N 63 '4250

·NASA TN D-1574

TECHNICAL NOTE
0-1574
FATIGUE BEHAVIOR OF MATERIALS UNDER STRAIN CYCLING

IN LOW AND INTERMEDIATE LIFE RANGE

By Robert W. SInith, Marvin H. Hirschberg,

and S. S. Manson

Lewis Research Center

Cleveland, Ohio

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

WASHINGTON April 1963
 NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

TECHNICAL NOTE D-1574

FATIGUE BEHAVIOR OF MATERIALS UNDER STRAIN CYCLING

IN LOW AND INTERMEDIATE LIFE RANGE

By Robert W. Smith) Marvin H. Hirschberg)

and S. S. Manson

SUMMARY

A series of constant strain range tests was made for a wide variety of mate-
rials producing fatigue lives varying from a few cycles to about one million cy-
cles. The specimens were subjected to axial, compression-tension, low-frequency
fatigue about a zero mean strain. Load range was measured periodically through-
out each test, enabling an analysis of fatigue results in terms of elastic, plas-
tic, and total strains. Materials tested were AlSI 4130 (soft and hard),
AlSI 4340 (annealed and hard), AlSI 52100, AlSI 304 ELC (annealed and hard),
AlSI 310 (annealed), AM 350 (annealed and hard), Inconel X, titanium (6Al-4V),
2014-T6, 5456-H311, and 1100 aluminum, and beryllium.

During strain cycling, load range generally changes during the very early
part of the test and then settles down to a fairly constant value for most of the
fatigue life. Cyclic strain hardening or softening causes the observed load
change and produces cyclic stress-strain relations that often differ substan-
tially from the virgin tensile flow curve. These comparisons are made for each
of the test materials.

Fatigue-life relations between elastic, plastic, and total strain components

were established. For metallurgically stable materials, straight-line fits of
the logarithmic elastic strain-life and plastic strain-life data produce a rela-
tion that agrees well with the total strain-life data. The strain-life data are
also used to explain changes in susceptibility to stress concentrations over a
large life span and to rate correctly the relative notch sensitivities of the
four test materials that experienced .nontest-section failures in comparison with
the other materials.

Relative performance of the test materials is illustrated on the basis of

both strain range and stress range over a life span ranging from a few cycles to
about one million cycles.
 INTRODUCTION

In recent years there has been an appreciable effort to incorporate low-

cycle fatigue data obtained at various laboratories into fatigue design proce-
dures. Manson (ref. 1) and Coffin (ref. 2) independently suggested that low-
cycle fatigue life for a specific material is directly proportional to a power of
the cyclic plastic strain (a straight line on a log-log plot of plastic strain
against cycles to failure). Coffin (ref. 3) discussed the application of this
criterion to design after extensive testing of type 347 stainless steel in both
constrained thermal cycling and constant-temperature strain cycling. Correlated
experimental data were developed at four laboratories to establish parameters
governing pressure-vessel design with respect to the plastic fatigue characteris-
tics of the material (ref. 4). In this analysis use was made of tota~ (elastic
plus plastic) strain range test data. Manson (ref. 5) has related fatigue life
to the elastic as well as the plastic strain range components of the total me-
chanical strain range) thereby producing one relation suitable for cyclic lives
of approximately 10 to 10 6 cycles. It has also been pointed out (ref. 6) that in
elastic-plastic stress analysis of fatigue problems there is a definite need for
knowledge of the relation between stress range (or amplitude) and strain range
(or amplitude) during strain cycling. Most recently, Langer (ref. 7) has de-
scribed a pressure-vessel design procedure using a stress amplitude-life equation
based on two factors: an empirical relation between plastic strain and tensile
ductility, and the endurance limit.

In order to evaluate present design procedures, to develop new methods if

necessary, and to increase the understanding of the stress-strain - life rela-
tions during fatigue, it was believed desirable to obtain detailed fatigue test
data for a wide variety of ductile materials using axial, compression-tension,
low-frequency fatigue machines in which both load and deformation were measured
periodically throughout the test. The first phase of such a program) reported
herein, provides the basic information obtained from room-temperature constant
diametral strain range tests with zero mean strain. The following desired test
information was obtained:

(1) Behavior of load range during cycling at constant strain amplitude

(2) Fatigue behavior for span of fatigue life ranging from a few cycles to
about one million cycles

(3) Data for a selection of materials in which there is a wide variation of

chemical and metallurgical composition

(4) Data for a selection of materials in which there is a wide variation in

elastic and mechanical properties (elastic modulus, yield strength, ul-
timate strength, and ductility)

Such information is then used to

(1) Determine the cyclic stress-strain relations necessary to the stress

analyst for fatigue analysis

2
 (2) Determine the elastic strain (or stress), the plastic strain, and the
total strain range as a function of fatigue life

(3) Compare relative performance of materials on a basis of strain range and

stress range

(4) Illustrate the use of strain~life relations to indicate relative notch

sensitivity of the materials

MATERIALS , APPARATUS, .AND PROCEDURE

Materials Tested

The nominal chemical composition, processing condition, and hardness of each

test material are tabulated in table I. The test materials were three ferritic
alloy steels: AISI 4130, AlSI 4340, and AISI 52100; three austenitic heat-
resisting steels: AlSI 304 (extra low carbon), AlSI 310, and AM 350; one heat-
resisting nickel-base alloy: Inconel X; three types of aluminum: 2014-T6,
5456-H3ll, and 1100; a 6Al-4V titanium alloy; and structural grade QMV beryl-
lium. The steels AlSI 4130, 4340, and 304 ELC, and AM 350 were tested in both
soft and hard conditions. Mechanical properties for these materials (table II)
were measured at room temperature.

Specimen Configuration

The fatigue test specimens (£igs. lea) and (b» were bars, circular in cross
section having an hourglass-shaped test section with a minimum diameter of 0.25
inch, unless otherwise noted in table III. AlSI 4130 and 52100 steels were fab-
ricated from 1/2-inch-diameter blanks into fatigue specimens as shown in fig-
ure l(b). All the rest of the materials were machined out of 3/4-inch-diameter
blanks into buttonhead fatigue specimens as shown in figure lea) (or fig. l(c)
for beryllium only). Separate buttonheads were screwed onto the threaded-head
specimens so that the same style of grips could be used for all materials. It
was necessary to use a modified specimen configuration for certain very short
life tests to reduce the buckling problem that developed in the more ductile
materials at large diametral strain ranges. The most common modification in-
volved the reduction of the hourglass radius to 1. 0 inch and the overall specimen
length to 2.25 inches. In a few tests of AlSI 4130 (annealed and hard) and
Inconel X the hourglass radius was ~urther reduced to 0.5 inch. Another modifi-
cation was necessary to prevent failures outside the test section for some mate-
rials under certain test conditions. In this case the minimum test-section diam-
eter was reduced to 0.21 or 0.18 inch as circumstances dictated.

A cylindrical test section (fig. l(c» was used to make the longitudinal
strain measurements necessary for elastic modulus determinations. Beryllium fa-
tigue specimens were made of this configuration also to allow longitudinal strain
control instead of diametral. This procedure was necessary because the extremely

3
low value of Poisson's ratio (~ = 0.024) for beryllium means that the elastic di-
ametrical strains are very small and, therefore, more difficult to measure accu-
rately.

Test Apparatus

Four low-frequency mechanical fatigue testing machines (fig. 2) designed and

built at Lewis Research Center were used for this test program. Alternate push
and pull forces on the specimen were supplied by a 6-inch-diameter hydraulic cyl-
inder. The buttonhead specimens were attached rigidly to the loading rods with
split-cone and wedge-type grips. A commercial die set was used to maintain
alinement under the action of compressive loads. The lower loading rod was at-
tached in series to the lower movable die-set platen, the commercial load cell,
and the hydraulically operated piston rod.

Strain range control was obtained by a lO-to-l deflection lever attached to

the moving platen. This lever actuated microswitches which energized relays that
control a solenoid-operated four-way hydraulic valve. The valve transferred
high-pressure oil to the opposite side of the double-acting piston and vented the
unpressurized side to the supply tank. Cycling rate could be varied from as low
as 2 or 3 cycles per minute to about 30 cycles per minute.

A strip-chart recorder was used to make either continuous or periodic rec-

ords of load amplitude. A special circuit built into the recorder for this pur-
pose powered the load cell which, in turn, supplied the load signals to the re-
corder for amplification.

Strain measurements were made with Tuckerman optical strain gages for all
tests in which the diametral strain range fell below the maximum practical capa-
bility (Ed = 0.034 in./in. or less) of the l-inch-gage-length extensometer.
(Symbols are defined in appendix A.) For larger strain ranges, a dial indicator
type of diametral extensometer was used. Tuckerman optical strain gages were
selected because of their reliability for a practically unlimited number of cy-
cles at large as well as small strains, and because their excellent sensitivity
is particularly desirable for diametral strain measurements. With these gages it
is possible to discern a strain of 8 microinches per inch over a l/4-inch gage
length. Diametral strain measurements were made using two l-inch extensometers
fastened with piano-wire springs to a special diametral strain-gage fixture
(figs. 3(a) and (b)). Two diametrically opposed aluminum bearing edges were rig-
idly fastened to a U-shaped leaf spring which pressed the bearing edges against
the test section. Long flexible loops of piano wire attached to the loading rod
stabilized the strain-gage fixture against tipping during the fatigue test.
Longitudinal strain measurements, used for the beryllium fatigue tests and for
all elastic modulus measurements, were made by fastening two l-inch-gage-length
extensometers directly to specimens having cylindrical test sections. All opti-
cal strain readings were made with the use of the Tuckerman autocollimators.

4
 Procedure

Determination of mechanical properties. - The usual procedure involved the

measurement of tensile and elastic properties before fatigue testing each test
material. For tensile testing, standard hourglass fatigue specimens were in-
stalled in a Rhiele universal testing machine and were fitted with a dial-gage-
type diametral extensometer. .AB the tensile load was increased, simultaneous
readings of the dial gage and load indicator were recorded. With these data the
true stress - true strain relation beyond the yield point was established. Other
quantities measured and used to calculate ultimate strength, fracture strength,
and ductility were maximum load, fracture load, and the minimum test-section di-
ameter after fracture. In all cases the reported tensile properties (table II)
represent the average of at least three tests. Cylindrical test-section speci-
mens (fig. l(c) were used to determine elastic modulus and Poisson's ratio. Two
l-inch-gage-length optical extensometers were attached in the axial direction to
opposite sides of the test section for strain measurement. Simultaneous strain
and load readings taken at regular increments of both increasing and decreasing
load provided the information nece~sary to plot the stress-strain curve from
which the elastic modulus was determined by averaging the slope of the straight-
line portions of this curve. The optical strain gages were then removed from the
test section and replaced with the specially designed diametral strain-gage fix-
ture for use with optical gages (fig. 3). A similar series of simultane6us load
and diametral-strain readings was used to plot this relation. The slope of the
straight-line portions of this curve represents the ratio E/~, so that Poisson's
ratio ~ can be established from this slope and the previously calculated elas-
tic modulus E. The values of E and ~ (table II) were rounded off to two sig-
nificant figures.

Fatigue testing. - The fatigue machine was operated manually during the
first few cycles of each test to set the platen displacement limit switches
(fig. 2) to produce the desired diametral strain range .6.Ed as measured at the
test section. Initial loading was chosen to be compressive rather than tensile,
because it was much easier to stop deformation of the test section at the desired
strain level by manipulation of the pressure regulator. The natural increase of
cross-sectional area during compressive flow reduces the required sensitivity of
initial load control, whereas the reduction of area associated with tensile de-
formations increases the required sensitivity. Also, the initial plastic flow in
compression tends to reduce the subsequent sharpness of the "knee" of the virgin
stress-strain curve, which makes the control of tensile strains easier. The
specimen was first compressed to produce an increase in strain of an amount
~.6.Ed by slowly increasing the oil pressure to the hydraulic cylinder. When
this strain was reached, the compression microsWitch was set to be tripped by the
deflection lever. The compression load was removed, and tension was slowly ap-
plied to produce a diametral decrease of strain of ~.6.€d from the initial diam-
eter. When this strain was reached, the tension microsWitch was adjusted. Thus,
the specimen was strain cycled about a zero mean strain over a strain range of
.6.Ed.

5
 Automatic operation of the fatigue machine was begun once the limit switches
had been positioned so that the deflection lever triggered the switches at the
,desired strain limits as observed with the optical strain gages. Repeated checks
were made of strain range throughout the test. Strain softening or strain hard-
ening of the test specimens necessitated minor adjustments in the deflection-
lever limit-switch settings in order to maintain constant ~€d. These adjust-
ments were necessary because the limit switches maintained constant displacement
between upper and lower die-set platens. The loading rods, grips, and specimen
heads were subjected to deflections entirely within the elastic range, but the
test section of the specimen flowed plastically, undergoing a gradual change in
material properties. Hence, as the test progressed, there was a change in strain
distribution; and, in order to maintain a constant strain range at the test sec-
tion, it was necessary to change the displacement limits between the platens.
For a strain-hardening material these limits had to be increased, and for a
strain-softening material they had to be decreased. Microswitch adjustments were
made whenever the diametral strain range deviated 40 to 80 microinches from the
desired value.

Continuous load recordings were taken during the early part of every test.
Periodic load measurements were taken thereafter throughout the test in conjunc-
tion with the strain measurements. The basic test information periodically re-
corded throughout each test comprised total diametral strain range ~€d, load
range 6P, and number of cycles N. Specimen life was defined as the number of
cycles causing separation of the test section.

Special precautions, described in appendix B, were taken while testing be-

ryllium to prevent distribution of toxic beryllium dust particles into the air.

RESULTS AND DISCUSSION

The basic test data, descriptive test information, and calculated stresses
and strains are tabulated in table III for each fatigue test. Equations used to
calculate these various quantities will be discussed later. Information from
this table was used to plot most of the curves used in the subsequent analysis.

Cyclic Stress-Strain Relation

When fatigue specimens are cycled between fixed strain limits, the stress
range generally changes during the test. Figure 4 shows the typical variations
of stress range ~0 = ~/A with cycles for each of the test materials at three
different values of applied strain range (zero mean strain in all cases). One
group of materials including the heat-treatable ferritic alloy steels (figs. 4(a)
to (e», hardened AM 350 stainless (fig. 4(j», and titanium (6Al-4V)
(fig. 4(l) is characterized by a stress range that decreases from the initial
value. Since the stress required to produce a fixed strain decreases in succes-
sive cycles, and since hardness tests indicate a coincident softening, these ma-
terials are described as cyclic strain softening. A second group of materials

6
including stainless steels (figs. 4(f) to (i», Inconel X (fig. 4(k», 5456-H311
aluminum (fig. 4(n», and beryllium (fig. 4(p) shows an increase of stress and
hardness during strain cycling. These are therefore called cyclic strain..
hardening materials. Commercial purity 1100 aluminum (fig. 4(0» displays a de-
crease in stress range with life, but this change is not simply a result of
strain softening. Visible macrocracks that appear very early in these test
specimens cause a reduction in stress range because the load-carrying ability in
the tensile portion of the strain cycle is reduced. Little change in stress
range takes place for 2014-T6 aluminum for the strain ranges shown in fig-
ure 4(m).

In addition to illustrating the strain-hardening and strain-softening char-

acteristics, these figures indicate that the most significant changes in stress
range usually occur within the first 20 percent of specimen life. During the re-
maining 80 percent or more of the life, the stress range remains relatively con-
stant. This latter value of stress range can be considered then as a character-
istic value corresponding to the applied range. For the purposes of subsequent
analysis the stress range 6a at one-half the number of cycles to failure ~ Nf
was selected as the characteristic value.

The stress analyst who wishes to calculate the stress and strain distribu-
tion during fatigue involving plastic flow needs to know the characteristic
stress range corresponding to an applied strain range. If the material undergoes
cyclic strain softening or hardening, then he cannot properly use the tensile
stress-strain relation of the virgin material, but he needs instead the stress-
strain curve that applies following shakedown in fatigue. (Shakedown is defined
herein as the condition that develops after a sufficient number of cycles have
been applied to II stabilize ll the stress range. Changes in stress range in succes-
sive cycles after shakedown are relatively small compared with those in the early
cycles of strain.) Such a curve is established from fatigue test data by plot-
ting the stress amplitude at shakedown in fatigue against the accompanying longi-
tudinal total strain amplitude. Figure 5 graphically illustrates the difference
between these two stress-strain relations and a method by Which the desired rela-
tion can be obtained. Consider, for example, a test for which the strain ampli-
tude is 0.018 inch per inch, corresponding to point A in figure 5(a). During the
first quarter cycle, the stress amplitude will be governed by the static stress-
strain curve, which indicates a required stress amplitude of 120 ksi. This
stress is transferred to point AI of figure 5(b) in which stress amplitude is
plotted against number of cycles. As the strain amplitude of 0.018 is repeti-
tively applied, the stress amplitude increases and then levels off at 140 ksi as
indicated by curve A'PANf • Point A" (fig. 5(a»), representing the applied
strain amplitude (0.018 in. lin.) and the stress range at ~ Nf (140 ksi) ,is then
one point on the desired cyclic stress-strain curve. A similar point B" is ob-
tained from another test of strain amplitude (0.036 in. lin.); All and B" represent
two points on the cyclic stress-strain curve which is fully obtained from a se-
ries of different constant-strain range tests.

7
 Data necessary to plot the cyclic stress-strain curve (in terms of stress
amplitude 0a against longitudinal strain amplitude €~) are readily calculated
using the applied diametral strain range 6€d and the stress range 60 at
21 Nf · It was observed during zero mean strain fatigue testing that the maximum
compressive and tensile stresses during anyone cycle remained approximately
equal to one another throughout the test. In other words, cyclic strain harden-
ing or softening affected the peak tensile or compressive stresses equally, and
the mean stress throughout a test was zero. By definition

(1)

where

applied load range at

A cross-sectional area

Total mechanical strain amplitudes €l

and €~ are equal to the sum of the
elastic and plastic strain components when creep and anelastic strains are negli-
gi ble . Thus,

€Z = €ZeZ + €Zp (2a)

a

•
€d == €d + €d (2b)
a eZ p

where

€~Z longitudinal elastic strain amplitude

€lp longitudinal plastic strain

€~2 diametral elastic strain amplitude

€d
p
diametral plastic strain

Also,

(3a)

(3b)

8
where E is the elastic modulus and ~ is Poissonts ratio. The plastic term
€~ must be related to ~€d through the total plastic diametral strain€~.
Figure·6 illustrates stresses and diametral strain components for a typical
stress-strain hysteresis loop during fatigue. The width of this hysteresis loop
is the measure of €~. The longitudinal plastic strain €~ is equal in magni-
tude to twice €~ (since volume is conserved during plastic flow) and is oppo-
site in sign, alternating each cycle from tension to compression. Since com-
parison is desired with the tensile stress-strain curve, only positive strains
need be considered. Hence,

(4a)

(4b)

(4c)

By substituting equations (3a) and (4c) into (28.), the desired relation is ob-
tained:

(5 )

The cyclic stress-strain relation was established for each material by plot-
ting 0 a against €~ as shown in figure 7. This curve defines the shakedown
stress amplitude that accompanies a given strain limit during strain cycling
loading. It is readily apparent, on comparison with the virgin tensile data,
that very appreciable changes to the stress-strain relation are caused by cyclic
straining. These figures also reveal cyclic strain-softening and strain-
hardening behavior by the vertical displacement of the fatigue curve from the
static tensile curve. Strain-softening materials exhibit a larger percentage re-
duction of stress at 1- or 2-percent strain (corresponding to intermediate life)
than at 7- or 8-percent strain (corresponding to very short life). Apparently
the softening characteristics are counteracted somewhat by large values of ap-
plied strain. On the other hand) strain-hardening materials tend to have a
greater increase of stress at the larger strains.

Great differences exist between the stress-strain curves of initially an-

nealed (or soft) and hardened conditions of an alloy. Figure 7 shows that
strain-cycling fatigue causes the cyclic stress-strain relations for the two ini-
tial hardness conditions to become much more alike. This effect is most pro-
nounced for materials that are customarily hardened by working such as AISI 304
and AM 350) though it does occur to a lesser degree with heat-treatable steels
such as AISI 4130 and 4340.

9
 Although the degree of softening or hardening varies somewhat with applied
strain, an approximate comparison of materials with respect to these characteris-
tics can be made by evaluating the ratio of the shakedown stress in fatigue at a
gi ven strain range to the virgin tensile stress at a corresponding value of
strain. This ratio is less than unity for cyclic strain softening and greater
than unity for cyclic strain hardening. Table IV lists this ratio, calculated at
I-percent strain, for each material (in order of increasing value). A fair cor-
relation exists between the degree of softening or hardening and the parameter
au/ays (ultimate tensile strength 0u divided by the conventional 0.2-percent
offset yield strength ays ' fig. 8). All materials for which au/oys = 1.2 or
less softened under cyclic straining and those for which 0u/OyS = l.4 or
greater strain hardened. At intermediate values, 1.2 < 0u/Oys < 1.4, both hard-
ening and softening were observed. This correlation may serve as a guide for es-
timating from tensile test data the degree of hardening or softening that may be
expected in the absence of the requisite fatigue data.

The practical use of cyclic stress-strain relations in conjunction with the

strain-life relations, which will be discussed next, for the application to the
solution of practical problems is demonstrated in reference 8.

Relation of Life to Total Strain and to Elastic

and Plastic Components

For the most general of design applications, it is necessary to know the

relation of fatigue life to both the total mechanical strain and the elastic and
plastic components of this strain. Basic experimental strain-life test data are
necessary also to evaluate and devise methods of predicting fatigue life and to
obtain a better understanding of notch behavior in fatigue. ~hese purposes are
served by the experimental strain-life data shown for each test material in fig-
ure 9. Plotted on log-log coordinates are the elastic strain range ~E~l'
plastic strain E~' and total strain range ~El against Nf for each fatigue
test. The elastic strain range and the plastic strain component are calculated
from the basic data using equations (3a) and (4c), respectively. The total
strain range ~l equals the sum of these elastic and plastic components.

Some general observations can be made about the behavior of the various
strain data. The plastic strain, for example, has a magnitude many times that of
·the elastic strain range at short life, but it decreases rapidly with increasing
life to become negligible in comparison with 6E~l at long life. This is true
for all the test materials except in the long-life region for AlSI 304 ELC an-
nealed and AM 350 annealed (figs. 9(f) and (i)), where the plastic strain remains
a significant quantity though less than the elastic component. l The relative
tolerance of different materials to plastic strain in the short-life region seems

IThe anomalous behavior of these strongly strain hardening stainless steels,

as well as the delayed hardening of these steels at the low strain levels (figs.

10
to correlate ~uite directly with the tensile ductility. Those materials with
larger ductilities such as 1100 aluminum and AlSI 304 annealed withstand a
greater plastic strain for a given life than those with low ductilities such as
beryllium and Inconel X.

Figure 9 also indicates that a log-log representation of plastic strain

against fatigue life for most materials can be fitted well with a straight line
(again except for the annealed states of AISI 304 ELC and AM 350). In determin-
ing the least-squares straight-line fit shown in figure 9, all data were used ex-
cept those for fatigue lives less than 10 cycles and those for which the plastic
strain is less than one-half the elastic strain range. The former exclusion is
made because, under these large load and plastic flow conditions, bending in the
test section often takes place or the diametral strain-gage fixture often indents
the test section or both. The strain and load readings for these very short
lives, therefore, may be open to some question. The latter exclusion is made be-
cause any error of fixed magnitude that enters into the calculated plastic strain
becomes an increasingly larger percentage of the strain as it decreases to low'
values. Therefore, for the purposes of subsequent detailed analysis (ref. 10)
the line was fitted to the more representative data. It is obvious that empiri-
cal mathematical relations such as this straight-line correlation may work out
very well in the short-life region where the plastic strain is large. However,
for small values of plastic strain, on the order of 0.001 inch per inch or less,
large errors in predicted life may result because of the uncertainty of the
straight-line extrapolation into this region and because of uncertainties in E,
~, and other factors that enter into the €p calculation. Consequently, such a
method, at best, would be limited to a maximum life of about 800 cycles for mate-
rials like AM 350 (hard) or 4000 cycles for those like AlSI 52100.

The elastic strain range, unlike the plastic strain, never decreases to di-
mini shingly small values that are impossible to measure or calculate accurately.
Hence, it could be practical to relate fatigue life with the elastic strain (or
stress) over the entire life span. This might be done in a way similar to that
used for the plastic component, that is, by also fitting the elastic data with a
single straight line. For many of the test materials such a correlation repre-
sents the data well over the range investigated, but for others there is a ten-
dency for the data points to falloff such a line in either the long- or short-
life region. The straight lines fitted to the elastic strain data of figure 9
represent the least-squares fit of all failure data except those cases in the
short-life region where the elastic strain is less than one-half of the plastic
strain. This straight-line method of representation again was chosen for pur-
poses used in reference 10 in which equations representing the elastic and plas-
tic components are summed up to develop a relation between total strain range and

4(f) and (i)), appears to be associated with the transformation of austenite to

the body-centered lattice that takes place during cyclic strain fatigue (ref. 9).
The transformation, which is evidenced by an increase in magnetic properties dur-
ing the test, proceeds at a much slower rate at the lower strain levels than the
high. It appears to go to completion at high strain levels (as evidenced by mag-
netic properties and stress range behavior) well before specimen failure, but
only partial transformation is accomplished at failure under low strain levels.

11
cyclic life. Such relations provide the designers with complete information con-
cerning stress, strain, and fatigue life.

Figures 9(a) to (p) also demonstrate the behavior of total longitudinal

strain range with cyclic life for the test materials. In most cases ~EZ de-
creases very rapidly with increasing Nf in the short-life region and then
changes much less rapidly in the intermediate life region. This behavior of
total strain reflects, of course, the overriding effect that the plastic strain
component has upon total strain in the short-life region and the predominant ef-
fect the elastic strain component has in the longer life region. By adding to-
gether, at a given life, the strain values denoted by the straight-line correla-
tion of the elastic and plastic strain components the corresponding correlating
line for total strain can be determined. This line represents test data well for
most of the materials, indicating that the total strain against life relations
can be obtained by summing up the elastic and plastic components as approximated
by straight-line relations. The biggest exceptions occur in the longer life re-
gion of the annealed states of AlSI 304 and AM 350, which is apparently due to
the previously mentioned transformation that takes place in these materials.

Relative Performance of Materials

Comparison on basis of diametral strain range. - Figure 10 illustrates the

effect of diametral strain range ~Ed on fatigue life for all the materials
where this was the independent test variable. At a specific life the spread in
6Ed is greatest in the short-life region, becomes small at about 10,000 cycles,
and then becomes fairly large again in the long-life region. A close examination
of the data reveals that the relative standings of materials change with Nf .
For example, 1100 aluminum and AlSI 4130 (soft) steel withstand the greatest 6 Ed
for short lives, but they rate among the poorest at long life. Conversely,
AM 350 (annealed) shows up poorly at short life, but it is one of the best for
fatigue lives of 105 to 106 cycles. Table V, derived from this figure, makes a
detailed comparison of the diametral strain range capability for the various ma-
terials at fatigue lives of 100, 3000, and 100,000 cycles. The strain range ca-
pability of the best material is about 6 times that of the poorest at 100 cycles,
only l~ times greater at 3000 cycles, and 3 times greater at 100,000 cycles. The
change in relative ratings in going from short to long life is plainly evident.
Titanium (6Al-4V) is an exception to the general trend since it ranks as one of
the better strain-resisting materials over the complete life range.

The fact that the relative strain-cycling performance of materials changes

with Nf is due to their different tolerances to plastic strain and elastic
strain (or stress). In the short-life region the materials with the greatest
plastic strain absorbing capacity (most ductile) will show up best, and in the
long-life region generally those with the greatest elastic strain absorbing ca-
pacity (best endurance limit) will excel. For moderate life, 1000 to 10,000 cy-
cles, the better materials are those with the best combination of plastic plus
elastic strain capacity.

12
 Stress range comparison. - Stress range at is plotted against
log Nf
in figure 11 for all test materials. There is less change in the relative rat-
ings of the materials with increasing life than there is when comparison is made
on a cyclic strain basis. Aluminum and beryllium fall well below the steels;
however, on a strength-to-density basis these materials would be comparable to
the steels, and titanium (6Al-4V) would rank highest. Table VI indicates the
small variation of relative ratings of materials on a stress range basis for fa-
tigue lives of 100, 3000) and 100,000 cycles, Comparison with table V shows
stress range ratings of materials to be almost the inverse of the strain range at
Nf = 100 cycles.

Notch sensitivity. - Under certain test conditions for some materials fa-
tigue failure took place outside the test section in a region of stress concen-
tration. Although it was not the purpose of this test program to investigate
systematically the effect of stress concentrations) the previously discussed
strain-life relations explain the observed behavior very well. These nontest-
section fractures never occurred during very high strain range tests (correspond-
ing to short lives) but) rather, appeared at some critical lower strain level and
at all strain ranges less than this critical value. In oth~r words, certain ma-
terials exhibited sensitivity to a particular type of stress raiser when sub-
jected to applied strain ranges below, but not above) some critical value.

Materials showing this behavior were AlSI 4130 (hard») AlSI 52100) AM 350
(hard), and titanium (6Al-4V). Two of these) AISI 4130 (hard) and 52100) were
tested in the threaded-head configuration (fig. l(b). Nontest-section failures
occurred at the innermost thread root at one end of the specimen. The only other
material tested with threaded ends) AlSI 4130 (soft») did not show notch sensi-
tivity. AM 350 (hard) and titanium) like most of the materials, were made into
buttonhead specimens (fig. l(a»). Nontest-section failures originated at the
small fillet joining the buttonhead at the 1/2-inch-diameter shoulder. Once
these failures were observed) it was necessary to reduce the test-section diam-
eter to obtain strain-life data in the longer life region. Such a change is ef-
fective) of course) because it reduces the ratio of strain at the notch to that
at the test section.

In order to use the strain-life relations to explain this behavior it is

necessary to relate the localized total strain range at the stress raiser ~€~r
to the uniform elastic strain range l::,.€~ 1- at the test section and then to com-
pare l::,.€~r with ~El) the total strain range at the test section. Since both
the test section and stress raiser are subject to the same load range .6P) the
nominal stress range at the stress raiser l::,.asr can be equated to the test-
section stress range ~a as follows:

~a sr = AA ~a = m ~a
sr

13
where

A cross-sectional area at test section

Asr cross-sectional area at stress raiser (thread root or shoulder)

m A/Asr ' a constant for a given specimen configuration

Localized total strain range at the stress raiser ~€2 is assumed to be

approximately e~ual to the elastically computed strain ran~~ at this point multi-
plied by the elastic stress concentration factor k. This is a commonly used as-
sumption that is sufficiently accurate for the present purpose. The total strain
range at the stress raiser ~€~r can then be related to the elastic strain range
at the test section ~€~2:

where C is a constant e~ual to kIn.

If C is less than unity, the total strain range at the stress raiser is
less than the elastic strain range at the test section and therefore must be less
than the total strain range at the test section. Conse~uently, for C < 1, fail-
ure must occur at the test section. When C is greater than unity, the total
strain range at the stress raiser is greater than the elastic strain range at the
test section, and therefore mayor may not be greater than the total strain range
at the test section. Location of failure is dependent upon which of these total
strain ranges is the larger. For this latter case consider figure 12, which pre-
sents elastic and total strain range as a function of life for AlSI 52100 steel.
A dotted line is drawn parallel to the elastic strain range line but is verti-
cally displaced by the factor C = 1.5 to represent a hypothetical localized
total strain at the thread root ~€1r = 1.5 ~€~2' This line intersects the test-
section total strain range line at 700 cycles. For lives less than 700 cycles,
~€2 is greater than ~€2 ; conse~uently, failure is expected at the test sec-
sr
tion. For lives greater than 700 cycles, f:,.€~r is greater than ~€2, and frac-
ture should take place at the notch. Qualitatively, this is the type of behavior
that was observed in the abovementioned materials.

These strain-life relations can be used further to make comparisons of the

relative notch sensitivities of all test materials. An effective way of doing
this is to compare characteristic strain ratio curves in which the ratio of total
strain range divided by elastic strain range at the test section ~€2/~€~2 (ob-
tained from the strain-life test data) is plotted against log Nf , as has been
done in figure 13 for the three materials tested in the threaded-head configura-
tion. Now, whenever the total strain at the stress raiser is less than that at
the test section, failure should occur at the test section. In other words, if
~€~r/~€~2 = C is less than the characteristic strain ratio ~€2/~€~2' failure

14
takes Flace at the test section; if C is greater than the characteristic strain
ratio, failure takes Flace at the stress raiser. As an examFle, consider again
the hypothetical case where C = 1.5; that is, the total strain range at the
stress raiser is ~ times the elastic strain range at the test section. The or-
2
dinate on the right side of the graph is used to represent ~€~r/~€!z = C, which,
for a sFecific sFecimen geometry, is a fixed value independent of Nf' The
dashed line at C = 1.5 intersects the strain ratio - life curve of AlSI 52100
steel at about 700 cycles. If failure is to occur anywhere in the specimen at
less than 700 cycles, it should take place in the test section because ~€Z is
greater than ~€~r in this life region. If failure occurs at more than 700 cy-
cles, it should take place at the stress raiser since ~€k exceeds !::J.€Z in
this region. This intercept occurs at about 7000 cycles for AlSI 4130 (hard) and
17,000 cycles for AlSI 4130 (soft). The value of life at which this transition
from test-section to stress-raiser-type failure occurs is then a measure of fa-
tigue notch sensitivity. Notch sensitivity decreases with increasing values of
IItransition ll life.

These characteristic strain ratio curves indicate that AlSI 52100 would be
more notch sensitive than AlBI 4130 (hard), which in turn is more notch sensitive
than AlBI 4130 (soft) regardless of C, since the curves do not cross on~ an-
other. Tests results for the standard 0.25-inch-diameter test section substanti-
ate this observation. The least number of cycles causing a thread failure was
225 for AlSI 52100 steel and 15,500 for AlSI 4130 (hard). No thread failures
took place in AlSI 4130 (soft) even for fatigue lives exceeding one million cy-
cles. Similar characteristic strain ratio comparisons (fig. 14) for the mate-
rials tested in the buttonhead configuration indicate for 1 < C < 1.2 that
nontest-section failures should occur at the lowest life for AM 350 (hard) and
next lowest life for titanium. Experiment indicated buttonhead failures in as
few as 23,200 cycles for AM 350 (hard) and 192,000 cycles for titanium. No such
failures developed in the other materials.

A more quantitative aPFlication of this analysis could not be accurately ap-

plied to these data for several reasons. First, the value of stress concentra-
tion factor k is not accurately known. Roark (ref. 11) quotes k for similar
configurations; however, he does not give the values for geometries as severe as
those used. Second, this method of analysis should be useful for the prediction
of the initiation of a fatigue crack at the stress raiser, but not necessarily
for the prediction of complete separation of the entire specimen cross section.
Crack propagation may have an appreciable effect upon life, depending uFon the
duration of the macrostage of crack growth (ref. 12, PF. 22-25). The appearance
of test-section fracture surfaces indicated that the size of the fatigued area
was less for high-yield-strength materials than for low; also the fatigued area
for a given material tended to be less at high strain levels than at low levels.
The quantitative analysis can be particularly difficult when C is close to
unity for these reasons and other reasons such as the uncertainty in the
~€Z/~€lz ratio and the material effect upon the k value.

15
 CONCLUSIONS

The following conclusions are based upon extensive strain-cycling tests

about a zero mean strain and apply to AlSI 4130 (soft and hard conditions),
AlSI 4340 (annealed and hard), AlSI 52100 (hard), AlSI 304 ELC (annealed and
hard), AlSI 310 (annealed), AM 350 (annealed and hard), Inconel X, titanium
(6Al-4V), 20l4-T6, 5456-H3ll, and 1100 aluminum, and beryllium materials:

1. The cyclic stress-strain relations have been established and are shown to
be substantially different from the virgin tensile data in many instances.

2. These relations are based upon the general observation that the stress
range, following the initial period in which cyclic strain hardening or softening
occurred, remains relatively unchanged throughout the majority of the specimen
life.

3. A fair correiation was obtained between the degree of cyclic strain hard-
ening and softening and the ratio of ultimate strength over yield strength.
Hardening always took place when this ratio exceeded 1.4, and the softening oc-
curred when the ratio was less than 1.2.

4. The life relation between elastic, plastic, and total strain components
has been established. Straight-line fits of the logarithmic elastic strain-life
and plastic strain-life data, when summed together, provide good agreement with
the total strain range - life data for metallurgically stable materials.

5. The test data can be used to evaluate relative performance of all test
materials both on a strain range and stress range basis over a life span of a few
cycles up to one million cycles.

6. In general, the materials with the better total strain-absorbing capacity

at low life (100 cycles or less) were among the poorest at high life (100,000 cy-
cles or more). Titanium (6Al-4V), which had good performance throughout the en-
tire life span, was an exception to this behavior.

7. The strain-life relations can be used to explain the changes in suscepti-

bility to stress concentrations over a large life span and to rate correctly the
relative notch sensitivities of the four test materials that experienced nontest-
section failures in comparison with the other test materials.

Lewis Research Center

National Aeronautics and Space Administration
Cleveland, Ohio, October 10, 1962

16
 APPENDIX A

SYMBOLS

A cross-sectional area of test section, s~ in.

C dimensionless constant

E modulus of elasticity, Ibjs~ in.

k stress concentration factor

m area ratio

N number of cycles

N number of cycles to failure

f
bP load range, Ib

€ strain, in.jin.

~€ strain range, in.jin.

~ Poissonts ratio

o stress, Ibjs~ in.

~0 stress range, Ibjs~ in.

Subscripts:

a amplitude

e"L elastic component of total strain

f fracture

max maximum value during a fatigue cycle

\
min minimum value during a fatigue cycle

p plastic component of total strain

s r a t stress raiser

u ultimate

ys O.2-percent offset yield strength

17
Superscripts:

d diametral

I longitudinal

18
 APPENDIX B

PRECAUTIONS TAKEN DURING BERYLLIUM TESTING

Since beryllium particles ingested into the body or deposited into open
wounds are known to be toxic) special precautions were taken during beryllium
testing to prevent dispersion of such dust particles into the test-room atmos-
phere. One likely source of beryllium dust in addition to that caused by test-
section failure is due to fretting that can take place between the specimen
alinement shoulders and the s~lit-cone grips. To minimize the chances of devel-
oping airborne particles from these sources) the s~ecimen was thoroughly coated
with Lubriplate grease. Figure 15 illustrates the enclosure built around the
specimen assembly and used to further prevent beryllium dust particles from en-
tering the atmosphere. The enclosure was made of a longitudinally split plastic
cylinder) 8 inches in outside diameter by 13 inches long by 1/4 inch thick)
sealed at both ends with 1/16-inch-thick sheet rubber. Two diametrically oppo-
site l~-inch-diameter holes in the plastic were provided for the passage of in-
take air into the chamber and for straight-through sighting of the Tuckerman gage
with the autocollimator. Another l~-inch-diameter hole near the base of the cy-
linder served as an air exhaust port by the attachment of a vacuum cleaner hose.
Exhaust air was filtered with Hollinsworth and Boss number 70 filter pa~er. The
vacuum cleaner was operated continuously throughout the test and during disassem-
bly of the fractured specimen at an approximate flow rate of 0.05 cubic meter per
minute. Test operators wore protective clothing while assembling the setup)
taking strain-gage readings) disassembling the apparatus) and washing down test
specimen) grips) plastic cylinders) rubber seals) and so forth) after the test.
A respirator was worn together with rubber gloves and a laboratory coat.

In order to obtain data that might be used to evaluate the degree of hazard
involved) emission spectrographic analyses of air filter and smear swabs taken
from various parts of the apparatus were made following a fatigue test o£ about
l~ hour duration (1376 cycles). Whatman number 4lH filter paper disks about 2~6
inch in diameter were used for the swabs. Results of the analysis made by a com-
mercial spectrographic company) which reported a sensitivity of 0.01 microgram of
beryllium) are the following:

Sample source Be content)

micrograms/sample

Air filter <0.1

Test specimen (fractured portion 15.00
and shoulders)
Rubber pad beneath cylinder 3.5
Inside surfaces of plastic cylinder .30
Lubriplate as it comes from supply .15
can

19
 The amount of beryllium trapped out from the exhausted air is low, and, with
further substantiating data, might indicate that the respirator and specimen en-
closure is not necessary. The greater amounts of apparently larger beryllium
particles found on the greased specimen and the rubber pad suggest that a careful
cleanup after each test is desirable,

20
 REFERENCES

1. Manson, S. S.: Behavior of Materials Under Conditions of Thermal Stress.

NACA Rep. 1170,1954. (Supersedes NACA TN 2933, 1953.)

2. Coffin, L. F., Jr. z A Study of the Effects of Cyclic Thermal Stresses on a

Ductile Metal. Trans. ASME, vol. 76, no. 6, Aug. 1954, pp. 931-949; dis-
cussion, pp. 949-950.

3. Coffin, L. F., Jr.: Design Aspects of High Temperature Fatigue with Particu-
lar Reference to Thermal Stresses. Trans. ASME, vol. 78, no. 3, Apr. 19S6,
pp. 527-532.

4. KOOistra, L. F.: Effect of Plastic Fatigue on Pressure Vessel Materials and

Design. Welding Jour. {Welding Res. Supplement), vol. 36, Mar. 1957,
pp. 120S-130S.

5. Manson, S. S.: Thermal Stresses in Design. Ft. 19 - Cyclic Life of Ductile

Materials. Machine Design, vol. 32) no. 14, July 7, 1960, pp. 139-144.

6. Manson, S. S.: Thermal Stresses in Design. Ft. 20 - Thermal Cycling with

Steady Stress. Machine Design, vol. 32, no. 15, July 21, 1960, pp. 161-
167.

7. Langer, B. F.: Design of Pressure Vessels for Low-Cycle Fatigue. Paper

6l-WA-18, ASME, 1961.

8. Manson, S. S.: Thermal Stresses in Design. Ft. 18. Working Stresses for
Ductile Materials. Machine Design, vol. 32, no. 13, June 23, 1960, pp.
153-159. (See also pt. 19, vol. 32, no. 14, July 7,1960, pp. 139-144.)

9. Brick, R. M., and Phillips, Arthur! structure and Properties of Alloys.

McGraw-Hill Book Co., Inc., 1942, p. 160.

10. Tavernelli, J. F., and Coffin, L. F., Jr.: Experimental Support for General-
ized Equation Predicting Low-Cycle Fatigue (Discussion by S. S. Manson).
Jour. Basic Eng. (Trans. ASME), sere D, vol. 84, no. 4, Dec. 1962, pp.
537-541.

11. Roark, Raymond Jefferson: Formulas for Stress and Strain. Third ed.,
McGraw-Hill Book Co., Inc., 1954, pp. 348; 353.

12. Smith, Robert W., and Smith, Gordon T.: Thermal-Fatigue Crack-Growth Charac-
teristics and Mechanical Strain Cycling Behavior of A-286, Discaloy, and
16-25-6 Austenitic Steels. NASA TN D-479, 1960.

21
ro TABLE I. - MATERIAL DESCRIPTION
ro
Material Nominal composition, Condition Hardness,
percent Rockwell -
AISI 4130 C 0.30, Mn 0.50, P 0.040, S 0.040, 17000 F; 1/2 hr in salt, water quench C-25 to 27
(soft) Si 0.28, Cr 0.95, Mo 0.20, Fe 12000 F; 1/2 hr in salt, air cool
remainder
AISI 4130 Same heat as above 1600 0 F; 1/2 hr in salt, water quench C-39 to 40
(hard) 750 0 F; 1 hr in salt, air cool
AISI 4340 C 0.40, Mn 0.70, P 0.040, S 0.040, Hot rolled and annealed by supplier C-22 to 24
(annealed) Si 0.28, Ni 1.82, Cr 0.80,
Mo 0.25, Fe remainder
AISI 4340 Same heat as above 15200 F; 1/2 hr in salt, water quench C-44
(hard) 800 0 F; 2 hr in salt, air cool
AISI 52100 C 1.02, Mn 0.35, P 0.025, S 0.025, 16500 F; 1 hr in salt, air cool to room temp. C-52 to 53
8i 0.28, Cr 1.45, Fe remainder 1535 0 F; 1 hr in salt, oil quench to 1800 F
Immediately temper 3200 F, 35-min total
time, air cool
775 0 F; I-hr total time in salt, air cool
AISI 304 ELC a C 0.026, Mn 0.45, P 0.03, S 0.014, Hot rolled and annealed by supplier B-82 to 84
(annealed) Si 0.40, Cr 18.67, Ni 8.50,
Mo 0.02, Cu 0.10, Fe remainder
AISI 304 ELC Same heat as above Cold drawn by supplier C-34 to 36
(hard)
AISI 310 C 0.25 max., Mn 2.00 max., Si 1.50 Hot rolled, pickled, and annealed by B-75 to 81
(annealed) max., Cr 25.0, Ni 20.5, Fe supplier
remainder
AM 350 C 0.10, Mn 0.75, Si 0.35, Cr 16.50, Hot rolled, pickled, and annealed by
(annealed) Ni 4.25, Mo 2.75, N 0.10, Fe supplier
remainder
AM 350 Same heat as above Cold drawn 25-35% reduction by supplier C-50 to 52
(hard)
Inconel X C 0.04, Mn 0.70, Si 0.30, Cr 15, 2100 0 F 4 hr, air cool C-34 to 35
Cb 1, Ti 2.5, Al 0.9, Fe 7, 1550 0 F 24 hr, air cool
Ni remainder 13000 F 20 hr, air cool
Titanium C 0.029, Fe 0.22, N2 0.017, Al 6.1, Centerless ground, solution treated, and C-40 to 41
(6Al-4V) V 4.0, H2 0.007, Ti remainder aged by supplier
2014-T6 Si 0.85, Fe 1.0, Cu 4.5, Mn 0.80, Bar stock as received B-81 to 84
aluminum Mg 0.50, Cr 0.10 max., Zn 0.25
max., Ti 0.15 max., Al remainder
5456-H311 Si + Fe 0.40 max., Cu 0.10 max., Bar stock as received B-56
aluminum Mn 0.75, Mg 5.1, Cr 0.12, zn 0.25,
Ti 0.20, Al remainder

1100 Si + Fe 1.0, Cu 0.20 max., Mn 0.05 Bar stock as received BRN -26
aluminum max., Zn 0.10, Al remainder
Beryllium C 0.15 max., Mg 0.08 max., Si 0.12 Vacuum hot-pressed QMV beryllium B-75 to 90
max., Al 0.16 max., Fe 0.18 max.,
BeO 2.0 max., Be 98.0 min., other
metal 0.04 max.
aActual analysis.
 TABLE II. - MECHANICAL AND ELASTIC

PROPERTIES OF TEST MATERIALS

[Data represent the average of three test specimens unless otherwise specified. 1

Material Yield strength Ultimate Fracture Reduction True Modulus of Poisson's

0.2-percent strength, strength, in area, ductili ty elas ti ci ty, ratio,
offset strength, au' a , percent E jl.
a ys , f
ksi
ksi ksi

r0ftl
4130 hard
4340 ann.
113
197
92
130
207
120
245
302
174
67.3
54.7
43.4
1.12
.79
.57
32XIO- 6
29
28
0.29
.28
.32
4340 hard 199 213 278 38.1 .48 29 .30
52100 279 292 323 11.2 .12 30 .29

304 ELC ~ann. ~ 3.7 108 278 74.3 1.37 27 .27

304 ELC hard 108 138 295 68.8 1.16 25 .34
310 32 93 197 63.5 1.01 28 .30
350 tnn.j
ann. 64 191 339 52.1 .74 28 .32
350 hard 270 276 328 20.3 a. 23 26 .30
Inconel X 102 176 219 19.7 .22 31 .31
Titanium 172 179 249 41.0 .53 17 .33
2014-T6 Al 67 74 91 25.0 .29 10 .33
5456-H311 Al 34 58 82 34.6 .42 10 .33
1100 Al 14 16 (b) 87.6 2.09 cIO ~.33
Beryllium 38 46.9 47.7 1.7 .017 42 .024

a Data for five specimens ranged from 0.08 to 0.38.

b
Could not be accurately measured.
cAssumed equal to value obtained for other aluminum alloys.
dBeryllium datum from commercial supplier.

[\)
CfI
 TABLE III. - SUMMARY OF TEST DATA AND STRAIN
CALCULATIONS FOR EACH TEST SPECIMEN

Stress range I Fatigue

life,
Diametral
elastic
Diametral
plastic dstrain,
Diametral
total strain
Longi tudina1
elastic strain
Longitudinal
plas tl c strain,
Longi tudina1
total strain
Comments
at '2 Nf,
. 1
t,a, I Nf,
cycles
strain range,
&E~'L'
€p, range,
6.e d ,
range,
b.€~'L '
E~, range,
b.e 1.,
ksi in./in. in·/in. in·/in.
in·/in. in·/in. in·/in.
(a)
AlSl 4130 (soft)
307 12 0.0028 0.1252 0.1280 0.0096 0.2504 0.2600
299 17 .0027 .0773 .0800 .0093 .1546 .1639
293
274
238

221
27
63
138

169
.0027
.0025
.0022

.0020
.0773
.0455
.0250

.0220
.0800
.0480
.0272

.0240
.0092
.0086
.0074

.0069
.1547
.0910
.0501

.0440
.1638
.0996
.0575

.0509
Ii!
215 245 .0019 .0191 .0210 .0067 .0381 .0448
214 455 .0019 .0128 .0147 .0067 .0255 .0322
189 672 .0017 .0096 .0113 .0059 .0192 .0251
186 1,050 .0017 .0083 .0100 .0058 .0166 .0224

179 1,860 .0016 .0050 .0066 .0056 .0100 .0155

174 5,630 .0016 .0022 .0038 .0054 .0044 .0099
169 7,260 .0015 .0023 .0038 .0053 .0045 .0098
155 24,400 .0014 .0008 .0022 .0048 .0016 .0064
155 36,300 .0014 .0006 .0020 .0048 .0012 .0060

145 82,000 .0013 .0001 .0014 .0045 .0002 .0047

150 86,400 .0014 0 .0013 .0047 0 .0047
132 326,000 .0012 0 .0012 .0041 0 .0041 (el' (f)
145 329,000 .0013 0 .0013 .0045 0 .0045 e)

f:U~l
110 11,200,000 .0010 0 .0010 .0034 0 .0034
120 12,000,000 .0011 0 .0011 .0037 0 .0037

AISl 4130 (hard)

f~l
437 4 0.0042 0.1958 0.2000 0.0151 0.3916 0.4066
403 16 .0039 .0761 .0800 .0139 .1522 .1661
336 49 .0032 .0448 .0480 .0116 .0895 .1011
303 86 .0029 .0211 .0240 .0104 .0421 .0526
261 3.'56 .0025 .0075 .0100 .0090 .0150 .0240
282 360 .0027 .0120 .0147 .0097 .0240 .0337
244 1,320 .0024 .0042 .0066 .0084 .0085 .0169
226 7,890 .0022 .0016 .0038 .0078 .0032 .0110
229 15,500 .0022 .0006 .0028 .0079 .0012 .0091 (h), (i)
206 27,700 .0020 0 .0020 .0071 0 .0071 (h)
222
200
146
157
29,900
37,110
884,000
1,030,000
.0021
.0019
.0014
.0015
0
0
0
0
.0020
.0015
.0013
.0015
.0077
.0069
.0050
.0054
0
0
0
0
.0077
.0069
.0050
.0054
fj
Hl~ ~l
l
AlSI 4340 (annealed)

!~l
292 7 0.0033 0.1167 0.1200 0.0104 0.2333 0.2438
271 14 .0031 .0769 .0800 .0097 .1538 .1635
232 43 .0027 .0373 .0400 .0083 .0747 .0830
192 165 .0022 .0178 .0200 .0069 .0356 .0425
171 635 .0020 .0120 .0140 .0061 .0241 .0302
159 1,050 .0018 .0082 .0100 .0057 .0164 .0220
143 2,780 .0016 .0040 .0056 .0051 .0079 .0130
124 14,000 .0014 .0018 .0032 .0044 .0036 .0080 (f)
114 51,200 .0013 .0007 .0020 .0041 .0014 .0055
109 166,000 .0012 .0004 .0016 .0039 .0007 .0046
104 524,000 .0012 .0002 .0014 .0037 .0004 .0041

AISl 4340 (hard)

I~l
470 2 0.0049 0.1951 0.2000 0.0162 0.3903 0.4065
424 4 .0044 .1956 .2000 .0146 .3912 .4058
396 13 .0041 .0759 .0800 .0137 .1518 .1655 (k/, (1)
~l
398 51 .0041 .0359 .0400 .0137 .0718 .0855
378 52 .0039 .0361 .0400 .0130 .0722 .0852
333 192 .0034 .0166 .0200 .0115 .0331 .0446
296 886 .0031 .0069 .0100 .0102 .0139 .0241
254 2,850 .0026 .0030 .0056 .0088 .0059 .0147
238 9,810 .0025 . )007 .0032 .0082 .0015 .0097
230 30,700 .0024 0 .0020 .0079 0 .0079
203 63,100 .0021 0 .0020 .0070 0 .0070
175 162,000 .0018 0 .0017 .0060 0 .0060
175 176,000 .0018 0 .0018 .0060 0 .0060
150 974,200 .0016 0 .0014 .0052 0 .0052 (g)

24
 TABLE III. - CONTINUED. SUMMARY OF TEST DATA AND STRAIN
CALCULATIONS FOR EACH TEST SPECIMEN

stress range Fatigue Diametral Diametral Diametral Longitudinal Longitudinal Longitudinal Comments
elastic total strain elastic strain plastic strain, total strain
at ~ Nf'
life,
N , strain range,
plas tic dB train,
range, ! range,
lOa,
f €p' r~~~:, €p'
leSl
cycles b.€~p in./in. in./in. ~€~lJ in·/in.
D.€ l,
in./in. (a) in./in. in·/in.

AISI 52100

621 2 0.0060 0.0420 0.0480 0.0207 0.0840 0.1047

608 3 .0059 .0501 .0560 .0203 .1002 .1205

~~ l
610 4 .0059 .0341 .0400 .0203 .0682 .0885
605 5 .0058 .0302 .0360 .0202 .0603 .0805
579 10 .0056 .0344 .0400 .0193 .0688 .0881

571 15 .0055 .0225 .0280 .0190 .0450 .0640 (b)

560 18 .0054 .0266 .0320 .0187 .0532 .0718
553 18 .0053 .0187 .0240 .0184 .0373 .0557 (b)
514 40 .0050 .0150 .0200 .0171 .0301 .0472
520 44 .0050 .0150 .0200 .0173 .0299 .0473
483 73 .0047 .0113 .0160 .0161 .0227 .0388
456 98 .0044 .0076 .0120 .0152 .0152 .0304
468 159 .0045 .0083 .0128 .0156 .0166 .0322
--- 225 ---- .0080 .0080 ---- .0160 .0160 (h)
386 728 .0037 .0031 .0068 .0129 .0061 .0190

374 1,110 .0036 .0016 .0052 .0125 .0032 .0156

---
350
1,470
3,430
----
.0034
.0040
.0006
.0040
.0040
----
.0117
.0080
.0012
.0080
.0129
--- ----
331
5,640
7,840 .0032
.0024
0
.0024
.0030
----
.0110
.0048
0
.0048
.0110 ItI
--- ----
f~l
9,910 ---- .0026 .0026 .0052 .0052
273 30,200 .0026 0 .0024 .0091 0 .0091
182 351,000 .0018 0 .0017 .0061 0 .0061 (j), (h)
217 799,000 .0021 0 .0020 .0072 0 .0072 ( j)
AISI 304 ELC (annealed)

I' f l
~b),
343 25 0.0034 0.0446 0.0480 0.0127 0.0891 0.1018
315 55 .0031 .0288 .0320 .0117 .0577 i
.0694
329 110 .0033 .0167 .0200 .0122 .0334 .0456 b b) 1
274 210 .0027 .0101 .0128 .0101 .0201 .0303
250 596 .0025 .0039 .0064 .0093 .0078 .0171
209 1,970 .0021 .0019 .0040 .0077 .0038 .0116
174 6,010 .0017 .0015 .0032 .0064 .0029 .0094
162 9,460 .0016 .0016 .0032 .0060 .0032 .0092
110 40,400 .0011 .0013 .0024 .0041 .0026 .0067
114 46,500 .0011 .0013 .0024 .0042 .0025 .0067
103 143,000 .0010 .0010 .0020 .0038 .0019 .0058
100 174,000 .0010 .0011 .0021 .0037 .0022 .0059
90 932,000 .0009 .0009 .0018 .0033 .0018 .0051
~- I
AISI 304 (hard)

i~U~l
350 31 0.0048 0.0452 0.0500 0.0140 0.0905 0.1045
343 46 .0047 .0353 .0400 .0137 .0707 .0844

1~l
311 138 .0042 .0158 .0200 .0124 .0315 .0440
314 208 .0043 .0117 .0160 .0126 .0235 .0360
274 1,320 .0037 .0035 .0072 .0110 .0069 .0179
262 1,660 .0036 .0028 .0064 .0105 .0057 .0162
206 7,780 .0028 .0008 .0036 .0082 .0016 .0098
187 27,500 .0025 .0005 .0030 .0075 .0009 .0084
150 95,000 .0020 .0004 .0024 .0060 .0007 .0067
135 121,000 .0018 .0002 .0020 .0054 .0003 .0057
125 474,000 .0017 .0001 .0018 .0050 .0002 .0052

AISI 310 (annealed)

i~U~l
284 12 0.0030 0.1170 0.1200 0.0101 0.2339 0.2441
282 23 .0030 .0770 .0800 .0101 .1540 .1640

~~ l
260 32 .0028 .0572 .0600 .0093 .1144 .1237
218 91 .0023 .0277 .0300 .0078 .0553 .0631
222 95 .0024 .0276 .0300 .0079 .0552 .0632
163 335 .0017 .0123 .0140 .0058 .0245 .0303
134 1,400 .0014 .0066 .0080 .0048 .0131 .0179
107 5,120 .0011 .0033 .0044 .0038 .0065 .0103
83 30,000 .0009 .0011 .0020 .0030 .0022 .0052 (g)
79 55,200 .0008 .0010 .0018 .0028 .0019 .0047
75 61,000 .0008 .0008 .0016 .0027 .0016 .0043
70 173,000 .0008 .0005 .0012 .0025 .0009 .0034

f~l
65 223,000 .0007 .0005 .0012 .0023 .0010 .0033
63 274,000 .0007 .0003 .0010 .0022 .0007 .0029
64 363,000 .0007 .0002 .0009 .0023 .0005 .0027

25
 TABLE III. - CONTINUED. SUMMARY OF TEST DATA AND STRAIN
CALCULAT~ONS FOR EACH TEST SPECIMEN

Stress range Fatigue Diametral Diametral Diametral Longi tudinal Longi tudinal Longi tudinal Comments
at ~ Nr,
life,
Nr,
elastic
s train rang e,
plastic strain,
d
total strain
range,
elastic strain
range,
plastic strain, total strain
range,
I!.a, .p' 6.e: d , €~,
ksi
cycles nE~z' in·/in. in./in. ll€~1.' in·/in.
l::.e 1.,
in./in. (a) in./in. in·/in.

AM 350 (annealed)
499 7 0.0057 0.0343 0.0400
f~U~l
0.0178 0.0686 0.0864
501 18 .0057 .0183 .0240 .0179 .0365 .0544
462 34 .0053 .0087 .0140 .0165 .0174 .0339
422 125 .0048 .0032 .0080 .0151 .0064 .0214
367 228 .0042 .0018 .0060 .0131 .0036 .0167
288 1,134 .0033 .0015 .0048 .0103 .0030 .0133
234 3,220 .0027 .0013 .0040 .0084 .0027 .0110
193 15,410 .0022 .0012 .0034 .0069 .0024 .0093
141 42,700 .0016 .0010 .0026 .0050 .0020 .0070
152 98,300 .0017 .0011 .0028 .0054 .0021 .0076
159 137,000 .0018 .0012 .0030 .0057 .0024 .0080
134 170,000 .0015 .0011 .0026 .0048 .0021 .0069
117 1,020,000 .0013 .0007 .0020 .0042 .0013 .0055 (g)
AM 350 (hard)
579 3 0.0067 0.0453 0.0520 0.0223 0.0906 0.1129
567 14 .0065 .0235 .0300 .0218 .0469 .0667
536 46 .0062 .0176 .0240 .0206 .0356 .0562
525 71 .0061 .0139 .0200 .0202 .0279 .0481
523 85 .0060 .0060 .0140 .0201 .0159 .0360
524 90 .0060 .0140 .0200 .0202 .0279 .0481
488 169 .0056 .0024 .0060 .0188 .0047 .0235
522 176 .0060 .0080 .0140 .0201 .0160 .0360
466 190 .0054 .0014 .0068 .0179 .0028 .0208
436 450 .0050 .0010 .0060 .0168 .0019 .0187
316 5,971 .0036 .0004 .0040 .0122 .0007 .0129
272 11,500 .0031 .0001 .0032 .0105 .0001 .0106
237 13,920 .0027 0 .0028 .0091 .0001 .0092
225
241
16,300
23,200
.0026
.0028
0
0
.0025
.0028
.0087
.0093
0
0
.0087
.0093 W
213 35,940 .0025 0 .0024 .0082 0 .0082 (h)
234

PU~l
40,050 .0027 .0001 .0028 .0090 .0002 .0092
220 56,600 .0025 0 .0025 .0085 0 .0085
223 133,000 .0026 0 .0026 .0086 .0001 .0087
219
227
759,000
,010,000
.0025
.0026
0
0
.0025
.0026
.0084
.0087
0
0
.0084
.0087 Hl;i~U~l
Inconel X

I~I
382 2 0.0038 0.0682 0.0720 0.0123 0.1364 0.1487
422 4 .0042 .0518 .0560 .0136 .1036 .1172
396 6 .0040 .0400 .0440 .0128 .0801 .0929
406 9 .0041 .0379 .0420 .0132 .0758 .0890
378 37 .0038 .0322 .0360 .0122 .0644 .0766

f~l
368 41 .0037 .0303 .0340 .0119 .0606 .0725
356 50 .0036 .0304 .0340 .0115 .0609 .0724
336 79 .0034 .0266 .0300 .0108 .0533 .0641

f~l
359 89 .0036 .0244 .0280 .0116 .0488 .0604
342 137 .0034 .0206 .0240 .0110 .0412 .0522
322 193 .0032 .0148 .0180 .0104 .0296 .0399 (i)
312 220 .0031 .0129 .0160 .0101 .0258 .0358
314 261 .0031 .0169 .0200 .0101 .0337 .0438
317 282 .0032 .0128 .0160 .0102 .0257 .0359
298 296 .0030 .0070 .0100 .0096 .0140 .0237
288 779 .0029 .0071 .0100 .0093 .0142 .0235
281 813 .0028 .0072 .0100 .009l .0144 .0234
281 1,080 .0028 .0040 .0068 .0091 .0080 .0170
264 1,580 .0026 .0024 .0050 .0085 .0047 .0132
233 5,160 .0023 .0011 .0034 .0075 .0021 .0097
232 5,210 .0023 .0017 .0040 .0075 .0034 .0108
230 6,440 .0023 .0005 .0028 .0074 .0010 .0084
231 7,800 .0023 .0005 .0028 .0075 .0010 .0084
211 9,400 .0021 .0001 .0022 .0068 .0002 .0070
215 9,940 .0021 .0008 .0029 .0069 .0015 .0065
204 18,400 .0020 0 .0019 .0066 0 .0066
188 36,400 .0019 0 .0016 .0061 0 .0061
169 47,500 .0017 0 .0016 .0055 0 .0055
155 99,800 .0015 0 .0014 .0050 0 .0050
137 125,000 .0014 0 .0012 .0044 0 .0044
124 267,000 .0012 0 .0012 .0040 0 .0040
124 395,000 .0012 0 .0011 .0040 0 .0040
114 688,000 .0011 0 .0010 .0037 0 .0037

26
 TABLE III. - CONTINUED. SUMMARY OF TEST DATA AND STRAIN
CALCULATIONS FOR EACH TEST SPECIMEN

stress range Fatigue Dlametral Diametral D1ametral Longi tudinal Longi tudinal Longi tudinal Comments
elastic total strain elas tl c strain plastic strain, total strain
~t ~ Nf,
life, plastic dstrain,
Nr, s train range,
'p' r~~~~, range, €~, rC;nge,
60,
ksi
cycles A€~Z' in./'in. in./in. Aekz.J in·/in.
b,.e ~,
in./in.
in·/in. (a) in·/in.
Titanium (6Al-4V)

i~l
430 6 0.0083 0.1117 0.1200 0.0253 0.2233 0.2486
382 27 .0074 .0646 .0720 .0225 .1292 .1516
338 86 .0066 .0294 .0360 .0199 .0589 .0788
326 188 .0063 .0177 .0240 .0192 .0353 .0545
313 453 .0061 .0079 .0140 .0184 .0158 .0343

281 1,470 .0055 .0025 .0080 .0165 .0051 .0216

216 11,800 .0042 0 .0040 .0127 0 .0127
175 '37,000 .0034 0 .0032 .0103 0 .0103
156 70,200 .0030 0 .0028 .0092 0 .0092
149 192,000 .0029 0 .0026 .0088 0 .0088 (h)

134 881,000 .0026 0 .0024 .0079 0 .0079

(hi
Ul;i~l; ;U~l
150 4,270,000 .0029 0 .0029 .0088 0 .0088
172 5J590~OOO .0033 0 .0033 .0101 0 .0101

2014-T6 aluminum

166 4 0.0055 0.0825 0.0880 0.0166 0.1650 0.1816

147 10 .0048 .0552 .0600 .0146 .1103 .1250
155 42 .0051 .0249 .0300 .0155 .0498 .0653
142 203 .0047 .0093 .0140 .0142 .0186 .0328
127 608 .0042 .0038 .0080 .0127 .0076 .0203

117 1,610 .0039 .0009 .0048 .0117 .0019 .0136

117 1,980 .0039 .0009 .0048 .0117 .0019 .0136
107 4,180 .0035 .0001 .0036 .0106 .0002 .0108
72 61,700 .0024 0 .0022 .0071 0 .0071
92 95,550 .0030 0 .0029 .0091 0 .0091

62 229,000 .0020 0 .0062 0 .0062

67 285,000 .0022 0 .0020 .0066 0 .0066
58 394,000 .0019 0 .0018 .0058 0 .0058 (g)
58 418,000 .0019 0 .0058 0 .0058

5456-H311 aluminum

138 7 0.0046 0.0554 0.0600 0.0138 0.1109 0.1247 (c)

133 12 .0044 .0476 .0520 .0133 .0952 .1085
127 54 .0042 .0258 .0300 .0127 .0516 .0643
118 175 .0039 .0101 .0140 .0118 .0202 .0320
111 494 .0036 .0044 .0080 .0110 .0087 .0198

100 1,920 .0033 .0015 .0048 .0100 .0030 .0130

88 5,950 .0029 0 .0028 .0087 0 .0087
72 22,800 .0024 0 .0022 .0072 0 .0072
65 65,100 .0021 0 .0020 .0065 0 .0065
59 90,300 .0019 0 .0018 .0059 0 .0059

52 130,000 .0017 0 .0016 .0052 0 .0052

47 366,000 .0015 0 .0014 .0047 0 .0047
44 917,000 .0015 0 .0014 .0044 0 .0044

1100 aluminum
37 2 0.0012 0.2388 0.2400 0.0037 0.4776 0.4813
37 2 .0012 .2388 .2400 .0037 .4776 .4812
42 5 .0014 .2186 .2200 .0042 .4372 .4414
42 11 .0014 .1986 .2000 .0042 .3972 .4014
36 27 .0012 .1188 .1200 .0036 .2376 .2412

30 73 .0010 .0590 .0600 .0030 .1180 .1210

24 282 .0008 .0292 .0300 .0024 .0584 .0608
22 784 .0007 .0133 .0140 .0021 .0266 .0287
18 2,120 .0006 .0054 .0060 .0018 .0108 .0126
18 6,320 .0006 .0024 .0030 .0018 .0048 .0066

17 12,800 .0006 .0014 .0020 .0017 .0029 .0046

15 104,000 .0005 .0005 .0010 .0015 .0010 .0025
15 205,000 .0005 .0003 .0008 .0015 .0006 .0021
15 417,000 .0005 .0002 .0006 .0015 .0003 .0018
14 1,000,000 .0005 0 .0005 .0014 .0001 .0014 (g)

27
I:\:)
en

TABLE III. - CONCLUDED. SUMMARY OF TEST DATA AND STRAIN

CALCULATIONS FOR EACH TEST SPECIMEN

Stress range Fatigue Diametral Diametral Diametral Longitudinal Longitudinal Longitudinal Comments
1 lif'e, elastic plastic strain, total strain elastic strain plastic strain, total strain
at "2 Nf', strain range, d range, range, range,
Nf"
€p' /',Ed, E1 ,
/',cr,
ksi cycles /',E~L ' in·/in. in./in. /',E ~ L' p /',EL
in·/in. ( a') in./in. in./in. in./in.

Beryllium
117 6 0.0001 0.0066 0.0067 0.0028 0.0132 0.0160
103 73 .0001 .0028 .0028 .0025 .0055 .0080
82 138 0 .0010 .0011 .0020 .0020 .0040
79 260 0 .0008 .0008 .0019 .0015 .0034
66 4,260 0 .0004 .0005 .0016 .0008 .0024
-- --- ---------

aSmall discrepancies which appear upon comparison of the summation of elastic and plastiC strain components with total
strain arise because of small inaccuracies of the measured quantities E and ~. These cause errors in the fourth
decimal place and sometimes result in the calculation of small negative plastic strains. In the table all negative
plastiC strains are reported as zero, but no change was made to the other calculated strains.
bHourglass radius, 1 in.
cS ome buckling of test section.
dApproximate strain range.
eLoad cycled with high-frequency push-pull fatigue machine.
f'Approximate number of cycles to failure.
gNo f'ailure.
hFailure outside test section (threads or at buttonhead).
iApproximate stress range.
jTest-section diameter, 0.18 in.
kHourglass radius, 0.5 in.
LTest-section diameter, 0.21 in.
ffiPolished test section.
 TABLE IV. - QUANTITATIVE }1EA.SURE OF CYCLIC

STRAIN HARDENING AND SOFTENING

:Material Ratio of stress to produce l-percent

strain after strain cycling to
that for the virgin material
.....
AlSI 4130 (hard) 0.60
AlSI 4340 (annealed) .63
AlSI 4340 (hard) .64
1100 aluminum .68
AlSI 52100 .72 >Cyclic strain
AlSI 4130 (soft) .75 softening
Titanium (6Al-4V) .78
AM 350 (hard) .89
20l4-T6 aluminum .90 ....

AlSI 304 ELC (hard) 1.07"'"

Beryllium 1.19
Inconel X 1.24
5456-H3ll aluminum 1.39 >- Cyclic strain
AlSI 310 (annealed) 1.61 hardening
AM 350 (annealed) 2.42
AlSI 304 ELC (annealed) 2.90

29
 TABLE V. - DIAMETRAL STRAIN RANGE RATINGS OF MATERIALS FOR

FATIGUE UVES OF 100, 3000, .AND 100,000 CYCLES

Material Nf = 100 Nf = 3000 Nf = 100,000

.6.€d Rank .6.€d Rank .6.€d Rank

1100 aluminum 0.050 1 0.0048 9 0.0010 15

AlSI 4130 (soft) .036 2 .0052 5 .0014 12
Titanium (6Al-4V) .033 3 .0062 1 .0026 2
AlSI 4340 (annealed) .030 4 .0054 4 .0017 9
AlSI 4130 (hard) .030 4 .0050 8 .0016 11

AISI 310 (annealed) .029 6 .0056 2 .00138 14

AISI 4340 (hard) .028 7 .0056 2 .0018 8
Inconel X .027 8 .0051 7 .0014 12
AlSI 304 ELC (hard) .025 9 .0052 5 .0023 4
AISI 304 ELC (annealed) .021 10 .0037 15 .0022 5

2014-T6 aluminum .020 11 .0040 13 .0021 6

5456-H311 aluminum .019 12 .0039 14 .0017 9
AlSI 52100 .014 13 .00"16 10 .0020 7
AM 350 (hard) .012 14 .0042 11 .0026 2
AM 350 (annealed) .009 15 .0041 12 .0028 1

30
 TABLE VI. - STRESS RANGE RATINGS OF MATERIALS FOR FATIGUE

LIVES OF 100, 3000, AND 100,000 CYCLES

Material Nf ::: 100 Nf ::: 3000 Nf ==. 100,000 Density,

Ib/cu in.
6.0 ) Rank. 6.0, Rank 6.0) Rank
ksi ksi ksi

AM 350 (hard) 510 1 330 1 202 2 0.30

AlSI 52100 469 2 321 2 253 1 .30
AM 350 (annealed) 426 3 246 5 148 8 .30
AlSI 4340 (hard) 354 4 261 3 189 3 .30
Inconel X 353 5 248 4 155 5 .30

Titanium (6Al-4V) 327 6 241 6 155 6 .16

AlSI 304 ELC (hard) 325 7 236 7 145 9 .30
AlSI 304 ELC (annealed) 311 8 189 9 105 11 .30
AlSI 4130 (hard) 293 9 223 8 178 4 .30
AlSI 4130 (soft) 252 10 173 10 150 7 .30

AlSI 4340 (annealed) 213 11 143 11 112 10 .30

AlSI 310 (annealed) 213 12 120 13 72 12 .30
2014-T6 aluminum 144 13 123 12 72 13 .10
5456-H311 aluminum 122 14 95 14 58 15 .10
Beryllium 102 15 81 15 64 14 .07
1100 aluminum 29 16 18 16 15 16 .10

31
 "r -3'25~"'1
0.02-0.03 Rad.

----r-f-
0.75 Di81ll.

* ''-0.25 "-
Diam. "-~0.50 Diam.
1.5 Rad.

(a) Buttonhead fatigue specimen.

II-.--~--4'OO-~l

--. t .§. ~

I
L
I
I

1/2 " -
-

20 NF - 3
-- 1
-

t -
'-0.25 Diam.
-

t '-0.47 Diam.
2.0 Rad.

(b) Threaded-head fatigue specimen.

~-------------3.25--------------~

~----1.50 • !

t
0.75 Diam.-

~ \
\
'-- 0.25 Diam. ",'--'-0.50 Diam.
0.31

(c) Specimen used for moduli determinations and

beryllium fatigue tests.

Figure 1. - Test specimen configurations. (All dimen-

sions in inches.)

32
 Limit Platen
displacement

.
00
"-

.
~.

L Four-way
hydraulic
valve

Figure 2 . - Low- frequency fatigue machine.

33
 (a) Horizontal view of assembly during test .

Figure 3. - Diametral strain- gage fixture.

34
 " Piano - wire
I \ s prings
I \
I \
Aluminum

U spr ing ,
\
\
\
........ \
........
...L Tuckerman
extensom-
eters

I
L Piano -
wire
loops - _

C- 5890l
(b) Vertical view of assembly after specimen failure .

Figure 3 . - Concluded . Diametral strain- gage fixture.

35

L
();I
0)

350 400

~
-- -
I
flEd

---
300 350
~
!'.Ed
0.0l!,8
250 300 "--
~ ~010
~
.,;

..."
~
200

150
-----= ------ .010

.002
250

200 -
.002
-

(a) AISI 4130 (soft). (b) AISI 4130 (hard).

ru'
M 350 450

~
~
ctl
H
! flEd
400
""ru 300

~~
,
H
-P
Ul flEd
250 350

~
.........-

200
-
0.040

r----- r-- .010

300
- --1.010

150 250
i--- .002 .002
.002
100 200 - '--- - L---
1 10 100 1,000 10,000 100,000 1 10 100 1,000 10,000 100,000
Number of cycles, N

(c) AISI 4340 (annealed). (d) AISI 4340 (hard).

Figure 4. - Variation of stress range with cycles for strain cycling about a zero mean strain.
 60 ) 350
l; E d

)~
.---
/"
/0.020
55 300
~
50 )~ ~
.M d

250 / ---t

)~ V- 0064

45 200 / /

40 )
~ ~-
150 / /
...
'"
~

.g

H
OJ
till
~
35

30

25 n
.0024
100

50

o
---
/ V ______ f-'""
/0024

'"OJ
rD
(e) AISI 52100. (r) AISI 304 ELC (annealed).
H
-P 350 300
Cfl
~
1;~
l; E d

0.020 0.080
300 250

250

200
~
-- - --.0072
200

150
/
II

I-
.008
150
.0024
100

50
V ---- .0018
100
1 10 100 1,000 10,000 100,000 1 10 100 1,000 10,000 100,000
Number or cycles, N

(g) AISI 304 ELC (hard). (h) AISI 310 (annealed).

Figure 4. - Continued. Variation or stress range with cycles ror strain cycling about a zero mean strain.

():I
-...l
c.N
CD

50 r 550
LO,Ed
lIEd

4501) ~74 500

~
/1
400 ,r----r---+-+------t-- 450

350
) I /L--:066 400
.006

300 ) 1 / 350

250'
) I / 300

/ /
·rl
OJ
-'<
200) 250

)~ ~)/
b
<l

OJ .0025
b!I 150 200
>::
Ctl ~ .0028
H
Ul
OJ
OJ
100 150
H
.p
CIl
(i) AM 350 (annealed). (j) AM 350 (hard) .
450

LO,,,d
400I ~

350I ~~
l>€d
---- 0.072

II 0.034

300I

250
/
V
~
-=----=-- -- .008
-i

'V
200
.0016 ,
.0028
150
1 10 100 1,000 10,000 100,000 1 10 100 1,000 10,000 100,000
Number of cycles, N

(k) Inconel X. (1) Titanium (6Al-4V).

Figure 4. - Continued. Variation of stress range with cycles for strain cycling about a zero mean strain.
 250 150,
"'Ed
0.030
200 125
"'Ed
0.03
V
-----
150 100 --I
-rl ~048
.0048

----
Ul
-'"
100 75 t-
,;
<I .0022
~--------- -,
,; .0018
bll 50 50 1

<=:
cO
H (m) 2014-T6 aluminum. (n) 5456-H311 aluminum.
Ul 75 12 )
Ul L
ill
H
"'E
0.0080
""
(I)
50 10) (

25 75_
V .0034

.0024
o 5 rJ
1 10 100 1,000 10,000 100,000 1 10 100 1,000 10,000 100,000
Number of cycles, N

(o) 1100 aluminum. (p) Beryllium.

Figure 4. - Concluded. Variation of stress range with cycles for strain cycling about a zero mean strain.

~
to
 180
Elastic l'
modulus '2 Nf
160
- 1-- ~~Nf
~-

'r!
, ~ B 1

...
til
..!.:l
140 ~ '2 Nf
<IJ"' - - ~- I- - -~ .iN"f
~~PA
----
rd
E ,
'r!
r-I 120
~
til
til
<IJ 100
1-1
+>
m

80r---ff-+----.~-----r--~~

60~~--~--~~----~--~~

o .01 .02 .03 .04

Strain amplitude Cycles

(a) Static and cyclic stress- (b) Stress range as a function

strain characteristics. of applied cycles, logarithmic
scale for cycles.

Figure 5. - Static and cyclic stress characteristics of a hypo-

theticalcyclic strain hardening material.

40
 ~~ ____________ ~€d ____________~

o diametral

Figure 6. - Hypothetical axial stress - diametra1 strain relation during

strain cycling in which mean strain is zero.

41
~
N

160 320

---
II
140 - 280
J---
L---
120

100 r ~(
fQ
...9---
I-;::--~
o 0
240

200

80 160
sP
60 120

40 Virgin tensile tests -- 80

- - - Cyclic tests
orl
ro
..>: 20 I I 40
~

'0

ro~ 0 o
ro
(jJ
H
(a) AISI 4130 (soft). (b) AISI 4130 (hard).
-1-0
C!l 140 280

120 240

100 200

r - -
...--c f------
_ _ ----c

L---o-~ ----
80 160

60 120
g
40

20
80

40 r
II
0 .02 .04 .06 .08 .10 0 .02 .04 .06 .08 .10
Longitudinal strain

(c) AISI 4340 (annealed). (d) AISI 4340 (hard).

Figure 7. - Comparison of cyclic and virgin tensile stress-strain curves.
 -
360 180
__ ~II

320 160

----
IJ>-
°f - - - -°
280

/ ~
u
_-----0
140
0'",,--
~ - 1-----r::J'

240 120

200 I
yo ,,/
V
100 I
160 80

120 1 60
I

80 t 40
$ - I--
-
oM

'"
~
40
II Virgin tensile tests
- - - Cyclic tests
_ 20 lL
b 0 I I o
"'~
(e) AISI 52100. (f) AISI 304 ELC (annealed).
'"
ill
180

----
....,H
CI) _ --0-- I-

160
V·..o--
140 /"
. ?"
120
.....-
~
100
f ~
f ~ ----

---- ---
80

60
t /
V
f-

40 ~ ~

20 1'---
0 .02 .04 .06 .08 .10 0 .02 .04 .06 .08 .10
Longitudinal strain

(g) AISI 304 ELC (hard). (h) AISI 310 (annealed).

11>- Figure 7. - Continued. Comparison of cyclic and virgin tensile stress-strain curves.
CJ.l
l!>-
I!>-

32

28\r

(~
_ -0---- 0 0

24r L--o-
20 0
p
~..-o-
#'
16 0
, l
f I
-rl
'"
.=<

b
12 0

4
oj
ot-
~
-- v
I
~
I

'"'"
Q)
~

H
-P
(i) AM 350 (annealed). (j) AM 350 (hard).
Ul
240

200

r- - .f\.
160

120
V-
i
80

40 r
II .
Virgin tensile tests __
- - - Cyclic tests
~ I
o .02 .04 .06 .08 .10 0 .02 .04 .06 .08 .10
Longitudinal strain
(k) Inconel X. (1) Titanium (6Al-4V).
Figure 7. - Continued. Comparison of cyclic and virgin tensile stress-strain curves.
 90

80
~ -U-
-
70
~
!:/'
r
60

50 I

40

30

orl 20
rll
~

b 10
;;
rll
ill
H
0 o .02 .04 .06 .08 .10
.p
[/)
(m) 2014-T6 aluminum. (n) 5456-H311 aluminum.

24 60

--,-
--
50 --0--
20 ~
..(y'-
16 40
~
12 lL 1---- -- 30 /
If~,o- /
_--0---

8 20

4i
It§
--l J -l 10 I Virgin tensile tests
II - - - - - Cyclic tests
I I
0 .02 .04 .06 .08 .10 0 .002 .004 .006 .008 .010
Longitudinal strain

(0) 1100 aluminum. (p) Beryllium.

Figure 7. - Concluded. Comparison of cyclic and virgin tensile stress-strain curves.

If'>.
CJl
I!>-
m r I I I I I
~
I \
-\
AM 350 (annealed)
I
0 AISI 4130 (sof't)
•
0
AISI
AISI
4130 (hard)
4340 (annealed)
~
'\l
AM 350 (hard)
Inconel X
•
~
AISI
AISI
4340 (hard)
52100
0
Cl
D
Ti tanium (6Al-4V)
20l4-T6 aluminum
5456-H3ll aluminum
0 AISI 304 ELC (annealed)

•
LI
AISI
AISI
304 EIC (hard)
310 (annealed)
0
0
llOO aluminum
Beryllium

3.0 Nontailed symbols denote ~

N
v
strain-softening materials ~
Tailed symbols denote
Ul
strain-hardening materials
J'>
---t? 2.6
"'
:S
~
(])
H
+>
Ul

cd
.-I
(])
'r!
2.2
?>
0
+> - Strain
softening
Strain hardening ~
.-.
.s::
+>
bO
i=1
(])
H
+> 1.8
Ul if C
(])
.-I
'r!
Ul
i=1
(])
+>
G-i
0
1.4
6 0
or! .. C
+>
&! 00
o· ~ q Cl
~
1.0
o .4 .8 1.2 1.6 2.0 2.4 2.8 3.2
Stress ratio at l-percent strain

Figure 8. - Comparison of stress ratio at l-percent strain with virgin tensile proper-ties.
 101,----.-----,-----,-----.-----,----~ ,-----,-----,-----,-----,-----,-----,10°

100·~--~--~~----_+----_+----~----~ r-----r-~&-r-----r-----r-----T---~10-1

10- 2 10- 3
~
.r<
~ "-
-rl ~
.r<
"-
~ ;-
-rl
<I
;- 10- 3 0- 10- 4
<I oJ
ao
OJ
bO
- (a) AISI 4130 (soft) . (b) AISI 4130 (hard) .
~
cO
H
~
cO 101 10° ~
H .r<
cO
~ H
-rl 0 /',E L +>
cO eL Ul
H
+> L rl
CI) 0 Ep cO
+>
0
L'. /',E L 8

10 0 10- 1

r-----+-~--+-----~----~~---r----~10-2

o 0

10- 3 10- 4
1
10° 10 10 2 103 10
4 105 10
6 100 101 10
2 103 104 105 10 6
Cycles to failure, Nf
(c) AISI 4340 (annealed). (d) AISI 4340 (hard) •
Figure 9. - Elastic, plastic, and total strain range variations with cycliC life.

47
 101.-----r-----r-----r-----.-----.-----,

100~~~~----+-----+-----+_----+_--~

10- 3
C
0,-{
C
0,-{
'-.
:-
~
0,-{
C
0,-{
.;;
.;; <l
<l 10- 3 0 10- 4 a!bO
a! ( e) AISI 52100. (r) AISI 304 ELC (annealed) .
~
ro
bO
~ 1 H
ro
H
10 100 ~
0,-{
~ ro
0,-{

i;€~L
.,H
.,ro
H
U1
0
€L
00

rl
0 p .,ro
L 0
t:,. i;€ E-I

10° 10-1

10-1r-----+-~--+_----+_--~~----1_----~

(g) AISI 304 ELC (hard). (h) AISI 310 (annealed).

Figure 9. - Continued. Elastic, plastic, and total strain range variations with
cyclic life.

48
 101 ,-----.-----.-----,-----,-----,------

100~---t;~----_+----~----~----~----~

0
0

10- 2 10- 3
.:
.r<

.: :-
~
.r<
:-
~
0
or{

<JJ~
or{ 0
<I
~

<JJ
<l 10- 3 10- 4 OJ
b()
~
(j) AM 350 (hard).
" (i) AM 350 (annealed). aJ
b() H
~
1 10 0
aJ 10 ~
or{
H aJ
~ 0 LEt ...,
H
.r< eZ til
aJ L
...,
H 0 Ep rl
CI) ...,aJ
t; t; LEL 0
E-<
t;
100 10- 1

10- 3 10- 4
0 1 2 3 4 5 6 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10 10 10 10 10
Cycles to failure, Nf
(k) Inconel X. (1) T1 tan1um (6Al-4V) .
Figure 9. - Continued. Elastic, plastic, and total strain range variations with cyclic life.

49
 101 ,----.-----.-----.-----.----.-----, ,----,-----.----.-----.-----,----.100

10°r---_+~--~----+-----r---_+----~

10- 2

.::
.,;
.:: -......
.,;
.::
':- .,;
>: 0
.,; .;;
<I
tv' 10- 3 10- 4
<I oJ
bD
(m) 2014-T6 aluminum. (n) 5456-H311 aluminum. >:
oJ oj

~ I'-<

10 1
oj
I'-< 100 >:
.r<
>:
I oj
.,;
0 {:,El, ..,
I'-<

..,
oj
6.
I'-<
CI)
6. el,
1,
"
rl
E oj
0 -I-'
P 0
{:,E 1, 8
6.
-1
100 10

0
0

'A..,
10-1r---~r_--~----_+--~~----~----~ ""- ~
~
~~
tI",

~
~
o 0
-0-: =- ..n
~
--0-
0.....
100 10 1
failure, Nf
(0) 1100 aluminum. (p) Beryllium.
Figure 9. - Concluded. Elastic, plastic, and total strain range variations with cyclic life.

50
 0
10

I II I I I I II ~ I I II I I
4130 !SOft~ /),. AM 350 ~annealed)
0
< •
0 AISI
AISI
AISI
4130 hard
4340 annealed)
... AM 350 hard)
Inconel X

~
()
0
•
0

~
AISI
AISI
4340 hard)
52100
"
Q
Cl
Titanium (6Al-4V)
2014-T6 aluminum
10- 1 0 AISI 304 ELC ~annealed) D 5456-H3l1 aluminum

-=
'r-.
•
LI
AISI
AISI
304 ELC hardl
310 (annealed
0 1100 aluminum

:5
'"
<l
to.
" v
I....
<;;~
[:,
,; "'t>.lJ 0
Q()
c I." AV~ "'-
m
~<;; ~I€
H
C
/),. ~D
-rl
m 10- 2
Ie
H
.p
rn
~.
rl 11
m In I!!l
H
.p

'm" ~ ~ ,;. r-A ..,

"'
-rl
q
[ -r I'ro.
A'"',
b~ ~ ...
.... a;
A -
" 'i
<;;2:1'
~" LI1!J S\ 14]
~~ ~

~
10- 3 " LI ,

10- 4
0 1 2 3 4 5 6 7
10 10 10 10 10 10 10 10
Cycles to failure, Nf

Figure 10. - Effect of diametral strain range on fatigue life.

U1
I-'
CJl
N

II111 II II II II J II III
JI II I
I I II
IIIII I I I I IIIII I I I I IIIII I I I

o AISI 4130 1SOft) • AISI 304 ELC (hard) Q Titanium (6Al-4V)

• AISI 4130 hard LI AISI 310 (annealed Cl 2014-T6 aluminum
o AISI 4340 annealed) ~ AM 350 (annealed) D 5456-H311 aluminum
• AISI 4340 hard) A AM 350 hard) o 1100 aluminum
10 3 t>. AISI 52100 'V Inconel X o Beryllium
-rl
OAISI 304 ELC (annealed)
Ul
.>: ~~
1:>'
...
~I!I!II A
.; ..vP41 1"- -~
<l -L>.
(> IA:iL ~ .1:> I~

~
:<':
~O I~
.::I P 00
'J
n -In rt*, .~
~ I~~ l&:l '..AI
t:..
A
u
P Ll
~Ll
C I~ 0" ~~~ ~~ ~~ p-
~
rllOJ
D
O .~ ~
0 •-4 I ~
+'
cO
Q)
102 ~
D
J Cl~ ~ ,rl
" r>-
bO
>::
cO
N l.o. '"' '"
Ul 0
Ul
Q)
N <;
+'
(1)
P 1" 0 p
P ~
~
101
100 101 10 2 10 3 10 4 105 10
6
10
7
Cycles to failure, Nf

Figure 11. - Variation of stress range at ~ Nf with fatigue life for constant strain range cycling.
 10-
Failure in threads
i'..
"'-.....
" .6E7,
i'.
Test-section
failure ..
'r-.
Z = 1 5 &7,
.6Esr
~
-- • e7,

"-
-~ r- -I- \ Transition Nf ;: 700 cycles
. :-1_t--_ "-
~~
>=1

----.
·rl
r-- r-, r- r- -- --
--- I" --
I--
.6E7- -/ 1-- I"-

---
>=1
e7-
·rl

"
~
"'- ........
I--
- -l-
-- -
~
(])"
b.O
10- ::;
I-- r--
- r--. .....
r- I"-
~
:-- ~
'~-I~

>=1
al
H
>=1
·rl
al
r--
-
r- - I -

H
.p
(J)

10- :5 --- ---

0
101
2 3 4 5 6
10 10 10 10 10 10
Cycles to failure~ Nf

(Jl
Figure 12. - Graphical illustration of failure transition in test speclirren based upon strain-life
~ relations for AISI 52100 steel and an assumed stress concentration factor.
 ~

10 2 10 2

N
NQ) N
NQ)
IJJ
IJJ
~ <I
N
IJJ ~
NUl
<I
~
IJJ
r AISI 4130 <I
o ~hard)
..-l
.p H
H
ro
~ ~ 1'--. ""I'---. r'/ AISI 4130
(soft)
Q)
Ul
..-l
(lj
H
"@H 101 ~ 10
1
Ul
.p Ul
Ul
....... .......... Q)
I".: ......... Nf == 700 H
.p
<:) ....... ........ 1-...
..-l Ul
.p ..........
Ul "I-- ........ .p
(lj
........
~
..-l
H
Q) ..... t-... Nf = 7,000
Nf = 17,000 0
.p ~ ~ISI ........ t--

--
..-l
52100
<:)
ro r-............. I'-- ........ llE~r
.p
(lj
H

5
ro t--1-
t- " ....... r--..... r-.. r-~-=
llEe~
1.5 H
~
~
I
I--- l"- ..-l
ro
~ r-- I--- l-

~~~- r-.-
H
I---
t: r=====-==--- -::r- ::...
.p
f-- -b, lZl

r---I---t-- t---~
100
101 10
2
'- '----
10 3
i--

104
- r- N-
105
10
0

Cycles to failure, Nf

Figure 13. - Relative notch sensitivity behavior of threaded specimens as indicated by strain ratio against
life curves.

/\
 2 2

10 PS S421 I III/II Ii] I 111111 I I I 111111 I I I 111111

~
I I I 11111110 co

~Q)
~

~ ~Q)
~

'" {
>'"
<I r AISI 310 ~
1100 aluminum ~ ~~B
I I I II
-.l
I
i I I I I III I I I I I I III ~"'

~ (annealed) 1'--.. '"

<I
o
..-{ I.............~ f"-.... 11111 I I I 111111 I I 1111111 fu
-I-'>
ttl ...... ~~ r AISI 4340 r---)--.. "'ttl
r
H
~1 ~/ (hard) N I ..-{

<=: ..... , I ~"" ~ AISI 304 , '''\- H

..-{
101 ..... / ~ (hardl"--J::: 5456-H311 Il ' 101
~ .... / aluminum ....... "'"'
Q)
-I-'>
~:--.. ...... / _1111 :::--, H
" ........... ~ .... ...: ......... .......... ~AISI 4340
-I-'>

"
..-{
-I-'> , ........ ~~ I::::::" 7-r....... . . / (annealed) r-.... "
-I-'>
!!.:ryllium, .~ -...... I'>-.... 1'-,'1-...1' , , , ttl
"'
..-{
H -_. I \ ..........:::: ~
7"--.-
.... b,.. 't--:J: .... o
Q) ..-{
-I-'> -I-'>

"~ ~
ttl <=:

~:::f~~~~~~~~~~~~i:~~~~r+~~~~-±~~~~~~:tjjHl------II--;/r-(annealed)
.<:! Titanium , AISI 304 ..-{
co
~
-I-'>
C/l

100
1 10 3 lb 4 10~
10 102
Cycles to failure, Nf
Figure 14. - Relative notch sensitivity behavior of buttonhead specimen as indicated by strain ratio against life curves.

(J1
(J1
 ------ ~ -. - _. --- _.. _,- - - ---- - - .- -- - .

Figure 15 . - Protective assembly for testing beryllium.

56 NASA-Langley . 1963 E-1590