Professional Documents
Culture Documents
By far, the majority of engineering design projects involve machine parts subjected to fluctuating or
cyclic loads. Such loading induces fluctuating or cyclic stresses that often results in failure by fatigue.
Usually, the stresses are fluctuating and their nominal magnitude is well within the elastic range of the
material. About 80% of all mechanical component failure under service are due to fatigue.
1853 F Braithwaite, First to introduce the term fatigue in a research paper “On the Fatigue and
Consequent Fracture of Metals#, Proceedings of the Institution of Civil
Engineers ICE v.13, pp.463-475, 1853-1854
1860 August Wohler Carried out a systematic investigation on the failure of steel rail axles
subjected to alternating bending moment. He introduced the stress
amplitude-life (S-N) curve (Wohler-curves) to characterize the fatigue
behavior of the steel, he showed that higher stress amplitudes lead to
shorter fatigue life, he described the concept of endurance limit that is a
stress amplitude value below which there is no fracture and he stressed the
importance of amplitude and mean stresses.
1886 J. Bauschinger, 1886 Change in the elastic limit by often repeated stress cycles. ”Uber die
Veranderung der Elastizitatsgrenze und der Festigkeit des Eisens und Stahls
durch Streckenund Quetschen, durch Erwarmen und AbkuJhlen und durch
oftmals wiederholte Beanspruchung”. Mitt.
Mech.-Tech.Lab. Munch. 13, 1-115.
1910 Basquin O. H.,1910 “The exponential law of endurance tests”. Represented the finite life region of the "Wöhler curves"
Proc. Annual Meeting, American Society for Testing Materials, in the form (log σa) on the ordinate (log N) on the
Vol. 10, pp. 625-630 abscissa and describe
1903 J.A. Edwin and J.C.W. Humfrey “The Fracture of Metals under Used optical microscopy to study the onset of fatigue
Repeated Alterations of Stresses” Phil. Trans. Roy. Soc, London cracks. They observed that localized slip lines and slip
v.CC, 1903 p241 bands lead to the formation of microcracks
1874 H. Gerber (Gerber, H. (1874), Bestimmung der zulassigen Investigated the effect of mean stress on the life of the
Spannungen in Eisen-konstructionen. Zeitschrift des Bayerischen component
Architeckten und Ingenieur-Vereins 6, 101),
1899 Goodman, J. (1899). Mechanics Applied to Engineering. London: Investigated the effect of mean stress on the life of the
Longmans Green component
1924 Palmgren, A. (1924). Die Lebensdauer von Kugellagern. Zeitschrift Damage accumulation model developed by Palmgren and
1945 des Vereins Deutscher Ingenieure 68, 339-41. later by Miner
Miner, M.A. (1945). Cumulative damage in fatigue. Journal of
Applied Mechanics 12, 159-64.
1954 Coffin, L.F. (1954). A study of the effects of cyclic thermal stresses Provided an empirical relationship between the number
on a ductile metal. Transactions of the American Society of of load reversals to fatigue failure and the plastic strain
Mechanical Engineers 76, 931 and Manson, S.S. (1954). Behavior amplitude.
of materials under conditions of thermal stress. National
Advisory Commission on Aeronautics: Report 1170. Cleveland:
Lewis Flight Propulsion Laboratory.
1921 Griffith, A.A. (1921). The phenomenon of rupture and flow in Provided a mathematical tool for quantitative treatment
1957 solids. Philosophical Transactions of the Royal Society, London of solids with cracks.
1961 A221, 163-97. Showed that the amplitude of the stress ahead of a crack
Irwin, G.R. (1957). Analysis of stresses and strains near the end of could be expressed in terms the stress intensity factor, K .
a crack traversing a plate. Journal of Applied Mechanics 24, 361-4. Suggested that the increment of fatigue crack advance
Paris, P.C, Gomez, M.P. & Anderson, W.P. (1961). A rational per stress cycle, da/dN, could be related to the range of
analytic theory of fatigue. The Trend in Engineering 13, 9-14. the stress intensity factor, AK, during constant amplitude
cyclic loading
Fluctuating Stresses
All fluctuaDng stresses can be decomposed into a stress amplitude and a stress mean.
From ASTM E1823
fatigue—the process of progressive localized permanent
structural change occurring in a material subjected to
conditions that produce fluctuating stresses and strains
at some point or points and that may culminate in cracks
or complete fracture after a sufficient number of
fluctuations.
cycle—in fatigue, one complete sequence of values of
force (strain) that is repeated under constant amplitude
loading (straining).
test whose loading mimics the expected cyclical loading that will be experienced by the material in the
0.25 application. This means that if the part will experience cyclical uniaxial tension, then the test data should
be generated in uniaxial tension. For expected bending or torsional stresses, the test data should be
The0 fatigue characterization of a material must
S-N Diagram begenerated
related toorthe
in bending torsion,application
respectively. that is the data for
0 0.5 1 1.5 2 2.5
fatigue characterization must be obtained Stress Ratio
from aNext, type of loading
the stress ratio, or R ratio,that mimics
of the test the expected
must be determined. This is simply thecyclical
ratio of the minimum
The loading that will be experienced by the
most extreme case would be if each cycle consisted of loading the test specimenmaterial inthisFatigue
to
(R Ratio) a given stress levelthe application.
stress experienced during a cycle to the maximum stress experienced during a cycle. It is important to note
(in bending or torsion), and then loading to an equal and opposite stress level in the other direction. In that and Stress Ratios (continued) At other times, a part such as an electro-mechanical
the stress values can be positive (tensile stress) or negative (compressive stress). In the easiest case to
case,This means
the minimum that
stress would if negative
be the the part will experience
of the maximum cyclical
stress, and the R ratio would be exactlyuniaxial
relay maythebetest
visualize, tension,
preloaded
specimen to astarts
certain
outthen
stress levelthe
unloaded test
that isstress),
(zero isdata
never removed toshould
loadedduringthethe life of be
required the part. The
maximum part
positive
Unidirectional
equal to -1. This is known as fully reversed (R=-1) testing. A rotating axle or shaft subjected stress to a level, and then unloaded to start the next cycle. In this case, the R ratio would be exactly zero. the
may then be loaded to the maximum stress level and then released back to the preload. In this case,
generated in uniaxial tension. For expected
(R=0) Testing bending
constant bending stress would be an example of this condition, as each point on the circumference ofisthe or
minimum torsional
stress would be stresses,
some small positive the
value test
associateddata
with theshould
preload,
called unidirectional (R=0) testing. A spring contact in a modular jack would be an example of this and be
the maximum
This
shaftgenerated in bending
would alternately experience or torsion,
tension and compression as it rotatesrespectively.
around the center. stress would
loading be a(assuming
condition larger positive
that thevalue. In isthisnotcase,
contact the R ratio
preloaded in anywould
way.)be greater than 0 and less than 1.
Fully Reversed Another possibility is to have a minimum compressive stress of lower magnitude than the maximum
1.5 (R=-1) Testing stress. This would result in an R ratio between -1 and 0. This could be the case in some switch and
R=-1 Fully Reversed Stress relay1.25contacts. R=0 Unidirectional Stress
1
1
0.5
Mininum Stress
1.25
0.75
R=0.5 Fluctuating Stress
Mininum Stress
Cyclical Stress
Fatigue
0 and Stress Ratios (continued) At other
Cyclical Stresstimes, a part such as an electro-mechanical 1
Maximum Stress
0 0.5 1 1.5 2 2.5 0.5 Mininum Stress
relay-0.5
may be preloaded to a certain stress level that is never
Maximum removed during the life of the part. The part
Stress
may then be loaded to the maximum stress level and then released back to the preload. In this case, the 0.25 References:
0.75 Cyclical Stress
Maximum Stress
minimum -1 stress would be some small positive value associated with the preload, and the maximum 0.5
stress would be a larger positive value. In this case, the R ratio would be greater than 0 and less than 1. 0
-1.5
Another possibility is to have a minimum compressive stress of lower magnitude than the maximum 0.25 0 0.5 1 1.5 2 2.5
stress. This would result in an R ratio between -1 and 0. This could be the case in some switch and 0
relay contacts. The most 0
extreme
0.5
case 1would1.5be if each
2
cycle
2.5
consisted of loading the test specimen to a given stress level
(in bending or torsion), and then loading to an equal and opposite stress level in the other direction. In this
case, the minimum stress would be the negative of the maximum stress, and the R ratio would be exactly
ush Inc.
1.25
R=0.5 Fluctuating Stress equal 1.25
to -1. This R=-0.5
is knownFluctuating
as fully reversed
Stress(R=-1) testing. A rotating axle or shaft subjected to a
1
constant1 bending stress would be an example of this condition, as each point on the circumference of the
Mininum Stress
shaft would alternately experience tension and compression as it rotates around the center.
0.75 Mininum Stress
0.75 Cyclical Stress
0.5 Cyclical Stress
Maximum Stress 1.5
0.5 0.25 R=-1 Fully Reversed Stress
Maximum Stress
10
0.25 The next issue of -0.25
0 0.5 1 1.5 2 2.5
Technical Tidbits will 0.5
-0.5 Mininum Stress
0
0 0.5 1 1.5 2 2.5
continue the discussion Please
0
-0.75 contact your local Cyclical Stress
on Fatigue. sales
0 representative
0.5 1 for 1.5 2 2.5
Maximum Stress
-0.5
further information on
One important point to note is that the all sources of stress will factor into the fatigue performance of a
1.25 R=-0.5 Fluctuating Stress the
-1 fatigue
component. or other
This includes the design stress from mechanical sources (applied forces, pressures, or
Manufacture
Geometry Load Environment
1. Theaccepted
For life predicDon, four different models are generally nominal stress-life ( S - N ) model, first fo
and 1870s.
Choosing the fatigue life model is a significant decision. Currently four
such models exist for design engineers. These are:
The nominal stress-life model (S-N) (1850-1870)
The local strain-life model (e-N) (1960)
2. The localstress-life
1. The nominal strain-life ( Efirst
( S - N ) model, - Nformulated
) model, betweenfirst form
the 1850s
and 1870s.
3. The
The linear elastic fracture mechanics LEFM or fatigue fatigue
crack
2. The local strain-lifecrack
growth model growth
(1960)
( E - N ) model, ( d d dinNthe- A1960s.
first formulated K ) mo
3. The fatigue crack growth ( d d d N - A K ) model, first formulated in the
The two-stage model or a combination of the local1960s.
strain-life
1960s.
model and the LEFM model
4. The two-stage model, which consists of combining models 2 and 3 to
4. The two-stage
incorporate model,
both macroscopic fatiguewhich consists
crack formation of and
(nucleation) co
Methods to Determine Fatigue Life
There are two different material behaviors (regimes)
(a) Low-cycle FaDgue (LCF – Number of life-cycles < 103) Material subjected to large stresses.
(b) High-cycle FaDgue (HCF – Number of life-cycles >103) Material subjected to low stresses.
One of the objecDve of failure analysis is to predict the faDgue life of a mechanical component, subjected to specific
operaDng condiDons. There are three major faDgue life methods used in design, analysis and component-life
esDmaDon:
(a) Stress-life method (S-N Curves): It is used in HCF. It is a stress-based model. The model uses nominal stresses and
it relates them to the local faDgue strengths for a notched and unnotched members.
(b) Strain-life method (ε-N Curve): It is used in LCF. It is a strain-based model used when the cyclic load is high enough
to produce localized plasDc deformaDon. It relates the local strain at a notch, with the faDgue behavior of a smooth
specimen under strain-controlled faDgue.
(c) Linear elasDc fracture mechanics method (LEFM): Used to model the propagaDon of an exisDng crack (during
stage II). It is used when there is an exisDng crack or when the stresses are high enough to cause the formaDon of
localized cracks. It requires the use of fracture mechanics concepts, and the integraDon of the faDgue crack growth
rate equaDon to determine the number of cycles required to growth a crack from a given length to another length
and/or to fracture.
Fatigue failure of a bicycle crank spider arm
Stage I: Initiation/nucleation
Stage II: Stable growth
Stage III: Final Fracture
ethod, the specimen is sectioned along a plane which is oriented at a small
a (i.e. a few degrees) to the specimen surface. The profile of the surface, as
Crack Initiation (Stage I):
ed on the sectioned plane, is magnified by a factor of (1/ sin a), over and above
gnification obtained by other visual methods such as scanning electron micro-
From the point of view of a material scienDst, they are likely to consider the nucleaDon of flaws along
optical microscopy and optical interferometry. Hunsche & Neumann (1986)
persistent slip bands as the iniDaDon stage of faDgue failure, whilst a mechanical engineer may
efined this technique to obtain sections with sharp edges normal to the speci-
urface so that the surface features can be recorded to a resolution of 20 nm in
associate the resoluDon of crack detecDon with the threshold for crack nucleaDon.
anning electron microscope. Basinski & Basinski (1984), Hunsche & Neumann
, and Ma & Laird (1989a,b) have utilized this method to examine the details of
Wood's postulated (1958) is that repeated cyclic straining of the material leads to different amounts of
e roughening in fatigued Cu. Their results collectively indicate the following
net slip on different glide planes. The irreversibility of shear displacements along the slip bands then
l trends:
results in the 'roughening' of the surface of the material. This roughening is manifested as microscopic
The surface of the fatigued crystal is covered with PSB extrusions, intrusions
and protrusions. A protrusion is a surface uplift (a large extrusion), many
'hills' andin'valleys'
micrometers height, whereatasites where
macro-PSB, tens slip bands emerge
of micrometers wide and at the free surface. The valleys so generated function
as micro-notches and the effect of stress concentration at the root of the valleys promotes additional
containing tens of matrix and/or PSB lamellae, emerges at the free surface.
A protrusion may contain several intrusions and extrusions. Figure 4.2 is a
slip and
scanning fatigue
electron crackobtained
micrograph nucleation
using the sectioning method which
142 Fatigue crack initiation in ductile solids
in crack initiation, removal of surface layer at New crack initiates from the surface. Firstly, because the surface grains are i
Fig. 10. Crack initiation sites [21, p. 16]
contact with the atmosphere, thus if environment is a factor in the fatigue damag
frequent intervals enhances the fatigue life
manifold.
The ASTM Committee E-08 recently adopted the
www.intechopen.com
use of either “nucleation” or “formation” to
describe the early stages of the fatigue process,
and dropped the use of the word “initiation”.
Crack Propagation (Stage II):
Since growth mechanism and propagaDon rate of a crack
differs according to its size, it is essenDal to define the crack
based on its size in increasing order.
i) Metallurgically or micro-structurally small crack which is
small as compared to a metallurgical variable such as the
grain size.
ii) Physically small crack in which the resistanceMetal toFatigue
crackand Basic Theoretical Models: A Review 217
This empirical approach was introduced by WohlerThe stress range, AS, is twice the alternating stress. Tensile
in 1860.
Wide-spread use in fatigue stresses are taken algebraically asstresses
positive and negative, respec
mean,analysis in applications
S,, maximum, where
S, , minimum, low-amplitude
&in,
Standard and range,
fatigue cycled
AS, of and
test methods stress induce
are indi-
procedures primarily
for metals are availa
ing stress isthese
an[4].absolute value.standards
The stress ratio, R, and the a
elastic deformation, incated.
a TEST
4.1 FATIGUE LOADING, component
MACHINES, that
The algebraic
AND is designed
relationships
SPECIMENS 61 for a
among
from long
ASTM life (HCF).
terms
These are:
pertinent are under the direction of AST
ratio, A, are used frequently in fatigue literature, where
Committee E-08 and are given in Table 4.1. Both ASTM standard designatio
+ The stress range, AS,and is titles
twice the in
are given alternating stress. Tensile
Table 4.1. International Organization orforcompressiv
Standardizati
stresses are taken algebraically as positive
(ISO) draft standards and negative,
on fatigue respectively.
testing of metals are availableAlterna
R is thecurrently
through t
stressavaila
ISO, Geneva, Switzerland. Three separate draft standards
ing stress is an absolute value.
are given The4.2stress
in Table ratio,
with both R, and
designations andthe
ratioalternating
titles. and A is
Additional I S 0stres
fatig
ratio, A, are useds frequently
, = Smax +2 Smin
testing draftin fatigue
standards literature,
are being where
developed.
y;: the alternating
e R = -1 and R = 0 are two common reference test con
6i
4.1.3 Fatigue
obtaining fatigueTest SpecimensR = - 1 is called
properties. stressthe ratio
“fully reve
0
Time Smin
sinceCommon is test
equal to -Smax;
specimens R = 0 fatigue
for obtaining wheredata
Smin 0 is called
are=shown “puT
in Fig. 4.5.
YLx I
-I
Onespecimens
cycle isshown in Fig. 4.5~-f
the smallest have beenofused
segment thetostress
obtain total
versusfatigue
timli
Stress ratios: [ =
A = S IS,,, that includes crack nucleation life and crack growth life. These specime
R =AND-1SPECIMENS
4.1 FATIGUE LOADING, TEST MACHINES, and R 65repeated are periodically,
= 0usually two common
have finely as shown
polished surfacesintoFig.
reference 4.2.conditions
test
minimize Under variable am
usedeffec
surface roughness fo
The stress range, AS, is twice the the definition
alternatingofstress. Tensile or compressive of
Figure 4.2 Nomenclature for constant amplitudeobtaining
cyclic loading.fatigue[4] properties. = one
- 1 cycle
R negative, is not
is called theclear andreversed”
“fully hence reversals
conditio
TABLE 4.1 ASTM Standard Practices stresses are
Related to taken
Fatigue algebraically
Testing of Metals as positive
considered. and
In respectively.
constant amplitude Alternat-
loading, one cycle equal
E466 since Smin is equal
ing stress is an absolute value.while
Conducting Force Controlled Constant Amplitude Axial Fatiguc Tests to -Smax;
The stress R = 0 where Smin
ratio,amplitude =
R, and theloading0 is called
alternating “pulsating
stress tension
in variable a defined cycle may contain
E467
of Metallic Materials
ratio, A, areOneused cycle is
frequently the
Verification of Constant Amplitude Dynamic Forces in an Axial Fatigue insmallest
fatigue segment
literature, of
wherethe stress versus time history that
TABLE 4.2 I S 0 Draft Standards Related to Fatigue Testing of Metals
Testing System repeated periodically, as shown in Fig. 4.2. Under variable amplitude loadin
, S,,
E468maximum, S, ,of
Presentation minimum, andFatigue
Constant Amplitude
&in, range,Test of stress
AS, Results for are indi-
Metallic ISO/DIS 12106 Metallic Materials-Fatigue Testing-Axial Strain-Controlled
. The algebraic the
relationships among these terms are:
Materials definition of one cycle is not clear and hence reversals of stress are ofte
Method
E606 Strain-Controlled Fatigue Testing considered. In constant amplitude
ISO/DIS 12107 loading,
Metallic one cycle
Materials-Fatigue equals two
Testing-Statistical reversal
Planning and
E647 Measurement of Fatigue Crack Growth Rates Analysis of Data
E739 while in( Svariable
Statistical Analysis of Linear or Linearized Stress-Life - N ) and amplitude
ISO/DISloading
12108 a defined
Metallic cycle mayTesting-Fatigue
Materials-Fatigue contain many reversal
Crack Growth
Strain-Life ( E - N ) Fatigue
R Data
= -1 and R = 0 are two common referenceMethod test conditions used for
El012 Verification of Specimen Alignment Under Tensile Loading (under the
s, Smax
jurisdiction of Committee
=
+ Smin
obtaining fatigue
E-28 on Mechanical Testproperties.
Methods) R = - 1 is called the “fully reversed” condition
El049 2
Cycle Counting in Fatigue Analysis
El 823
since Smin is equal to -Smax;R = 0 where Smin= 0 is called “pulsating tension.”
Standard Terminology Relating to Fatigue and Fracture Testing
One cycle is the smallest segment of the stress versus time history that is
repeated periodically, as shown in Fig. 4.2. Under variable amplitude loading,
the definition of one cycle is not clear and hence reversals of stress are often
Stress-Life Method
(a) based on stress levels only
(b) Based on the amplitude stress –number of cycles S-N diagram
(c) Not very accurate prediction for low-cycle applications.
(d) Accurate prediction at high-cycle applications
(e) Most traditional method with plenty of supporting data.
S-N Diagram
The strength of the materials under fatigue
load are determined in the laboratory
€
The criterium for yielding can
be modified to take into
account the yield strength (Sy)
in monotonic tensile test and
the cyclic yield strength (Sy’).
Miner’s Rule
Most of the fatigue testing has been carried out at a constant amplitude fatigue loading. The problem
faced by the designer is how to use constant amplitude fatigue data in the prediction of fatigue lives
when the component is subjected to a wide range of variable amplitude load history (load spectrum).
In predicDng the faDgue life of ball bearings Palmgren (Palmgren A (1924) Die Lebensdauer von
Kugellagern (The durability of ball bearings). Verein Deutscher Ingenieure, 68: 339–341.) assumed that
the damage accumulaDon due to faDgue was linear with the number of stress cycles. Miner (Miner MA
(1945) CumulaDve damage in faDgue. J. Appl. Mech., 12: A159–A164. ) also suggested that the faDgue
damage at a given stress level could be considered to accumulate linearly with the number of stress
cycles. (Palmgren-Miner rule).
If a component that is subjected to stress S1 has a life of N1 cycles, then the damage accumulated per
cycle is 1/N1 and the damage after n1 cycles is n1/N1
n1 n2
For a two stress level that ends with failure at the second level, then + =1
N1 N 2
The generalization of this approach is called Miner's Law, and can be
written : nj
where nj is the number of cycles applied at a load corresponding to a S =1
lifetime of Nj . NJ
Example Example
Example7.7: 7.7: Cumulative
Consider a component Damage
subjected to the following loading:
Cumulative Damage
20% of its life at175MPa,
30% of its life at 220MPa…… ;
Loading 20% @ 175MPa 30% @ 220MPa 40% @ 250MPa 10% @ 275MPa
Loading Example
Example 7.7:
20% @ 175MPa Cumulative
30% @ 220MPa
7.7:268,564 40%Damage
@ 250MPa
Cumulative 46,048
Damage 10% @ 275MPa
Life @ Load ∞ 12,367
Life @ Load
Loading 20%∞@ 175MPa 268,564
30% @ 220MPa 40%46,048
@ 250MPa 10% 12,367
@ 275MPa
Loading 20% @ 175MPa 30% @ 220MPa 40% @ 250MPa 10% @ 275MPa
Life
S-N Diagram @forLoad
Steel Specimen with ∞ut ==440MPa,
withSS
268,564 46,048 12,367
Life
S-N Diagram @for
Load
Steel Specimen ∞ 440MPa, Axial
ut
Load
Axial 268,564
Load 0 . 2 L
0.2 L + 0.3L 46,048
0 .3 L 0+.4 L0 .4 L 0+.112,367
L0.1L = 1
0 450
+ + + =1
S-N Diagram for Steel Specimen with Sut = 440MPa, Axial Load
S-N Diagram for Steel Specimen with Sut = 440MPa, Axial Load
∞ ∞
0.2 L 268268
,564
0.3L
,564 46 ,
0.4 L
46
048 , 04812 ,
0.1L
12
367 ,367
450 0.+2 L 0 .3 +
L 0 .4 +
L 00.1.=1
L1
0 400 450 0.02∞.2 268 0+.30,.564
3 46 0.4 120+.,1367
0+.4,048 1=1 1
++ ∞ 268+,564 + 46,+048+ 12,367 = =
400 ∞0∞.2 268 0,564
268 .3 ,564460,.048 46
4 ,048 120,.367
112,367 1L L
0 350 400 0+.2 0.3+ 0.4+ 0.1= 1
∞ + ,564 46+,048 12+,367 L=
268 1
350 00+ 1.117E
∞ 268
+ 1.117E + 8.687E
- 06-,564
06 - 06 -+12
46,048
+ 8.687E 8.086E
,367
06 -L06 =- 06 = 1
+ 8.086E
300 350 1L
1 L
Fatigue Strength (Sf)
300 1.78899E - 05 = L
Fatigue Strength (Sf)