The 615th Lecture to be read before the' Royal Aeronautical Society since

524

"N. A. DE BEUYNE.

The F a t i g u e Characteristics of Aerolite,
( i ) Static Fatigue.
(2) Dynamic Fatigue (Wohler T e s t s ) .
(3) Impact F a t i g u e .
Resistance of Aerolite to Corrosion.
(1) W a t e r Absorption.
(2) W e a t h e r i n g Tests.
(3) Resistance to Sea W a t e r .
(4) Effect of Oil and Petrol.
(5) Heat and Fire Resistance.
Glued Joints.
Bolted Joints.
Part II.
Applications.
Monoplane W i n g Root Joints.
Shear Bracing.
Cowlings.
Aircraft Structures.
Advantages of Low Density.
Bearings.
Spinning Pots.
Airscrews.
INTRODUCTION.

An account that is to be of any value of a subject in such an early stage of
development as the use of synthetic resin materials in aircraft construction, must
necessarily contain a great many first thoughts and rough guesses, and anyone
writing such an account must be prepared to take the risk of having to eat his
words at a later date. But I believe it is better to have blundered than never to
have thought at all, and it is my hope that what I have to say will at least arouse
interest enough among those with special knowledge to contribute to the discussion
at the end of the lecture.
The study of the mechanical behaviour of synthetic resins is largely the study
of the properties of matter in an amorphous state. In contrast to that of single
crystals, which has attracted so much interest and been so productive of results
in recent years, our knowledge of matter in its most extreme polycrystalline
form is meagre. Practically all that we have we owe to Professor L. Prandtl
(a name more familiar perhaps in connection with concepts such as the boundary
layer and induced drag) who, by differentiating between true hysteresis and
elastic after-effect (1), and by devising a mechanical model (2) of a molecular
system capable of behaving like an amorphous material, has given us at least
a satisfactory basis for discussion.
Our work at Duxford has been largely concerned with practical proble; ms and,
.despite the need, the urgency of day-to-day requirements has made it difficult
during the past year to give as much attention to fundamental research as we
should have liked. It is therefore with real gratitude that I have to thank the
Department of Scientific and Industrial Research for making us a g r a n t (beginning
from last October) which will enable some research to be done into the nature of
things without regard to immediate utility.
It ha"s been my privilage to work in co-operation with the de Havilland
Aircraft Co., Ltd., and Messrs. Bakelite, Ltd., on the development of variable
pitch propellers, but, since I was not associated with that development before it
had reached ari advanced stage, I should not like to be given credit for any
considerable part of it. I mention this particularly because I am afraid newspaper reports may have given a contrary impression. Actually it is Mr. E. P .
King and Mr. C. D. Philippe who have done the lions' share of the work and it is
Mr. C. C. W a l k e r who has been responsible for its direction.

PLASTIC MATERIALS FOE AIRCRAFT CONSTRUCTION.

525

It is a pleasure to acknowledge the help I have received in the preparation of
this lecture from the staff of Aero Research, Ltd., and particularly from
Dr. R. E. D. Clark, Mr. H. Leaderman, and Mr. G. Newell.
PART I.
THERMOPLASTIC AND THERMOSETTING

PROPERTIES.
RESINS.

I suppose that the property of resins which first attracts attention is the
possibility of moulding them into intricate shapes. In addition to this property
common to all resins, some kinds of synthetic resins and to a limited extent one
natural resin (shellac) have the valuable characteristic of being " t h e r m o s e t t i n g , "
i.e., when once moulded they set to permanently infusible products. Most resins
are " thermoplastic," i.e., they become soft whenever the temperature exceeds a
certain value. It is thus convenient to divide all resins into two classes—the
thermosetting class and the thermoplastic class.
The thermoplastic resins are not of g r e a t interest for structural purposes,
because although it might be possible for the softening temperature to be so high
as to be above anything likely to be reached in service, it would in turn necessitate
the use of an uncomfortably high moulding t e m p e r a t u r e ; in fact, none of the
thermoplastic resins appears to have a sufficiently high melting point for structural
purposes. Two thermoplastic resins, cellulose acetate and methyl methacrylate,
are, however, widely used in aircraft for windscreens and fairings.
This division of resins into thermoplastic and thermosetting classes is a fundamental one because the two classes are formed in different ways. Thermosetting
synthetic resins are essentially condensation products formed by linkages between
molecules with the elimination of water, while thermoplastic synthetic resins are
polymerisation products formed by the simple process of joining up similar molecules as pearls are strung on a thread. In the condensation products, cross
linkages take place as well as chain linkages. The process of growth by polymerisation is a common one in nature, and whereas for some reason or other our
synthetic polymers stop growing after a certain chain length is reached, nature
is able to grow much longer chains such as the cellulose micelle and the gigantic
genes found in the salivary glands of Drosophila flies, which are actually visible
under a microscope. Nature evidently uses catalysts to prevent the process of
"Steric H i n d r a n c e " (3) which keeps down the sizes of synthetic chains and which
makes the synthetic products so weak. Thus whereas most synthetic resin
products consist of a chaotic tangle of comparatively short macro-molecules,
chromosomes are flat ribbon-like structures of enormous size but probably not
more than one or two molecules thick. Each part of this structure appears to be
able by means of the surface electric charges set up by changes of hydrogen ion
contentrations to select its own kind of unit from the surrounding medium and
then when the hydrogen ion concentration changes again, to leave another ribbon
of molecules juxtaposed and ready for further combination (4). The catalysts
used by nature are apparently far more specific in their action than any of the
catalysts with which we are familiar in inorganic chemistry. Their synthesis
would put into our hands tools of a most powerful kind.
THERMOSETTING

RESINS.

There are two main groups of thermosetting resins (1) the phenol formaldehyde
group (2) the urea formaldehyde group. All synthetic resins are weak in tension
and need reinforcement. The best reinforcement on a strength-to-weight basis
is cellulose, and it is an unfortunate fact the urea formaldehyde resins have an
affinity for cellulose which results in a brittle structure. This affinity is so
marked that it is believed that a chemical compound is formed for which the
name " Glucanure " (5) has been suggested. This property, though useful in
making crease-resisting cotton goods, is disadvantageous for structural purposes
Such resins have also a somewhat higher density than the phenol formaldehyde

526

N. A. DE BHUYNE.

group. F r o m the manufacturing point of view they have the great attraction of
being water soluble in their initial stages, and the expense, unpleasantness, and
danger of working with organic solvents is absent.
Although not at present of direct structural importance, the urea formaldehyde
resins, by the addition of suitable retarders and catalysts, can be made into
excellent cements. These cements, because of their great strength and water
resistance, are valuable for plywood manufacture, and can also be used in a cold
setting form in general wooden aircraft construction. Unlike casein cements,
they are proof against the attacks of mould and fungus.
In "this lecture we will be concerned with phenol formaldehyde resins only. I
propose for convenience to refer to such resins as " Bakelite," though I know
that in so far as the word is a trade name its use in such a way is objectionable;
but it has the merit of doing honour to Dr. Baekeland and of being known already
to many of those here.
T H E CHEMICAL CONSTITUTION OF BAKELITE.

Phenol and formaldehyde interact to form resinous products whose properties
vary considerably with the conditions under which the reaction takes place.
Analogy with well-known substitution reactions (such as that of a halogen in
acetanilide (6) or of the phenyl diazo-group in aniline) (7) suggests that
formaldehyde will attack phenol in two ways simultaneously
A _ O H + CH20

AoCH2OH

(I)
and ortho- and para- H O . C 6 H 4 . C H 2 O H

(ID
producing a phenol ether (I) and a hydroxyphenylmethanol (II), (II) will not
readily undergo further change but (I) will easily hydrolyse back to its progenitors. These will react again to give a mixture of (I) and (II) and (I) will be
hydrolysed once more. In this way, given infinite time, the whole of the product
may be converted into (II) but whether or not this will occur in fact will depend
upon the various reaction rates involved and in the time allowed for the
condensation.
In addition t o these simple derivatives the diphenyl di-ether (III) and the three
isomeric dihydroxy-diphenyl methanes (IV) may be formed in a similar way to
(I) and ( I I ) . Again it is to be expected that a mixed ether-aryl methane .(V)
will be formed as a by-product in the formation of ( I V ) .

CH

(T 'l
an)
H O . C6H4. CH2. C6H4. OH
(IV)
H O . C 6 H 4 . C H 2 . O. C„H 5 .
(V)
If the formaldehyde is in excess then this will be able to attack the benzene
nucleus in more than one position and a large number of further compounds will
result of which phenol trimethylol (VI) is an example. So far, experiment is in
accordance with theory since most of these compounds have been isolated from the

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

527

reaction mixture. In addition there are others (such as VII) (8) for which it is
,not so easy to account.
OH

CH,OHACH 2 OH

V

CH2OH
(VI)
It might be supposed that the substances actually formed would be largely
determined in an easily predictable manner by conditions such as the relative
proportions of phenol and formaldehyde. But unfortunately this is not so. In
acid solution, for instance, equimolecular proportions of phenol and formaldehyde
react rapidly until all the phenol and one half of the formaldehyde have disappeared
and thereafter the reaction goes on more slowly, suggesting that compounds of
types (HI) to (V) are formed under such conditions. Clearly then the main
reactions depend upon the relative speeds with which different groups of compounds are attacked, and even though the probability of the molecular collisions
necessary for the formation of one compound is low, this particular compound
may be formed if the rate of its subsequent interactions is large. In such a wild
scramble for the pancake it is difficult to predict from initial conditions alone who
will be the winner.
This all too brief account of the beginnings of the reaction will at least indicate
the obscurity surrounding the formation and constitution of these resins. H o w
the compounds initially formed combine to form macro-molecules is even more
obscure.
In what follows two of the better known theories will be briefly
reviewed.
(i) During the formation of a phenolic resin there are, as Baekeland pointed
out (9) three main stages to be distinguished. In the first or (A) stage a liquid
or pasty product is obtained. Further heating produces partial polymerisation,
and a (B) stage is reached in which the product is thermoplastic and
swells in solvents. W i t h further heating some varieties of the (B) stage resin
give a (C) stage in which the true bakelite, unaffected by heat or solvents, is
formed.
Baekeland found that in the transition from (A) to (B) water is eliminated, i.e.,
condensation takes place, but that in going from (B) to (C) no loss of water
occurs, so that probably the change is one of polymerisation. The evidence is
perhaps hardly convincing in the light of our present knowledge of the properties
of colloids, because if the molecular weight is large in the (B) stage the transition
to macro-molecules might take place by condensation and the water could be
absorbed in the molecular interstices. If, however, the (B) to (C) transition
is one of polymerisation, then in view of the evidence that polymerisation is
peculiarly apt to occur when ethylenic bonds are present, Baekeland and Bender
(10) suggested the following mechanism.
Ph
CH,0 + H,C<°
- ^

< = >

(C stage)
-!>•

OH

^CH
C <C°.PHh. O H
> CH
* =C

ly)
!?cS
C^ C « H «° H > CH
* = C>C H OH
6

C6H4OH
C6H4OH C6H4OH
^
,
1 CH„—C—CH,—C—CH„—C—CH, .
C K6 H
H,4 O H C 0 H 4 O H
C6H4OH
OPh

and

(inp

. . .

CcH4OH

C H , — C—CH„—C— CH 2 .
C„H d OH

OPh

.

4

528

N. A. DE BEUYNE

According to this theory the use of a ketone
such as V I I I

(e.g., acetone)

to give a compoun

OPh

CH3

V
c
A
CH 3
C6H4OH
(VIII)
should prevent the formation of the (C) stage after treatment with
formaldehyde.
But in fact (VIII) does give a (C) stage resin.
Quinonoid formula; have also
been Invoked in order that ethylenic formula? may be used, but they do not seem
very plausible.
In brief, every attempt to explain the formation of bakelite in terms of the
polymerisation of substituted ethylenes appears to be open to serious objections,
and it seems that despite the attractions of the theory, ethylenic compounds may
have no part in the reaction.
(2) The possibility remains that bakelite in its final or (C) stage is the result
of a true condensation and not a polymerisation. A mechanism of this nature
was suggested by Raschig (11) and although based on little sound experimental
evidence at the time it has steadily gained in favour in recent years. Raschig
believed that the Novolac resins consisted of various isomerides of type ( I V ) .
These by themselves were incapable of further condensation but on treatment
with formaldehyde (e.g., by heating with paraformaldehyde) they yielded alcohols
of which (IX) is one of the many possible isomerides.
OH

OH

OH

These isomerides then reacted with phenol or with themselves to give compounds
such as (X) after which the molecule could grow indefinitely by successive
reactions of the same kind. Raschig's final formula for the (C) stage was an
elaborate polycyclic structure, not free from objections from a stereochemical point
of view, yet much research has tended to support his views on the fundamental
stages of the reaction.
The relevant literature is so voluminous that a critical account of the experimental evidence in support of Raschig's theory cannot be given here. This
would in any case be unnecessary as other reviews are available (12). But one
recent confirmation is well worth mentioning: Megson" (13) in an important paper
.(which also gives references to earlier work) has analysed the products obtained
by the dry distillation of the resins and has shown that they are in accord with
theoretical anticipations. He finds moreover that the resins obtained by acid and
alkaline condensations do not differ in any fundamental way.
The molecular constitution of these resins may therefore with some confidence
be pictured as follows:—During the transition from the (B) to the (C) stage, an
immense number of benzene nuclei become linked together by methylene groups.
Such linkages take place both in the ortho- and para- positions; the evidence as
to which preponderates is conflicting. In this way chains are built up, and there
can be little doubt that the formaldehyde attacks the ends of the chains more
easily at first than at other positions where they are better protected sterically.
But later the proportion of chain ends becomes small and the formaldehyde begins
to attack the chains along their length ; cross linking occurs and the chains become

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

529

joined together. These cross linkages are responsible, as Staudinger (14) has
shown in the case of the polyvinyl benzenes, for the enormous resistance to
chemical attack which is shown by thermosetting resins.
W e may therefore imagine the structure of bakelite in the (C) stage to be
something like that shown in Fig. 1 (15). The structure will in reality of course
be three dimensioned.

Suggested

F I G . 1.
structure- of

bakelite.

T H E MANUFACTURE OF BAKELITE.

The manufacture of bakelite, as will be gathered from the above discussion, is
an empirical process whose success depends upon the exact control of temperature,
concentration, time of reaction and the like.
Condensation is carried out either with acid or with alkali catalysts, and may
be done either in one or two steps. In the " one-step " process, phenol and
formaldehyde are heated with ammonia in a steam jacketed digestor to
a temperature of 8o° to 90 0 for about half an h o u r ; as the reaction is exothermic
careful control of the temperature is necessary. The reaction is complete when
all the free formaldehyde has disappeared; water is then removed by vacuum
evaporation. The one step process though simple is not very controllable, and it

530

N. A. DE BKUYNE.

is usual not to allow resins produced by this process to g o further than a
comparatively low state of polymerisation or condensation, so that they take
some time to change to the C form. A slow setting resin is advantageous
when it is to be used for impregnating fabric or wood as it gives time for
impregnation to take place and the process is much better controlled.'
W h e n quick setting resins are required the " two-step " process is used, as
in this process the product can be taken much nearer to the C stage without
danger of reaching it. In the " two step " process the formaldehyde is added
in two s t a g e s ; the second charge is not added until the initial reaction has been
completed and excess water removed. Usually an acid catalyst is used in the
two-step process while an alkali is used in the single step process. Cresol is
frequently used instead of phenol as it is cheaper, though the resulting resins
are not as strong as the phenolic products.
If the resin is to be used for impregnating fabric, it is dissolved in methylated
spirits to form a varnish; if it is used for m a k i n g moulding powders it is ground
and mixed in a rolling mill with fine sawdust (woodflour) and catalysts and
lubricators.
FUTURE P O S S I B I L I T I E S .

In reviewing our knowledge of the formation and constitution of the phenol
formaldehyde resins the impression left on our minds is one of our ignorance
and inability to control the reactions.
The final C stage appears to be a
muddled pattern of consequent low strength, very different from the orderly
arrays of atoms or molecules found in cellulose or the proteins.
If we could orientate the molecules so as to increase the number of secondary
links or van der W a a l ' s forces we should be able to improve the mechanical
properties. W e can carry out such an orientation by mechanical means on
many substances. T h u s artificial silk (which is regenerated cellulose in which
the micelles are irregularly arranged, in contrast to natural cellulose, in which
they lie side by side) (16) can be much improved in strength if it is swollen in
water or acid and then stretched ; the experiment is easily carried out with a
strip of cellophane. Regenerated natural silk can be improved in the same
way (17). Vinyl acetate synthetic resins can be drawn into threads whose
strength is increased by stretching after softening by heat.
Of particular
interest is the recent work of Carothers and van N a t t a (18), who have found that
there is a definite relation between the lengths of molecules and their ability to be
drawn into t h r e a d s ; a product of small original strength was in this way given
a strength of 8.3 tons per sq. inch.
All attempts t o improve the properties of bakelite in this way have been fruitless
(15). Bakelite in its initial and in its final state of condensation has been drawn
into threads and rolled and pressed without affecting its properties or its X-ray
diagram, and it seems probable that the micelles are spherical in shape. This is
confirmed by measurements of the viscosity of bakelite in solution (19, 20).
Polystyrene, a thermoplastic polymerisation product, although known to consist
of long chains of molecules, also refuses to undergo molecular orientation, probably
because polystyrene in bulk has a tangled ' ' cotton wool '' structure (15).
But it is not impossible that some orientation could be achieved during the
early stages of the reaction between phenol and formaldehyde by the use of
electric fields. Adcock (21), for instance, has recently confirmed the results of
earlier workers, that changes in the viscosity of common liquids such as acetone
can be produced by electric fields; also long molecules such as p-azo oxyanisole
have been shown to orientate themselves under such conditions (22). K n a g g s
and Schryver (23), have found that in gelatine solutions coagulated by electric
fields, the " non-amino " nitrogen was increased, thus suggesting ring closures.
W o o d s (24) has been able to orientate wool cells and rebuild them into a macrostructure by electric fields.

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

531

In a recent patent (25), it is claimed that condensation of phenol and formaldehyde can be effected in alternating fields without the use of catalysts. It is not
therefore too much to hope, that we may obtain some control of the molecules in
thermosetting resins so as to obtain products without reinforcement of a strength
equal to that of cotton or silk.
T H E PHYSICAL STRUCTURE OF BAKELITE.

If an X-ray analysis be made of bakelite by the Debye Scherrer method, photographs such as those shown in Fig. 2 (reproduced by kind permission of Dr.
Houwink of the N . V. Philips Gloeilampenfabrieken, Eindhoven) are obtained.
T h a t on the left is of phenol formaldehyde in a low state of condensation, while
the other photograph was obtained with the final product. Apart from an increase
in the sharpness of the halo there is no difference between the two. The diameter
of the ring corresponds to a diffraction spacing of 4.6 A n g s t r o m s ; benzene and
phenol give diffraction rings of the same diameter. Exactly similar results are
obtained with a solution of B stage resin in acetone. From these photographs,

FIG.

2.

X-ray positives of phenol formaldehyde
resin, in a low state
(on the left) and a high state (on the right) of
polymerisation.
it can at once be deduced that bakelite is a completely amorphous material, and
that the chains of molecules cannot be arranged according to any regular plan.
The work of R. Houwink (15) has made it extremely probable that phenolformaldehyde resins (at least in their initial stages of condensation) are
" isogels " ; that is to say, consist of a " sponge " whose pores are filled with
the resin in a less condensed state. W e have not space here t o describe the
numerous and exhaustive investigations (15) which point to this conclusion, but
mention should be made of the work of Filon and Harris (26), who, from investigations with polarised light, have deduced that glass has a binary structure, one
part being in a state of compression and the other in tension.
As condensation takes place the proportion of solid in the isogel increases, but
it is improbable that the material ever becomes a complete solid, because even in
fully hardened resins it is easy to detect the presence of free phenol. The strength
and modulus of elasticity increase with the degree of condensation; R. Houwink
(27) states that in the A stage the strength is about 13 l b s . / s q . in., and E
(determined by bending) is about 0.41 x io 6 lbs./sq. in., while in the C stage the

532

N.

1

1

i

A. DE BRUYNE.

1

ASTM METHOD
14000

i
Z

o

^

o

lOOOO
VARIATION OF MODULUS OF RUPTURF

/

}

6000

a

/
'

J

1

6000

8000

10000

12000

WOO

8000

KTOOO

IZOOO

WOTCHCPBAW

14000

14000

16000

16000

M%MM«T

VARIATION OF IMPACT STRENGTH

8000

10000

12000

14000

16000

c

AST M MET 00
410000

0

390000

/
/
/
350000

/

VAH ATION OF M DputY *? OF 5l,AS (CITY

1
6000

8000

10000

12000

14000

16000

MOLECULAR WEIGHT

FIG. 3.
Variation of mechanical properties with molecular weight
of mixed vinyl chloride and vinyl acetate resins.

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

53.1

strength rises to 4.25 tons/sq. in. and E (determined by bending) is about
0.84 x 106 lbs./sq. in. These are the values at room t e m p e r a t u r e ; at liquid air
temperature the strength and elasticity in the C stage rise to 4.95 tons/sq. in., and
2.13 x io 6 lbs./sq. in. respectively.
It is possible in the case of vinyl resins, to connect the mechanical properties
with the number of molecules in the micelles (28) ; the molecular weight of the
polymer is determined from measurements of viscosity, and the use of Einstein's
(29) equation. F i g . 3 (for which I am indebted to the Editor of the Industrial
and Engineering Chemistry) gives such a comparison.
The isogel is held together by two kinds of b o n d s : (1) the primary chemical
bonds holding the molecules together, (2) secondary bonds, also called " van der
W a a l ' s " forces. Primary bonds are the familiar links of organic chemistry; their
magnitude can be computed from a knowledge of the heat of combustion associated
with a particular linkage and of the variation of the energy of the link with the
separation between the component atoms.
Secondary bonds are cohesive forces which can hold molecules together where
there is no actual chemical combination ; they can arise in various ways. For
instance, molecules or atoms which may be electrically neutral over a period of
time will produce alternating electric fields having finite instantaneous values ;
these electric fields will induce charges on surrounding molecules and cohesion
will result. Or again, some molecules, although electrically neutral as a whole,
have a permanent dipole characteristic because their positive and negative
charges are separated by a certain distance. It is possible to compute the magnitude of these van der W a a l ' s forces, and as may be expected they are much smaller
than those of the primary bonds.
J. H . de Boer (30) has calculated the strength and modulus of elasticity of
phenol formaldehyde assuming that (1) the bonds are all primary (2) the bonds are
all secondary. The results are given below.
TABLE

Primary Bonds only

...

Secondary Bonds only...
As

determined by
experiments

our

I.

Strength.
4300 kg./sq. mm.
(2720 t o n s / s q . in.)
39 k g . / s q . mm.
(25 tons/sq. in.)
12.0 k g . / s q . mm.
(7.6 tons/sq. in.)

Modulus of Elasticity.
11,000 k g . / s q . mm.
(6970 t o n s / s q . in.)
45 k g . / s q . mm.)
(28 tons/sq. in.)
544 k g . / s q . mm.
(345 tons/sq. in.)

If we compare the theoretical figures with those given by experiment, we see
that the resin is much weaker than it should be, even assuming that there were no
primary bonds. On the other hand, its modulus of elasticity is far greater than if
secondary bonds alone were acting.
Now a discrepancy between the theoretical and experimental tensile strengths
of materials is quite general and is not confined to resins and glasses. Similar
discrepancies are found in metals and in crystals, and the beautiful work of
A. A. Griffith (31) (chiefly with glass) suggests that the discrepancy, at least as
far as brittle materials are concerned, is due to long thin submicroscopic cracks
causing local concentrations of stress far above the mean value. It should be
noted, however, that cellulose shows good agreement between theoretical and
observed strengths (32) and elastic moduli (33), and that the stress concentration
theory cannot be true for materials such as zinc, where to explain the weakness
of a wire of 1 mm. diameter it would b.e necessary to assume the existence of a
crack 1 cm. long (34).
W i t h all diffidence I should like to ask whether the existence of initial strains in
the isogel structure, as found by Filon and Harris (26), might not provide a
supplementary, or possibly an alternative, explanation of the fact that resins and
glasses may have a low strength yet a high elastic modulus. The weakness of

mmmkm&*M*&m*iii^&t,..:,^ .

STRESS

CURVES

FOR

/o

STRAIN

SMALL COMPRESSIVE

STRAIN
STRESS

BAKELITE

UNDER

a

K
68
58

a

>

CO

en

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

535

the material in tension might be explained by assuming (in accordance with the
phenomenon of syneresis) that the solid portion of the isogel is in a state of
contraction relative to the less condensed portions. Provided that these less
condensed portions of the isogel remain under pressure up to the point of fracture
the tensile strength of the resin as a whole will be correspondingly reduced. But
by the principle of superposition the value of Y o u n g ' s Modulus will not be affected
since both condensed and uncondensed portions suffer deformation under load.
Thus while the strength will be much below the theoretical value the magnitude
of the elastic modulus will lie between that for purely secondary and purely
primary binding according to the degree of condensation.
BEHAVIOUR OF BAKELITE UNDER

COMPRESSION.

I . Small Compressive
Stresses.
F i g . 4 (curve ABC) shows that as we compress bakelite (cresol resin) its elastic
modulus steadily rises. If, however, instead of continually increasing the applied
load we unload and then reload again, the stress-strain curve describes a loop of
which C D E H is a typical example. Experiment shows t h a t the shape of a loop
such as C D E H , is entirely unaffected by what may have been done previously in
the way of loading and unloading, provided that the stress C at which unloading
begins is higher than any previously applied stress. In mathematical language
we can say that the shape of the characteristic C D E H is uniquely determined by
the magnitude of C.
One is at first inclined to explain the existence of a permanent deformation
such as A E , by supposing that plastic flow has taken place and to compare the
effect with that shown by a metal when stressed above its yield point. But
further consideration shows that such an analogy is false.
First of all, we find that if after the bakelite is given a permanent deformation
AE it is heated for about an hour at n o 0 C. a recovery of quite considerable
magnitude takes place, as shown in F i g . 5. On reloading the specimen the curve
approximates to the original curve ABC and is quite unlike E H C . The deformation is not therefore of the nature of an irreversible permanent' flow but is
apparently due to a " freezing up " o f the elastic restoring forces.
Secondly, the fact that the shape of the unloading and reloading curve is
uniquely determined by the stress at which unloading begins, differentiates the
phenomenon from that of plastic flow. In a ductile material stressed above its
elastic limit, it is quite true that a different characteristic will be obtained on
unloading and that there will be a permanent deformation at zero stress. But on
reloading, instead of getting a loop like C D E H , the curve will retrace its shape
back to the strain at which plastic flow took place. W e can only g e t a loop if
after unloading we reverse the stress so that the material is taken above its plastic
point for tensile stress so that the specimen is thereby stretched back to its original
length. But the size of this loop will depend not only on the stresses applied, but
on the time during which plastic flow is allowed to take place.
The behaviour of bakelite is a true hysteresis effect, exactly of the same kind as
the magnetic hysteresis of ferro-magnetic materials. Stone and cast iron behave
in the same way and H. Schlechtweg (35) has described these materials as being
" non-linear elastic " materials, since though they d o not obey Hooke's Law even
approximately the strain is uniquely determined by the stress, while in plastic
materials the strain is also a function of the time. T h e hysteresis of bakelite,
stone, and cast iron may similarly be described as " elastic hysteresis " in contrast
to the " plastic hysteresis " of ductile materials.
2. Large Compressive
Stresses.
F i g . 6 shows stress against strain up t o about i6,ooolbs. per sq. in. The
experimental results are shown by the circles. The curve is derived from an

2,000 H

THERMAL

OF

BAKELITE

%

STRAIN

COMPRESSION

RECOVERY
AFTER

sa
a
w

W

H

o

05
OS

PLASTIC MATERIALS FOR AIRCRAFT-CONSTRUCTION.

537

I50OO

IO0OO-I
z

o

STRESS-STRAIN
(CRESOL)

^-

THEORETICAL

STRAIN

i
o
U

IN

FOR

spoo-|

•02
STRAIN

6.

PURE

BAKELITE

COMPRESSION

CURVE

L
O

FIG.

CURVE

EXPERIMENTAL

VJ
POINTS

J

538

N.

dS
d£

A.

DE

BRUYNE

L B S . A Q . IN.

•60«IO"H

g~g

SHOWINC

CURVE

VARIATION
WITH

2AOO

4POO

FOR

RJRE

OF YOUNCS

BAKELITE, (CRESOL)

MODULUS

PRESSURE

6JOOO

8,000

COMPRESSIVE

STRESS

IO.OOO

12,000

U.OOO

L B S . / S Q . IN.

7.
Variation of Young's Modulus of cresol resin under
FIG.

compression.

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

539

expression, due to H . Schlechtweg (35), obtained by a generalisation on dimensional principles of Griibler's (36) empirical equation, which has been fully confirmed by investigations of the elastic properties of stone and cast iron. Its
general form is
s/e = E0[i + { B - C y ( 2 / 3 ) } s - L ( 2 / 3 ) i s 3 / 2 ]
.
.
(1)
where s is the compressive stress.
e is the strain.
E0 is the value of Y o u n g ' s Modulus at zero stress.
B is a constant called the compressive sensitivity.
C is a constant called the shear sensitivity.
L is a constant called the combined shear compressive sensitivity.
For cresol resin B — G\J (2/3) = 1.99 x i o - 4 and L = 1.406 x i o - 6 (inch-pound
units). Since L is small in comparison with B — C 1/(2/3) its effect only
becomes appreciable when the stress is large. For small stresses the shape of
the curve is therefore a hyperbola and the ratio of the stress to the strain is
proportional to the stress.
By drawing tangents to the curve shown in Fig. 6 it will be seen that E at
first increases to a maximum value and then decreases. The exact variation
of E with stress (obtained by differentiating Schlechtweg's expression) is shown
in Fig. 7.

Fragments

of specimen

F I G . 8.
of cresol which was originally

cylindrical.

Up to about 6,ooolbs. per sq. in. the value of E steadily increases so that the
elastic modulus of bakelite for small stresses could be improved by giving it in
some way or another a permanent compression of magnitude somewhat less than
this amount.
It should be noted that all the above figures refer to cresol resin. Phenol resins
have better characteristics and the experimental results given in Table I refer to
such resins.
3. Compressive
Failure.
Shortly before failure takes place the stress-strain curve departs from the
Schlechtweg expression. The failure is of a sudden explosive kind and the
specimen splits up into a number of columns or needles whose major axes arc

540

N. A. DE BRUYNE

parallel to the direction of the applied load (see Fig. 8). Results of tests on
cresol resin cylinders iin. high by £in. diameter gave ultimate strengths of
24,35olbs. per sq. in. and 2i,ooolbs. per sq. in.
BEHAVIOUR OP BAKELITE UNDER T E N S I O N .

Cresol resin is weak in tension and breaks at about 5,ooolbs. per sq. in. The
value of Y o u n g ' s Modulus continually decreases with increase in tension and the
nature of the variation can again be expressed by Schlechtweg's equation
modified to take into account the opposite sign of the applied stress. It then
becomes
s/e = E0[i-

{ £ + 0^/(2/3) } s + Ls 3 /2 ( 2 /3)l]

COMPRESSIVE

.

.

(2)

STRENCTH AND SPECIFIC GRAVITY

CF BAKELITE WITH W O O D - FLOUR FILLER(CRESOL RESIN)

28,000

z

o
•*

Z

COMPRESSIVE STRENCTH

26:000

en
(C

a.
2

o
U

24POO--

22)000--1-25

21,000
PERCENTAGE

OF WOOD FLOUR

9.
of wood flour on compressive strength
specific gravity of cresol resin.
FIG.

Effect

and

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

541

The nature of failure is characteristic of a brittle material; there is no flow
and any concentration of stress causes premature breakdown. The tensile
strength of stone is so small that the term in L in equation (2) can be neglected,
but this is not the case with bakelite even with cresol resin.
DISCUSSION OF BEHAVIOUR OF BAKELITE.

The most striking property of pure bakelite is the variation of Y o u n g ' s Modulus
with the sign and magnitude of the applied stress. Prandtl (2) has pointed out
that such a variation can be simulated by a structure composed of a pile of interlocked beams (like a carriage spring). Increase of pressure will tend to flatten
them out and cause an increase in Y o u n g ' s Modulus up t o a maximum when
they are quite flat, after which it will decrease again as the beams are bent in
the opposite direction. Tension, on the other hand, will cause a continuous
increase in curvature and a continuous decrease in Y o u n g ' s Modulus.
The nature of the fracture, both in tension and compression, suggests that the
existence of a maximum principal stress is the cause of failure
The investigations of M. M. Frocht (37) are in agreement with this view.
W e shall see that the type of failure is profoundly modified by the addition of
powdered fillers or of continuous reinforcement and that shear failure then
becomes the usual type.
PROPERTIES OF BAKELITE WITH

FILLER.

Phenol formaldehyde is not usually used without a " filler " and wood flour
is commonly used for this purpose. T h e effect of such a filler is t o increase the
compressive strength with some increase in density. F i g . 9 shows the effect
of variation in the proportion of wood flour to resin on the compressive strength
and density. W e have not yet had time t o study the corresponding variation in
the constants B, C, and L of Schlechtweg's expression, but we have observed
that the type of fracture changes at about 10 per cent, of wood flour. Below
this percentage the break is of an explosive kind; above this percentage failure
takes place by shearing along a plane at approximately 45 0 to the direction of
compressive failure, after a good deal of plastic deformation.
The properties of the resin can thus be considerably improved by the addition
of a filler. Very similar results are obtained when carbon black is added to
rubber (38). It is clear that there is scope here for a considerable amount of
research.
TENSION

REINFORCEMENT.

Although the use of a powdered filler improves the mechanical properties of
bakelite resin, greater strength in tension can be obtained by the use of a
continuous reinforcement, and it is then possible to obtain a material having
comparable compressive and tensile strengths. Table II below (taken chiefly
from Ref. 39) shows the tensile strengths of a number of common substances.
TABLE

Material.
Egyptian cotton (dry)
Indian cotton (dry)
American cotton (dry)
Bamboo
Flax

Jute

Hemp
Ramie
Apocynum sibiricium

II.

Tensile Strength.
Tons/sq. in.
.. 18-23
••• I8.5-23-5
... Up t o 28
... 24
• • 23-70
••

24

..
••
••

58.2
49-i
73-4

Density
1-55
i-55
i-55
—
—
•—
•

—

—
—

Max.
Strength/Density.
14.8
i5-i
18.0

542

N . A. DE BRDYNE.

TABLE II

(continued).

Tensile Strength.
Material.
Tons/sq. in.
Agave Americana
... 24.8
... 18.3
Cocos mucifera
... 28.4
Raw silk
10.6
Merino wool (scoured)
... 11.7
Camel hair
...
Chitin
••• 37
Steel
Up to 100
Spruce
5
Aluminium ...
9

Density.

Max.
Strength/Density.

7.85

12.7

0.5

10

2 6

-

3-5

It will be seen that cotton appears t o be one of the most suitable of the common
materials. It is easily obtainable woven into fabric and can easily be impregnated ; the elastic modulus of the thread is comparable with that of bakelite
so that when the composite material is strained the adhesion forces between the
bakelite and the reinforcement d o not reach high values. T h e use of a textile
reinforcement is advantageous because of the endurance of such materials to
repeated b e n d i n g ; this is illustrated by some results obtained by A. Schopper (40)
and reproduced in Table I I I below.
TABLE

Material.
...
Wool
Cotton fabric
Artificial silk
Paper
...
Jute fabric ...
Aluminium foil

...
...
...
...
...

III.

No. of bendings
required to cause fracture.
... 100,000,000
...
3,170,000
...
631,000
...
320,000
...
10,000
...
32

The figures m Table II refer to fibres. T o obtain a thread these fibres have
to be twisted together and the resulting mechanical system is a complicated
one (41). The central fibres remain untwisted and the surface fibres get the
greatest amount of twist. A tensile load applied t o the thread is taken chiefly
by the core as the outermost fibres have a considerable inclination to the direction
of the applied load. W i t h a proper amount of twist, failure eventually takes
place by breakage of the core fibres, because the friction produced by the twist
in the outer fibres is sufficient t o prevent slipping in the core. The core can be
defined as the central portion of the thread bounded by a surface in which the
force of friction is just equal to the true breaking strength of the fibres. Fig. 10
shows that the strength of the thread never exceeds about 75 per cent, of the
strength possible if all the fibres were parallel and prevented from slipping by
some agency other than t w i s t ; it will also be observed that practically the whole
strength of the thread is due t o the core. F i g . 10 refers t o unimpregnated
thread which has not yet been woven into fabric.
It will be seen t h a t a solid rod of cellulose such as can be produced by the
coagulation of viscose should be more suitable for our purpose than a thread of
twisted fibres. Unfortunately, preliminary experiments show that such coagulated
material (familiarly known as rayon) is not wetted by phenol formaldehyde. The
strength of rayon is also much reduced by moisture (42) (see Table IV below),
though this might not be important if the material were well protected by
impregnation. Experiments are being continued, however, t o see if these
difficulties cannot be overcome.

PLASTIC MATERIALS FOR AIRCRAFT

543

CONSTRUCTION.

ST REN GT i C »F N1AT ERI M_.
QO

SO

ST RE

*? H C )FT HRE AD
•^P

7C\

/

/

/

fiO

/ /

'r

50

/

/si "RE ^Gl H DF

cc>RE

/
/

c\J

^

j
y
IC>

2C>

3(3

4C>

5<D

TWIST IN TURNS PER INCH.
F I G . IO.

Effect of twist on strength of thread.

60

Xi

544

N.

A. DE BR13YNE

15,000

EXPERIMENTAL

THEORETICAL
lO

FOR

AEROLITE

STRESS-STRAIN
CORD

(CRESOL

5,000

FAINTS

RESIN)

|- - .273«IO*[l*l-&55*IO*S-«/V878xlo"'s

STRAIN
FIG.

Stress-strain

CURVE

MATERIAL

11.

characteristics of aerolite cord material
under compression.

]

545

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

0-6*10 -

57

CURVE

FOR

AEROLITE

(CRESOL

2,000

4.000
STRESS

8,000

10,000

CORD MATERIAL

RESIN)

12,000

L B S / S Q IN
FIG.

12.

Variation of Young's Modulus of aerolite cord material
under compression.

546

N . A. DE BETJYNE,
TABLE

IV.

VARIATION OF STRENGTH WITH MOISTURE CONTENT.

Material.
Ramie
Cotton
Camel hair
Wool
Silk
Artificial silk

Ultimate Strength in tons/sq. in. at humidity of
o%
5°%
70%
100%

18.0
18.4

25.3
26.1

24.6
24.0

5.9
49-6

11.8
9.2
22.6

10.3

10.7

9-3
26.9
5- 2

12.1

10.3
4.2

9.9

24.0
6.0

33-o
3-i

PROPERTIES OF CORD AEROLITE IN COMPRESSION.

T o distinguish bakelite reinforced with a continuous tension reinforcement from
other kinds, I propose t o call such a material " Aerolite." I am aware that
this practice is not free from objections since the word is a registered trade mark
and flatters my own firm (which is not, of course, the originator of such materials),
yet we have found it necessary in our work to have some distinctive name for such
products.

Failure under compression

• F I G . 13.
of aerolite cord material
cf. Fig. 8.

(crcsol resin

impregnation),

Now as Caldwell and Clay (43) pointed out in 1924 in their pioneer work on
synthetic resin airscrews, by having a material in which 90 per cent, of the
reinforcement is in the warp, an extremely strong product in tension along the
warp is obtained. Such material we will designate as Cord Aerolite; its production in this country is due t o Mr. C. D. Philippe, of Messrs. Bakelite, Ltd. Cord
aerolite gives the strongest reinforcement whenever the applied tensile loads
are limited t o directions parallel to the cords. It represents the simplest possible
type of reinforcement.
This kind of reinforcement has very little effect on the ultimate compressive
strength, which remains about the same as for the pure resin. Fig. 11 shows,
however, that the presence of the cord profoundly alters the stress strain characteristic above 6,ooolbs. per sq. in. and the nature of the ultimate failure as i«
shown in F i g . 13.
Up to about 6,ooolbs. per sq. in. Schlechtweg's equation is obeyed and the
constants are not very different from those of pure cresol resin. Above this

PLASTIC MATERIALS FOE AIRCRAFT

CONSTRUCTION.

547

stress, however, the strain increases rapidly for a g-iven stress and something
similar to plastic deformation takes place; the ds/de curve (see F i g . 12) shows
a very rapid falling off in the value of Young's Modulus. W e see, therefore,
that apart from eliminating' the brittleness (i.e., the liability for failure t o take
place under the greatest principal stress instead of the greatest shear) no increase
in the ultimate strength or in the value of Y o u n g ' s Modulus is obtained by the
use of cord reinforcement.
If we regard the cords as a series of columns stabilised in compression by the
surrounding- resin (see F i g . 18), then it appears that the resin ceases to have a
stabilising influence at about 6,6oolbs. per sq. in. Assuming a value of Poisson's
ratio of one-third, this corresponds t o a lateral stress of about 2,20olbs. per sq. in.
This is the value of the moulding pressure under which the material was made.
Experiments are now in progress to test whether such a relationship between
moulding- pressure and plastic flow holds good in general.

LTIMAT6
STRENGTH 2S,30OLBs/sQ IN.
15,000

STRESS

% 10,000

FOR

STRAIN

CORD
IN

CURVE

AEROLITE

TENSION

t1/)

1
-008

—1—
•004

1
•012

STRAIN
FIG.

14.

PROPERTIES OF AEROLITE IN T E N S I O N .

The cord reinforcement has a remarkable effect on the properties in tension.
It increases the tensile strength to about 26,ooolbs. per sq. in. and Young-'s
Modulus to at least 2 . 0 x io 6 lbs. per sq. in. (up to 0.2 per cent, strain). F i g . 14
shows a stress-strain curve for material moulded at one ton per sq. in. Up t o
6,ooolbs. per sq. in. hysteresis, but no " elastic after effect " is p r e s e n t ; above
this stress the strain rises gradually on applying- the load and does not reach a
final steady value until a few minutes after the time of application. Again,

548

N.

A.

DE

BKUYNE.

therefore, as in compressive behaviour, it appears as if the resin and cord reinforcement are able to act together so long as the lateral stress is less than the
moulding pressure. Above this stress the cord reinforcement shrinks away from
the surrounding resin which then falls out of action. This view is in agreement
with the fact that the ultimate strength of the material is equal to that of the
reinforcement alone. This was verified by determining the strength of the cord
material only (before impregnation) and then counting under a microscope the
number of cords in the cross section of a tensile test specimen. The strength of
the specimen was i,i5olbs., while that of the threads alone was i,245lbs.
(118 threads at io.55lbs. for the strength of each thread).
EFFECT

OF

INITIAL

COMPRESSION

ON STRESSES

IN BENDING

TENSION

ppP^o;

STRESSES
TO

- a «^dill

DUE

BENDINC

COMPRESSION

SUPERIMPOSED

-tr. I l l
11

INITIAL

COMPRESSION

TENSION (crk-o;)
COMBINED
&

COMPRESSION

INITIAL

BENDINC
COMPRESSION

(-o;-oi)
FIG.

15

Not only is the ultimate strength dependent only on the reinforcement, but
the value of Y o u n g ' s Modulus even at small stresses is chiefly determined by it.
Thus, if we take fabric having twice as many threads in the warp as in the
weft we shall find that the value of Y o u n g ' s Modulus, as well as the ultimate
tensile strength, are approximately in the ratio of 2 : 1 . W e have verified this
not only for cresol resin impregnation, but also for urea formaldehyde and
methyl methacrylate resins.
The process of moulding is one which we should expect would leave the resin
in a state of compression relative to the fabric, because when the resin softens
in the press it will experience a uniform hydrostatic pressure equal to the
moulding pressure and the fabric will be correspondingly extended. When the
resin hardens it will keep the fabric in this state of tension.
A number of other facts are explicable on such a view. Briefly they are as
follow:—1. Moulding at a low pressure gives a brittle material having properties like
unfilled bakelite.
A state of high initial compression,
on the other hand, would

PLASTIC MATERIALS FOR AIRCRAFT

549

CONSTRUCTION.

(exactly as in " Securit " glass) prevent fracture at surface cracks by displacing
the neutral axis towards the tension side of the beam as shown in Fig. 15.
2. W h e n fabric instead of cord material is used the stress at which elastic
after effect becomes appreciable is numerically equal to the moulding pressure.
Here, the bakelite, instead of being in continuous lengths parallel to the warp

Effect

of Varia.Uott. of Mouldi^gr fiosurg on, >somt physical properties

Urc* formaldehyde

rtsiv,. muforced witk coHon

of

ikltcc,

(as in the cord material), is broken up into a series of beads by the weft. These
beads will clearly be pulled apart at such a strain as corresponds to the initial
compression.
3. The value of Young's Modulus at low stresses increases somewhat with
increase in moulding pressure. This is what we should expect from Schlechtweg's

550

N.

A. DE BRUYNE.

equation if the resin is in a state of initial compression determined by the moulding
pressure.
Fig. 16 shows results obtained with urea formaldehyde resin. Similar results
have been obtained with bakelite and with methyl methacrylate (" Perspex " ) .
A mechanical model illustrating the effect of variation of moulding pressure
on the properties of the reinforced resin is shown in Fig. 17. Here the cylindrical
blocks represent the beads of resin in the fabric and the springs represent the
warp. By inserting additional cylindrical blocks the initial tension in the springs
is increased, and correspondingly the load (measured by the weights on the hook),
which must be applied to cause the blocks to fall out, must be increased. It will
also be found that at this load the ratio of load to extension undergoes a sudden
reduction because when the blocks fall out of action all the load is taken by the
springs.

FIG.

17.

The springs representing
the cord reinforcement
are
given an initial tension by the wood blocks.
At an
applied load equal to the initial tension the wood blocks
jail out of the model.
MlCROPHOTOGRAPHS.

Microphotographs of sections of cord aerolite
mitted light are shown in Figs. 18 and 19. In
sections of the cords which are seen t o consist
These cords are surrounded by the transparent
stices and forms columns parallel to the cords.

under normal and polarised transFig. 18 the dark objects are cross
of three groups of three threads.
resin which fills up all the inter-

PLASTIC MATERIALS FOE AIRCRAFT

CONSTRUCTION.

551

Fig. 19 was obtained with a crossed analyser and polariser using " Polaroid "
discs, whose great diameter in comparison with that of Nicol prisms makes them
advantageous for this work. In the absence of double refraction no light would
reach the c a m e r a ; cotton is always doubly refracting, whether strained or not,
but bakelite only becomes doubly refracting under strain. In F i g . 19 the doubly
refracting cotton appears grey and the resin is dark in comparison. But round
each cord there is a line of bright light suggesting a high adhesion stress in the
resin such as might be produced if resin and cord were in a state of initial stress
relative to one another.

F I G . 18.
Micro photo graph by transmitted
light of section of aerolite cord
material.
Each cord has nine threads and is seen to be
embedded in the resin which forms continuous columns parallel
to the cords.
AEROLITE AND BAKELITE BEAMS.

Since the value of Young's Modulus depends both on the magnitude and sign
of the stress, the ordinary relation (generally written p/y = M/I = E/R) between
the deflection, the stress, and the applied bending moment does not apply.
Provided that we know the relation between stress and strain it is possible,
however, t o compute by graphical means the magnitude of the deflection for a
given bending moment. Conversely, by measuring the extreme stresses on
the beam and knowing the applied bending moment we can deduce the relation
between stress and strain in compression and tension.
This method of solution is due t o C. Bach (44), and takes a particularly simple
form for rectangular beams (45) as follows:—

M= {h*b/(ei

+

e2r

f (e) ede

(3)

552

N.

A. DE BRUYNE.

where M = applied bending m o m e n t ;
h = height of beam ;
b = breadth of beam ;
and ej and e2 are the extreme tensile and compressive strains.
The integral is determined by graphical means.
H. Deutler (46) has solved equation (3) analytically for the particular case of
a material obeying the Schlechtweg relation, but so weak in tension that the
term L can be neglected. This is not the case with bakelite where the tensile
strength is sufficiently great for the L term to have a considerable effect even in
tension.
The position of the neutral axis, in beams where Hooke's Law is inapplicable,
will vary with the magnitude of the applied bending moment or, what is equivalent, with the deflection. The hysteresis of the material will also, therefore, as

Microphotograph
discs.

F I G . 19.
of cord aerolite, between crossed " Polaroid "
A stress pattern surrounds the cords.

Deutler has pointed out, affect the position of the neutral axis even in a test
where a continually increasing bending moment is applied. Thus in a beam
made of a material like bakelite, where the neutral axis continually moves
towards the compressive side, the existence of hysteresis must affect the place
where the stress changes from compressive t o tensile stress. The effect of
hysteresis is, however, not capable of any straightforward analytical treatment
and will probably always have to be neglected.
W e have not yet had an opportunity to investigate the behaviour of bakelite
beams, but from the above it will be clear that the " modulus of rupture " is not
a quantity of any fundamental significance and the fact that its magnitude depends
so markedly. on the shape of the beam and that it is always greater than the
real tensile strength is easily understood.

PLASTIC MATERIALS FOR AIHCBAFT CONSTRUCTION.

553

E F F E C T OF NOTCHES ON T E N S I L E S T R E N G T H .

A noteworthy illustration of the fundamental modification that takes place in
the properties of bakelite when reinforced is given by the behaviour of notched
tension specimens made of aerolite. These notches cause an apparent increase
in the tensile strength—a phenomenon characteristic of • ductile materials and
apparently first noticed by Kirkaldy (47) in 1862. F i g . 20 shows the results of
tests on aerolite material having 26 threads per inch in warp and weft (and
therefore rather low ultimate tensile strength). It will be seen that the apparent
tensile strength rises from 9,38olbs. per sq. in. t o i 3 , i o o l b s . per sq. in. as the
depth of the notch is increased.
This behaviour characteristic of ductile materials is generally explained as
due to the resistance t o deformation in the notched portion caused by the adjacent
unnotched material.
u.ooo

EFFECT OF
TENSILE
z
O 12.000

NOTCHES ON

STRENCTH

AEROLITE

FABRIC

OF
MATERIAL

(A
to
_i

I

E

10

10,000
UJ
-J

(/»
z

UJ

tpoo
O

.2

*

.6

.8

l.O

b/fe
FIG.

20.

Similar tests with pure cresol resin gave widely scattered results and failure
generally took place in regions remote from the notch.
GENERAL ELASTIC PROPERTIES OF AEROLITE.

Aerolite is an aeolotropic material. W h e n reinforced with ordinary fabric its
axes of symmetry correspond to the orthorhombic system of crystal symmetry
and nine elastic constants are necessary to describe a general state of stress.
Aerolite cord material corresponds t o the tetragonal system (one axis of symmetry
being parallel t o the cords, with t w o other axes, both of the same length, at
right angles to each other and the first axis). Six elastic constants are necessary
to describe a general state of stress in such a system. These considerations are
of importance because errors may arise in the determination of the elastic constants
of the materials if formulae derived for isotropic materials are applied without
modification (48).
The determination of G (the modulus of rigidity) may in particular give rise to
errors if carried out in the usual way by twisting a rod of the material. Only
in the case of cord aerolite when the axis of the rod is parallel t o the cords does
such an experiment give a true result.

554

N . A. DE BHUYNE.
TABLE

V.

SOME MECHANICAL PROPERTIES OF AEROLITE.

Tensile

strength

Compressive strength ...
Shear strength

...

Y o u n g ' s Modulus
Modulus' of rigidity
Impact
(on
standard
specimen as used for
spruce)
Density ...

Cord Material Cresol Resin
(33 cords/inch).
25,ooolbs./sq. in. along
cords.
27,ocolbs./sq. in. perpendicular to cords.
5,8oolbs./sq. in. parallel to cords.
2 . o x io 6 lbs./sq. in.
0.3 x io 6 lbs./sq. in.
35ft. lbs.
i-34-

Fine Fabric Material
(80 threads/inch).
i8,ooolbs./sq. in.
26,ooolbs./sq. in.
9,ooolbs./sq. in. perpendicular to fabric layers.
1.1 x io 6 lbs./sq. in.

8.7ft. lbs.
'•34-

T H E ENERGY ABSORBING PROPERTIES OF AEROLITE.

The energy absorption of metals was studied by Lord Kelvin (49) many years
a g o and he showed that the law of damping could not be one in which the damping
force is proportional t o the velocity. It is only in recent years, however, that the
damping properties of materials have attracted interest; the Izod, Charpy, and
similar impact tests give some kind of information about the energy required to
fracture a material, but none about the energy absorbed under working conditions—a quantity which must in general be of greater significance. The energyabsorption of a material may be said t o have the same kind of meaning in
estimating its value for conditions of alternating stress as the elongation in estimating quality under static loading. Probably it is the difficulty of measuring
such a small quantity that has led t o its neglect.
W h e n a material is strained, energy is used up partly in overcoming elastic
forces and partly in doing work against the internal friction forces ; the strain
energy due to the elastic stresses is reversible, but the energy absorbed by the
internal friction is dissipated as heat. The ratio of the energy absorbed to the
elastic or strain energy is called the energy absorption of the material. This
ratio becomes large if the material is stressed beyond the yield point, but it still
has a finite value for indefinitely small displacements.
The energy absorption of a material is most easily determined from observations of the decay of torsional oscillations in a strip of the material. W e
found (50) that the magnitude of energy absorption in aerolite is higher than that
in wood or metal. This, of course, is in accordance with the marked hysteresis
effects shown under static loading, because each torsional oscillation corresponds
to a hysteresis loop. Some values, which must be regarded as giving the order
of magnitude only and not as accurate determinations (since energy absorption is
in general a function also of temperature) are given below.
TABLE

VI.

ENERGY ABSORPTION IN TORSIONAL

OSCILLATION.

Material.
Percentage Energy Absorption.
Aerolite with fine fabric reinforcement (50)
... 24
Cord aerolite (50)
...
...
...
...
... 20
Bakelite with paper reinforcement (50) ...
... 18
" Novotext " (51)
12.5
Mahogany and walnut (50) ...
...
...
... 12
Zinc (52)
11.7
Aluminium (52)
...
...
...
...
...
1.1
0.9 per cent, carbon steel (52)
...
..
... ,0.17
Nickel (52)
...
...
...
...
...
...
0.021

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

555

The energy absorption at small displacements is independent of the length or
cross section of the specimen and of the amplitude or period of oscillation. It
is a characteristic only of the material.
Energy absorption is not connected with fatigue, creep, or with flow in the
material though, of course, if the amplitude of oscillation is sufficiently large to
produce plastic deformation the absorption is much increased. The absorption
continues indefinitely and after many millions of oscillations shows no signs of
diminution. Single crystals appear to have no energy absorption (53) so that
we conclude that the effect is connected with the existence of grain boundaries,
and for pure metals the logarithm of the energy absorption at constant temperatures appears to be a linear function of the appropriate Debye temperature,
suggesting that the process is due t o scattering of the phonons of elastic energy
by atoms at the Debye frequency.
The fact that the energy absorption is unaffected by practically every variable
except the nature and state of the material means that we cannot represent the
motion of a system with this type of damping by an equation of the form
mx + k'x + ex = P sin ut
.
.
.
.
(4)
A modification of this equation capable of mathematical development and in
better agreement with the facts is that suggested by von Schlippe (54) as
follows:—
mx + k/u>x+cx = P sin o>t
.
.
.
.
(5)
Equation (5) can be written in vector notation as
.
.
.
.
(6)
(c - a)2m) X + /fcX = P
which gives the amplitude of displacement as
A = Pjsf { {c-^mY
+ k2 }
.
.
.
.
(7)
whereas equation (4) gives the expression
A' = P/</'{ (c - <o2m)2 + fc'V } .
.
.
.
(8)
The maximum amplitude An is obtained when c = (u2ra. This corresponds to
resonance between the externally applied vibration and the natural frequency
of the svstem.
AR = P/k
(9)
If V is the dynamic magnifier, i.e., the ratio of the displacement at resonance
to the displacement under an equal static load, then
V = AJAU3Lt=AJ(P/c)
= c;k .
.
.
.
(10)
It will be seen that the value of the dynamic magnifier is not affected by the
frequency of the vibration at resonance as is the case for a system governed by
equation (4).
Fig. 21 shows the variation of amplitude with applied frequency when
fc = o. 153 c (corresponding t o aerolite) and fc = o.ooio8 c (corresponding t o steel).
The curves give an idea of the relative behaviour of bars made of aerolite and
steel having the same natural frequencies. The dynamic magnifier for steel is
thus about one hundred and forty times greater than that of aerolite.
Fig. 21 illustrates a characteristic of aerolite of importance in aircraft construction. N o one who has flown a metal Moth after a wooden one can fail to
have noticed the increase in the amount of engine vibration which reaches the
cockpit of the metal machine. Yet, apart from the fact that in one the engine
mounting and fuselage are of wood while in the other they are of steel, the
machines are identical. The use of aerolite should result in additional freedom
from vibration. Measurements of the energy absorption of metal and wooden
wings carried out by the D . V . L . (55) confirm this difference. Of course, in such
composite structures energy absorption will also be caused by friction between

556

N? A.

DE BKUYNE.

VERTEX

AT

92t

RESONANCE

CURVES

FOR

STEEL A N D AEROLITE

FREQUENCV/U.

FIG.

21.

Comparative response of steel and aerolite in forced vibration.

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

557

the components, nevertheless the results obtained showed that the torsional
damping (where energy absorption in the material may be expected to predominate) of a metal monoplane wing was about one-fifth of that of a wood
monoplane wing.
E F F E C T OF ENERGY ABSORPTION ON TORSIONAL

SHOCK.

The beneficial effects of high energy absorption can also be illustrated by
considering a member such as an airscrew blade which is free at one end and is
given a sudden twist at the other. The effect of the sudden twist will be to send
a torsional wave down the m e m b e r ; this torsional wave will be- reflected at the
free end (tip of the airscrew blade) and will return to the fixed end with practically undiminished energy if the material has small energy absorption. As the
free end and the fixed end form an antinode and a node respectively the original
and reflected waves will interact to produce a maximum stress at the fixed end
of twice the value which would obtain in the absence of reflection.
If, however, the material has finite energy absorption the reflected wave will
return with diminished energy and the maximum stress will be correspondingly
reduced.
Now let us disregard considerations of energy absorption and treat aerolite
as a metallic material. The maximum shear stress (56) q developed if a circular
shaft of radius r is suddenly given an angular velocity co is given by
q~ro> (Gp)'
.
.
.
.
.
(11)
Thus the ratio of shear stresses developed in airscrew blades of duralumin and
aerolite will be
[(^PD)/(GA/>J]i
(12)
Putting
pD=2.85
(? D = 4 . 2 x i o 6
/>A=I-34

«A =

0

-35X1O6

the ratio of shear stresses in duralumin and aerolite will be found to be 5 : 1 .
This is also about the ratio of the shear strengths of the materials. Even on
this basis of comparison, therefore, aerolite is seen t o have a high resistance
to torsional shocks.
E F F E C T OF ENERGY ABSORPTION ON W I N G

FLUTTER.

The effect of damping, due to internal energy absorption, on wing flutter has
been investigated by H. G. Kussner (55) and R. Kassner (57). In general we
should expect such energy absorption to raise the critical speed at which flutter
takes place, because the energy thus used up has to be supplied by aerodynamic
forces. It is only, however, when the ratio of torsional to bending stiffness is
low that energy absorption has any marked effect. In monoplanes the ratio is
high and energy absorption has no considerable effect.
As will be seen later, although an aerolite wing would necessarily have a low
bending stiffness it can be made (weight for weight) to have a greater torsional
stiffness than a metal or wooden wing. The energy absorbing powers of
aerolite would not therefore have any marked effect on the critical speed, but
since the critical speed is primarily dependent on the ratio of torsional to flexural
stiffness, such an aerolite wing should have a higher critical speed than a wing
of wood or metal of equal weight.
T H E C R E E P OF AEROLITE.

In addition to elastic hysteresis bakelite and aerolite also show " elastic after
effect." That is to say, when a stress is applied the strain does not instan-

558

N. A. DE BKUVNE.

taneously reach its limiting value. The magnitude of this elastic after effect
increases with the stress, but it can be detected in cord aerolite at stresses as
low as soolbs./sq. in. This after effect is largely reversible and becomes very
nearly so after the load has been applied and removed four or five times.
The irreversible component
of the creep becomes more noticeable at high
stresses. Even at 6,ooolbs. per sq. inch, however, in cord aerolite, the irreversible component practically disappears after four or five successive loadings and
unloadings. W e hope to study the creep of bakelite and aerolite in greater
detail in the near future.

T H E FATIGUE CHARACTERISTICS OF AEROLITE.

I . Static
Fatigue.
The strength of aerolite is greater under constant loads of short period than
under loads applied for a long time. Fig. 22 shows results obtained by the
STATIC ENDURANCE TESTS UNDER TENSILE
LOAD ON

REINFORED

SYNTHETIC

RESIN MATERIAL.

2 IOC

60

40

a

20

#

°

' 2 ^ 4 5 6 7 8 9 lb
TIME IN DAYS

20

REQUIRED TO FRACTURE SPECIMEN

F I G . 22.

de Havilland Aircraft Company, Ltd., on the variation of tensile strength with
time during which the load is applied. It will be seen that there appears to be a
static fatigue limit at about 75 per cent, of the strength to instantaneous loads.
W o o d (58) behaves in exactly the same wav as is shown bv Fig. 23 (see also
Ref. 59).
2. Dynamic Fatigue (Wohler
Tests).
Since the process of static fatigue must be going on during the time taken to
carry out dynamic fatigue tests, the determination of the dynamic fatigue limit
(if it exists) is difficult. The fact that the fatigue limit as determined by dynamic
tests seems to be somewhere about the same as the static limit suggests that
perhaps only a static limit exists. The behaviour of aerolite in rotating beam
tests is certainly very different from that of metals, because it is possible, as
both G. Parzich (51) and Gough and Cockroft (60) have pointed out, for the

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

559

— STATIC ENDURANCE TESTS ON BEAMS OF PINE —

to
UJ

'PER DAUERFESTIGKEIT PER WEftKSTOFFE* Q. GRAF BERLIN 1-929

-oa»
O
I
CO

©

8C

•

o
O
z 6C

•

0

-e-

LU

>

"GT

S>
©

-e-

cc
I—

to 4 0
u_
O
2C

z
LU

u
Lt

LU

a

O

SPECIMEN FRI £ OF KNOTS.

X

SPECIMEN WIT-I KNOT AT P'.ACE OF FRACT JRE.
IO
20
30
4 0 50 6 0 70 8 0 90
TIME IN PAYS REQUIRED TO FRACTURE SPECIMEN
F I G . 23.
Static fatigue of pine.

FATicue

TESTS ON FABRIC REINFORCED
BAKELITE

MATERIAL
(UNBROKEN O—')

to

©

o

55-

50

—TT
10

10*
NUMBER

OF CYCLES

F I G . 24.
Wohler fatigue tests on aerolite (coarse grade).
Reproduced by courtesy of Messrs. G. S. Gough and N. W. W. Cockcroft.

560

N. A

DE BRUYNE.

specimen t o continue t o hold together for many millions of revolutions after a
split has first appeared. The amorphous character of the material seems t o
prevent any violently progressive crack formation.
A difficulty in the interpretation of rotating beam tests is to know what stresses
are being caused by the applied bending moment. T h e stress-strain characteristics of the material make the ordinary Navier relation invalid and in addition
energy absorption will reduce the magnitude of the alternating stress.
Gough and Cockroft (60) obtained the results shown in F i g . 24, indicating the
existence of a fatigue limit at about 60 per cent, of the instantaneous static
strength. The speed of rotation was 1,000 r.p.m. so that ro 7 cycles corresponds
to nearly seven days—a period of time sufficient for static fatigue to make itself
apparent. O. Kraemer (61) found the fatigue limit to be 35 to 40 per cent, of
the instantaneous static tensile strength. G. Parzich (51) found the torsional
fatigue limit to be 5 0 + 1 0 per cent, of the static torsional strength. It is possible
that the moulding pressure may affect the fatigue limit since initial compression,
at least in brittle metals, raises the fatigue limit (62).
3. Impact
Fatigue.
W e have compared the behaviour of specimens of the same shape and size of
aerolite, R.R. 56 alloy, duralumin, and Y alloy, under repeated impact tensile
loads, using an Amsler repeated impact testing machine. These results, together
with others obtained by Messrs. High Duty Alloys, Ltd., and reproduced by their
kind permission, are shown in F i g . 25. It will be seen that the fatigue limit in
impact is about the same for aerolite cord as for R.R. 56. These results emphasise
the ability of aerolite t o resist shock.
Actually they are in accordance with elementary considerations of the process
of impact. Let K be the kinetic energy of the hammer on striking a specimen
of length I and area A. The strain energy produced in the specimen will be
E\?A/2l,
where E is Y o u n g ' s Modulus and A is the extension.
Provided no energy escapes from the specimen
K = E\2A/2l
(13)
The maximum tensile stress p in the specimen will be EX/l.
Thus
K=p2l/E.A/2
K = p*lA/E
or

p=V(KEJlA)

(14)

The maximum stress developed is therefore proportional t o the square root of
the elastic modulus. The ratio of the stresses set up in R.R. 56 and aerolite
by a given energy of impact will therefore be ^ 1 0 . 6 / 2 . 0 = 2.3. T a k i n g the
strength of aerolite as 26,ooolbs./sq. in. this would give the strength of R.R. 56
as 26.7 tons/sq. in. which is of the right order of magnitude.
Of course, a g r e a t deal of the energy of impact escapes from the machine as
is evident from the vibration set up in the floor. W h e n testing aerolite specimens
there is a considerable decrease in vibration for the same impact energy due to
the energy absorbing properties of the material.
RESISTANCE OP AEROLITE TO CORROSION.

W e have not carried out any systematic experiments on the resistance of
aerolite t o attack by chemical agents, but the following tests carried out at the
D . V . L . (61) on bakelite reinforced with paper and with fabric show what may
be expected.
1. Water
Absorption.
Test specimens were immersed in water for 24 hours. Paper-filled specimens
showed the greatest water absorption with 0.85 per c e n t . ; the pure resin showed

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

561

'•2,

IQ

R.R. 56

z
o

a

X .6
AEROLI

oo

.4.

O

R.R.56

0

Y ALLOY

OBSERVATIONS

BY

(> DURALUMIN

.2

R.R.56
j*

2

^OBSERVATIONS

AEROLITE J

3

H I C H DUTY
ALLOYS

AEROLITE UNBROKEN
O—
R.R. 56 UNBROKEN

BY

AERO RESEARCH

4

5

L o c , „ or

NUMBER

25.
fatigue

OF

6

7

BLOWS

FIG.

Comparative

impact

characteristics.

practically zero absorption. It appears that the greater the proportion of resin,
the greater is the resistance t o water.
2. Weathering
Tests.
The first signs of weathering in specimens were noticed after about three months
in a 1 mm. paper-filled specimen, which showed fraying at the edges. After six
months the surface, originally smooth, of the majority of the specimens had
become matt. The edges of t mm. paper-filled material showed distinct disintegration. At this stage, however, there w a s n o diminution in the strength
properties.

562

N. A. DE BRUYNE.

After 15 months it was observed that some of the specimens had become
somewhat bleached by the action of s t r o n g sunlight. The fraying of the thin
paper material, due chiefly t o the mechanical action of dust and rain, had
proceeded further. The strength reduction of these specimens amounted to
14 per cent. ; in the case of the thick fabric-filled boards, there was hardly any
reduction in strength. In the test pieces the area of edges constituted a large
proportion of the total surface" of the specimen. It is just these surfaces which
are especially sensitive t o the effects of weather and damp. The effect of
weathering on specimens of larger size should therefore be much less.
By repeated bending of several of the test pieces it was shown that the flexibility
of the material had in no way been impaired by the weathering process.
3. Resistance to Sea
Water.
T h e tests were carried out in tanks in which a 3 per cent, solution of common
salt was agitated. Alterations in surface, weight, and strength were studied
There was n o alteration in surface of the fabric material after eight m o n t h s ; on
the other hand, the paper material frayed slightly at the cut surfaces after one
month, but this was so small that even after eight months it could hardly be
detected by eye. Strength tests after eight m o n t h s ' immersion showed that the
paper material had lost 12 per cent, of its original strength while the fabric
material after drying had suffered n o loss.
4. Effect of Oil and Petrol.
Fabric and paper specimens 1 mm. thick h u n g in a petrol benzene mixture for
10 days, then exposed to air for three days, showed no change in appearance or
loss in strength properties. Specimens were also immersed in motor oil for
10 days without effect.
5. Heat and Fire
Resistance.
Commercial synthetic resins, are generally guaranteed up t o i5o°C. At 200 0 C.
paper-filled material began to char, while the fabric material after a long time
changed slightly in colour. Test pieces I O O X I O O X I mm. were held over a
bunsen burner so that the edges were not in contact with the flame. After
15 seconds the upper surface of the fabric material swelled up suddenly, and after
20 seconds began t o burn with much smoke. The flame was immediately extinguished on removal of the bunsen burner. The paper material behaved similarly ;
on removal of the burner it glowed weakly, and was immediately extinguished
by the slightest d r a u g h t . By comparison, a piece of plywood blazed up after
20 seconds.
GLUED J O I N T S .

The problem of glueing or cementing aerolite is one of fundamental importance
because the development of a successful method would enormously simplify the
construction and repair of aircraft structures. Riveted or bolted joints can, of
course, be made, and much more easily and effectively than in wood, but any kind,
of joint which involves the making of holes must be inferior to a practical method
of glueing.
O u r knowledge of the action of glues is small and mostly empirical. But there
is evidence (63) that the strength of a glued joint depends on
(1) A mechanical interlocking between the cement and the material to be
joined.
(2) A specific adhesion due either to primary or secondary bonds between the
cement and the material.
Concerning (1) it should be noted that though microphotographs (64) of cemented
joints between veneers of plywood show that it is not necessary that the cement
should penetrate far into the wood, the smooth glass-like surface of aerolite

PLASTIC MATERIALS FOR AIRCRAFT

CONSTRUCTION.

563

effectively prevents any kind of mechanical interlocking. As regards (2) the
amorphous structure of aerolite provides few anchorages for primary bonds.
H . Stager (65), of Messrs. Micafil A.G. of Zurich, has carried out an investigation on the tensile strength of cemented joints between smooth surfaces which is
of g r e a t significance in regard t o (2). His results are shown in F i g . 26. It will
be seen that phenol formaldehyde resin at first gives strong joints, but that as
polymerisation or condensation proceeds the strength falls off because of the
decrease in the number of available bonds. Long continued heating causes a
gradual recovery of tensile strength due probably to the fact that the bonds
causing cross linkages have time t o make themselves effective. Cresol resins
show a less marked variation in strength with time, while shellac (in which
heating can in time produce some cross-linking) shows a steady rise in strength.
VARIATION
JOINTS

IN

STRENGTH

BETWEEN

METALS

UNDER

TENSILE

LOAD

OF

CEMENTED

H. STA'CER. M I C A F I L - NACHRICHTEN (1931) pisl

It is also relevant in connection with the existence of specific adhesion that
gelatine (animal glue) loses its strength when the free amino groups are made
to disappear by the action of formaldehyde (66).
W e have carried out a number of investigations t o obtain a good specific
adhesion between fully cured aerolite using all the common cements and then
adding strongly polar compounds. The results were disappointing. W e then
tried etching the aerolite in such a way as t o break up the phenol formaldehyde
structure, but t o leave the cotton reinforcement untouched. W e have not yet
had time t o carry out any extensive investigations, but some of our results, such
as they are, are reproduced in F i g . 27. In the light of these results we believe
that the problem of making cemented joints and of building up structures in situ
can be regarded as solved in principle. W e have also obtained good results by
sandblasting and have constructed a simple device by which the fabric can easily
be exposed along the directions in which the glued joint is t o be made.

6

I.OOO •]

I5O0 -

z 2,Soo H

3.000-I

AT

ETCHED

CONTROLS

ROOM TEMPERATURE

HAND CLAMPED

3.500-I

CEMENT

CASEIN

4,00 O -I

Effect on strength

PER

INCH

HAND

CLAMPED

27.

AND WEFT.

-LLLL

CONTROLS

|ETCHED

aerolite.

PRESSED AT J$ TON
PER SQ. INCH
s MINS AT no' c

of glued joints of etching

FIG.

CONTROLS

L

s MINS. AT IIO'C

WARP

UREA

FORMALDEHYDE CEMENT

CONTROLS

ETCHED

PRESSED AT h TON / I N ' FOR IO MINS AT 9O*C

BETWEEN AEROLITE

CEMENT

IN

JOINTS

METHYLMETHACRYLATE

THREADS

AND UNTREATED

ETCHED

,

80

OF ETCHED

HAVINC

STRENGTHS

MATERIAL

SHEAR

m

w

o
B

PLASTIC

MATERIALS

FOR AIRCRAFT

CONSTRUCTION.

565

BOLTED JOINTS.

A fundamental rule in aircraft design is to avoid as far as possible the use of
fittings and connections; how far and with what success that rule can be applied
was first brought home to me after a close examination of a Fokker machine.
But some fittings and bolted joints are unavoidable.
Results on some tests on bearing strength are shown in Fig. 28. In these

O

1/)

£0

s
2

2

o

w

o
O

Ni/ssi NI ss3uig oNiavag IVNIWO^

566

N. A. DE BRUYNE.

tests the material was sufficiently thin to prevent any deformation or bending
of the bolt under l o a d ; they represent pure bearing strength unaccompanied by
any frictional contribution such as always exists in practice. Roughly speaking,
the bearing strength is over five times greater than spruce along the grain and
about forty times greater than spruce across the grain. One of the g r e a t troubles
in making bolted connections in wood is the low shearing strength which necessitates a large separation betvyeen the bolts. In aerolite a much closer spacing
can be used. The strength of such joints is dependent not only on the bearing
stresses in the holes, but also on the amount of friction between the metal fitting
and the wood or aerolite. The low compressive strength of wood perpendicular
t o the grain and the possibility of shrinkage make the attainment of any considerable amount of friction difficult and uncertain. In aerolite, however, the great
hardness of the material (particularly perpendicular to the layers of reinforcement) allows considerable frictional forces t o be set up.
The mean bearing strengths for the bolt diameters shown in F i g . 28 are given
in Table V I I below.
TABLE
BEARING

Diameter.
fin.
...
fV in

jin.
l%-i

n

...

VII.

STRENGTHS.

...

Bearing Strength,
lbs. per sq. in.
... 26,300
29,400

S1^00 '

,

37,ooo

These figures refer t o short bolts. The permissible bearing stress will decrease
with increase in length of the bolt (67).
These results show that the apparent maximum bearing stress is less with
large diameter bolts than with small ones. The conception of a bearing stress
equal to the applied load divided by the projected bolt area is, of course, a convenient but crude one, quite unrelated t o the stresses which in fact are set u p .
Actually the maximum load appears to be proportional t o the square root of the
bolt diameter as shown in the table below.
TABLE V I I I .

Maximum Load,
L lbs.
1,250
1,117
1,000
883
PART

Diameter of Bolt,
D ins.
f
-A
J
^
II.

£/£>*
2,040
I.998
2,000
2,040

APPLICATIONS.

MONOPLANE W I N G ROOT J O I N T S .

The most troublesome joints in an aircraft structure are the attachments of the
outer wings to the centre section. In a cantilever machine the bending moments
at the wing roots are extremely high and necessitate the use of heavy fittings
with numerous bolts. As an example we may consider an aeroplane of one ton
g r o s s weight. The factored bending moment a t the front spar joint, after
allowing for all relief loads, might be 285,ooolbs ins. with a shear load of
3,4oolbs. Airworthiness requirements stipulate that at a load of 62.5 per cent,
of the full load there shall be no permanent set. The distance between the top
and bottom attachments might be I2in. resulting in a factored load in the top
and bottom flanges of 23,75olbs. (10J tons). Assuming the spar to be 3m. wide
we should require, say, eight fin. bolts in each flange; thus in the centre section
front spar there will be 32 bolts. T a k i n g into account the rear spar and the

567

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

outer wings we shall need one hundred and twenty-eight bolts. These bolts could
be lightened with -^-in. holes, but even so the weight of the front and rear spar
centre section bolts alone amounts to 7.761bs.
A saving in weight can be obtained by using aerolite for the spar ends, not
only through the reduction made possible in the number of bolts and the size
of the fittings, but also through the reduction in the sizes of packing blocks.
These aerolite ends could be glued with a spliced joint to the w o o d ; but for the
centre section, which might be only three feet wide, it would hardly be worth
while making such glued joints and it would be better t o use aerolite entirely for
the centre section spar. The spar width could then be reduced to iin., thus
reducing the length of each bolt to one-third. Due to the more uniform bearing
pressure along the length of the bolts the fittings could simply be held on to
each side of the spar instead of in saw cuts as in the wooden centre section.
The reduction in flange area also makes possible an increase in the distance
between top and bottom attachments with a consequent reduction in the forces
produced by the applied bending moment.
Sketches of the different methods of construction are shown in Figs. 29 and 30.
Details of the weights are given below. As an equal saving in weight should
result for the rear spar the total saving in weight should be 24lbs.

O

-r

€ II II 5

..

8

--t PLYWOOD

WEBS
3,.... THICK

~T~1

- ¥s BRIGHT MILD STEEL BOLTS

V

MILD STEEL PLATE ^ 8 THICK

(COMMERCIAL)

JOINT

IN

WOOD

CENTRE

FRONT

SPAR

FIG.

SECTION

29.

W E I G H T OF W O O D CENTRE SECTION ( F R O N T

Eight steel plates (by weighing)
32 drilled bolts (by weighing) ...
32 nuts
...
...
...
...
Plywood
...
...
...
...
Spruce flanges and packing ...
Stiffeners ...
...
...
...

...
...
...
...
...
...

...
...
...
...
...
...

SPAR).

...
...
...
...
...
...

8.80 lbs.
3.30
0.58
3.25
7.09
1.50
24.52 lbs.

W E I G H T OF AEROLITE CENTRE SECTION ( F R O N T

Eight steel plates
16 -r^-in. solid bolts (1.2m. long)
Nuts
Webs
Aerolite flanges and packing . .
Stiffeners ...

SPAR).

0.50
...

,,

O.II

,,

2.87
2.64

,,
,,

4-o

,,

11.95 lbs.

568

N.

AEROLITE FOB SHEAR

A.

DE BKUYNE.

BRACING.

W e have seen t h a t the high bearing strength of aerolite makes possible a
considerable saving of weight in components where bearing loads predominate.
But this is not the only instance where the use of aerolite can effect a considerable
saving in weight. A special type of reinforcement makes it possible t o obtain
both increased stiffness and increased strength for a given weight (or alternatively
reduced weight for an equal stiffness or strength) in torsional and vertical shear
in comparison with alclad or plywood a s ordinarily used.

-4o-o-e-o-*~5)—r
WEBS OF
45* MATERIAL
M»" THICK
I3lfe"

14-

-jo-e-e-o- ID-

4 - f o " M'LD STEEL BOLTS

SPEC. A.I.

H.T. STEEL PLATES
3fe'' THICK

JOINT

AEROLITE
FRONT

CENTRE

SECTION

SPAR
FIG.

30.

Because of the great importance of torsional stiffness in modern aircraft, I
believe that this is one of the applications of aerolite of most immediate interest
to designers. As is well known, the critical flutter speed of a wing is in general
determined not by the separate magnitudes of flexural and torsional stiffness,
but by the ratio of torsional t o flexural stiffness and the greater this ratio the
higher is the critical speed, provided the centre of gravity lies behind the elastic
axis (68).
The method of making aerolite t o give a high torsional strength and stiffness
is to fold and press the fabric in such a way that the warp and weft a r e parallel
to the lines of principal stress.
Plywood can, of course, be made also with the plies a t 4 5 0 t o the edge of t h e
sheet, but the material is expensive and considerable waste is incurred in its
production.
In most aircraft structures the ratio of shear load to depth of structure is
such that we are concerned only with W a g n e r beams in which the web is in a
state of tension at about 45° t o the necessary stiffeners (69). Thus in comparing
the relative merits of plywood and aerolite the proper bases of comparison for
strength and flexibility are the tensile strength and the Young's Modulus respec-

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

569

tively measured at 45 0 to the edge of the sheets. It can be shown that the factor
governing the torsional stiffness of a structure made from thin sheet with
stiffeners is not G (the rigidity modulus) which ceases to have a meaning in such
cases, but E / 4 where E is the value of Y o u n g ' s Modulus at 45 0 (70).

.,

1mm. Pnwooo

«0

Stress-strain

F I G . 31.
curves for special aerolite material
plywood at 45° to the grain.

and for

F i g . 31 shows comparative stress-strain curves for 1 mm. three ply to specification 3V4 made with " T e g o . " cement, measured at 45° t o the grain, and also for
the special aerolite material. It will be seen that the stiffness is increased 2.3
times, the strength 2.5 times and the weight 1.7 times by the use of aerolite
instead of plywood of the same thickness.
COWLINGS OF AEROLITE.

The primary requisite of cowling is stiffness, the second, freedom from liability
to fatigue failure. It can easily be shown that for equal weight aerolite can give

570

N . A. DE BRUYNE.

the same stiffness as duralumin ; the static strength of the aerolite will, however,
be greater, as also will the energy absorption.
Consider a rectangular plate simply supported on all its edges under" a uniformly
distributed load w lbs. per sq. in.
Then the deflection is given by (71)
^CCua'/Et3
(,5)
where a = length of the shorter side.
E = Y o u n g ' s Modulus of the material.
t = thickness of the sheet.
Then for equal stiffness of aerolite and duralumin plates of the same dimensions,
but different thickness,
waiIEKtK3 = waiIE1)tJ)3
.
.
.
.
(16)
where Es, tA, and ED, tD, refer to the elastic modulus and thickness of aerolite
and duralumin respectively. T h e ratio of corresponding weights WA and WD
will be
WJWD=pAtJPl>tD
.
.
.
.
(17)
Substituting values for EA and E D we get
WJWn = (PJpD)(EJEJi
.
.
. . .
(18)
Putting p A = i . 3 4 , ^ = 2.85,
EA=i.ixiof, ED=io.5xio6,
WA/WD = (i.34/2.8s)(io.S/i.i)*=i.o
Hence for equal stiffness the weights of duralumin and aerolite are equal.
On the other hand, the ratio of strengths for equal weights will be
PA/PV=(PJPB)[PI>IPAY

.

.

.

.

(19)

where p A and p D are the ultimate strengths.
Putting p A = i6,ooolbs. per sq. in., p D = 56,ooolbs. per sq. in.
•
PJPD= 1.26
Half hard aluminium is frequently used for cowlings. This material has the
following characteristics :—•
/ , AI =

2

- 7 ° 5 ) PAI = i8,ooolbs./sq. in., E= 10.1 x io 6 lbs./sq. in.

This gives for equal stiffness
WA,'WA1 = (i.34/2.705) [10.1/1.1]*= 1.0
For equal weight the ratio of strengths of aerolite t o that of aluminium will be
P

A / P A I = ( I 6 . 0 0 0 / " 1 8 . 0 0 0 ) [ 2 - 7 0 5 / I - 3 4 ? = 3-6

Thus the material should be far less liable t o develop troublesome fatigue
cracks than aluminium.
The following table summarises the above results and includes figures for
electron sheet:—
TABLE I X .
COMPARISON OF AEROLITE WITH SOME OTHER MATERIALS FOR C O W L I N G .

Ratio of weight
Ratio of strength
of aerolite to
of aerolite to
weight of material strength of material
Material.
for same stiffness.
for same weight.
...
...
1.0
1.26
Duralumin
Aluminum (half hard)
1.0
3.6
Electron sheet
...
1.37
0.80
These results are of greater significance than appears at first sight, because
so often the thickness of an aeroplane covering is determined not by the structural loads but by its strength under handling loads ( " Griffestigkeit " ) . Thus

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

571

the covering of a fuselage is often made much thicker than is required by flying
and landing loads, because it has to be capable of resisting handling loads applied
at right angles to the plane of the sheet.
AEROLITE IN STRUCTURES.

Although the strength of aerolite in relation t o its density can compare
favourably with the values for other materials, a comparison on the basis of
Young's Modulus divided by density suggests that for equal weights the aerolite
will result in a far less rigid structure. Some representative figures are given
below.
TABLE

X.

Tensile Compressive
Strength, Strength, Specific
tons/sq. in . tons/sq. in. Gravity.
Material.
7.92
H i g h tensile steel
90
•

—

E
lbs./
Tensile Compressive
sq. in. Strength jp Strength /p
Ejp
30.0
11.4
3 . 8 x 10
—
X IO6

Spruce
Duralumin
Aerolite

4-5

2.2

5
11.6

25
11.2

2

cord

o-45
2.85

i-5
10.5

i-34

2.0

4.9

3 . 3 x 10

8.78

8.78

3.7 x 10

8.66

8.4

1.5 x io*

10.0

Actually, however, this is not a proper basis of comparison of stiffness or of
strength, because both involve not only constants characteristic of the material
but also of the geometrical features of the structure. N o general basis of comparison is therefore possible, and we must specify in each case the type of
structure considered and what quantities are t o be regarded as variable. In
aircraft, aerodynamic considerations often fix e x t e r n a l . dimensions and the
designer is able t o vary internal dimensions only. T h u s , in a wing or fuselage
the external dimensions can generally be regarded as fixed and the thickness of
the covering as variable. W e will now consider a few particular cases where
the use of aerolite may be advantageous.
(a) FLAT PLATE IN COMPRESSION (E

AND t VARIABLE).

The critical stress of an isotropic flat plate in compression is (72)
p0=kE(t/by
(20)
Thus for equal strengths of plates of same dimensions, but different thicknesses,
. . . . . .
(21)
E ^ =E ^
The ratio of weights will therefore be

'HVWWpjp.HE./E,]*
which is the same condition as for equal stiffness
Comparing aerolite and duralumin
^ , = 1.34,

Then W1/W2=i.o,
strength.

p 2 = 2.85,

B , = I.IXIO1,

.

.

.

.

(»)

in cowling, c/. eq. (18).

E3=IO.5XIO6.

i.e., aerolite and duralumin have equal weights for equal

(b) COMPRESSION OF A CYLINDER WITHOUT LONGITUDINAL STIFFENERS (E
VARIABLE).

+3.3 E{t/H)'
.
.
.
.
Po = o.2 E{t/B)
if the edges are held rigidly (73). Strength will therefore be given by
pa27rRt = KEt2 (0.4 + 6.6 Rt/H3)
.
.
.
.

AND t

(23)
(24)

572

N.

A.

DE

BRVYNE.

P u t t i n g i? = 4oin. and H=i2m.
(so as to make conditions approximate to those
of a fuselage such as that of the " Vultee " in which there are circular stiffeners,
say i2in. apart, but no longitudinal stiffeners)
.
.
.
(25)
Strength = irEt2 (0.4+ 1.83 t) .
Since t will in general be of the order of 1 mm. the term 1.83 t may be neglected.
Thus the ratio of weights for equal strength will be
W1/Wa = pli1/p3ta = {Pl/pa)(El/E1)i
Comparing normal aerolite and duralumin
Ty i /TF 2 = ( i . 3 4 / 2 . 8 S ) ( i o . 5 / i . i ) J = I . 4 4
If cord aerolite is compared with duralumin
TFJ ^ = ( 1 . 3 4 / 2 . 8 5 ) (10.5/2.0)*= 1.08
(c)

RECTANGULAR STRUT OF GIVEN W I D T H (E

T o get some idea
longitudinal stiffeners
stabilised about its
" Griffestigkeit " we
thick.
F o r equal stiffness

.

.

.

(26)

AND d VARIABLE).

of the relative behaviour of aerolite and duralumin for
we will assume the use of a simple rectangular section
minor axis by the skin. Then from considerations of
should not use material in either case less than 1 mm.

EJ^EJ,

(27)

or since
I^bdS/12,
I, =
3
E1d1 = Ead„3

bda'/12
(28)

This condition is the same as for cowling. But whereas for cowling we should
use aerolite with equal warp and weft for a stiffener it would be feasible to use
cord material. The ratio of weights for equal stiffness would then be
(1.34/2.85) (10.5/2.0)1 = 0.817
(d)

BEAM OF GIVEN D E P T H .

In a wing the depth of the spar is determined by aerodynamical considerations;
the width of the flange is variable. In this case for a given strength in bending
the ratio of weights for a given strength will be seen to be

WJW2 = (pJp.2)(p2/Pl)

.

.

.

.

(29)

For spruce ^, = 0.45, p 2 = 5,500. For cord aerolite ^ = 1.34, p 1 = 25,oco. Thus
WJW3 = {i.34/o.4?>) (5,500/25,000) = 0.655
On the other hand, in wing spars of spruce and aerolite designed for equal
strength the spruce spar will be much stiffer than the aerolite, since the ratio of
stiffnesses will be
Ej>2lE2Vi = {2lI-$) (5,000/25,000) = 0.27
W e have already noted, however, that the torsional stiffness of an aerolite wing
can be greater than that of a wooden wing for a given weight. Thus the ratio
of torsional t o flexural stiffness can be greater for an aerolite than for a wooden
wing.
ADVANTAGES OF L O W

DENSITY.

It will be seen from the above that generally a low density is more important
than a high Y o u n g ' s Modulus (as was, I think, first pointed out by F . R.
Shanley (74). A decrease in density, even though accompanied by a corresponding
decrease in Young's Modulus, would therefore be beneficial. W i t h this object
in view we have carried out experiments t o produce a light resin by giving it a

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

573

sponge-like structure and we find that the density can be reduced in this way t o
0.3 without resulting in too friable a material.
Such a light material should make possible the use of very thick walls so t h a t
a pure stressed skin construction should be possible even in fuselages and wings
of large diameter. Under these circumstances the second term, involving the
square of the skin thickness, in equation (27), will become important and t h e
strength will be increased at a rapidly increasing rate.
AEROLITE

BEARINGS.

Reinforced bakelite is used with g r e a t success for bearings under conditions
of slow speed and g r e a t pressure. In rolling mills the use of this material h a s
resulted in considerable economies both in power consumed, and due to the long
life of such bearings (about 13 times that of metal bearings) (75) in replacement
expenses. The bearings have the additional advantage of being *' n o n - s e i z i n g ; "
should lubrication fail, the bearing gets hot and produces an intensely strong
smell of burning phenol without d a m a g i n g the shaft. W h e r e a s it is not generally
possible to start up under full load with metal bearings because of seizing troubles
due to interruption under static load of the oil film, no such difficulties are found
with aerolite. In addition, dust and dirt do not have the deleterious effect on
shafts running in aerolite bearings that is observed when white metal or bronze
is used.
Many fittings in aircraft are subjected to rather similar conditions of slow
movement, very often reciprocating in character, and sometimes of considerable
pressure, and it may therefore be advantageous to use bushings of aerolite instead
of brass in control levers and the like.
Conversely the elimination of the necessity for metal bushings in aerolite
fittings leads to a saving in weight, and in a control lever we have made (to
replace one in metal where trouble had been experienced with corrosion) the
saving in weight and cost was quite considerable. In c.p. aerolite airscrew blades
the aerolite can be made to bear directly on to metal and the conditions of
operation (slow reciprocating motion) are just those suited to such a combination.
Research has not yet reached a point where the precise field of utility of such
bearings can be defined. Their low heat conductivity (about 1/500 of that of a
metal) rules them out for operation in engines, for example, though bakelite with
a graphite filler may be of service under conditions of high speed operation where
the bearing loads are light.
T h e following values for the coefficient of friction were obtained for a surface
pressure of 7.1 lbs. per. sq. in. (76) representing very light loads such as might
be produced for instance in a " fair l e a d . "
TABLE

Material.
Bakelite with
Fabric
Reinforcement

Bakelite
reinforced with
Fabric Snippings

Lubrication.
Dry
Grease
Oil
1 part Oil
5 parts water
Water
Grease
Oil
1 part Oil
5 parts water
Water

XI.

Coefficient of friction
for Synthetic Resin for
increasing times of running in
O.25
0.38
0.42

Comparative figures
for coefficient of
friction for brass.

0.14

0.12

0.12

0.7

0.07

O.O7

I0.08

0.08

O.08

O.08

0.08

O.08

0.085

O.IO

O.I I

O.I5

0.045

0.063

O.O75

o. 10
0.063

J-0.089

0.089

O.689

0.122

6.122

O.I22

0.10
0.063

0.085

574

N. A. DE BRUYNE.

It will be seen that under normal conditions of lubrication (grease or oil) the
use of reinforcement composed of moulded snippings of fabric gives results of
about the same value as brass. The coefficient of friction apparently decreases
ihyperbolically with increase of bearing pressure for constant velocity. T h e
optimum velocity appears to be about i metre per second (75).

FIG.

Ball

32.
races.

Ball and roller bearing cages are quite commonly made of synthetic material.
F i g . 32 shows examples manufactured by Messrs. Ellison Insulations, Ltd.
SPINNING

POTS.

Although not an aeronautical application the use of synthetic resin materials
for spinning pots in the artificial silk industry is of interest, because of the severe

Spinning

F I G - 33pot for viscose (artificial

silk).

PLASTIC MATERIALS FOR AIRCRAFT CONSTRUCTION.

575

conditions under which they operate, and the similarity of these conditions to
those affecting supercharger rotors. In such applications the material is required
to withstand a certain velocity of revolution and the resulting stress will be directly
proportional to the density. The low density and freedom from corrosion of
bakelite makes its use particularly suitable for such conditions. F i g . 33 shows a
spinning pot of wood-filled bakel