879 686u6i6
879 686u6i6
Fundamentals
SYMBOLS
A area, m
2
A
a
amplitude ratio,
a
/
m
a fatigue strength exponent
b width, m
b
s
slope
C constant used in Equation (7.47)
C
1
, C
2
integration constants
C intercept
c distance from neutral axis to outer ber of
solid, m
d diameter, m
E modulus of elasticity, Pa
e, f factors used in Equation (7.21)
f exponent
g gravitational acceleration, 9.807 m/s
2
(386.1 in/s
2
)
H height including two notch radii, m
h height without two notch radii, m
I area moment of inertia, m
4
I
m
impact factor
K stress intensity factor, MPa
m
K
c
stress concentration factor
K
f
fatigue stress concentration factor
k spring rate, N/m
K stress intensity range, MPa
m
k
f
surface nish factor
k
m
miscellaneous factor
k
r
reliability factor
k
s
size factor
k
t
temperature factor
k
0
stress concentration factor
l length, m
l
c
crack length, m
M bending moment, N m
m constant used in Equation (7.47)
N fatigue life in cycles
N
c
number of stress cycles
N
t
total number of cycles to failure
n
< N
n
s
safety factor
P force, N
P
max
maximum force, N
q
n
notch sensitivity factor
R
a
arithmetic average roughness, m
R
s
stress ratio,
min
/
max
r notch radius, m
S
e
modied endurance limit, Pa
S
e
endurance limit, Pa
S
f
modied fatigue strength, Pa
S
f
fatigue strength, Pa
S
i
strength at 10
3
cycles for ductile material, Pa
S
l
fatigue strength at which high-cycle fatigue
begins, Pa
S
u
ultimate strength, Pa
S
uc
ultimate strength in compression, Pa
S
ut
ultimate strength in tension, Pa
S
y
yield strength, Pa
S
yt
yield strength in tension, Pa
u sliding velocity, m/s
V shear force, N
W weight, N
Y plate size correction factor
x, y Cartesian coordinates, m
fatigue ductility exponent
i
cyclic ratio n
i
/N
i
deection, m
max
maximum impact deection, m
st
static deection, m
f
fatigue ductility coefcient
total strain
stress, Pa
s
f
stress at fracture, Pa
r
stress range, Pa
SUBSCRIPTS
a alternating
f nal
i integer
m mean
max maximum
min minimum
ref reference
7.1 INTRODUCTION
Most machine element failures involve load conditions that uctuate with time. However,
the static load conditions covered in Chapter 6 are important, because they provide the foun-
dation for understanding this chapter. This chapter focuses on understanding and predicting
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
Figure 7.1 On the Bridge, an illustration from Punch magazine in 1891 warning the
populace that death was waiting for them on the next bridge. Note the cracks in the
iron bridge. (Punch magazine, 1891)
extending these semiempirical rules to other ferrous and nonferrous metals, as well as
ceramics, polymers, and composite materials.
3. For the most part, fatigue involves the accumulation of damage within a material.
Damage usually consists of cracks that can grow by a small distance with each stress
cycle.
4. Experimenters have found that fatigue cracks generally begin at a surface and propagate
through the bulk. Therefore, much attention is paid to the quality of surfaces in fatigue-
susceptible machine elements. (However, if large subsurface aws or stress raisers exist
in the substrate, fatigue cracks can initiate below the surface.)
5. Fatigue cracks begin at several sites simultaneously and propagate when one aw
becomes dominant and grows more rapidly than the others.
6. Fatigue testing is imperative to conrm safe mechanical design.
The last of these points cannot be overemphasized, especially since this book con-
centrates on theoretical approaches to fatigue design. This is not intended to suggest that
theoretical approaches are sufcient; experimental verications are essential, but that is not
the focus of this book.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
Time
T
e
n
s
i
o
n
+
C
o
m
p
r
e
s
s
i
o
n
S
t
r
e
s
s
1 cycle
0
max
min
Figure 7.2 Variation in nonzero cyclic mean
stress.
dened as the average of the maximum and minimum stresses in the cycle, or
m
=
max
+
min
2
(7.1)
The stress range
r
is the difference between
max
and
min
, namely,
r
=
max
min
(7.2)
The stress amplitude
a
is one-half of the stress range, or
a
=
r
2
=
max
min
2
(7.3)
The stress ratio R
s
is the ratio of minimum to maximum stress amplitudes, or
R
s
=
min
max
(7.4)
Finally, the amplitude ratio A
a
is the ratio of the stress amplitude to the mean stress, or
A
a
=
m
=
max
min
max
+
min
=
1 R
s
1 + R
s
(7.5)
The four most frequently used patterns of constant-amplitude cyclic stress are
1. Completely reversed (
m
= 0, R
s
= 1, A
a
= )
2. Nonzero mean (as shown in Fig. 7.2)
3. Released tension (
min
= 0, R
s
= 0, A
a
= 1,
m
=
max
/2)
4. Released compression (
max
= 0, R
s
= , A
a
= 1,
m
=
min
/2)
EXAMPLE 7.1 Given: Before the thread-formed additions at the top third of high chimneys were invented,
chimneys swayed perpendicularly to the motion of the wind. This swaying motion appeared
when the wind speed and direction were constant for some time. In a certain chimney the bending
stress was 100 MPa.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
m
=
max
+
min
2
=
100 100
2
= 0
The range of stress is
r
=
max
min
= 100 (100) = 200 MPa
The alternating stress is
a
=
r
2
=
200
2
= 100 MPa
The stress ratio is
R
s
=
min
max
=
100
100
= 1
The amplitude ratio is
A
a
=
m
=
max
min
max
+
min
=
1 R
s
1 + R
s
=
1 (1)
1 +(1)
=
2
0
=
7.4 STRAIN LIFE THEORY OF FATIGUE
Fatigue is a damage accumulation process that manifests itself through crack propagation,
but no crack propagation is possible without plastic deformation at the crack tip. Although
the volume stressed highly enough for plastic deformation can be extremely small, if the
stress elds remain elastic, no crack propagation is possible. Using material properties,
such as yield strength or ultimate strength, presents difculties because cyclic loadings
can change these values near a crack tip. They may increase or decrease depending on the
material and its manufacturing history. Thus, the material strength at the location where
cracks are propagating can differ from the bulk material strength listed in handbooks or
obtained from tension tests.
Given the difculties in expressing material strengths near crack tips, several ap-
proaches have been suggested for dealing with the strain encountered at a crack tip. One of
the better known is the MansonCofn relationship, which gives the total strain amplitude
as the sum of the elastic and plastic strain amplitudes
2
=
f
E
_
2N
_
a
+
f
_
2N
(7.6)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
272 PART 1
Fundamentals
Table 7.1 Cyclic properties of some metals
Yield Fracture Fatigue Fatigue Fatigue
strength strength ductility strength ductility
Material Condition Sy, MPa
f
, MPa coefcient
f
exponent a exponent
Steel
1015 Normalized 228 827 0.95 0.110 0.64
4340 Tempered 1172 1655 0.73 0.076 0.62
1045 Q&T
80
f
= stress at fracture in one stress cycle, Pa
E = elastic modulus of material, Pa
N
f
= fatigue ductility coefcient (true strain corresponding to fracture in one
stress cycle)
a = fatigue strength exponent
= fatigue ductility exponent
Table 7.1 gives some typical values of the material properties a and .
Landgraf (1968) and Mitchell (1978) have surveyed the industrial applications of the
strain life approach to fatigue, but as pointed out by Shigley and Mitchell (1983), the
MansonCofn relationship is difcult to use because the total strain is difcult to
determine. Stress concentration factors, such as those presented in Chapter 6, are readily
available in the technical literature, but strain concentration factors, especially in the plastic
range, are nowhere to be found. The advantage of the MansonCofn equation is that it gives
insight into important properties in fatigue strength determination. It shows the importance
of strength as well as ductility, and it leads to the conclusion that as long as there is a cyclic
plastic strain, no matter how small, eventually there will be failure.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
8
Figure 7.3 R. R. Moore machine
fatigue test specimen. Dimensions in
inches.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
274 PART 1
Fundamentals
at a specic stress level, the test is conducted until failure occurs. The procedure is repeated
on other identical specimens, progressively decreasing the maximum stress amplitude.
7.5.2 REGIMES OF FATIGUE CRACK GROWTH
Figure 7.4(a) shows the size of a fatigue crack as a function of number of cycles for two
stress ratios. Figure 7.4(b) illustrates the rate of crack growth and more clearly shows three
different regimes of crack growth:
1. Regime A is a period of very slow crack growth. Note that the crack growth rate can
be smaller than an atomic spacing of the material per cycle, a situation that would be
corrected by the preferred lattice structure of the metal. Regime Ashould be recognized
as a period of noncontinuum failure processes. The fracture surfaces are faceted or
serrated in this regime, indicating crack growth is primarily due to shear deformations
within a grain. The crack growth rate is so small that cracks may be negligible over
the life of the component if this regime is dominant. Regime A is strongly affected by
material microstructure, environmental effects, and stress ratio R
s
.
2. Regime B is a period of moderate crack growth rate, often referred to as the Paris
regime or Paris power law. In this regime, the rate of crack growth is inuenced by
several processes, involving material microstructure, mechanical load variables, and the
environment. Thus it is not surprising that crack propagation rates cannot be determined
for a given material or alloy from rst principles, and testing is required to quantify the
growth rate.
3. Regime C is a period of high growth rate, where the maximum stress intenstity fac-
tor for the fatigue cycle approaches the fracture toughness of the material. Material
microstructural effects and loadings have a large inuence on crack growth, and addi-
tional static modes such as cleavage and intergranular separation can occur.
10
-2
10
-4
10
-6
10
-8
C
r
a
c
k
g
r
o
w
t
h
r
a
t
e
d
a
/
d
N
,
m
m
/
c
y
c
l
e
log K
1 mm/min
1 mm/h
1 mm/day
1 mm/week
C
r
a
c
k
g
r
o
w
t
h
r
a
t
e
a
t
5
0
H
z
= C(K)
m da
dN
Regime A Regime B
Regime C
one lattice
spacing
per cycle
m
1
K
c
C
r
a
c
k
l
e
n
g
t
h
a
Number of cycles N
2
>
1
1
da
dN
(a) (b)
Figure 7.4 Illustration of fatigue crack growth. (a) Size of a fatigue crack for two different stress ratios as a function of the
number of cycles; (b) rate of crack growth, illustrating three regimes. [Source: Suresh (1998).]
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
High Nominal Stress Low Nominal Stress
No stress
concentration
Mild stress
concentration
Severe stress
concentration
No stress
concentration
Mild stress
concentration
Severe stress
concentration
T
e
n
s
i
o
n
-
t
e
n
s
i
o
n
o
r
t
e
n
s
i
o
n
-
c
o
m
p
r
e
s
s
i
o
n
U
n
i
d
i
r
e
c
t
i
o
n
a
l
b
e
n
d
i
n
g
R
e
v
e
r
s
e
d
b
e
n
d
i
n
g
R
o
t
a
t
i
o
n
a
l
b
e
n
d
i
n
g
Beachmarks Fracture surface
Figure 7.6 Typical fatigue fracture surfaces of smooth and notched
cross-sections under different loading conditions and stress levels.
[From Metals Handbook, American Society for Metals (1975).]
7.5.4 S N DIAGRAMS
Data from reversed-bending experiments are plotted as the fatigue strength versus the
logarithm of the total number of cycles to failure N
t
for each specimen. These plots are
called SN diagrams or W ohler diagrams, after August W ohler, a German engineer who
published his fatigue research in 1870. They are a standard method of presenting fatigue data
and are common and informative. Two general patterns for two classes of material, those
with and those without endurance limits, emerge when the fatigue strength is plotted versus
the logarithm of the number of cycles to failure. Figure 7.7 shows typical results for several
materials. Figure 7.7(a) presents test data for wrought steel. Note the large amount of scatter
in the data, even with the great care in preparing test specimens. Thus, life predictions based
on curves such as those in Figure 7.7 are all subject to a great deal of error. Figure 7.7(a)
also shows a common result. For some materials with endurance limits, such as ferrous
and titanium alloys, a horizontal straight line occurs at low stress levels, implying that an
endurance limit S
e
is reached below which failure will not occur. This endurance limit S
e
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
Not broken
1.0
0.9
0.8
0.7
0.6
0.5
0.4
10
3
10
3
10
4
10
3
10
5
10
6
10
7
10
4
10
5
10
6
10
7
10
8
10
9
10
4
10
5
10
6
10
7
Number of cycles to failure N
Number of cycles to failure N
Number of cycles to failure N
F
a
t
i
g
u
e
s
t
r
e
s
s
r
a
t
i
o
S
f
/
S
u
t
80
70
60
50
40
35
30
25
20
18
16
14
12
10
8
7
6
5
A
l
t
e
r
n
a
t
i
n
g
s
t
r
e
s
s
s
a
,
k
s
i
A
l
t
e
r
n
a
t
i
n
g
s
t
r
e
s
s
s
a
,
M
P
a
A
l
t
e
r
n
a
t
i
n
g
s
t
r
e
s
s
s
a
,
p
s
i
0 0
2
4
6
8
10
20
30
40
50
60
PTFE
Phenolic
Epoxy
Diallyl-phthalate
Alkyd
Nylon (dry)
Polycarbonate
Polysulfone
(a)
(b)
(c)
W
ro
u
g
h
t
S
a
n
d
c
a
st
P
e
rm
a
n
e
n
t m
o
ld
c
a
st
Figure 7.7 Fatigue strengths as a function of number of loading cycles.
(a) Ferrous alloys, showing clear endurance limit [adapted from Lipson and
Juvinall (1963)]; (b) aluminum alloys, with less pronounced knee and no
endurance limit [adapted from Juvinall and Marshek (1991); (c) selected
properties of assorted polymer classes [adapted from Norton (1996).].
277
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
278 PART 1
Fundamentals
represents the largest uctuating stress that will not cause failure for an innite number of
cycles. For many steels the endurance limit ranges between 35% and 60% of the materials
ultimate strength.
Most nonferrous alloys (e.g., aluminum, copper, and magnesium) do not have a signif-
icant endurance limit. Their fatigue strength continues to decrease with increasing cycles.
Thus, fatigue will occur regardless of the stress amplitude. The fatigue strength for these
materials is taken as the stress level at which failure will occur for some specied number
of cycles (e.g., 10
7
cycles).
Determining the endurance limit experimentally is lengthy and expensive. The
MansonCofn relationship [Eq. (7.6)] demonstrates that the fatigue life will depend on the
materials fracture strength during a single load cycle, suggesting a possible relationship
between static material strength and strength in fatigue. Such a relationship has been noted
by several researchers (e.g., Fig. 7.8). The stress endurance limits of steel for three types of
loading can be approximated as follows:
Bending S
e
= 0.5S
u
Axial S
e
= 0.45S
u
Torsion S
e
= 0.29S
u
(7.7)
These equations can be used to approximate the endurance limits for other ferrous
alloys, but it must be recognized that the limits can differ signicantly from experimentally
determined endurance limits. Since the ultimate stress and the type of loading are known
for various materials, their endurance limits can be approximated.
160 10
3
140
120
100
80
60
40
20
0
0 20 40 60 80 100 140 180 220 260 300 10
3
120 160 200 240 280
S
e
_
_
_
S u
=
0
.6 0
.5
0
.4
Carbon steels
Alloy steels
Wrought irons
Tensile strength S
ut
, psi
E
n
d
u
r
a
n
c
e
l
i
m
i
t
S
e
,
p
s
i
100 10
3
psi
Figure 7.8 Endurance limit as function of ultimate strength for wrought steels. [Adapted from
Shigley and Mitchell (1983).]
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
e
= 0.35Su
Copper alloys 10
8
0.25Su < S
e
< 0.5Su
Nickel alloys 10
8
0.35Su < S
e
< 0.5Su
Titanium 10
7
0.45Su < S
e
< 0.65Su
Aluminum alloys 5 10
8
S
e
= 0.40Su (Su < 48 ksi)
S
e
= 19 ksi (Su 48 ksi)
SOURCE: Adapted from Juvinall and Marshek (1991).
Other materials, for which there is much less experience, are nding increasing uses
in fatigue applications. Table 7.2 gives the approximate strengths in fatigue for various
material classes. Figure 7.7(c) gives some stresslife curves for common polymers. Because
polymers have a much greater variation in properties than metals do, Figure 7.7(c) should
be viewed as illustrative of fatigue properties and not used for quantitative data.
7.6 FATIGUE REGIMES
The SN diagram [Fig. 7.7(a)] shows different types of behavior as the number of cycles
to failure increases. Two basic regimes are low-cycle fatigue (generally below 10
3
stress
cycles) and high-cycle fatigue (more than 10
3
but less than 10
6
stress cycles). The slope of
the line is much lower in low-cycle fatigue than in high-cycle fatigue.
Another differentiation can be made between nite life and innite life. Figure 7.7(a)
shows a clear endurance limit for ferrous alloys, below which any repeating stress will
lead to innite life in the component. Although a distinction between the nite-life and
innite-life portions of the curve is not always clear, for steels it occurs between 10
6
and
10
7
cycles. Thus, nite-life classication is considered for any loading below 10
7
cycles.
7.6.1 LOW-CYCLE FATIGUE
Low-cycle fatigue is any loading that causes failure below approximately 1000 cycles. This
type of loading is common. A number of devices, such as latches on automotive glove
compartments, studs on truck wheels, and setscrews xing gear locations on shafts, cycle
fewer than 1000 times during their service lives. Surviving 1000 cycles means that these
devices will last as long as intended.
For components in the low-cycle range, either designers ignore fatigue effects entirely,
or they reduce the allowable stress level. Ignoring fatigue seems to be a poor approach.
However, the low-cycle portion of the curve in Figure 7.7(a) has a small slope (i.e., the
strength at 1000 cycles has not been reduced a great deal). Further, the y intercept for the
curve is the ultimate strength, not the yield strength. Since static design often uses the yield
strength and not the ultimate strength in dening allowable stresses, static approaches are
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
280 PART 1
Fundamentals
acceptable for designing low-cycle components. In fact, the safety factor compensates for
the uncertainty in material strength due to cyclic loading.
Taking low-cycle effects into account allows modifying the material strength based on
experimental data. The fatigue strength for steel at which high-cycle fatigue begins can be
approximated as follows:
Bending S
l
= 0.9S
u
Axial S
l
= 0.75S
u
Torsion S
l
= 0.72S
u
(7.8)
Again, the fatigue strengths of other ferrous alloys can be approximated by these equations,
but they may differ from what is experienced in practice.
7.6.2 HIGH-CYCLE, FINITE-LIFE FATIGUE
In many applications the number of stress cycles placed on a component during its service
life is between 10
3
and 10
7
. Examples include car door hinges, aircraft body panels, and
aluminum softball bats. Because the strength drops rapidly in this range (Fig. 7.7), an
approach not accounting for this drop is inherently awed.
The fatigue strength at any location between S
l
and S
e
can generally be expressed as
log S
f
= b
s
log N
t
+
C (7.9)
where b
s
= slope
C = intercept
At the endpoints, Equation (7.9) becomes
log S
l
= b
s
log 10
3
+
C = 3b
s
+
C (7.10)
log S
e
= b
s
log 10
6
+
C = 6b
s
+
C (7.11)
Subtracting Equation (7.11) from Equation (7.10) gives
b
s
=
1
3
log
S
l
S
e
(7.12)
Substituting Equation (7.12) into Equation (7.11) gives
C = 2 log
S
l
S
e
+log S
e
= log
(S
l
)
2
S
e
(7.13)
Thus, by using Equations (7.7) and (7.8) the slope b
s
and the intercept
C can be determined
for a specic type of loading. Knowing the slope and the intercept from Equation (7.9)
yields the fatigue strength as
S
f
= 10
C
(N
t
)
bs
for 10
3
N
t
10
6
(7.14)
If the fatigue strength is given and the number of cycles until failure is desired, then
N
t
=
_
S
f
10
C
_
1/bs
for 10
3
N
t
10
6
(7.15)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
e
= 0.45S
u
and S
l
= 0.75S
u
. Note
that S
l
is for 1000 cycles and S
e
is for a life of 10
6
cycles. Equation (7.12) gives the slope as
b
s
=
1
3
log
S
l
S
e
=
1
3
log
0.75S
u
0.45S
u
= 0.07395
From Equation (7.13) the intercept is
C = log
(S
l
)
2
S
e
= log
(0.75)
2
(1080)
2
(0.45)(1080)
= 3.130
Given the slope and intercept, Equation (7.14) gives the fatigue strength as
S
f
= 10
C
(N
t
)
bs
= 10
3.13
(10,000)
0.07395
= 682.7 MPa
Thus, the stress has to be decreased to 682.7 MPa for 10,000 cycles.
7.6.3 HIGH-CYCLE, INFINITE-LIFE FATIGUE
A number of applications call for innite life, dened for steels as the number of cycles
above which an endurance limit can be dened, usually taken as 10
6
cycles. If a material
does not have an endurance limit, it cannot be designed for innite life. Thus, aluminum
alloys, for example, will always be designed for nite life (using the approach given in
Sec. 7.6.2).
For ferrous and titanium alloys, however, an innite-life design approach can be fol-
lowed. Basically, the designer determines an endurance limit and uses this strength as the
allowable stress. Then sizing and selection of components can proceed just as in static
design. This approach, which is fairly complex, is described next.
7.7 ENDURANCE LIMIT MODIFICATION FACTORS
Fatigue experiments assume that the best circumstances exist for promoting long fatigue
lives. However, this situation cannot be guaranteed for design applications, so the compo-
nents endurance limit must be modied or reduced fromthe materials best-case endurance
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
282 PART 1
Fundamentals
limit. The endurance limit modication factors covered in this text are for completely re-
versed loading (
m
= 0). The modied endurance limit can be expressed as
S
e
= k
f
k
s
k
r
k
t
k
m
S
e
(7.16)
where S
e
= endurance limit from experimental apparatus under idealized conditions, Pa
k
f
= surface nish factor
k
s
= size factor
k
r
= reliability factor
k
t
= temperature factor
k
m
= miscellaneous factor
The type of loading has been incorporated into S
e
, as presented in Equation (7.7).
Many effects can inuence the fatigue strength of a part. Some of those most frequently
encountered are described in this section.
7.7.1 STRESS CONCENTRATION EFFECTS
The MansonCofn equation [Eq. (7.6)] showed that the life of a component has a direct
correlation with the strain to which it is subjected. Because locations of stress concentration
are also locations of strain concentration, these locations can be seen as prime candidates
for the promotion of fatigue crack initiation and growth. However, the stress concentration
factor developed in Chapter 6 cannot be directly applied to fatigue applications since many
materials will relieve stresses near a crack tip through plastic ow. That is, because some
materials ow plastically near crack tips, fracture is avoided and the cracks growth is
retarded.
For static loading the stress concentration factor K
c
is used, and for fatigue loading the
fatigue stress concentration factor K
f
is used, where
K
f
=
endurance limit for notch-free specimen
endurance limit for notched specimen
(7.17)
A notch or stress concentration may be a hole, llet, or groove. Recall from Section 6.3 that
the stress concentration factor is simply a function of geometry. In this section the fatigue
stress concentration factor is not only a function of geometry but also a function of the
material and type of loading. The consideration of the material is often dealt with by using
a notch sensitivity factor q
n
, dened as
q
n
=
K
f
1
K
c
1
(7.18)
or K
f
= 1 +(K
c
1)q
n
(7.19)
From Equation (7.18) note that the range of q
n
is between 0 (when K
f
= 1) and 1 (when
K
f
= K
c
). From Equation (7.19) observe that obtaining the fatigue stress concentration
factor requires knowing the materials notch sensitivity and the type of loading.
Figure 7.9 is a plot of notch sensitivity versus notch radius for some commonly used
materials and for various types of loading. For all the materials considered, the notch
sensitivityapproaches zeroas the notchradius approaches zero. Also, the harder andstronger
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
60
(414)
8
0
(5
5
2
)
6
0
(4
1
4
)
5
0
(3
4
5
)
Figure 7.9 Notch sensitivity as function of notch radius for several materials and types
of loading. [Adapted from Sines and Waisman (1959).]
steels tend to be more notch-sensitive (have a large value of q
n
). This is not too surprising,
since notch sensitivity is a measure of material ductility and the hardest steels have limited
ductility. Figure 7.9 also shows that a given steel is slightly more notch-sensitive for torsional
loading than for bending and axial loading.
The endurance limit modication factor taking stress concentrations into account is
k
0
=
1
K
f
(7.20)
where K
f
is obtained from Equation (7.19).
To apply the effects of stress concentrations, designers can either reduce the endurance
limit by k
0
or increase the stress by K
f
; either approach is equally valid. In this text neither
factor is included in the modied endurance limit equation [Eq. (7.16)]. It is more conve-
nient to deal with these factors separately, since in later chapters, such as that on shafting
(Chap. 11), these effects are treated differently for ductile and brittle materials.
EXAMPLE 7.3 Given: The driveshaft for a Formula One racing car has a diameter of 30 mmand a half-circular
notch with a 1-mm radius. The shaft was dimensioned for a coefcient of friction between the
tires and the ground of 1.5 for equal shear and bending stresses. By mounting spoilers and a wing
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
284 PART 1
Fundamentals
on the car, the load on the tires can be doubled at high speed without increasing the cars mass.
The shaft material has an ultimate tensile strength of 965 MPa. Assume fromthe distortion energy
theory that the equivalent stress is proportional to the square root of the bending stress squared
plus 3 times the shear stress squared (
e
=
2
+3
2
).
Find: Determine the fatigue stress concentrationfactors for bendingandtorsionof the driveshaft.
Also, determine if increased acceleration or increased curve handling will give the higher risk of
driveshaft failure.
Solution: From Figure 7.9 for a notch radius of 1 mm and ultimate strength of 965 MPa, the
notch sensitivity is 0.825 for bending and 0.853 for torsion. From Figure 6.9(b) when r/d =
1/28 = 0.0357 and D/d = 30/28 = 1.0714, the stress concentration factor is 2.2 for bending,
and from Figure 6.9(c) the stress concentration factor is 1.8 for torsion.
From Equation (7.19) the fatigue stress concentration factor due to bending is
K
f
= 1 +(K
c
1)q
n
= 1 +(2.2 1)(0.825) = 1.99
The fatigue stress concentration factor due to torsion is
K
f
= 1 +(1.8 1)(0.853) = 1.68
Let
e1
be the equivalent stress for increased curve handling and
e2
be the equivalent stress
for increased acceleration. Doubling the load and using the distortion energy theory result in
e1
e2
=
_
(2)
2
+3
2
_
2
+3(2)
2
=
2
_
1 +0.75(/)
2
_
1 +12(/)
2
Recall that the shaft was dimensioned such that the shear and bending stresses are equal
( = ).
\
e1
e2
=
2
1 +0.75
1 +12
= 0.7338
e2
=
e1
0.7338
= 1.363
e1
Therefore, increased acceleration gives a higher risk of driveshaft failure than increased curve
handling.
7.7.2 SURFACE FINISH FACTOR
The specimen shown in Figure 7.3 has a highly polished surface nish with nal polishing
in the axial direction to smooth any circumferential scratches. Most machine elements do
not usually have such a high-quality nish. The modication factor to incorporate the nish
effect depends on the process used to generate the surface and on the ultimate strength.
Once that process is known, Figure 7.10(a) enables the surface nish factor to be obtained
when the ultimate strength in tension is known, or else the coefcients from Table 7.3 can
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
Table 7.3 Surface nish factor
Manufacturing
process
Factor e
MPa ksi Exponent f
Grinding 1.58 1.34 0.085
Machining or cold 4.51 2.70 0.265
drawing
Hot rolling 57.7 14.4 0.718
As forged 272.0 39.9 0.995
SOURCE: Shigley and Mitchell (1983).
If the process used to obtain the surface nish is not known but the quality of the surface is
known from the arithmetic average surface roughness R
a
, the surface nish factor can be
obtained from Figure 7.10(b).
These approaches are all approximate and are based on the processes given in the
charts and used only for well-controlled processes. It is misleading to apply Table 7.3 for
other circumstances or operations. For example, plasma spray operations tend to provide
an extremely rough surface, but the fatigue properties are determined by the surface layer
beneath the plasma-sprayed coating. Further, the data in Table 7.3 are dated. With modern
numerically controlled machine tools and improvements in tooling materials, superior n-
ishes are routinely produced today that will give slightly better performance from a fatigue
standpoint.
7.7.3 SIZE FACTOR
The high-cycle fatigue apparatus used to obtain the endurance limit S
e
was for a specic
diameter, namely, 0.30 in, and uses extruded or drawn bar stock for metals. For metals
such extrusions have pronounced grain elongation in the direction opposite to the fatigue
crack growth. Also, the degree of cold work is high, and the likelihood of large aws is
low. Similar effects are seen for ceramics and castings but for different reasons (smaller
shrinkage pores, etc.). Regardless of the manufacturing process, larger parts are more likely
to contain aws and will not demonstrate the strength of the 0.3-in-diameter shaft. The size
factor for a round bar is affected by the method of loading. For bending or torsion the size
factor is
k
s
=
_
_
_
0.869d
0.112
0.3 in < d < 10 in
1 d < 0.3 in or d 8 mm
1.189d
0.112
8 mm < d 250 mm
(7.22)
For axial loading k
s
= 1.
For components that are not circular in cross-section, the size factor is difcult to
determine. However, because the reasons for strength reductions hold for such components,
it is reasonable to dene an effective diameter based on an equivalent circular cross-section.
The effective diameter can be approximated by using equivalent cross-sections dened
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
f
in one loading cycle. It is therefore reasonable
to follow one of two approaches. The designer can either (1) modify the ultimate strength
of the material based on its properties at the temperature of interest before determining a
material endurance limit in Equation (7.16) or (2) use a temperature factor:
k
t
=
S
ut
S
ut,ref
(7.23)
where S
ut
= ultimate tensile strength of material at desired temperature, Pa
S
ut,ref
= ultimate tensile strength at reference temperature, usually room
temperature, Pa
Table 7.4 Reliability factors for six
probabilities of survival
Probability of Reliability factor
survival, % kr
50 1.00
90 0.90
95 0.87
99 0.82
99.9 0.75
99.99 0.70
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
288 PART 1
Fundamentals
7.7.6 MISCELLANEOUS EFFECTS
Several other phenomena can affect a components strength. Whereas the preceding sections
have outlined methods for numerically approximating some effects, other considerations
defy quantication. Among these are the following:
1. Manufacturing history. Manufacturing processes play a major role in determining
the fatigue life characteristics of engineering materials. This role is manifested in the
size factor discussed in Section 7.7.3, but there are other effects as well. Because fatigue
crack growth is more rapid along grain boundaries than through grains, any manufac-
turing process affecting grain size and orientation can affect fatigue. Because some
forming operations, such as rolling, extrusion, and drawing, lead to elongated grains,
the materials fatigue strength will vary in different directions (anisotropy). With extru-
sion and drawing this effect is usually benecial, since the preferred direction of crack
propagation becomes the axial direction and crack propagation through the thickness
is made more difcult by grain orientation and elongation in metals. Annealing a metal
component relieves residual stresses, causes grains to become equiaxed, and may cause
grain growth. Relieving tensile residual stresses at a surface is generally benecial, but
equiaxed or larger grains are detrimental from a fatigue standpoint.
2. Residual stresses. These can result frommanufacturing processes. Aresidual stress
is caused by elastic recovery after nonuniformplastic deformation across a components
thickness. Compressive residual stress in a surface retards crack growth; tensile residual
stress can encourage crack growth.
Compressive residual stresses can be obtained through shot peening and roller
burnishing and may be obtained in forging, extrusion, or rolling. Shot peening is a cold-
working process in which the surface of a part is bombarded with small spherical media
calledshot. Eachimpact leads toplastic deformationat the workpiece surface, leadingto
compressive residual stress after elastic recovery. The layer under compressive residual
stress is usually less than 1 mm (0.04 in) thick, and the material bulk properties are
unaffected. Crack development and propagation is severely retarded by compressive
residual stresses; for this reason shot peening is a common surface treatment for fatigue-
susceptible parts such as gears, springs, shafts (especially at stress concentrations), and
connecting rods.
The benecial effect of shot peeningonfatigue life canbe seeninFigure 7.11. Similar
behavior can be found for other materials. This is an important tool for fatigue design,
because it represents one of the only strategies that increases the fatigue strength of
materials, and this increase can be very large. For example, consider an aircraft landing
gear, produced from steel with a 2068 MPa (300 ksi) strength. Figure 7.11(a) shows
that shot peening can increase the fatigue strength by a factor of 3 over a polished
surface. Similar benets are possible with other materials, but as seen in the gure, the
more typical fatigue strength improvement is 15% to 30%.
3. Coatings. These can greatly affect fatigue. Some operations, such as carburizing,
lead to a high carbon content in steel surface layers (and thus a high fracture strength)
and impart a compressive residual stress on the surface. Electroplated surfaces can be
extremely porous and promote crack growth, reducing fatigue strengths by as much
as 50%. Zinc plating is the main exception where the fatigue strength is not seriously
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
a
,
M
P
a
k
s
i
10
4
10
5
10
6
10
7
10
8
Number of cycles to failure, N'
Al 7050-T7651
Ti-6Al-4V
Shot peened
Machined
Polished
(a) (b)
Figure 7.11 The use of shot peening to improve fatigue properties. (a) Fatigue strength at 2 10
6
cycles for high-strength
steel as a function of ultimate strength; (b) typical SN curves for nonferrous metals. [Source: From Electronics, Inc.]
affected. Anodized oxide coatings are also usually porous, reducing fatigue strength.
Coatings applied at high temperatures, such as in chemical vapor deposition processes
or hot dipping, may thermally induce tensile residual stresses at the surface.
4. Corrosion. It is not surprising that materials operating in corrosive environments
have lowered fatigue strengths. The main culprits in corroding metals are hydrogen
and oxygen. Hydrogen diffuses into a material near a crack tip, aided by large tensile
stresses at the tip, embrittling the material and aiding crack propagation. Oxygen causes
coatings to form that are brittle or porous, aiding crack initiation and growth. High
temperatures in corrosive environments speed diffusion-based processes.
EXAMPLE 7.4 Given: Figure 7.12 shows a tensile-loaded bar. In Figure 7.12(a) the bar is not notched, and in
Figure 7.12(b) the bar is notched; but the smallest height is the same in both gures. The bar is
machine-made of low-carbon steel (AISI 1020).
1 in
1 in
(a) (b)
P
P
P
P
1 in
1.2 in
1 in
r = 0.1 in
r = 0.1 in
Figure 7.12 Tensile loaded bar. (a) Unnotched; (b) notched.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
290 PART 1
Fundamentals
Find: The modied endurance limits for the notched and unnotched bars as well as the fatigue
strengths at 10
4
cycles.
Solution: From Table A.1 for low-carbon steel (AISI 1020) the ultimate strength is
S
u
= 57 ksi
The loading is axial, so from Equations (7.7) and (7.8)
S
e
= 0.45S
u
= (0.45)(57)(10
3
) psi = 25.65 ksi
S
l
= 075S
u
= (0.75)(57)(10
3
) psi = 42.75 ksi
From Equations (7.12) and (7.13) the slope and intercept are, respectively,
b
s
=
1
3
log
S
l
S
e
=
1
3
log
42.75
25.65
= 0.07400
C = log
(S
l
)
2
S
e
= log[(1.667)(42.75)(10
3
)] = 4.853
From Figure 7.10(a) for machined surfaces the surface nish factor k
f
= 0.84. For axial loading
the size factor k
s
= 1. For the unnotched bar the fatigue stress concentration factor K
f
= 1. For
the notched bar with a groove Figure 6.4 is to be used. From Figure 7.12
H = 1.2 in r = 0.1 in b = 1 in h = 1 in
\
H
h
=
1.2
1
= 1.2 and
r
h
=
0.1
1
= 0.10
From Figure 6.4(a) for H/h = 1.2 and r/h = 0.1, the stress concentration factor K
c
is 2.35.
From Figure 7.9 for S
u
= 57 ksi and r = 2.5 mm, the notch sensitivity is 0.73. From Equation
(7.19) the fatigue stress concentration can be expressed as
K
f
= 1 +(K
c
1)q
n
= 1 +(2.35 1)(0.73) = 1.986
From Equation (7.16) the modied endurance limit for the notched bar is
S
e
K
f
=
k
f
k
s
S
e
K
f
=
(0.84)(1)(25.65)(10
3
)
1.986
psi = 10.85 ksi
For the unnotched bar, the modied endurance limit is
S
e
K
f
=
(0.84)(1)(25.65)(10
3
)
1
psi = 21.55 ksi
Thus, even though the notched and unnotched bars have the same height, width, and length, the
presence of the notch decreases the modied endurance limit by one-half.
The fatigue strength at 10
4
cycles from Equation (7.14) is
S
f
= 10
C
(N
t
)
bs
= 10
4.853
(10
4
)
0.074
= 10
4.557
psi = 36.06 ksi
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
f
K
f
=
(0.84)(1)(36.06)(10
3
)
1.986
psi = 15.25 ksi
and for the unnotched bar
S
f
K
f
=
(0.84)(1)(36.06)(10
3
)
1
psi = 30.29 ksi
7.8 CUMULATIVE DAMAGE
In constructing the solid curve in Figure 7.7(c), it was assumed that the cyclic variation
was completely reversed (
m
= 0). Furthermore, for any stress level between the strengths
S
l
and S
e
, say S
1
, the maximum stress level in the completely reversed variation was kept
constant until failure occurred at N
1
cycles. Operating at stress amplitude S
1
for a number
of cycles n
1
< N
1
produced a smaller damage fraction. Because cyclic variations are not
constant in practice, engineers must deal with several different levels of completely reversed
stress cycles. Operating over stress levels between S
l
and S
e
, say S
i
, at a number of cycles
n
i
< N
i
resulted in the damage fraction n
i
/N
i
. When the damage fraction due to different
levels of stress exceeds unity, failure is predicted. Thus, failure is predicted if
n
1
N
1
+
n
2
N
2
+ 1 (7.24)
This formulation is frequently called the linear damage rule (sometimes called Miners
rule), since it states that the damage at any stress level is directly proportional to the number
of cycles (assuming that each cycle does the same amount of damage). The rule also assumes
that the stress sequence does not matter and that the rate of damage accumulation at a
particular stress level is independent of the stress history. Despite these shortcomings, the
linear damage rule remains popular, largely because it is so simple.
If N
t
is the total number of cycles to failure when there are different cyclic patterns
(all of which are completely reversed), the ratio of the number of cycles at a specic stress
level to the total number of cycles to failure is
i
=
n
i
N
t
or n
i
=
i
N
t
(7.25)
Substituting Equation (7.25) into Equation (7.24) gives that failure will occur if
i
N
1
N
t
(7.26)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
292 PART 1
Fundamentals
EXAMPLE 7.5 Given: For the unnotched bar in Example 7.4, the fatigue stress is 25 ksi for 20% of the time,
30 ksi for 30%, 35 ksi for 40%, and 40 ksi for 10%.
Find: The number of cycles until cumulative failure.
Solution: Note that S
1
= 25 ksi is less than S
e
, which was found to be 25.65 ksi. Therefore,
N
1
= , implying that at this stress level, failure will not occur. From Equation (7.14) for the
other three fatigue stress levels
N
2
=
_
S
f
10
C
_
1/bs
= [(30,000)(10)
4.853
]
1/0.074
= 1.201 10
5
cycles
N
3
= [(35,000)(10)
4.853
]
1/0.074
= 1.495 10
4
cycles
N
4
= [(40,000)(10)
4.853
]
1/0.074
= 2.461 10
3
cycles
Making use of Equation (7.26) gives
1
N
1
+
2
N
2
+
3
N
3
+
4
N
4
=
1
N
t
0.2
+
0.3
1.201 10
5
+
0.4
1.495 10
4
+
0.1
2.461 10
3
=
1
N
t
1
N
t
= 0 +2.498 10
6
+2.676 10
5
+4.063 10
5
N
t
= 14,310 cycles
The cumulative number of cycles to failure is 14,310.
7.9 INFLUENCE OF NONZERO MEAN STRESS
Other than in classifying cyclic behavior, completely reversed (
m
= 0) stress cycles have
been assumed. Many machine elements involve uctuating stresses about a nonzero mean.
The experimental apparatus used to generate the results shown in Figure 7.7 could not apply
mean and alternating stresses. Because data are not available, the inuence of nonzero mean
stress must be estimatedbyusingone of several empirical relationships that determine failure
at a given life when alternating and mean stresses are both nonzero.
7.9.1 DUCTILE MATERIALS
Figure 7.13 shows how four empirical relationships estimate the inuence of nonzero mean
stress on fatigue life for ductile materials loaded in tension. On the ordinate, the yield
strength S
yt
is plotted, reminding us that yielding rather than fatigue might be the failure
criterion. On the abscissa, the yield strength in tension S
yt
and the ultimate strength in
tension S
ut
are plotted. Failure is predicted for values of
a
and
m
above the curves.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
a
Mean stress
m
Soderberg line
0
S
yt
S
yt
S
ut
S
e
Figure 7.13 Inuence of nonzero mean stress on
fatigue life for tensile loading as estimated by four
empirical relationships.
Gerber Line
The Gerber line is sometimes called the Gerber parabolic relationship because the equation
is
K
f
n
s
a
S
e
+
_
n
s
m
S
ut
_
2
= 1 (7.27)
where S
e
= modied endurance limit, Pa
S
ut
= ultimate strength in tension, Pa
n
s
= safety factor
a
= alternating stress, Pa
m
= mean stress, Pa
K
f
= fatigue stress concentration factor
This line passes through the central portion of the experimental failure points and hence
should be the best predictor of failure. The parabolic nature of Equation (7.27) poses
problems for the practical implementation of this equation.
Goodman Line
The Goodman line proposes connecting the modied endurance limit on the alternating
stress axis with the ultimate strength in tension on the mean stress axis in Figure 7.13 by a
straight line, or
K
f
a
S
e
+
m
S
ut
=
1
n
s
(7.28)
Note the linearization of Equation (7.28) relative to Equation (7.27). Equation (7.28) ts
experimental data reasonably well and is simpler to use than Equation (7.27). The starting
and ending points for the Goodman and Gerber lines are the same in Figure 7.13, but
between these points the Goodman line is linear and the Gerber line is parabolic.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
294 PART 1
Fundamentals
EXAMPLE 7.6 Given: A straight, circular rotating beam with a 30-mm diameter and a 1-m length has an axial
load of 30,000 Napplied at the end and a stationary radial load of 400 N. The beamis cold-drawn,
and the material is AISI 1040 steel. Assume that k
s
= k
r
= k
t
= k
m
= 1.
Find: The safety factor for innite life by using the Goodman line.
Solution: From Table A.1 for AISI 1040 steel S
u
= 520 MPa. From Figure 7.10(a) for the
cold-drawn process and S
u
= 520 MPa, the surface nish factor is 0.78. From Equation (7.16)
the modied endurance limit while making use of Equation (7.7) is
S
e
= k
f
k
s
k
r
k
t
k
m
S
e
= (0.78)(1)(1)(1)(1)(0.45)(520) = 182.5 MPa
The bending stress from Equation (4.48) gives the alternating stress that the beam experiences as
a
=
Mc
I
=
(64)(400)(1)(0.03/2)
(0.03)
4
Pa = 150.9 MPa
The mean stress due to the axial load is
m
=
P
a
A
=
(30,000)(4)
(0.03)
2
Pa = 42.44 MPa
For an unnotched beam K
f
= 1. From Equation (7.28)
K
f
a
S
e
+
m
520
=
1
n
s
\
(1)(150.9)
182.5
+
42.44
520
=
1
n
s
n
s
=
1
0.9084
= 1.101
Using the Goodman line, the safety factor for innite life is 1.101.
Soderberg Line
The Soderberg line is conservative and is given as
K
f
a
S
e
+
m
S
yt
=
1
n
s
(7.29)
Note from Figure 7.13 and Equations (7.28) and (7.29) that the ultimate strength in
the Goodman relationship has been replaced with the yield strength in the Soderberg
relationship.
Yield Line
To complete the possibilities, the yield line is given. It is used to dene yielding on the rst
cycle, or
a
S
yt
+
m
S
yt
=
1
n
s
(7.30)
This completes the description of the theories presented in Figure 7.13.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
max
max
min
min
m
+
c d
45
45
S
u
S
y
S
u
S
e
/K
f
S
y
S
y
S
e
/K
f
Figure 7.14 Complete modied Goodman diagram, plotting stress as
ordinate and mean stress as abscissa.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
296 PART 1
Fundamentals
Table 7.5 Equations and range of applicability for construction of complete
modied Goodman diagram
Line Equation Range
AB max =
Se
Kf
+m
_
1
Se
Su Kf
_
0 m
Sy Se/Kf
1 Se/(Kf Su)
BC max = Sy
Sy Se/K f
1 Se/(K f Su)
m Sy
CD min = 2m Sy
Sy Se/K f
1 Se/(K f Su)
m Sy
DE min =
_
1 +
Se
K f Su
_
m
Se
K f
0 m
Sy Se/K f
1 Se/(K f Su)
EF min = m
Se
K f
Se
K f
Sy m 0
FG min = Sy Sy m
Se
K f
Sy
GH max = 2m + Sy Sy m
Se
K f
Sy
HA max = m +
Se
K f
Se
K f
Sy m 0
As an example of the way the modied Goodman diagram aids in visualizing the
various combinations of uctuating stress, consider the mean stress indicated by point L in
Figure 7.14. The Goodman criterion indicates that this stress can uctuate between points
M and N. The nonzero mean cyclic uctuation is sketched on the right of the gure.
Also shown in Figure 7.14 are the four regions of mean stress on the abscissa. Table 7.6
gives the failure equation for each of these regions as well as the validity limits for each
equation.
Table 7.6 Failure equations and validity limits of equations for four regions
of complete modied Goodman relationship
Region in Failure Validity limits
Figure 7.14 equation of equation
a max 2m =
Sy
ns
Sy m
Se
K f
Sy
b max m =
Se
ns Kf
Se
K f
Sy m 0
c max +m
_
Se
Kf Su
1
_
=
Se
ns Kf
0 m
Sy Se/K f
1 Se/(K f Su)
d max =
Sy
ns
Sy Se/K f
1 Se/(K f Su)
m Sy
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
max
= b
s
m
+
C (a)
where b
s
= slope and
C = intercept. But
m
= 0
max
= 30 ksi \
C = 30 ksi
\
max
= b
s
m
+30 ksi (b)
But
max
= 90 ksi when
m
= 90 ksi
Substituting this into Equation (b) gives b
s
=
2
3
. Therefore, line AB is given as
max
=
2
3
m
+30 ksi (c)
\ Point A = (0, 30 ksi)
Point B
max
= 60 ksi
From Equation (c)
60 ksi =
2
3
m
+30 ksi
\
m
= 45 ksi
Point B = (45 ksi, 60 ksi)
Point C = (60 ksi, 60 ksi)
Line EF (Goodman line) is
min
= b
s
m
+
C (d)
Given that
m
= 0
min
= 30 ksi
\
C = 30 ksi (e)
and
min
= b
s
m
30 ksi
Given that
m
= 90 ksi
min
= 90 ksi
\ 90 ksi = b
s
(90) ksi 30 ksi (f)
\ b
s
=
4
3
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
298 PART 1
Fundamentals
Substituting Equations (e) and ( f ) into Equation (d) gives
min
=
4
3
m
30 ksi
\ Point E = (0, 30 ksi)
Point D = (45 ksi, 30 ksi)
Figure 7.15 shows a plot of these points as well as the shaded nonfailure region.
90
60
B C
F
D
A
30
120
45
S
t
r
e
s
s
,
k
s
i
0
30
E 45
Mean stress
m
ksi
90
Figure 7.15 Modied Goodman diagram for
Example 7.7.
EXAMPLE 7.8 Given: For the same two bars given in Example 7.4, the load is cyclically varying between 2
and 10 ksi.
Find: Using the modied Goodman relationship, determine the safety factor guarding against
fatigue failure.
Solution: From Table A.1 for low-carbon steel (AISI 1020)
S
u
= 57 ksi and S
y
= 43 ksi (a)
From Example 7.4 the modied endurance limit is
S
e
K
f
= 10.85 ksi for notched bar (b)
S
e
K
f
= 21.55 ksi for unnotched bar (c)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
m
=
max
2
= 6 ksi (d)
Because the loading is tensile, the region in Figure 7.14 is either c or d. To establish which,
evaluate as follows:
Notched bar:
S
y
S
e
/K
f
1 S
e
/(S
u
K
f
)
=
(43 10.85)10
3
1 10.85/57
psi = 39.71 ksi
\ 0
m
S
y
S
e
/K
f
1 S
e
/(S
u
K
f
)
or 0 6 ksi 39.71 ksi
\ Region c
Unnotched bar:
S
y
S
e
/K
f
1 S
e
/(S
u
K
f
)
=
(43 21.55)10
3
1 21.55/57
psi = 34.49 ksi
\ 0
m
S
y
S
e
/K
f
1 S
e
/(S
u
K
f
)
or 0 6 ksi 34.49 ksi
\ Region c
From Table 7.5 for region c failure will occur if
max
m
=
S
e
K
f
_
1
n
s
m
S
u
_
Notched bar: 10 6 = 10.85
_
1
n
s
6
57
_
n
s
= 2.110
Unnotched bar: 10 6 = 21.55
_
1
n
s
6
57
_
n
s
= 3.438
Neither bar has a safety factor that indicates an innite life cannot be achieved. However, if there
are additional effects (such as corrosion) that apply equally to the two bars, then it is expected
that the notched bar will fail before the unnotched bar.
7.9.2 BRITTLE MATERIALS
Figure 7.16 shows the alternating stress ratio as a function of the mean stress ratio for axially
loaded cast iron. The gure is skewed since the compressive strength is typically several
times greater than the tensile strength. Until recently, the use of brittle materials in a fatigue
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
300 PART 1
Fundamentals
4.0
Mean stress ratio
m
/S
u
A
l
t
e
r
n
a
t
i
n
g
s
t
r
e
s
s
r
a
t
i
o
a
/
S
u
0
0.5
1.0
1.5
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 0.5 1.0
(0.4)(0.9) = 0.36
S
e
S
u
Figure 7.16 Alternating stress ratio as function of mean stress ratio
for axially loaded cast iron.
environment has been limited to gray cast iron in compression. Now, however, carbon bers
and ceramics have had signicant usage in fatigue environments.
InFigure 7.16the dimensionalizationof the alternatingandmeanstresses is withrespect
to the ultimate strength rather than to the yield strength, as done for ductile materials. Also,
the compressive mean stress permits large increases in alternating stress.
For brittle materials a stress raiser increases the likelihood of failure under either steady
or alternating stresses, and it is customary to apply a stress concentration factor to both. Thus,
designers apply the fatigue stress concentration factor K
f
to the alternating component of
stress for ductile materials but apply the stress concentration factor K
c
to both the alternating
and mean components of stress for brittle materials.
For a single normal stress in brittle materials, the equation for the safety factor with a
steady stress
m
and with the ultimate tensile strength S
ut
as the basis of failure is
n
s
=
S
ut
K
c
m
(7.31)
With an alternating stress
a
and a modied endurance limit S
e
n
s
=
S
e
K
c
a
(7.32)
For a single shear stress on a brittle component and with a steady shear stress
m
, the safety
factor is
n
s
=
S
ut
K
cs
m
(1 + S
ut
/S
uc
)
(7.33)
With an alternating shear stress
a
and a modied endurance limit S
e
,
n
s
=
S
e
K
cs
a
(1 + S
ut
/S
uc
)
(7.34)
7.10 INFLUENCE OF MULTIAXIAL STRESS STATES
The previous sections have considered fatigue failures for uniaxial stress states. Most ma-
chine element applications encounter more complicated loading conditions. Two special
cases are important. The rst situation exists where the applied stresses are in phase, a
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
e
=
_
2
a,1
+
2
a,2
+
2
a,3
a,1
a,2
a,2
a,3
a,1
a,3
(7.35)
If the stress state is two-dimensional (with
a,3
= 0), this equation can be written as
e
=
_
2
a,1
+
2
a,2
a,1
a,2
(7.36)
The safety factor can then be calculated from
n
s
=
S
e
or n
s
=
S
e
(7.37)
Here S
f
is used for nite-life applications and is the fatigue strength at the desired life; S
e
is used for innite-life applications.
All fatigue strength reduction factors, including stress concentration effects, should be
used in the application of Equations (7.35) to (7.37).
Simple Multiaxial Stresses with Nonzero Mean
A number of studies have addressed the situation where the applied stresses are in phase,
but the mean stress is non-zero. Two of the more common theories are referred to as the
Sines method and the von Mises method. These theories use the common approach of
dening an effective alternating and mean stress and then inserting these effective stresses
in the modied Goodman failure criterion given in Table 7.5.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
302 PART 1
Fundamentals
The approach of Sines (1955) uses the following equivalent stresses:
a
=
_
(
a,x
a,y
)
2
+(
a,y
a,z
)
2
+(
a,z
a,x
)
2
+6
_
2
a,xy
+
2
a,yz
+
2
a,zx
_
2
(7.38)
m
=
m,x
+
m,y
+
m,z
(7.39)
For a two-dimensional stress state (
z
=
xz
=
yz
= 0), these equations can be simplied
to, respectively,
a
=
_
2
a,x
+
2
a,y
a,x
a,y
+3
2
a,xy
(7.40)
m
=
m,x
+
m,y
(7.41)
Note that the mean component of shear stress does not appear in these equations. This
is acceptable for some circumstances, but is nonconservative for situations where a stress
concentration such as a notch or llet is present. For this reason, another approach using
the von Mises effective stresses denes the effective alternating and mean stresses as
a
=
_
(
a,x
a,y
)
2
+(
a,y
a,z
)
2
+(
a,z
a,x
)
2
+6(
2
a,xy
+
2
a,yz
+
2
a,zx
)
2
(7.42)
m
=
_
(
m,x
m,y
)
2
+(
m,y
m,z
)
2
+(
m,z
m,x
)
2
+6(
2
m,xy
+
2
m,yz
+
2
m,zx
)
2
(7.43)
or, for a two-dimensional stress state with
a,z
=
m,z
=
a,xz
=
a,yz
=
m,xz
=
m,yz
= 0,
a
=
_
2
a,x
+
2
a,y
a,x
a,y
+3
2
a,xy
(7.44)
m
=
_
2
m,x
+
2
m,y
m,x
m,y
+3
2
m,xy
(7.45)
One of the difculties in applying these theories is that conicts may arise in determin-
ing stress concentration factors and fatigue strengths depending on the loading condition
selected. Recall from Equation (7.8) that S
l
will vary depending on whether the loading
is bending, axial, or torsional. If the loading is a combination of these loadings, it is not
clear how to calculate S
l
. Similarly, stress concentration factors can be dened based on
the loading; thus it is not obvious which value to use in Table 7.5. Recognizing that an
experimental verication of a design is imperative for critical applications, it is reasonable
to follow any of the following approximations:
1. Perform a worst-case scenario, using the smallest resulting strengths and largest stress
concentrations that result from the loading.
2. Since mode I failure is usually most critical, calculate strengths and stress concentra-
tions based on axial loads when they are present. If normal stresses are present due to
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
SEQA
=
2
_
1 +3
_
_
2
+
_
1 +6
cos 2 +9
_
_
4
_
1/2
(7.46)
where = bending stress amplitude including stress concentration effects
= torsional stress amplitude including stress concentration effects
= phase angle between bending and torsion
Equation (7.46) can be used to obtain both mean and alternating components of stress.
It can be shown by comparing Equations (7.46) and (7.42) to (7.43) that the von
Mises approach is conservative for any phase difference or stress ratio. However, this is
true only for high-cycle fatigue and can be nonconservative for nite-life applications.
Since shafts are usually designed for innite life, this is rarely a concern.
3. Behavior for a given loading and material combination may be well quantied within
an organization or can be found in the technical literature.
7.11 FRACTURE MECHANICS APPROACH TO FATIGUE
With the increasing interest in materials without clear endurance limits, special attention
must be paid to damage accumulation and replacing fatigued components before cata-
strophic failure can occur. Indeed, this is a main design challenge in the aircraft industry,
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
304 PART 1
Fundamentals
where aluminum alloys, although they have no endurance limit, are used because of their
high strength/weight ratios. Routine nondestructive evaluation to determine the size of aws
in stress-bearing members is conducted to identify and remove suspect components. This
approach is called fault-tolerant design since it recognizes the presence of defects and
allows the use of material as long as the defects remain smaller than a critical size.
Parris et al. (1961) hypothesized that crack growth in a cyclic loading should follow
the rule
dl
c
dN
= C(K)
m
(7.47)
where dl
c
/dN is the change in crack length per load cycle (l
c
is the crack length and N is
the number of stress cycles), C and m are empirical constants, and K is the stress intensity
range, dened as
K = K
max
K
min
(7.48)
where K
max
and K
min
are the maximum and minimum stress intensity factors, respectively,
around a crack during a loading cycle. For a center-cracked plate with crack length 2l
c
K
max
= Y
max
_
l
c
(7.49)
where
max
is the maximum far-eld stress and Y is a correction factor to account for nite
plate sizes. Appendix Cgives values of Y. Equation (7.47) is known as the Paris power law
and is the most widely used equation in fracture mechanics approaches to fatigue problems.
Suresh (1991) has derived the life of a component based on the Paris power law as
N =
2
(m 2)CY
m
()
m
m/2
_
1
(l
c0
)
(m2)/2
1
(l
cf
)
(m2)/2
_
(7.50)
unless m = 2, when the fatigue life is
N =
1
CY
2
()
2
ln
l
cf
l
c0
(7.51)
where N = fatigue life in cycles
C, m = material constants
Y = correction factor to account for nite plate sizes
= range of far-eld stresses to which component is subjected
l
c0
= initial crack size
l
cf
= critical crack size based on fracture mechanics
EXAMPLE 7.9 Given: An aluminumalloy aircraft component in the formof a 100-mm-wide plate is subjected
to a 100-MPa stress during pressurization of the aircraft cabin. Superimposed on this stress is
a uctuation arising from vibration, with an amplitude of 10 MPa and a frequency of 45 Hz.
Nondestructive crack detection techniques do not detect any aws, but the smallest detectable
aw is 0.2 mm.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
m and a fracture stress of 260 MPa, and assume m = 2.5 and C = 6.9
10
12
m/cycle for in megapascals.
Solution: The critical crack length l
cf
is given by Equation (6.5) and is found to be 1.60 mm. A
worst-case scenario would occur if the largest undetectable aw resulting in the largest value of
Y were located in the geometry. This aw occurs for the double-edge cracked tension specimen,
where Y equals 2.0. Therefore, the life is found from Equation (7.50) as
N =
2
(m 2)CY
m
()
m
m/2
_
1
(l
c0
)
(m2)/2
1
(l
cf
)
(m2)/2
_
= 324 h
If l
c0
= 1.1 mm, the life until fracture is approximately 47 h.
7.12 LINEAR IMPACT STRESSES AND DEFORMATIONS
Thus far this chapter has focused on cyclic load variation; this section focuses on impact
loading. Throughout most of this text, a load is applied to a body gradually such that when
the load reaches a maximum, it remains constant or static. For the material covered thus far
in this chapter, the loading is dynamic (i.e., it varies with time). When loads are rapidly
applied to a body as in impact loading, the stress levels and deformations induced are often
much larger than those in static or cyclic loading.
The properties of the material are a function of the loading speed; the more rapid
the loading, the higher both the yield strength and the ultimate strength of the material.
Figure 7.17 shows the variation of the mechanical properties with loading speed for a
typical mild steel. For average strain rates from 10
1
to 10
3
s
1
the yield strength increases
signicantly.
If no energy is assumed to be lost during impact, the mechanics of impact can be studied
by using conservation of energy. Consider a simple block falling a distance h and striking
a spring compressed a distance
max
before momentarily coming to rest. If the mass of the
spring is neglected and it is assumed that the spring responds elastically, conservation of
energy requires that the kinetic energy be completely transformed to elastic strain energy:
W(h +
max
) =
1
2
(k
max
)
max
(7.52)
where k = spring constant, N/m, and W = weight of block, N. Equation (7.52) can be
expressed as a quadratic equation:
2
max
2W
k
max
2
_
W
k
_
h = 0 (7.53)
Solving for
max
gives
max
=
W
k
+
_
_
W
k
_
2
+2
_
W
k
_
h (7.54)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
306 PART 1
Fundamentals
100
80
U
l
t
i
m
a
t
e
a
n
d
y
i
e
l
d
s
t
r
e
s
s
e
s
S
u
a
n
d
S
y
,
k
s
i
E
l
o
n
g
a
t
i
o
n
,
p
e
r
c
e
n
t
S
y
/
S
u
,
p
e
r
c
e
n
t
60
40
20
Average strain rate, s
1
0
100
80
60
40
20
0
10
6
10
5
10
4
10
3
10
2
10
1
1 10 10
2
10
3
Ratio Sy
/Su
Ultimate strength
Su
Yield strength S
y
Total elongation
Figure 7.17 Mechanical properties of mild steel at room temperature as a
function of average strain rate [Manjoine (1944)].
If the weight is applied statically (or gradually) to the spring, the static displacement is
st
=
W
k
(7.55)
Substituting Equation (7.55) into Equation (7.54) gives
max
=
st
+
_
(
st
)
2
+2
st
h
or
max
=
st
_
1 +
_
1 +
2h
st
_
(7.56)
From the maximum displacement in Equation (7.56) the maximum force is
P
max
= k
max
(7.57)
Recall that in dropping the block from some distance h the maximum force P
max
on impact
is instantaneous. The block will continue to oscillate until the motion dampens and the
block assumes the static position. This analysis assumes that when the block rst makes
contact with the spring, the block does not rebound from the spring (does not separate from
the spring). Making use of the spring constant k in Equations (7.55) and (7.57) relates the
static and dynamic effects.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
max
st
=
P
max
W
= 1 +
_
1 +
2h
st
(7.58)
Note that once the impact factor is known, the impact load, stresses, and deections can be
calculated. The impact stress is
=
P
max
A
(7.59)
where A = area of surface impacting spring, m
2
.
If, instead of the block dropping vertically, it slides with a velocity u on a surface that
provides little frictional resistance (so that it can be neglected) and the block impacts the
spring, the blocks kinetic energy Wu
2
/(2g) is transformed to stored energy in the spring,
or
1
2
_
W
g
_
u
2
=
1
2
k
2
max
(7.60)
where g =gravitational acceleration, 9.807 m/s
2
. Note that the right side of Equation (7.52)
is identical to the right side of Equation (7.60). Solving for
max
in Equation (7.60) gives
max
=
_
Wu
2
gk
(7.61)
By using Equation (7.55), Equation (7.61) becomes
max
=
_
st
u
2
g
(7.62)
In both situations the moving body (the block) is assumed to be rigid, and the stationary
body (the spring) is assumed to be deformable. The material is assumed to behave in a
linear elastic manner. Thus, whether a block falls a distance h and impacts on a spring, or a
block moves at a velocity u and impacts on a spring, the formulation for the deformation,
the impact force, or the impact stress can be determined.
EXAMPLE 7.10 Given: A diver jumps up 2 ft on the free end of a diving board before diving into the water.
Figure 7.18 shows a sketch of the diver and the dimensions of the diving board. The supported
end of the diving board is xed. The modulus of elasticity is 10.3 10
6
psi, and the yield strength
is 30 ksi. The weight of the diver is 200 lb.
Find: The safety factor for impact loading based on yielding.
Solution: Equations (7.54) and (7.55) should be used to determine the maximum deection at
impact at the end of the diving board. The spring rate is not given and will need to be determined.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
308 PART 1
Fundamentals
5 ft
(a) (c) (b)
2 ft
18 in
1.5 in
y
x
P
V
M
Figure 7.18 Diver impacting diving board, used in Example 7.10. (a) Side
view; (b) front view; (c) side view showing forces and coordinates.
The forces acting are shown in Figure 7.18(c). From Equation (5.3)
d
2
y
dx
2
=
M
EI
=
Px
EI
Integrating this equation gives
EI
dy
dx
=
Px
2
2
+C
1
where C
1
= integration constant. Integrating again gives
EIy =
Px
3
6
+C
1
x +C
2
where C
2
= integration constant. The boundary conditions are
1. x = l, y
= 0 C
1
=
Pl
2
2
2. x = l, y = 0 C
2
=
Pl
3
3
The deection at the end of the board is of interest when x = 0.
\ EIy =
Px
3
6
+
Pl
2
x
2
Pl
3
3
or
6EIy
P
= x
3
+2l
2
x 2l
3
\ = y|
x=0
=
Pl
3
3EI
The spring constant is
k =
P
=
3EI
l
3
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
st
=
W
k
=
200 lb
724.3 lb/in
= 0.2761 in
From Equation (7.58) the impact factor is
I
m
=
max
st
= 1 +
_
1 +
2h
st
= 1 +
_
1 +
2(24)
0.2761
= 14.22
\
max
= I
m
st
= (14.22)(0.2761) = 3.927 in
From Equation (7.57)
P
max
= k
max
= (724.3)(3.927) = 2844 lb
The maximum bending stress from Equation (4.48) is
max
=
Mc
I
=
(2844)(5)(12)(0.75)
5.063
psi = 25.28 ksi
The yield stress is given as 30 ksi. The safety factor for yielding is
n
s
=
S
y
max
=
(30)(10
3
)
(25.28)(10
3
)
= 1.187
Thus, n
s
> 1 and thus failure should not occur; however because the safety factor is just above 1,
the margin of safety is a minimum.
Case Study 7.1 POWER PRESS BRAKE STUD DESIGN REVIEW
Given: An engineer working for a manufacturer of me-
chanical power presses is assisting in the development of a
new size of machine. When starting to design a new brake
stud, the engineer nds the information in Figure 7.19
from an existing computer-assisted drawing (CAD) of a
machine with the closest capacity to the new project. The
brake stud supports the brake on the camshaft of the me-
chanical power press. The brake actuates with every cycle
of the press, stopping the ram at the top-dead-center posi-
tion so that an operator can remove a workpiece from the
dies and insert another workpiece for the next cycle. If the
stud fails, the press could continue to coast downward and
could result in a serious injury.
FocusingonsectionA-A, the engineer recognizes that
no llet radius had been specied for the brake stud for ma-
chines sold previously. Conversations with machinists who
routinely worked on the part led to the conclusion that the
common practice was to undercut the llet to make sure
assembly was complete; the llet radius in effect was the
radius of the machine tool insert, a value as low as
1
8
in.
The immediate concern is whether a product recall is in
(continued)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
310 PART 1
Fundamentals
Case Study (Continued)
2.25 in 1 in
1.375 in. 3.0625 in
R = 0.375 in
Machine frame
AA
P
x
Shoulder
Figure 7.19 Dimensions of existing brake stud design.
order, since a brake stud failure could result in a machine
operators hands being in a die while the machine fails
to stop after a cycle. The supervisor shares the engineers
fears but assigns one immediate task.
Find: Is the machine safe as manufactured? What con-
cerns exist for this machine in service? Is a recall neces-
sary? The carbon steel used for the stud has a minimum
ultimate strength of 74.5 ksi, and no yield strength was
prescribed in the drawing.
Solution:
(a) Applied stresses. Figure 7.19 shows that the power
press frame and the brake stud share the loading if
the stud-retaining nut is tight. If the nut is loose, the
stud can conceivably carry the entire loading. Exist-
ing brake design calculations reveal that the brake
acts over one-quarter of the camshaft revolution and
has a relatively constant peak load of almost 1000 lb.
Figure 7.20 shows the shear and bending moment di-
agrams for the stud and the loading with respect to
time. The critical loading is bending (shear must also
be considered, but in this case, as with most shafts,
bending stresses are the critical consideration). The
maximum stress in the absence of stress raisers is
given by
max
=
Mc
I
=
(1000)(3.0625/2 +1.375)(0.5)
(/4)(0.5)
4
psi
= 29.6 ksi
The loading is not completely reversing; the mean
stress is 14.8 ksi, and the stress amplitude is 14.8 ksi
(Fig. 7.20).
The stress raiser cannot be ignored in this case.
From Figure 6.5 the theoretical stress concentration
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
e
= 0.5S
ut
=
37.25 ksi. The following correction factors are
obtained:
Surface nish: A machined nish so that from
Equation (7.21)
k
f
= 2.70(74.5)
0.265
= 0.86
Size factor: From Equation (7.22)
k
s
= 0.869(1)
0.112
= 0.869
Reliability factor (chosen from Table 7.3 for a
99% reliability)
k
r
= 0.82
The endurance limit for the steel brake stud is then
given by Equation (7.16) as
S
e
= (37.25)(0.86)(0.869)(0.82) = 22.8 ksi
(c) Fatigue criterion. Using the Goodman line
[Eq. (7.28)] gives the safety factor as
K
f
a
S
e
+
m
S
ut
=
1.6(14.8)
22.8
+
14.8
74.5
=
1
n
s
n
s
= 0.8
(continued)
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
312 PART 1
Fundamentals
Case Study (Concluded)
Thus, the designis inadequate. However, if there were
no stress concentration at all, the safety factor would
increase to 1.17, certainly a low safety factor regard-
less.
Discussion: The machine manufacturer had not experi-
enced failures in the brake stud, even though the existing
design had been on the market for over three decades.
Through laboratory testing it was determined that if the
brake-stud-retaining bolt were loose, the brake would
make a noticeably loud squeal and the stud would clank.
Undoubtedly, such problems were promptly identied by
maintenance personnel, and the stud bolt was retightened
whenever it became loose in service. As long as the brake
stud was snug against the machine support, or stud shoul-
der, the safety factor for innite life was high. It was ul-
timately determined that a recall was not in order, but a
service memorandumwas issued to all previous customers
informing them of the need to properly maintain the tight-
ness of the stud-retaining bolt. Subsequent service manuals
also mentioned this important consideration.
7.13 SUMMARY
Failures in components or structures are often caused by uctuating stresses. If the stress
variation sequence repeats itself, it is called cyclic. The various cyclic patterns were de-
scribed. The fatigue strength versus the logarithm of the number of cycles to failure was
presented for materials with and without an endurance limit. Ferrous materials tend to have
endurance limits, and nonferrous materials tend not to have endurance limits. The endurance
limit for ferrous materials is a function of the type of loading the component is subjected
to. Also, low-cycle fatigue failure has been classied as occurring at less than 10
3
cycles
and high-cycle fatigue failure as occurring above 10
3
cycles.
The endurance limit has been experimentally determined for a specimen of specic
size with a mirrorlike surface nish. The specimen was precisely prepared and tested under
controlled conditions. In practice, conditions differ signicantly from those in a test situa-
tion. To more accurately characterize the conditions that prevail in practice, the modied
endurance limit is used. Correction factors for the endurance limit include effects of surface
nish, size, stress concentration, and other common effects.
Various approaches for estimating when failure will occur under nonzero mean cyclic
stresses were also considered. The Soderberg, Gerber, Goodman, and modied Goodman
theories were presented. These theories allowthe variation of mean and alternating stresses.
The construction of a complete modied Goodman diagram was shown, and fatigue and
yield criteria were used. Failure equations were given for specic regions of the modied
Goodman diagram.
Most of the results presented were for ductile materials, but the behavior of brittle
materials was briey described. The major difference between brittle and ductile materials
is that the compressive and tensile strengths are nearly identical for ductile materials whereas
for brittle materials the compressive strengthis several times greater thanthe tensile strength.
Impact loading has to be considered when loads are rapidly applied. The stress levels
and deformations induced are much larger than with static or cyclic loading. The more rapid
the loading, the higher the yield and ultimate strengths of a material. By equating the kinetic
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
shot peening cold working process to surfaces that severely retards crack development and
propagation.
simple multiaxial stress multiaxial fatigue loading where the forces and torques are in phase.
Sines method non-conservative method used for simple multiaxial stresses with non-zero mean.
SN diagram plot of stress level versus number of cycles before failure.
Soderberg line theory connecting modied endurance limit and yield strength on plot of
alternating stress versus mean stress.
stress amplitude one-half of stress range.
stress range difference between maximum and minimum stresses in cycle.
stress ratio ratio of minimum to maximum stresses.
von Mises method conservative method used for simple multiaxial stresses with non-zero mean.
W ohler diagram same as SN diagram.
yield line failure criterion that postulates yielding on rst cycle of cyclic loading with nonzero
mean.
RECOMMENDED READINGS
Juvinall, R. C. (1967) Engineering Considerations of Stress, Strain, and Strengths, McGraw-Hill,
New York.
Madayag, A. F. (1969) Metal Fatigue: Theory and Design, Wiley, New York.
Rice, R. C., ed. (1988) Fatigue Design Handbook, Society of Automotive Engineers, Warrendale,
PA.
Suresh, S. (1998) Fatigue of Materials, 2nd ed., Cambridge University Press, Cambridge,
United Kingdom.
Zahavi, E., and Torbilo, V. (1996) Fatigue Design, CRC Press, Boca Raton, FL.
REFERENCES
American Society of Materials International, (1975), Metals Handbook, vol. 11: Failure Analysis and
Prevention, Metals Park, Ohio.
Dowling, N. E. (1993) Mechanical Behavior of Materials, Prentice-Hall, Englewood Cliffs, NJ.
Johnson, R. C. (1967) Predicting Part Failures, Part I, Machine Design, vol. 45, p. 108.
Juvinall, R. C. (1967) Engineering Considerations of Stress, Strain, and Strength, McGraw-Hill,
New York.
Juvinall, R. C., and Marshek, K. M. (1991) Fundamentals of Machine Component Design, Wiley,
New York.
Landgraf, R. W. (1968) Cyclic Deformation and Fatigue Behavior of Hardened Steels, Report no. 330,
Department of Theoretical and Applied Mechanics, University of Illinois, Urbana.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
m
= 0,
r
= 4 MPa, R
s
= 1.
Section 7.5
7.2 The jack for a Volvo consists of a mechanismin which the lift screwextends horizontally through
two corners of the mechanism while the other two corners apply a force between the ground
and the car to be lifted. The maximum compressive stress in the jack is 190 MPa when the car
is jacked up so high that both wheels on one side of the car are in the air and the load on the
jack is 8000 N. How many times can the jack be used for a small truck that weighs 6 tons and
loads the jack to 17,000 N before it fails from fatigue? The jack material is AISI 1080 steel.
Ans. 6050 cycles.
Section 7.6
7.3 The shaft shown in sketch a rotates at high speed while the imposed loads remain static. The
shaft is machined from ground, high-carbon steel (AISI 1080). If the loading were sufciently
large to produce a fatigue failure after 1 million cycles, where would the failure most likely
occur? Show all necessary computations and reasoning.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
316 PART 1
Fundamentals
1 in
1 in. 2 in.
2 in.
3 in. 3 in.
1 in. 1 in.
r = in.
16
r = in.
8
4
r = in.
8
P
P/2 P/2
y
x
1
1 1 1
Sketch a, for Problem 7.3
Section 7.7
7.4 For each of the AISI 1040 cold-drawn steel bars shown in sketches b and c, determine
a) The static tensile load causing fracture
b) The alternating (completely reversing) axial load P that would be just on the verge of
producing eventual fatigue failure
25 mm
25 mm
(b) (c)
P
P
P
P
25 mm
30 mm
25 mm
r = 2.5 mm
r = 2.5 mm
Sketches b and c, for Problem 7.4
7.5 A stepped shaft, as shown in sketch d, was machined from high-carbon steel (AISI 1080). The
*
loading is one of completely reversed torsion. During a typical 30 s of operation under overload
conditions, the nominal stress in the 1-in-diameter section was calculated to be as shown in
sketch e. Estimate the life of the shaft when it is operating continually under these conditions.
Ans. 4.79 h.
2 in
r = 0.1 in
1 in
Sketch d, for Problem 7.5
Fundamentals
11-mm-diameter central shaft in the pedal. The shaft is made of AISI 4340 alloy steel tempered
at a low temperature to give an ultimate stress of 2 GPa. Find the fatigue stress concentration
factor for the llet, and calculate the maximum allowable pedal force for innite life if the force
from the foot is applied 70 mm from the llet. Ans. 1090 N.
7.13 During the development of a new submarine a 1-to-20 model was used. The model was tested
to nd if there was any risk of fatigue failure in the full-size submarine. To be on the safe side,
the stresses in the model were kept 25% higher than the stresses in the full-size submarine. Is
it possible to conclude that the full-size submarine will be safe if the model was safe at a 25%
higher stress level?
7.14 During the development of a newcar it was found that the power fromthe motor was too high for
*
the gearbox. When maximumtorque was transmitted through the gearbox, 50%of the gearboxes
failed within the required lifetime of 800 h. How much stronger should the gearbox be made to
ensure that only one gearbox out of 1000 would fail within 800 h?
7.15 A tension member in service has been inspected, and a fatigue crack has been discovered, as
shown in sketch f . A proposed solution is to drill a hole at the tip of the crack, the intent being
to reduce the stress concentration at the crack tip. The material is AISI 1040 medium-carbon
steel and was produced through forging. The load is a completely reversing 3500 lb, and the
original design is as shown.
a) What was the original factor of safety?
b) What is the smallest drilled hole that restores the safety factor to that of the original
design? (Use the nearest
1
32
-in increment that is satisfactory.) Use a reliability of 90%.
1.25 in 1 in
r = 1/8in
Fatigue crack
Hole to be
drilled
1.25 in
0.75 in
r
+
16
1
8
Sketch f , for Problem 7.15
Section 7.8
7.16 Truck gearboxes are dimensioned for innite life at maximum torque regarding contact stresses
and bending stresses for all gear steps. Car gearboxes are dimensioned for nite life according
to typical running conditions. Maximum torque for the rst gear can typically be maintained
only 3 to 6 s for each acceleration before the maximum speed is reached. If a driver accelerates
at full power 20 times per day for 20 years on the rst gear, the required life is only 60 to 120 h.
The normal load spectrum gives a life of 200,000 km for 99% of the gearboxes. A driver uses
the car twice a year to move his 10-ton boat 50 km (the distance between his home and the
harbor) during the 10-year life of the car. Calculate how much of the 200,000-km nominal life
of the gearbox is consumed by the boat moving if the life is inversely proportional to the load
raised to the 3.5 power. Assume that during the move the gearbox load is 4 times higher than
normal.
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
4
in
4
1 in
8
3
7 3
1
Sketch g, for Problem 7.21
7.22 The at bar shown in sketch h is made of cold-drawn, high-carbon steel (AISI 1080). The
*
cyclic, nonzero, mean axial load varies from a minimum of 2 kN to a maximum of 10 kN. Using
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
320 PART 1
Fundamentals
the Goodman failure theory, determine the safety factors for the hole, llet, and groove. Also,
indicate where the at bar will rst fail. Ans. n
s,hole
= 2.97, n
s,llet
= 3.25.
P
P
B
C
A
4-mm R
25 mm
d = 5 mm
t = 5 mm
35 mm
3-mm R
Sketch h, for Problem 7.22
7.23 A straight, circular rotating beam with a diameter of 30 mm and a length of 1 m has an axial
load of 3000 N applied at the end and a stationary radial load of 400 N. The beam is cold-drawn,
and the material is titanium. Note that k
s
= k
r
= k
m
= 1. Find the safety factor n
s
for innite
life by using the Goodman line.
7.24 The cold-drawn AISI 1040 steel bar shown in sketch i is subjected to a tensile load uctuating
between 800 and 3000 lb. Estimate the factor of safety, using the Goodman criterion.
P P
-in through hole
1 in
3/8 in
16
1
4
16
3
8
Sketch i , for Problem 7.24
7.25 The cantilever shown in sketch j carries a downward load F that varies from 300 to 700 lb.
a) Compute the resulting safety factor for static and fatigue failure if the bar is made from
AISI 1040 steel.
b) What llet radius is needed for a fatigue failure safety factor of 3.0 (use a constant notch
sensitivity)?
Sketch j , for Problem 7.25
P1: GIG
ham65328_Ch07 Hamrock-0402T PB466-Hamrock-v28.cls April 10, 2004 14:50
CHAPTER 7
Fundamentals
25 m
500 mm
2
Elevator shaft
Sketch l , for Problem 7.31
7.32 Modern kitchen drawers have small rubber springs mounted onto the sides of the inside of the
*
front plate to take up the force and stop the drawer when it is being closed. The spring constant
for each of the two rubber springs is 400 kN/m. The drawer is full of cutlery, which weighs 5 kg,
and is closed with a speed of 0.5 m/s. Calculate the maximum force in each rubber spring if the
drawer itself weighs 1 kg and
a) The cutlery is in a container that is xed to the drawer so that it moves with the drawer.
b) The cutlery is in a plastic container that can slide 80 mm with a coefcient of friction of
0.25 inside the drawer.
7.33 Car doors are easy to slam shut but difcult to press shut by hand force. The door lock has two
latches, the rst easily engaged and the second requiring the rubber seal around the door to be
quite compressed before it can engage. The rubber seal has a spring constant of 50,000 N/m at
the locked position for the door, and the mass moment of inertia for the door around its hinges
is 2.5 kg m
2
. The distance between the lock and the hinges is 1 m. Calculate the force needed
to press the car door shut at the lock if a speed of 0.8 m/s at the lock slams it shut. Ans. 282.8 N.