Fatigue of metals

Subjects of interest
• Objectives
• Introduction
• Stress cycles
• The S-N curve
• Cyclic stress-strain curve
• Low cycle fatigue
• Structural features of fatigue
• Fatigue crack propagation
• Factors influencing fatigue properties
• Design for fatigue
Objectives

•This topic provides fundamental aspects of fatigue in

metals and the significance of fatigue failure.
•Different approaches for the assessment of fatigue
properties, i.e., fatigue S-N curve and fatigue crack growth
resistance will be introduced.
•Discussion will be made on factors influencing fatigue
properties of metals, for example, mean stress, stress
concentration, temperature
• Finally design against fatigue failure will be highlighted.
 Introduction

Fatigue failure in a bolt

www.corrosionlab.com
Fatigue initiation

Beach mark
 Introduction

www.btinternet.com

Fatigue failure occurs at the outer rim

of the wheel
Fatigue fracture area in a shaft
caused by corroded inside area
 Introduction
Fatigue failures are widely studies
because it accounts for 90% of all service
failures due to mechanical causes.

Characteristics
•Fatigue failures occur when metal is mmd.sdsmt.edu

subjected to a repetitive or fluctuating Failure of crankshaft journal

stress and will fail at a stress much lower
than its tensile strength.
•Fatigue failures occur without any plastic
deformation (no warning).
•Fatigue surface appears as a smooth
region, showing beach mark or origin of
fatigue crack.
www.capcis.co.uk
Fatigue failure of a bolt
 Factors causing fatigue failure
Basic factors 1) A maximum tensile stress of sufficiently high value.
2) A large amount of variation or fluctuation in the
applied stress.
3) A sufficiently large number of cycles of the applied
stress.

Additional factors
• Stress concentration • Residual stress
• Corrosion • Combined stress
• Temperature
• Overload
• Metallurgical structure
 Stress cycles
Different types of fluctuating stress

max = - min
(a) Completely reversed cycle of (b) Repeated stress cycle
stress (sinusoidal)

Tensile stress +
Compressive stress -

(c ) Irregular or random stress cycle

 Stress cycles
Maximum stress, max
Nomenclature of stress parameter
in fatigue loading Minimum stress,  min
Stress range
+
 or  r =  max −  min Eq.1
a
 Alternating stress
  max −  min
a = = Eq.2
 max
2 2
m
Mean stress

 min  +  min
 m = max Eq.3
2
_ cycles
Stress ratio Amplitude ratio

R=
 min A=
 a = 1− R
Fatigue stress cycle  max  m 1+ R
Eq.5
Eq.4
 The S-N curve
•Engineering fatigue data is
normally represented by means of
S-N curve, a plot of stress S
against the number of cycle, N.
• Stress can be → a, max, min
• m , R or A should bementioned. Typical fatigue curves

•S-N curve is concerned chiefly with fatigue failure at high numbers

of cycles (N > 105 cycles) → high cycle fatigue (HCF).
• N < 104 or 105 cycles → low cycle fatigue (LCF).
• N increases with decreasing stress level.
•Fatigue limit or endurance limit is normally defined at 107
or 108 cycles. Below this limit, the material presumably can
endure an infinite number of cycle before failure.
• Nonferrous metal, i.e., aluminium, do not have fatigue limit
→ fatigue strength is defined at ~ 108 cycles.
 Basquin equation

•The S-N curve in the high-cycle region is sometimes described by

the Basquin equation

N ap = C Eq.6

Where a is the stress amplitude

p and C are empirical constants

LCF
HCF High cycle (low strain) fatigue

Stress level
LCF Low cycle (high strain) fatigue HCF

Log Nf
 Construction of S-N curve
•The construction of S-N curve normally requires ~ 8-12 specimens
by first testing at a high level of stress ~ 2/3 of the tensile strength of
the material.
• The test is then carried out at lower levels of stress until runout.
www.statisticalengineering.com

•The data obtained is normally

scattered at the same stress level
by using several specimens.
• This requires statistic approach
to define the fatigue limit.

S-N fatigue curve

 Statistical nature of fatigue
•Because the S-N fatigue data is
normally scattered, it should be
therefore represented on a
probability basis.
•Considerable number of
specimens are used to obtain
statistical parameters.
•At 1, 1% of specimens would be
expected to fail at N1 cycles.
Fatigue data on a probability basis
•50% of specimens would be Note: The S-N fatigue data is more
expected to fail at N2 cycles. scattered at lower stress levels. Each
specimen has its own fatigue limit.
•For engineering purposes, it is sufficiently accurate to
assume a logarithmic normal distribution of fatigue life in
the region of the probability of failure of P = 0.10 to P = 0.90.
 Effect of mean stress, stress range and stress
intensity (notch) on S-N fatigue curve
 m1 loc
 m4 > m3 > m2>  m1 Kt =
m2 app
R = 0.3
 m3
a  max a
m4 R=0
Kt = 1
R = -0.3
R = -1.0 K t = 1.5

Log Nf Log Nf Log Nf

Mean stress Stress range Stress intensity

Fatigue strength Fatigue strength Fatigue strength

Goodman diagram

Goodman diagram

•Goodman diagram shows the variation of the limiting range

of stress (max - min) on mean stress.
•As the mean stress becomes more tensile the allowable
range of stress is reduced.
• At tensile strength, u , the stress range is zero.
 Haig-Solderberg diagram

•In Haig-Solderberg diagram is

a plot of alternating stress a and
mean stress m.
•The Goodman relationship may
be expressed by

𝑎
𝑥
𝑚 Eq.7
= 𝑒 1 −
𝑢
Haig-Solderberg diagram
Where x = 1 for the Goodman line,
x = 2 for the Gerber parabola,
e = the fatigue limit for completely reversed loading.

• If the design is based on the yield strength o, (based on

Solderberg line), then the u is replaced by o in this equation.
Master diagram for establishing influence
of mean stress in fatigue

Ex: at max = 400 MPa, min = 0, a fatigue limit of the notched

specimen is less than 106 cycles.
For the unnotched specimen is below the fatigue limit.
 Example: A 4340 steel bar is subjected to a fluctuating axial load that
varies from a maximum of 330 kN tension to a minimum of 110 kN
compression. The mechanical properties of the steel are:
u = 1090 MPa, o = 1010 MPa, e = 510 MPa
Determine the bar diameter to give infinite fatigue life based on
a safety factor of 2.5.

Cylindrical cross section of the bar = A, the variation of stress will be

0.110
 max = 0.330 MPa,  =−
min
MPa
A A
 +  min = 0.330 / A + (−0.110 / A) = 0.110 MPa
 mean = max
2 2 A
 −  min = 0.330 / A − (−0.110 / A) = 0.220 MPa  m 
 a = max  =  ,  e = 510 = 204 MPa
2 2 A e 1−

 u 
a
2.5
0.220 / A 0.110 / A
Using the conservative Goodman line =1−
204 1090
and Eq.7.
A = 1179 mm2
4A
D= = 38.7 mm

Cyclic stress-strain curve
•Cyclic strain controlled fatigue occurs when the strain amplitude
is held constant during cycling.
•Found in thermal cycling where a component expands and
contracts in response to fluctuations in the operating temperature
or in reversed bending between fixed displacements.
•During the initial loading, the stress-strain curve
is O-A-B.
•Yielding begins on unloading in compression at a
lower stress C due to the Bauschinger effect.
•A hysteresis loop develops in reloading with its
dimensions of width,  and height .
•The total strain range  consists of the elastic
strain component plus the plastic strain
component.
Stress strain loop for  =  e +  p Eq.8
constant strain cycling
Cyclic hardening and cyclic softening

•Cyclic hardening
would lead to a
decreasing peak
strain with increasing
cycles. (n>0.15)
•Cyclic softening
would lead to a
continually increasing
strain range and early
fracture. (n<0.15)
 Comparison of monotonic and cyclic
stress-strain curves of cyclic hardened
materials
•The cycle stress-strain curve may
be described by a power curve as
follows
 = K ' (  p )n
'
Eq.9

Where n’ is the cyclic strain-hardening

exponent
K’ is the cyclic strength
coefficient
Since
strain amplitude   e  p
= +
2 2 2
'

  1   
1n
Monotonic and cyclic stress-strain curves
= +  ' 
2 2E 2  K 

For metals n’ varies between 0.10 -0.20.

Low cycle fatigue
•Low cycle fatigue (LCF) (high strain) is concerned about fatigue
failure at relatively high stress and low numbers of cycles to failure.
•Ex: in the nuclear pressure vessels, steam turbines and power
machinery. Usually concerned with cyclic strain rather than cyclic
stress.
•LCF data is normally present as a
•On the log scale, this relation
plot of strain range p against N.
can be best described by
 p
=  'f (2N ) c Eq.10
2
Where
p/2 = plastic strain amplitude
’f = fatigue ductility coefficient
2N = number of strain reversals
to failure.
Low-cycle fatigue curve ( p vs. N). c = fatigue ductility exponent
varies between -0.5 to -0.7.
 Example: For the cyclic stress-strain curve, B =75 MPa and
B = 0.000645. If f = 0.30 and E = 22x104 MPa.
Determine

(a) e and p

 2(75)
 e = = = 6.81810 −4

E 22 10 4
 p =  −  e = (2  0.000645) − 0.0006818 = 6.08210 −4

(b) The number of cycles to failure.

 p
From the Coffin-Manson relation =  'f (2N ) c
2
If c = -0.6 and ef ~ e’f
6.08210 −4
= 0.30(2N) −0.6
2
N = 49,000 cycles
 Strain-life equation
•For the high-cycle (low strain) fatigue (HCF) regime, where the
nominal strains are elastic, Basquin’s equation can be
reformulated to give

 a = e E =  'f (2N ) b
2 Eq.11

  e +  p
=
2 2 2
 =  f
'

(2N ) b +  'f (2N ) c

2 E
Where a = alternate stress amplitude
e/2 = elastic strain amplitude
E = Young’s modulus
 ’f = fatigue strength coefficient defined by the stress
intercept at 2N=1.
2N = number of load reversals to failure (N = number of
cycles to failure)
b = fatigue strength exponent, which varies between –
0.05 and -0.12 for most metals.
 Fatigue strain-life curve
Ductile materials
High cyclic strain condition

Strong materials

Low cyclic strain condition

The fatigue life value at which this

transition occurs is
1 (b−c)
  'f E 
2N t =  '  Eq.12
 
 f 

The fatigue strain-life curve

• tends toward the plastic curve at large total strain amplitudes
• tends toward the elastic curve at small total strain amplitudes.
Structural features of fatigue
The fatigue process can be divided into the following processes;

1) Crack initiation: the early development of fatigue damage

(can be removed by a suitable thermal anneal).
2) Slip band crack growth: the deepening of the initial crackon
plane of high shear stress (stage I crack growth)
3) Crack growth on planes of high tensile stress: growth of
well-defined crack in direction normal to maximum tensile stress
4) Ultimate ductile failure: occurs when the crack reaches
sufficient length so that the remaining cross section cannot
support the applied load.
 Initiation of fatigue crack and slip
band crack growth (stage I)

• Fatigue cracks are normally initiated at a free surface. Slip lines

are formed during the first few thousand cycles of stress.
•Back and forth fine slip movements of fatigue could build up notches
or ridges at the surface. → act as stress raiser → initiate crack.

Extrusion
Intrusion
•In stage I, the fatigue crack tends to
propagate initially along slip planes
(extrusion and intrusion of persistent
PSB PSB
slip bands) and later take the direction
normal to the maximum tensile stress
(stage II).
•The crack propagation rate in stage I
Model for fatigue initiation by is generally very low on the order of
extrusions and intrusions
caused by cyclic slip during
nm/cycles → giving featureless
fatigue loading. surface.
 Stable crack growth (stage II)
Crack closed

Crack opening

Crack at
maximum
load

Crack closing
Fatigue striations
Crack closed

•The fracture surface of stage II crack

Plastic blunting model of fatigue striation
propagation frequently shows a pattern of
ripples or fatigue striations. •Crack tip blunting occurs
during tensile load at 45o and
•Each striation is produced by a single crack grows longer by plastic
stress cycle and represents the shearing.
successive position of an advancing crack •Compression load reverses the
front normal to the greatest tensile stress. slip direction in the end zones →
crushing the crack surface to form
a resharpened crack tip.
 Fatigue crack propagation
Stage I Non-propagating fatigue crack (~0.25nm/cycle)
Fatigue crack
Stage II Stable fatigue crack propagation- widely study
propagation
Stage III Unstable fatigue crack propagation → failure

•For design against fatigue failure, fracture

mechanics is utilised to monitor the fatigue crack
growth rate in the stage II Paris regime.

= A(K )m
da a Monolithic
Eq.13 ,h
dN t
g
n
•Where the fatigue crack growth rate e
lk
c
a
da/dN varies with stress intensity r
factor range K, which is a function of
C

stress range  and crack length a. ao

K = K max − K min Number of cycles to failure

Eq.14
K =  max a −  min a
FCG curve
A log scale plot gives Paris exponent m as the slope
 Fatigue crack propagation
Stage I Non-propagating fatigue crack (~0.25nm/cycle)
Fatigue crack
Stage II Stable fatigue crack propagation- widely study
propagation
Stage III Unstable fatigue crack propagation  failure

Non continuum Continuum behaviour Static mode of

behaviour (striations) or transition behaviour

Fatigue crack growth rate da/dN (log scale)

from non continuum (cleavage,
Large influence of behaviour with intergranular and
• microstructure Small to large influence of dimples)
• mean stress • microstructure, Large influence of
• environment depending on the • microstructure
material • mean stress
• thickness
Large influence of
• certain combination of Little influence of
environment, mean stress • environment
and frequency
Non propagating

Unstable crack growth

fatigue cracks

1
= a (K )m
da
dN
for linear portion

Kth

Stage I Stage II Stage III

Fatigue crack
Stress intensity factor range, K (log scale) growth behaviour
 Fatigue crack growth propagation in
stage II regime
Stage II fatigue crack growth propagation has been widely investigated
in order to determine the fatigue crack growth life from the
representing stable fatigue crack growth rate.
m
= A(K )
da m 1e-1

dN da/dN R = Pmin/Pmax = 0.1

Frequency = 0.25 Hz
1e-2
The fatigue crack growth TIMET bar III
m = 2.02

life Nf (stage II) canbe BuRTi EBW01

da/dN, mm/cycle
m = 9.05
1e-3
determined by BuRTi EBW02
m = 2.35
Nf Ti679-BuRTi-TI679 TIG01

N f =  dN
m = 6.84
1e-4
Ti679-BuRTi-Ti679 TIG02
m = 10.18
0

a −(f m / 2)+1 − a i−(m / 2)+1

1e-5

Nf = Eq.15
(−(m / 2) + 1)A  m
r
m/2
 m
1e-6
1 10 100
where m ≠2 K, MPa.m1/2
 is the crack geometry factor Fatigue crack growth in base metal and welded materials
Example: A mild steel plate is subjected to constant amplitude uniaxial
fatigue loads to produce stresses varying from max = 180 MPa to
min = -40 MPa. The static properties of the steel are o = 500 MPa,
u = 600 MPa, E = 207 MPa, and K c = 100 MPa.m1/2. If the plate contains
an initial through thickness edge crack of 0.5 mm, how many fatigue
cycles will be required to break the plate?

For through thickness edge crack,  = 1.12, and for ferritic-pearlitic

steels, A = 6.9 x 10-12 MPam1/2 and m = 3.0.
r = (180-0), since compressive stress are ignored, and neglect the
influence of mean stress on the crack growth.

1  Kc 
2
1  100 
2

ai = 0.0005 m, a f =   = = 0.078 m
   max   180 1.12 
From Eq.15
a −(f m / 2)+1 − a i−(m / 2)+1
Nf =
(−(m / 2) + 1)A rm  m / 2 m
(0.078)−(3 / 2)+1 −(0.0005)−(3 / 2)+1
Nf= = 261,000 cycles
(−(3 / 2) + 1)(6.9 10 −12
)(180) ( )
3 3/2 3
(1.12)
 P/2 P/2

S-N curve fracture surfaces

•S-N curve test involves crack initiation and crack 20 mm

propagation to failure. → overall fatigue life. 40 mm

•Fatigue testing normally uses plain specimens of

Crack
different specimen surface conditions, i.e., polished, length a
ground, machined, etc. under tension or bending.
•Crack initiation might be due to inclusions, second Corner
crack
phases, porosity, defects.

Porosity
Fractured
carbides

Fatigue crack initiation Fatigue crack

from porosity initiation from
inclusion/particle.
 Crack growth direction
FCG fracture surfaces
•Fatigue crack growth (FCG) test involves only crack
propagation stage but excludes crack initiation stage.
•Specimen has an initial short or small crack and this
Fatigue striation
crack will propagate under cyclic loading.

(a) Crack growth direction

Fatigue crack growth rate da/dN (log scale)

Brittle facet (slip) +
fatigue striation
Brittle facets

Stage I (b)

Non propagating

Unstable crack growth

Fatigue

fatigue cracks
m
striation 1
= a (K )m
da
dN
(c) for linear portion

Stage II Kth

Stage I Stage II Stage III

Brittle facets
(cleavage) + Stress intensity factor range, K (log scale)
microvoids
tyStage III
of Technology
Factors influencing fatigue properties

• Stress concentration
• Size effect
• Surface effects
• Combined stresses
• Cumulative fatigue damage and
sequence effects
• Metallurgical variables
• Corrosion
• Temperature
 Effect of stress concentration on fatigue
Stress raiser Fatigue strength

Should avoid stress raisers from machining and fabrication processes.

•The effect of stress raiser or notch on fatigue strength can be
determined by comparing the S-N curve of notched and
unnotched specimens. (based on the net section of specimen).
• The notch sensitivity factor q in fatigue is determined from

K f −1
q= Eq.16
K t −1
Where
K t is theoretical stress-concentration
factor, depending on elasticity of
crack tip
K f is fatigue notch factor, ratio of fatigue
strength of notched and unnotched
specimens.
S-N curve of notched and unnotched specimens
 Size effect on fatigue
Due to
Fatigue properties Experimental scale ≠ Industrial scale size effect

• Fatigue property is better in the small sized specimens.

• Problem: the machine cannot accommodate large specimens.
• Larger specimens → increases surface area subjected to cyclic load,
higher possibility to find defects on surface.
→ decrease the stress gradient and increases the
volume of material which is highly stressed.

Explain: It is usually impossible to duplicate the same stress

concentration and stress gradient in a small-sized laboratory
specimens.

Solution: Use statistic approaches, i.e., Weibull statistics.

Surface effects on fatigue

• Fatigue properties are very sensitive to surface conditions,

• Fatigue initiation normally starts at the surface since the
maximum stress is at the surface.

The factors which affect the surface of a fatigue specimen can

be roughly divided into three categories;

• Surface roughness
• Changes in surface properties
• Surface residual stress
 Surface roughness
• Different surface finishes produced by different machining processes
can appreciably affect fatigue performance.
•Polished surface (very fine scratches), normally known as ‘par bar’
which is used in laboratory, gives the best fatigue strength.

Reduction factor for fatigue

limit of steel due to various
surface treatments
 Changes in surface properties
•Changes in surface properties due to Change in fatigue
surface treatments strength/properties.
Treatments which reduces fatigue performance

Decarburization Ex: decarburization of surface of heat-treated steels.

Soft coating Ex: Soft aluminium coating on an age-hardenable Al alloy.
Electroplating Might reduces fatigue strength due to changes in
residual stress, adhesion, porosity, hardness.
Treatments which improves fatigue performance
Carburizing
• Forming harder and stronger surface →
Nitriding introducing compressive residual stress.
Flame hardening •The strengthening effect depends on the
diameter of the part and the depth of the
Induction hardening surface hardening.
 Surface residual stress

•Residual stresses arise when plastic

deformation is not uniform throughout the
entire cross section of the part being deformed.
Loading Unloading
Part undergone Compressive
plastically deformed in residual stress
tension
Part undergone
Tensile residual
plastically deformed in
stress
compression
Superposition of applied andresidual
stresses
(a) Shows the elastic stress distribution in a beam with no residual
stress.
(b) Typical residual stress distribution produced by shot peening where
the high compressive stress is balanced by the tensile stress
underneath.
(c) The stress distribution due to the algebraic summation of the
external bending stress and the residual stress.
Commercial methods introducing
favourable compressive stress

• Surface rolling
- Compressive stress is introduced in between
the rollers during sheet rolling.

• Shot peening Sheet rolling

- Projecting fine steel or cast-iron shot
against the surface at high velocity.

• Polishing
- Reducing surface scratches

• Thermal stress
- Quenching or surface treatments introduce volume change
→ giving compressive stress.
 Effect of combined stresses on fatigue
Few data has been made on fatigue test with different
combinations of types of stresses.
•Ductile metals under combined bending and torsion fatigue
follow a distortion-energy (von Mises).
•Brittle materials follows the maximum principal stress theory
(Tresca).

Sines has proposed expressions for

Low strain (

a1 −  a2 )2 + ( a2 −  a3 )2 + ( a3 −  a1 )
2

1/ 2
+ C2 (m1 + m 2 + m3) 
2 a
Kf
Eq.17

High strain

 q = oct =
2
1
3

(1 −  2 )2 + ( 2 −  3 )2 + ( 3 − 1) 2 1/ 2
Eq.18

Note: Effects of residual stress and triaxial stress are included.

 Cumulative fatigue damage and
sequence effects on fatigue
Practically, levels of stress are not held constant as in S-N tests,
but can vary below or above the designed stress level.

•Overstressing : The initial applied stress level

is higher than the fatigue limit for a short period
of time beyond failure, then cyclic stressing
below the fatigue limit. This overstressing
 Overstressing
reduces the fatigue limit.
Cycle
•Understressing : The initial applied stress

Stress level
level is lower than the fatigue limit for a period
of time, then cyclic stressing above the fatigue
Fatigue limit
limit. This understressing increases the 
fatigue limit (might be due to strain hardening Understressing Cycle
on the surface.

Log Nf
 Cumulative damage rule
The percentage of fatigue life consumed by operation at one operating
stress level depends on the magnitude of subsequent stress levels
 the cumulative rule called Miner’s rule.

n1 n 2 + nk j=k n
+ = 1 or 
j
+ ... Eq.19
N1 N 2 Nk j =1 N j

Where n1, n2,..nk = the number of cycles of operation at

specific overstress levels.
N1, N2,..Nk = the life (in cycles) at this same
overstress level.

Note: for notched specimen, the fatigue strength is reduced much

more than it would be predicted from the Miner’s linear damage rule.
This is due to the effect of residual stress produced at the notch by
overload stresses in the plastic region.
Effects of metallurgical variables
on fatigue
•Fatigue property is normally greatly improved by changing the
designs or, reducing stress concentration, introducing compressive
stress on the surface.
•Few attempts have paid on improving metallurgical structure to
improve fatigue properties but it is still important.

• Fatigue property is frequently correlated with tensile properties.

Tensile strength
Fatigue strength
Fatigue ratio =
Tensile strength Fatigue strength

Note: for smooth and polished specimen.

Ex: fatigue ratio ~ 0.5 for cast and wrought steels,
~ 0.35 for non-ferrous.
 Fatigue strength improvement by
controlling metallurgical variables

By increasing tensile strength

• Grain boundary strengthening

By
strengthening • Fibre strengthening
mechanisms • Second phase strengthening
• Cold working

Note: not for all cases and

Stress level

not proportionally.

•Grain size has its greatest

effect on fatigue life in the
Fatigue limit
low-stress, high cycle regime.

Log Nf
 Fatigue strength improvement by
controlling metallurgical variables
By controlling microstructure
•Promote homogeneous slip /plastic deformation through
thermomechanical processing  reduces residual stress/ stress
concentration.
•Heat treatments to give hardened surface but should avoid stress
concentration.
• Avoid inclusions → stress concentration → fatigue strength
•Interstitial atoms increase yield strength , if plus strain aging → fatigue
strength
Strain aging from
interstitials

Solid solution

Pure metal

Effect of interstitial atoms

 Effect of corrosion on fatigue
•Fatigue corrosion occurs when material is subjected to cyclic
stress in a corrosive condition.

• Corrosive attack produces pitting

on metal surface. Pits act as notches
→ fatigue strength .
• Chemical attack greatly accelerates
the rate of fatigue crack propagation.

Corrosion fatigue of brass

Role of a corrosive
environment on fatigue
crack propagation
 Corrosion fatigue test
Corrosion fatigue test can be carried out similar to fatigue test but in
a controlled corrosive environment.
•Since corrosion process is a time-dependent phenomenon, the
higher the testing speed (frequency), the smaller the damage
due to corrosion.
•The action of the cyclic stress causes localised breakdown of
the surface oxide film → corrosion pits.
Corrosion fatigue test www.mtec.or.th
www.corrosion-doctors.org

S-N curve in various condition

 Minimization of corrosion fatigue

•Select corrosion-resistant materials for the desired application.

Ex: stainless steel, bronze, would give better service than heat-
treated steel.
•Protection of the metal from contact with the corrosive
environment by protective metallic or non-metallic coatings.
•Introducing compressive residual stresses bynitriding, shot
peening → eliminating surface defects.
 Effect of temperature on fatigue
Temperature
If testing temp < RT → low temperature fatigue.
(Increasing TS)
If testing temp > RT → high temperature fatigue.
Fatigue strength

•In high temperature fatigue, there is a transition from fatigue failure

to creep failure as the temperature increases (creep dominates at
high temperatures).
•Coarse grained metal has higher fatigue strength – where creep
dominates.
• Fine grained metal has higher fatigue strength at low temperatures.
Temp

Fatigue failure Creep failure

Transcrystalline Intercrystalline
fatigue failure creep failure
 Thermal fatigue
Thermal fatigue occurs when metal is subjected
to high and low temperature, producing
fluctuating cyclic thermal stress.

Volume change Thermal fatigue failure

Thermal cycle
•Normally occurs in high
Cold Hot
temperature equipment.
•Low thermal conductivity and high
thermal expansion properties are
critical.
• The thermal stress developed by a temperature change T is

 = ET Eq.19

Where  is linear thermal coefficient of expansion

E is elastic modulus
If failure occurs by one application of thermal stress,
the condition is called thermal shock.
 Stress level
Design for fatigue
Fatigue limit
There are several distinct philosophies Allowable stress
concerning for design for fatigue Log N f

1) Infinite-life design: Keeping the stress at some fraction of the

fatigue limit of the material.
2) Safe-life design: Based on the assumption that the material has
flaws and has finite life. Safety factor is used to compensate for
environmental effects, varieties in material
production/manufacturing.
3) Fail-safe design: The fatigue cracks will be detected and repaired
before it actually causes failure. For aircraft industry.
4) Damage tolerant design: Use fracture mechanics to determine
whether the existing crack will grow large enough to cause failure.
References

• Dieter, G.E., Mechanical metallurgy, 1988, SI metric edition,

McGraw-Hill, ISBN 0-07-100406-8.
• Suresh, S., Fatigue of materials, 1998, 2nd edition,
Cambridge university press, ISBN 0-521-57847-7.
•Lecture note, MRes 2000, School of Metallurgy and Materials,
Birmingham University, UK