You are on page 1of 20

FATIGUE OF METALS

ABSTRACT:

In most of the laboratory experiments we must have studied the stress-


strain diagram which is obtained by loading a component in UTM
(Universal testing machine) and subjecting it to both tension and
compression at higher and lower strain rates for studying the structural
and machine components. These conditions are very static in nature and
do not approximate the actual loading conditions which we encounter in
actual field. In actual dynamic conditions stresses vary with time and also
they fluctuate between different levels as in the case of bridge deck with
variation in traffic with time, stresses also vary accordingly. Sometimes
they are reverse in nature for components like automobile differential
shaft. When these stresses are repeated over a large number of times, the
material fails at stresses below the ultimate strength of material. Hence the
failure occurred due to this type of loading is termed as “Fatigue failure”,
these are sudden in nature with no sign s of warning. As a design engineers
we are expected to predict these types of failures in static as well as
dynamic environment to tackle various fatigue problems that we encounter
in today’s challenging environment. Here we are discussed an integrated
approach to deal with fatigue problems. The whole fatigue domain is
divided into six zones that include different fatigue regimes. The
propagating crack length is considered as one of the important parameter
to evaluate safe fatigue life. Life, including the use of an ‘‘equivalent crack
Propagation rate”, which averages the intense variations of CPR in the
vicinity of grain boundaries. For each fatigue zone a unique prediction
relation is presented and discussed in detail. Surface fatigue strength and
the ways of increasing fatigue life of the component also discussed.
1)Key words:
The fatigue domain
Equivalent crack propagating length
Reversed stresses
Ultimate tensile strength
Endurance limit
S-N curve
Stress amplitude
2) Introduction:
The main goal of design engineers in fatigue study is to reduce fatigue
failure occurrence in various structural and machines components under
static and dynamic loading conditions. Many of the static failures gives
visible advance warnings in nature, but fatigue failures are not. It occurs all
of sudden and dangerous. Fatigue is complicated phenomenon and fatigue
failure has an appearance similar to a brittle fracture, as the fracture. The
fracture features of a fatigue failure, however, are quite different from a
static brittle fracture arising from three stages of development. In first
stage there is initiation of one or more micro cracks due to cyclic plastic
deformation followed by crystallographic propagation extending from two
to five grains about the origin. Stage I cracks are not normally discernible to
the naked eye. In stage two, cracks progresses from micro cracks to macro
cracks. During cyclic loading, these cracked surfaces open and close, the
stage three occurs during the final stress cycle when the remaining
material cannot support the loads, resulting in a sudden fracture. The aim
of a design engineer is to introduce a relatively simple and effective fatigue
design tool that captures or predicts the fatigue failures of various
structures and machine components. The current study divides whole
fatigue domain into distinct zones, each with a specific fatigue regime, and
particular calculation procedures are discussed for each zone.

Courtesy of Star Bulletin/Dennis Oda/© AP/Wide World Photos.

The above image describes the importance of fatigue study. The above
Photograph is of a Boeing 737-200 commercial aircraft (Aloha Airlines
flight 243) that experienced an explosive Decompression and structural
failure on April 28, 1988. An investigation of the accident concluded that
the cause was metal fatigue aggravated by crevice corrosion in as much as
the plane operated in a coastal (humid and salty) environment. Stress
cycling of the fuselage resulted from compression and decompression of
the cabin chamber during short hop flights. A properly executed
maintenance program by the airline would have detected the fatigue
damage and prevented this accident.
As I said earlier, the fatigue failures are sudden in nature without any signs
of failure, so in order to prevent that we must be prepared with proper
counter measures. It is nothing more than analyzing the failure and
developing fault detecting mechanism.
3) Fatigue domain:
In the classical approach to fatigue problems we deal with various types of
alternating stresses and numbers of cycles to failure. Parameters like
ultimate tensile strength – Su, endurance limit – Se (where it exists), high
and low cycle fatigue, fatigue damage and some more are the popular
factors to deal with. In the more recent fracture mechanics approach one
deal with cracks, crack length “a” and stresses at the crack tip, remote
stresses, stress intensity factors – K and stress intensity factor limits like
fracture toughness – K1C. Definition of the whole fatigue domain should
include and refer to all, or to most of the above mentioned parameters. Not
all these parameters are consistent, therefore a diagram of the fatigue
domain will have to include the main ones directly, and some more as
depending parameters.
In S-N diagram nominal stress required to cause a fatigue failure in
some number of cycles. This test result in data presented as a plot of stress
(S) against the number of cycles to failure (N), which is known as an S-N
curve. A log scale is almost always used for N. typical S-N curve for
aluminium and 1045 steel shown in below figure.
In general loading conditions the stress/load oscillation may be sinusoidal,
but the mean stress/load may be such that the stress state during the
entire cycle is tensile. Needless to say, for a given stress amplitude this type
of loading is more severe (as maximum stress max is min+ r). Various
parameters are defined in the equations below.
 r   max   min

r  max   min  max   min  min


a   m  Stress ratio  R 
2 2 2  max

 a 1 R  a 1 R
Amplitude ratio  A   Amplitude ratio  A  
 m 1 R  m 1 R
The first parameter we choose to perform fatigue test is the remote stress
amplitude  a . It is used in all classical fatigue tests, like in tension
compression fatigue of a round specimen. Obviously a crack is developed
and propagates during such tests, but the results are depicted on an S–N
curve without recalculating the reduction of the net tension area, namely
with the remote stress amplitude. The cycle by cycle crack growing size, in
a fatigue test in a classical smooth specimen, that has been performed for
more than 150 years, has usually not been measured nor been reported. So
by taking the alternating remote stress can serve as one main parameter of
the sought simple diagram, as the ordinate.

The other parameter will be the crack length, as it can almost always be
measured or estimated and used for calculations, and depicted as the
abscissa.

The whole fatigue domain is shown in the above figure with the whole
fatigue domain can be depicted on a diagram with these two parameters
and is shown in figure above. In the diagram the maximal values are
depicted as limits of the fatigue domain. But ultimate tensile strength (Su)
for the relevant material is constant, and gross yielding will be reached
when the real engineering stress and not the remote stress will reach Su.
This can happen when a ductile material is used with high fracture
toughness (K1C), and the gross yielding will happen before reaching
fracture toughness (K1C). So the ultimate strength (Su) line, which on the
given figure scales is a function of the crack length, should not stay
horizontal and it has to be bended downwards and somewhat reduce the
fatigue presented domain. In the above figure the ultimate tensile strength
(Su) line is depicted to the right of the fracture toughness (K1C) line,
therefore in this specimen shape, material and type of loading, most
failures will happen by critical crack propagation. In some ductile materials
the K1C line will be to the right of the ultimate strength (Su) line, and
failure will take place in gross yielding. This issue will be discussed in due
course. The lower left line is the Kitagawa and Takahashi line that defines
the upper limit, under which cracks don’t propagate, therefore the fatigue
domain is above it and is depicted by the crosshatched area. The fatigue
domain is the whole possible combination of parameters where fatigue of
materials may occur, and has to be predicted and taken into consideration
when designing parts and structures. If a material is stressed above the
ultimate tensile strength, the materials fails due to gross yielding and no
fatigue concept applies here.
In engineering problems, by applying S-N diagram approach or by
any of the classical approach stress profile is applied and life is predicted
from it and if possible we measure the existing crack lengths. By
considering ultimate strength, fracture toughness and propagating crack
length, we get to know that a fatigue problem exists or not. It will not
indicate how to tackle fatigue problem. As we could from the previous
graph, crack propagation plays crucial role in determining the fatigue life of
the object. To enable the next step in handling fatigue problem issue, we
require a clear calculation method for crack propagation in the loading
cycle. The fatigue domain has to be divided into separate fatigue regimes
that describe distinctive behaviour of the part under fatigue loading. Each
fatigue regime will include a prediction method of crack propagation for
every one loading cycle. When the crack propagation will be calculated, it
can be added to the previous crack length and so obtain the new crack
length for the next loading cycle. Summation of the crack length will
eventually generate a new length that may cause the crossing from one
regime to another, and require using the prediction method for the new
regime. This will proceed until eventually the combination of the current
crack length and the alternating stress amplitude will bring the part to
cross the fatigue domain boundary and then the part will fail. A division of
the fatigue domain into six fatigue zones enable us to concentrate on each
regime. This method was first introduced by Weiss in 1992, and its early
version is depicted in below figure. But here the stress amplitude  a was
defined as the true, corrected engineering stress.
The above figure includes three horizontal constant stress amplitude lines,
the endurance limit (Se), the yield stress (SY) the ultimate tensile stress
(Su) , and two constant stress intensity factor range lines, the effective
threshold range ktheff And the fracture toughness (K 1c ) lines. These lines

divide the fatigue domain into 6 zones, including zone 1 which is out of the
domain. The attributes of each zone will be described in detail. One can see
that for each combination of any single stress amplitude and for a chosen
crack length “ai” the diagram indicates a certain fatigue zone. If we had an
equation to calculate the crack propagation for each zone will be defined,
then the crack extension in the current loading cycle “da” can be calculated,
and the subsequent new crack length (ai+1 ) will be the addition of the crack
length at the start of the “i-th” cycle “ai ” and the extension, namely “ai+1 =
ai+da” In this way the crack extension for additional loading cycles can be
calculated and summated and the crack extension can be displayed on the
diagram. As mentioned, the stress amplitude in the diagram  a is the
engineering stress, namely stress that is calculated by actuation of the
external load “P” on the net residual area of the specimen. But using the
engineering stress amplitudes as a calculation parameter was inconvenient
for designers and therefore the diagram needs an update actual stress.

4)The enhanced fatigue diagram:

The fatigue tests that have been performed on specimens were portrayed
on remote stress amplitudes as parameter. The ‘‘engineering stress” that
considers the reduction of the specimen’s net residual area as the crack
propagates, and which is the ‘‘true” stress was only rarely used. Therefore
an enhanced fatigue diagram will also have the remote stress amplitude as
ordinate, instead of the engineering stress amplitude.
The above is the enhanced fatigue diagram. The abscissa is measured in log
of the crack length, and in the fatigue it starts from one micron. Micro
cracks are produced in components while manufacturing. It can propagate
or stay dormant as a non propagating crack. The diagram is divided into six
regimes and they are conceptually explained in the due course.
The material constants “S’Y” “S’e” and “S’U” are illustrated on the above figure
are differently than they were in the old version of the fatigue diagram. In
the old version, due to the use of the engineering stresses, they were
constant horizontal lines. In the enhanced diagram, where the stress
amplitudes are the remote stresses, the material stress constant lines,
which are absolute values but a function of the crack length, have to be
calculated and shown differently and the diagram has to be amended
accordingly. The material constants have to be
Depicted based on the real acting stresses, namely have to be calculated on
the net residual specimen area. Therefore on the diagram they start as
horizontal lines. They are depicted as semi horizontal when the crack
length is small, but when the crack size increases, these lines change due to
the smaller residual net area and are turned downward as depicted with
the blue lines.
When the propagating crack becomes large, the material constants
“S’Y” “S’e” and “S’U” lines decrease, depending on the geometry and the
loading. The “S’U” falls either to the left or to the right of the “K 1C “and this
fact determines whether the failure will be by gross yielding or by critical
crack propagation.

Zone: 1
This zone is below the Kitagawa line. Kitagawa line is the maximum limit
under which cracks do not propagate. Hence it is termed as the non
propagating cracks zone. It is defined as a reference line to identify the
crack propagating and non propagating zones. No fatigue events happen in
zone 1.
Zone: 2
This region is bounded below the fatigue limit, above the stress intensity
threshold range line and below the fracture toughness line. This is the
Linear Elastic Fracture Mechanics (LEFM) zone. Numerous LEFM relations
and studies that were introduced to fit relatively well to experimental
values in this zone. Crack propagation in zone 2 is represented as (da/dn)2.
Zone: 3
This zone is the High Cycle Fatigue (HCF) zone within the elastic regime.
This region is bounded above the fatigue limit, below the yield stress and
below the stress intensity factor range threshold. Most of the fatigue tests
are performed in this region. But the great majority of the stress cycles in
each test took place in zone 3. The common view and a used expression
used in fatigue tests is
‘‘more than 90% of the fatigue life took place in crack initiation”, that was
considered to reach the threshold stress intensity factor range “DKth”
namely the right boundary of zone 3. There is no generally accepted
relation for step by step crack extension calculation exists for zone3, and
such crack extension from an initial micro-crack that emanates from the
surface finish, till failure of the specimen has not been found in the
literature. Crack propagation in zone 3 is represented as (da/dn)3.
Zone 4:
This zone is unique from the all other three zones. Two fatigue regimes are
active here in parallel, namely High cycle fatigue (HCF) and Linear elastic
fracture mechanics (LEFM). Zone 4 is above the fatigue limit and below the
Yield stress, as same like zone 3 does, and above the stress intensity factor
range threshold and below the fracture toughness, as is zone 2. Therefore
the two fatigue regimes, HCF and LEFM, both contribute to crack
propagation. For simplicity, and as a first approximation, it is assumed that
the two contributions are generated independently. Crack propagation in
zone 4 will be assigned as (da / dn)4 and will be calculated as a linear
combination of the two expressions,
It can be represented as (da/dn)4 = (da/dn)2+(da/dn)3 .
Zone 5:
This zone is within the plastic regime and is the Low Cycle Fatigue (LCF)
zone. It is also above the yield stress, below the ultimate tensile strength
and below the stress intensity factor threshold. This zone a specimen
cannot take too many cycles of fatigue loading to failure will occur in zone
5. Crack propagation will be named as (da/dn)5. It is very rare to design
structures in Zone 5, therefore for a designer it can practically be ignored
and as a first approximation.
Zone 6:
This zone is also in the Low cycle fatigue ( LCF) like zone 5, but also above
the stress intensity factor (SIF) threshold and below the fracture
toughness. This area is referred to as Elasto-Plastic Fracture Mechanics
(EPFM). Crack propagation will be named as (da/dn)6 so as assumed
before for zone 4, zone 6 is also a combination of two regimes, Low cycle
fatigue(LCF) and Elasto-plastic fracture mechanics( EPFM) exists. So it can
be assumed as a linear combination of two fatigue regimes can be active
here, they are represented as
(da/dn)6 = (da/dn)5 + (da/dn)2 .
Like zone 5, It is extremely rare for a designer to design structures in zone
6, therefore for a designer it can be ignored.
5) Calculation of the crack propagating by zones:

The current study is aimed at design engineers, the only parameter for any
damage due to fatigue will be the crack length, even in regimes where just
tiny micro-cracks do exist and their apparent size must be estimated. The
combination of stress function and cracks parameters will be the basis for
calculations of crack propagation in all the 6 fatigue zones. For each zone a
way of calculation will be proposed.
5.1. The “equivalent crack propagation” concept:
We know cracks present in many of the manufacturing processes. Small
cracks that are smaller than the grain sizes of the material, the crack
formation is different within the grains and much slower when it has to
cross the grain boundaries. The crack growth rate in this region is varying,
and it increases and decreases as shown in the figure below.

It is quite complicated, or even impossible to measure the crack


propagation rate (CPR) in the initiation phase of a real specimen. But for
design purposes, the exact CPR values for each crack length are not needed.
For design purposes the exact knowledge of the crack propagation
rate (CPR) in every instant of the short cracks formation is not really
important, as these crack sizes are far from the critical failure values of the
cracks at ultimate strength(Su) and fracture toughness (K1c). As an
alternative we will define an ‘‘Equivalent crack propagation rate (CPR)”. As
we can see from the above figure, it is concluded that equivalent crack
propagation rate monotonically increases and connects the crack
propagating line at other end. It simply takes care of the initial phase of the
crack propagation. For design purpose equivalent crack propagation rate is
fully accepted to evaluate fatigue life of a component.
5.2: Zone 2
The zone 2 is named as linear elastic fracture mechanics zone. The crack
propagation in this zone will be calculated by an expression.
(da / dn)2  C2a(keff / Klc )m ((1  (kth / keff ) z ) / (1  (keff / K1c ) z ) 
1

Where
C2 =dimensionless material parameter

a =crack length
keff =Effective stress intensity factor

kth = Stress intensity threshold.

K1c =fracture toughness

m =2 (Always)
z =8 (Always)
The exponent m is always selected as m = 2, so that an analytical derivation
will be possible in the future. The expressions in the other parentheses are
there to imitate the boundary conditions, namely to zero the propagation
rate when the ( keff ) gets close to the stress intensity range threshold

( kth ) and to accelerate the propagation when keff gets close to the fracture

toughness ( K1c ). The exponent z manages the leaning toward the boundary
condition lines, and z = 8 is just a worthy estimate, as the residual number
of cycles till failure in this stage is small. The above equation holds good to
predict the fatigue crack propagation within zone 2.
5.3: Zone 3:
As we can see from fatigue diagram, zone 3 is high cycle fatigue zone. We
know that micro cracks exists in components due to manufacturing
processes more than 90% of the fatigue life took place in the crack
initiation. So crack propagation even for micro cracks in contemplated. The
average crack propagation rate in zone-3 is represented by below relation.

(da / dn)3  C3a( a  Se / Su ) m (1/1  ( a / Su ) z ) 


2

Where
 a = Stress amplitude

C3 =Dimension less material parameter

a = Crack length
 a = Stress amplitude

S e = Endurance limit

S u = Ultimate strength

m = 2 (Always)
z =8
The expression in the parentheses is there to imitate the boundary
condition when Stress amplitude (  a ) gets close to Su. Here also the
exponent z manages the leaning toward the boundary condition line, and z
= 8 is just a realistic estimate.
5.4: Zone-4
Zone 4 is more complicated because two fatigue regimes active here
namely High cycle fatigue and linear elastic fracture mode. Hence the crack
propagation will be calculated as a linear combination of the previous two
expressions. It is just a linear summation.

(da / dn)4  (da / dn)2  (da / dn)3 


3

5.5: Zone 5&6:


The zone 5 &6 are in plastic region, . It is very rare to design structures in
Zone 5 therefore for a designer it can practically be ignored. To calculate
fatigue crack propagation it should be based on plasticity including strain
rate effects.
The above introduced equations can be used to calculate the crack
extension, cycle by cycle, till failure. Obviously the straight forward
calculations will be for the simple loading cases, like tension– compression,
bending, one dimension crack propagation etc, that for design engineers
will be the vast majority of cases. When the crack propagates due to cyclic
loading, it will pass from one zone to another, like from zone 3 to zone 4, or
from zone 2 to zone 4 when the crack size crosses the boundary values. But
using the proposed equations will enable to calculate the case where each
loading cycle has different stress amplitude, and in such case each loading
cycle may take place in a different fatigue zone.
The application of the fatigue diagram by the designer will be as
follows: when confronting a fatigue problem in a designed element the
designer calculates the operating stress amplitude, estimates the initial
crack length (that is based on the surface finish of the element) and inserts
them into the fatigue diagram. In the diagram the zone and therefore the
fatigue regime are defined. Calculation of the crack size extension is based
on the relations that were introduced above. If the zone is 1, even if cracks
exist, the element is in the safe zone and cracks will not propagate.
If the zone is 2and 3 then use the equations stated above for crack
propagation estimate.

6) Crack propagation calculation:

The above graph shows crack propagation for varying stress values. Crack
propagation can be calculated, step by step; by use of the equations we
have it for each zone. The loading modes are clear from the definition of the
test or the design problem. The metal properties have to be known by the
material selected. The size of the initial crack that will later propagate has
to be evaluated based on the surface finish of the material. Once we have all
these values, step by step crack propagation can be calculated. The
individual loading cycles can be quite different one from another, or can be
in sets of cycles with the same loading function, namely block loading.
The initial micro-crack length ai is estimated here to be of the size of
about 2 microns. The initial stress amplitude is  j as denoted by the red

point 1. The specimen is loaded with a block of loading cycles with stress
amplitude. Crack propagates in zone 3 from point 1 to point 2 to the length
Ai+1.The crack propagating rate (CPR) can be calculated by the equation
2,for zone 3. Now the stress amplitude is elevated to  j 1 and another block

of cycles is actuated. The crack propagates from point 3 to point 4 in the


low cycle plastic regime zone 5. The propagation was calculated, as a first
approximation, by using Equation 2. But a more proper relation has to be
developed for this regime. Now the stress amplitude is lowered to point 5
and the propagation proceeds till point 6. When the crack propagates from
point 7 to 8, the process crosses the threshold line and therefore the
calculation have to start with Equation 2 in zone 3 and switch to Equation 3
at the transition point.
The next stress amplitude is lower than the endurance limit, as we
can see in point 9. Therefore the crack arrests and does not propagate in
this block of loading. The phenomenon of crack arresting is clearly shown
on the diagram, as the loading takes place in zone 1 – the non-propagating
cracks zone. Then again, the propagation proceeds in blocks 10–11 in
Zone 3 and 12–13 in zone 4. The following block 14–15 is in the linear
elastic fracture mode (LEFM) regime zone 2 and Equation 1is used to
calculate crack propagating rate. The stress amplitude can now be elevated
to point 16 where fracture by gross yielding occurs, or alternatively to
point 17 and the specimen will fracture by critical crack propagation at
point 18. If the ultimate strength line falls to the left of the fracture
toughness line, then fracture will always be by gross yielding.
7) Conclusions:
The suggested method gives powerful approach for evaluating the fatigue
life by measurable property called crack length. We have divided the whole
fatigue regime into six components for better understanding and calculated
crack propagation rates for each of them by applying block of cyclic loads
to know where the material fails. We have also introduced equivalent crack
to analyze and predict exact behaviour of micro cracks that have constantly
varying propagation rates.
8) References:
1) Shigley’s. Mechanical engineering design. McGraw-Hill; 2014.
2) Kitagawa H, Takahashi S. Applicability of fracture mechanics to very
small cracks or the cracks in the early stage. In: Proceedings of 2nd
international conference on mechanical behaviour of materials. Metals Park
(OH): ASM; 1976.p. 627–31.

You might also like