FILE COpy

j NO I-W

l
r.1ECENICAL 1G::,:ORANDUMS

~~ATlmJAL ADVISORY C011EITT=-::: FOR .A 1f ROJAUTICS

No. 538

CANTILEVER ~VIlJGS FOR MOJERN AIRCRAFT

Some Aspect s of C2..:1ti1ever WinG Construot ion with

Special Reference to Weight ruld Torsional StiffnsS3
By H. J. Stieger
J
From Aircraft Engineering, August, 1929

i7aahington
:~overuber, 19~9
 NATIONAL ADVISORY CO~~ITTEE FOR AERONAUTICS.

TECHNICAL MEMORANDUM NO. 538.

CANTILEVER WINGS FOR MODERN AIRCRAFT. *

Some Aspects of C&~tilever Wing Construction ~ith

Special Reference to Weight and Torsional Stiffness.

By H. J. Stieger.

The relative merits of the monoplane and the biplane have

of~en been argued, a,nd the respective advantages which, uP. to
the present, have been claimed on either side still leave the
solution of the question in doubt. Until now the bias in this
country has been toward the biplane but, as the knowledge with
regard to methods of obtaining torsionally stiff structures
grows, the pendulum is bound to swing the other way.
The success of an airplane naturally depends on the skill-
ful combination of the elements of one or the other methods _f
construction incorporated with the design _specification, and
m~ch depends on the size and purpose of the airplane. The prob·~

lem really reduces, in so far as wing structure is concerned, to

the use of:
(1) An externally braced "thin" wing section, thickness
ratio 6.5 per cent to 12 per cent.
(2)
A cantilever "thick" wing section, thickness ratio
12 per cent to 18.per cent.
*From Aircraft Engineering, August, 1929 •

•
i~.A.C.A. Tecllnical lvieil1or-andum No. 538 2

In a fast s~out type or 2_ racing airplane," where minimum

drag is of the utmost importance, the aclv8.lltage is usually with
the externally braced "thinl! wing , either' monopl3Jlc or bipl3Jle.
But if ',7e consider 8, type of airpl3Jle of the long range, night
bomber, or commercial'class -, that. is, a type in which the total
lift/drc:tg ratio, aspect ratio, etc., are paramount - the second
syste,n of construction i 8 the one which scores.
It must bc realized that, with normal present":"dEl-Y airplenes,
about two-third,s of the available b. hp is expended in overcoming
turbulent air flow which has often been unnecessarily created •.
Assuming a..'1 efficiency of 25 per cent for a gasoline engine,
the energy transfol'med into useful work is about 8 .per cent of
that initially ava.ilable from the fuel used. Obviously there
is still much to be done by both engine cilld aircraft designers.

These may be briefly sumrned up as follows!.

(1) Distortion of the wing under air load causing reduc-
tion, or even rever sal, of aileron .contl'ol, weak-
ness in torsion, and the probability of wing flutter.

(2) Increased profile drag.

(3) Weight.

Dealing 'wi th these i terns separately::

Distortion and Flutter.- A moderate deflection at the wing
tip, the outcome of pure bending due to air load,. is Quite per-
missible and does not affect the aerodynamic properties of the
i-I. A. C. A. Techni c E'.J. Memorandum No. 538 3

1.'!ring, but it is absolutely imperative that torsiono..l stiffness

is maint8.ined. Several failures have recently occurred in this
country as the result of making the wing structure too flexible.
This may have been due to inexperience in the particular type
of d.esign and lack of design data. As the result of subsequent
investigations,* the exact nature of the phenomenon of wing
flutter has been well defined. Providing that certain funda-
mental r'ules are observed, there should be now no difficulty in
constructing a cantilever wing in which the likelihood of flut-
ter is very remote.

Profile Drag.- It has often been claimed that the increased

profile dre..g of a "thick" wing is balanced by the parasite drag
due to the struts, wire, etc., in an externally braced wing
system.
For the purpose of compari son, assume a biplane structure
using an 8 per cent section, and a tapered cL.U'_tilever "thick"
wing having rul 18 per cent section at, the root, and a 12 per
cent section at the tip (i.e., a 15 per cent average section),
both having the SaIae stalling speed. It follnws th.J.t the wing
area in each case is proportional to the maXimUlll lift coeffi-
cient. The profile dIng depends on the actual shape of the sec-
tion, being smaller rOl' a "thin" wing. The following figures

may be taken as representing average values:

*See British A.R.C. Reports and Memorilnda No. 1177: "A Brief Sur-
vey of Wing Flutter with CLl1 Abstl'act of Design Recommendations,"
by R. A. Frazer Q...Tld W. J. -Duncan. (1928)
N.A.C.A. Te8hnical MemorDndum No. 538 4

Biplnne Monoplane
. per per cent
cent
Thickness rat io 8.0 ·15.0
Profile drag (c.t maximum speed) 0.004 0.006
Parasi te drag due to struts
and wires 0.002
Interference due to struts
an.d ·wires 0.001
Maxir,lUm lift ooeffir:ient 0.55 0.68
Profile drag coeffioient re-
duced to wing area basia 0.004 0.00485
CompClIative values for total
drag of wing structure 0.007 0.00485

Note.- The induced drag is almost zero at low values ()f lift
coefficient, corresponding to the maximum speed flight 00ndition.

It is obvious from these figures that the increased profile

drag in the case of the monoplane is reduced to very near equal-
ity with that of the biplane when wing area is considered, and
that, when the parasite drag of the biplane structure is takeb
into account as well, there is a very real advantage to the mon-
oplane.
Assuming the total drag coefficient fOl' a normal biplane to
be 0.018, the drag due to s.truts and wires, with the accompany-
ing interference effect, is 0.003, or 16.7 per cent of the total
drag~ If the biplane has wing tanks thi s percent age for para-
site drag is greater. In a thiok wing· such tanks ca..'1 be fitted
internally.
N. A. c.. A. Technical Memorandum Ne. 538 5

Weight.- Up to the present the weight (f the cantilever

wing constructirm has been the main drawback and, in certain
circumstances, it can eliminate the aerodynamic advantages.
The ·weight naturally depends on the air· 10ad for the maj or part,
but saving can be effected by skillful arrangement of the struc-

tural memoers. There will alway3 bA, however, the inevitable

torsion load to be provided against. That, consequent upon
travel of center pressure due to ailel'on iTI:)vement, must al1.i)'ays

be present; but for the actual wing the variation of center of

pressure with angle 0f attack, embracing the whole flight range,
can be reduced by the evolution of a suitable sestioL. In the
ideal case the torsion due to aileron movement should be the
governing factf)r. As regards the actual struoture there are

several schools of thought.

Utilization of Wing Covering for strength.-- The ideal solu-

t ion seems to be to stabilize the wing covering so tilat it can

trans;·nit .the air loads, and to i.1ake full use of it in taking

both the bending and tcrsinnal stresses. ThiR method ta8 been
used by both Rohrbach and Fokker. In actual praotioe it is very
difficult to build such a structure without considerable addi-
tion to the weight. Actually it is found that the rigid skin
for the ;'nost part stabil izes the spar s , transmit t ing the tor si on-

al loads to them. As a ;nedium for resisting cending, the wing

cover~ng is inefficient.
N.A.C.A. Technic<ll }Jie:-floT;mdum No- 538 6

Multi-Spar ~ystem.- This method has been developed to a

fine art by Junker s and others'. The bending loads, di stri but ed
as they- are over the whole chord, instead of being entirely tnk-
en by two strength members, are resisted by a number of ~ight

spars which are comple~ely inter'uraGed 8nd Gtabiliz~d~ The main

disadvantage of this type of structure) apart from th.A j=\:y.:\~ro.mF>ly

light spar section required, is the lack of torsional stiffness.

In most cases a rigid wing cov~ring has to be used to supply
this deficiency, with the resultant sacrifice in weight.

Two-S~ar System._ With the standard method of using tW0

spars to take the bending loads the wing suffers from abnormal
torsional deflection. Even the use of box spars and braced
bulkheads between them, as well as complete lateral bracing,
does not suffice to reduce the torsion within reasonable limits.
The wing weight involved is prohibitive, as the system depends
for its success on there being sufficient depth to the airfoil
section to enable bracing of economic si~e to be used.

Principle of the Mono-Spar System

The fundamental principles of the system'are:

(1) To segregate the main and secondary structures'so that
each performs its functi~n independently of the other •
. (2) To transmit the load by the shortest path, and, in
doing so, to make the structure as simple and straightforward as
possible, in order that the strength and deflection can be accu-
F.A.C.A. Technical Memorandum No. 532 7

rat ely calculated.

(3) To stabilize the main struoture as fully as possi ble,
so that a high stress can be developed in the material, with the
resultant saving in weight.

The practical r9suit is as follcws:

(1) The adoption of -a single spar pI-aced attb:e maximJ.m
depth of the wing sention - approximately at one-third of the
chord - to take all bending loaq._s.
(2) The provision of "pyramids"forming continuous tension
spirals round and along the spar. These pyramids, together with
drag and antidragmembers, transmit all torsi~n and drag loads
back to the spar, and at the same time stabilize it at conven-
iently short intervals. In virtue of this, it is possible to
use an 11k ratio of about 20 on the compression flange of the
spar, so that a very high stress can be developed without fear
of buckling.
(3) In its simplest form the system is practically non-
redundant, so that it can be easily rigged ffild corrected for in-
cidence along the spar. In the event of a torsion wire being
cut, there is admi tt edly more, but not at all undue, twi st. The
torque so liberated is taken by the spar itself, drag members,
and also by the secondary structure over a length corresponding
to the extent of half the pyramid, when it is absorbed by the
next pyraJnid. This case has been satisfactorily investigated
in an actual test.
N.A.C.A. Technical Memorandum No. 538 8

The spiral tension members may be adjustable tie rods or

simply channel or tube se8tions. As their main purpose is to
provide torsional st iffness, the advaTltage of adjust able tte. rods
lies in the fa8t that by introdusing initial tension the torsion-
al deflection may be haJ.ved. There is no need f·')r frequent ad-
justment of t.hisbracing, as in the. external bra8ing of a biplane,
since the tie rods are comparatively short in relation to their
sectional area - more like the spokes in a bicycle wheel.
There are c,bviously various possible combinations of pyramid
bracing, and a redundant stru8ture might be used, but on actual
test the simplest method of bracing has given the most satisfac-
tory results from strength, stiffness, and cost point of view.
(4) The main structure is very robust and, the.refore, not
easily damaged, and the pyra~id points, together with the spar,
form excellent points cf attachment for engines, landing gear,
struts, etc~

(5) The torsion brac·ing is not affected by the bending

deflection of the spar and is, therefore, efficient under all
flight conditions.
(6) Any required torsional stiffness can be achieved at
the expense of a small increase in weight (See Fig. 7).
(7) The secondary structure is very light and, if damaged,
-
does not affect the strength of the main structure; it is also
easily replaced.
N.A.C.A. TeGhnical Me;norandum No. 538 9

Comparison of Single and Two-Spar Structures

spar weights have been cal,:::tllated for the two cases by con-
sidering only ~he bending loads and a movement of center of pres-
sure of flO per cent, representing the effect of aileron move-
ment. I'am indebted to Mr. Duncanson ("Aircraft Engineer,"
Flight, March 28, 1929) for the method of rapidly obtaining the
icleal spar weight. In practice, of course, it is impossible to
live up to this ideal, but the figures are good enough for rela-
tive comparison.
Under ideal conditions a single spaJ.' weighs only 60 per cent
of the two-spar system. Actually there is alwaya ext'ra weight
since the spar il1USt be stabilized against buckling, it must have
fastenings, and it is not practical to have the ideal sectional
area and moment of inertia everywhere along the spar. The amount
of extra weight is obviously greater in proportion for two sPars
than for one. In fact, the actual single spar weight will be
less than 60 per cent of that of the actual two-spar system, as-
surning the extra weight per foot run to be the same for each
spar (See Fig. 2)~
The torsional deflection of a two-spar wing only designed
to take bending loads is prohibitive and it is necessary to use
some method for stiffening it. This may be done:
(1) By increasing the spar strength so as to reduce the
-
unequal bending deflection of the spars ~, this is very heavy
and inefficient,.
N.A.C.A. Technical Memorandum·No. 538 10

(2) By the use of toxsionally stiff box spars intercon-

nected by substantial main ribs ox bulkheads - this also suffers
from excessive weight and ineffic:iencywi th a wing of large as-
pect ratio.
(3) By using'"ciouble" dxag bracing - this slackens and
becomes less operative under bending loads.
(4) By employing a rigid wing covering eithex of plywood
-
or metal- this is effective but heavy.
(5) By the adoption of some type' of pyraiJid bracing between
the spars, using either tension or 00mpression. i!lembexs.

Analyzing the weight figures for these various methods, it

is found that efficient torsion bxa~ing used in conjunction
with a single spar is at least as light as, and in most cases
lighter than, a system of equal efficiency used with two spars.
Of the secondary structure, only the question of xib de-
sign needs comment. The xibs may be built up as cantilevers in
front of and behind the spar. However, by utilizing the dxag
and antidrag members, it is possible to obtain three points of
support for each rib (See Fig. 2), nrunely, at the leading edge,
the spar, and at the apex of the reaX pyramid (thi s embraces 60
per cent of the chord). The antidrag bracing at the rear can
take the fo~m of a light auxiliary membex on which the intexme-
diate. ribs can xest. This latter method gives the lightex con-
stxuction, and the actual weight is no more thaT). that of a two.,..
or eyen three.- spar wing.
N. A. C~ A. Technical lJ e :llorandum No. 538 11

Experimenta~ Test Wing

A wing has been specially designed for test purposes in or-

der to check the calculations, and the agreement between pre-
dicted and actual deflection and twist was remarkably close.
The wing weight was made up as follows:
Test Complete.
section . wing
lb./sq.ft. lb./sq.ft.
Spar 0.340 0.450
Torsion and drag bracing 0.1~9 0.269

Complete primary structure 0.539 0.719

Secondary structure 0.560 0.560

Tot al weight 1.099 1.279

Figure 4 gives the weight distribution along the span. It

should be noted that the secondary structure weight is directly
proportional to the wing area and that the weight of the torsion
bracing is a linear function of the span. The criterion weight
per square foot of wing area was chosen because of the tapered
plan form of the wing.
The complete wing was designed for an airplane of the fol-
lowing particulars:
All-up weight - 13,200 lb.
Wing area - 780 sq. ft.
Load factors - C.P.F. : 4.
C.P. B. :. 3.25.
Wing loading - 17 lb./sq.ft.
N.A.C.A. 'Technicnl Memorandum £,10. '538 12

The complete wing, would weigh 1000 lb. ~ "that is, 7.6 per cent of

the all-up weight. Further saving coulQ be effected by incorpo-

rating the compression tUbes of the torsion bracing in the bulk-
head l'ibs.,

The ideal spar weight of the tes~, section Was calculated.

The actual spar, by compari son, WaS very heavy - nearly 200 per

cent of the ideal. The inefficient spar taper -:an 18 per cent
section v'laS used throughout - was pal'tly responsible for this,
and it should be po s sible to produce a spar such what the id.eal
is of the crcler of 70 per cent of the actual weight. Incidental-
ly, th.e spar can be constructed nearer to the iCleal 'weight at
the root th~~ at the tip (See Fig. 5).
"

Tl1e test spar, d.esigned for pure bending with the C.P.F.
factor in operation, WaS also strong enough to cover -the C.,P,.B.

case and the accompanying torsion.

The follo\'Jing curve eho'ws the torsional deflection of the
test section under one of the test conditions (See Fig. 6).

The twist is 'propor~ional to the span. This is explained

by the fact that it i's approximately proportion31 to the length

of the torsion bracing. The magnitude of -the twist depends on

the thickness of wing section used, ~nd the design stress.

The deflection quoted in Figure 6 was' obtained with a cer-
~

tain amount of initial tension in the torsion bracing. By re-

ducing the allowable stres's in the :Wire~, _t:q.at is, by increasing

their weight, the torsional deflection can be reduced to any de-
- ,
N.A.C.A. Technical Memo:L[,<.1ldum No. 538 13

sired degree at the expense of very little extra weight (See Fig.
7) •
Wing Tip Construction

The mono-spar torsion bracing system, as vIi th all other

ty,pes of construction, is inefficient when the airfoil is of

IIthin" section, as at the wing tip. Here the aileron spar and

the aileron itself can be made to contribute appreciably to the

torsional stiffness.
Figure 8 shows one method by which this can be done'. iJote
the torsionally st iff triangle formed by the main spar, aileron
spar, lli1d bulkhead rib.
Some of the following figures are quoted from W. S. Farren1s
lecture (R.Ae.S., January 31,1929) in order to give a compal'a-
t i ve idea of the way in which the saving in weight conferred by

the mono-spar system c&~ be utilized:

----------.,----------r-------,----- -----------------r-------
Actual Corre- II Nominal Pay 108.d I Relative
M/c type wing sponding pay VIi th mono- increase
wt. mono-spar load spar 'V'Ting in
wing wt. - ____..!o:p--=.a="--y_l_o--=-a,_d_
per per per per per
cent cent cent cent cent
Fokker F VII 31£ 16.1 9.3 22.2 29.0 30.6

Argosy 18.0 10.Q 25.0 33.0 32.0

The mono-spar figures were based on the wing area necessary

to keep the stalling speed the swne as that of each prototype.

It should. be noted that for the Fokker F VII 3M wing the corre-
sponding lilono-spar wing is practically identical with that used
N. A. C. A. Technical fAel:1l01'andum No~' 538 14

in tb~ test hitherto described. In this particular case, there-

fore, the weight figu!'es are extremely accurate. The test wing,
therefore, only needed reduct ion in the rat io of the wing 11reas -
780 sq. ft. to 728 sq. ft. - to obtain the above figure. The re-
lluced rling loading per squ<1:Le foot of the Fokker airplane brings'
•
the load factor on the 'corresponding mono-spo.!' wing ul? to 5.3.
When :-educed to thesaJ11e load factor of 4, the 'Ning weight is
reduced from 9.3 I)er cent to about 8 per cent. The correspond-
ing pay load is, there,fore, 30.3 per cent, or an increase of
36.5 per cent.
In the foregoing remarks I have made an atteinpt to touch on
some of the structural problems met with in cant ilever vlings,
and dealt rather fully with a certain type of single-spar'con-
struction. The experimental test wing was a first attempt to
demonstrate the 'principles of this departure from orthodox meth-
ods. The result was a I'Jing both torsionally stiff and of light
weight - lighter than a corresponding bipla.'1e construction.
Several new airplanes incorperating the system of construc-
tion referred to are about to be built, and aeronautical engi-
neers will thus be able to get final confirmation of the claims
made.
l~ . A. 0 . A. Tec hn ical ~,C e{,1orandUJ{l l.~ o. 538 Fig.l
f-;
cd
P,
ro
I
0
s:::
0
S
f-l
ro
co
ill
.r!
+'
UJ
ill
,q
+'
ct--I
0
+'
s:::
Q)
eill
cO
~
,J
f-;
f-;
.
s:::
cdo
.r!
r1+'
d 0
f-; ;:j
ill f-;
S:::+'
ill CD
bO s:::
0
ill 0
..c:
.p t>D
s:::
cD .r!
r-! ~
.r!
~; ct--I
0 0
rCl
ill
co e
<1,)
S ~+'
(l) ro
QO .r! :»
cd P- ro
f-;
'd
s:::
.r!
.r!
.p
rl
bO
.
\ (j ~ .r!
"-'.'
,..~ "l! rr..
 .z:
a,Cross-sectional nrea of :.>

-----
J

1:
spar in mono-spar wing, o
. -:. ~ - ---
~
'------: . .
60'~
~-_. . - .- I ----- _ _
1 · _ _ _ _ _ _ _ 1 _ _ _ _ ---= b,Saving in weight, 40%
~

.'3 .6 c, Cross-sect ional area of ~

C.P. travel 20 'J~ of chord sum of spars in tvlO spar 0
wing , 100% g
~t )r.~ -------
j(_
'----- -,-
I. - / <-L-
___.
__ .._ Actual
d, Cross-sectional area of
sp.:tr iT.. ]r.o r~o-spar ":!ing ,
58 1/2/0
~
~­

-
e" S aVIng . ht 41 1/2,-> d/ J
- i'7eIg,
In l~
(lJ

Fig . .2 Spar weigi1t s compared in mono-spar

f,Cross-s ectional area of g
and two-spar system. ~in8 ,100 ~
. ~ .
sum of spars in tT'70 snar ~

b
I-i

Q,

70.85 ft. ~
~ ,----.::....;.~~
>
~
_ 24 .93' ft . • o

~-t-->.----- .-----.------
(J1
c.N
co
+,'\ +> f
;1: ~
'H 'H
/ '
A.R.6.5/1

to ~ \

---J-'
.--l Y -----
Y-- _ _ --===-=--=====-=-=
Plan
L oo-un --~~--~--
".rJ
1-'-
()"l
rJ)

l:\)

Ro
c.N
Fig.3 A specially d e si gned test wing showed remarkably accurate results.
If . A. C . A. Technical l\.~ e Dorandum iio . 538 Figs.4 & 5

rSeil1i-span = 35,5 ft '\

. 1 6 1')::::::rT' - JL~i- '--'- - ~ , , ,' - -- - ~! \ ~
r.;:-+_-~t - ,- -~_ '~'
-' ,-- - +, +-+t-+-+ i~1
-- ::-!=l+::f:::+=f-c:1
-P
, ,1 4 ,// // v/1V/ ._,' 17'- " 1/ /
7 -/' ;;;:; z
-'Y:
'/, v ' ~
j

'H
d , 12
W
-1-lJ'--,
I
!

=t--1- --r-- +-t

-

-+--:r ,- ~j
-j-
.'1. {,~ - h-
' -C"'r- ~ -
-I
t-t- -t'-H"!'
i
---, r- - - - ~
r - -

.......... 10
++
-.l-~.Lf-- r-r- ' +'!+-'+-1++·-+-.ti~
-,*·~-'--"'-f,~++-+-l-++-.+-t--++-l

.ml- '.- '.

P fi r- __ --rail. ~.., ,-" -
H .
c.:l
08 .
,06 ~ It- I
=r-'J~f--'-~ I -jf-+,-H---'-'-r-f--
',.J" ~ -i-+-H+"::h,
_.:: _'1';--:"<:l+-+-+-t--1
Q) I 'I 1 -, -
-r--tt--ji-+-t - f-- -j.-- '- ~:- ,- -
~ .LL ++-+-1-4--il-l--- -t ~~ >-
--j- +
~ . 04 -t+ - _.1
- r- ' -I -t-- ~
t--t--t-.-t-+-t:+
~-d--lt--+--t-++t--j-jl
,-+-:-:~~±+--+-~-. 'J--t+r-. ---
~ ,02 =: . i H- f-- i-' , n~ -ti:-r--~ --+--±-rr-- -- --
00 il II III I IV ! I~
l I r-;;p
Bulkheads
a,Drag and torsion wir es
b,Compressiol1 tubes and fittings ,
c, Wiri ng fittings .
d, Bulkh3ads .
Fig.4 We i ght distribution of var i ous components
along the s par.
-- !
-
So,,; l1i - spc ~n - ---
- 1--- - _. --

I 1_.

--

~
a, Actual
b, I deal
--'

---r '~
..;,a
bY '-""""": :::t-l~
.l l l
10 20 30 40
Distance in feet

Fig . 5 Comparison between actual and ideal

wei ghts of spar ,
 Fi2,s . t3, 7 & B

jH~~p+*~_~
t=± =f :±~
·rl 0
~ ~
s 30 '---_-.-J_ _--L_ _---L_ _- '
..
o 0';'; 0 5 10 15 20
E-< 'd DistO-nce from e nd of test
Note . section, fcet .
structure ~ e sign e d to a fnctor of 3.25 F.Tcst
section tapor3d in plan form and a ti1ickn\~8s /
chord ratio of 0 . 18 usod throu~hout .
Fig . 6 Tor s ional deflection for 1
-1
~. loading and
± 10 /0 C . P. r,lOVJ:'i1cnt .

1.20 \-1
I-~- - - r- --
- Note
-
LLni tatiol1 of deflection
-- effected by increasing
. 80 -~ f--- s i zo of torsion bracing
only , bulkhond members
~
-1 ~
-------.,
-- remQining constant .

.40 --
o i

1 • 28 1 . 40 1 . 52
i1fing w3i ght, lb . / sq. ft.
Fig . ? Variation of torsional deflection ~ith wing ~eight .

spar

Fig . 8