THE RESSSTANCE 0F CASEBN GM}? :czm's
LN \f‘fflt-QL‘ ‘50 FATiGUL‘ STRESfiNG

 

Thesis {at m De‘gm 53$ 5A. 53.
Naifiiulifiki‘a‘ STATE Uééi‘J’ERSZTY
Aii As‘hm‘? Mafiami’
i965}

 
   

LIBRARY L

Michigan Stats
University

  
 

TE RESISTANCE 01' CASEIN GLUE JOINTS
IN FOOD T0 FATIGUE SIRESSM

By

A11 ”brat below!

A THESIS

Suinittod to the School of Graduate Studio: of Michigan State
Uninrsity of Agriculture and Applied Science in partial
fulfillment of tho requirements for the deer» of

mm 01' SCIENCE

 

Departnont of Ioroat Products

1960 , A 1 l

ABSTRACT

In this investigation, the strength of casein glue joints in
wood to fatigue stressing was studied. A specimen which suited
test conditions to an appreciable extent, was designed. This speci-
men, according to its design, was subjected to a stress pattern
consisting of torsional and simple shear components in which the
previous was dominant in the resultant stress. In order to deter-
mine the strength of the glue joint to fatigue stressing, its
strength to statioal shear stresses created by the employed stress
pattern had to be found. The results of the investigation are
smarized in the following list:

1. Casein glue joints in wood withstand higher statical
torsional shear stresses than the standard nple block
conventionally used for determination of shear stress
in the glue line.

2. Failure by fatigue could be induced in the casein glue
joints in wood.

3. Casein glue joints resist to a considerably high number
of cycles in fatigue stressing in shear when the stress
approaches about 30 per cent of average ultimate static
shear stress of the joint.

4. The endurance limit of casein glue joints in fatigue
stressing is expected to be approached if the magnitude
of applied stress falls in a range between 20-25 per cent
of average ultimate static shear stress of the joint.

THE RESISTANCE 0P CASEIN GLUE JOINTS
IN HOOD T0 FATIGUE SIRESSIM}

By

Ali Ashraf Hoslemi

A TESIS

Submitted to the School of Graduate Studies of lichigan State
University of Agriculture and Applied Science in partial
fulfillment of the requirements for the degree of

IASI'ER OF sumac:

Department of Forest Products

1960

11

ACKNIWIEDGELENI'S

The author wishes to express his appreciation to Dr.
A. J. Panshin and Dr. R. Hejasi. Grateful acknowledgement
is given to Dr. Otto Suchsland for his guidance and criticism
in the preparation of this report.

The writer desires to acknowledge his indebtedness to
Dr. G. 0. Harris, Dr. C. A. retro, Dr. I. D. Baton and

Dr. A. I. Sliker for their generous assistance.

l.

3.

5.

iii

gngNrg

Acmwmm S C O 0 O O O C O O O

LIS.‘ 0? TABLES

LISIOFELUSI‘RAIIONS.......

III'RODUCT ION

1.1 - Definition of’ratigue

1.2 - Historical Background

1.3 - Fatigue resting and its Importance

O

1. 4.- Statement of the Problem.. . . . .
1.5'D081gn0f8p801fl0n. e e e e e e e e

WALEO-URESeeeeeeeeeeeee

2.1 - Selection and Preparation of Specimens
2.2 - Static Testing . . . .

2.3 - Fatigue Testing.
ESUIJ'S

3.1 - Static Testing . .
3.2 - Fatigue Testing. .
3.3 - loisture Content .
ANALYSISOFESULTSo . . . .

wmsmns O O C O O O O O O

5.1 - cowl-“ions mum. 0 e e e e e
5.2 - Present Concepts on.ratigue Failure-

mmn. O O O O O O O O O O O O O O O O O O O

mmm cm. 0 O O O O O O O O O O O O O O O

Page

ii
iv

wun-IHH l-'

iv

mflm

Table Page
1. Comparative Stress Magnitudes for the Two

Component Stresses in Stress Pattern Employed. . . 13
2. Data Obtained in Static Testing. . . . . . . . . . 33
3. Data Obtained in Fatigue Testing . . . . . . . . . 35

LIST OF ILLUSTRATIONS
Illustration Page
1.AsketchoftheSpecimenUsed.............7
2. Detailed Dimensions of the Members of Each Specimen . . 8
3. Members of Each Specimen (photograph) . . . . . . . . .13
1.. Jig used for Manufacturing the Specimens (photograph) .11.
5. Baldwin Static Testing Machine (photograph) . . . . . .16

6. Specimen Mounted on the Static Testing Machine
(pmtograph)eeeeeeeeeeeeeeeeeeeeeel?

7. Comparative Magnitude of Torsional and Simple Shear
Components of the Stress Pattern Employed . . . . . . .19

8. Stress-Strain Curves in Static and Fatigue Testing. . .21
9. Sonntag Fatigue Testing Machine (photograph). . . . . .22
10. Loading Fixture in Fatigue Testing (photograph) . . . .25
11. Stress Drop Curves due to Wood Crushing and Creep . . .28
12. Adjustment Curves for Fatigue Testing . . . . . . . . .29

13. Strain Oscillation as Recorded by Oscilloscope
(motograph) O O O O O O O O O O O O O O O O O I O O C .30

14. A Failed Specimen in Static Testing (photograph). . . .31.

15. A Failed Specimen in Fatigue Testing . . . . . . . . .36
(photograph)

16. S-N Curve Derived for Aircraft Casein . . . . . . . . .39
17. Hypothetical S-ll Curves for Glue-Wood Combination . . .U.

1 - INTRODUCTION

1.1 - Definition of Fatigue

Fatigue is the phenomenon of fracture under repeated stresses.
The number of cycles to which the specimen resisted before failure
has been termed the ”life" of the specimen.

The term "fatigue" is not admittedly a proper descriptive
term (7), but it has become so widely established in the literature
that it will be wise to adhere to it throughout this paper.

1.2 - Historical Background

Fatigue testing was first undertaken on a large scale by
August l8h1er in Germany in 1852 (7). Prior to I‘d'hler, however,
similar experiments had been carried out in 1829 on mine-hoist
chains (11) .

Today, workers are still engaged in studying this unsolved
problem all over the world in order to enable engineers and de-
signers to predict the life of a material under repeated stresses,
which is an important factor in the design of certain structures
as well as machine parts.

1.3 - Fatigue Testing and its Importance

The fatigue strength of materials is one of the most challeng-
ing fields for engineering progress (8). This subject is of
considerable concern since materials rupture at a unit stress not
only less than the ultimate static strength of the material, but
even less than the elastic limit, provided the stress is repeated
a sufficient number of cycles .

Fatigue causes difficulties in a tremendous number of appli-
cations such as earth-moving equipment, railroad rolling stock,
farm machinery, automobiles, as well as structures which are
exposed to vibrating stresses.

Fatigue testing, primarily involves the application of cyclic
loads on the material under investigation. The method of load
application, depending on the investigator's purpose of analysis,
is different. Fatigue tests can be made under conditions such as
(a) complete reversal of tension and compression loads of equal
magnitude, (b) complete reversal of tension and compression loads
of unequal magnitudes, (c) fatigue tests in tension only and (d)
fatigue tests in compression only. A material may or may not have
the same behavior when subjected to these different forms of re-
peated stresses.

Ih.fatigue research, usually the inmostigator finds out the
life of the material in question, in terms of number of cycles to
failure, at different stress levels and plots the calculated stress,
8, as ordinate against the number of cycles, I, as abcissa. These
diagrams are known as S-l’curves (5) (6) (8).

In collecting fatigue data, a wide scatter is encountered in
both homogenuous and heterogenuous materials (3). Although attempts
have been made to investigate the character and range of such scatter
in order to improve procedures of presentation and interpretation
of test results, fatigue tests are in general planned and interb
preted as if the fatigue life of a material were a fairly well

controlled physical characteristic that can be determined, with

reasonable accuracy, by a small number of individual tests perb
formed at various stress levels (3). The ASTM manual on Fatigue
Testing (9), for instance, recommends, as a normal procedure of
Svfl'curve determination, testing at least ten specimens selecting
stress values so as to yield a more or less uniform distribution
of test points throughout the elected intervals of cycles-to-
failure.

1.4.- Statement of the Problem

The purpose of this study was to find characteristics of ad-
hesive bond to fatigue stressing. Data were to be represented by
using statistical methods and interpreted. Conclusions were to

be drawn.

1.5 - Design of Specimen

Frequently the shear strength of the glue line exceeds that
of the wood, consequently resulting in appreciable wood failure
on the glue bond. Getting a significant amount of wood failure
on the bond would not obviously determine the true shear strength
of the adhesive used. In addition, in.many specimens used for
shear strength determination of the glue line, such as standard
maple block, the calculation of shear stress is not precisely
performed.

To eliminate as much wood failure as possible throughout this
inmostigation, a wood species which had high strength values had
to be used. Hard maple seemed satisfactorily suitable for this

purpose, since it possesses relatively high strength values (15).

To choose the adhesive for this study, one point had to be
thoroughly considered. This was the strength properties of glue
relative to those of the wood species employed. In a considerable
number of adhesives, for many practical uses the glue bond yields
higher strength values than the wood used. This phenomenon would
consequently result in failure in the wood adjacent to the adhesive
bond. In order to induce the failure on the glue bond and not in
the wood, a kind of adhesive had to be used which had lower strength
values than the wood species. Among commercially produced adhesives,
casein seemed apprOpriate for the purpose since this type of adhesive
yields lower strength values than those of other adhesives such as
phenol formaldehyde, urea formaldehyde, resoroinol, etc.

The shape of the specimen was to be such that errors caused
by bending moments would be entirely, or at least to a great extent
eliminated. This elimination of bending moments would determine the
true shear stress of the glue bond.

First, it was decided to make specimens composed of tee
members which, after gluing, would form a small beam with rectan-
gular cross section. The glue line in this case would be located
attheneutralaxiscfthebeam. Inthistypecfbeam, the shear
stress has a parabolic variation and it approaches its maximum
magnitude at the neutral axis according to the following formula:

7,; .3. 315 < 1 - 4 g )
in which V is the shearing force, b and h are the width and depth
ofthebeamandycouldbeanypointcntheyaxis, providedthe
origin is at the neutral axis. In the design of such a specimen
two difficulties were encountered.

1. If the specimen mas designed so that the failure would
be likely to occur on the glue bond, a great amount of
load had to be applied and in this study, the amount
of load was limited to only 200 pounds in the fatigue
machine.

2. In the case of staying within the fatigue machine
capacity, the specimen would fail by bending before
the glue line fails in shear.

Because of the following reasons, it was decided to design

a specimen in which the failure would be induced by torsional
stresses in the glue line:

1. Failure could be induced by loads within the capacity
of the fatigue machine.

2. The shear stress exerted on the glue line is well
defined.

3. The influence of bending moments which would yield
erroneous data, would be greatly eliminated.

The glue area had to be as small as feasible due to the

following reasons:

1. load by nature is variable in its physical structure
and-behavior. In glue joint of the test specimens, the
effects of variation in anatomy, specific gravity,
moisture content of the different points on the
glue line, and alignment of annual rings are of great
significance. The degree of effectiveness of each of
these variables is a moot question. Many of the above-
mentioned variables could not easily be entirely

eliminated whereas their effectiveness could be de-
creased by reducing the glue area as much as practicable.
2. A small glue area will serve to reduce the development
of internal stresses to an appreciable extent (14).

3. The shear stresses developed on the Joint by the appli-
cation of stress pattern subsequently described will
be fairly uniform (1) (5).

The axis of rotation was chosen to be perpendicular to the
glue line for simplicity and practical reasons. }

A circular glue area Md to be used because in a plane perpen-
dicular to the axis of rotation, shear stresses act in a direction
at right angles to the radial lines and therefore would tend to
distort the shear plane if it was not circular. In the case of a
square or rectangular glued area, the mathematical expressions for
variation of stresses along x-and y-axes are:

7a.: 1‘1 (1 - 522) §

and 73': f2(1-é)5

in which 73' is the shear stress component along the y-axis and

‘2: the second component along the x-axis. After a mathematical

process based on the definition of twisting moment and by using

double integration, the maximum magnitude of the two shear stress

1

inuhich fzwill tendtodistort the glue area. Inthecase of

anelliptical glue line therewill againbef1 :% “323.21%;
fla T! b

in which the same discussion as in the case of the rectangular bar

°°IP°nents which-Ears f and f2 turn out to be f1 = g: and f2 = 5
a

holds. In other words, in only one special case in which the tangent
at every point on the boundary is at right angles to a line passing

3.5% one do
assoc on» scum hose semen.“ ma 3 moavsoufias coca no venom

use one» one. advocates: an: S canons—o nodule one A 8&2

 

 

 

 

 

 

 

 

[i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

u 5;
L. " 5/16 A/'6
H-——d / ,
// //
i T
' F , “It“? 1"“?
_ ‘ I I ‘ I,
. to: :—‘_:B~f5§u‘. ‘N r‘“* 1
ppppp — - .— f...- :>‘ m.» i e--—_. 0‘
er;
' l
' I
I
!
__ . .1
3
5' HIP/l\\ u
I \3/8

— ——_—..._.___.V._._. .

 

 

 

 

 

3/

 

Figure 2. Detailed dimensions of the three wooden members of
each specimen. Note the circular glue line.

 

through the center of the glue line, the glue line will not tend
to warp and this must consequently be a complete circle (see
Figures 1 “d 2)e
. As previously noted, a stress pattern was applied to the joint

designed for this study. This pattern was a combination of tor-
sional and simple shear stresses in.which the first one was
dominant. The torsional component of the stress pattern is computed
as follows:

C s 252

L r ' radius of circle
9" angle of twist
L 3 thickness of glue line
(:= shear strain
According to Ecok's law :
’6'; Gt

Z;=:toreional stress
G = shear’mcdulus

the torque
T = J{r7;da

I: as,
L
therefore
1' = 9—? It?“
the last integral is 43

jrzda=2|1 r3dr- 19-1:

32
a
da = ZTIrdr (see Figure 2a)
In the specimen used in this study, the glue area was confined
to an area located between two concentric circles with the diameters

d : 0.8' and D 3 1'. For this case:

10

 

 

Figure 2a.

Jrzda = £- (D4 - d4)
therefore
.-. can 1:4 - d"
r “32L ( )
The maximum torsional stress will be:

G_9.:. - a
242...: L rm" 9-ng

or finally
T = 173.6?
1‘mqalt D ’ d4)

The second part of the stress concerned, as previously
mentioned, is simple shear which is equal to the applied load
divided by the area subjected to stress, that is §= §.

The algebraic addition of the two above-mentioned stresses

would determine the final stress pattern as:

- -2 6
s. - gm; analyse,

in which the torque T would be the product of the load and the
torque arm chosen as 15 inches in this paper. The length of the
horizontal member for convenience of handling was 18 inches. The
vertical members which serve as supports for the Joint were 5%
inches long each. Both vertical and horizontal members were 2 inches
wide and B/B-inch thick. The thickness of 3/8-inch was chosen to
reduce the inertia effect of the horizontal member on the joint
due to lower weight, shich may particularly influence the accuracy
'of the fatigue data and also to minimise the magnitudes of moments
not considered in calculations. The magnitudes of these moments
were very small compared to the magnitude of total shear stress
applied, and they were negligible.

2 - EXTERIMENTAL PBDGEDQEE
2.1 - Selection and Preparation of Specimens

The lumber used for the specimens was airbdried and most of it
defect-free. This airbdried lumber was put through cutting and
planing process and the final dimensions of the members were pre-
.cisely obtained within the different machine accuracies employed.

Careful scrutinization was exercised to discard all abnormal
specimens before, as well as after they were assembled and glued
together. However, the only area of the members possessing great
importance in influencing the strength.was the first four squared

12

inches at the top of the L-shaped specimen where the circular glue
area was being cut out (see Figure 3). At this stage, the members
were marked and conditioned at a temperature of 70° F. and 65 percent
relative humidity.

The second step in preparing test samples, which obviously had
appreciable influence on the strength of the Joint, was to establish
the circular glue area as precise as practicable. The joint areas of
the horizontal member (4 squared inches each) were covered with a 2-inch
wide cellophane tape. The thickness of the ce110phane tape was in the
order of magnitude of a normal glue line (.002—inch). After the tape.
was applied the circular glue line was established (see Figure 3).

Care was taken to get the circular cut-outs located exactly
opposite each other. Otherwise, different torque arms would result
which comequently would produce erroneous data.

The glue under question was an "Aircraft Casein" manufactured by
National Casein Company. The glue was carefully prepared according
to manufacturer's instructions (13). Although the recommended pot
life of this brand was five hours, the glue used for this study was
applied from 20 minutes to almost 1&- hours after mixing.

lhen specimens were to be assembled, some of them contained some
defects that were missed in the first rejection period, or some imper-
fections were observed in the tape covering the joint such as tin
torn out spots visible with the naked eye. Embers missing in the
data, therefore, are discarded non-defect-free specimens.

Specimens were glued together and assembled in a Jig (Figure 1.).
The necessary pressure was applied on the joint by tightening a bolt

13

Figure 3. The members of the specimen. A part of the long member
has been cut off. The area of the Joint except the
circular area was covered by cellophane tape.

 

 

11.

Figure 4. Jig used to assemble the specimens.

 

 

 

15

which fitted through a hole at the center of the joint area with a torque
much. The diameter of this hole was limited to 5/16-inch in order to
elieinate any interference with the glue area. To improve the pressure
distribution over the joint area, a 2-inch diameter washer was used on
each side of the assembly. The pressure applied on the glue line was
200 pounds per square inch according to the following formula (2):
FL : M
2 11 D - H
where l‘ - force applied, in pounds

I. 3 lever arm, in inches

l = total load, in pounds

D = mean diameter of the screw,

in inches

K 8 pitch of thread, in inches
f s coefficient of friction

2.2 - Static Itesting

To accomplish the purpose of determining the fatigue characteris-
tic of the adhesive bond, the static strength of the Joint subjected
to the stress pattern previously described had to be found.

All static testings carried out in this investigation were perforsed
on a Baldwin Emery, SR-l. Testing lachine (See Appendix Ho. 11), at a
speed of .018 inches per sinute (see Figure 5). Since the platen of
the aachine was not large enough, a piece of hard saple lunber 1-3/4-
inch thick was properly bolted to the lower platen of the testing na-
chine extending almost five inches out. To one side of the maple lumber,
a clasp was bolted which would hold up the specimen. The clamp was care-
fully inspected prior to use. Since the pressure applied to vertical
members of the specimen would crush the wood, two pieces of rubber were
provided and used on both sides of the seaber subjected to crushing.

To record the ultimate load applied to the specinens to the
nearest half pound, a load cell (see Appendix No. 12) was enployed
in which the hundred pound range was used. On this load cell which
was overhanging from the tap, a belt which was sharpened at the

16

Figure 5. Baldwin her: 83% Testing Machine. Strain
Amplifier measures the strain on the top
edge of the horizontal member of the specimen
at 13.5 inches from the point of load
application.

 

 

 

17

Figure 6. Specilen mounted on the testing nachine.
lo'te load cell and clamp fixture.

 

 

 

18

lower end to a knife edge was used, which.would apply the load

at the determined spot of the horizontal member (see Figure 6).

As the lower platen started to move upward, at the specified speed,
the load would apply through the load cell and the bolt.

In static testing as well as fatigue testing, as illustrated
in.Tahle I, the torsional shear is considerably greater in magni-
tude than the sinple shear. Figure 7 shows the comparative
magnitude of component stresses as functions of applied load.

Table I - Comparative stress magnitudes fer the two
component stresses in stress pattern employed

 

 

 

 

fizz? EE- 1(zgzxg dz; ' Total Stress
_p.s.i. p.g&i. p.s,i,
5 8.93 324.6 413.33
10 17.86 61.8.8 666.66
15 26.79 973.2 999.99
20 35.71 1297;6 1333.31
25 44.64 1622.0 1666.61.
30 53.57 19%6.4 1999.97
35 62.50 2270.8 2333.30

40 71.43 2595.2 2666.63

.oohoaose shopped ensues on» no
upsenoosoo geese sands was 3333 no noun—pass seasoned—goo .N. 0553
Hmm 2H mmmfih 44.89

m3 a e H m3

 

19

\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0H

ON

on

0.x

SGHflCH HI CIVO'I

20

In testing a considerable number of specimens, adjustments
were to be made. Not infrequently, the specimens' two vertical
members were not exactly parallel and this would cause some de-
viations from the exact position of the specimen. It was decided
to reject these specimens prior to testing.

After failure at ultimate load, the failed joint was inspected
regarding the type of failure and the amount of mood torn out of
the wooden members. Note should be made of the fact that the
members of the failed specimens were still loosely attached
together.

Strain measurements were performed on the tension side of the
horizontal members at 13%- inches from the point of load application.
These measurements were taken in order that they might be compamd
with corresponding measurements on the fatigue specimens for
reasons to be explained later (see Figure 8).

2.3 - Fatigue Testing

Fatigue testing was carried out on a Sonntag Fatigue Testing
machine (Appendix No. 11.). This type of machine subjects the speci-
men to an oscillating force (constant amplitude).

Basically, the, machine consists of a heavy stationary platen,
a reciprocating assembly and a cabinet (see Figure 9). The station-
ary platen attached to the heavy frame of the machine is freely
suspended by springs with no solid mechanical connection with the
cabinet.

333.33 one 35.... 505 name». songs.“ use eggs m.“ age awakening...» .m 993m
moon mum madam—”80H! 2H Edy—um

I‘l

 

02. 80 00m 08 com com 8H 0
d
l
.

   

 

 

 

l r

-. w --i...|i..:-J 2
. .

_

“

 

l

L

 

 

21

 

 

 

 

 

rm...“

 

 

 

 

 

I
s...“_--..—-...._— 0..—__ - ._ .-_

I

l

T

1

I

F____ . v
i

l

I

L._._...

SGNflOcI NI (NO?

Figaro 9. Sonntag fatigue Testing Machine used in this
study. The bridge-Amplifier and oscilloscope were
used to check the vibration and measure the strain
created at 13.5 inches from the point of load
application.

23

The alternating force is produced by an unbalanced rotating
mass in a cage-like frame. This rotating mass would apply the de-
sired force, below or at the machine capacity, through the recip-
rocating frame at a constant speed of 1800 cycles per minute. The
rotating mass in the machine is directly driven by an 1800 rpm
synchronous electric motor which is connected to a 60 cycle power
supply.

The specimen was subjected only to the vertical component of
the centrifugal force since the horizontal component was absorbed
by four flexpplates which would guide the reciprocating assembly
in the vertical direction.

A compensator spring is positioned between the stationary'frame
and the vibrating assembly and absorbs all inertia forces. These
forces, in absence of the compensator spring, would be transmitted
to the specimen. The rigidity of this spring is not adjustable but
the machine could be brought to exact proper resonance by adding
weights to the reciprocating assembly. weights to be added were
specified by the manufacturer.

The machine is designed to be used with or without a pro-load.

In order to use the type of specimen described previously, a
special fixture had to be made. The improper structure of the fix-
ture could drastically influence the test results. Therefore, great
care has been exercised in making the fixture used in this investi-
gation.

The fixture consisted primarily of two parts. One part was
fastened to the heavy stationary platen and the second part bolted

to the top of the reciprocating assembly (see Figure 10). The ‘
previous part of the fixture consisted of four steel plates assembled
together by bolts so the vertical members of the specimen could be
tightly fastened. The second part of the fixture which was bolted

on the vibrating platen consisted of seven pieces, discarding the
bolts used to fasten the fixture to the platen. The horizontal part
of the specimen was inserted into the rectangular frame and fastened
tightly by two semi-cylindrical pieces of steel and two bolts. Posi-
tioning two small pieces of rubber between each semi-cylindrical

piece of steel and the specimen was considered an excellent practice.
Preliminary testing revealed that after a few thousand number of
cycles, the semi-cylindrical pieces of steel would press the wood
fibers down, causing local crushings , which consequently would loosen
the fixture and drop the load. The small pieces of rubber mentioned
above greatly eliminated this loosening of the fixture. The upper
part of the fixture which consisted of a frame-like piece of steel,
would rotate around the base and this rotation point had to be regu-
larly lubricated in order to facilitate its free motion with minimum
friction.

The machine was capable of subjecting the specimen to different
stresses such as complete tension, complete compression, etc.,
which has been described earlier.

To stay within the machine capacity, which was 200 pounds, con-
sidering 100 pounds static and 100 pounds created by eccentric rotat-
ing mass, the specimen was so designed that it was very likely to
fail within a range of 1.0 - 50 pounds. The reason for this type
of sample design was, first, to reduce glued area which would

25

Figure 10. Loading.!ixture, consisting of two separate
parts; one mounted on reciprocating platen
and the other attached to heavy stationary
platen.

 

 

 

26

eliminate the variables described and second, to avoid using load
multiplier.

The static preload was applied through a spring mentioned
previously as compensator spring. This spring, under preload,
would deflect and this deflection would be measured in inches to
the nearest .0005 of an inch. Deflection measurement was performed
by a dial (see Figure 9) which was attached to the stationary frame
and as the reciprocating assembly moved due to compensator spring,
the plunger of the dial would register the deflection. To concert
inches into pounds, a conversion factor of .69 pounds was given, by
the manufacturer, for every .001 inches.

To apply preload, the dial reading was recorded prior to in-
serting the specimen in the fixture. This reading would correspond
to zero load. After inserting the specimen, the desired preload
would be applied to the specimen through two wheels constructed in
the machine for this purpose. Cut-off switches were provided in the
machine. These switches would turn the machine off when deflection
increased one-half inch. Preliminary testing showed that this
amount of deflection corresponds to a broken specimen.

The maximum allowable movements of the reciprocating platen
with and without full preload were i .375 and t 0.225 inches,
respectively, when the specimen was still rigid.

To accomplish a satisfactory fatigue test, it was decided to
determine the effect of creep as well as any wood crushing influence
under the semi-cylindrical parts of the fixture. This was performed

27

through observations in diminishing the static preload applied to

the specimens, as well as the magnitude of strain recorded by
oscilloscope. Eventually it was found out that some readjustments
were necessary in order to maintain the stress at the assigned level.
Figure 11 illustrates this effect on specimens subjected to different
stress levels and Figure 12 shows the adjustments made in order for
the stress to remain at the assigned level.

In fatigue testing, five distinct stress levels were predeter-
mined which ranged from almost 83 to about 28 per cent of average
static ultimate stress of the matched specimens. These five stress
levels were 2000 p.s.i., 1666.61. p.s.i., 1333.31 p.s.i., 1000 p.s.i
and 666.66 p.s.i. At each stress level 10 specimens were tested.
One half of the load was static and applied through the compensator
spring. The other half was generated by the rotating mass. The
total load applied to the specimens would oscillate from zero to
total load; Figure 13 shows that the load was oscillating in a
sinusoidal pattern ranging from core to maximum assigned load.

Each centimeter on the oscilloscope plate corresponded to 1m
microinohes per inch of strain. Aw drop in load would be revealed
by this curve because the amplitude of the curve would, of course,
decrease. The strain gages were positioned at the same spot as
static specimens, that is, 13.5 inches from the point of load
application.

Adjustment curves are shown by Figure 12. As illustrated,
the stress drop ranges from 11.0 to 70 pounds per square inch for
the first thousand cycles. All stress levels approached the
desired constant stress level after 15,000 cycles. The lowest

28

.3th: nachos escape condenseuosn 33 no.“ 35.3 on» 5 nacho
was sneak—G one .3 mousse mafia—fie woos 53 o» 26 no.8 weaken .4...” 056.2

mum—owe ho mg

 

 

 

.3 . ., a
ofo j 3 43m :1” o
. a

- ,..---i L", i 03

 

 

 

 

 

 

 

 

 

 

coma

 

 

 

/
/
/

 

 

 

 

 

 

 

Doom

 

19:! HI 893318 '1le

29

08.5

coofi

.893. 22mg so

assume moose use «massage woos no scavenged-cc weakens assume psoaanswu< .NH shaman

macs no mamas:
823

 

000m 060m coma .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ooq

OONH

ISH HI SSEHIS TVIOL

30

Figure 13. Oscillation of strain as recorded by oscilloscope.
Each side of a square corresponds to 100 micro-
inches per inch of strain.

 

 

 

 

 

31

stress level employed, that is, 666.66 pounds per square inch,
approach the constant magnitude equal to the amount mentioned,
after 10,000 cycles.

Figure 11 shows the drop of stress without adjustment. As is
readily concluded from this figure, if the adjustment was not per-
formed, the stress would drop as much as almost 700 pounds per square
inch. The curves were predetermined by using two specimens at each
level and measuring the drop for five stress levels employed. Note
that at all stress levels the stress drop readings for two specimens
used at each level were very close. Average of stress drop readings
was obtained and Figure 12 was established.

Strain measurements (figure 8) for fatigue testing recorded the
same readings as those taken through static testings. The coincident
of the strain measurements for both fatigue and static testing implies
that there was no load tabs-up in the fatigue machine.

32

- RESUDT

3.1 - Static Testing
Twenty-five specimens were subjected to static testing in order

to determine the static strength of the joint. Results obtained

are shown in Table II.

3.2 - Fatigue Testing
Table III shows the number of cycles, that is, the life of each

individual specimen subjected to fatigue stressing at the predeter-
mined stress levels. The five stress levels, as shown in this Table,
range from 83.39 to 27.80 per cent of average ultimate static stress

of specimens.

33

A: enema eeav 93.5.3 cool. no M55: as encompass 3" sneak on

flu. eased. 23 33838.5? 3398. .5. co a3- 5 2.53 e85.

 

 

 

 

33 3348 13:8... sacs

3.39.3

 

 

 

 

 

.9533. 393.. 5 seasona-
ufleaoagoo has... n: 35% 3.3.5.80 use 8282" $8 - H .28

o o «5.8? 3.3mm «N13 0.0m onca
3v 0 seamen $.88 name «8m 23
3V 0 Sofia 3.2.3 .860 «fin wmua
QC 0.0 3.3.3 3.3mm S. 3 m .3 band
0.? ed $.33 3. 3.x... 3.5 06m on...”
Q? 3v 923$ ms. «mum 3.3 «.mm mm...”
o o 3.8%“ onfimmu mu. .3 «:3 3A
0." v o 00.3mm «QRNN in. mém main
0 QC 86.3“ 3”. mama 333 men mu..."
oav 0.0 «0.3.3 «0.0qu 3.3 mfim and
ed o3 3.33 3.3.)... 3.3 9mm omna
ca v . o 8.3% 3.33 323 m .3 3....”
ca v 3v oo.om- on.mn- 8. .3 «:3 3..”
o 0.3 3.33 «N. menu on. no n .3 5nd
ed 3v $.me 3.88 3.8 9mm 3A
3v o 8.88 8.3m: Ran can «TH
SV SV 8.38 8:83 «1:. 8.3 «.3
0.9 3v 3.3.3 3.3mm S. .3 men flu...“
cm 6qu Nnémmm 3.3 9% Swan
3.3mm «wanna main» 0.0m o...”
be .98 on :Rom .933 N. am a..."
oo.mm..N 3.3 No.3 0.3 b..."
3.33 aflommw 49$ 9% can
8.2mm 00.3mm 360 >.mm .TH
3.33.. 3. Ram 3.3 men mu."
eHeOem eHeUeQ eHeCe“
30.3. asses deck—s seen. 393. seen- . 3H .33 needs:
a .35. 32:88 558%

34

Figure 14. A specimen failed in static testing.

”no-

 

 

 

nil

35

Table III. Number of cycles obtained tor five different
stress levels in fatigue testing. Preload was
equal to half of the total load applied.*

 

 

 

 

 

 

 

 

 

Specimen Total Shear‘gtress Number of cycles
number’ i of average ultimate to failure
static stress .p.s.i. “

2-2 83.39 2000 16,000
2~4 I “ 6,000
2-5 ' n 39 COO
2-6 " " 6,000
2-7 " " 17,000
2-8 ' " 5,000
2-9 . a 7.000
2-10 " ” 7,000
2-11 " " 7,000
2.12 II n 7, 000
2-13 69.49 1666.64 23,000
2-u. I " 34,000
2-1 " n 9: 000
2-1 " “ 27,000
2-17 I ll 69, m0
2.18 II a 36, m0
2-19 " " 12,000
2-20 " ' 67,000
2-21 ' ' 899 M
2-22 " ' 21,000
2-24 55.92 1333.31 14,000
2-28 " ' 27,000
2-29 ' " 49,000
2-31 I ' 76,000
2-32 I n m, m
2-33 " ' 86,000
2-35 ” ‘ 102,000
2-36 II fl “6’ m0
2'37 :1 ' 39, m
2-38 ' ' 64,000
2-39 41.69 1000 110, 000
2-40 " ' 550,000
2-41 ' ' 415,000
2.42 ' ' 1,228,000
2-43 ' I 236,000
2-44. ' ' 1,051,000
2:45 " ” 458,000
2-46 ' ' 1,358,000
2947 ' ' 3,850,000
2a48 ' ' 80,

2-49 27.80 666.66 2,640,000
2-51 " " 6,841,000
2-52 " ' 4,861,000
2-55 " " 4,911,000
2-56 I ' 6,226,000
2-57 " ' 8,039,000
2-58 - .. 9,438,000
22%; I I 2, 50,000
2 I! I! 2, , 000
2- n n 5, ,

 

 

 

 

 

 

*In almost all specimens subjected 3: fatiEEe testing, no weed
failure could be observed (see Figure 15).

36

Figure'l5. A. specimen failed in fatigue testing,
shoving pure glue failure.

 

 

«'5 have".

37

3.3 - moisture Content

The specimens were conditioned at 70° F. and 65 per cent rela-
tive humidity. moisture content measurements showed that the speci-
mens after being in.the conditioning room for one week to ten days,
approached a moisture content of 6 - 7.8 per cent. At this stage,
they were taken to the static testing room in polyethylene bags and
tested within the next hour. But in the case of fatigue testing,
sometimes the specimens would stay in the machine for as long a
time as about eighty hours in the laboratory where the temperature»
and humidity ‘were not controlled. It was,therefore,decided to
measure the moisture content change during the fatigue test. From
the measurements it was determined that the moisture content would
drop from 6 - 7.8 per cent at the beginning to 5.3 - 5.8 per cent
after staying in the machine for 80 hours; from 6 - 7.8 per cent to
5.6 - 6 percent after 48 hours. Attention should be paid to the fact
that all mentioned figures are averages. The moisture content deter-
mination samples were taken from the ends of horizontal and vertical
members, which would add up to three moisture content determination
samples for each specimen; the average of these three has been consi-
dered as the moisture content of the specific fatigue or static test—
ing specimen. It would have been more desirable to determine the
moisture content of the specimens closer to the glue line but no
feasible way of performing such a task could be devised. All
moisture content measurements were taken through oven drying method

at102t2°c.

38

- ANALISI 0F RESUETS

One of the important results of this study is the fact that
fatigue stressing induces failure in the glue line even at rela-
tively low stress levels.

In fatigue testing, all specimens showed little or no wood failure
after'being failed in the machine. Among the specimens which showed
wood failure, it had occurred at the point of maximum stress. There
‘were, however, some specimens which had a few wood fibers torn.out at
other areas of the circular adhesive line.

In fatigue testing, two variables were considered. These two
variables were stress level and the number of cycles to failure. The
first variable was held constant and the second variable was dependent
on the stress level. The functional dependence of two variables can
be represented in the form of a so-called SPN'curve (Figure 16).

As noted in Table III, when the stress level drops from about
56 per cent to about 42 per cent of the ayerage ultimate static stress,
the life of the specimen increases appreciably and when.the stress is
almost 28 per cent of the total ultimate static stress of the joint,
the resistance of the joint to fatigue stressing approaches as high a
magnitude as 9.4 million cycles. In many fatigue experiments, after
ten million cycles, the specimen is taken away from the machine
assuming the endurance limit of the material. On this basis, it is
very likely to approach such a limit fer casein if the stress level
drops to about 20 per sent. This, however, is only an.estimation and
in this study such a limit haauot been approached. The life of the

specimen at higher stresses is relatively low.

O
H

mango no 556... 3....» «Sense no.“ cocoaeeoo ego sum .3 enema

 

.__..__4

“—5

 

 

 

Y
D
. __ __ .__.._ _1-_.______..._.

_.. —-_..——.——.~-—— T___.__._—_.._.

 

hm: . Loa- Lmoa - 0.03

E ._ .. ....-
“_ . m w
.

 

w-._-——— __..

 

 

 

 

 

 

 

 

 

-4-— .__... a-4.- 4

.,__

.mmm

 

 

ii-.
_ n

 

 

——— --_——. —--{y - _—*

 

 

 

  

 

 

 

 

 

 

 

39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 -45 r -5LI. .m- fi55 - _-.- L- -“v-5.r-55.-55-.. 55 5... L a.
a “.i: 4.. . . .T _. . . g . l
. . . _ _.. i” . . . . . . J m
fl _ _ . d_ in l M . _. _
r" H V H . . . M _.w m H .1 m._ .
:7 . ., _._ , .
- e75- 5.7.5,----.5---.-:.i ,. ...... - 5.. .2...
. _. . . _ , .__. . 1 . . .1
_ . _ _ fl . w ... l * H. _ m
. . . _ _. l . _ H. . . I
_ . . m L. _ r _ .. l .
. . fl. . . .. u . . .1 4 w.
. A . - 81., _ we .-e of... .17.. i... .: -5-..--.-.-..- 82 m
. . w .... _. . H. ._ w _ m . u; . . m
. . . . .7 . . _ .1. . . . _ , .: m x
...w .f _ f... 1.... m
. _ _ _. _ . .L... . ;_ _
u _ . _. .. .. _ .w_ . .. . .u. . M
4 a T 5. 5 .75...--5..F--T-.8 - - -. ¢.¢--.¢ .0 -5- Av- . 1--..515-5 m.nmmHI
. .. ._._.
l " fl n H _ m ._ .w. . n . _ . .1“ _
. M . 1 W ... u W W . T H . . : . _
... .:.._ .f... e ..._
55-15.55- 14-111....H T- i -- 4..o Neon-18% o . - bell-P .83
. . LT... 1.. . ..;.
.. _ . " . _ ._ u a ., .
. .a:. . :_: . .j
-. i.....--: -_ - .-: 7 :5 .-
. . . . g.. , ....
n _ . w . __ ..
. fl . _ . . .
.cifi --.-9- .fi._ 1#-. - L.. 5 t--- - a; -w

 

 

 

 

 

 

 

 

 

 

 

 

 

 

40

Figure 16 illustrates the S-N curve derived. Acceptable statisti-
cal procedures (10) were employed in derivation of this curve. lots
that the slope of this curve is decreasing as the stress level steps
down and this slope may approach zero if the stress level drops down
to 400 to 500 pounds per square inch.

It is deserving of mention that sometimes the fatigue data
could be represented by a so-called response curve if the experiment
is properly designed. In this type of representation, the probabili-
ties of specimen survival (or failure) to a constant number of cycles,
at pro-determined stress levels, are computed. Response curves are
particularly useful in determination of endurance limit of the ma-
terials to fatigue stressing. In this study the time limitation
did not permit such determination.

During fatigue testing, the machine oscillation was constantly
measured by the sine curve previously described, which simultaneously
would record the strain created at the point. Another way of checking
the load was the dial mounted on the compensator spring which would
well reveal aw decrease in the preload applied to the specimen.

The standard deviation in static testing was ‘9: 160.8 pounds per
square inch. The average ultimate static stress was 2398.37 pounds
per square inch. It could readily be concluded that this value is
sipificantly higher than the value given by standard maple block test
(see Table II). Furthermore, in the latter, very often there is a con-
siderable percentage of wood failure whereas in specimens used in this
investigation the percentage of wood failure is appreciably lower.

The grain direction on the Joint was not parallel on the glue
line and this has been concluded as having some iderable influence

41

on the glue behavior and resistance (6) to load but without further
interpretation of this effect, it is obvious from the data taken
that this method of testing is very likely to give higher values
of shear stress for adhesives. .

The members of each failed specimen were loosely attached to-
gether and to see the Joint, the members were to be pulled apart by
hand. On this basis, each static-tested specimen member was taken
apart and the amount of glue and wood failure was estimated. Table
II, Columns 6 and 7, represent the per cent wood failure which
occurred in the Joint.

The wood failure was very often a few fibers torn off from both
the vertical and horizontal membersand most of the time from B to D
on the circular glue line and, not infrequently, around point e (see
Figure 2). This seems to be due to the fact that the two components
of the stress pattern are added together at this point and this small
area around 0 has very significant effect on the resistance of the
Joint to the load applied. The line across the glue line at A is
subjected to least amount of shear, since the two components of the
stress pattern, that is, torsional and simple, are subtracted at this
point. At point A, there was very little evidence of the existence
of wood failure.

5 - CONCLU§ICm§

5.1 - Conclusion Derived

0n the basis of tests discussed herein:

a) using a specimen so designed that failure could be induced in
the glue line both in static and fatigue loading

b) employing a stress pattern composed of torsional and simple shear
in which the latter had very low magnitude in comparison with the
torsional shear component 5

c) using an Aircraft Casein glue _

d) restricting the glue line to a very small area

e) testing specimens in fatigue stressing by means of a constant
amplitude fatigue machine with special fixture made to suit the
designed specimen and staying within the machine capacity

f) testing the specimens in static loading

it appears that the following conclusions are warranted:

1. That casein bonded wooden Joints could withstand
combined shear stresses.

2. That the casein bonded wooden Joints as described in
this paper gave a higher torsional shear stress than
that of standard maple block shear specimen conven-
tionally used.

3. That an exponential curve is resulted in fatigue
testing as stress level is plotted versus the number
of stress cycles applied to the specimen to failure.

4. That at high stresses the fatigue life of the specimen

is relatively low.

43

5. That at low stresses the fatigue life is appreciably
high. _

6. That stepping down the stress levels from about 83
to 28 per cent of average ultimate static stress at
five equal increments, endurance limit or fatigue
strength of the material was not approached.

7. Load adJustments may have had a small influence on
the fatigue life at the beginning of the test.

There did not seem to result a significant amount of wood fail-
ure in static and fatigue testings. The fact that there was little
or no wood failure on the Joints indicates that in this method of
specimen design, the wood species could resist more cycles of fatigue
stressing than the glue at all stress levels employed. It could
therefore be concluded that the S-N curve for wood, if plotted, falls
somewhat above that of glue (see Figure 17A) .

If the Joint was made of a strong adhesive such as phenol for-
maldehyde, the situation shown in Figure 178 would be likely to occur.
The fatigue failure in this case would have occurred in the wood ad-
Jacent to the bond. Figures 170 and 170 illustrate other possibilities
in which the wood, up to some extent, yields longer life in fatigue

stressing, and from there on, glue produces longer life, or vice versa.

5.2 - Present Concept on Fatigue Failure

There are many schools of thought concerning fatigue failure
in engineering materials, particularly metals, because this class of
materials has been of great concern to this age of modern civilisa-

tion. There has, therefore, been a tremendous number of tests

.Comumfinaoo oooaiosflm new metro ziw 303653? .3 when:

ADV .

 

 

Am.

 

 

52

B.

 

[#3 NO

 

:3

 

 

1.5

carried out on metals since the last century. There is not, however,
a very consistent concept on the cause of this phenomenon as yet and
due to statistical nature of fatigue, there is not an entirely satis-
factory limit at which the life of the structure or machine parts
could be confidently guaranteed for a long time.

This part of the paper is devoted to some present theories on
the cause of fatigue failure and their possible relation to the ma-
terials used throughout this investigation.

Hystersis is a term referring to the loop formed when a non-
perfect elastic member is subjected to load and its removal. In
other words, the points of stress-strain curve in loading do not
coincide with those of unloading the member; consequently they form
a loop (7) (11) (8). This phenomenon caused by the loss of energy
in the load-unloading cycle will form a pennanent deformation. Barstow
(7) «messed the opinion that when a material acts in a purely elas-
tic manner, that is, forms no hystersis loop in the load-deformation
.cycle, fatigue would not occur. he did not , however, subject his
specimens to a larger number of cycles. In materials in which hyster-
sis loop results, in each cycle, the loss of energy would cause gradual
deterioration effect of repeated stresses.

Molecular slip is another explanation of fatigue failure. Ihen
the elastic strength of individual crystals in metals exceeds by load
application, localised deformations develop. These deformations
within the crystal are referred to as slip bands (8). These bands, as
the repetition of stress proceeds, cover the entire surface of the
crystal (11). There is not a well-expressed opinion on the occurance

1.6

of such phenomenon in the case of adhesives. It is, however, believed
that such slippage may take place between the molecules of the region
under repeated stresses on the glue line.

The existence of secondary stresses is another fatigue failure
interpretation expressed by experts.

According to this theory the actual strength of solid materials
does not have the same magnitude as the strength they should possess
by reason of molecular cohesive forces (11). Note should be made of
the fact thatnaterials are apt to contain very'tiny defects when they
are made, either naturally, as in trees, or artifically, as in metals
or adhesives. These subnmicroscopic defects form favorable spots for
localizing fatigue failure. These defects could. also present some
irregularities in the material and eventually raise the stress at
that point. Internal stresses definitely contribute to the failure
of the material below the expected point.

This last theory could be applicable to a considerable extent on
adhesives. It has been determined in experiments that the thinner
glue line gives a higher strength value than a thick line. One of
the reasons claimed for this phenomenon is that of increased possi-
bility of getting internal defects on the glue bond. In other words,
as a thicker glue line is applied, more defects develop, which would
fail the bond at a lower strength value than the thin bond. Through-
.out this paper it has been assumed that the glue line possesses a
thickness almost equal to that of the tape covering the entire joint
except for the circular line, because as the pressure was applied, the
members were practically brought tight together. This thickness was

0.002 inch. The thickness in this case is relatively small.

47

Consequently there exists less possibility of establishing defects
in the glue line. But there are,undoubtedly, still sub-microscopic
defects on the adhesive line. In addition, in the preparation of
the adhesive which was to be applied to the specimens, foreign parti-
cles may have been introduced which would subsequently contaminate
the prepared glue. These foreign rigid particles would function as
stress raisers on the glue line.

The surface of wood is never smooth. This non-smoothness of
the wood surface on the glue area would cause irregularity on the
borders between the member and the glue, eventually establishing
stress concentration at these areas.

The stress raisers mentioned above will actually cause the bond

to fail when subjected to repeated stresses.

1.

2.

4.
5.

10.

11.

12.

13.

15.

16..

17.

m1!
war or APPARATUS

Hard maple of stock from which specimens were prepared
contained about 5 growth increments per inch.

Casein glue manufactured by National Casein Company, sold
under the name "Aircraft Casein".

Cellophane tape,0.002-inch thick; manufactured by Minnesota
Mining and Manufacturing Company.

Compass. Included in drawing set, made by Eugene Dietzgen Co.

Band Saw. Model 3-18388, 36-inch bandsaw made by Iates-
American Machine Company, Beloit, Wisconsin.

Circular Saw. lanufactured by the Tannewitz Works,
Grand Rapids, Michigan.

Planer. lenufactured by Boice-Crane Company.
J ointer. Squared Industrial Control Division, Kilwaukeefllisconsin

Drying Oven. Temperature range up to 200° C, manufactured by
Precision Scientific Company.

Strain Gages. SR-l. strain gages, type A-l, with a resistance
of 119.61: 0.2 ohms and gage factor of 2.04i 11. Gages made
by the Baldwin-Lina—Hamilton Corporation, Waltham 54, Mass.

Static Testing Machine. Baldwin-Emery, SR-l. Testing hehine,
lodel F6T, made by the Baldwin-Lima-Hamilton Corporation,
Philadelphia, Pennsylvania.

Load Cell. Baldwin 83-1. load cell, manufactured by Baldwin-
Lima-Hamilton Corporation, Philadelphia.

Strain Amplifier. Baldwin SR-A, type u, Portable Strain Indi-
cator manufactured by the Baldwin-Lima-Hamilton Corporation,
Philadelphia.

Fatigue Testing hehine. Sonntag Universal Fatigue Testing
Machine, Model SP-Ol-U, manufactured by the Sonntag Scientific
Corporation, Greenwich, Connecticut.

Bridge-Amplifier. Iodel BA-l, manufactured by Ellis Associates,
Pelhau, New York.

Oscilloscope. Type 53/54 n, Plug-in Unit, manufactured by
Tetronix Incorporated, Portland, Oregon.

Calculator. Merchant Calculator, manufactured by Merchant
Division of with-Corona Marchant, Inc., Oakland, California

 

49

LIT URE CH

1. Clark, D.A.R., Advanced Strength of Materials, Blackie and Son,
Limited, London and Glasgow, 1951.

2. Press, A.D. and Selbo, M.L., "Fabrication and Design of Glued
Laminated Wood Structural Members." Tech. Bull. No. 1069, U.S.
Forest Products Laboratory, 1951..

3. Freudenthal, A. 11., "Planning and Interpretation of Fatigue Tests."
Symposium on Statistical Aspects of Fatigue, American Society
for Testing Materials, SIP No. 121, 1951.

1.. . "Adhesives and Sealants in Building.", Building
Research Institute, Division of Eng. and Ind. Research,

Washington, D.C., 1958.

5. Harris, 0. 6., Introduction to Stress Analysis, The Macmillan
Company, New York, 1959.

 

6. Lewis, N. 0., “Fatigue of Wood and Glued Joints used in -
Laminated Construction, " Proceedings Forest Products Research
Society, p. 221, 1951.

7. Moore, M. F. and Kommers, J.B., The Fatigue of Metals.
McGraw-Hill Book Company, Inc., New York, 1927.

8. Sines, G. and Heisman, J .L., Metal Fatigue. The University
of California' s Engineering Extension Series, McGraw-Hill
Book Company, New York, 1959.

9. . "Resistance of Several Types of Glue in Hood
Joints to Fatigue Stressing'. United States Forest Products
Laboratory, F.P.L. Mimeo. 1539, 1955.

10. . 'Ianual on Fatigue Testing,‘ American Society
for Testing Materials, Sl'P lo. 91, 1950.

11. Sullivan, J. D., "The Effect of Fatigue Stress on Shear in wood.”
Dissertation for the degree of Doctor of Philosoplw, Michigan
State University, 1958.

12. leibull, I. and Odquist, K. F. G., Colloquium on Fatigue, *
International Union of Theoretical and Applied Mechanics.
Harlin, 1956.

13. . Instructions for Use of National Aircraft Casein Glue,
National Casein Compazw, Chicago, 1959.

11.. Iavorsky, J. M., Cunningham, J. 3., and Hundley, N. G., "Survey
of Factors Affecting Strength Tests of Glued Joints,‘ Journal of
Forest Products Research Society, Vol. 5, No. 5, 1955.

15. . Food Handbook, Forest Products Laboratory, Agricul-
tural Handbook No. '72. U. S.Government Printing Office,
“8W3, D.C., 1955s

 

 

I.I| , a. ' i . 15‘ 2‘ fi
RQOM UEE Ova-L3]

 

W“bh

n1 '

J. L I‘VV‘ I: .33.: fl? Pv". . ‘W‘u‘: g v
E _. " r ‘ v '

 

 

 

muumlwmuuumuImlumwugluuymm:I

293 0