ME Web Module

Online Support Module
MECHANICAL ENGINEERING
1
CONTENTS
Learning Objectives 2
M.1 Introduction 2
Materials Selection for a Torsionally
Stressed Cylindrical Shaft (Case Study) 2
M.2 Strength ConsiderationsTorsionally
Stressed Shaft 3
M.3 Other Property Considerations and the Final
Decision 8
Fracture 9
M.4 Principles of Fracture Mechanics 9
M.5 Flaw Detection Using Nondestructive
Testing Techniques 21
M.6 Fracture Toughness Testing 25
M.7 Impact Fracture Testing 29
Fatigue 33
M.8 Cyclic Stresses 33
M.9 The S-N Curve 35
M.10 Crack Initiation and Propagation 37
M.11 Crack Propagation Rate 41
M.12 Factors That Affect Fatigue Life 45
M.13 Environmental Effects 47
Automobile Valve Spring (Case Study) 48
M.14 Mechanics of Spring Deformation 48
M.15 Valve Spring Design and Material
Requirements 50
M.16 One Commonly Employed Steel Alloy 52
Investigation of Engineering Failures 54
M.17 Reasons for Failure 55
M.18 Root Causes 56
The Failure Analysis 57
M.19 What Exactly Is the Failure Problem? 57
M.20 What Is the Root Cause of the Failure
Problem? 57
M.21 What Are Possible Solutions? 70
M.22 Which of These Is the Best
Solution? 71
M.23 Effective Evaluation of Corrective
Actions 71
M.24 The Final Report 71
Failure of an Automobile Rear Axle
(Case Study) 72
M.25 Introduction 72
M.26 Testing Procedure and Results 73
M.27 Discussion 79
Summary 80
Equation Summary 84
Important Terms and Concepts 85
References 86
Questions and Problems 86
Design Problems 90
Problem Duplication Guide 95
Glossary 96
Answers to Selected Problems 97
Index for Support Module 98
Due to constraints on book length, several topics especially suited to the discipline of
mechanical engineering were either not discussed in sufficient detail or omitted from the
print textbook. Therefore, it was decided to provide this online web module supplement,
which includes the following:
A materials selection case studyMaterials Selection for a Torsionally Stressed
Cylindrical Shaft.
Alternative (and more detailed) versions of Sections 9.5 and 9.8 Principles of
Fracture Mechanics and Fracture Toughness Testing.
An alternative (and more detailed) treatment of the topic of fatigueSections 9.9
through 9.14.
A case study on constraints and materials used for an automobile valve spring.
A submodule Investigation of Engineering Failures that outlines a protocol that
may be used to analyze the failure of engineered components.
Another case study that details an investigation that was conducted to determine
the cause of failure of an automobile rear axle.
Learning Objectives
After studying this web module, you should be able to do the following:
M.1 INTRODUCTION
1. Describe the manner in which materials
selection charts are employed in the materials
selection process.
2. Explain why the strengths of brittle materials
are much lower than predicted by theoretical
calculations.
3. Define fracture toughness in terms of (a) a
brief statement and (b) an equation; define all
parameters in this equation.
4. Make distinctions between stress intensity
factor, fracture toughness, and plane strain
fracture toughness.
5. In a qualitative manner, describe how a condi-
tional value of plane-strain fracture toughness
is determined using ASTM Standard E 399-09.
6. Name and briefly describe the two techniques
that are used to measure impact energy (or
notch toughness) of a material.
7. Define fatigue and specify the conditions
under which it occurs.
8. From a fatigue plot for some material,
determine (a) the fatigue lifetime (at a speci-
fied stress level) and (b) the fatigue strength
(at a specified number of cycles).
9. Cite five measures that may be taken to
improve the fatigue resistance of a metal.
10. Briefly describe the steps that are used to
ascertain whether a particular metal alloy is
suitable for use in an automobile valve spring.
11. List and briefly explain the three root causes of
failure.
12. List the four questions that a typical failure
investigation seeks to answer.
13. Make a list of procedures/analyses that were used
to determine the cause of failure described in the
Failure of an Automobile Rear Axle case study.
Materials Selection for a Torsionally Stressed
Cylindrical Shaft (Case Study)
We begin this web module for mechanical engineers by presenting a case study on ma-
terials selection. This process of materials selection involves, for some specified applica-
tion, choosing a material having a desirable or optimum property or combination of
2
M.2 Strength ConsiderationsTorsionally Stressed Shaft 3
M.2 STRENGTH CONSIDERATIONSTORSIONALLY STRESSED SHAFT
For this portion of the design problem, we will establish a criterion for selection of
light and strong materials for this shaft. We will assume that the twisting moment and
length of the shaft are specified, whereas the radius (or cross-sectional area) may be
varied. We develop an expression for the mass of material required in terms of twist-
ing moment, shaft length, and density and strength of the material. Using this expres-
sion, it will be possible to evaluate the performancethat is, maximize the strength
of this torsionally stressed shaft with respect to mass and, in addition, relative to
material cost.
Consider the cylindrical shaft of length L and radius r, as shown in Figure M.1. The
application of twisting moment (or torque), M
t
, produces an angle of twist . Shear stress
at radius r is defined by the equation
(M.1)
Here, J is the polar moment of inertia, which for a solid cylinder is
(M.2)
Thus,
(M.3)
A safe design calls for the shaft to be able to sustain some twisting moment without frac-
ture. In order to establish a materials selection criterion for a light and strong material,
we replace the shear stress in Equation M.3 with the shear strength of the material
f
divided by a factor of safety N, as
1
(M.4)
t
f
N

2M
t
pr
3
t
2M
t
pr
3
J
pr
4
2
t
M
t
r
J
properties. Selection of the proper material can reduce costs and improve performance.
Elements of this materials selection process involve deciding on the constraints of the
problem and, from these, establishing criteria that can be used in materials selection to
maximize performance.
The component or structural element we have chosen to discuss is one that has rele-
vance to a mechanical engineer: a solid cylindrical shaft that is subjected to a torsional
stress. Strength of the shaft will be considered in detail, and criteria will be developed for
maximizing strength with respect to both minimum material mass and minimum cost.
Other parameters and properties that may be important in this selection process are also
discussed briefly.
1
The factor of safety concept as well as guidelines for selecting values are discussed in Section 7.20.
Figure M.1 A solid cylindrical shaft that
experiences an angle of twist in response to
the application of a twisting moment M
t
.
f
r
L
M
t
4 Online Support Module: Mechanical Engineering
It is now necessary to take into consideration material mass. The mass m of any
given quantity of material is just the product of its density () and volume. Since the
volume of a cylinder is r
2
L, then
(M.5)
or, the radius of the shaft in terms of its mass is
(M.6)
Substituting this r expression into Equation M.4 leads to
(M.7)
Solving this expression for the mass m yields
(M.8)
The parameters on the right-hand side of this equation are grouped into three sets of paren-
theses. Those contained within the first set (i.e., N and M
t
) relate to the safe functioning of
the shaft. Within the second parentheses is L, a geometric parameter. Finally, the material
properties of density and strength are contained within the last set.
The upshot of Equation M.8 is that the best materials to be used for a light shaft that
can safely sustain a specified twisting moment are those having low ratios. In terms
of material suitability, it is sometimes preferable to work with what is termed a performance
index, P, which is just the reciprocal of this ratio; that is
(M.9)
In this context we want to use a material having a large performance index.
At this point it becomes necessary to examine the performance indices of a variety of
potential materials. This procedure is expedited by the use of materials selection charts.
2
These are plots of the values of one material property versus those of another property.
Both axes are scaled logarithmically and usually span about five orders of magnitude, so
as to include the properties of virtually all materials. For example, for our problem, the
chart of interest is logarithm of strength versus logarithm of density, which is shown in
Figure M.2.
3
It may be noted on this plot that materials of a particular type (e.g., woods,
and engineering polymers) cluster together and are enclosed within an envelope delin-
eated with a bold line. Subclasses within these clusters are enclosed using finer lines.
P
t
2
3
f
r
r
t
23
f
m 12NM
t
2
23
1p
13
L2
a
r
t
23
f
b
2M
t

B
pL
3
r
3
m
3

t
f
N

2M
t
pa
B
m
pLr
b
3
r
B
m
pLr
m pr
2
Lr
2
A comprehensive collection of these charts may be found in M. F. Ashby, Materials Selection in Mechanical Design,
4th edition, Butterworth-Heinemann, Woburn, UK, 2011.
3
Strength for metals and polymers is taken as yield strength; for ceramics and glasses, compressive strength; for
elastomers, tear strength; and for composites, tensile failure strength.
For a cylindrical
shaft of length L and
radius r that is
stressed in torsion,
expression for mass
in terms of density
and shear strength of
the shaft material
Strength
performance index
expression for a
torsionally stressed
cylindrical shaft
performance index
materials selection
chart
Now, taking the logarithm of both sides of Equation M.9 and rearranging yields
(M.10)
This expression tells us that a plot of log t
f
versus log r will yield a family of straight and
parallel lines all having a slope of each line in the family corresponds to a different per-
formance index, P. These lines are termed design guidelines, and four have been included in
Figure M.2 for Pvalues of 3, 10, 30, and 100 All materials that lie on one of
these lines will perform equally well in terms of strength-per-mass basis; materials whose
(MPa)
23
m
3
/Mg.
3
2
;
log t
f
3
2
log r
3
2
log P
0.1 0.3 1 3 10 30
10,000
1000
100
10
1
0.1
Mg
Alloys
Ash
Si
Glasses
Density (Mg /m
3
)
Engineering
ceramics
Engineering
composites
Engineering
alloys
Porous
ceramics
Engineering
polymers
Woods
Elastomers
Polymer
foams
S
t
r
e
n
g
t
h

(
M
P
a
)
P = 100
P = 30
P = 10
P = 3
Diamond
Cermets
Sialons
B
MgO
Al
2
O
3
ZrO
2
Si
3
N
4
SiC
Ge
Nylons
PMMA
PS
PP
MEL
PVC
Epoxies
Polyesters
HDPE
PU
PTFE
Silicone
LDPE
Soft
Butyl
Wood
Products
Ash
Balsa
Balsa
Oak
Oak
Pine
Pine
Fir
Fir
Parallel
to Grain
Perpendicular
to Grain
Cork
Cement
Concrete
Engineering
alloys
KFRP
CFRP
Be
GFRP
Laminates
KFRP
Pottery
Ti
Alloys
Steels
W Alloys
Mo Alloys
Ni Alloys
Cu Alloys
Cast
Irons
Zn
Alloys
Stone,
Rock
Lead
Alloys
CFRP
GFRP
UNIPLY
Al Alloys
Figure M.2 Strength versus density materials selection chart. Design guidelines for performance indices of 3,
10, 30, and 100 (MPa)
23
m
3
/Mg have been constructed, all having a slope of
(Adapted from M. F. Ashby, Materials Selection in Mechanical Design. Copyright 1992. Reprinted by permission of
Butterworth-Heinemann Ltd.)
3
2
.
positions lie above a particular line will have higher performance indices, whereas those
lying below will exhibit poorer performances. For example, a material on the P 30 line
will yield the same strength with one-third the mass as another material that lies along
the P 10 line.
The selection process now involves choosing one of these lines, a selection line
that includes some subset of these materials; for the sake of argument let us pick P
10 m
3
/Mg, which is represented in Figure M.3. Materials lying along this line or
above it are in the search region of the diagram and are possible candidates for this
1MPa2
23
Engineering
alloys
KFRP
CFRPBe
GFRP
Laminates
KFRP
Pottery
Mg
Alloys
Ti
Alloys
Steels
W Alloys
Mo Alloys
Ni Alloys
Cu Alloys
Cast
Irons
Zn
Alloys
Stone,
Rock
Al Alloys
Lead
Alloys
CFRP
GFRP
UNIPLY
10,000
1000
100
10
1
Glasses
Engineering
composites
Engineering
alloys
Porous
ceramics
Engineering
polymers
Woods
Elastomers
Polymer
foams
S
t
r
e
n
g
t
h

(
M
P
a
)
P = 10
Diamond
Cermets
Sialons
B
MgO
Al
2
O
3
ZrO
2
Si
3
N
4
SiC
Si
Ge
Nylons
PMMA
PS
PP
MEL
PVC
Epoxies
Polyesters
HDPE
PU
PTFE
Silicone
Cement
Concrete
LDPE
Soft
Butyl
Wood
Products
Ash
Ash
Balsa
Balsa
Oak
Oak
Pine
Pine
Fir
Fir
Parallel
to Grain
Perpendicular
to Grain
Cork
300 MPa
(MPa)
2/3
m
3
/Mg
0.1 0.3 1 3 10 30
0.1
Density (Mg /m
3
)
Engineering
ceramics
Figure M.3 Strength versus density materials selection chart. Materials within the shaded region are acceptable
candidates for a solid cylindrical shaft that has a mass-strength performance index in excess of 10
and a strength of at least 300 MPa (43,500 psi).
(Adapted from M. F. Ashby, Materials Selection in Mechanical Design. Copyright 1992. Reprinted by permission of
Butterworth-Heinemann Ltd.)
(MPa)
2
3
m
3
/Mg
rotating shaft. These include wood products, some plastics, a number of engineering
alloys, the engineering composites, and glasses and engineering ceramics. On the basis of
fracture toughness considerations, the engineering ceramics and glasses are ruled out as
possibilities.
Let us now impose a further constraint on the problemnamely that the strength of
the shaft must equal or exceed 300 MPa (43,500 psi). This may be represented on the ma-
terials selection chart by a horizontal line constructed at 300 MPa, see Figure M.3. Now
the search region is further restricted to the area above both of these lines. Thus, all wood
products, all engineering polymers, other engineering alloys (viz., Mg and some Al alloys),
and some engineering composites are eliminated as candidates; steels, titanium alloys,
high-strength aluminum alloys, and the engineering composites remain as possibilities.
At this point we are in a position to evaluate and compare the strength performance
behavior of specific materials. Table M.1 presents the density, strength, and strength per-
formance index for three engineering alloys and two engineering composites, which
were deemed acceptable candidates from the analysis using the materials selection
chart. In this table, strength was taken as 0.6 times the tensile yield strength (for the
alloys) and 0.6 times the tensile strength (for the composites); these approximations
were necessary because we are concerned with strength in torsion, and torsional
strengths are not readily available. Furthermore, for the two engineering composites, it
is assumed that the continuous and aligned glass and carbon fibers are wound in a helical
fashion (Figure 15.15) and at a 45 angle referenced to the shaft axis. The five materials
in Table M.1 are ranked according to strength performance index, from highest to lowest:
carbon fiber-reinforced and glass fiber-reinforced composites, followed by aluminum,
titanium, and 4340 steel alloys.
Materials cost is another important consideration in the selection process. In real-
life engineering situations, economics of the application often is the overriding issue and
normally will dictate the material of choice. One way to determine materials cost is by
taking the product of the price (on a per-unit mass basis) and the required mass of
material.
Cost considerations for these five remaining candidate materialssteel, aluminum,
and titanium alloys, and two engineering compositesare presented in Table M.2. In
the first column is tabulated The next column lists the approximate relative
cost, denoted as this parameter is simply the per-unit mass cost of material divided by
the per-unit mass cost for low-carbon steel, one of the common engineering materials.
The underlying rationale for using is that although the price of a specific material will
vary over time, the price ratio between that material and another will, most likely,
change more slowly.
c
c;
r/t
23
f
.

f

f
23
/ P
Material (Mg/m
3
) (MPa) [(MPa)
2/3
m
3
/Mg]
Carbon fiber-reinforced composite 1.5 1140 72.8
(0.65 fiber fraction)
a
Glass fiber-reinforced composite 2.0 1060 52.0
a
Aluminum alloy (2024-T6) 2.8 300 16.0
Titanium alloy (Ti-6Al-4V) 4.4 525 14.8
4340 Steel (oil-quenched 7.8 780 10.9
and tempered)
a
The fibers in these composites are continuous, aligned, and wound in a helical fashion at
a 45 angle relative to the shaft axis.
Table M.1
Density (), Strength
(
f
), and Strength
Performance Index (P)
for Five Engineering
Materials
Finally, the right-hand column of Table M.2 shows the product of and . This
product provides a comparison of these materials on the basis of the cost of materials
for a cylindrical shaft that would not fracture in response to the twisting moment M
t
. We
use this product inasmuch as is proportional to the mass of material required
(Equation M.8) and is the relative cost on a per-unit mass basis. Now the most econom-
ical is the 4340 steel, followed by the glass fiber-reinforced composite, the carbon fiber-
reinforced composite, 2024-T6 aluminum, and the titanium alloy. Thus, when the issue of
economics is considered, there is a significant alteration within the ranking scheme. For
example, inasmuch as the carbon fiber-reinforced composite is relatively expensive, it is
significantly less desirable; in other words, the higher cost of this material may not out-
weigh the enhanced strength it provides.
c
r
t
23
f
r
t
23
f
c
Table M.2 Tabulation of the
f
23
Ratio, Relative Cost (c
), and Product of
f
23
and c
for Five Engineering

Materials
a
f
23
c
(
f
23
)
Material 10
2
[Mg(MPa)
23
m
3
] ($$) 10
2
($$)[Mg(MPa)
23
m
3
]
4340 Steel (oil-quenched 9.2 3.0 27
and tempered)
Glass fiber-reinforced composite 1.9 28.3 54
b
Carbon fiber-reinforced composite 1.4 43.1 60
b
Aluminum alloy (2024-T6) 6.2 12.4 77
Titanium alloy (Ti-6Al-4V) 6.8 94.2 641
a
The relative cost is the ratio of the price per unit mass of the material and a low-carbon steel.
b
The fibers in these composites are continuous, aligned, and wound in a helical fashion at a angle relative to the
shaft axis.
45
M.3 OTHER PROPERTY CONSIDERATIONS AND THE FINAL DECISION
To this point in our materials selection process we have considered only the strength of
materials. Other properties relative to the performance of the cylindrical shaft may be
importantfor example, stiffness, and, if the shaft rotates, fatigue behavior (Sections M.9
through M.13). Furthermore, fabrication costs should also be considered; in our analysis
they have been neglected.
Relative to stiffness, a stiffness-to-mass performance analysis similar to the one just dis-
cussed could be conducted. For this case, the stiffness performance index is
(M.11)
where G is the shear modulus. The appropriate materials selection chart (log G versus
log r) would be used in the preliminary selection process. Subsequently, performance
index and per-unit-mass cost data would be collected on specific candidate materials;
from these analyses the materials would be ranked on the basis of stiffness perform-
ance and cost.
In deciding on the best material, it may be worthwhile to make a table employing the
results of the various criteria that were used. The tabulation would include, for all candidate
materials, performance index, cost, and so forth for each criterion, as well as comments rel-
ative to any other important considerations. This table puts in perspective the important
issues and facilitates the final decision process.
P
s

1G
r
P
s
Fracture
M.4 PRINCIPLES OF FRACTURE MECHANICS
4
Brittle fracture of normally ductile materials, such as that shown in the chapter-opening
Figure b (of the oil barge) for Chapter 9, has demonstrated the need for a better under-
standing of the mechanisms of fracture. Extensive research endeavors over the last cen-
tury have led to the evolution of the field of fracture mechanics. This subject allows
quantification of the relationships between material properties, stress level, the presence
of crack-producing flaws, and crack propagation mechanisms. Design engineers are now
better equipped to anticipate, and thus prevent, structural failures. The present discus-
sion centers on some of the fundamental principles of the mechanics of fracture.
Stress Concentration
The fracture strength of a solid material is a function of the cohesive forces that exist
between atoms. On this basis, the theoretical cohesive strength of a brittle elastic solid
has been estimated to be approximately E/10, where E is the modulus of elasticity. The
experimental fracture strengths of most engineering materials normally lie between 10
and 1000 times below this theoretical value. In the 1920s, A. A. Griffith proposed that
this discrepancy between theoretical cohesive strength and observed fracture strength
could be explained by the presence of microscopic flaws or cracks that always exist un-
der normal conditions at the surface and within the interior of a body of material. These
flaws are a detriment to the fracture strength because an applied stress may be ampli-
fied or concentrated at the tip, the magnitude of this amplification depending on crack
orientation and geometry. This phenomenon is demonstrated in Figure M.4, a stress
profile across a cross section containing an internal crack. As indicated by this profile,
fracture mechanics
Figure M.4 (a) The geometry of surface and internal cracks. (b) Schematic stress profile along the line X-X in (a),
demonstrating stress amplification at crack tip positions.
r
t
s
0
s
0
X' X
x' x
2a
a
Position along X
X'
s
0
s
m
S
t
r
e
s
s
x' x
4
This section is an expanded and more detailed version of Section 9.5.
(a) (b)
the magnitude of this localized stress decreases with distance away from the crack tip.
At positions far removed, the stress is equal to the nominal stress
0
, or the applied load
divided by the specimen cross-sectional area (perpendicular to this load). Because of
their ability to amplify an applied stress in their locale, these flaws are sometimes called
stress raisers.
If it is assumed that a crack has an elliptical shape (or is circular) and is oriented per-
pendicular to the applied stress, the maximum stress at the crack tip, , is equal to
(M.12a)
where
0
is the magnitude of the nominal applied tensile stress,
t
is the radius of curvature
of the crack tip (Figure M.4a), and a represents the length of a surface crack, or half of the
length of an internal crack. For a relatively long microcrack that has a small tip radius of
curvature, the factor may be very large (certainly much greater than unity); under
these circumstances Equation M.12a takes the form
(M.12b)
Furthermore,
m
will be many times the value of
0
.
Sometimes the ratio
m
0
is denoted the stress concentration factor K
t
:
(M.13)
which is simply a measure of the degree to which an external stress is amplified at the
tip of a crack.
Note that stress amplification is not restricted to these microscopic defects; it may
occur at macroscopic internal discontinuities (e.g., voids or inclusions), sharp corners,
scratches, and notches. Figure M.5 shows theoretical stress concentration factor curves for
several simple and common macroscopic discontinuities.
The effect of a stress raiser is more significant in brittle than in ductile materials. For
a ductile metal, plastic deformation ensues when the maximum stress exceeds the yield
strength. This leads to a more uniform distribution of stress in the vicinity of the stress
raiser and to the development of a maximum stress concentration factor less than the
theoretical value. Such yielding and stress redistribution do not occur to any apprecia-
ble extent around flaws and discontinuities in brittle materials; therefore, essentially the
theoretical stress concentration will result.
Griffith then went on to propose that all brittle materials contain a population of
small cracks and flaws, which have a variety of sizes, geometries, and orientations.
Fracture will result when, upon application of a tensile stress, the theoretical cohesive
strength of the material is exceeded at the tip of one of these flaws. This leads to the for-
mation of a crack that then rapidly propagates. If no flaws were present, the fracture
strength would be equal to the cohesive strength of the material. Very small and virtu-
ally defect-free metallic and ceramic whiskers have been grown with fracture strengths
that approach their theoretical values.
Griffith Theory of Brittle Fracture
During the propagation of a crack, there is a release of what is termed the elastic strain
energy, some of the energy that is stored in the material as it is elastically deformed.
K
t
s
m
s
0
2 a
a
r
t
b
1 2
s
m
2s
0
a
a
r
t
b
1
2
1a
r
t
2
1 2
s
m
s
0
c 1 2 a
a
r
t
b
1
2
d
s
m
stress raiser
For tensile loading,
computation of
maximum stress at
a crack tip
Furthermore, during the crack extension process, new free surfaces are created at the
faces of a crack, which give rise to an increase in surface energy of the system. Griffith
developed a criterion for crack propagation of an elliptical crack (Figure M.4a) by perform-
ing an energy balance using these two energies. He demonstrated that the critical stress s
c
required for crack propagation in a brittle material is described by the expression
(M.14) s
c
a
2Eg
s
pa
b
1 2
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n
f
a
c
t
o
r

K
t
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n

f
a
c
t
o
r

K
t
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n

f
a
c
t
o
r

K
t
w
d
w
3.4
3.0
2.6
2.2
1.8
3.8
3.4
3.0
2.6
2.2
1.8
1.4
1.0
3.2
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0 0.2 0.4 0.6 0.8 1.0
0 0.2 0.4 0.6 0.8
0 0.2 0.4 0.6 0.8 1.0
d
2r
b
b
h
w
w
h
r
+
r
h
r
h
b
r
1
2
=
b
r
1 =
w
h
2.00 =
w
h
1.25 =
w
h
1.10 =
b
r
4 =
Figure M.5
Theoretical stress
concentration factor
curves for three simple
geometrical shapes.
(From G. H. Neugebauer,
Prod. Eng. (NY), Vol. 14,
pp. 8287, 1943.)
Critical stress for
crack propagation in
a brittle material
(a)
(b)
(c)
where
a 5 one-half the length of an internal crack
Worth noting is that this expression does not involve the crack tip radius r
t
, as does the
stress concentration equations (Equations M.12a and M.12b); however, it is assumed that
the radius is sufficiently sharp (on the order of the interatomic spacing) so as to raise the
local stress at the tip above the cohesive strength of the material.
The previous development applies only to completely brittle materials, for which
there is no plastic deformation. Most metals and many polymers do experience some
plastic deformation during fracture; thus, crack extension involves more than producing
just an increase in the surface energy. This complication may be accommodated by re-
placing g
s
in Equation M.14 by g
s
g
p
, where g
p
represents a plastic deformation en-
ergy associated with crack extension. Thus,
(M.15a)
For highly ductile materials, it may be the case that g
p
>> g
s
such that
(M.15b)
In the 1950s, G. R. Irwin chose to incorporate both g
s
and g
p
into a single term, G
c
, as
(M.16)
G
c
is known as the critical strain energy release rate. Incorporation of Equation M.16
into Equation M.15a after some rearrangement leads to another expression for the
Griffith cracking criterion as
(M.17)
Thus, crack extension occurs when exceeds the value of G
c
for the particular
material under consideration.
ps
2
a
E
G
c
ps
2
a
E
G
c
21g
s
g
p
2
s
c
a
2Eg
p
pa
b
1
2
s
c
c
2E1g
s
g
p
2
pa
d
1
2
g
s
specific surface energy
E modulus of elasticity
EXAMPLE PROBLEM M.1
Maximum Flaw Length Computation
A relatively large plate of a glass is subjected to a tensile stress of 40 MPa. If the specific sur-
face energy and modulus of elasticity for this glass are 0.3 J/m
2
and 69 GPa, respectively, deter-
mine the maximum length of a surface flaw that is possible without fracture.
Solution
To solve this problem it is necessary to employ Equation M.14. Rearranging this expression
such that a is the dependent variable, and realizing that and
, leads to
m 8.2 10
6
m 0.0082 mm 8.2

122 169 10
9
N/m
2
2 10.3 N/m2
p 140 10
6
N/m
2
2
2
a
2Eg
s
ps
2
E 69 GPa
g
s
0.3 J/m
2
, s 40 MPa,
Stress Analysis of Cracks
As we continue to explore the development of fracture mechanics, it is worthwhile to
examine the stress distributions in the vicinity of the tip of an advancing crack. There are
three fundamental ways, or modes, by which a load can operate on a crack, and each will
affect a different crack surface displacement; these are illustrated in Figure M.6. Mode I
is an opening (or tensile) mode, whereas modes II and III are sliding and tearing modes,
respectively. Mode I is encountered most frequently, and only it will be treated in the en-
suing discussion on fracture mechanics.
For this mode I configuration, the stresses acting on an element of material are
shown in Figure M.7. Using elastic theory principles and the notation indicated, tensile
(s
x
and s
y
)
5
and shear (t
xy
) stresses are functions of both radial distance r and the angle u
as follows:
6
(M.18a)
(M.18b)
(M.18c)
If the plate is thin relative to the dimensions of the crack, then s
z
0, or a condition of
plane stress is said to exist. At the other extreme (a relatively thick plate), s
z
n(s
x
s
y
), and the state is referred to as plane strain (since 0); n in this expression is
Poissons ratio.
In Equations M.18, the parameter Kis termed the stress intensity factor; its use pro-
vides for a convenient specification of the stress distribution around a flaw. It should be
noted that this stress intensity factor and the stress concentration factor K
t
in Equation
M.13, although similar, are not equivalent.
P
z
t
xy
K
22pr
f
xy
1u2
s
y
K
22pr
f
y
1u2
s
x
K
22pr
f
x
1u2
5
This
y
denotes a tensile stress parallel to the y-direction and should not be confused with the yield strength
(Section 7.6), which uses the same symbol.
6
The f() functions are as follows:
f
xy
1u2 sin
u
2
cos
u
2
cos
3u
2
f
y
1u2 cos
u
2
a 1 sin
u
2
sin
3u
2
b
f
x
1u2 cos
u
2
a 1 sin
u
2
sin
3u
2
b
Figure M.6 The three
modes of crack surface
displacement. (a) Mode I,
opening or tensile mode;
(b) mode II, sliding mode; and
(c) mode III, tearing mode.
(a) (b) (c)
stress intensity factor
plane strain
The stress intensity factor is related to the applied stress and the crack length by the
following equation:
(M.19)
Here Y is a dimensionless parameter or function that depends on both the crack and
specimen sizes and geometries, as well as on the manner of load application. More will
be said about Y in the discussion that follows. Moreover, it should be noted that K has
the unusual units of [alternatively ]).
Fracture Toughness
In the previous discussion, a criterion was developed for the crack propagation in a brittle
material containing a flaw; fracture occurs when the applied stress level exceeds some critical
value s
c
(Equation M.14). Similarly, since the stresses in the vicinity of a crack tip can be
defined in terms of the stress intensity factor, a critical value of Kexists that may be used to
specify the conditions for brittle fracture; this critical value is termed the fracture toughness
K
c
, and, from Equation M.19, is defined by
(M.20)
Here, s
c
again is the critical stress for crack propagation, and we now represent Yas a func-
tion of both crack length (a) and component width (W)i.e., as Y(aW).
Relative to this Y(aW) function, as the aWratio approaches zero (i.e., for very wide
planes and short cracks), Y(a/W) approaches a value of unity. For example, for a plate of
infinite width having a through-thickness crack, Figure M.8a, Y(a/W) 1.0; whereas for a
plate of semi-infinite width containing an edge crack of length a (Figure M.8b), Y(aW)
1.1. Mathematical expressions for Y(aW) (often relatively complex) in terms of aW
are required for components of finite dimensions. For example, for a center-cracked
(through-thickness) plate of width W(Figure M.9),
(M.21) Y1a
W2 a
W
pa
tan
pa
W
b
1
K
c
Y1a
W2 s
c
1pa
ksi 1in. MPa1m 1psi 1in.
K Ys 1pa
Fracture toughness
dependence on
critical stress for
crack propagation
and crack length
fracture toughness
Figure M.7 The stresses acting
in front of a crack that is loaded
in a tensile mode I configuration.
y
r
x
z
q
s
y
s
x
s
z
t
xy
Stress intensity
factordependence
on applied stress and
crack length
Here the pa/Wargument for the tangent is expressed in radians, not degrees. It is often
the case for some specific component-crack configuration that Y(a/W) is plotted versus
a/W(or some variation of a/W). Several of these plots are shown in Figures M.10a, b, and
c; included in the figures are equations that are used to determine K
c
s.
By definition, fracture toughness is a property that is the measure of a materials
resistance to brittle fracture when a crack is present. Its units are the same as for the
stress intensity factor (i.e., or
For relatively thin specimens, the value of K
c
depends on and decreases with increas-
ing specimen thickness B, as indicated in Figure M.11. Eventually, K
c
becomes
independent of B, at which time the condition of plane strain is said to exist.
7
The constant
K
c
value for thicker specimens is known as the plane strain fracture toughness K
Ic
, which
is also defined by
8
(M.22)
It is the fracture toughness normally cited since its value is always less than K
c
. The I sub-
script for K
Ic
denotes that this critical value of K is for mode I crack displacement, as
illustrated in Figure M.6a. Brittle materials, for which appreciable plastic deformation is
not possible in front of an advancing crack, have low K
Ic
values and are vulnerable to
catastrophic failure. On the other hand, K
Ic
values are relatively large for ductile mate-
rials. Fracture mechanics is especially useful in predicting catastrophic failure in materi-
als having intermediate ductilities. Plane strain fracture toughness values for a number
K
Ic
Ys1pa
psi 1in. 2 . MPa1m
Figure M.8 Schematic representations of (a) an interior
crack in a plate of infinite width, and (b) an edge crack in a
plate of semi-infinite width.
2a
a
Figure M.9 Schematic
representation of a flat
plate of finite width
having a through-
thickness center crack.
W
2
W
2a
B
s
s
(a) (b)
Minimum specimen
thickness for a
condition of plane
strain
plane strain fracture
toughness
Plane strain fracture
toughness for mode
I crack surface
displacement
7
Experimentally, it has been verified that for plane strain conditions
(M.23)
where s
y
is the 0.002 strain offset yield strength of the material.
8
In the ensuing discussion we will use Y to designate Y(a/W) in order to simplify the form
of the equations.
B 2.5 a
K
Ic
s
y
b
2
W
F
Y
F
a
B
F
B
a
W
S
1.00
2.00
3.00
4.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6
a/W
K
c

YF
WB
a
Y
1.10
1.12
1.14
1.16
1.18
1.20
1.22
0.0 0.1 0.2 0.3 0.4 0.5 0.6
2a/W
K
c

YF
WB
a
Y
0.9
1.1
1.3
1.5
1.7
1.9
0.0 0.1 0.2 0.3 0.4 0.5 0.6
a/W
K
c

S/W 8
S/W 4
3FSY
4W
2
B
a
W
F
F
a a
B
Figure M.10 Y calibration curves for three simple crack-plate geometries.
(Copyright ASTM. Reprinted with permission.)
(a)
(b)
(c)
of different materials are presented in Table M.3; a more extensive list of K
Ic
values is
given in Table B.5 of Appendix B.
The stress intensity factor Kin Equations M.18 and the plane strain fracture toughness
K
Ic
are related to one another in the same sense as are stress and yield strength. A material
may be subjected to many values of stress; however, there is a specific stress level at which
the material plastically deformsthat is, the yield strength. Likewise, a variety of Ks are
possible, whereas K
Ic
is unique for a particular material, and indicates the conditions of flaw
size and stress necessary for brittle fracture.
Several different testing techniques are used to measure K
Ic
; one of these is described
later in Section M.6. Virtually any specimen size and shape consistent with mode I crack dis-
placement may be utilized, and accurate values will be realized, provided that the Y scale
parameter in Equation M.22 has been properly determined.
Figure M.11 Schematic representation showing
the effect of plate thickness on fracture toughness.
F
r
a
c
t
u
r
e

t
o
u
g
h
n
e
s
s

K
c
Plane stress
behavior
Thickness B
Plane strain
behavior
K
Ic
Yield Strength K
Ic
Material MPa ksi MPa ksi .
Metals
Aluminum alloy
a
(7075-T651) 495 72 24 22
Aluminum alloy
a
(2024-T3) 345 50 44 40
Titanium alloy
a
(Ti-6Al-4V) 910 132 55 50
Alloy steel
a
1640 238 50.0 45.8
(4340 tempered @ 260C)
Alloy steel
a
1420 206 87.4 80.0
(4340 tempered @ 425C)
Ceramics
Concrete 0.21.4 0.181.27
Soda-lime glass 0.70.8 0.640.73
Aluminum oxide 2.75.0 2.54.6
Polymers
Polystyrene (PS) 25.069.0 3.6310.0 0.71.1 0.641.0
Poly(methyl methacrylate) 53.873.1 7.810.6 0.71.6 0.641.5
(PMMA)
Polycarbonate (PC) 62.1 9.0 2.2 2.0
1in 1m
Table M.3
Room-Temperature
Yield Strength and
Plane Strain Fracture
Toughness Data for
Selected Engineering
Materials
a
Source: Reprinted
with permission,
Advanced Materials
and Processes, ASM
International, 1990.
The plane strain fracture toughness K
Ic
is a fundamental material property that
depends on many factors, the most influential of which are temperature, strain rate, and
microstructure. The magnitude of K
Ic
decreases with increasing strain rate and decreasing
temperature. Furthermore, an enhancement in yield strength wrought by solid solution
or dispersion additions or by strain hardening generally produces a corresponding
decrease in K
Ic
. Furthermore, K
Ic
normally increases with reduction in grain size as com-
position and other microstructural variables are maintained constant. Yield strengths are
included for some of the materials listed in Table M.3.
Design Using Fracture Mechanics
According to Equations M.20 and M.22, three variables must be considered relative to
the possibility for fracture of some structural componentviz. the fracture toughness
(K
c
) or plane strain fracture toughness (K
Ic
), the imposed stress (s), and the flaw size
(a), assuming, of course, that Y has been determined. When designing a component, it is
first important to decide which of these variables are constrained by the application and
which are subject to design control. For example, material selection (and hence K
c
or
K
Ic
) is often dictated by factors such as density (for lightweight applications) or the cor-
rosion characteristics of the environment. Alternatively, the allowable flaw size is either
measured or specified by the limitations of available flaw detection techniques. It is
important to realize, however, that once any combination of two of the preceding param-
eters is prescribed, the third becomes fixed (Equations M.20 and M.22). For example,
assume that K
Ic
and the magnitude of a are specified by application constraints; therefore,
the design (or critical) stress s
c
is given by
(M.24)
On the other hand, if stress level and plane strain fracture toughness are fixed by the
design situation, then the maximum allowable flaw size a
c
is given by
(M.25) a
c
1
p
a
K
Ic
sY
b
2
s
c
K
Ic
Y1pa
Computation of
design stress
Computation of
maximum allowable
flaw length
EXAMPLE PROBLEM M.2
Determination of the Possibility of Critical Flaw Detection
A structural component in the form of a very wide plate, as shown in Figure M.8a, is to be fab-
ricated from a 4340 steel. Two sheets of this alloy, each having a different heat treatment and
thus different mechanical properties, are available. One, denoted material A, has a yield
strength of 860 MPa (125,000 psi) and a plane strain fracture toughness of 98.9 MPa
. For the other, material Z,
y
and K
Ic
values are 1515 MPa (220,000 psi) and
respectively.
(a) For each alloy, determine whether or not plane strain conditions prevail if the plate is 10
mm (0.39 in.) thick.
(b) It is not possible to detect flaw sizes less than 3 mm, which is the resolution limit of the flaw
detection apparatus. If the plate thickness is sufficient such that the K
Ic
value may be used,
determine whether or not a critical flaw is subject to detection. Assume that the design stress
level is one half of the yield strength; also, for this configuration, the value of Y is 1.0.
60.4 MPa1m 155 ksi 1in. 2 ,
190 ksi 1in. 2
1m
Solution
(a) Plane strain is established by Equation M.23. For material A,
0.033 m 33 mm (1.30 in.)
Thus, plane strain conditions do not hold for material A because this value of B is greater
than 10 mm, the actual plate thickness; the situation is one of plane stress and must be
treated as such.
And for material Z,
which is less than the actual plate thickness, and therefore the situation is one of plane strain.
(b) We need only to determine the critical flaw size for material Z because the situation for
material A is not plane strain, and K
Ic
may not be used. Employing Equation M.25 and
taking s to be s
y
/2,
0.002 m 2.0 mm (0.079 in.)
Therefore, the critical flaw size for material Z is not subject to detection since it is less
than 3 mm.
a
c
1
p
a
60.4 MPa1m
112 11515/22 MPa
b
2
B 2.5 a
60.4 MPa1m
1515 MPa
b
2
0.004 m 4.0 mm 10.16 in. 2
B 2.5 a
K
Ic
s
y
b
2
2.5 a
98.9 MPa1m
860 MPa
b
2
DESIGN EXAMPLE M.1
Material Specification for a Pressurized Spherical Tank
Consider a thin-walled spherical tank of radius r and thickness t (Figure M.12) that may be used
as a pressure vessel.
(a) One design of such a tank calls for yielding of the wall material prior to failure as a result of
the formation of a crack of critical size and its subsequent rapid propagation. Thus, plastic distor-
tion of the wall may be observed
and the pressure within the tank
released before the occurrence of
catastrophic failure. Consequently,
materials having large critical
crack lengths are desired. On the
basis of this criterion, rank the
metal alloys listed in Table B.5,
Appendix B, as to critical crack
size, from longest to shortest.
(b) An alternative design that is
also often utilized with pressure
vessels is termed leak-before-
break. On the basis of principles
of fracture mechanics, allowance
is made for the growth of a crack
through the thickness of the
t
p
p
p
p
p
p
p
p
r
2a
s
s
Figure M.12 Schematic diagram showing the cross section
of a spherical tank that is subjected to an internal pressure p,
and that has a radial crack of length 2a in its wall.
vessel wall prior to the occurrence of rapid crack propagation (Figure M.12). Thus, the crack will
completely penetrate the wall without catastrophic failure, allowing for its detection by the leak-
ing of pressurized fluid. With this criterion the critical crack length a
c
(i.e., one-half the total
internal crack length) is taken to be equal to the pressure vessel thickness t. Allowance for a
c
t instead of a
c
t/2 ensures that fluid leakage will occur prior to the buildup of dangerously
high pressures. Using this criterion, rank the metal alloys in Table B.5, Appendix B as to the
maximum allowable pressure.
For this spherical pressure vessel, the circumferential wall stress s is a function of the pressure
p in the vessel and the radius r and wall thickness t according to
(M.26)
For both parts (a) and (b), assume a condition of plane strain.
Solution
(a) For the first design criterion, it is desired that the circumferential wall stress be less than
the yield strength of the material. Substitution of s
y
for s in Equation M.22, and incorporation
of a factor of safety N leads to
(M.27)
where a
c
is the critical crack length. Solving for a
c
yields the following expression:
(M.28)
Therefore, the critical crack length is proportional to the square of the K
Ic
-s
y
ratio, which
is the basis for the ranking of the metal alloys in Table B.5. The ranking is provided in Table
M.4, where it may be seen that the medium carbon (1040) steel with the largest ratio has the
longest critical crack length and, therefore, is the most desirable material on the basis of this
criterion.
(b) As stated previously, the leak-before-break criterion is just met when one-half the internal
crack length is equal to the thickness of the pressure vessel (i.e., when a t). Substitution of
a t into Equation M.22 gives
a
c

N
2
Y
2
p
a
K
Ic
s
y
b
2
K
Ic
Y a
s
y
N
b 1pa
c
s
pr
2t
Material
(mm)
Medium carbon (1040) steel 43.1
AZ31B magnesium 19.6
2024 aluminum (T3) 16.3
Ti-5Al-2.5Sn titanium 6.6
4140 steel (tempered @ ) 5.3
Ti-6Al-4V titanium 3.7
17-7PH stainless steel 3.4
7075 aluminum (T651) 2.4
4340 steel (tempered @ ) 0.93 260C
370C
425C
482C
a
K
Ic
S
y
b
2
Table M.4
Ranking of Several Metal
Alloys Relative to Critical
Crack Length (Yielding
Criterion) for a Thin-Walled
Spherical Pressure Vessel
M.5 Flaw Detection Using Nondestructive Testing Techniques 21
(M.29)
And, from Equation M.26,
(M.30)
The stress is replaced by the yield strength because the tank should be designed to contain the
pressure without yielding; furthermore, substitution of Equation M.30 into Equation M.29, af-
ter some rearrangement, yields the following expression:
(M.31)
Hence, for some given spherical vessel of radius r, the maximum allowable pressure con-
sistent with this leak-before-break criterion is proportional to The same several
materials are ranked according to this ratio in Table M.5; as may be noted, the medium
carbon steel will contain the greatest pressures.
Of the eleven metal alloys listed in Table B.5, the medium carbon steel ranks first ac-
cording to both yielding and leak-before-break criteria. For these reasons, many pressure
vessels are constructed of medium carbon steels when temperature extremes and corro-
sion need not be considered.
K
2
Ic
s
y
.
p
2
Y
2
pr
a
K
2
Ic
s
y
b
t
pr
2s
K
Ic
Ys1pt
Material
(MPa-m)
Medium carbon (1040) steel 11.2
Ti-5Al-2.5Sn titanium 5.8
2024 aluminum (T3) 5.6
17-7PH stainless steel 4.4
AZ31B magnesium 3.9
Ti-6Al-4V titanium 3.3
4140 steel (tempered @ C) 2.4
7075 aluminum (T651) 1.2
260C
370
425C
482C
K
Ic
2
S
y
Table M.5
Ranking of Several Metal
Alloys Relative to Maximum
Allowable Pressure (Leak-
Before-Break Criterion) for a
Thin-Walled Spherical Pressure
Vessel
M.5 FLAW DETECTION USING NONDESTRUCTIVE TESTING TECHNIQUES
A number of nondestructive testing (NDT) techniques have been developed that per-
mit detection and measurement of both internal and surface flaws.
9
Such techniques are
used to examine structural components currently in service for defects and flaws that
could lead to premature failure; in addition, NDTs are used as a means of quality con-
9
Sometimes the terms nondestructive evaluation (NDE) and nondestructive inspection (NDI) are also used for these
techniques.
nondestructive
testing
trol for manufacturing processes. As the name implies, these techniques must not destroy
or damage the material/structure being examined nor impair its future serviceability.
Some testing techniques are capable of detecting only surface defects, some only subsur-
face (interior) defects, while other tests detect defects at both surface and subsurface
sites. Furthermore, in some instances location of testing is important. Some testing meth-
ods must be conducted in a laboratory setting; others may be adapted for use in the field.
Several commonly employed NDT techniques are visual inspection, optical microscopy,
scanning electron microscopy, dye (or liquid) penetrant, magnetic particle, radiographic,
ultrasonic, and acoustic emission. A listing of these techniques and their characteristics
are presented in Table M.6.
The following discussions on some of these techniques are very brief and con-
densed. More detailed treatments are provided in the end-of-module references.
Visual Inspection
Visual inspection is probably the most common detection technique and the easiest to
conduct; of course, only cracks and defects found on surfaces may be observed visually.
Only relatively large cracks/defects are subject to detection with the unaided eye or a
magnifying glass. For inspection of inaccessible/remote regions, mirrors, fiberscopes,
and borescopes may be used. Fiberscopes and borescopes are optical devices composed
of an eyepiece on the inspection end and a lens on the observation end, which are
linked by either a rigid or flexible tube (normally of an optical fiber bundle and a pro-
tective outer sheath), which acts as the optical relay system, and in some cases is used
to illuminate the remote object. An imaging device (e.g., video camera) may also be
incorporated.
Portable video inspection camera systems are used to inspect the interiors of large
structures (e.g., containers, railroad tank cars, sewer lines) that are inaccessible and/or
hazardous. A video camera (with a zoom lens) is mounted on a pole, cable, or trolley for
deployment into the structure that is to be inspected.
Visual inspection of some confined (and normally horizontal) and long structures
(such as pipelines, air ducts, and reactors) is possible using robotic crawlers. These
devices typically include a sensor (or video camera) mounted on a mobile support car-
riage that is capable of traveling through the system to be inspected. The crawler must
be steerable, capable of both forward and reverse motions, as well as acceleration,
deceleration, and stopping; in addition, it must have the ability to negotiate bends in
piping and to pass through different diameters of pipes. An illumination system is also
provided, and the camera must deliver full-directional viewing as well as have a remote
adjustable focus.
Defect Size
Technique Defect Location Sensitivity (mm) Testing Location
Visual inspection Surface 0.1 Laboratory/in-field
Optical microscopy Surface 0.10.5 Laboratory
Scanning electron Surface 0.001 Laboratory
microscopy (SEM)
Dye penetrant Surface 0.0350.25 Laboratory/in-field
Magnetic particle Surface 0.5 Laboratory/in-field
Radiography Subsurface 2% of Laboratory/in-field
(x-ray/gamma ray) specimen thickness
Ultrasonic Subsurface 0.50 Laboratory/in-field
Acoustic emission Surface/ 0.1 Laboratory/in-field
subsurface
Table M.6
A List of Several
Common
Nondestructive Testing
Techniques
Source: Adapted from
ASM Handbook,
Vol. 19, Fatigue and
Fracture. Reprinted
with permission of
ASMInternational. All
rights reserved. www.
asminternational.org.
M.5 Flaw Detection Using Nondestructive Testing Techniques 23
Optical and Scanning Electron Microscopy Inspections
For detection of small surface cracks (less than about 0.5 mm in size), employment of
optical and/or scanning electron microscopic techniques is necessary. Normally inspec-
tions of these types are conducted in a laboratory setting (as opposed to in the field).
Brief discussions of these two techniques are presented in Section 5.12.
Dye (Liquid) Penetrant Inspection
This common and low-cost technique is used to detect surface cracks in nonporous ma-
terials. In essence, a liquid is used to enhance the visual contrast between a defect and
the bulk solid material. This liquid must have high surface wetting characteristics (i.e., a
low surface tension) and is applied (by spraying, brushing, or dipping) onto the surface
of the part to be inspected. After adequate time has been allowed for this liquid to pen-
etrate (by capillary action) into any surface-breaking defects that are present, excess liq-
uid is removed, and a powder developer is applied that draws the penetrant out of any
defects and to the surface, making possible their observation. Visual inspection is per-
formed using a white light. The penetrant may also be loaded with a fluorescent dye to
increase detection sensitivity; under these circumstances an ultraviolet (or black) light
(in a darkened environment) is used to reveal the defects.
Figure M.13 shows a surface crack that was exposed using this technique.
Magnetic Particle Inspection
A variation of the dye penetrant technique is used to detect both surface and near-surface
defects in ferrous alloys that may be magnetized. For such a material that has been magnet-
ized, in the vicinity of a surface flaw (or discontinuity) there is a distortion of the magnetic flux
(or field) such that the flux leaksfrom (or passes out of) the solid. This inspection technique
utilizes fine iron particles that are suspended in a suitable liquid carrier (e.g., kerosene); these
particles are often coated with a fluorescent dye. The article to be inspected is first magnet-
ized, and when such a suspension is applied (sprayed or painted) onto its surface, iron parti-
cles become attracted to regions where any leakage fields are presenti.e., they cluster in the
vicinity of any surface defects. Visual detection of these clusters (and defects) is possible
under proper lighting conditions (e.g., a fluorescent light and a dark ambient).A surface crack
that was made visible using magnetic particles is presented in Figure M.14.
After observation, demagnetization of the inspected object is possible.
Figure M.13 Photograph showing a surface crack in
an automobile engine connecting rod that was made
visible using a dye penetrant.
(Photograph courtesy of Center for NDE, Iowa State University.)
Figure M.14 Photograph showing a surface crack in
a crane hook that was revealed using the magnetic
particle inspection technique.
The final three nondestructive testing techniques (radiographic, ultrasonic [pulse
echo], and acoustic emission) are utilized to detect subsurface (interior) defects. The first
two techniques (radiographic and ultrasonic testing) employ some type of signal or energy
source (e.g., ultrasonic waves, x-rays) to probe the object to be examined. An interaction
between the incoming signal and a defect or crack causes some type of signal disruption,
and a response in the form of an image or signature that may be sensed and recorded (such
as a photograph or a blip on a screen).
Radiographic Testing
The radiographic testing technique utilizes x-rays or gamma radiation (from a radioac-
tive isotope such as iridium-192 or cobalt-60) as a signal source. This radiation is directed
upon, penetrates, and passes through the object to be inspected. An image is generated
on the remote side by radiation that is transmitted through the object. Photographic film
(that is sensitive to the type of radiation) is most frequently employed as the detection
device to record this image (Figure M.15); fluoroscopic screens and digitized systems
with video monitors may also be used. An image results from variations in the transmit-
ted radiation intensity over the cross-section of the object. Defects and cracks will ap-
pear as part of this image inasmuch as transmission intensity will be different through
regions containing defects than those regions that are defect-free. This technique is com-
monly used to assess the integrity of welds.
It should be noted that health hazards are associated with radiographic testing; ex-
posure to both x-rays and gamma rays must be avoided since both are forms of ionizing
radiation.
Ultrasonic (Pulse-Echo) Inspection
For this inspection technique, the input energy (or signal) is in the form of ultrasonic
wavesi.e., sound waves having high frequencies, normally in the range of 0.1 to 50 MHz.
These waves are emitted from a transducer as intermittent pulses, which are introduced
into and propagate through the object that is being inspected. Normally both transducer
and test object are immersed in a coupling medium (a liquid such as water or oil) so as
to promote the transfer of ultrasonic waves. These waves experience reflection (or
echo) whenever they encounter some type of interface or discontinuity, such as the
back face of the test object or the surface of an interior defect. A reflected wave is
received by the same transducer, which then converts the wave signal into an electrical
signal. Test results are displayed on a screen as reflected signal strength amplitude versus
travel time (i.e., the time between when the pulsed signal was sent and when the
reflected signal was received). A high intensity peak of signal strength represents the
Figure M.15 Photograph generated using x-radiation
that shows an internal defect (dark wedge-shaped region)
in an automobile motor mount casting.
time at which a reflected wave was received. Signal travel time may be converted into
travel distance inasmuch as the velocity of the ultrasonic waves is known. Measurement
of travel time is important in order to distinguish between back-surface and defect
echos, and, in addition, to determine the location (depth) of a defect. A complete inspec-
tion involves passing the transducer probe over the entire region of the test object.
Aerospace, aviation, and automotive industries utilize this NDT technique extensively.
Acoustic Emission
Acoustic emission testing also utilizes ultrasonic waves to detect the presence of cracks
usually in the frequency range between 30 kHz and 1MHz. However, unlike conventional
ultrasonic testing, this technique monitors sound (acoustic) waves that are emitted during
the failure process (i.e., as a crack forms and then propagates) and while a structure is in
servicethat is, no ultrasonic signal is artificially generated and then collected. Associated
with the formation and extension of cracks is the release of elastic strain energy, in the
form of sound waves, that propagate throughout the material and ultimately to its surface
where they may be recorded using some type of sensor (i.e., a transducer). This signal is
converted into an electrical signal and then displayed on a screen for analysis.
One advantage of acoustic emission testing (over other NDT techniques) is that it
monitors failure processes that are dynamic (i.e., crack formation and growth). As such,
instantaneous information relative to the status of and risk of failure is provided. This
technique is frequently used on aircraft. For example, a group of transducers mounted
in a highly stressed area can detect the presence of a crack the instant that it forms, and,
in addition, very precisely determine its origin by measurement of the time it takes for
the signal to reach different transducers.
M.6 FRACTURE TOUGHNESS TESTING
As noted earlier, fracture toughness is defined as a materials resistance to crack propaga-
tion and ultimately to brittle fracture. In Section M.4 we used the symbol K
Ic
to represent the
fracture toughness for the condition of plane strain (i.e., when specimen thickness is greater
than crack length) and also when stress application is such as to promote mode I crack sur-
face displacement (Figure M.6). Inasmuch as K
Ic
is such an important material property with
regard to fracture prevention, it seems reasonable to explore the manner in which it is meas-
ured.A variety of standardized tests have been devised. In the United States, these test meth-
ods are developed by ASTM; for the international marketplace, standards are established by
the International Organization for Standardization (ISO). Most of these techniques are de-
signed for measuring fracture toughness values for metals and their alloys; in addition, some
have also been developed for ceramics, polymers, and composite materials.
In essence, a typical fracture toughness test is conducted on a standard specimen that
contains a preexisting crack. A testing apparatus loads the specimen at a prescribed rate,
and continuously records load magnitude and crack displacement. Resulting data are ana-
lyzed and fracture toughness parameters are determined. These parameters are then sub-
jected to qualification procedures in order to ensure they meet established criteria before
the fracture toughness values are deemed acceptable. We have chosen to describe one of
the earliest and least complicated fracture toughness test standards that was developed:
ASTM Standard E 399-09, Standard Test Method for Linear-Elastic Plane-Strain Fracture
Toughness K
Ic
of Metallic Materials.
First of all, as the title of this standard suggests, the test is used to measure K
Ic
when
the crack-tip region is exposed to a condition of plane-strain upon load application. In ad-
dition, the material being tested should exhibit linear-elastic behaviorthat is, a plot of
load versus crack displacement is linear, and virtually all deformation to the point of frac-
ture is elastic (i.e., the material has limited ductility). Furthermore, it may be recalled
(Section M.4), that the elastic stress field near a crack tip can be described in terms of the
stress intensity factor K; as will be seen we use this parameter in the development of the
methodology for this testing technique.
Two specimen geometries permitted by Standard E 399single-edge notched bend
and compact tensionare represented in Figures M.16a and M.16b, respectively. As
noted in these illustrations, a three-point loading scheme is used for the bend specimen,
whereas the compact specimen is loaded in tension. Specimen size is not specified by this
standard, which must be selected. Test validity is dependent on specimen size, which is
not subject to evaluation until after the conclusion of the test; therefore, unless an ade-
quate size is chosen, the test will need to be repeated. A notch is machined in each spec-
imen, after which a very sharp precrack of length a is introduced at the notch root using
cyclic fatigue-loading. As noted in Figures M.16a and M.16b, initial crack length includes
both notch depth as well as precrack length. Details relating to specimen size selection,
geometrical tolerances, notch configuration, and precracking procedures are contained
in the ASTM standard.
During testing, load is applied at a specified rate and measured using a load cell, which
is one component of the testing apparatus. Furthermore, a clip gage, mounted on the test
specimen across the open end of the notch (Figure M.17), monitors crack displacement.
Results are plotted as load (P) versus displacement (v). The test is continued until fracture,
after which the initial crack length (a) (Figures M.16a and M.16b) is physically measured
on the broken specimen halves. From these data a conditional load P
Q
is measured, from
which a conditional K
Ic
may be determined (and labeled K
Q
); this K
Q
is then evaluated as
to its validity as we explain below.
Three different types of load versus displacement curves have been observed, which
are presented in Figure M.18 (and labeled I, II, and III). The procedure for determina-
tion of this conditional P
Q
value is described as follows: For each curve type, a tangent is
constructed at the initial linear portion of the curve (OA) and its slope is determined. A
straight-line segment having a slope 5% less than this initial tangent is then constructed
from the origin; the intersection of this secant (OP
5
) with the load-displacement curve is
indicated by the point labeled P
5
for each of the curves shown in this plot. If, on a load-
displacement curve, every force point that precedes P
5
is less than P
5
(as is the case for
only curve I in Figure M.18), then P
Q
P
5
. On the other hand, when there is a sharp
drop in load just past the termination of the linear-elastic region, such that maximum
S
B
P
W
Support pin
a
Precrack
Loading pin
B
W
Precrack
P
P
a
Figure M.16 Configuration of (a) single-edge notched bend and (b) compact-tension specimens used for
fracture toughness tests (ASTM Standard E 399-09).
(From V. J. Colangelo and F. A. Heiser, Analysis of Metallurgical Failures, 2nd edition. Copyright 1987 by John Wiley & Sons,
New York. Reprinted with permission of John Wiley & Sons, Inc.)
(a)
(b)
load on the resultant cusp precedes and is greater than P
5
(Figure M.18, curves II and
III), then this maximum load is taken as P
Q
.
At this time it becomes necessary to impose the first validity criterionto determine
whether the specimen is too ductile to be tested using this technique. This criterion is
expressed quantitatively by the expression
(M.32)
where P
max
is the maximum force that the test specimen is able to sustain (Figure M.18).
If this criterion is not satisfied, then another fracture toughness testing technique must
be employed.
10
However, if the criterion specified by Equation M.32 is realized, the next step is to
determine a value for the conditional K
Q
. For the single-edge notched bend specimen,
the following equation is employed:
(M.33)
In this expression (and from Figure M.16a)
P
Q
the conditional load value, determined as described above
S distance between support points
B specimen thickness
W specimen width (or depth)
a precrack length
K
Q
P
Q
S
BW
32
f a
a
W
b
P
max
P
Q
1.10
W
Test fixture
Test
specimen
Displacement
gage
P
O O O
Displacement,
L
o
a
d
,

P
Type I
A P
max
P
5
P
Q
P
max
P
Q
P
5 P
5
P
max
A A
Type II Type III
95% Secant
P
Q
Figure M.17 Schematic diagram showing a
displacement gage that has been installed on a
single-edge notched bend specimen in
preparation for a fracture toughness test.
(Adapted with permission from Figure A2.1 in ASTM
E 399-09 Standard Test Method for Linear-Elastic
Plane-Strain Fracture Toughness K
Ic
of Metallic
Materials. Copyright ASTM International, 100 Barr
Harbor Drive, West Conshohocken, PA 19428. A copy
of this standard may be obtained from ASTM
International, www.astm.org.)
Figure M.18 Three principal types of load versus
displacement curves that may be generated during a fracture
toughness test. (ASTM Standard E 399-09).
(Adapted with permission from Figure 7 in ASTM E 399-09 Standard
Test Method for Linear-Elastic Plane-Strain Fracture Toughness K
Ic
of Metallic Materials. Copyright ASTM International, 100 Barr
Harbor Drive, West Conshohocken, PA 19428. A copy of this
standard may be obtained from ASTM International, www.astm.org.)
10
For example, ASTM Standard E 1820.
K
Ic
validity
criterionmaximum
degree of ductility
In Equation M.33, is a calibration function that depends on the a/Wratio as
(M.34)
Similarly, the following equation is used to compute K
Q
for the compact-tension
specimen configuration:
(M.35)
in which B and W are the specimen thickness and width (depth), respectively, and a is
the precrack length (Figure M.16b). In this case
(M.36)
Again, K
Q
is conditional, and before it can be accepted as a valid K
Ic
value, verifica-
tion of a condition of plane-strain must be established. Such is possible when the follow-
ing criterion is satisfied (for both specimen geometries)
11
:
(M.37)
Here a and B are, respectively, crack length and specimen thickness, and s
y
is the 0.2%
offset yield strength (measured in tension).
By way of summary:
When the criteria specified by Equations M.32 and M.37 are met, then K
Q
is a
valid value for K
Ic
, and may be reported as such.
If the condition of Equation M.32 is not satisfied, then another testing technique
must be employed.
And, finally, when the criterion of Equation M.32 is met, while at the same time
Equation M.37 is not realized, then the test must be repeated using a thicker
specimen. A new specimen thickness may be estimated by incorporating the
measured value of K
Q
into Equation M.37.
a and B 2.5 a
K
Q
s
y
b
2
a 2
a
W
b c 0.866 4.64 a
a
W
b 13.32 a
a
W
b
2
14.72 a
a
W
b
3
5.6 a
a
W
b
4
d
a 1
a
W
b
32
f a
a
W
b
K
Q
P
Q
B2W
f a
a
W
b
3
B
a
W

1.99 a
a
W
b a 1
a
W
b c 2.15 3.93 a
a
W
b 2.7 a
a
W
b
2
d
2c 1 2 a
a
W
b d c 1 a
a
W
b d
3
2

f a
a
W
b
f a
a
W
b
K
Ic
validity
criterionminimum
crack length and
minimum specimen
thickness
11
Note the similarity between Equations M.23 and M.37, the former of which was cited earlier as a minimum
thickness for the condition of plane strain.
M.7 IMPACT FRACTURE TESTING
12
12
This section is virtually identical to Section 9.8.
13
ASTM Standard E 23, Standard Test Methods for Notched Bar Impact Testing of Metallic Materials.
Prior to the advent of fracture mechanics as a scientific discipline, impact testing techniques
were established to ascertain the fracture characteristics of materials at high loading rates.
It was realized that the results of laboratory tensile tests (at low loading rates) could not be
extrapolated to predict fracture behavior. For example, under some circumstances normally
ductile metals fracture abruptly and with very little plastic deformation under high loading
rates. Impact test conditions were chosen to represent those most severe relative to the po-
tential for fracturenamely, (1) deformation at a relatively low temperature, (2) a high
strain rate (i.e., rate of deformation), and (3) a triaxial stress state (which may be introduced
by the presence of a notch).
Impact Testing Techniques
Two standardized tests,
13
the Charpy and the Izod, are used to measure the impact
energy (sometimes also termed notch toughness). The Charpy V-notch (CVN) technique
is most commonly used in the United States. For both the Charpy and the Izod, the spec-
imen is in the shape of a bar of square cross section, into which a V-notch is machined
(Figure M.19a). The apparatus for making V-notch impact tests is illustrated schemati-
cally in Figure M.19b. The load is applied as an impact blow from a weighted pendulum
hammer that is released from a cocked position at a fixed height h. The specimen is
positioned at the base as shown. Upon release, a knife edge mounted on the pendulum
strikes and fractures the specimen at the notch, which acts as a point of stress concentra-
tion for this high-velocity impact blow. The pendulum continues its swing, rising to a
maximum height which is lower than h. The energy absorption, computed from the
difference between h and is a measure of the impact energy. The primary difference
between the Charpy and Izod techniques lies in the manner of specimen support, as
illustrated in Figure M.19b. These are termed impact tests because of the manner of load
application. Variables including specimen size and shape as well as notch configuration
and depth influence the test results.
Both plane strain fracture toughness and these impact tests determine the fracture
properties of materials. The former are quantitative in nature, in that a specific property
of the material is determined (i.e., K
Ic
). The results of the impact tests, on the other hand,
are more qualitative and are of little use for design purposes. Impact energies are of
interest mainly in a relative sense and for making comparisonsabsolute values are of
little significance. Attempts have been made to correlate plane strain fracture toughnesses
and CVN energies, with only limited success. Plane strain fracture toughness tests are
not as simple to perform as impact tests; furthermore, equipment and specimens are
more expensive.
Ductile-to-Brittle Transition
One of the primary functions of the Charpy and the Izod tests is to determine whether
a material experiences a ductile-to-brittle transition with decreasing temperature and, if
so, the range of temperatures over which it occurs. As may be noted in the chapter-opening
photograph (of the oil barge) for Chapter 9, widely used steels can exhibit this ductile-
to-brittle transition with disastrous consequences. The ductile-to-brittle transition is
related to the temperature dependence of the measured impact energy absorption. This
transition is represented for a steel by curve A in Figure M.20. At higher temperatures
the CVN energy is relatively large, corresponding to a ductile mode of fracture. As the
h,
h,
Charpy test, Izod test
impact energy
ductile-to-brittle
transition
temperature is lowered, the impact energy drops suddenly over a relatively narrow tem-
perature range, below which the energy has a constant but small value; that is, the mode
of fracture is brittle.
Alternatively, appearance of the failure surface is indicative of the nature of frac-
ture and may be used in transition temperature determinations. For ductile fracture, this
surface appears fibrous or dull (or of shear character) as in the steel specimen of Fig-
ure M.21 which was tested at 79C. Conversely, totally brittle surfaces have a granular
10 mm
(0.39 in.)
Izod
Scale
Charpy
Starting position
Pointer
End of swing
Specimen
Anvil
8 mm
(0.32 in.)
10 mm
(0.39 in.)
Notch
h'
Hammer
h
Figure M.19 (a)
Specimen used for
Charpy and Izod
impact tests. (b) A
schematic drawing of
an impact testing
apparatus. The
hammer is released
from fixed height h
and strikes the
specimen; the energy
expended in fracture
is reflected in the
difference between h
and the swing height
Specimen
placements for both
Charpy and Izod
tests are also shown.
[Figure (b) adapted
from H. W. Hayden,
W. G. Moffatt, and J.
Wulff, The Structure and
Properties of Materials,
Vol. III, Mechanical
Behavior, p. 13.
Copyright 1965 by
John Wiley & Sons,
New York. Reprinted
by permission of John
Wiley & Sons, Inc.]
h.
(a)
(b)
(shiny) texture (or cleavage character) (the 59C specimen in Figure M.21). Over the
ductile-to-brittle transition, features of both types will exist (in Figure M.21, displayed
by specimens tested at 12C, 4C, 16C, and 24C). Frequently, the percent shear frac-
ture is plotted as a function of temperaturecurve B in Figure M.20.
For many alloys there is a range of temperatures over which the ductile-to-
brittle transition occurs (Figure M.20); this presents some difficulty in specifying a single
ductile-to-brittle transition temperature. No explicit criterion has been established, and so
this temperature is often defined as the temperature at which the CVN energy assumes
some value (e.g., 20 J or 15 ft-lb
f
), or corresponding to some given fracture appearance
(e.g., 50% fibrous fracture). Matters are further complicated by the fact that a different
transition temperature may be realized for each of these criteria. Perhaps the most conser-
vative transition temperature is that at which the fracture surface becomes 100% fibrous;
on this basis, the transition temperature is approximately 110C (230F) for the steel alloy
that is shown in Figure M.20.
Structures constructed from alloys that exhibit this ductile-to-brittle behavior
should be used only at temperatures above the transition temperature to avoid brittle
and catastrophic failure. Classic examples of this type of failure occurred with disastrous
Figure M.20 Temperature
dependence of the Charpy V-notch
impact energy (curve A) and percent
shear fracture (curve B) for an A283
steel.
(Reprinted from Welding Journal. Used by
permission of the American Welding
Society.)
I
m
p
a
c
t

e
n
e
r
g
y

(
J
)
100
80
60
40
20
0
100
80
60
40
20
0
40 0 40 80 120 160 200 240 280
40 20 0 40 20 80 120 60 100 140
Temperature (F)
Temperature (C)
S
h
e
a
r

f
r
a
c
t
u
r
e

(
%
)
Impact
energy
Shear
fracture
A
B
Figure M.21 Photograph of
fracture surfaces of A36 steel Charpy
V-notch specimens tested at indicated
temperatures (in C).
(From R. W. Hertzberg, Deformation and
Fracture Mechanics of Engineering
Materials, 3rd edition, Fig. 9.6, p. 329.
Copyright 1989 by John Wiley & Sons,
Inc., New York. Reprinted by permission
of John Wiley & Sons, Inc.)
59 12 4 16 24 79
consequences during World War II when a number of welded transport ships away from
combat suddenly split in half. The vessels were constructed of a steel alloy that possessed
adequate ductility according to room-temperature tensile tests. The brittle fractures
occurred at relatively low ambient temperatures, at about 4C (40F), in the vicinity of
the transition temperature of the alloy. Each fracture crack originated at some point of
stress concentration, probably a sharp corner or fabrication defect, and then propagated
around the entire girth of the ship.
In addition to the ductile-to-brittle transition represented in Figure M.20, two other
general types of impact energy-versus-temperature behavior have been observed; these
are represented schematically by the upper and lower curves of Figure M.22. Here it
may be noted that low-strength FCC metals (some aluminum and copper alloys) and
most HCP metals do not experience a ductile-to-brittle transition (corresponding to the
upper curve of Figure M.22) and retain high impact energies (i.e., remain ductile) with
decreasing temperature. For high-strength materials (e.g., high-strength steels and tita-
nium alloys), the impact energy is also relatively insensitive to temperature (the lower
curve of Figure M.22); however, these materials are also very brittle, as reflected by their
low impact energy values. The characteristic ductile-to-brittle transition is represented
by the middle curve of Figure M.22. As noted, this behavior is typically found in low-
strength steels that have the BCC crystal structure.
For these low-strength steels, the transition temperature is sensitive to both alloy
composition and microstructure. For example, decreasing the average grain size results
in a lowering of the transition temperature. Hence, refining the grain size both stren-
gthens (Section 8.9) and toughens steels. In contrast, increasing the carbon content,
although it increases the strength of steels, also raises their CVN transition, as indicated
in Figure M.23.
Izod or Charpy tests are also conducted to assess the impact strength of polymeric
materials. As with metals, polymers may exhibit ductile or brittle fracture under impact
loading conditions, depending on the temperature, specimen size, strain rate, and mode
of loading, as discussed in the preceding section. Both semicrystalline and amorphous
polymers are brittle at low temperatures and both have relatively low impact strengths.
However, they experience a ductile-to-brittle transition over a relatively narrow temper-
ature range, similar to that shown for a steel in Figure M.20. Of course, impact strength
undergoes a gradual decrease at still higher temperatures as the polymer begins to
soften. Typically, the two impact characteristics most sought after are a high impact
strength at the ambient temperature and a ductile-to-brittle transition temperature that
lies below room temperature.
Most ceramics also experience a ductile-to-brittle transition, which occurs only at
elevated temperatures, ordinarily in excess of 1000C (1850F).
I
m
p
a
c
t

e
n
e
r
g
y
Low-strength (FCC and HCP) metals
Low-strength steels (BCC)
High-strength materials
Temperature
Figure M.22 Schematic curves
for the three general types of
impact energy-versus-temperature
behavior.
M.8 Cyclic Stresses 33
Fatigue
Temperature (C)
Temperature (F)
I
m
p
a
c
t

e
n
e
r
g
y

(
J
)
I
m
p
a
c
t

e
n
e
r
g
y

(
f
t
-
l
b
f
)
200 100
200 0 200 400
0 100 200
100
0
40
80
120
160
0.01 0.11
0.22
0.31
0.43
0.53
0.63
0.67
200
240
0
200
300
Figure M.23 Influence of carbon
content on the Charpy V-notch energy-
versus-temperature behavior for steel.
(Reprinted with permission from ASM
International, Metals Park, OH 44073-9989,
USA; J. A. Reinbolt and W. J. Harris, Jr.,
Effect of Alloying Elements on Notch
Toughness of Pearlitic Steels, Transactions of
ASM, Vol. 43, 1951.)
Fatigue is a form of failure that occurs in structures subjected to dynamic and fluctuat-
ing stresses (e.g., bridges, aircraft, and machine components). Under these circumstances
it is possible for failure to occur at a stress level considerably lower than the tensile or
yield strength for a static load. The term fatigue is used because this type of failure normally
occurs after a lengthy period of repeated stress or strain cycling. Fatigue is important
inasmuch as it is the single largest cause of failure in metals, estimated to be involved in
approximately 90% of all metallic failures; polymers and ceramics (except for glasses)
are also susceptible to this type of failure. Furthermore, fatigue failure is catastrophic
and insidious, occurring very suddenly and without warning.
Fatigue failure is brittlelike in nature even in normally ductile metals, in that there
is very little, if any, gross plastic deformation associated with failure. The process occurs
by the initiation and propagation of cracks, and typically the fracture surface is perpen-
dicular to the direction of an applied tensile stress.
M.8 CYCLIC STRESSES
14
The applied stress may be axial (tension-compression), flexural (bending), or torsional
(twisting) in nature. In general, three different fluctuating stress-time modes are possi-
ble. One is represented schematically by a regular and sinusoidal time dependence in
Figure M.24a, where the amplitude is symmetrical about a mean zero stress level, for
example, alternating from a maximum tensile stress (s
max
) to a minimum compressive
stress (s
min
) of equal magnitude; this is referred to as a reversed stress cycle. Another
type, termed a repeated stress cycle, is illustrated in Figure M.24b; the maxima and min-
ima are asymmetrical relative to the zero stress level. Finally, the stress level may vary
randomly in amplitude and frequency, as exemplified in Figure M.24c.
14
fatigue
Also indicated in Figure M.24b are several parameters used to characterize the fluc-
tuating stress cycle. The stress amplitude alternates about a mean stress s
m
, defined as
the average of the maximum and minimum stresses in the cycle, or
(M.38)
The range of stress s
r
is the difference between s
max
and s
min
, namely,
(M.39)
Stress amplitude s
a
is one-half of this range of stress, or
(M.40) s
a
s
r
2

s
max
s
min
2
s
r
s
max
s
min
s
m
s
max
s
min
2
0
min
max
Time
+
S
t
r
e
s
s
T
e
n
s
i
o
n
C
o
m
p
r
e
s
s
i
o
n
0
min
max
Time
+
S
t
r
e
s
s
T
e
n
s
i
o
n
C
o
m
p
r
e
s
s
i
o
n
r
Time
+
S
t
r
e
s
s
T
e
n
s
i
o
n
C
o
m
p
r
e
s
s
i
o
n
Figure M.24 Variation of stress with time that
accounts for fatigue failures. (a) Reversed stress
cycle, in which the stress alternates from a
maximum tensile stress (+) to a maximum
compressive stress of equal magnitude.
(b) Repeated stress cycle, in which maximum and
minimum stresses are asymmetrical relative to
the zero stress level; mean stress range
of stress and stress amplitude are indicated.
(c) Random stress cycle.
s
a
s
r
,
s
m
,
()
Mean stress for cyclic
loadingdependence
on maximum and
minimum stress levels
Computation of
range of stress for
cyclic loading
Computation of
stress amplitude for
cyclic loading
(a)
(b)
(c)
M.9 The S-N Curve 35
Finally, the stress ratio R is the ratio of minimum and maximum stress amplitudes:
(M.41)
By convention, tensile stresses are positive and compressive stresses are negative.
For example, for the reversed stress cycle, the value of R is 1.
R
s
min
s
max
Computation of
stress ratio
Bearing housing
Bearing housing
Load
Specimen
Load
Flexible coupling
High-speed
motor
Counter
+
Figure M.25 Schematic diagram of a fatigue-testing apparatus for making rotating-bending tests.
(From KEYSER, MATERIALS SCIENCE IN ENGINEERING, 4th, 1986. Printed and Electronically reproduced by
permission of Pearson Education, Inc., Upper Saddle River, New Jersey.)
M.9 THE S-N CURVE
15
As with other mechanical characteristics, the fatigue properties of materials can be
determined from laboratory simulation tests.
16
A test apparatus should be designed to
duplicate as nearly as possible the service stress conditions (stress level, time frequency,
stress pattern, etc.). A schematic diagram of a rotating-bending test apparatus commonly
used for fatigue testing is shown in Figure M.25; the compression and tensile stresses are
imposed on the specimen as it is simultaneously bent and rotated. Tests are also fre-
quently conducted using an alternating uniaxial tension-compression stress cycle.
A series of tests is commenced by subjecting a specimen to the stress cycling at a rel-
atively large maximum stress amplitude (s
max
), usually on the order of two thirds of the
static tensile strength; the number of cycles to failure is counted. This procedure is re-
peated on other specimens at progressively decreasing maximum stress amplitudes. Data
are plotted as stress S versus the logarithm of the number N of cycles to failure for each
of the specimens. The values of S are normally taken as stress amplitudes (s
a
, Equation
M.40); on occasion, s
max
or s
min
values may be used.
Two distinct types of SN behavior are observed, which are represented schemati-
cally in Figure M.26. As these plots indicate, the higher the magnitude of the stress, the
smaller is the number of cycles the material is capable of sustaining before failure. For
some ferrous (iron base) and titanium alloys, the SN curve (Figure M.26a) becomes
horizontal at higher N values; there is a limiting stress level, called the fatigue limit (also
sometimes the endurance limit), below which fatigue failure will not occur. This fatigue
limit represents the largest value of fluctuating stress that will not cause failure for
essentially an infinite number of cycles. For many steels, fatigue limits range between
35% and 60% of the tensile strength.
15
16
See ASTM Standard E 466, Standard Practice for Conducting Force Controlled Constant Amplitude Axial
Fatigue Tests of Metallic Materials, and ASTM Standard E 468, Standard Practice for Presentation of Constant
Amplitude Fatigue Test Results for Metallic Materials.
fatigue limit
Most nonferrous alloys (e.g., aluminum, copper, magnesium) do not have a fatigue
limit, in that the SNcurve continues its downward trend at increasingly greater Nvalues
(Figure M.26b). Thus, fatigue will ultimately occur regardless of the magnitude of the
stress. For these materials, one fatigue response is specified as fatigue strength, which is
defined as the stress level at which failure will occur for some specified number of
cycles (e.g., 10
7
cycles). The determination of fatigue strength is also demonstrated in
Figure M.26b.
Another important parameter that characterizes a materials fatigue behavior is
fatigue life N
f
. It is the number of cycles to cause failure at a specified stress level, as
taken from the SN plot (Figure M.26b).
Unfortunately, there always exists considerable scatter in fatigue datathat is, a
variation in the measured N value for a number of specimens tested at the same stress
level. This variation may lead to significant design uncertainties when fatigue life and/or
fatigue limit (or strength) are being considered. The scatter in results is a consequence
of the fatigue sensitivity to a number of test and material parameters that are impossible
to control precisely. These parameters include specimen fabrication and surface prepa-
ration, metallurgical variables, specimen alignment in the apparatus, mean stress, and
test frequency.
Cycles to failure, N
(logarithmic scale)
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
,

S
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
Fatigue
limit
(logarithmic scale)
Fatigue life
at stress S
1
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
,

S
10
3
10
4
10
7
N
1
10
8
10
9
10
10
Fatigue strength
at N
1
cycles
S
1
(a)
(b)
fatigue strength
fatigue life
Figure M.26 Stress amplitude (S) versus
logarithm of the number of cycles to fatigue
failure (N) for (a) a material that displays a
fatigue limit, and (b) a material that does
not display a fatigue limit.
Fatigue SNcurves similar to those shown in Figure M.26 represent best-fit curves
that have been drawn through average-value data points. It is a little unsettling to real-
ize that approximately one-half of the specimens tested actually failed at stress levels
lying nearly 25% below the curve (as determined on the basis of statistical treatments).
Several statistical techniques have been developed to specify fatigue life and fatigue
limit in terms of probabilities. One convenient way of representing data treated in this
manner is with a series of constant probability curves, several of which are plotted in
Figure M.27. The P value associated with each curve represents the probability of fail-
ure. For example, at a stress of 200 MPa (30,000 psi), we would expect 1% of the speci-
mens to fail at about 10
6
cycles, 50% to fail at about 2 10
7
cycles, and so on. Remember
that SN curves represented in the literature are normally average values, unless noted
otherwise.
The fatigue behaviors represented in Figures M.26a and M.26b may be classified
into two domains. One is associated with relatively high loads that produce not only elas-
tic strain but also some plastic strain during each cycle. Consequently, fatigue lives are
relatively short, this domain is termed low-cycle fatigue and occurs at less than about 10
4
to 10
5
cycles. For lower stress levels where deformations are totally elastic, longer lives
result. This is called high-cycle fatigue because relatively large numbers of cycles are
required to produce fatigue failure. High-cycle fatigue is associated with fatigue lives
greater than about 10
4
to 10
5
cycles.
70
60
(logarithmic scale)
40
50
30
20
10
400
300
200
100
S
t
r
e
s
s

(
1
0
3

p
s
i
)
S
t
r
e
s
s
,

S

(
M
P
a
)
P = 0.99
P = 0.90
P = 0.50
P = 0.01
P = 0.10
10
4
10
5
10
6
10
7
10
8
10
9
Figure M.27 Fatigue SN probability
of failure curves for a 7075-T6 aluminum
alloy; P denotes the probability of failure.
(From G. M. Sinclair and T. J. Dolan, Trans.
ASME, 75, 1953, p. 867. Reprinted with
permission of the American Society of
Mechanical Engineers.)
M.10 CRACK INITIATION AND PROPAGATION
17
The process of fatigue failure is characterized by three distinct steps: (1) crack initiation,
in which a small crack forms at some point of high stress concentration; (2) crack prop-
agation, during which this crack advances incrementally with each stress cycle; and
(3) final failure, which occurs very rapidly once the advancing crack has reached a criti-
cal size. The fatigue life N
f
, the total number of cycles to failure, therefore can be taken
as the sum of the number of cycles for crack initiation N
i
and crack propagation N
p
:
(M.42)
The contribution of the final failure step to the total fatigue life is insignificant since
it occurs so rapidly. Relative proportions to the total life of N
i
and N
p
depend on the
N
f
N
i
N
p
17
This section is an expanded and more detailed version of Section 9.12.
particular material and test conditions. At low stress levels (i.e., for high-cycle fatigue),
a large fraction of the fatigue life is utilized in crack initiation. With increasing stress
level, N
i
decreases and the cracks form more rapidly. Thus, for low-cycle fatigue (high
stress levels), the propagation step predominates (i.e., N
p
> N
i
).
Cracks associated with fatigue failure almost always initiate (or nucleate) on the
surface of a component at some point of stress concentration. Crack nucleation sites in-
clude surface scratches, sharp fillets, keyways, threads, dents, and the like. In addition,
cyclic loading can produce microscopic surface discontinuities resulting from dislocation
slip steps which may also act as stress raisers, and therefore as crack initiation sites.
Once a stable crack has nucleated, it then initially propagates very slowly and, in
polycrystalline metals, along crystallographic planes of high shear stress; this is some-
times termed stage I propagation (Figure M.28). This stage may constitute a large or
small fraction of the total fatigue life depending on stress level and the nature of the test
specimen; high stresses and the presence of notches favor a short-lived stage I. In poly-
crystalline metals, cracks normally extend through only several grains during this prop-
agation stage. The fatigue surface that is formed during stage I propagation has a flat and
featureless appearance.
Eventually, a second propagation stage (stage II) takes over, wherein the crack ex-
tension rate increases dramatically. Furthermore, at this point there is also a change in
propagation direction to one that is roughly perpendicular to the applied tensile stress
(see Figure M.28). During this stage of propagation, crack growth proceeds by a repeti-
tive plastic blunting and sharpening process at the crack tip, a mechanism illustrated in
Figure M.29. At the beginning of the stress cycle (zero or maximum compressive load),
the crack tip has the shape of a sharp double-notch (Figure M.29a). As the tensile stress
is applied (Figure M.29b), localized deformation occurs at each of these tip notches
along slip planes that are oriented at 45 angles relative to the plane of the crack. With
increased crack widening, the tip advances by continued shear deformation and the
Stage II
Stage I
s
s
representation showing stages I
and II of fatigue crack
propagation in polycrystalline
metals.
(Copyright ASTM. Reprinted with
permission.)
Figure M.29 Fatigue crack propagation mechanism (stage II) by
repetitive crack tip plastic blunting and sharpening; (a) zero or
maximum compressive load, (b) small tensile load, (c) maximum
tensile load, (d) small compressive load, (e) zero or maximum
compressive load, (f ) small tensile load. The loading axis is vertical.
(Copyright ASTM. Reprinted with permission.)
(a)
(d)
(e)
(f )
(b)
(c)
assumption of a blunted configuration (Figure M.29c). During compression, the direc-
tions of shear deformation at the crack tip are reversed (Figure M.29d) until, at the cul-
mination of the cycle, a new sharp double-notch tip has formed (Figure M.29e). Thus, the
crack tip has advanced a one-notch distance during the course of a complete cycle. This
process is repeated with each subsequent cycle until eventually some critical crack dimen-
sion is achieved that precipitates the final failure step and catastrophic failure ensues.
The region of a fracture surface that formed during stage II propagation may be
characterized by two types of markings termed beachmarks and striations. Both of
these features indicate the position of the crack tip at some point in time and appear
as concentric ridges that expand away from the crack initiation site(s), frequently in
a circular or semicircular pattern. Beachmarks (sometimes also called clamshell
marks) are of macroscopic dimensions (Figure M.30) and may be observed with the
unaided eye. These markings are found for components that experienced interrup-
tions during stage II propagationfor example, a machine that operated only during
normal work-shift hours. Each beachmark band represents a period of time over
which crack growth occurred.
On the other hand, fatigue striations are microscopic in size and subject to observation
with the electron microscope (either TEM or SEM). Figure M.31 is an electron fracto-
graph which shows this feature. Each striation is thought to represent the advance distance
of the crack front during a single load cycle. Striation width depends on, and increases
with, increasing stress range.
It should be emphasized that although both beachmarks and striations are fatigue
fracture surface features having similar appearances, they are nevertheless different in
both origin and size. There may be thousands of striations within a single beachmark.
Often the cause of failure may be deduced after examination of the failure surfaces.
The presence of beachmarks and/or striations on a fracture surface confirms that the
cause of failure was fatigue. Nevertheless, the absence of either or both does not exclude
fatigue as the cause of failure. Striations are not observed for all metals that experience
fatigue. Furthermore, the likelihood of the appearance of striations may depend on
stress state. Striation detectability decreases with the passage of time because of the for-
mation of surface corrosion products and/or oxide films. Also, during stress cycling,
Figure M.30 Fracture surface of a rotating steel
shaft that experienced fatigue failure. Beachmark
ridges are visible in the photograph.
(Reproduced with permission from D. J. Wulpi,
Understanding How Components Fail, American Society for
Metals, Materials Park, OH, 1985.)
striations may be destroyed by abrasive action as crack mating surfaces rub against one
another.
One final comment regarding fatigue failure surfaces: Beachmarks and striations will
not appear on the region over which the rapid failure occurs (which region is noted in
Figure M.32). Rather, the rapid failure may be either ductile or brittle; evidence of plastic
deformation will be present for ductile, and absent for brittle, failure.
1 m
Figure M.31 Transmission electron fractograph
showing fatigue striations in aluminum. 9000.
(From V. J. Colangelo and F. A. Heiser, Analysis of
Metallurgical Failures, 2nd edition. Copyright 1987 by
John Wiley & Sons, New York. Reprinted by permission of
John Wiley & Sons, Inc.)
Region of slow
crack propagation
Region of rapid failure
2 cm
Figure M.32 Fatigue failure surface. A crack
formed at the top edge. The smooth region also
near the top corresponds to the area over which
the crack propagated slowly. Rapid failure
occurred over the area having a dull and fibrous
texture (the largest area). Approximately 0.5.
(Reproduced by permission from Metals Handbook:
Fractography and Atlas of Fractographs, Vol. 9, 8th
edition, H. E. Boyer, Editor, American Society for
Metals, 1974.)
M.11 CRACK PROPAGATION RATE
Even though measures may be taken to minimize the possibility of fatigue failure, cracks
and crack nucleation sites will always exist in structural components. Under the influence
of cyclic stresses, cracks will inevitably form and grow; this process, if unabated, can
ultimately lead to failure. The intent of the present discussion is to develop a criterion
whereby fatigue life may be predicted on the basis of material and stress state parame-
ters. Principles of fracture mechanics (Section M.4) will be employed inasmuch as the
treatment involves determination of a maximum crack length that may be tolerated
without inducing failure. It should be noted that this discussion relates to the domain of
high-cycle fatigue, that is, for fatigue lives greater than about 10
4
to 10
5
cycles.
Results of fatigue studies have shown that the life of a structural component may be
related to the rate of crack growth. During stage II propagation, cracks may grow from
a barely perceivable size to some critical length. Experimental techniques are available
which are employed to monitor crack length during the cyclic stressing. Data are
recorded and then plotted as crack length a versus the number of cycles N. A typical plot
is shown in Figure M.33, where curves are included from data generated at two differ-
ent stress levels; the initial crack length a
0
for both sets of tests is the same. Crack growth
rate da/dN is taken as the slope at some point of the curve. Two important results are
worth noting: (1) initially, growth rate is small, but increases with increasing crack length;
and (2) growth rate is enhanced with increasing applied stress level and for a specific
crack length (a
1
in Figure M.33).
Fatigue crack propagation rate during stage II is a function of not only stress level
and crack size but also material variables. Mathematically, this rate may be expressed in
terms of the stress intensity factor K(developed using fracture mechanics in Section M.4)
and takes the form
(M.43)
The parameters A and m are constants for the particular material, which will also de-
pend on environment, frequency, and the stress ratio (R in Equation M.41). The value of
m normally ranges between 1 and 6.
Furthermore, K is the stress intensity factor range at the crack tip, that is,
(M.44a)
or, from Equation M.19,
(M.44b) K Ys1pa Y1s
max
s
min
2 1pa
K K
max
K
min
da
dN
A1 K2
m
Cycles N
C
r
a
c
k

l
e
n
g
t
h

a
a
0
a
1
s
2
> s
1
s
2
s
1
da
dN

a
1
,

s
2
da
dN

a
1
,

s
1
Figure M.33 Crack length
versus the number of cycles at
stress levels s
1
and s
2
for
fatigue studies. Crack growth
rate da/dN is indicated at crack
length a
1
for both stress levels.
Dependence of stage
II crack propagation
rate on stress
intensity factor range
at a crack tip
Since crack growth stops or is negligible for a compression portion of the stress cycle, if
s
min
is compressive, then K
min
and s
min
are taken to be zero; that is, K K
max
and s
s
max
. Also note that K
max
and K
min
in Equation M.44a represent stress intensity factors,
not the fracture toughness K
c
nor the plane strain fracture toughness K
Ic
.
The typical fatigue crack growth rate behavior of materials is represented schemat-
ically in Figure M.34 as the logarithm of crack growth rate da/dN versus the logarithm
of the stress intensity factor range K. The resulting curve has a sigmoidal shape which
may be divided into three distinct regions, labeled I, II, and III. In region I (at low stress
levels and/or small crack sizes), preexisting cracks will not grow with cyclic loading.
Furthermore, associated with region III is accelerated crack growth, which occurs just
prior to the rapid fracture.
The curve is essentially linear in region II, which is consistent with Equation M.43.
This may be confirmed by taking the logarithm of both sides of this expression, which
leads to
(M.45a)
(M.45b)
Indeed, according to Equation M.45b, a straight-line segment will result when log (da/dN)-
versus-log K data are plotted; the slope and intercept correspond to the values of m
log a
da
dN
b m log K log A
log a
da
dN
b log 3 A1 K2
m
4
Stress intensity factor range, K (log scale)
F
a
t
i
g
u
e

c
r
a
c
k

g
r
o
w
t
h

r
a
t
e
,

(
l
o
g

s
c
a
l
e
)
d
a
d
N
da
dN
A(K)
m
Region I
Non-
propagating
fatigue
cracks
Region III
Unstable
crack
growth
Region II
Linear relationship
between
log K and log
da
dN
representation of logarithm
fatigue crack propagation rate
da/dN versus logarithm stress
intensity factor range DK. The
three regions of different crack
growth response (I, II, and III)
are indicated.
(Reprinted with permission from ASM
International, Metals Park, OH 44073-
9989. W. G. Clark, Jr., How Fatigue
Crack Initiation and Growth Properties
Affect Material Selection and Design
Criteria, Metals Engineering Quarterly,
Vol. 14, No. 3, 1974.)
and log A, respectively, which may be determined from test data that have been repre-
sented in the manner of Figure M.34. Figure M.35 is one such plot for a Ni-Mo-V steel
alloy. The linearity of the data may be noted, which verifies the power law relationship
of Equation M.43.
One of the goals of failure analysis is to be able to predict fatigue life for some com-
ponent, given its service constraints and laboratory test data. We are now able to develop
an analytical expression for N
f
, due to stage II, by integration of Equation M.43.
Rearrangement is first necessary as follows:
(M.46) dN
da
A1 K2
m
10
6
10
4
10
3
10
2
10
5
10
4
10
3
10 20
20
4000
5000
6000
7000
8000
9200
9900
40 60 80 100
40 60 80 100
Stress intensity factor range, K
(10
3
psi in.)
MPa m
C
r
a
c
k

g
r
o
w
t
h

r
a
t
e
,

(
i
n
.
/
c
y
c
l
e
)
d
a
d
N
C
r
a
c
k

g
r
o
w
t
h

r
a
t
e
,

(
m
m
/
c
y
c
l
e
)
d
a
d
N
0.2% yield strength = 84,500 psi
Test temp. = 24C (75F)
Test frequency = 1800 cpm (30 Hz)
Max. cyclic load, lb
f
Figure M.35 Plot of
logarithm crack growth
rate versus logarithm
range for a Ni-Mo-V
steel. These data provide
experimental verification
for the linearity of region
II of Figure M.34, and, in
addition, for Equation
M.45b [i.e., a straight-line
segment when log (da/dN)
is plotted versus log K].
(Reprinted by permission of
the Society for Experimental
Mechanics, Inc.)
which may be integrated as
(M.47)
The limits on the second integral are between the initial flaw length a
0
, which may be
measured using nondestructive examination techniques, and the critical crack length a
c
determined from fracture toughness tests.
Substitution of the expression for K (Equation M.44b) leads to
(M.48)
Here it is assumed that s (or s
max
s
min
) is constant; furthermore, in general Y will
depend on crack length a and therefore cannot be removed from within the integral.
A word of caution: Equation M.48 presumes the validity of Equation M.43 over the en-
tire life of the component; it ignores the time taken to initiate the crack and also for final
failure. Therefore, this expression should only be taken as an estimate of N
f
.

1
Ap
m/2
1 s2
m
a
c
a
0
da
Y
m
a
m/2
N
f
a
c
a
0
da
A1Ys1pa2
m
N
f
N
f
0
dN
a
c
a
0
da
A1 K2
m
DESIGN EXAMPLE M.2
Fatigue Life Prediction
A relatively large sheet of steel is to be exposed to cyclic tensile and compressive stresses of
magnitudes 100 MPa and 50 MPa, respectively. Prior to testing, it has been determined that the
length of the largest surface crack is 2.0 mm (2 10
3
m). Estimate the fatigue life of this sheet
if its plane strain fracture toughness is and the values of m and A in Equation M.43
are 3.0 and 1.0 10
12
, respectively, for s in MPa and a in m. Assume that the parameter Y
is independent of crack length and has a value of 1.0.
Solution
It first becomes necessary to compute the critical crack length a
c
, the integration upper limit
in Equation M.48. Equation M.25 is employed for this computation, assuming a stress level of
100 MPa, since this is the maximum tensile stress. Therefore,
We now want to solve Equation M.48 using 0.002 m as the lower integration limit a
0
, as stipu-
lated in the problem. The value of s is just 100 MPa, the magnitude of the tensile stress, since
s
min
is compressive. Therefore, integration yields
N
f
1
Ap
m/2
1 s2
m
a
c
a
0
da
Y
m
a
m/2

1
p
c
25 MPa1m
1100 MPa2 112
d
2
0.02 m
a
c
1
p
a
K
Ic
sY
b
2
25 MPa1m
Computation of
predicted fatigue life
M.12 Factors That Affect Fatigue Life 45
5.49 10
6
cycles

2
11 10
12
2 1p2
3
2
11002
3
112
3
a
1
10.002

1
10.02
b

2
Ap
3
2
1 s2
3
Y
3
a
1
1a
0
1
1a
c
b

1
Ap
3
2
1 s2
3
Y
3
122 a
1
2
`
a
c
a
0

1
Ap
3/2
1 s2
3
Y
3
a
c
a
0
a
3 2
da
M.12 FACTORS THAT AFFECT FATIGUE LIFE
18
As mentioned in Section M.9, the fatigue behavior of engineering materials is highly sensi-
tive to a number of variables. These factors include mean stress level, geometrical design,
surface effects, and metallurgical variables, as well as the environment. This section is devoted
to a discussion of these factors and to measures that may be taken to improve the fatigue
resistance of structural components.
Mean Stress
The dependence of fatigue life on stress amplitude is represented on the S-N plot.
Such data are taken for a constant mean stress s
m
, often for the reversed cycle situation
(s
m
0). Mean stress, however, will also affect fatigue life; this influence may be repre-
sented by a series of S-N curves, each measured at a different s
m
, as depicted schemati-
cally in Figure M.36. As may be noted, increasing the mean stress level leads to a
decrease in fatigue life (as well as a decrease in fatigue strength).
Empirical equations have been developed that express the dependence of fatigue
strength on mean stress. One of these, the Goodman equation, is written as follows:
(M.49)
In this expression
Surface Effects
For many common loading situations, the maximum stress within a component or struc-
ture occurs at its surface. Consequently, most cracks leading to fatigue failure originate
at surface positions, specifically at stress amplification sites. Therefore, it has been
observed that fatigue life is especially sensitive to the condition and configuration of the
component surface. Numerous factors influence fatigue resistance, the proper management
TS tensile strength
s
fs
0
fatigue strength for s
m
0
s
fs
fatigue strength for s
m
0
s
m
mean stress
s
fs
s
fs
0

a 1
s
m
TS
b
Goodman equation
computation of the
nonzero-mean-stress
fatigue limit for a
material using tensile
strength and zero-
mean-stress fatigue
limit values
18
This section is a slightly modified version of Section 9.13.
of which will lead to an improvement in fatigue life. These include design criteria as well
as various surface treatments.
Design Factors
The design of a component can have a significant influence on its fatigue character-
istics. Any notch or geometrical discontinuity can act as a stress raiser and fatigue crack
initiation site; these design features include grooves, holes, keyways, threads, and so on.
The sharper the discontinuity (i.e., the smaller the radius of curvature), the more severe
is the stress concentration. The probability of fatigue failure may be reduced by avoiding
(when possible) these structural irregularities or by making design modifications by
which sudden contour changes leading to sharp corners are eliminatedfor example,
calling for rounded fillets with large radii of curvature at the point where there is a change
in diameter for a rotating shaft (Figure M.37).
Surface Treatments
During machining operations, small scratches and grooves are invariably introduced
into the workpiece surface by cutting-tool action. These surface markings can limit the fa-
tigue life. It has been observed that improving the surface finish by polishing will en-
hance fatigue life significantly.
One of the most effective methods of increasing fatigue performance is by imposing
residual compressive stresses within a thin outer surface layer. Thus, a surface tensile
stress of external origin will be partially nullified and reduced in magnitude by the resid-
ual compressive stress. The net effect is that the likelihood of crack formation and there-
fore of fatigue failure is reduced.
Residual compressive stresses are commonly introduced into ductile metals mechan-
ically by localized plastic deformation within the outer surface region. Commercially, this
is often accomplished by a process termed shot peening. Small, hard particles (shot) hav-
ing diameters within the range of 0.1 to 1.0 mm are projected at high velocities onto the
surface to be treated. The resulting deformation induces compressive stresses to a depth
of between one-quarter and one-half of the shot diameter. The influence of shot peening
on the fatigue behavior of steel is demonstrated schematically in Figure M.38.
Case hardening is a technique by which both surface hardness and fatigue life are en-
hanced for steel alloys. This is accomplished by a carburizing or nitriding process by which
a component is exposed to a carbonaceous or nitrogenous atmosphere at an elevated
Fillet
Figure M.37 Demonstration of how design
can reduce stress amplification. (a) Poor
design: sharp corner. (b) Good design:
fatigue lifetime is improved by incorporating
a rounded fillet into a rotating shaft at the
point where there is a change in diameter.
(a) (b)
(logarithmic scale)
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
,

a
m
1
m
1
m
2
m
2
> >
m
3
m
3
Figure M.36 Demonstration of
influence of mean stress s
m
on SN
fatigue behavior.
case hardening
M.13 Environmental Effects 47
temperature. A carbon- or nitrogen-rich outer surface layer (or case) is introduced by
atomic diffusion from the gaseous phase. The case is normally on the order of 1 mm deep
and is harder than the inner core of material. (The influence of carbon content on hard-
ness for Fe-C alloys is demonstrated in Figure 11.30a.) The improvement of fatigue pro-
perties results from increased hardness within the case, as well as from the introduction
of residual compressive stresses that attends the carburizing or nitriding process. A carbon-
rich outer case may be observed for the gear shown in the top chapter-opening photograph
for Chapter 6; it appears as a dark outer rim within the sectioned segment. The increase
in case hardness is demonstrated in the photomicrograph in Figure M.39. The dark and
elongated diamond shapes are Knoop microhardness indentations. The upper indentation,
lying within the carburized layer, is smaller than the core indentation.
Cycles to failure
(logarithmic scale)
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
Shot peened
Normal
Case
Core
region
Figure M.38 Schematic SN fatigue
curves for normal and shot-peened steel.
Figure M.39 Photomicrograph showing both
core (bottom) and carburized outer case (top)
regions of a case-hardened steel. The case is
harder as attested by the smaller microhardness
indentation. 100.
(From R. W. Hertzberg, Deformation and Fracture
Mechanics of Engineering Materials, 3rd edition.
Copyright 1989 by John Wiley & Sons, New York.
Reprinted by permission of John Wiley & Sons, Inc.)
M.13 ENVIRONMENTAL EFFECTS
19
Environmental factors may also affect the fatigue behavior of materials. A few brief
comments will be given relative to two types of environment-assisted fatigue failure:
thermal fatigue and corrosion fatigue.
Thermal fatigue is normally induced at elevated temperatures by fluctuating thermal
stresses; mechanical stresses from an external source need not be present. The origin of
these thermal stresses is the restraint to the dimensional expansion and/or contraction
that would normally occur in a structural member with variations in temperature. The
magnitude of a thermal stress developed by a temperature change T depends on the
coefficient of thermal expansion a
l
and the modulus of elasticity E according to
(M.50) s a
l
ET
19
thermal fatigue
Thermal stress
dependence on
coefficient of
thermal expansion,
modulus of elasticity,
and temperature
change
(The topics of thermal expansion and thermal stresses are discussed in Sections 17.3 and
17.5.) Thermal stresses will not arise if this mechanical restraint is absent. Therefore, one
obvious way to prevent this type of fatigue is to eliminate, or at least reduce, the restraint
source, thus allowing unhindered dimensional changes with temperature variations, or to
choose materials with appropriate physical properties.
Failure that occurs by the simultaneous action of a cyclic stress and chemical attack
is termed corrosion fatigue. Corrosive environments have a deleterious influence and
produce shorter fatigue lives. Even normal ambient atmosphere will affect the fatigue
behavior of some materials. Small pits may form as a result of chemical reactions between
the environment and the material, which may serve as points of stress concentration, and
therefore as crack nucleation sites. In addition, the crack propagation rate is enhanced
as a result of the corrosive environment. The nature of the stress cycles will influence the
fatigue behavior; for example, lowering the load application frequency leads to longer
periods during which the opened crack is in contact with the environment and to a
reduction in the fatigue life.
Several approaches to corrosion fatigue prevention exist. On one hand, we can take
measures to reduce the rate of corrosion by some of the techniques discussed in Chapter 16
for example, apply protective surface coatings, select a more corrosion-resistant material,
and reduce the corrosiveness of the environment. Instead, or in addition, it might be
advisable to take actions to minimize the probability of normal fatigue failure, as out-
lined previouslyfor example, reduce the applied tensile stress level and impose residual
compressive stresses on the surface of the member.
corrosion fatigue
M.14 MECHANICS OF SPRING DEFORMATION
The basic function of a spring is to store mechanical energy as it is initially elastically
deformed and then recoup this energy at a later time as the spring recoils. In this section
helical springs that are used in mattresses and in retractable pens and as suspension springs
in automobiles are discussed. A stress analysis will be conducted on this type of spring, and
the results will then be applied to a valve spring that is used in automobile engines.
Consider the helical spring shown in Figure M.40, which has been constructed of
wire having a circular cross section of diameter d; the coil center-to-center diameter is
denoted as D. The application of a compressive force F causes a twisting force, or mo-
ment, denoted T, as shown in the figure. A combination of shear stresses result, the sum
of which, t, is
(M.51)
where K
w
is a force-independent constant that is a function of the D/d ratio:
(M.52)
In response to the force F, the coiled spring will experience deflection, which will be
assumed to be totally elastic. The amount of deflection per coil of spring, d
c
, as indicated
in Figure M.41, is given by the expression
K
w
1.60 a
D
d
b
0.140
t
8FD
pd
3
K
w
Automobile Valve Spring (Case Study)
The following submodule is a case study that discusses the valve spring found in a typical
automobile engine. Issues addressed include mechanics of the deformation of helical
springs, constraints imposed on the deformation of a typical valve spring, and, in addition,
one of the steel alloys that is commonly used for these springs and the rational for its use.
M.14 Mechanics of Spring Deformation 49
(M.53)
where G is the shear modulus of the material from which the spring is constructed.
Furthermore, d
c
may be computed from the total spring deflection, d
s
, and the number
of effective spring coils, N
c
, as
(M.54)
Now, solving for F in Equation M.53 gives
(M.55)
and substituting for F in Equation M.51 leads to
(M.56)
Under normal circumstances, it is desired that a spring experiences no permanent
deformation upon loading; this means that the right-hand side of Equation M.56 must
be less than the shear yield strength t
y
of the spring material, or that
(M.57) t
y
7
d
c
Gd
pD
2
K
w
t
d
c
Gd
pD
2
K
w
F
d
4
d
c
G
8D
3
d
c
d
s
N
c
d
c
8FD
3
d
4
G
D
F
F
T
d
Figure M.40 Schematic diagram of a helical spring
showing the twisting moment T that results from the
compressive force F.
(Adapted from K. Edwards and P. McKee, Fundamentals of
Mechanical Component Design. Copyright 1991 by McGraw-
Hill, Inc. Reproduced with permission of The McGraw-Hill
Companies.)
D
2
D
2
D
2
F
c
Figure M.41 Schematic diagrams of one coil of a helical spring, (a) prior to being compressed, and (b) showing
the deflection d
c
produced from the compressive force F.
(Adapted from K. Edwards and P. McKee, Fundamentals of Mechanical Component Design. Copyright 1991 by McGraw-Hill,
Inc. Reproduced with permission of The McGraw-Hill Companies.)
Condition for
nonpermanent spring
deformationshear
yield strength and its
relationship to shear
modulus, number of
effective coils, and
spring and wire
diameters
(a) (b)
M.15 VALVE SPRING DESIGN AND MATERIAL REQUIREMENTS
We shall now apply the results of the preceding section to an automobile valve spring. A
cut-away schematic diagram of an automobile engine showing these springs is presented
in Figure M.42. Functionally, springs of this type permit both intake and exhaust valves to
alternately open and close as the engine is in operation. Rotation of the camshaft causes a
valve to open and its spring to be compressed, so that the load on the spring is increased.
The stored energy in the spring then forces the valve to close as the camshaft continues its
rotation. This process occurs for each valve for each engine cycle, and over the lifetime of
the engine it occurs many millions of times. Furthermore, during normal engine operation,
the temperature of the springs is approximately 80C (175F).
A photograph of a typical valve spring is shown in Figure M.43. The spring has a total
length of 1.67 in. (42 mm), is constructed of wire having a diameter d of 0.170 in. (4.3 mm),
has six coils (only four of which are active), and has a center-to-center diameter D of
1.062 in. (27 mm). Furthermore, when installed and when a valve is completely closed,
its spring is compressed a total of 0.24 in. (6.1 mm), which, from Equation M.54, gives an
installed deflection per coil d
ic
of
The cam lift is 0.30 in. (7.6 mm), which means that when the cam completely opens a
valve, the spring experiences a maximum total deflection equal to the sum of the valve
d
ic

0.24 in.
4 coils
0.060 in./coil 11.5 mm/coil 2
Cam
Camshaft
Exhaust
valve
Piston
Valve
spring
Intake
valve
Crankshaft
Figure M.42 Cutaway drawing of a section of
an automobile engine in which various
components including valves and valve springs
are shown.
Figure M.43 Photograph
of a typical automobile valve
spring.
M.15 Valve Spring Design and Material Requirements 51
lift and the compressed deflection, namely, 0.30 in. 0.24 in. 0.54 in. (13.7 mm).
Hence, the maximum deflection per coil, d
mc
, is
Thus, we have available all of the parameters in Equation M.57 (taking d
c
d
mc
), except
for t
y
, the required shear yield strength of the spring material.
However, the material parameter of interest is really not t
y
inasmuch as the spring
is continually stress cycled as the valve opens and closes during engine operation; this
necessitates designing against the possibility of failure by fatigue rather than against the
possibility of yielding. This fatigue complication is handled by choosing a metal alloy
that has a fatigue limit (Figure M.26a) that is greater than the cyclic stress amplitude to
which the spring will be subjected. For this reason, steel alloys, which have fatigue lim-
its, are normally employed for valve springs.
When using steel alloys in spring design, two assumptions may be made if the stress
cycle is reversed (if t
m
0, where t
m
is the mean stress, or, equivalently, if t
max
t
min
,
in accordance with Equation M.38 and as noted in Figure M.44). The first of these as-
sumptions is that the fatigue limit of the alloy (expressed as stress amplitude) is 45,000
psi (310 MPa), the threshold of which occurs at about 10
6
cycles. Secondly, for torsion and
on the basis of experimental data, it has been found that the fatigue strength at 10
3
cycles
is 0.67TS, where TS is the tensile strength of the material (as measured from a pure ten-
sion test). The SN fatigue diagram (i.e., stress amplitude versus logarithm of the number
of cycles to failure) for these alloys is shown in Figure M.45.
Now let us estimate the number of cycles to which a typical valve spring may be sub-
jected in order to determine whether it is permissible to operate within the fatigue limit
regime of Figure M.45 (i.e., if the number of cycles exceeds 10
6
). For the sake of argu-
ment, assume that the automobile in which the spring is mounted travels a minimum of
100,000 miles (161,000 km) at an average speed of 40 mph (64.4 km/h), with an average
engine speed of 3000 rpm (rev/min). The total time it takes the automobile to travel this
distance is 2500 h (100,000 mi/40 mph), or 150,000 min. At 3000 rpm, the total number of
revolutions is (3000 rev/min)(150,000 min) 4.5 10
8
rev, and since there are 2 rev/cycle,
the total number of cycles is 2.25 10
8
. This result means that we may use the fatigue
limit as the design stress inasmuch as the limit cycle threshold has been exceeded for the
100,000-mile distance of travel (i.e., since 2.25 10
8
cycles 10
6
cycles).
Furthermore, this problem is complicated by the fact that the stress cycle is not com-
pletely reversed (i.e., t
m
0) inasmuch as between minimum and maximum deflections
d
mc
0.54 in.
4 coils
0.135 in./coil 13.4 mm/coil 2
Time
S
t
r
e
s
s

max
0
min
Figure M.44 Stress versus time for a reversed

cycle in shear.
0.67TS
45,000 psi
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
,

S
10
3
10
5
10
7
10
9
(logarithmic scale)
Figure M.45 Shear stress amplitude versus logarithm of the
number of cycles to fatigue failure for typical ferrous alloys.
the spring remains in compression; thus, the 45,000-psi (310-MPa) fatigue limit is not
valid. What we would now like to do is to make an appropriate extrapolation of the
fatigue limit for this t
m
0 case and then compute and compare with this limit the actual
stress amplitude for the spring; if the stress amplitude is significantly below the extrapo-
lated limit, then the spring design is satisfactory.
A reasonable extrapolation of the fatigue limit for this t
m
0 situation may
be made using the Goodman equation (Equation M.49) modified to take into account
the application of shear (rather than tensile) stresses. This modified expression takes the
form
(M.58)
where is the fatigue limit for the mean stress t
m
; is the fatigue limit for t
m
0 [i.e.,
45,000 psi (310 MPa)]; and, again, TS is the tensile strength of the alloy]. To determine
the new fatigue limit from this expression necessitates the computation of both the
tensile strength of the alloy and the mean stress for the spring.
t
fl
t
fl
0
t
fl
t
fl
t
f l
0

a 1
t
m
0.67TS
b
M.16 ONE COMMONLY EMPLOYED STEEL ALLOY
One common spring alloy is anASTM232 chromevanadiumsteel, having a composition of
0.480.53 wt%C, 0.801.10 wt%Cr, a minimum of 0.15 wt%V, and the balance being Fe.
Spring wire is normally cold drawn (Section 14.2) to the desired diameter; consequently, ten-
sile strength will increase with the amount of drawing (i.e., with decreasing diameter). For
this alloy it has been experimentally verified that, for the diameter d in inches, the tensile
strength is
(M.59)
Since d 0.170 in. for this spring,
Computation of the mean stress t
m
is made using Equation M.38 modified to the
shear stress situation as follows:
(M.60)
It now becomes necessary to determine the minimum and maximum shear stresses for
the spring, using Equation M.56. The value of t
min
may be calculated from Equations M.56
and M.52 inasmuch as the minimum d
c
is known (i.e., d
ic
0.060 in.). A shear modulus of
11.5 10
6
psi (79 GPa) will be assumed for the steel; this is the room-temperature value,
which is also valid at the 80C service temperature. Thus, t
min
is just
(M.61a)
41,000 psi 1280 MPa2
c
10.060 in. 2 111.5 10
6
psi 2 10.170 in. 2
p11.062 in. 2
2
d c 1.60 a
1.062 in.
0.170 in.
b
0.140
d

d
ic
Gd
pD
2
c 1.60 a
D
d
b
0.140
d
t
min
d
ic
Gd
pD
2
K
w
t
m
t
min
t
max
2
227,200 psi 11579 MPa2
TS 1psi 2 169,00010.170 in. 2
0.167
TS 1psi 2 169,0001d2
0.167
For an ASTM 232
steel wire,
dependence of
tensile strength on
drawn wire diameter
M.16 One Commonly Employed Steel Alloy 53
Now t
max
may be determined taking d
c
d
mc
0.135 in. as follows:
(M.61b)
Now, from Equation M.60,
The variation of shear stress with time for this valve spring is noted in Figure M.46; the
time axis is not scaled, inasmuch as the time scale will depend on engine speed.
Our next objective is to determine the fatigue limit for this t
m
66,600 psi
(460 MPa) using Equation M.58 and for and TS values of 45,000 psi (310 MPa) and
227,200 psi (1570 MPa), respectively. Thus,
Now let us determine the actual stress amplitude t
aa
for the valve spring using
Equation M.40 modified to the shear stress condition:
Thus, the actual stress amplitude is slightly greater than the fatigue limit [i.e.,
], which means that this spring design is marginal. t
aa
125,600 psi 2 7 t
fl
125,300 psi 2

92,200 psi 41,000 psi
2
25,600 psi 1177 MPa2
t
aa
t
max
t
min
2
25,300 psi 1175 MPa2
145,000 psi 2 c 1
66,600 psi
10.672 1227,200 psi 2
d
t
fl
t
fl
0
a 1
t
m
0.67TS
b
t
fl
0
1t
fl
2

41,000 psi 92,200 psi
2
66,600 psi 1460 MPa2
t
m
t
min
t
max
2
92,200 psi 1635 MPa2
c
10.135 in. 2 111.5 10
6
psi 2 10.170 in. 2
p11.062 in. 2
2
d c 1.60 a
1.062 in.
0.170 in.
b
0.140
d
t
max
d
mc
Gd
pD
2
c 1.60 a
D
d
b
0.140
d
100
80
60
40
20
0
Time
S
t
r
e
s
s

(
1
0
3

p
s
i
)
max
= 92,200 psi
aa
= 25,600 psi
min
= 41,000 psi

m
= 66,600 psi
Figure M.46 Shear stress versus time for an
automobile valve spring.
The fatigue limit of this alloy may be increased to greater than 25,300 psi (175 MPa)
by shot peening, a procedure described in Section M.12. Shot peening involves the intro-
duction of residual compressive surface stresses by plastically deforming outer surface re-
gions; small and very hard particles are projected onto the surface at high velocities. This
is an automated procedure commonly used to improve the fatigue resistance of valve
springs; in fact, the spring shown in Figure M.43 has been shot peened, which accounts for
its rough surface texture. Shot peening has been observed to increase the fatigue limit of
steel alloys in excess of 50% and, in addition, to reduce significantly the degree of scatter
of fatigue data.
This spring design, including shot peening, may be satisfactory; however, its ade-
quacy should be verified by experimental testing. The testing procedure is relatively
complicated and, consequently, will not be discussed in detail. In essence, it involves per-
forming a relatively large number of fatigue tests (on the order of 1000) on this shot-
peened ASTM 232 steel, in shear, using a mean stress of 66,600 psi (460 MPa) and a
stress amplitude of 25,600 psi (177 MPa), and for 10
6
cycles. On the basis of the number
of failures, an estimate of the survival probability can be made. For the sake of argument,
let us assume that this probability turns out to be 0.99999; this means that one spring in
100,000 produced will fail.
Suppose that you are employed by one of the large automobile companies that
manufactures on the order of 1 million cars per year, and that the engine powering each
automobile is a six-cylinder one. Since for each cylinder there are two valves, and thus
two valve springs, a total of 12 million springs would be produced every year. For the
preceding survival probability rate, the total number of spring failures would be approx-
imately 120, which also corresponds to 120 engine failures. As a practical matter, one
would have to weigh the cost of replacing these 120 engines against the cost of a spring
redesign.
Redesign options would involve taking measures to reduce the shear stresses on the
spring, by altering the parameters in Equations M.52 and M.56. This would include either
(1) increasing the coil diameter D, which would also necessitate increasing the wire
diameter d, or (2) increasing the number of coils N
c
.
Investigation of Engineering Failures
The entirety of Chapter 9 was devoted to discussions of the various forms of failure that
materials experience, of failure mechanisms, and, in some instances, of measures that
may be taken to prevent, or at least, minimize the possibility of failure. However, once
an unexpected failure has occurred an investigation may be conducted in order to deter-
mine the causes or factors that led to the failure, and to recommend courses of action
that, if taken, will prevent or at least reduce the likelihood of future events. In some
instances the primary purpose of organizing a failure investigation is to assign legal
responsibility for the consequences of the failure incidentwho is to be held accountable:
the company/individual that manufactured the failed component, or the company/indi-
vidual that was operating the component when it failed? Thus, the term forensic engi-
neering is sometimes used in the context of failure investigations and analyses.
Inasmuch as some engineers will be expected to conduct failure investigations, we have
included this submodule as a guide for planning and conducting effective and organized
investigations. The discussion that follows addresses the following topics: causes and
kinds of failure, planning a failure investigation, types of failure mechanisms, procedures
that may be used to ascertain root causes, and how to determine corrective actions. More
detailed treatments of the whys and hows of failure investigations are contained in the
reference list at the end of this module.
M.17 Reasons for Failure 55
M.17 REASONS FOR FAILURE
At the outset of such an investigation, one of the first issues to be addressed is why the
failure occurred. And as we shall see below, such a failure analysis is just one aspect of
the overall failure investigation. There are many possible reasons for engineering fail-
ures, and one way of classifying the various types is as follows:
Design errors
Fabrication/manufacturing defects
Assembly errors
Misuse during operation
Improper maintenance
Design Errors
Several aspects of design determine a products overall reliability. The shape, size, and con-
figuration of a component are important so that it will (1) perform the function intended,
(2) withstand any applied loads without deforming excessively or fracturing, and (3) not
fail as a result of unanticipated stress levels that result from the presence of stress raisers
sharp corners, configurational discontinuities, etc. Selecting materials that have an appro-
priate combination of properties (mechanical, electrical, etc.) is also an important aspect
of design; this also includes the specification of any treatments to which the materials are
to be subjected (e.g., heat treatments, cold working, etc.). Designation of manufacturing
and assembly procedures is also part of the design process, which also have an influence
on the lifetime of a product.
Fabrication/Manufacturing Defects
There are many possible types of fabrication/manufacturing defects, which normally
are relatively easy to identify as causes of failure. Virtually all of the fabrication tech-
niques discussed in Chapter 14 are prone to the introduction of defects. Some of the
more common fabrication/manufacturing defects include welding defects (porosity,
lack of penetration), improper heat treating, machining/grinding defects (gouges,
burns, tears, scratches, cracks), decarburization, and casting defects (porosity, shrink-
age cavities).
Assembly Errors
During a manufacturing process, the various components must be assembled together to
form the desired product. In todays world, in order to be economically competitive, in-
dustries have to devise faster and cheaper assembly processes. This, coupled with in-
creasingly more complicated products, leads to a greater likelihood that components
wont be assembled correctly. Furthermore, automated inspection techniques often do
not detect misassembled products. And, of course, a misassembled product has a greater
probability of failing prematurely than one that was assembled correctly.
Misuse During Operation
Most products and machines are designed to have a reasonable lifetime expectancy; this
life expectancy is often expressed in terms of a warranty by the manufacturer. A failure
occurs when the component/machine wears out sooner than expected. Many times this
type of failure results when the component/machine is operated improperly or is abused
during servicei.e., when operating procedures recommended by the manufacturer are
not observed. For example, the radiator in an automobile may fail if the appropriate water
level is not maintained. This type of failure is one of the most common, and should be one
of the first suspects in the investigation.
Improper Maintenance
In order to function properly, many products require periodic maintenancefor exam-
ple, automobiles (engine oil changes, tire rotation), lawn mowers (lubrication), aircraft
(inspection/replacement of high-stress structures), computers (virus checks), etc.
Improper maintenance can result in a premature failure of a component, structure, or
machine, and may be intentional or unintentional. Corrosion failures often result from
maintenance neglect.
M.18 ROOT CAUSES
We sometimes refer to the actual and true cause of failure as the root cause. This root
cause will most likely be related to one or more of the reasons for failure discussed in
the previous section (e.g., design errors, fabrication/manufacturing defects, etc.).
Furthermore, there are really three levels or classifications of root causesviz. physical,
human, and latentwhich are described as follows:
PhysicalFor physical, the primary cause for the failure of a component/structure
is related to one of the failure types or mechanisms discussed previouslyviz.
fracture due to overload, fatigue, creep, etc.
HumanFor human, a physical cause may be of secondary importance, in that the
actions of an individual led up to the failurefor example, a poorly written set of
instructions on how to use or properly maintain a product.
LatentA latent root cause relates to failures resulting from organizational policy
e.g., company cost-reduction measures with the elimination of critical testing
procedures.
The relationships among reasons for failure and root causes are presented in
Figure M.47.
Unfortunately, some failure investigations never discern the actual root cause.
Whereas the real root cause may be human or latent, the investigation stops at the phy-
sical cause level. It is essential that the failure investigation be conducted so as to include
the possibility of involvment of human and latent factors.
Another complicating issue is that a series of events may lead up to the eventual
failure. For example, a failure-producing crack is initiated by stress corrosion; this crack
then propagates in response to cyclic stresses (i.e., it becomes a fatigue crack); and final
failure results from a mechanical overload condition (when this crack reaches some crit-
ical length). Thus, three physical causes are involved in this failure. Determination of the
real root cause (i.e., the crack induced by stress corrosion) becomes a complex problem
for the failure investigator.
Physical
Design
errors
Assembly
errors
Misuse
during
operation
Improper
maintenance
Fabrication/
manufacturing
defects
Human
Root causes of failure
Latent
Reasons for failure
Figure M.47 Interrelationships
among the root causes of failure
and reasons for failure.
root cause
M.20 What Is the Root Cause of the Failure Problem? 57
The Failure Analysis
A failure investigation is essentially an exercise in problem solving, which can be broken
down into finding the answers to the following four questions:
1. What exactly is the failure problem?
2. What is the root cause of the failure problem?
3. What are possible solutions?
4. Which of these is the best solution?
These four steps are also commonly used by engineers to solve most general engineer-
ing problems.
A schematic diagram that outlines those procedures used to answer the above four
questions is shown in Figure M.48. The sections that follow discuss the implementation
of these protocols.
M.19 WHAT EXACTLY IS THE FAILURE PROBLEM?
The first question that should be asked in any failure investigation is: What event pre-
cipitated the malfunction of a component, machine, or process? The answer, in essence,
defines the purpose of the investigation. It will help also to determine what kind(s) of
expertise is (are) needed, as well as the time and resources required.
What is the failure
problem?
What is the
root cause?
Identification of
possible root causes
What are possible
solutions?
Which solution
is best?
Implementation of appropriate
analytical procedures
(NDT, microscopic examination,
mechanical testing, etc.)
Construction of
fault tree
Prove or disapprove
each possible root cause
Identification of
root cause
Construction of
corrective action tree
Figure M.48 Schematic outline of
procedures used to answer the four failure
investigation questions.
M.20 WHAT IS THE ROOT CAUSE OF THE FAILURE PROBLEM?
Ascertaining the root cause of the failure is one of the primary goals of a failure inves-
tigation. It is at this point that planning and organization of the investigation take place.
This includes the formation of an investigation team, which will be composed of techni-
cal experts who have appropriate expertise and experience.
As noted in Figure M.48 (the column below What is the root cause?), it is first
necessary to identify all possible root causes. This is accomplished by conducting a vari-
ety of analytical procedures so as to gain the best possible understanding of the failure.
One tool that may be implemented to help organize the investigation and discern the
actual root cause from the list of possibilities is a fault tree.
Failure Analysis Procedures
A well organized failure analysis will involve a number of procedures; some of the com-
mon ones are provided in the following list. The sequence followed in an actual analysis
need not be as given; furthermore, not all procedures are included in every investigation.
1. Collection of background data and selection of samples for examination
2. Preliminary visual examination of the failed part
3. Nondestructive testing
4. Mechanical testing (e.g., tensile, hardness, impact)
5. Selection, identification, preservation, and/or cleaning of critical specimens
6. Macroscopic examination and analysis of fracture surfaces, secondary cracks, and
other important surface features
7. Microscopic examination and analysis of fracture surfaces
8. Selection, preparation, examination, and analysis of metallographic sections
9. Determination of the actual stress state of the failed component
10. Determination of the failure mode
11. Chemical analyses (bulk, local, surface corrosion products, and deposits or coatings)
12. Application of fracture mechanics
We now present some discussion for each of these procedures.
Collection of Background Data and Samples
Background data should include, when available, information pertaining to the original
design (including all underlying assumptions), manufacture, processing, fabrication, and serv-
ice history of the failed component. Details regarding abnormal and unusual conditions such
as loading excursions, variations in temperature, the presence of a corrosive environment, and
any accidental events are part of the service record. Photographs of the failed component and
its surrounding environment are also essential background information. It may be necessary
to select samples for both macroscopic and microscopic examinations. These specimens
should be carefully chosen so as to include not only the region that encompasses the failure,
but also other locations both adjacent to and far removed from the failure site. Care should
be exercised so as to preserve any debris or oxide materials that are present.
Preliminary Visual Examination
The next step is to perform an examination, using the unaided eye, of the part that
failed as well as all of its broken fragments. Of particular interest are the features of and
changes in texture across the fracture surface, any evidence of corrosion, surface marks,
and angle of fracture. Details of this examination should be documented both in writing
and with photographs. When taking photographs, direction of light illumination may be
important so as to reveal critical surface characteristics. Examination of fine features of
the failure surface may be necessary using a magnifying glass.
Nondestructive Inspection
Some of the nondestructive testing techniques discussed in Section M.5 (Table M.6)
for detecting flaws in structural components may also be utilized in failure analysesto
detect small surface cracks and discontinuities in failed parts. Dye-penetrant, ultrasonic,
and radiographic are those most commonly used.
Mechanical Testing
Mechanical tests on failed parts are conducted for several reasons: to determine if
the material conforms to specifications; to determine the heat treatment; to detect any
alteration of mechanical properties due to cold working or overheating; and to detect
decarburization or any increase in carbon and/or nitrogen concentration. Hardness tests
are easiest to conduct, but tensile and impact tests are also possible provided that ade-
quate material is available for the fabrication of test specimens.
Specimen Preservation and Selection
This stage is important in order that evidence critical to the investigation is not de-
stroyed, obscured, or altered. Fracture surfaces may be susceptible to damage from me-
chanical forces or some chemical environments, and, therefore, should be protected dur-
ing the investigation. The investigator should not try to fit back together broken sections,
and touching and rubbing fracture surfaces should be avoided. The best way to prevent
chemical damage is to place the fracture specimen in a desiccator, or pack it with a desic-
cant material (one that removes water vapor from the air). In some cases it may be neces-
sary to dry the specimen, which may be accomplished using a jet of dry air (which will of
course blow away any surface foreign residue that may be important to the investigation).
In order to perform some tests and examinations (e.g., hardness, electron micro-
graphic, photomicrographic), it may be necessary to remove a portion of the fracture
specimen of convenient size. This is normally done using a cutting or sectioning proce-
dure. Measures to protect the area of fracture are necessary, and the location of any cut-
ting action should be chosen such that the fracture region itself as well as adjacent areas
are not damaged or altered. The cutting action attendant to sectioning will necessarily
heat neighboring regions with possible alteration of microstructure and properties; it is
essential that microstructural elements and properties of critical areas be preserved.
Macroscopic Examination
Macroscopic examinations are conducted with the unaided eye, and/or using a hand-
held magnifying glass, a low-power stereoscopic microscope, and/or a scanning electron mi-
croscope (SEM) (at low magnifications). In general, magnifications range between 1 and 50.
Reasons for conducting this type of examination include: to locate the crack origin, to
determine its shape and size as well as the path of crack propagation, to characterize the
texture of the fracture surface, and to note possible points of stress concentration (e.g.,
drilled holes, hammer marks, accidental dents, etc.) as well as any other gross features that
may provide clues as to the mode of failure. In addition, an attempt should be made to
determine if there is more than one crack origin.
Surface Topography. For failures that result from overload conditions, the
topography of the fracture surface depends on whether the material was ductile or brittle,
as well as the manner of loading (i.e., tensile, shear, torsional, bending, or combinations
of these loading modes). Figures M.49a and M.49b show schematic representations of
Brittle
Tension Torsion
Brittle Ductile Ductile
Figure M.49 Characteristic fracture surface contours for
(a) ductile and (b) brittle materials that are stressed in uniaxial tension,
and (c) ductile and (d) brittle materials that are stressed in torsion.
(Adapted from D. J. Wulpi, Understanding How Components Fail, ASM
International, 1985, p. 30. Reprinted with permission of ASM International
.
All rights reserved. www.asminternational.org.)
(a) (b) (c) (d)
fracture surfaces for cylindrical specimens of both ductile and brittle materials that failed
from overloading in uniaxial tension. The fracture surface for the ductile material
(normally a metal) has the typical cup-and-cone configuration (per the photograph of
Figure M.50a)i.e., central regions of both mating pieces are relatively flat, oriented
perpendicular to the stress direction, and have a rough and fibrous texture, whereas
the plane of the outer-periphery shear lips makes a 45 angle with the stress direction.
(The mechanism of crack formation and propagation for this situation is represented in
Figure M.51.) By way of contrast, for the brittle material (Figure M.49b), once formed,
a crack propagates within a plane that is oriented perpendicular to the stress axis, and
yields a flat failure surface. A photograph of a specimen that failed in this manner is
shown in Figure M.50b.
Consider now the situation in which the overload stress is torsional in nature. For
cylindrical specimens of ductile and brittle materials, schematic failure profiles are
shown, respectively, in Figures M.49c and M.49d. For the ductile material, the fracture
surface is flat and oriented parallel to the direction of the applied torsional stress. And
a helical fracture surface results when the material is brittle.
Failures resulting from other mechanisms may have yet other surface configura-
tions. For example, Figure M.52 is a photograph of the surface of a shaft that failed by
fatigue. Important features shown here include the crack origin [on the outer surface
(near the top edge)], the region of slow crack propagation during cycling (that appears
light and has a smooth texture), and the area of rapid failure (having a dull and fibrous
texturecorresponding to the region of largest cross-sectional area).
At this time, it is appropriate to make a distinction between brittle materials and
brittle fractures. A brittle fracture is one in which there is little or no gross plastic defor-
mation on a macroscale. Of course, when brittle materials are overloaded, they fracture
in a brittle manner. On the other hand, under some circumstances, there may be very little
evidence of any macroscale deformation on the failure surface of a ductile metali.e.,
the mode of fracture is brittle. For example, for a failure mechanism in which cracks form
and then propagate relatively slowly (i.e., as with fatigue or stress-corrosion cracks),
crack growth proceeds until the intact cross-sectional area of the part reaches a state of
overloading, at which time rapid crack propagation and sudden failure occur. This type
of failure may be observed in the photograph of Figure M.52the rotating steel shaft
that experienced a fatigue failure. In this case, a condition of overload was achieved once
the crack had propagated through that cross-sectional region that appears light in the
photograph. Furthermore, the area of rapid failure (the largest cross-sectional area that
appears dark) has a dull and fibrous texture (and no evidence of plastic deformation).
These features are characteristic of a brittle fracture, in spite of the fact that this steel is
a ductile material.
Figure M.50 (a) Cup-and-
cone fracture in aluminum.
(b) Brittle fracture in a mild
steel.
(a) (b)
Surface Features. In addition to failure surface topography, surface features
that are present may also provide valuable information about the failure mode. For
example, Figure M.53a is a photograph (unmagnified) that shows matching cross sections
of a structure that failed in a brittle manner. The fracture surfaces are relatively flat
(indicative of a brittle fracture), and V-shaped chevron markings may be observed
that point back toward the crack origin.
Another type of brittle fracture surface is presented in the photograph of Figure
M.53b; here fan-shaped ridges may be observed that radiate from the crack origin.
The photograph of Figure M.54 shows the fracture surface of a rotating steel shaft
that experienced fatigue failure. In addition to the point of crack origin (at the corner of
a keyway on the shaft) and site of final rupture, beachmark ridges may be observed;
beachmarks, features found on some fatigue failure surfaces, are discussed briefly in
Section M.10.
In this discussion, we have treated some of the more common types of macroscopic
configurations and features found on surfaces of failures; of course, others are also possible.
It should be noted that accurate interpretation of them is a skill acquired only through
experience; and space does not allow a more thorough treatment here.
Fibrous
Shear
Figure M.51 Stages in the cup-and-cone
fracture. (a) Initial necking. (b) Small cavity
formation. (c) Coalescence of cavities to form
a crack. (d) Crack propagation. (e) Final
shear fracture at a 45 angle relative to the
tensile direction.
(From K. M. Ralls, T. H. Courtney, and J. Wulff,
Introduction to Materials Science and Engineering,
p. 468. Copyright 1976 by John Wiley & Sons,
New York. Reprinted by permission of John Wiley
& Sons, Inc.)
(a) (b)
(d) (e)
(c)
Region of slow
crack propagation
Region of rapid failure
2 cm
Figure M.52 Fatigue failure surface. A crack formed at the
top edge. The smooth region also near the top corresponds to
the area over which the crack propagated slowly. Rapid
failure occurred over the area having a dull and fibrous
texture (the largest area). Approximately 0.5.
(Reproduced by permission from Metals Handbook: Fractography
and Atlas of Fractographs, Vol. 9, 8th edition, H. E. Boyer, Editor,
American Society for Metals, 1974.)
Microscopic Examination
Microscopic (or fractographic) examinations are conducted at higher magnifica-
tions than macroscopic ones; normally a scanning electron microscope is used.
Magnifications as high as 200,000 times are possible, as are also large depths of field.
Depth of field is important in order to adequately observe topographical features of the
failure surface at these magnifications. Some SEMs are equipped with energy-dispersive
x-ray spectroscopes, which permit semiquantative and quantitative chemical analyses of
selected areas. This capability is useful in ascertaining the chemistry of microstructural
features. The most significant limitation of SEM analysis is specimen size; in order to fit
within the examination chamber, a specimen must have a diameter less than about 200
mm (8 in.)consequently, it is necessary to section pieces that are larger than this.
10 mm
Figure M.53 (a) Photograph showing V-shaped chevron markings characteristic of brittle fracture. Arrows
indicate the origin of cracks. Approximately actual size. (b) Photograph of a brittle fracture surface showing radial
fan-shaped ridges. Arrow indicates the origin of the crack. Approximately 2.
[(a) From R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edition. Copyright 1989 by John
Wiley & Sons, New York. Reprinted by permission of John Wiley & Sons, Inc. Photograph courtesy of Roger Slutter, Lehigh
University. (b) Reproduced with permission from D. J. Wulpi, Understanding How Components Fail, American Society for Metals,
Materials Park, OH, 1985.]
(a)
(b)
A microscopic examination may also provide valuable evidence regarding the
mechanism of failure. For example, an SEM micrograph for a ductile metal that failed
due to overloading in tension will appear as that shown in Figure M.55a; that is, spheri-
cal dimples will be present. Whereas, for a shear overloading failure (also of a ductile
metal), dimples will have a parabolic shape (Figure M.55b).
As noted in Section 9.4, brittle failures of metals may be transgranular (crack prop-
agation is through interiors of grains) or intergranular (crack propagation is along grain
boundaries). For transgranular, an SEM micrograph will reveal cleavage facets, Fig-
ure M.56a, whereas a grainy or faceted texture (characteristic of the three-dimensional
nature of the grains) will exist when the failure is intergranular (Figure M.56b).
Figure M.54 Fracture surface of a rotating steel
shaft that experienced fatigue failure. Beachmark
ridges are visible in the photograph.
(Reproduced with permission from D. J. Wulpi, Understanding
How Components Fail, American Society for Metals,
Materials Park, OH, 1985.)
(a)
(b)
5 m
4 m
Figure M.55 (a) Scanning electron fractograph showing spherical dimples characteristic of ductile fracture
resulting from uniaxial tensile loads. 3300. (b) Scanning electron fractograph showing parabolic-shaped dimples
characteristic of ductile fracture resulting from shear loading. 5000.
(From R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edition. Copyright 1989 by John
Wiley & Sons, New York. Reprinted by permission of John Wiley & Sons, Inc.)
Intergranular fracture is caused by some process that weakens or embrittles grain bound-
ary regionse.g., segregation of embrittling species (viz., hydrogen, liquid metals) along
grain boundaries, intergranular stress-corrosion cracking, etc.
On occasion, more than one fracture mode is involved in a failure process. For
example, a mixed-mode fracture is shown in the SEM fractograph of Figure M.57. Here
may be observed both circular dimples (characteristic of ductile fracture, Figure M.55a),
as well as cleavage facets (characteristic of transgranular brittle fracture, Figure M.56a).
For some (but not all) failures resulting from fatigue, electron fractographs will reveal
the presence of striations (Figure M.58)closely spaced and parallel lines or curves. These
striations are oriented perpendicular to the direction of crack propagation, and each stri-
ation represents the advance distance of the crack front during a single loading cycle.
Fractographs of brittle ceramic materials have their own distinctive features.
Figure M.59, an optical photomicrograph that shows the fracture surface of a fused silica
rod that was fractured in four-point bending, represents one possibility. As noted, a
flat, smooth, and highly reflective mirror region is present; it corresponds to the fracture
surface that formed during the initial stage of crack propagation. The outer perimeter
of this region is circular, with the crack origin located near its centerin this case on
the surface of the rod. The measured radius of this circle may be used to approximate
the stress level that caused fracture; fracture stress level is inversely proportional to
the square root of radius (Equation 9.14). Also shown in Figure M.59 are the mist and
hackle, annular areas that reside outside the mirror region. The advancing crack began
to branch and change propagation direction within the mist and hackle regions, which
gave rise to rougher surface textures.
Metallographic Examination
For metallographic examination, a specimen is first removed from the failed piece
by sectioning; this specimen is then polished and etched so as to reveal details of its
200 m
Figure M.56 (a) Scanning electron fractograph of ductile cast iron showing a transgranular fracture surface.
Magnification unknown. (b) Scanning electron fractograph showing an intergranular fracture surface. 50.
[(a) From V. J. Colangelo and F. A. Heiser, Analysis of Metallurgical Failures, 2nd edition. Copyright 1987 by John Wiley & Sons,
New York. Reprinted by permission of John Wiley & Sons, Inc. (b) Reproduced with permission from ASM Handbook, Vol. 12,
Fractography, ASM International, Materials Park, OH, 1987.]
(a) (b)
Figure M.57 Scanning electron micrograph of a
mixed-mode fracture surface, which is composed of
both cleavage and dimpled regions. 570.
(Reproduced with permission from Handbook of Case Studies
in Failure Analysis, Vol. 1 (1992), ASM International,
Materials Park, OH, 44073-0002.)
1 m
Figure M.58 Transmission electron fractograph
showing fatigue striations in aluminum. 9000.
(From V. J. Colangelo and F. A. Heiser, Analysis of Metallurgical
Failures, 2nd edition. Copyright 1987 by John Wiley & Sons,
New York. Reprinted by permission of John Wiley & Sons, Inc.)
microstructuree.g., grains, the various phases present, etc. Microstructural observa-
tions are normally conducted using an optical microscope.
A wealth of microstructural data regarding probable causes of failure may be
gleaned from a metallographic examination. Types of information that are available in-
clude the following:
Grain size and shape provide some indication as to thermal and mechnical history.
A coarse grain structure (i.e., large grains) indicates that the material was likely
subjected to an annealing heat treatment or perhaps overheating during service
(with grain growth). Whereas an elongated grain structure means that the speci-
men was deformed by some mechanical process (e.g., forging, rolling, drawing).
Also, deformation direction may be deduced from the grain-elongation direction.
Identification of the microconstituents present is helpful in determining whether
the material was propertly heat treated, as well as ascertaining other aspects of its
thermal historye.g., was the microstructure produced intentionally during manu-
facturing, or did it result inadvertently while in service? The presence of a grain
boundary phase may explain intergranular brittle fracture. Also, determination of
heat treatment deficiencies, such as surface decarburization, is possible.
Internal imperfections such as pores, inclusions, and welding defects, that may
have played a role in the failure process, may also be revealed.
Other effects resulting from service conditions may be investigated. These include the
occurrence of oxidation, corrosion, and severe surface strain hardening.
And, determination of the mode of crack propagation (viz. transgranular or intergran-
ular) is possible when both crack and grain structure are shown in a photomicrograph.
In some investigations it is imperative to ascertain whether or not the microstructure of
the failure region is indeed representative of the microstructure of the component in its
as-manufactured condition (i.e., did the service conditions alter the microstructure, and,
if so, what was the nature of the alteration?). This determination is possible by compar-
ing photomicrographs of specimens taken both from a region adjacent to the failure and
from a remote area.
Stress Analyses
When an excessively high load level (i.e., a condition of overload) is suspected as the
direct cause of failure, a stress analysis is warranted so as to ascertain if it is indeed the
root cause. Determinations of stress magnitude as well as type (tension, torsion, bend-
ing, static, fluctuating) are the goals of such an analysis. The failure analyst should also
endeavor to obtain records from the manufacturer relating to allowable stress levels that
were predicted during the original design of the part. Experimental verification of these
data is possible by taking strain gauge measurements on other identical (or similar) com-
ponents during exposure to in-service stresses. The above information is useful in deter-
mining whether the failed part was sized properly, if properties of the materials called
for in the design of the part met specifications, or if failure was a result of overloading.
Mist region
Origin Mirror region
Hackle region
Figure M.59 Photomicrograph of the fracture surface of a 6-mm-diameter fused silica rod that was fractured in
four-point bending. Features typical of this kind of fracture are notedthe origin, as well as the mirror, mist, and
hackle regions. 60.
(Courtesy of George Quinn, National Institute of Standards and Technology, Gaithersburg, MD.)
Determination of Failure Mode
Of course, one critical element of an investigation involves determining the mecha-
nism(s) that was (were) responsible for the failure. There exist a large number of possi-
ble failure modes or mechanisms; some (but certainly not all) are listed as follows:
Ductile fracture
Brittle fracture
Fatigue (high-cycle, low-cycle, corrosion, thermal)
Corrosion (uniform, galvanic, pitting, crevice, etc.)
Stress-corrosion cracking
Distortion (elastic and plastic)
Creep and creep rupture
Liquid-metal embrittlement
Hydrogen embrittlement
Radiation damage
Preceding treatments (viz. macroscopic, microscopic, and metallographic examinations)
as well as sections in the printed text discuss characteristics of some of these failure
modes that may be used for making a reliable identification. References at the end of
this module provide additional instruction on the determination of failure mode.
Chemical Analyses
A chemical analysis of the failed material may also be necessary in the investigation.
Normally, this is one of the last procedures conducted inasmuch as it requires the
destruction of some of the failure specimen. One reason for doing a chemical (or com-
positional) analysis is to certify that the failed part was fabricated from the correct
material; in most instances, slight deviations from the specified composition are not crit-
ical. Chemical analyses may also be performed on corrosion products and other deposits
associated with the failure. The presence of gaseous elements (e.g., hydrogen, oxygen,
nitrogen) may have deleterious effects on the mechanical properties of some metal alloys;
detection of these elements in concentrations above acceptable limits is desirable.
Techniques utilized for chemical analyses include the following: wet chemistry
routes, emission spectrography, spectrophotometry, atomic-absorption spectroscopy,
x-ray diffraction, x-ray fluorescence spectroscopy, infrared and ultraviolet spectroscopy,
Auger electron spectroscopy, energy and wavelength-dispersive x-ray spectroscopy
(with SEMs), and electron microprobe analysis.
Application of Fracture Mechanics
Principles of fracture mechanics (Section M.4) are used to quantify the relationship
between the fracture toughness of a material, applied stress level, and the size of crack-
producing flaws (Equation M.20). Utilization of these principles allows the evaluation of
structural reliability and prediction of service lifetimes when there is the possibility of
ductile or brittle fracture, as well as failure by fatigue.
Identification of Possible Root Causes
Once an investigator has acquired a clear understanding of the failure (using the tech-
niques described in the preceding paragraphs), the next step is to clearly and objectively
identify and analyze all possible root causes. One organizational tool that is especially use-
ful to accomplish this goal is a fault tree. A fault tree is simply a taxonomic chart that shows
the interrelationships among a single major root cause and all possible sources (or causes)
that can lead up to (or are responsible for) this root cause. For example, Figure M.60 is a
fault tree for a dye penetrant defect (a root cause) that was observed in a failed forging;
shown here is a hierarchy of possible causes for the root cause. There are three primary
possible sources (or causes) of this defect [in the boxes labeled (1) through (3)], and
for each of these defects there is a least one (secondary) cause, and in some instances also
tertiary and quaternary causes. To each cause (or box) is assigned a number (or numbers)
and in some cases a letter; this scheme helps to organize the failure investigation.
There are a number of approaches that may be used to design and construct
a comprehensive and appropriately organized fault tree; we suggest the following pro-
cedure:
Define all possible root causes. Techniques that may be employed are discussed
previously in this section.
Analyze each proposed root cause as to its possible causes; this will generate a set
of primary causes. Such analyses are possible only if all information about the
failed part has been gatherede.g., details of its design, the manner in which it
was fabricated, specifics of its operational history. On occasion, experts in these
areas may need to be consulted.
Brainstorming sessions are also necessary when performing these analyses.
Team members should meet, question, and thoroughly discuss the reason(s) why
each primary cause could have occurred. This process will generate a set of secondary
causes, which are then entered on the fault tree. Continuation of this procedure
will produce sets of tertiary, quaternary, etc. causes, and, hopefully, the ultimate
discovery of all causing effects (and the generation of a complete fault tree).
Casting Process
Defects
(1)
Forging Process
Defects
(2)
Burst
(2A)
Hot
Short
(2B)
Defect Cavity
Enhancement
(2C)
Heat
Treatment
Process
Defects
(3)
Quench
Cracking
(3A)
Adiabatic
Heating
(2B1)
Localized
Chemistry
(2B2)
Forging
Temperatures
(2A1 and 2C1)
Forging
Strain Rate
(2A2 and 2C2)
Forging
Strain Direction
(2A3 and 2C3)
Nonmetallic
(1A)
Shrinkage
Porosity
(1C)
Hydrogen
Porosity
(1B)
Hydrogen
Content
(1B1)
Casting
Temperature
(1C1) Chemical
Composition
(1A1)
Filtering
Processes
(1A2)
Ingot Drop
Speed
(1C2)
Metal Flow
Rate
(1C3)
Why Do the Forgings Have
Penetrant Defects?
Figure M.60 Fault tree
diagram for a dye penetrant
defect in a forging that failed.
(Adapted from D. P. Dennies,
How to Organize and Run a
Failure Investigation, ASM
International, 2005, p. 109.
Reprinted with permission of ASM
International
. All rights reserved.

www.asminternational.org.)
The fault tree is not to be treated as a static document. Rather, it should continue
to change and improve as new information is uncovered throughout the course of
the investigation.
Once the construction of a fault tree has pretty much stabilized, two additional
questions should still be addressed: (1) What is unique or different about this
failure? and (2) Are there any things that are still missing from our investigation?
There are several distinct advantages for constructing and utilizing a fault tree in failure
investigations:
First and foremost, it is a problem-solving guide. A fault tree is essentially an out-
line that may be used to generate a set of procedures and evaluations designed
and planned to either prove or disprove each possible root cause.
The fault tree organizes the complex problem of determining each root cause of a
failure into a series of simpler and more manageable components.
It is a form of documentation or a record of how the failure investigation was
organized and executed.
Identification of the Root Cause Responsible for Failure
Once the fault tree (or trees) has (have) been constructed, it next becomes necessary
to objectively evaluate the likelihood that each of the possible root causes was respon-
sible for the failure. The first step in this process is to assess the probability of each of
the root causes in the fault treeis it likely, possible, or unlikely? In this regard, it is
important to rely on the technical expertise of members of the investigation team. The
next step is prioritize the order in which each of the possible root causes is to be
investigated; in all likelihood, time, financial, and/or personnel resources will not permit
testing of them all. These investigations will undoubtedly involve some of the analytical
procedures discussed in the previous section (e.g., microscopic/metallographic exami-
nations, NDT, mechanical testing, etc.). At this time the rationale that was used to assign
the probability and priority to each root cause should be documented. It is sometimes
convenient to summarize and record the information formulated in these three steps
in chart form.
It is now time to create a plan that will either prove or disprove each of the possi-
ble root causesthat is, prepare a schedule of appropriate tests and analyses based on
the probability/prioritization scheme described in the preceding paragraph. Then, for
each root cause, it is necessary to make a list of what physical evidence is required to
substantiate that this root cause occurred or was present. In preparing this list, one
should keep in mind that it is sometimes easier to disprove a root cause than to prove it.
The next step is to decide which of the procedures given under the Failure Analysis
Procedures heading are to be used to provide this physical evidence. Responsible per-
sonnel are then assigned to perform these test/analysis procedures, and completion
deadlines are set. And, finally, upon completion of the tests/analyses, results are tabu-
lated; these results are the basis for the final decision as to which of the possible root
cause(s) is (are) most likely responsible for the failure. This planning and testing phase
of the investigation may take weeks or even months to carry out.
The above discussion has dealt with the determination of the physical root cause.
As noted above, there is also the possibility that human or latent root causes are
responsible for this physical one; furthermore, it is important that the investigation be
designed to also ascertain whether or not either of these factors is the real cause of
failure.
At this point we have finally answered the second question that was presented at
the outset of our discussion on failure analysis: What is the root cause of the failure
problem?
Now, after the completion of this laborious and time-consuming root-cause-determination
phase, the investigation is at the stage where possible corrective actions may be
ascertainedthat is, it is possible to answer the What are possible solutions? question
(Figure M.48). One approach is to identify corrective actions for the most likely root
cause(s) responsible for the failure. This is possible by creation of a corrective action
tree, which is formatted after the manner of the fault tree. For example, for the forging
penetrant defects fault tree, Figure M.60, the corresponding corrective action tree
might appear as in Figure M.61. Here it is assumed that the primary root cause was a
casting process defect (which is the reason that most of the other box lines in the dia-
gram are dashed). The proposed corrective actions listed in Figure M.61 would come
from brainstorming sessions involving members of the investigation team.
This action tree approach may also be used for suggesting preventative actions.
Whereas a corrective action is taken to ensure that the root cause of failure does not
happen again, the purpose of a preventative action is to reveal the presence of a root
cause so that failure does not occur. For example, in Figure M.61 there is a fourth
primary corrective action NDE Improvements (i.e., nondestructive evaluation im-
provements), whereas there are only three primary root causes in the fault tree of
Figure M.60. Here, nondestructive evaluation improvements is really a preventative
action.
M.21 WHAT ARE POSSIBLE SOLUTIONS?
Casting Process
Improvements
(1)
Forging Process
Improvements
(2)
Heat Treatment
Process
Improvements
(3)
NDE
Improvements
(4)
Prevent
Hydrogen
Pickup
(3A)
Fumace
Atmosphere
(2C1 and 3A1)
Coat the
Forgings
(2C2 and 3A2)
Develop
Rating
System
(4A2)
Visual, Dye
Penetrant or
Ultrasonic
Inspection
(4A1)
Improved
Ultrasonic
Inspection of
Forgings
(4B)
Immersion
Tank
(4B1)
Add
Shear Wave
(4B2)
Develop Ingot
Slice Exam
(4A)
Measure H
2
in
All Ingots and
Forgings
(4C)
Improve
Forging
Reduction
(2A)
Add
Double
Cross
Work
(2B)
Forging
Temperatures
(2A1)
Improve Forging
Strain Rate
(2A2)
Improve Forging
Strain Direction
(2A3)
Improve
Forging Heat-
Up Process
(2C)
Nonmetallic
Inclusions
Content
Reduction
(1A)
H
2
Content
Reduction
(1B1)
Casting
Temperature
(1C1)
Ingot Drop
Speed
(1C2)
Metal Flow
Rate
(1C3)
LIMCA
Device
(1A1)
Improve
Filtering
Processes
(1A2)
Use New
Filters
(1A2A)
Increase
Fluxing and
Settling
Time
(1B1A)
Hydrogen
Porosity
Reduction
(1B)
Shrinkage
Porosity
Reduction
(1C)
What Is the Corrective Action for the
Forgings with Penetrant Defects?
Figure M.61 Corrective action tree diagram for the fault tree diagram of Figure M.60.
(Adapted from D. P. Dennies, How to Organize and Run a Failure Investigation, ASM International, 2005, p. 119. Reprinted with
permission of ASM International
. All rights reserved. www.asminternational.org.)

M.24 The Final Report 71
Finally, the investigation has progressed to a point such that the last of our four ques-
tions may be answeredWhich of the possible solutions is the best? The investigation
team should first objectively evaluate each of the corrective actions in the corrective ac-
tion tree of Figure M.61, with regard to its likelihood of remedying the root cause, and,
in addition, feasibility, cost effectiveness, and implementation time. On the basis of this
evaluation a decision is made as to the optimal corrective action(s) that should be taken.
Of course, if the real root cause of failure was determined to be due to a human or
latent factor (as opposed to a physical cause), the proposed solution should include cor-
rective actions appropriate to remedy the fundamental problem.
A set of recommendations should be carefully formulated for the purpose of elimi-
nating future failures. These recommendations might include some of the following
measures:
Design changes
Metallurgical alterations
Changes in manufacturing
Improvement of quality control
Improvement of repair procedures
Use of warning labels
M.22 WHICH OF THESE IS THE BEST SOLUTION?
Follow-up plans should be made for evaluating the effectiveness of the corrective
action(s) called for. This evaluation should come after a specified time period subsequent
to implementation as called for by the investigation team. Furthermore, it may be nec-
essary to conduct periodic evaluations.
M.23 EFFECTIVE EVALUATION OF CORRECTIVE ACTIONS
After completion of the failure investigation, it is imperative that a final report be pre-
pared. This should be done as soon as possible so that important details regarding the
investigation are not forgotten and, therefore, omitted. Furthermore, the customer is
entitled to and should expect to receive a report.
The first section should be a summary of the investigation, its results, conclusions,
and recommendations. It has been suggested that the remainder of the report be divided
into the following sections:
1. Description of the failed component
2. Service conditions at the time of failure
3. Prior service history
4. Manufacturing and processing history of the component
5. Mechanical structural analysis of the failed part
6. Metallurgical study of the failure
7. Metallurgical evaluation of quality
8. Summary of mechanisms that caused failure
9. Recommendations for prevention of similar failures or for correction of similar
components in service
10. Appendixto include any figures, tables, etc.
M.24 THE FINAL REPORT
This final report should be concise, clearly written, and logically organized. Further-
more, those parties on the investigation team who were responsible for making recom-
mendations should be provided the opportunity to review the report for accuracy and
any omissions.
We now demonstrate the use of some of these investigative techniques and principles
with the following case study on the failure of an automobile rear axle.
20
This case study was taken from Lawrence Kashar, Effect of Strain Rate on the Failure Mode of a Rear Axle,
Handbook of Case Histories in Failure Analysis, Vol. 1, pp. 7478, ASM International, Materials Park, OH, 1992.
Failure of an Automobile Rear Axle (Case Study)
20
M.25 INTRODUCTION
After an accident in which a light pickup truck left the road and overturned, it was noted
that one of the rear axles had failed at a point near the wheel mounting flange. This axle
was made of a steel that contained approximately 0.3 wt% C. Furthermore, the other
axle was intact and did not experience fracture. An investigation was carried out to
determine whether the axle failure caused the accident or whether the failure occurred
as a consequence of the accident.
Figure M.62 is a schematic diagram that shows the components of a rear axle
assembly of the type used in this pickup truck. The fracture occurred next to the bearing
lock nut, as noted in this schematic. A photograph of one end of the failed axle shaft is
presented in Figure M.63a, and Figure M.63b is an enlarged view of the other fractured
piece that includes the wheel mounting flange and the stub end of the failed axle. Here
(Figure M.63b), note that a keyway was present in the area of failure; furthermore,
threads for the lock nut were also situated next to this keyway.
Figure M.62 Schematic diagram showing typical components of a light truck axle and the fracture site for the
failed axle of this case study.
(Reproduced from MOTOR Auto Repair Manual, 39th Edition Copyright 1975. By permission of the Hearst Corporation.)
Axle shaf
Point of failure
Upon examination of the fracture surface it was noted that the region correspon-
ding to the outside shaft perimeter [being approximately 6.4 mm (0.25 in.) wide] was
very flat; furthermore, the center region was rough in appearance.
Let us take a moment to apply some of the failure investigation principles discussed
in Sections M.19 through M.22specifically in addressing the questions that were
posed.
The answer to the first question (What exactly is the failure problem?) is a
fractured rear axle of a pickup truck.
For the second question (What is the root cause of the failure problem?) there
are really only two possible answers: either (1) the accident caused the axle
failure, or (2) the axle failure caused the accident.
Inasmuch as the purpose of this investigation is to identify the root cause, it is not
necessary to find answers to the last two questions: What are possible solutions?
and Which of these is the best solution?
Let us now explore the analytical procedures that were utilized to determine the
root cause of the axle failure.
Figure M.63 (a) Photograph of one section of the failed axle. (b) Photograph showing wheel mounting flange
and stub end of failed axle.
[Reproduced with permission from Handbook of Case Histories in Failure Analysis, Vol. 1 (1992), ASM International, Materials
Park, OH, 44073-0002.]
(a) (b)
Details of the fracture surface in the vicinity of the keyway are shown in the photograph
of Figure M.64; note that the keyway appears at the bottom of the photograph. Both the
flat outer perimeter and rough interior regions may be observed in the photograph.
There are chevron patterns that emanate inward from the corners of and parallel to the
sides of the keyway; these are barely discernible in the photograph but indicate the
direction of crack propagation.
Fractographic analyses were also conducted on the fracture surface. Figure M.65
shows a scanning electron micrograph taken near one of the keyway corners. Cleavage
features may be noted in this micrograph, whereas any evidence of dimples and fatigue
striations is absent. These results indicate that the mode of fracture within this outer
periphery of the shaft was brittle.
M.26 TESTING PROCEDURE AND RESULTS
An SEM micrograph taken of the rough central region (Figure M.66) revealed the
presence of both brittle cleavage features and also dimples; thus, it is apparent that the
failure mode in this central interior region was mixed; that is, it was a combination of
both brittle and ductile fracture.
Metallographic examinations were also performed. A transverse cross section of the
failed axle was polished, etched, and photographed using the optical microscope. The
microstructure of the outer periphery region, as shown in Figure M.67, consisted of tem-
pered martensite.
21
On the other hand, the microstructure in the central region was com-
pletely different, per the photomicrograph of Figure M.68. It may be noted that the
microconstituents are ferrite, pearlite, and possibly some bainite.
22
In addition, transverse
microhardness measurements were taken along the cross section; Figure M.69 is a plot
of the resulting hardness profile. Here it may be noted that the maximum hardness of
approximately 56 HRC occurred near the surface, and that hardness diminished with
radial distance to a hardness of about 20 HRC near the center. On the basis of the
observed microstructures and this hardness profile, it was assumed that the axle had
been induction hardened.
23
The results of these fractographic/metallographic analyses and hardness tests are
summarized in Table M.7.
Figure M.64 Optical micrograph of failed section of axle
that shows the keyway (bottom), as well as the flat outer-
perimeter and rough core regions.
[Reproduced with permission from Handbook of Case Histories in
Failure Analysis, Vol. 1 (1992), ASM International, Materials Park, OH,
44073-0002.]
2 m
Figure M.65 Scanning electron
micrograph of failed axle outer-perimeter
region near the keyway, which shows
cleavage features. 3500.
[Reproduced with permission from Handbook of
Case Histories in Failure Analysis, Vol. 1 (1992), ASM
International, Materials Park, OH, 44073-0002.]
21
For a discussion of tempered martensite, see Section 11.8.
22
Ferrite, pearlite, and bainite microconstituents are discussed in Sections 11.5 and 11.7.
23
With induction hardening, the surface of a piece of medium-carbon steel is rapidly heated using an induction
furnace. The piece is then quickly quenched so as to produce an outer surface layer of martensite (which is subsequently
tempered), with a mixture of ferrite and pearlite at interior regions.
Figure M.66 Scanning electron micrograph of the
failed axle rough core region, which is composed of
mixed cleavage and dimpled regions. 570.
[Reproduced with permission from Handbook of Case
Histories in Failure Analysis, Vol. 1 (1992), ASM International,
Materials Park, OH, 44073-0002.]
20 m
Figure M.67 Optical photomicrograph of the failed
axle outer-perimeter region, which is composed of
tempered martensite. 500. [Note: Although the
microstructure for tempered martensite in this figure
appears to be different from that shown in Figure 11.34,
they are nevertheless the same. One reason for this
disparity in appearance is due to the difference in
magnification of the micrographs: Figure 11.34 is an
electron micrograph with a magnification that is
approximately twenty times greater than this optical
micrograph. Furthermore, the dark regions of this
photomicrograph are clusters of Fe
3
C particles (which
are unresolved) that stand in relief above the etched
a-ferrite matrix, which appears light.]
[Reproduced with permission from Handbook of Case Histories
in Failure Analysis, Vol. 1 (1992), ASM International, Materials
Park, OH, 44073-0002.]
Ferrite
Pearlite
20 m
Figure M.68 Optical
photomicrograph of the failed
axle core region, which is
composed of ferrite and pearlite
(and possibly bainite). 500.
[Reproduced with permission from
Handbook of Case Histories in
Failure Analysis, Vol. 1 (1992), ASM
International, Materials Park, OH,
44073-0002.]
At this point in the investigation it was not possible to ascertain irrefutably whether
the accident caused the axle failure (scenario 1) or whether the axle fracture caused the
accident (scenario 2). The high hardness and, in addition, the evidence of cleavage of the
outer surface layer indicated that this region failed in a brittle manner as a result of
being overloaded (i.e., as a result of the accident, scenario 1). On the other hand, the
evidence of a mixed ductile-brittle mode of fracture in the central region neither sup-
ported nor refuted either of these two failure scenarios.
24
Assuming the validity of the first scenario, it was hypothesized that the core region
was strain-rate sensitive to fracture; that is, at high strain rates, as with the truck rollover,
the fracture mode would be brittle. By contrast, if failure resulted from loads that were
applied relatively slowly, as under normal driving conditions (the second scenario), the
mode of failure would be more ductile.
In order to explore the feasibility of scenario 1 (i.e., the strain-rate sensitivity of the
core region), it was decided to fabricate and test impact specimens taken from both
outer-perimeter and core regions; in addition, a tension test was to be conducted on a
core-region specimen. Failure surfaces for all three specimens were to be subjected to
SEM examinations. The following results from these tests/examinations would be
expected if the core region of the axle were sensitive to the rate of straining:
The failure of the core-region specimen to be impact (high strain rate) tested
would not be totally ductile in nature.
The core-region specimen to be tensile (low strain rate) tested would display at
least a moderate degree of ductility.
Analytical Result for Result for
Technique Outer Region Core Region
Fractographic Cleavage features Cleavage features/dimples
(brittle fracture) (ductile/brittle fracture)
Metallographic Tempered martensite Ferrite, pearlite (bainite?)
Hardness tests (profile) Induction hardened
(i.e., heat treatment)
Table M.7
Tabulation of Test
Results on Specimens
Taken from Failed
Rear Truck Axle
24
It is at this stage of an investigation that a fault tree may be implemented. We have chosen not to use a fault tree
due to the simplicity of this failure problem.
Distance from outer surface (in.)
H
a
r
d
n
e
s
s

(
c
o
n
v
e
r
t
e
d

t
o

R
o
c
k
w
e
l
l

C
)
0.0 0.2 0.4 0.6 0.8
10
20
30
40
50
60 Figure M.69 Transverse hardness profile across
the axle cross section. (Microhardness readings were
converted to Rockwell C values).
[Reproduced with permission from Handbook of Case
Histories in Failure Analysis, Vol. 1 (1992), ASM
Figure M.70 Fracture surface of the
Charpy impact specimen that was taken
from the outer-perimeter region.
[Reproduced with permission from Handbook
of Case Histories in Failure Analysis, Vol. 1
(1992), ASM International, Materials Park, OH,
44073-0002.]
Figure M.71 Fracture surface of the
Charpy impact specimen that was taken
from the core region.
[Reproduced with permission from Handbook
of Case Histories in Failure Analysis, Vol. 1
(1992), ASM International, Materials Park,
OH, 44073-0002.]
The outer-perimeter specimen to be impact tested would fail in a totally brittle
manner.
Impact Tests
For the impact tests, small [~2.5-mm- (0.1-in.-) wide] Charpy V-notch test specimens
were prepared from both outer perimeter and interior areas. Because the hardened
outer region was very thin (6.4 mm thick), careful machining of these specimens was re-
quired. Impact tests were conducted at room temperature, and the energy absorbed by
the surface specimen was significantly lower than for the core specimen [4 J (3 ft-lb
f
)
versus 11 J (8 ft-lb
f
)]. Furthermore, the appearances of the fracture surfaces for the two
specimens were dissimilar. Very little, if any, deformation was observed for the outer-
perimeter specimen (Figure M.70); conversely, the core specimen deformed significantly
(Figure M.71).
Fracture surfaces of these impact specimens were then subjected to examination
using the SEM. Figure M.72, a micrograph of the outer-perimeter specimen that was im-
pact tested, reveals the presence of cleavage features, which indicates that this was a brit-
tle fracture. Furthermore, the morphology of this fracture surface is similar to that of the
actual failed axle (Figure M.65).
For the impact specimen taken from the core region the fracture surface had a much
different appearance; Figures M.73a and M.73b show micrographs for this specimen,
which were taken at relatively low and high magnifications, respectively. These micro-
graphs reveal the details of this surface to be composed of interspersed cleavage features
and shallow dimples, being similar to the failed axle, as shown in Figure M.66. Thus, the
fracture of this specimen was of the mixed-mode type, having both ductile and brittle
components.
Tensile Tests
A tensile specimen taken from the core region was pulled in tension to failure. The
fractured specimen displayed the cup-and-cone configuration, which indicated at
least a moderate level of ductility. A fracture surface was examined using the SEM,
and its morphology is presented in the micrograph of Figure M.74. The surface was
5 m
micrograph of the fracture surface for the
impact specimen prepared from the outer-
perimeter region of the failed axle. 3000.
Case Histories in Failure Analysis, Vol. 1 (1992), ASM
100 m
Figure M.73 (a) Scanning electron micrograph of the fracture surface for the impact specimen prepared from the
core region of the failed axle. 120. (b) Scanning electron micrograph of the fracture surface for the impact specimen
prepared from the core region of the failed axle taken at a higher magnification than (a); interspersed cleavage and
dimpled features may be noted. 3000.
[Reproduced with permission from Handbook of Case Histories in Failure Analysis, Vol. 1 (1992), ASM International, Materials
Park, OH, 44073-0002.]
(a) (b)
composed entirely of dimples, which confirms that this material was at least moder-
ately ductile and that there was no evidence of brittle fracture. Thus, although this
core material exhibited mixed-mode fracture under impact loading conditions, when
the load was applied at a relatively slow rate (as with the tensile test), failure was
highly ductile in nature.
A summary of the results of these impact and tensile tests is presented in Table M.8.
M.27 Discussion 79
2 m
micrograph of the fracture surface for the
core specimen that was tensile tested; a
completely dimpled structure may be noted.
Approximately 3500.
Case Histories in Failure Analysis, Vol. 1 (1992),
ASM International, Materials Park, OH,
44073-0002.]
Table M.8 Tabulation of Impact and Tension Test Results for Specimens Taken from Core
and Outer Regions of Failed Truck Rear Axle
Specimen/Test Fracture Mode Fractographic Features
Core region/impact Some ductility Dimples and cleavage features
Outer region/impact Brittle Cleavage features
Core region/tension Ductile Dimples
In light of the previous discussion, it was supposed that the truck rollover was responsible
for the axle failure (i.e., scenario 1 was valid). Reasons for this supposition are as follows
(see Tables M.7 and M.8):
1. The outer-perimeter region of the failed axle shaft failed in a brittle manner, as did
also the specimen taken from this region that was impact tested. This conclusion
was based on the fact that both fracture surfaces were very flat and that SEM
micrographs revealed the presence of cleavage features.
2. The fracture behavior of the core region was strain-rate sensitive and indicated that
axle failure was due to a single high strain-rate incident. Fracture surface features
for both the failed axle and impact-tested (i.e., high-strain-rate-tested) specimens
taken from this core region were similar: SEM micrographs revealed the presence
M.27 DISCUSSION
of features (cleavage features and dimples) that are characteristic of mixed-mode
(brittle and ductile) fracture.
In spite of evidence supporting the validity of the accident-caused-axle-failure scenario,
the plausibility of the other (axle-failure-caused-the-accident) scenario (scenario 2) was
also explored. This latter scenario assumes that a fatigue crack or some other slow-crack
propagation mechanism initiated the sequence of events that caused the accident. In this
case it is important to consider the mechanical characteristics of the portion of the spec-
imen that was last to failin this instance, the core region. If failure was due to fatigue,
then any increase in loading level of this core region would have occurred relatively
slowly, not rapidly as with impact loading conditions. During this gradually increasing
load level, fatigue crack propagation would have continued until a critical length was
achieved (i.e., until the remaining intact axle cross section was no longer capable of sus-
taining the applied load); at this time, final failure would have occurred.
On the basis of the tensile tests (i.e., slow strain-rate tests) performed on this core
region, the appearance of the axle fracture surface would be entirely ductile (i.e., dim-
pled, as per the SEM micrograph of Figure M.74). Inasmuch as this core region of the
failed shaft exhibited mixed-mode (ductile and brittle) fracture features (both cleavage
features and dimples, Figure M.66) and not exclusively dimples, the axle-failure-caused-
the-accident scenario was rejected.
For a torsionally stressed cylindrical shaft, an expression for strength performance
index was derived (Equation M.9).
Using the appropriate materials selection chart (log strength versus log density, Figure
M.2) a preliminary candidate search was conducted. From the results of this search,
several candidate engineering materials were ranked on both strength-per-unit mass
and cost bases.
The significant discrepancy between actual and theoretical fracture strengths of brittle
materials is explained by the existence of small flaws that are capable of amplifying an
applied tensile stress in their vicinity, leading ultimately to crack formation. Fracture
ensues when the theoretical cohesive strength is exceeded at the tip of one of these flaws.
The maximum stress that may exist at the tip of a crack (oriented as in Figure M.4a)
is dependent on crack length and tip radius, as well as the applied tensile stress
according to Equation M.12a.
An expression was developed by A. A. Griffith for a crack propagation critical stress
in brittle materials (Equation M.14), which is a function of elastic modulus, specific
surface energy, and crack length. Fracture ensues when this critical stress is exceeded
at the tip of a preexisting flaw or crack.
Sharp corners may also act as points of stress concentration and should be avoided
when designing structures that are subjected to stresses.
There are three different crack displacement modes (Figure M.6): opening (tensile),
sliding, and tearing.
A condition of plane strain exists when specimen thickness is much greater than crack
lengththat is, there is no strain component perpendicular to the specimen faces.
Stress distributions in front of an advancing crack may be expressed in terms of posi-
tion (as radial and angular coordinates) as well as stress intensity factor, K.
The fracture toughness of a material (or critical value of the stress intensity factor) is
indicative of its resistance to brittle fracture when a crack is present. For the plane strain
SUMMARY
Materials Selection
for a Torsionally
Stressed Cylindrical
Shaft
Fracture
Principles of
Fracture Mechanics
Summary 81
situation (and mode I loading) it is dependent on applied stress, crack length, and the
dimensionless scale parameter Y as represented in Equation M.22:
The units for K
Ic
are .
Minimum specimen thickness for the condition of plane strain (B) is a function of
fracture toughness and yield strength according to Equation M.23.
K
Ic
is the parameter normally cited for design purposes; its value is relatively large for
ductile materials (and small for brittle ones) and is a function of microstructure, strain
rate, and temperature.
With regard to designing against the possibility of fracture, consideration must be given
to material (its fracture toughness), the stress level, and the flaw size detection limit.
Several nondestructive testing (NDT) techniques for detecting and measuring preex-
isting cracks were discussed brieflyviz.
Visual
Optical and scanning electron microscopic
Dye penetrant
Magnetic particle
Radiographic
Ultrasonic (pulse-echo)
Acoustic emission
The procedure and details of ASTM Standard E 399-09 for plane-strain fracture
toughness testing of single-edge notched bend and compact tension specimens were
discussed.
A conditional value of plane-strain fracture toughness is determined from a plot of
load versus crack displacement.
Validation of this conditional value depends on the satisfaction of two criteria:
One criterion (Equation M.32) ascertains whether the material being tested
is too ductile to yield a valid value.
The second criterion (Equation M.37) determines if a condition of plane
strain is met.
Three factors that may cause a metal to experience a ductile-to-brittle transition are
exposure to stresses at relatively low temperatures, high strain rates, and the presence
of a sharp notch.
Qualitatively, the fracture behavior of materials may be determined using Charpy and
Izod impact testing techniques (Figure M.19).
On the basis of the temperature dependence of measured impact energy (or appear-
ance of the fracture surface), it is possible to ascertain whether a material experiences
a ductile-to-brittle transition and the temperature range over which such a transition
occurs.
Low-strength steel alloys typify this ductile-to-brittle behavior and, for structural
applications, should be used at temperatures in excess of the transition range.
Furthermore, low-strength FCC metals, most HCP metals, and high-strength materials
do not experience this ductile-to-brittle transition.
Fatigue is a common type of catastrophic failure in which the applied stress level fluc-
tuates with time; it may occur when the maximum stress level is considerably lower
than the static tensile or yield strength.
MPa1m 1ksi 1in. 2
K
Ic
Ys1pa
Fracture Toughness
Testing
Impact Fracture
Testing
Fatigue
Fluctuating stresses are categorized into three general stress-versus-time cycle modes:
reversed, repeated, and random (Figure M.24). Reversed and repeated modes are
characterized in terms of mean stress, range of stress, and stress amplitude.
In-service conditions that should be replicated in a fatigue test are stress type, time
frequency, and stress pattern.
Test data are plotted as stress (normally stress amplitude) versus the logarithm of the
number of cycles to failure.
For many metals and alloys, stress decreases continuously with increasing number of
cycles at failure; fatigue strength and fatigue life are parameters used to characterize
the fatigue behavior of these materials (Figure M.26b).
For other metals (e.g., ferrous and titanium alloys), at some point, stress ceases to de-
crease with, and becomes independent of, the number of cycles; the fatigue behavior
of these materials is expressed in terms of fatigue limit (Figure M.26a).
Fatigue cracks normally nucleate on the surface of a component at some point of
stress concentration.
Crack propagation proceeds in two stages (Figure M.28), which are characterized by
propagation direction and rate. The mechanism for the more rapid stage II corre-
sponds to a repetitive plastic blunting and sharpening process at the advancing crack
tip (Figure M.29).
Two characteristic fatigue surface features are beachmarks and striations.
Beachmarks form on components that experience applied stress interruptions;
they normally may be observed with the naked eye.
Fatigue striations are of microscopic dimensions, and each is thought to represent
the crack tip advance distance over a single load cycle.
An analytical expression (Equation M.43) was proposed for fatigue crack propaga-
tion rate in terms of the stress intensity factor range at the crack tip. Integration of the
expression yields an equation whereby fatigue life may be estimated.
The Goodman equation (Equation M.49) may be used to estimate fatigue life for a
nonzero mean stress.
Measures that may be taken to extend fatigue life include the following:
Reducing the mean stress level
Eliminating sharp surface discontinuities
Improving the surface finish by polishing
Imposing surface residual compressive stresses by shot peening
Case hardening by using a carburizing or nitriding process
Thermal stresses may be induced in components that are exposed to elevated tem-
perature fluctuations and when thermal expansion and/or contraction is restrained;
fatigue for these conditions is termed thermal fatigue.
The presence of a chemically active environment may lead to a reduction in fatigue
life for corrosion fatigue. Measures that may be taken to prevent this type of fatigue
include the following:
Application of a surface coating
Utilization of a more corrosion-resistant material
Reducing the corrosiveness of the environment
Reducing the applied tensile stress level
Imposing residual compressive stresses on the surface of the specimen
Cyclic Stresses
The SN Curve
Crack Initiation and
Propagation
Factors That Affect
Fatigue Life
Environmental
Effects
Summary 83
A stress analysis was first performed on a helical spring, which was then extended to
an automobile valve spring. Since the valve spring is subjected to cyclic loading, the
possibility of fatigue failure is crucial to its performance.
The results of this analysis included computation of the shear stress amplitude, the
magnitude of which was almost identical to the calculated fatigue limit for a
chromevanadium steel that is commonly used for valve springs. It was noted that the
fatigue limit of valve springs is often enhanced by shot peening.
Finally, a procedure was suggested for assessing the economic feasibility of this spring
design incorporating the shot-peened chromevanadium steel.
The five possible reasons for failure were presented and explained briefly:
Design errors
Fabrication/manufacturing defects
Assembly errors
Misuse during operation
Improper maintenance
In addition, the three root (or actual) causes of failure were discussed; these root causes are
Physicalfailure is related to a type of mechanism discussed in the module (e.g.,
fracture, fatigue)
Humanthe actions of an individual are the primary causes of failure
Latentfailure is a result of organizational policy
A failure investigation seeks to find answers to four questions:
What exactly is the failure problem?
What is the root cause of the failure problem?
What are possible solutions?
Which of these is the best solution?
With regard to determination of the root cause, a number of procedures are available,
including:
Nondestructive testing
Mechanical testing (tensile, hardness, impact)
Microscopic and macroscopic examination of fracture surfaces
Metallographic examinations of microstructure
Determination of the stress state
Determination of fracture mode
Chemical analyses
Application of fracture mechanics
The complex problem of determining possible root causes of failure is simplified by
using a fault treea taxonomic chart that displays the interrelationships among a
major root cause and its subordinate root causes.
Identification of the root cause may be determined by evaluating the likelihood of
each of the possible root causes in the fault tree. This is possible using the tests/analy-
ses detailed above.
Formulation of possible solutions (including the best solution) to a failure problem may be
accomplished utilizing a corrective action treea chart constructed from (and in the form
of) the fault tree, which consists of corrective actions for each of the fault tree entries.
Several procedures were employed in this failure investigation in order to determine
whether the accident caused the axle failure; these are listed as follows:
Fracture surfaces were visually inspected
SEM fractographic inspections were made on both outer perimeter and interior
core regions of the failed axle
Automobile Valve
Spring (Case Study)
Investigation of
Engineering
Failures
Failure of an
Automobile Rear
Axle (Case Study)
Metallographic examinations of both outer surface and interior regions, were
performed and the microconstituents present were identified
A hardness profile across the axle cross section was generated
Impact specimens taken from both surface and interior regions were prepared
and tested; fractographic examinations of both fractured specimens were also con-
ducted
A tensile specimen taken from the core region was prepared and tested to fracture.
Furthermore, its fracture surface was examined with an SEM.
On the basis of the outcomes of these procedures/analyses, it was concluded that
the automobile accident (i.e., truck rollover) was responsible for the rear axle
failure.
Equation Page
Number Equation Solving for Number
M.9 4
M.12a 10
M.12b 10
M.14 11
M.19 Stress intensity factor 14
M.20 Fracture toughness 14
M.22 15
M.23 15
M.24 Design (or critical) stress 18
M.25 18
M.38 Mean stress (fatigue tests) 34
M.39 34
M.40 34
M.41 Stress ratio (fatigue tests) 35
M.43 41
da
dN
A1 K2
m
R
s
min
s
max
s
a
s
max
s
min
2
s
r
s
max
s
min
s
m
s
max
s
min
2
a
c
1
p
a
K
Ic
sY
b
2
s
c
K
Ic
Y1pa
B 2.5 a
K
Ic
s
y
b
2
K
Ic
Ys1pa
K
c
Ys
c
1pa
K Ys1pa
s
c
a
2Eg
s
pa
b
1
2
2s
0
a
a
r
t
b
1
2
s
m
s
0
c 1 2 a
a
r
t
b
1
2
d
P
t
2
3
f
r
Equation Summary
Strength performance index for a torsionally
stressed cylindrical shaft
Maximum stress at tip of elliptically-shaped
crack
Critical stress for crack propagation in a brittle
material
Plane-strain fracture toughness, mode I crack
surface displacement
Minimum thickness for condition of
plane strain
Maximum allowable flaw size
Range of stress (fatigue tests)
Stress amplitude (fatigue tests)
Fatigue crack propagation rate
Summary 85
Equation Page
Number Equation Solving for Number
M.48
44
M.49 45
M.50 Thermal stress 47
M.52 and M.56 48, 49 t
d
c
Gd
pD
2
c 1.60a
D
d
b
0.140
d
s a
l
E T
s
fs
s
fs
0

a 1
s
m
TS
b
1
Ap
m
2
1 s2
m
a
c
a
0
da
Y
m
a
m
2
N
f
a
c
a
0
da
A1Ys2pa2
Fatigue life
Goodman equationfatigue strength for
nonzero mean stress
Shear stress experienced by a spring
List of Symbols
Symbol Meaning
A Material constant
a Length of a surface crack
D Spring coil center-to-center diameter
d Spring wire diameter
E Modulus of elasticity
G Shear modulus
N Number of stress cycles (fatigue)
TS Tensile strength
T Temperature difference or change
Y Dimensionless parameter or function
l
Linear coefficient of thermal expansion
c
Amount of deflection per spring coil
s
Specific surface energy
Density
t
Crack tip radius
Applied stress
Fatigue strength for zero mean stress
0
Applied tensile stress
max
Maximum stress (cyclic)
min
Minimum stress (cyclic)
f
Shear strength
s
fs
0

Important Terms and Concepts
case hardening
Charpy test
corrosion fatigue
ductile-to-brittle transition
fatigue
fatigue life
fatigue limit
fatigue strength
fracture mechanics
REFERENCES
Principles of Fracture Mechanics
Griffith, A. A., The Phenomena of Rupture Flow in Solids,
Philos. Trans. R. Soc. London, 221A, 163 (1920). This paper
was republished with annotations in Trans. ASM, 61, 871
(1968).
Inglis, C. E., Stresses in a Plate Due to the Presence of Cracks
and Sharp Corners, Trans. Inst. Nav. Archit., 55, 219
(1913).
Irwin, G. R., Fracture, in Encyclopedia of Physics, S. Flgge
(Editor), Vol. VI, Springer, Berlin, 551 (1958).
Nondestructive Testing (NDT)
Cartz, L., Nondestructive Testing: Radiography, Ultrasonics,
Liquid Penetrant, Magnetic Particle, Eddy Current, ASM
International, Materials Park, OH, 1995.
Hellier, C. J., Handbook of Nondestructive Evaluation,
McGraw-Hill, New York, 2001.
Mix, P. E., Introduction to Nondestructive Testing: A Training
Guide, 2nd edition, Wiley, Hoboken, NJ, 2005.
Shull, P. J. (Editor), Nondestructive Evaluation: Theory,
Techniques, and Applications, Marcel Dekker, New York,
2002.
General References on Failure and Fatigue
ASM Handbook, Vol. 11, Failure Analysis and Prevention, ASM
International, Materials Park, OH, 2002.
ASM Handbook, Vol. 12, Fractography, ASM International,
Materials Park, OH, 1987.
Colangelo, V. J., and F. A. Heiser, Analysis of Metallurgical
Failures, 2nd edition, Wiley, New York, 1987.
Collins, J. A., Failure of Materials in Mechanical Design, 2nd
edition, Wiley, New York, 1993.
Courtney, T. H., Mechanical Behavior of Materials, 2nd edition,
McGraw-Hill, Burr Ridge, IL, 2000.
Dieter, G. E., Mechanical Metallurgy, 3rd edition, McGraw-Hill,
New York, 1986.
Esaklul, K. A., Handbook of Case Histories in Failure Analysis,
ASM International, Materials Park, OH, 1992 and 1993. In
two volumes.
Hertzberg, R. W., Deformation and Fracture Mechanics of
Engineering Materials, 4th edition, Wiley, New York,
1996.
McEvily, A. J., Metal Failures: Mechanisms, Analysis, Prevention,
Wiley, Hoboken, NJ, 2002.
Murakami, Y. (Editor), Stress Intensity Factors Handbook,
Elsevier Science, New York, 2001.
Wulpi, D. J., Understanding How Components Fail, 2nd edition,
ASM International, Materials Park, OH, 1999.
Automobile Valve Spring
Juvinall, R. C., and K. M. Marshek, Fundamentals of Machine
Component Design, 4th edition, Chapter 12, Wiley,
Hoboken, NJ, 2005.
Shigley, J., C. Mischke, and R. Budynas, Mechanical Engineering
Design, 7th edition, Chapter 10, McGraw-Hill, New York,
2004.
Investigation of Engineering Failures
Dennies, D. P., How to Organize and Run a Failure Investigation,
ASM International, Materials Park, OH, 2005.
Lewis, P. R., K. Reynolds, and C. Gagg, Forensic Materials
Engineering Case Studies, CRC Press, Boca Raton,
FL, 2004.
QUESTIONS AND PROBLEMS
Note: Those problems having an asterisk (*) by their
numbers are identical to (or refreshed versions of)
problems found in the print text. A problem conver-
sion guide is provided in the section that follows the
design problems.
*M.1 What is the magnitude of the maximum stress
that exists at the tip of an internal crack having a
radius of curvature of 2.5 10
4
mm (10
5
in.)
and a crack length of 2.5 10
2
mm (10
3
in.)
when a tensile stress of 170 MPa (25,000 psi) is
applied?
*M.2 Estimate the theoretical fracture strength of a
brittle material if it is known that fracture occurs
by the propagation of an elliptically shaped sur-
face crack of length 0.25 mm (0.01 in.) and having
a tip radius of curvature of 1.2 10
3
mm (4.7
10
5
in.) when a stress of 1200 MPa (174,000 psi)
is applied.
*M.3 If the specific surface energy for soda-lime glass
is 0.30 J/m
2
, then using data in Table 7.1, compute
the critical stress required for the propagation of a
surface crack of length 0.05 mm.
*M.4 A polystyrene component must not fail when
a tensile stress of 1.25 MPa (180 psi) is applied.
fracture toughness
impact energy
Izod test
materials selection chart
nondestructive testing
performance index
plane strain
plane strain fracture
toughness
root cause (of failure)
stress raiser
thermal fatigue
Determine the maximum allowable surface
crack length if the surface energy of polystyrene
is 0.50 J/m
2
(2.86 10
3
in.-lb
f
/in.
2
). Assume a
modulus of elasticity of 3.0 GPa (0.435 10
6
psi).
M.5 The parameter Kin Equations M.18a, M.18b, and
M.18c is a function of the applied nominal stress s
and crack length a as
Compute the magnitudes of the normal stresses
s
x
and s
y
in front of a surface crack of length 2.0 mm
(0.079 in.) (as depicted in Figure M.7) in response
to a nominal tensile stress of 100 MPa (14,500 psi)
at the following positions:
(a) r 0.10 mm (3.9 10
3
in.), 0
(b) r 0.10 mm (3.9 10
3
in.), u 45
(c) r 0.50 mm (2.0 10
2
in.), u 0
(d) r 0.50 mm (2.0 10
2
in.), u 45
M.6 The parameter K in Equations M.18a, M.18b, and
M.18c is defined in Problem M.5.
(a) For a surface crack of length 2.0 mm (7.87
10
2
in.), determine the radial position at an angle
u of 30 at which the normal stress s
x
is 100
MPa (14,500 psi) when the magnitude of the nom-
inal applied stress is 150 MPa (21,750 psi).
(b) Compute the normal stress s
y
at this same
position.
M.7 Below is shown a portion of a tensile
specimen.
K s1pa
(a) Compute the magnitude of the stress at point
P when the externally applied stress is 140 MPa
(20,000 psi).
(b) How much will the radius of curvature at point
Phave to be increased to reduce this stress by 25%?
M.8 A cylindrical hole 19.0 mm (0.75 in.) in diameter
passes entirely through the thickness of a steel
plate 12.7 mm (0.5 in.) thick, 127 mm (5 in.) wide,
and 254 mm (10 in.) long (see Figure M.5a).
(a) Calculate the stress at the edge of this hole when
a tensile stress of 34.5 MPa (5000 psi) is applied in
a lengthwise direction.
(b) Calculate the stress at the hole edge when the
same stress in part (a) is applied in a widthwise
direction.
M.9 For each of the metal alloys listed in Table M.3,
compute the minimum component thickness for
which the condition of plane strain is valid.
*M.10 A specimen of a 4340 steel alloy having a plane
strain fracture toughness of 45 MPa
is exposed to a stress of 1000 MPa (145,000 psi).
Will this specimen experience fracture if the largest
surface crack is 0.75 mm (0.03 in.) long? Why or
why not? Assume that the parameter Y has a value
of 1.0.
*M.11 An aircraft component is fabricated from
an aluminum alloy that has a plane strain fracture
toughness of 35 MPa It has
been determined that fracture results at a stress of
250 MPa (36,250 psi) when the maximum (or criti-
cal) internal crack length is 2.0 mm (0.08 in.). For
this same component and alloy, will fracture occur
at a stress level of 325 MPa (47,125 psi) when the
maximum internal crack length is 1.0 mm (0.04 in.)?
Why or why not?
*M.12 Suppose that a wing component on an air-
craft is fabricated from an aluminum alloy that
has a plane strain fracture toughness of 40
MPa . It has been determined
that fracture results at a stress of 365 MPa
(53,000 psi) when the maximum internal crack
length is 2.5 mm (0.10 in.). For this same compo-
nent and alloy, compute the stress level at which
fracture will occur for a critical internal crack
length of 4.0 mm (0.16 in.).
*M.13 A large plate is fabricated from a steel alloy that
has a plane strain fracture toughness of
. If, during service use, the plate is ex-
posed to a tensile stress of 200 MPa (29,000 psi),
determine the minimum length of a surface crack
that will lead to fracture.Assume a value of 1.0 for Y.
150 ksi 1in. 2
55 MPa1m
1m 136.4 ksi 1in. 2
ksi 1in. 2 . 131.9 1m
141 ksi 1in. 2 1m
s
s
6.5 mm
20 mm
4 mm
40 mm
P
*M.14 Calculate the maximum internal crack length
allowable for a 7075-T651 aluminum alloy (Table
M.3) component that is loaded to a stress one-half its
yield strength. Assume that the value of Y is 1.35.
*M.15 A structural component in the form of a wide
plate is to be fabricated from a steel alloy that has
a plane strain fracture toughness of 77.0 MPa
and a yield strength of 1400 MPa
(205,000 psi).The flaw size resolution limit of the flaw
detection apparatus is 4.0 mm (0.16 in.). If the
design stress is one-half the yield strength and the
value of Y is 1.0, determine whether a critical flaw
for this plate is subject to detection.
M.16 A structural component in the shape of a flat
plate 25.4 mm (1.0 in.) thick is to be fabricated
from a metal alloy for which the yield strength and
plane strain fracture toughness values are 700
MPa (101,500 psi) and 49.5 MPa ,
respectively; for this particular geometry, the value
of Y is 1.65. Assuming a design stress one-half of
the yield strength, is it possible to compute the
critical length of a surface flaw? If so, determine
its length; if this computation is not possible from
the given data, then explain why.
Impact Fracture Testing
*M.17 The following tabulated data were gathered
from a series of Charpy impact tests on a ductile
cast iron:
Temperature (C) Impact Energy (J)
25 124
50 123
75 115
85 100
100 73
110 52
125 26
150 9
175 6
(a) Plot the data as impact energy versus temper-
ature.
(b) Determine a ductile-to-brittle transition tem-
perature as the temperature corresponding to the
average of the maximum and minimum impact
energies.
(c) Determine a ductile-to-brittle transition tem-
perature as the temperature at which the impact
energy is 80 J.
1m 145 ksi 1in. 2
170.1 ksi 1in. 2
1m
*M.18 The following tabulated data were gathered
from a series of Charpy impact tests on a tem-
pered 4140 steel alloy:
Temperature (C) Impact Energy (J)
100 89.3
75 88.6
50 87.6
25 85.4
0 82.9
25 78.9
50 73.1
65 66.0
75 59.3
85 47.9
100 34.3
125 29.3
150 27.1
175 25.0
(a) Plot the data as impact energy versus
temperature.
(b) Determine a ductile-to-brittle transition temper-
ature as the temperature corresponding to the aver-
age of the maximum and minimum impact energies.
(c) Determine a ductile-to-brittle transition tem-
perature as the temperature at which the impact
energy is 70 J.
Cyclic Stresses
The SN Curve
*M.19 A fatigue test was conducted in which the
mean stress was 50 MPa (7250 psi) and the stress
amplitude was 225 MPa (32,625 psi).
(a) Compute the maximum and minimum stress
levels.
(b) Compute the stress ratio.
(c) Compute the magnitude of the stress range.
*M.20 A cylindrical 1045 steel bar (Figure M.75) is
subjected to repeated tension-compression stress
cycling along its axis. If the load amplitude is 22,000
N (4950 lb
f
), compute the minimum allowable bar
diameter to ensure that fatigue failure will not
occur. Assume a factor of safety of 2.0.
*M.21 An 8.0-mm- (0.31-in.-) diameter cylindrical rod
fabricated from a red brass alloy (Figure M.75) is
subjected to reversed tension-compression load
cycling along its axis. If the maximum tensile and
compressive loads are 7500 N (1700 lb
f
) and
7500 N (1700 lb
f
), respectively, determine its
fatigue life. Assume that the stress plotted in
Figure M.75 is stress amplitude.
*M.22 A 12.5-mm- (0.50-in.-) diameter cylindrical rod
fabricated from a 2014-T6 alloy (Figure M.75) is
subjected to a repeated tension-compression load
cycling along its axis. Compute the maximum and
minimum loads that will be applied to yield a fatigue
life of 1.0 10
7
cycles. Assume that the stress plot-
ted on the vertical axis is stress amplitude, and data
were taken for a mean stress of 50 MPa (7250 psi).
*M.23 The fatigue data for a brass alloy are given as
follows:
Stress Amplitude
(MPa) Cycles to Failure
310 2 10
5
223 1 10
6
191 3 10
6
168 1 10
7
153 3 10
7
143 1 10
8
134 3 10
8
127 1 10
9
(a) Make an SN plot (stress amplitude versus
logarithm of cycles to failure) using these data.
(b) Determine the fatigue strength at 5 10
5
cycles.
(c) Determine the fatigue life for 200 MPa.
*M.24 Suppose that the fatigue data for the brass alloy
in Problem M.23 were taken from torsional tests,
and that a shaft of this alloy is to be used for a cou-
pling that is attached to an electric motor operating
at 1500 rpm. Give the maximum torsional stress
amplitude possible for each of the following lifetimes
of the coupling: (a) 1 year, (b) 1 month, (c) 1 day,
and (d) 2 hours.
*M.25 The fatigue data for a ductile cast iron are given
as follows:
Stress Amplitude Cycles to
[MPa (ksi )] Failure
248 (36.0) 1 10
5
236 (34.2) 3 10
5
224 (32.5) 1 10
6
213 (30.9) 3 10
6
201 (29.1) 1 10
7
193 (28.0) 3 10
7
193 (28.0) 1 10
8
193 (28.0) 3 10
8
(a) Make an SNplot (stress amplitude versus log-
arithm of cycles to failure) using these data.
(b) What is the fatigue limit for this alloy?
(c) Determine fatigue lifetimes at stress ampli-
tudes of 230 MPa (33,500 psi) and 175 MPa
(25,000 psi).
(d) Estimate fatigue strengths at 2 10
5
and 6
10
6
cycles.
70
80
60
40
50
30
20
10
0 0
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e
,

S

(
M
P
a
)
S
t
r
e
s
s

a
m
p
l
i
t
u
d
e

500
400
300
200
100
1045 steel
2014T6 aluminum alloy
Red brass

(
1
0
3

p
s
i
)
Figure M.75 Stress magnitude S versus the
logarithm of the number N of cycles to fatigue
failure for red brass, an aluminum alloy, and a
plain carbon steel.
(Adapted from H. W. Hayden, W. G. Moffatt, and J.
Wulff, The Structure and Properties of Materials, Vol.
III, Mechanical Behavior, p. 15. Copyright 1965 by
John Wiley & Sons, New York. Reprinted by permission
of John Wiley & Sons, Inc. Also adapted from ASM
Handbook, Vol. 2, Properties and Selection: Nonferrous
Alloys and Special-Purpose Materials, 1990. Reprinted
by permission of ASM International.)
*M.26 Suppose that the fatigue data for the cast iron
in Problem M.25 were taken for bending-rotating
tests and that a rod of this alloy is to be used for
an automobile axle that rotates at an average ro-
tational velocity of 750 revolutions per minute.
Give maximum lifetimes of continuous driving
that are allowable for the following stress levels:
(a) 250 MPa (36,250 psi), (b) 215 MPa (31,000
psi), (c) 200 MPa (29,000 psi), and (d) 150 MPa
(21,750 psi).
*M.27 Three identical fatigue specimens (denoted A,
B, and C) are fabricated from a nonferrous alloy.
Each is subjected to one of the maximum-minimum
stress cycles listed in the following table; the fre-
quency is the same for all three tests.
Specimen
max
(MPa)
min
(MPa)
A 450 350
B 400 300
C 340 340
(a) Rank the fatigue lifetimes of these three
specimens from the longest to the shortest.
(b) Now justify this ranking using a schematic
SN plot.
*M.28 Cite five factors that may lead to scatter in
fatigue life data.
Crack Initiation and Propagation
Factors That Affect Fatigue Life
*M.29 Briefly explain the difference between fatigue
striations and beachmarks in terms of (a) size and
(b) origin.
*M.30 List four measures that may be taken to
increase the resistance to fatigue of a metal
alloy.
DESIGN PROBLEMS
Solving some of these design problems may be
expedited by using the Engineering Materials
Properties component of VMSE found on the
books Web site [www.wiley.com/college/callister
(Student Companion Site).] We have noted these spe-
cific problems by inclusion of the following icon in
one of the margins by the problem statement:
Materials Selection Using Performance Indices
M.D1 (a) Using the procedure as outlined in Section
M.2, ascertain which of the metal alloys listed in
Appendix B, have torsional strength performance
indices greater than 10.0 (for t
f
and r in units of
MPa and g/cm
3
, respectively), and, in addition,
shear strengths greater than 350 MPa. (b) Also
using the cost database in Appendix C, conduct a
cost analysis in the same manner as in Section
M.2. For those materials that satisfy the criteria
noted in part (a), and, on the basis of this cost
analysis, which material would you select for a
solid cylindrical shaft? Why?
M.D2 In a manner similar to the treatment of
Section M.2, perform a stiffness-to-mass per-
formance analysis on a solid cylindrical shaft that
is subjected to a torsional stress. Use the same
engineering materials that are listed in Table
M.1. In addition, conduct a material cost analysis.
Rank these materials on the basis of both mass
of material required and material cost. For glass
and carbon fiber-reinforced composites, assume
that the shear moduli are 8.6 and 9.2 GPa,
respectively.
M.D3 (a) A cylindrical cantilever beam is subjected
to a force F, as indicated in the accompanying fig-
ure. Derive strength and stiffness performance
index expressions analogous to Equations M.9
and M.11 for this beam. The stress imposed on
the unfixed end s is
(M.62)
L, r, and I are, respectively, the length, radius, and
moment of inertia of the beam. Furthermore, the
beam-end deflection d is
(M.63)
where E is the modulus of elasticity of the beam.
d
FL
3
3EI
s
FLr
I
L
r
F
(b) From the properties database presented in
Appendix B, select the metal alloys with stiffness
performance indices greater than 3.0 (for E and r
in units of GPa and g/cm
3
, respectively).
(c) Also using the cost database in Appendix C,
conduct a cost analysis in the same manner as in
Section M.2. Relative to this analysis and that in
part (b), which alloy would you select on a stiff-
ness-per-mass basis?
(d) Now select the metal alloys with strength
performance indices greater than 14.0 (for s
y
and
r in units of MPa and g/cm
3
, respectively), and
rank them from highest to lowest P.
(e) Using the cost database, rank the materials in
part (d) from least to most costly. Relative
to this analysis and that in part (d), which
alloy would you select on a strength-per-mass basis?
(f) Which material would you select if both stiff-
ness and strength are to be considered relative to
this application? Justify your choice.
M.D4 (a) Using the expression developed for stiff-
ness performance index in Problem M.D3(a) and
data contained in Appendix B, determine stiff-
ness performance indices for the following poly-
meric materials: high-density polyethylene,
polypropylene, poly(vinyl chloride), polystyrene,
polycarbonate, poly(methyl methacrylate),
poly(ethylene terephthalate), polytetrafluoroeth-
ylene, and nylon 6,6. How do these values com-
pare with those of the metallic materials? (Note:
In Appendix B, where ranges of values are given,
use average values.)
(b) Now using the cost database in Appendix C,
Section M.2. Use cost data for the raw forms of
these polymers.
(c) Using the expression developed for the
strength performance index in Problem M.D3(a)
and data contained in Appendix B, determine
strength performance indices for these same poly-
meric materials.
M.D5 (a) A bar specimen having a square cross sec-
tion of edge length c is subjected to a uniaxial ten-
sile force F, as shown in the accompanying figure.
Derive strength and stiffness performance index ex-
pressions analogous to Equations M.9 and M.11 for
this bar.
(b) From the properties database presented in
Appendix B, select the metal alloys with stiffness
performance indices greater than 26.0 (for E and
r in units of GPa and g/cm
3
, respectively).
(c) Also using the cost database in Appendix
C, conduct a cost analysis in the same manner
as in Section M.2. Relative to this analysis and
that in part (b), which alloy would you select on a
stiffness-per-mass basis?
performance indices greater than 120 (for s
y
and r in units of MPa and g/cm
3
, respectively),
and rank them from highest to lowest P.
(e) Using the cost database, rank the materi-
als in part (d) from least to most costly.
Relative to this analysis and that in part (d),
which alloy would you select on a strength-per-
mass basis?
M.D6 Consider the plate shown in the accompanying
figure (on the next page) that is supported at its
ends and subjected to a force F that is uniformly
distributed over the upper face as indicated. The
deflection d at the L/2 position is given by the ex-
pression
(M.64)
Furthermore, the tensile stress at the underside and
also at the L2 location is equal to
(M.65) s
3 FL
4 wt
2
d
5 FL
3
32 Ewt
3
F
L
c
c
F
(a) Derive stiffness and strength performance in-
dex expressions analogous to Equations M.9 and
M.11 for this plate. (Hint: Solve for t in these two
equations, and then substitute the resulting ex-
pressions into the mass equation, as expressed in
terms of density and plate dimensions.)
(b) From the properties database in Appendix B,
select the metal alloys with stiffness performance
indices greater than 1.40 (for E and r in units of
GPa and g/cm
3
, respectively).
(c) Also using the cost database in Appendix C,
Section M.2. Relative to this analysis and that in
part (b), which alloy would you select on a stiffness-
per-mass basis?
performance indices greater than 5.0 (for s
y
and
r in units of MPa and g/cm
3
, respectively), and
rank them from highest to lowest P.
(e) Using the cost database, rank the materials in
part (d) from least to most costly. Relative to this
analysis and that in part (d), which alloy would
you select on a strength-per-mass basis?
M.D7 Consider a flat plate of width 100 mm (4.0
in.) that contains a centrally positioned,
through-thickness crack (Figure M.9) of length
(i.e., 2a) 25 mm (1.0 in.). Determine the mini-
mum plane strain fracture toughness necessary
to ensure that fracture will not occur for a de-
sign stress of 415 MPa (60,000 psi). The pa/W
ratio is in radians.
M.D8 A flat plate of some metal alloy contains a
centrally positioned, through-thickness crack
(Figure M.9). Determine the critical crack length
if the plane strain fracture toughness of the alloy
is 50.0 MPa , the plate width is
60 mm (2.4 in.), and the design stress is 375 MPa
(54,375 psi). The pa/Wratio is in radians.
M.D9 Consider a steel plate having a through-
thickness edge crack similar to that shown in
Figure M.10a. If it is known that the minimum
crack length subject to detection is 3 mm (0.12
in.), determine the minimum allowable plate
width assuming a plane strain fracture toughness
of 65 MPa a yield strength of
1000 MPa (145,000 psi), and that the plate is to be
loaded to one-half of its yield strength.
M.D10 Consider a steel plate having a through-
thickness edge crack similar to that shown in
Figure M.10a; the plate width (W) is 40 mm (1.6
in.) and its thickness (B) is 6.0 mm (0.25 in.).
Furthermore, plane strain fracture toughness
and yield strength values for this material are
60 MPa and 1400 MPa (200,000
psi), respectively. If the plate is to be loaded
to a stress of 200 MPa (29,000 psi), would you ex-
pect failure to occur if the crack length a is 16 mm
(0.63 in.)? Why or why not?
M.D11 A small and thin flat plate of a brittle material
having a through-thickness surface crack is to be
loaded in the manner of Figure M.10c; the K
Ic
value
for this material is .
For a crack length of 0.5 mm (0.02 in.), determine
the maximum load that may be applied without fail-
ure for B 1.5 mm (0.06 in.), S 10 mm (0.39 in.),
and W2.5 mm (0.10 in.). Assume that the crack is
located at the S/2 position.
*M.D12(a) For the thin-walled spherical tank
discussed in Design Example M.1, on the basis
of critical crack size criterion [as addressed in
part (a)], rank the following polymers from
longest to shortest critical crack length: nylon
6,6 (50% relative humidity), polycarbonate,
poly(ethylene terephthalate), and poly(methyl
methacrylate). Comment on the magnitude
range of the computed values used in the rank-
ing relative to those tabulated for metal alloys
as provided in Table M.4. For these computa-
tions, use data contained in Tables B.4 and B.5
in Appendix B.
(b) Now rank these same four polymers relative
to maximum allowable pressure according to the
leak-before-break criterion, as described in the
(b) portion of Design Example M.1. Comment on
these values in relation to those for the metal
alloys that are tabulated in Table M.5.
0.60 MPa1m 10.55 ksi 1in. 2
ksi 1in. 2 1m 154.6
159.2 ksi 1in. 2 , 1m
145.5 ksi 1in. 2 1m
L
F
w
t
Crack Propagation Rate
M.D13 Consider a flat plate of some metal alloy that is
to be exposed to repeated tensile-compressive
cycling in which the mean stress is 25 MPa. If the
initial and critical surface crack lengths are 0.25 and
5.0 mm, respectively, and the values of mand Aare
4.0 and 5 10
15
, respectively (for s in MPa and
a in m), estimate the maximum tensile stress to
yield a fatigue life of 3.2 10
5
cycles. Assume the
parameter Yhas a value of 2.0, which is independent
of crack length.
M.D14 Consider a large, flat plate of a metal alloy that
is to be exposed to reversed tensile-compressive
cycles of stress amplitude 150 MPa. If initially the
length of the largest surface crack in this speci-
men is 0.75 mm and the plane strain fracture
toughness is , whereas the values of m
and A are 2.5 and 2 10
12
, respectively (for s
in MPa and a in m), estimate the fatigue life of
this plate. Assume that the parameter Y has a
value of 1.75 that is independent of crack length.
M.D15 Consider a metal component that is exposed
to cyclic tensile-compressive stresses. If the fa-
tigue lifetime must be a minimum of 5 10
6
cycles
and it is known that the maximum initial sur-
face crack length is 0.02 in. and the maximum
tensile stress is 25,000 psi, compute the critical
surface crack length. Assume that Y is independ-
ent of crack length and has a value of 2.25, and
that m and A have values of 3.5 and 1.3 10
23
,
respectively, for s and a in units of psi and in.,
respectively.
M.D16 (a) Calculate values for the A and m param-
eters in Equation M.43 (in both SI and customary
U.S. units) for the crack propagation rate of the
Ni-Mo-V steel for which the log da/dN versus log
K plot is shown in Figure M.35.
(b) Suppose that a metal component fabricated
from this Ni-Mo-V steel alloy is exposed to cyclic
tensile-compressive stresses. If the fatigue lifetime
must be a minimum of 3 10
5
cycles and it is
known that the critical surface crack length is 1.5
mm and the maximum tensile stress is 30 MPa,
compute the maximum initial surface crack length.
Assume that Y is independent of crack length and
has a value of 1.25.
M.D17 Consider a thin metal plate 25 mm wide that
contains a centrally positioned, through-thickness
crack in the manner shown in Figure M.9. This plate
is to be exposed to reversed tensile-compressive cy-
cles of stress amplitude 120 MPa. If the initial and
critical crack lengths (i.e., 2a
0
and 2a
c
) are 0.10 and
6.0 mm, respectively, and the values of mand Aare
35 MPa1m
4 and 6 10
12
, respectively (for s in MPa and
a in m), estimate the fatigue life of this plate.
M.D18 For an edge crack in a plate of finite width
(Figure M.10a), Y is a function of the crack
length-specimen width ratio as
(M.66)
Now consider a 60-mm-wide plate that is exposed
to cyclic tensile-compressive stresses (reversed
stress cycle) for which s
min
= 135 MPa.
Estimate the fatigue life of this plate if the initial
and critical crack lengths are 5 mm and 12 mm,
respectively. Assume values of 3.5 and 1.5
10
12
for the m and A parameters, respectively,
for s in units of megapascals and a in meters.
M.D19 The spherical tank shown in Figure M.12 is
alternately pressurized and depressurized be-
tween atmospheric pressure and a positive pres-
sure p; thus, fatigue failure is a possibility.
Utilizing Equation M.48, derive an expression for
the fatigue life N
f
in terms of p, the tank radius r
and thickness t, and other parameters subject to
the following assumptions: Y is independent of
crack length, , and the original and critical
crack lengths are variable parameters.
Design and Materials Selection for Springs
M.D20 A spring having a center-to-center diameter of
20 mm (0.8 in.) is to be constructed of cold-drawn
and annealed 316 stainless steel wire that is 2.5 mm
(0.10 in.) in diameter; this spring design calls for
eight coils.
(a) What is the maximum tensile load that may
be applied such that the total spring deflection
will be no more than 6.5 mm (0.26 in.)?
(b) What is the maximum tensile load that may
be applied without any permanent deformation
of the spring wire? Assume that the shear yield
strength is 0.6s
y
, where s
y
is the yield strength in
tension.
M.D21 You have been asked to select a material for
a spring that is to be stressed in tension. It is to
consist of ten coils, and the coil-to-coil diameter
called for is 15 mm; furthermore, the diameter of
the spring wire must be 2.0 mm. Upon application
of a tensile force of 35 N, the spring is to experi-
ence a deflection of no more than 12 mm, and not
plastically deform.
m 2
Y
1.1 a 1
0.2a
W
b
a 1
a
W
b
3
2
(a) From the materials included in the database
in Appendix B, make a list of candidate materials
that meet the preceding criteria. Assume that the
shear yield strength is 0.6s
y
, where s
y
is the yield
strength in tension, and that the shear modulus is
equal to 0.4E, E being the modulus of elasticity.
(b) Now, from this list of candidate materials, se-
lect the one you would use for this spring appli-
cation. In addition to the preceding criteria, the
material must be relatively corrosion resistant,
and, of course, capable of being fabricated into
wire form. Justify your decision.
M.D22 A spring having seven coils and a coil-to-coil
diameter of 0.5 in. is to be made of cold-drawn
steel wire. When a tensile load of 15 lb
f
is applied,
the spring is to deflect no more than 0.60 in. The
cold-drawing operation will, of course, increase
the shear yield strength of the wire, and it has
been observed that t
y
(in ksi) depends on wire di-
ameter d (in in.) according to
(M.67)
If the shear modulus for this steel is 11.5 10
6
psi,
calculate the minimum wire diameter required
t
y
63
d
0.2
such that the spring will not plastically deform
when subjected to the preceding load.
M.D23 A helical spring is to be constructed from a
4340 steel. The design calls for five coils, a coil-to-
coil diameter of 12 mm, and a wire diameter of 3
mm. Furthermore, a maximum total deflection of
5.0 mm is possible without any plastic deforma-
tion. Specify a heat treatment for this 4340 steel
wire in order for the spring to meet the preced-
ing criteria. Assume a shear modulus of 80 GPa
for this steel alloy, and that t
y
0.6s
y
. (Note:
heat treatment of the 4340 steel is discussed in
Section 11.8.)
Failure Analysis
*M.D24 Each student (or group of students) is to
obtain an object/structure/component that has
failed. It may come from the home, an automo-
bile repair shop, a machine shop, and so on.
Conduct an investigation to determine the cause
and type of failure (i.e., simple fracture, fatigue,
creep). In addition, propose measures that can be
taken to prevent future incidents of this type of
failure. Finally, submit a report that addresses
these issues.
Problem Duplication Guide 95
M.1 9.1
M.2 9.2
M.3 9.3
M.4 9.4
M.5
M.6
M.7
M.8
M.9
M.10 9.5
M.11 9.6
M.12 9.7
M.13 9.8
M.14 9.9
M.15 9.10
M.16
M.17 9.15
M.18 9.16
M.19 9.17
M.20 9.18
M.21 9.19
M.22 9.20
M.23 9.21
M.24 9.22
M.25 9.23
M.26 9.24
M.27 9.25
M.28 9.27
M.29 9.28
M.30 9.29
M.D1
M.D2
M.D3
M.D4
M.D5
M.D6
M.D7
M.D8
M.D9
M.D10
M.D11
M.D12 9.D2
M.D13
M.D14
M.D15
M.D16
M.D17
M.D18
M.D19
M.D20
M.D21
M.D22
M.D23
M.D24 9.D1
Problem Number Problem Number
Module Problem Fundamentals, 4th Module Problem Fundamentals, 4th
Number Edition Number Edition
PROBLEM DUPLICATION GUIDE
Due to the interrelated nature of the content, some
sections (and homework problems) in the ME
Module are identical to sections and homework
problems found elsewhere in the text. This guide
notes which problems appear both in the ME
Module and elsewhere, and which appear only in the
ME Module. In this way, instructors and students
may easily identify which problems appearing in the
ME Module may have already been assigned dur-
ing the course, if the related sections were covered
previously.

ME Web Module

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

ME Web Module

Transféré par

Droits d'auteur :

Formats disponibles

Online Support Module

for Five Engineering

Figure M.44 Stress versus time for a reversed

. All rights reserved.

. All rights reserved. www.asminternational.org.)

Vous aimerez peut-être aussi