centre line and .

y = O.-t67y, y2 = 0.02819, W2/A = 0.0846, ( u/'I = 0.0852, lqqJ'/ = O.Ol43f

for tho second. The second approxclmations to 8, according to cqn,(5),

are now 0.0943 and 0.00950 for the upper surface for the first and second centre
lines respcct~vely, and 0.1043 azd 0.1033 for the lower surface. The largest
ycrcent~~c error in the first approximation (about 9 per cent) is now on the
lower surface for the first centre lmc; if we inoludc the second term in (7)
thus error 1s reduced%?&s than h,slf, and the error on the upper surface for the
first centre l~nc is roduccd almost to zero; moreover, this term till ole~~ly
account for nn lncrcnslng fraction of the error as h increases. For the
second centre line, howvcr, computation shows that there is no substantial
ndvnntagc to be gained by including the second term of (7) unless WC also chgc
the first term to 7~~(2pL/h), and then we may as vmll solve (5) by suocessivo
approxmation.
mus/
__-__ --------_ ___l-_-l_l---
%OI- the wper surface these second approximations s.rc 0.1898 for the first
centre lint and 0.1974 for thz second.

+Ewn when 0 is as small as O.Cd+, is 0n1.y 0.0679. Eventually as c'* 0,

the whole basis of our approtimation % o q' will fd because of the
SlrpJbl-lty, ,md WC mst USC more nwrly oxact vnl.ws m (5). But the
nruneric,Q results zLbovc stem to show that such failure will not occur for
any practical vnluc of ?,.
 - 12 -

Thus we see that, ~ti:len @ is small but not very small, a satisfactory
approximation to the position of dmum velocity is given, quite simply, by
0 = 2QI/h ..... ‘(8)

in all cases. @ becomes very small when ( CL - cL ,+.I becomes very large;
in such cases Approximation II itself may not be a satisfactory basis, and we
shall briefly consider the matter later on the basis of Approximation III
(only briefly because the results are of no great practical interest);
meanwhile we note only that for centre lines without singularities at
0), eqn.(7) provides a better answer for very small's
and for such very small values of d we may as well use (4), which
is easier, in place of (7); but for centre lines Nith singularities (gi(0) 4 0)
it is safer to solve (5) by successive approximation,%sing (8) for the
fmst trial value.
We seek next for the simplestequation
" . to solve when @ is not small.
The second and third order terms in J \ in (I) are now small compared with
the first order terms; in order, h%ver, to ensure that the solution for @
should pass fairly smoothly into the value given by2(8) as CL increases
and 0 bcoomes smsll, we must include the term Jo oos 0 of the second order.
For the purpcsc for which it is included, however, WC may approximate to it by
2pL 00s 0; i.e. we neglect terms

(Y2 - 2pL) cos G- y7+Jt sin@

of the second or&r, end. all terms of the third order. The equation as/a@ = O
then becomes, approximately

2OL -2.3s 0 + sin3F [ep, + gJ(F)] + ; ‘+ ;; CLopt sin 6 (1 + 00s 0)

( 0 )
= 0.
Probably this equation till be most often used when it is required to study the
maxima of q over a range of values of CL, and the si@est way to carry
cut the calculation would appwr to be to write

. ..*. (9)

plot the function on the right against 0, and read off the values of G
for which it is equal to specified values of CL. If the function stays
practically constant over a considerable range of values of U, then for the
corresponding CL we have a 'flat' maximum.
We may note that for ac = 2f(, (9) becomes, more si@y, ,

CL= = I----cot Q + Cl - cos 83 [gg 6 1 + gp @11 *

2PL ..... (10)
29-r 1 + oar3

I
As/
.

- 13 -

As CL increases,' after a certain stage @ becomes sms.ller, and when h

(as given by cqn. (3)) is hrge compred with 2F 3 is given approximately
by (0) provided that 13 is also large compared $th

l?indly,we return to the cons am&Son of very large values of

1% " CL opt\ and very small values of' 0 for which it is advisable to use
Approximation III. At these very large v&es of 1 CL - cL optl the velocity
graph will have a very sharp peak ncsr the lending edge; if the aerofoil 1s not
completely sts.llcd, we should expect that at @ny rate a restricted boundary-layer
separation lnll occw and appreciably modify the high theoreticnl peak, utilcss
such separation is prevented by suction. Consequently the theoretical
oalcu.lations cannot bc eqccted to have much practical significanoe at present,
and they h&v0 therefore been relegated to an Amen&ix. It is there sholm
tint oqn.(5) still holds if the definition of h in cqn.(2) is altcrea to

h=CL

the mam effect is that in (2)

if Cr/a, is comparable with I.
deduced for eqn. (5), WC have the following rules for detenrdning the positions
$ rraxinnun velocity on the upper surface.

If 1 CL - CL optl is not loge, solve cqn. (9) by plotting the

right-hsnd side against 0 and reading off the values of 0 for which it is
equti to specified vshes of 0,; f or, if a0 = 2rf, solve eqn.(lO)
similclrly by plotting the right-hand side <urd reading off the values of G
for which it is equal to (CL - C If the right-hand side stays
practicnlly constant over a consi of values of e, then for the
oorrcspon&ng value of CL WC have c 'flat' &mum, dioh 1'70 do not attempt
t0 looate with nftcr n certain stage 6
bccomos smaller. is l‘arge comprcd with
2FL, ana h3 simply by (8). If,
howwAr, we proceed to wry large values of CL iind very small values of 0,
we must t&c the dcflnltion (12) of h in place of (2); 6 is given by (4)
for centre lines Tslthout singul<arities et 0 = 0 (gi(0) = 0), ‘and by
substituting 2 PJh into the right-h‘-& side of (5) ccand thon, if WCcsssry,
solving (5) by su0ccssive approximation] for centre lines with si@iLarities
(g;.(o) f 0).

The approximate n&hods of this section have been tested by

Iti-. E. J. Richard& %&o has applied them to N.A.C.A.16 series and ClarkY
ncrofoils, with satisfactory results for practical purposes.

3.,. The Approxinate Calculation of the Positions of Maximum Velocity on the

Lower Surfcce.

Our previous equations a:rply on the lower surface if we remember that

@ is negative, ES, even fmctlons of 6 and g$ qk, gi*
'1' 0 odd fundions of owever, convenient always to consider

these/
 - 14 -

these functions in the range 0 < o,( TT'. If we do this, then the eqn. (9)
d-h& we solve when 1 CL - CL opt) is not large, becomes

1
+- CL opt
2T()

..... (91)

and (IO), which is the form taken by this equation when aO

?a 2r(, is

CLo%t - CL oot 9‘
--------- = 2"L;-;cos-;3+ El - c0sE-j [g;(G) - g;(o)). . . . . . . (101)
2ti

If the sign of A is changed in the definition (2), so that

. . . . . (21)

then when h is large compared with 2?,, and ti large compared with

..... (111)

0 is given by

8= 2F&, ..,.. (81)

simply. For very large values of CL opt - CL, and very small values of 8,
h must be defined by (12) with the sign change&

a = cLopt(;+;) -oL[ -31” [-J+--&-.J, . ..*. (121)

and then, for centre lines without singularities, 8 -is given by (4),
G= 2pJx -JW L) y'(o), . . . .. (41)
and for centre lines with singularities we substitute @ = 2 pi/h into the
right-hand side of (5):

yJ2(Eu - hV(O) y(B)

O= -------------* . .. .. (51)
a+ y(G) u/' (El
and proceed if necessary by successive approximation; but whereas on the
upper surfaoe T+, 'yy' are given by

Y= Ys(Q) + . ~J~)9 -Y = I$@) + l+JpL . . . .. (13)

on the lower surface we must take
‘t/Z Y%(Q) - Uo(6), V’ = y;(6) - 7p’(G’). ..... (131)
c
4./
 - 15 -

4. The Theoretical C~rR~snSe of a Low-Drag Aerofoil

We recal.1 the definition of the theoretical CL-range gaven in the

antroductlon, as the range of values of CL for vrhlch the velocity continually
mcrcasos, on both the upper and lower surface, from the stagnation point to
the dosigncd posation of maximum+velocity at the desagn CL if, as is now usual,
the slopes of the graphs of gs - gi are dascontinuous at that position; ifs
thC graphs of g, +, g< are rounded off in the future, we shall require the
velocity to incrcaoe only to the beginning of the rounding off.

If 0, is the value of 8 up to vhach the velocity is to increase,

WC may immediately write lown.from cqn.(l) thc.oondition to be satisfied by CL.
The coefficient of -CL in the cz@ressaon in { ,I in (1) is positive for
o<c,<o,; so, in order that dq/dc should bc > 0 on the upper surface,
CL must not exceed the minimum vsSue of

LA 1
+-2 a,- +27f
Y” CL opt [sinO(l + :I2 + cost) + -y-q (1 t co9 ei) +
i: ) 1

. _’ .
in the range O<G<o,. Similarly in order that &q/d@ should be > 0
on the lower surface: -CL must not exceed the minimum value of the expression
obtained by changing tho sagns of gi, gi, CL opt in (14). On the wpcr
surface IJ,y' arc given by (IJ), and on the lower surface by (131).
The expression (14) is much too complicated to be of general use.
No17 as CL - CL opt ( increases, x&en it reaches a certain value the position of
maximum vcloc&ty may begin to move forward or a new m&mum may m&c Its
aJ?peaEu~cc somztierc near the nose of the aerofoil. In oithor case, as WC! seo
from the discussion in i%2, w my expect the V~LICS of 1x1 (defined by cqns.(2)
‘and 21))-at tho end of the CL-range to be such that the asswm&ons leading to
cqn. 19) /or eqn.(8) if a-becomes small enough] will be suffioicnt to provide
a fair approximation to dg/clQ. [We found thas to bc the case when
ICL - CL opt 1 = e/6, snnd the error was not prohibitively large even vrhcn
1% - CL opt 1 = 7f/3]. Consequently NC hove, approximately, that ,CL
must not exceed the manimum value of

2pLms& tsinW(g:,
+g;,+2 l/l I\
CL opt sir19 (1 + co9 0)
(i 1
2\,,Y4
cos c
I

j s3.n B ( -- + ---- ) ..,.. (15)

a0 2 7-c
id
*On the lowr surface the coofficicnt of CL, i.e. the denominator in (14), ny
be negative for wry small values of 0 If we USC the approximate values of'
For such vary small values of 8 we should, as mentioned in a previous
use‘ more accurate values for v', but such very small vslucs of 0 do not
oonczrn us here. &n-cover, the stagnation pomt does not

loading cdgc; accoraang to Approxzmation II it is at i

I
approximately, on the Lzcr or Icvc2 stu-face according as
negative or positive. X0 com$.icntions ,a.r$
conocmed tith the rsngo of 8 betwwn the 1sadlnC edge and the stagnation point,
 - 16 -

in the range 0 < 0 a,; and similwly -CL must not cxcecd the miW
value of the expressIon derived from (35) by changing the signs of g! and CL optS
The most important criteria for the validity of this amroxlmation are that, on
both the upper and. the lower surfaces, Iy Y1l should be smd.1 compared n+.th
sin 6 at the position of the minimum, and 1~~ not too different from 2p *
these criteria may be applied numerically after the vqXlucs of c at the rmh&l
h.?.vobeen found, but the numerical values in 82 arc sufficient to show that WC
nz~y expect them to be faxrly well satisfied.
When a0 = 27‘1, our approximate conditions for CL reduce to the'
1
simpler conditions that - CL opt) and f; (CL wt - CL)must not exceed
2 TI
the minimum v~alues of
cot 3
2 (7L I---.. + (1 - 00s 0 1 .gfi +, g; , ... .. (16)
I+ DOS @ ( )

respectively.
Let us now consider the case in which, as 1 CL - CL opt 1 i3 incrcascd,
a maximum of q makes its appearance near the nose of the aerofoil before the
maximum moves forward from o= Fj,. Then, if we suppose 0 small in (15),
we see that CL and -CL mat not cxcced the mininavn values of

reqwctively; i.e.

respcctlvely. Since &c/d6 = 2 sin 3, we may write (17) in the form

In order that the right-hand side of (17) may have a minimum for a
sdl value 0, of I‘, g; +, $9 ;h:hc "'+'P;~P~~~;~""~~;~~~~ I' >
must be large compared with '2pL
then the minims which occur when
or arc lower thnn others which my
occur when @ is not small.
satisfied, and g; f g! is not large compared
the former case (positive sign) or
case (negative sign) must not exceed a
quantity/

%ver the whole range of 6' up to 0 = t3,, except that very small values of @
are irrelevant, since for veq small values of @ the term in 2cL in (15) or
(16) clearly dominates; in other words the minima will not occur for very small
values of 0.
For CL = CL opf, dq/d 8 vi11 be smdl, on the upper or the lower surface
respectively, fcr all valves of F fi~h5.3 are not too md; our approximakons
~~11 be inadeauats to provide resuitc,,of even fair percentage acsuracy, bet it
ivill still be correct That ?<1/2fl)jC~ - CJ, cpt) must be O(2pL) or less.
 - 17 ”

qwl-clty which is of order 2pL, or less; whereas when g: f. g! is large

compcarcd neth 2pL in the relevant range of values of Q, the right-hand
side of (17) has a minimum of a somewhat higher order than 24L, the minimum
occurring when the two terms in ) in (17) are of the s- order of magnitude.
I
We have so far considered g' 2 g! to be positive, and we shjll new .
show it is advantageous that they shodd be'so. It is probably sufficient to
i'llwtrate the argument by using (16); similar deductions may, in fact, be made
from (15). Quite generally we may write (16) in the form
cot (9
$&..&J - 2p L T;,-F - (I - 00s ")(g; - g! -I,
>
I

< -I- CL - CL opt

I: 2ti >

cot 6
<ivIhimm 2CL -+(I-cosQ)g;+gi 7 . ..... (19)
I + WSP i )
! .J

If &-EC! is negative and gi + gi positive, and g$ +, g; are

large comparedwith 2~ for sll relevant values of 0, then the left-hs.nd
member of (19) is probab Jiy greater than the right-hand member, and there is nc
CL -range at all. For whereas the right-hard mcniwr has a minimum for 530~
fairly small value of 8, when both its terms are of the same order of magnitude,
the left-hand mcni!xr may either have nc maximum at all and we nay have to take
simply the greatest value for 0 < 6) < f.31, or a msxmm may occur v&en 8 is
not small; in either case the left-hand member till be greater than, or at least
nearly cqud to, the right-hand member. Similar conclusions follcw from (I 5);
in fact, If gl+ gA are positive end large compared with 2pL, r

(;+~)~L-cLcpt) must be positive and of the order of magnitude of

K; ” g; in order that the velocity on tic lower surface may be increasing;

land then there 1s probably some range of (fairly small) values of G for which
the velocity is decreasing on the upper surface.
Sim~ar statcwnts may be mn8.e if g; + G, or both g; + gi and
.!$ - .&cl are ncgativc. If they are lsrgc in absolute vnluc compared cith 2CL
ova- the rclcvont r‘ange of vnlucs of 0, then there is probably nc CL-range at
al, end at best a very small r<anSe.

If le; 2 $1 is small, of order 2eL, then, whet er gi z gi is

positive or negative, it still rema'ns ocrrect that +(1/2H) (7do - CL cpt ) must
be O(2fL) or less,

It follows that to obtain a CL-range of any practically significant

size, gts +I g; should be positive ‘and large compared with 2pL, except perhaps
for small values of C:. Hence g; should be positrv- and l‘orge compared with
2PL' oxccpt perhaps for small valws of 8.

It/
 -'I8 -

It dso appears that when a, 5 25f best results are obtained by

taking .$ = 0. In (16) let us write temporarily

cot (
F( c-1 E 2PL -+ (I -cosB)gk,
I + cosc:

G(cj) = (1 -cos f?)g$

L$ the vastvslue of F(6) occurtien F= @,, of F(o) + G(e) ~&en

= 1v2, and of F(k ) - G(!:) when fi = Cj3. Then, in general,

- F( 6+ + G(h’+ < -& ~L-CL$) <FF(c2)+G(g2),

and we require to show that

F(D2) t F(Fj)t G((:'2) - G(6 $ < 2-F(i;'o). . ... . (20)

Eut

F(c2) + G(c2) < F(C ,) + G('>*)

F($) - G(i'j) < F(cyo) - G( f!?,),

since FZ G is least when 0 = (‘ 2 or G3, rcspectivcly. Hence (20)

follows by addition, and our theorem is proved.

Similarly from (15) we may shew that, when a0 # 271, best results
are obtained by taking

to compensate for the variation of

CL opt (1 * 00s 6)

but usually, bn modern low-drag aerofoils, a, will be near enough to' 27T' for
thx suggested diffcrcnce of 9;. from zero to be negligible.

It should be mndc plain that, whereas the thexem on the best value
Of gl may be rlgcrously proved on the basis of our initid. approximations, the
previous discussion of the or&r of magnitude of g' is neither rigorous nor
conprehcnsive, since g' may be of widely differen: orders of magnitude in different
pnrts of' the range 0 <B&'CfJ,, and it is not practicable to discuss rigorously
aJ1 possible cases. In partloular WC did not discuss the case when gJ was
suffxiently small over the relevant range of values of 0% for the minimum of 8
(Ii'), for a fairly smxll value of 6$, to be awided, but when gk increased to i'
a different order of m~gn5.tud.e as 0 inorensod. All thnt wewere attempting ,
was a prclininsry gcncr3?. disoussion of (1) the ciroumstanccs likely to arise for
any given acrofoil; (2) th c oonditions necessary to obtain n CL-r‘ange of some
practical signil"icanc0. We dv3 not attelQ7t to find a formula for g3 to make

the/
 - IT -

the CL-r~ange as lnrgo as possible; to do this it would appear that the minimum
(17) must be avoided, and then, ns fnr as practicable, g: must be increased where
it 1s least, ana g, inorcased, in pnrticul~ar for small vciluos of 8 , in order
to inorensc 2p L h&therx?tically, for no = 2X, this problem may 'be
defined PB that of making the minimum of
cot f:

2C’L --w-w + (I - COSC) g:, . . . . . (21)

1 + 00s 6

for o< IGS((--, as lnrgo as possible, where5

d sntlsfactory solution for practical pur~ses has boon found by Thwaites 6, by

considering variatxons, involvin a sm?ll number of pammeters, of the 'roof-top'
aorofo&3 d.x3cussed in Ref.5, ~33% <SM. 7.
5. Displacement of the Eiddlc of the C,-Rcange from CT opt.
We break off the discussion of CL-ranges to refer bxaefly td a m&ta-
to which rcfcrcnoe had already boon nnde in Rcf.3, espoci~ally as the discussion
thcrc MS incomplete and mislca . - mmcly the skft of the middle of the
CL- range from CL opt when gi7 0. Wiih a0 = 277, in the notation of
cqn.(20), the middle of the CL-range is given by

--I opt = F(C)2) - F(v?3) + G(@2) +G( 93)

2/r
rind the right-h‘and side till not be zero unless g; = 0.

As nn cxamplc of both the diminution and the shift of the CL-range

when g! f 0, WC m-y cohsider the cane when d&&/&c and dgi/dx nro constant,
is lnrgc oomprcd 112th 2eL, Then the minimum of the expression
in (18) ooou~s when

..... (23)

,and, if wc wit0

dg&x, = s, agi/&K = As (o,c 1 <I), ..*.. (24)

the CL-range is given by

The/
 - 20 -

11
The totjl range of -- + ;;; CL is then
( no ' 1

2 0 -2 3/4 (%(Z L )3'4 $4 r 1 + x) l/4

\( +(1-h) l/4 I ,
3
which has its greatest value when h = 0, m.d the rrdddlc of the CL-rs.WC
ient

\;'2 refer deo to the results for the aerofoil EQH 1250/4050
(Rcf.3, Part II), a fnirly thin acrofoil with R loge cnnber (4 per cent).
According to the accurate results reported in Part II, there 1s a small CL-range,
For this acrofoil, g; is

to be md accurate than it is;

rather large. The mason lies in the large onnber, and the consoqucnt rapid
vnriation of JP. 1f we find the positlon of the minimun on the right in 17)
in the sme way as before, but substitute the value of v/ at the minimum 0.148)
for d(2pL) before finding its actual value, we obtain practically the accurate
result, CL ,( 0.66.
6. CT,-R’X~~ROS of 'Roof-Top' Acrofoils
the
CL-r‘ange of a 'roof-top' acrOfoi1,
centre lines designed for constant
see Ref.7 &O).
is y; c~~('!~ %e application of eqn.(18)
is inmediate; YE have, in fact, only to put h 3 0' in the results of the
preocding scctlon, rind we find that the CL-r,ange is given by

(,a -- + --2i )I OL ’ 0 -% opt s 2 23 3/4 (2q 3314 -- ’ . ,I)4 = $1;1$(2~33'4 s'b.

,,.d.L (25)

This formula has proved remarkably accurate for acrofoils with reasonably
large values of s and small. cnmber. (It 110s not been test& on any
acrofod with large camber, s~ncc no pmctlcol necessitv to do so hos
yet a-isen.) The results obtained, nil found In the CO"WSCof
lnvsstl,ations mode for other purposes, ilrc tcstcd bclo\J.
 - 21 -

Aerofoils 1 and 2 were symmetric&l; 3, 4 and 5 all had a centre line

desimned for constant approximate loading for 0 < x < 0.6, with the a~roximate
loading d3oreasing linearly to zero for 0.6< x<i, and
(T-C/R, + ;)c, o t = 0.126. With so L 5.5, CL opt = 0.118, 'but the
fiddle of the 8 L-renge was, in fact, ,s&what greater tiian 0.126, being 0.1305,
0.1295, 0.131 for aerof0i.k 3, 4 and 5, respectively. I'ne results for aerbfoils
1 and 2 we due to Mr. Ii. C. Garrer, and those for aerofoils 3, 4 snd 5 to
&s. I&acre. It will be seen that the formula (25) gives a satisfactory result
for a value of' s as low as 0.0545.

The shape of the falring depends on the four parsmeters a, b 'c, X. -

In place of a and o, we ?ntroduce the slope s of gs and A/ (2C~j, where
pr is the radius of curvature of the trailing edge. For a cusped aerofoil
h/ @py) = 0, and, more genersJ.ly, the degree of the concavity of the aerofoil
surface to~r~ds.,the~trailir~ Ldge is sensibly influenced by the value of h/ (2&).
We suppose X ad d(2(JT) 'h rive certain
- fzcd chosen 'values. in addition we
suppose the theoretical critical Mach number, or the azrofoil thickness at n
given chordwue position, or the msximum thickncss,is given. If the theoretical
cnticd Mach nuder is given, then vvlth a given qcntre line and design CL,
b will be fixed. If the theorcticsl critical. &ch number, for example, 1s 0.68,
and the centre lint 1s of the type previduslj: mentioned (for aerofoils 3, 4 snd 5
above) but with X = 0.5, and the low speed cquivslent of the top-qccd CL
is 0.2, then the xxtximm value of q/U on the surface must be 1.2525; with
ncoounts for 0.0667, so b is 0.1858. In any cast, if X,
or the tiiickcss for a given x, or the maximum thickness,
arc given, there will be a value For s which makes the CL-rsngc n maximum.

Cbnsidc;, for exaqle, the case X = 0.5. Then

S z 2(b - a), 2/(2pT) = 0.06831 a + 0.36338 b t 0.56831 o,

,/ (2eL) = 0.56831 a t 0.36338 b + O,Q6831 o

= O.@BOb- 0.28005 s t 0.12020 ,J (2&)'

T:le CL-range is proportional-to IJ 2p L)3/2 s'/'+; if J (2Q ) end b are

T
flxcd, and s varies, this exp:esslon has a maximum when
s = O&l+880 b + 0.0613,,/ (2pT).

If, how&r, the thickness is given at a given x, for exsmple x = 0.4, then
from the tables of Rcf.5,

0,10976 a c 0.3;936 b + 0.06077 c = ylr

where yl 1s the half-thickness, as a fraction of the chord, at x = 0.4.
Hence

0.38296 b - 0.05123 s + 0.10693 J (2pT) = y.j

end
J(2p,) = 2.29737 yl - 0.16236 s - 0.1254.6 /,/ (2pT).

Agaln (,/2eL)3!2 sl/fe *I when

has a maximum s varies, t'fiis tints when

s = 2.0214y, - 0.11% j(+).

me4
 - 22 -
When the maximum thickness is given, the matter is a little more complicated,
since the position of the maximum thickness varies as s varies, The vnriation,
however, is not large+ and there is still a maximum CL-range for which, if
JGPT) and the thickness are given, tho corresponding value of s may be
computed; and hence the values of a, b, c maybe found. For X=0.5 2
and 0.6, these values wore computed by b@. H. C. Garner; I find that his values
are satisfactorily reproduced by the formulae
a = 0.9453 t - 0.208 h/w Ip), b = I.4362 t - 0.335 ,,/(2&),
-0 = 1.0319 t - 2.00 J(2 (JT), s = o.pa10 t - o.s4 JP@

for x = 0.5, and

a = 0,8908t- O-156 ,J(2@ b- I 1.3765 t - 0.259 ,,/(22,),

-0 = 1.2121 t - 2.170 &\jT), 8 I o.8oy5-t- 0.172- J(q.3)

for x = 0.6. In these formulae t represents the maximum thickness

(not the half-thickness) as a fraction of the chord. Values have also bcon
roughly calculated for X = 0.4 by Mr. E. J. Richards; his V‘nlUcx are
represented by
a = 1.0125 t - 0.3 RlPP!r)I b q l..5125t- 0.4 J(2f+)'
-0 s 0.9085 t - 1.87 A@+ S = 1.25t- 0.25 &2&.
For all normal thicknesses and values of ,../(2c)T), the above
fo mulac lead to vary reasonable values of s. [For a ousped nerofoil
s (2pm) = 0, ‘~3 values would not normally exceed 0.02; O.CY+would be a
very Grgo va1ue.J Experimental evidence of the cffeot of s on tho'tolcrance
that can be allowed for waviness of the surface is still rather scanty, and not
at all systematic; but such ovidenoe as we have indicntcs that once a fair value
of s has boon reach@ any further increases need to be very large indeed
to make any practical difforcncc to the waviness, and the values obtained from
the above formulae are, for all normal values of t and +/((2e ), large
enough for this stage to have been reached. Thus once s = O?l, for exnmple,
it is very doubtful if it y?ould make any practical difference to the tolerance if
s wmo increased to 0.2. On the other hand, when the velocity gradient is made
very much bigger indeed, for example multiplied by a factor of IO, so that
instead of 0.1 it becomes 1.0, then it seems that the tolerance on waviness may
definitely be increased. Thus we may expect to be able to tolerate a larger
waviness very nenr ~1 aerofoil noso thnn els&hcre. Also if s is very much
dcoroased the tolerance on waviness ocrtainly bccomos less; but WChnve no
exact quanto.tative kno~vledgc, and systcmztio expwimonts are ocrtainly rcquirod.
Rough values of the maximum CL-ranges for a0 = 2If rind for vnrious values
of t nre given in the tables below. The figures give the complete CL-range
(2r( timos the right-hand side of (25)), not the half-r‘ange. For X = 0.5
and 0.6 they ‘arc derived from Mr. Gnmer's results; for X = 0.4 thoy havebeen
oomputed from the formulae given above as represcntlng Mr-. Richards' results.
I9h.m "/(2PT) = c, the CL-range is vary ne,arly proportional to
t&
CL-rsngcs/

%or X = 0.5, it ame,srs that the position of the nnximum thickness is given
quite closely by x = 0,3767 t 0.0576 s/b t O.O@g ,,/ (2eT)/b. There nre
similar fomulnc for other values of X.
 - 23 -

-I--- ---- ------ ____._ ---.--_\ _.- -

------------i
. J@PT) = 0
2..--- -- -----.
/
-.
I
J((2(q = 0.02 --

\I
100
------A--
a
t
X
0.4

0.111 t
--
0,103
0.5 0.6 I
L% ------ O.lCJ+
a
0.4

0.099
0.161 0. I+? 10 0.157
:z 0.227
0.165 0,222 0.203 12 0.218
14
15
48
20
0.297
0,375
0.291
0.367
0.451
0.543
~
0.264
0.333
0.408
0,490
14 ~~ 0.2E7
16
48
20
0.363
0.449
o.ybo
,
I
22 x I
x 22 0.640
--- -a-

1. The ~raof-top' aerofoils,' consideycc? m the preceding section,

a-se, of course, not the only ones for which the analysis can be fully carried
out. We might, for example, take

.ss = A+Btm$Q.
Then we fmd (still with g = 0) that the &nimm of (17) occurs appro-dmntoly
when

and zhe CL-range is given bv

1cL - cL ,& < L5(2pL)2'3 ES"~-

This fern of g, my lend to somewhat lcargcr CL-mngos than (25); but to

pursue.the matter fwther,.m should kwe to fmd a-d work out formdnc for the
fairmg ordinates, and 11%lewve the matter for the present.
2. In so fnr ns we my neglect -+J', ma replace y2W 2e,,
ti.1 our work could have been based on n simple form of Approximation II,
which we my call Approximtion IIa, and which is the simplest form neccss~ly if
TX nrc to make nny nttempt at nil at npproximdq to the velocity near the nose;

I
- =-
(1+$c,2 _
r_
\ (I + P- + I )’sin 0 +cL(;+::$) ' ,
u (2pL+ sin20)i; 1‘ -* "
1 1 1
Ye -+-
2 ( 2r( )
CL opt (1 + cos Q) (qa<s1 )
“0 3
/

-UC-
where
(1 + 4 $, sin 0 ,
K s -------
(2pL t sin* !'q)k'

On A&roximation III,
4 eYO(l+
- E:
C’) c:,gsin(O+ CL
f5-P)+-oos(B+E-P)
U W2 + &-g$ i(,' - ;; ) aO

)
27-7 eco
+ -w---m 1
ana
2$
(ly2 + sin* 0)3'2 eye ;;, 4-U 3 [ 1 I "; ] { h* + sin*0 7
aO

we still suppose e - p, . E,', E", U,, l.p' smdl, but CI/a,, Cdd(*~ecO),
though they are less then 1, may now be comparable in magnitude dth 1. We are
concerned only with cases in which C a,, is large enough for 8 to be small
compared with unity; and in order to 4a ve a comparison with the results from
Approximation II it will be convenient to introduce g, and gi, We havefi
11 1
g.3 + gi = co+ C't (F -P)cotG+- I + -- CL opt cot a(39
7 2 a0 27-l
( )
 - 25 -

If wx retain only the most important terms the equation as/a6 simpl2AY.es to

i I(
24
CL
1 -- y2 cos 0 - ?+ lpl sin Q + sin3 0 (g: + g!)
"20
,I 1 1
+- -+- CL opt sin G (1 + cos B ) - yy* ( C - P) Cos 0
2 ( a0 2Y-f) I

&art from the al.+m-ed.eqression for 1, this equation is tine same as

cqua.tion (5).

1. Jacobs, X.N., Abbott, Im H. and Dovldson, ?Xton.- Prilmmnry

Low-drag Amfoil and Flap Data fro17 Tests at Lorzc lieynolds Fmber
an.3 Low Turbuloncu. N.A.C.L.. Avlvrncc Confldcntlnl Report.
(I%Woh) 1942). A.R.C. 581+5m
3-. xooru. - II. J.- 1, Cmparmon of the Velocity D&crlbutlon OYW the
Dosqncd ,~erofo~l with thllt over a ,,cnsurcd ;,odcl of N..,.C....66, Z-015,
d1.R.C. 7841.

3. Goldstem, S.- Arpproxmatc Two Dmcns1onnl .icrofoA Theory. Part I.

Voloclty Distributions for Sy~~.letr~c:~l .arofolls. (iiay, 1942) s
Current Popcr No. 68.

4. Goldstcm, S.- Approxmatc Two Dimns~onaL ,:crofml Theory. Part II.

Vcloclty Dlstrlbutlons for Cmbcrud m~ofo~ls. (scptmlbcr, 1942).
Current Popcr No. 69.

Richards, E. J.- Thcorcticnl Crltlopl IIach N~.lbcrs for N.aeC.,i. I6 SOXCS

Acrofo11.

5. Goldstein, S. and Rlchzrds, Z. J.- ,,pproxxLmtc T&lo Dli8~onslonnl &rofo11

Theory. Part III. aipproxkmto Do&m of' S~lwtrlcol iim-OfO~lS for
Spccifwd Prcssurc Dlstrlbutrons. (October, 1942).
Current Paper No. 70.

6.

7. Goldstein, S.- d,pproximtc Two Dmcnsionnl .,crofoll Theory, Pm-t IV.

The Design of Cmtru Lmos. (Ihrch, 191~5). current Rpcr No. 71.

a. &on, P. R.- Gcncraliscd CIWVC:; for USC m Dcsl&nm:: Lo~-Drilg kng

scct10ns. ih.R.Co 6353.
 0 0-I 0.3 a.4 0.5
o.2 I/=

Velocity di&pibutions for NACA 66,2-015 aerofoil, to how

$ Egc when ao0’2 TTQ =o ( With a0 = 2 ff the graphs are

indistinqul’shabte
 C.P. No 72
(8549)
A R.C Technical Report

CROWN COPYRIGHT RESERVED -

S0. Code 23-9006-12 .