Fatigue of Materials

• Most of the failures in machinery are due to time-varying loads rather than to
static loads.

• These failures typically occur at stress levels significantly lower than the yield
strengths of the materials.

In 1800‟s railroad-car axles made of ductile steel began failing after limited time
in service (exhibited brittle-like failures)!!

Rankine published a paper in 1843, “On the causes of unexpected breakage of

journals of railway axles” - postulated that material had crystallized and become
brittle due to a fluctuating stress
Fatigue of Materials

Fracture surface
Railway axle
 Fatigue of Materials

 These axles were designed with all the engineering expertise at that time,
but design was based on static loading conditions !!

Dynamic loading conditions were a new phenomenon during those times!!!

August Wohler (a German engineer), made the first systematic

investigation/testing on axles to failure in the laboratory under
fully reverse loading !! (12 yrs of work)

Found that the number of cycles of time varying stresses - real reason for
failure
- Found the existence of Endurance Limit in steels (1870)

S-N diagram / Wohler diagram

 Fatigue of Materials

Time-varying stresses

The term Fatigue was first

coined by Poncelet in 1839.
Mechanism was not clearly
understood!!
 Chronology of fatigue failure events
1839 Poncelet used the term Fatigue
1837 Rankine discussed the theory of crystallization of fatigue
1871 Wohler published his results on axle failures by fatigue,
developed the rotating bending test, S-N diagram, and
defined endurance limit
1903 Ewing/Humfrey discover slip lines, fatigue cracks - disproved
crystallization theory
1910 Basquin developed the exponential law of endurance tests
(Basquin equation)
1915 Smith/Wedgewood separated cyclic plastic strain from total
plastic strain
1927 Moore quantified high cycle fatigue data
1930 Goodman/Soderberg independently determined the influence
of mean stress on fatigue.
 Chronology of fatigue failure events
1937 Neuber published Neuber equation for strain concentration in
notches
1955 Coffin/Manson independently published strain-based low
cycle fatigue law (Coffin-Manson law)
1961 Paris published fracture mechanics based Paris law for
fatigue crack growth

Definition of fatigue: ?
 Fatigue Crack Growth

 Once a crack is present in a material, it will tend to

grow under the influence of cyclic loading.

 The crack may be initiated by fatigue, or may be pre-

existing from manufacture, or may be caused by an
impact, or similar event (e.g., a thermal shock.)

 The crack will grow to a critical length then fracture of

the component will occur.
 Cyclic Loading

 max   min
 mean 
2
 range   max   min
 max   min
Rotating Machinery  amplitude 
2
 min
Stress Ratio, R 
 max

Airframes, Bridges, Tanks, etc,

 Three stages of fatigue failure

 Crack initiation stage

 At a microscopic level metals are not homogeneous and isotropic
(eg., inclusions, porosity),

 There always exists regions of geometric discontinuities (eg.,

notches) at locations of significant time-varying stresses.
 Local yielding at these stress concentrators (yet nominal stress is
well below the yield strength of the material).
 Localized plastic yielding causes distortion and creates slip
bands (regions of intense deformation due to shear) along the
crystal boundaries of the material.

 Coalescence of slip bands into microscopic cracks with continued

alternate loadings.
 Three stages of fatigue failure

 Crack propagation stage

 After a microcrack is initiated (or could have existed even before
the loading), the sharp crack creates stress concentrations larger
than that of the original geometrical discontinuity (eg., notch).

 Development of plastic zones at the tip of the crack each time a

tensile stress open the crack - blunting of the crack tip -
reduction of the effective stress concentration - small amount of
crack growth.
 Crack propagation rate is quite small, 10-8 - 10-4 inches/cycle.
 Observation of fatigue striations on the fractured surface of the
failed component.

 Effect of corrosive environment - rapid crack propagation

(corrosion fatigue)
 Three stages of fatigue failure
 Sudden fracture due to unstable crack growth

Fatigue striations on the crack

surface of an Aluminum alloy

12000X
 Fatigue striations/fracture
Steel keyed shaft failed in rotating Diesel engine crank shaft failed in
bending (crack started at key way) combined bending and torsion
 Fatigue-failure models
 Stress-life approach (S-N) - determine fatigue strength, endurance limit
 Strain-life approach ( - N) - crack initiation stage
 Linear elastic fracture mechanics approach (LEFM) - Crack
propagation stage

Fatigue regimes

Based on the number of stress or strain cycles that the component/machine

element is expected to undergo in its lifetime - it is relegated to either low-cycle
fatigue (LCF) or high-cycle fatigue (HCF).

N = 103 cycles

LCF HCF
 Stress-life approach
 Oldest of the three models - often used for high-cycle fatigue applications
(HCF) where the machine element/component is expected to last for more
than 103 cycles of stress.

 It is basically a stress-based model - the interest is to determine the fatigue

strength and/or endurance limit for the material so that the cyclic stresses
are kept below that level and avoid failure.

 This approach basically attempts to keep the local stresses at crack

initiation points (eg., notches) so low that the crack-initiation stage never
begins.

 Desired goal - Stresses/strains remain elastic everywhere and no local

yielding occurs to initiate a crack.
 S-N diagram/Wöhler diagram
 Wöhler‟s approach - Loading a rotating cantilever beam in bending to
achieve variation in stresses with time.

 R. R. Moore‟s approach - Loading a simply supported rotating beam in fully

reversed, pure bending.
(most of the information in literature on fatigue strength/endurance limit for various materials are
obtained from this approach)

- Highly polished specimen (0.3” dia) mounted in a fixture

- Constant magnitude, pure bending moment applied (specimen rotates at ~ 1725 rpm)

- Specimen subjected to fully reversed loading (bending stress) at any point on the
circumference

- Run the experiment at a particular stress until it fails; record the applied stress and no. of
cycles to failure

- Repeat the experiment for various specimens at various loads - generate S-N diagram
Fatigue-testing set up
 S-N diagram/Wöhler diagram

doesn’t break
(Endurance limit)

Fatigue test data now can be fit by the Basquin‟s equation of the form Nap = constant,
where a = stress amplitude, p = constant (slope) - (log-log scale).
 S-N curves and fatigue limit

Effect of notch & corrosive

environment on fatigue limit

S-N curves for wrought steels

S-N curve is not a smooth curve !

Relationship between fatigue strength and ultimate strengths (steels)

 S-N curves and fatigue limit

Influence of
processing ?

S-N curves for aluminum alloys (wrought, die-cast, sand-cast):

Unnotched-polished specimens:Rotating beam test
 Thickness Residual
Cr plating stress
(mm) (MN/m2)
0.0025 970
0.0075 510
0.025 77
0.15 77

Effect of chrome plating on the

fatigue strength of steel

Magnitude & sign of the

residual tensile stresses
will determine the fatigue
strength !

Effect of nickel plating & peening on the fatigue strength of steel

 Effect of environment

Environmental effects on the fatigue strength of steel

 Axial fatigue

Comparison with
axial fatigue data Difference between rotating-
bending test and axial fatigue
tests ?

10-30% lower values reported for axial fatigue

compared to rotating bending tests
 Strain-life approach

 Since in most of the machine elements/components the response of the

material at critical locations (eg., notches) is strain (deformation)dependent,
a strain-based approach seems more reasonable. In this approach the
plastic strain is measured directly and quantified.

 A strain-based model gives an accurate picture of the crack-initiation stage

and a more reasonable approach for fatigue life prediction. Crack-growth is
not explicitly accounted for in this method and hence this approach is more
or less for crack initiation life estimates.

 Combinations of fatigue loading and high temperature effects are better

handled by this method as creep effects can be included.

 Often applied to low-cycle fatigue (LCF) applications where the cyclic

stresses are high enough to cause local yielding.
Cyclic stress-strain behavior/strain
controlled fatigue
 If the loading process is reversed &
the specimen unloaded after yielding,
the stress-strain relationship will follow
a line parallel to the loading direction
(same slope) – AB

 Further loading, in the compressive

direction leads to the curve BC.

 In 1926, Massing observed that

stress-strain curve AC can be
obtained by doubling values of the
stress-strain curve OA. (assuming the
material exhibits symmetry along both
tension and compression
Cyclic stress-strain curve

 If the loading process continues

from the compressive zone to the
tensile zone (i.e from -max to +
max ) – one gets a hysteresis
loop (this forms one fatigue cycle
in the strain-life approach) –
equivalent to 2 strain reversals.

 Observe the strain and stress

amplitudes – they are half of the
total strain and stress ranges.
Cyclic stress-strain curve

 During strain-controlled cyclic

loading, the stress-strain response
of the material can change with the
no. of cycles.

 If the maximum stress keeps

increasing with each successive
cycle, then the material undergoes
cyclic hardening, & if the maximum
stress keep decreasing, then the
material undergoes cyclic softening.

 If the maximum stress does not

change with number of cycles –
then the material is supposed to be
cyclically stable
 Cyclic stress-strain curve

 The cyclic stress-strain curve is now comprised of two components of strain,

elastic strain (e) and the plastic strain (p)

 Now, the total strain can be expressed in terms of the stress (analogous to
an expression used for a non-cyclic stress strain curve)

where, K‟ = cyclic strength coefficient

n‟ = cyclic strain hardening exponent
 Cyclic stress-strain curve

 For a stabilized hysteresis loop (stabilization occurs say after about 100
cycles of imposed strain amplitude), the hysteresis equation would be:

 The above equation is valid only for those materials which exhibit symmetry
in tension and compression.
 Low-cycle fatigue (Strain-life approach)

For low-cycle fatigue applications, using the test

data, a plot of the plastic strain range (p) is
plotted as a function of number of cycles to failure.

LCF data for stainless

steel exhibiting a linear
dependence on a double
log scale

The above data can be fit using, Where, P/2 = plastic strain amplitude,
 P ‟f = regression intercept (fatigue ductility
  'f (2 N) C coefficient) and
2
C = regression slope (fatigue ductility
Coffin-Manson relation exponent)
 High cycle fatigue (low strains)/Modified
Basquin‟s equation

For high cycle fatigue – low strain cases, the Basquin‟s equation that is used
to fit S-N data can be modified as follows:

 e
a  E   'f (2 N ) b Strain-life equation
2

Where, a = stress amplitude

e/2 = strain amplitude (elastic)
‟f = fatigue strength coefficient
b = fatigue strength exponent
 Strain life curve
Now one can combine the Coffin-Manson equation (plastic strain amplitude)
and the modified Basquin‟s equation (elastic strain amplitude) to get the total
strain amplitude.
 Strain life curve

Influence of plastic Influence of elastic

strain on fatigue life strain on fatigue life  The transition life (2Nt) represent the
life at which the elastic and plastic
strain ranges are equivalent, which
can be expressed as

 Elastic strains have dominant

influence on fatigue lives above the
transition life and plastic strains
below the transition life (observe
figure)

Clear dilineation between low and high cycle fatigue

 Recap

 For long fatigue lives the fatigue strength controls the

fatigue performance and the Stress-life and Strain-life
approaches essentially yield the same results.

 For short fatigue lives, the fatigue ductility controls the

fatigue performance
 Linear elastic fracture mechanics(LEFM)
approach
 This is the best model for understanding the crack propagation stage during
fatigue process.

 Normally applied to low cycle fatigue (LCF), finite life problems where the
cyclic loadings are known to cause formation of cracks and is useful in
predicting the remaining life of cracked parts in service.

 This approach heavily relies on the expression for the geometric factor ()
of the stress-intensity factor and on the estimate of the initial crack size „a‟
K =  a

 Now, the crack propagation life can be determined from the assumed initial
crack size to a final crack size.
 Driving force for crack growth

 Driving force for crack growth is the range in the stress intensity
factor during cycling:

K  f (a/W)  a
   max   min for R  0
   max for R  0

Where, K = stress intensity factor range (Kmax – Kmin)

 = stress range
a = crack length
R = min/max (stress ratio)
 Crack growth rate (da/dN)

1 < 2 1

ac1 X
Crack Length, a

2
ac2 X

da/dN2
da/dN1

ao
Initial crack
length, ao

Cycles, N
 Paris Law
Paris Region
Sigmoidal curve
Stable Growth

Threshold Region
da
 A(K) m
dN

Rapid-unstable Growth
log(da/dN)

Slow Growth

Fast Fracture Region

m

A II

I III
Kth log(K)
KC
 Paris law

 Applied stress intensity factor (K) < Threshold (Kth) implies no fatigue crack growth.

 Fatigue crack growth life is found by integrating the Paris equation between a known
(or assumed) initial crack length and a maximum acceptable crack length.

da
 A(K) m
dN
ac
da
Np   m
a o A(K)

Initial crack size (ao) is usually found by inspection/assume a min. crack

length,
2
 Kc 
Critical crack size (ac) is found from a c  1  

  f (a/W)  max 
 Problem

When subjected to fatigue under  = 140 MPa, an alloy showed the

following Paris-type fatigue crack propagation relationship:

where K is in MPa.m1/2. Estimate the number of cycles required for the

crack to grow from 2 mm to 8.8 mm. The geometric factor can be taken as
1.12.

Answer: Nf = 1413 cycles

 Fatigue fracture surfaces

• Crack origin:

• Fatigue zone:

• Progression marks:

Macroscopic surface features

• Overload zone:
 Fatigue fracture surfaces

Progression marks
Ratchet mark - boundary
• indicative of direction of between two failure planes
propagation of crack
• indicative of multiple origins
• indicative of substantial
• relatively high stresses
changes in load
• not to be confused with
fatigue striations
 Origin of failures & river marks

• multiple ratchet marks indicative River marks - indicative of

of multiple origins of cracks direction of fatigue crack
• angle between ratchet marks • appear at fast-growing
indicative of primary origin sections of the fatigue zones
Torsional fatigue failures
Plane bending, reversed bending &
rotational bending fatigue failures