Escolar Documentos
Profissional Documentos
Cultura Documentos
c 2014 TheATLAS
ISSN: 1933-5423,
Table of Contents
7 Design for Fatigue
7.1 Failure Mechanisms . . . . . . . . . . . . . . . . . . . . . .
7.1.1 Fatigue Mechanism of Failure . . . . . . . . . . . .
7.1.2 Fatigue Interpretation . . . . . . . . . . . . . . . .
7.2 Low Cycle and High Cycle fatigue . . . . . . . . . . . . .
7.2.1 Low Cycle Fatigue . . . . . . . . . . . . . . . . . .
7.2.2 High Cycle Fatigue . . . . . . . . . . . . . . . . . .
7.3 Fatigue Strength Correction Factors . . . . . . . . . . . .
7.3.1 Surface Finish Correction Factor . . . . . . . . . .
7.3.2 Size Correction Factor . . . . . . . . . . . . . . . .
7.3.3 Temperature Correction Factor . . . . . . . . . . .
7.3.4 Reliability Correction Factor . . . . . . . . . . . .
7.3.5 Fatigue Stress Concentration Correction Factor . .
7.3.6 Miscellaneous Correction Factor . . . . . . . . . .
7.4 Fluctuating Stresses . . . . . . . . . . . . . . . . . . . . .
7.4.1 Fatigue Analysis for Fluctuating Normal Stress . .
7.4.2 Fatigue analysis for Combined Fluctuating Stresses
7.5 Miners Rule Cumulative Fatigue Damage . . . . . . . .
7.5.1 Offshore Pipeline Fatigue Application . . . . . . .
7.6 Fracture Mechanics Based Fatigue Analysis . . . . . . . .
7.6.1 Fracture Mechanics Analysis of Fatigue Resistance
ii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
4
5
6
6
10
11
11
13
13
13
15
20
20
26
30
32
40
42
MODULE
7.1
Failure Mechanisms
7.1.1
Figure 7.1:
Bracing failed due to
fatigue is on display
in the Norwegian
Petroleum Museum.1
Majority of the mechanical failures occur from fatigue and examples are almost
everywhere whether it is a minor maintenance failure such as shaft bearings or
a disaster of countrywide significance such as Alexander L. Kielland, Norwegian
semi-submersible drilling rig that capsized while working in the Ekofisk oil field in
March 1980 killing 123 people. The capsize was the worst disaster in Norwegian
waters since World War II. The investigative report settled that the rig collapsed
due to a fatigue crack in one of its six bracings (bracing D-6), which connected
the collapsed D-leg to the rest of the rig (see Figure 7.1 and Figure 7.2).1
1 The Alexander L. Kielland accident, Report of a Norwegian public commission appointed
by royal decree of March 28, 1980, presented to the Ministry of Justice and Police March, 1981
ISBN fig7-B0000ED27N.
Failure was traced to a small 6 mm fillet weld which joined a non-loadbearing flange plate to this D-6 bracing. The weak profile of the fillet weld
causing reduction in fatigue strength and cyclical stresses due to the sea waves
contributed the most in the rigs collapse.
A technical understanding of fatigue mechanisms (damage) and the development of constitutive equations for damage occurrence leading to a crack initiation
and propagation is an important and fundamental problem for engineers. Fatigue crack is a progressive fracture mechanism due to cyclic loading and occurs
in two stages (1) fatigue crack initiation and (2) fatigue crack propagation. Fatigue crack will initiate at a point of discontinuity in the mechanical components,
such as a change in geometry, a keyway, a hole, inspection or stamp marks, internal defects, and manufacturing imperfections. Once a fatigue crack initiates,
it will propagate into the material with each stress cycle and continues as long
as the material is subjected to cyclic stress. Depending on the magnitude and
frequency of stress cycles, the crack may grow over a time and eventually, the
crack advances to a point where the remaining undamaged cross section of the
material cant support any more stress cycles and catastrophic failure occurs.
As an example, shaft manufactured from 300M, ultra-high strength, alloy
steel fractured due to fatigue progression from an intergranular stress crack, initiated due to dot peen marking of the shaft as shown in Figure 7.3(a). The dot
peened engraving created a high stress concentration in the hard surface of the
shaft. The fracture crosses the 0 in the date stamp on the part shown in the
figure. Two events of fatigue progression (zone-1 and zone-2) were observed covering approximately 33% of the fracture surface, prior to final torsional overload
failure of the component. First two zones of events are due to the progressive
development of the crack, while the final torsional overload is due to the catastrophic failure. Intergranular cracking at the initiation site indicates a brittle
surface condition and may be an indicator of excessive residual stress in the
surface of the shaft.2
Thermo Mechanical Fatigue (TMF)
2 Metallurgical Technologies, Inc., Metallurgical Laboratory - Failure Analysis and Metallurgical Engineering Services, Mooresville, NC.
Figure 7.3: (a) Wiew of the fractured input shaft. (b) Wiew of the fracture surface. (Courtecy of Metallurgical Technologies, Inc., Metallurgical Laboratory).
Thermo mechanical fatigue (TMF) occurs when cyclic mechanical loading superimposes with cyclic thermal loading as shown in Figure 4. Since this type
of loading can be more damaging, that needs to be considered when designing
components and equipment operating under high temperatures such as turbine
rotors subjected to the combined influences of cyclic damage arising from strain
transients created during the start-up and shut-down and creep damage resulting
from cyclical thermal loading during the operation. Properties of materials play
an important role in the thermal mechanical fatigue damage formed in the material. TMF loading is often described to be in-phase (IP) or out-of-phase (OP)
as shown in Figure 7.4. In IP cyclic loading, the maximum temperature and
stress appear at the same time. In OP cyclic loading, the material suffers from
compression at high temperature and tension at lower temperatures. Thermal
mechanical fatigue incorporates damage accumulation due to failure mechanisms
of fatigue, environment (oxidation), and creep as explained thereafter.3
Figure 7.4: (a) TMF in-phase cyclic loading. (b) TMF out-of-phase cyclic
loading.
Figure 7.5:
Creep rupture.4
Creep mechanism Creep is the permanent plastic deformation at elevated temperatures and stresses (much less than the high-temperature
yield stress) and one of the most serious damage mechanisms in high temperature environment. Creep deformation occurs by grain-boundary sliding. That is, adjacent grains or crystals move as a unit relative to each
other and may end in a fructure called stress rupture or creep rupture (see
Figure 7.5).4 In-phase thermo mechanical loading is dominated by creep.
Surface Fatigue
Anywhere rolling contact occurs in mechanical parts there is a potential for
surface failure. This will include rolling-element bearings, gears, cams, and
rollers. There are several kinds of surface failures.
indentation Even though, nonrotating ball bearings can have surface
failure when they are under heavy loading, especially if vibration is also
exist. This type of situation will create surface damage, called indentation.
Pitting Pitting is a surface fatigue phenomenon arising in concentrated
rolling and sliding contacts; particles fall out of the surface, leaving surfaces
pockmarked with scattered holes as shown in Figure 7.6.5 Micro-pitting
and surface fatigue can be reduced by using the right lubricant. Speed and
operating temperature main factors effecting on pitting.
Figure 7.6:
Pitting and dent on the
rolling element surface.5
7.1.2
Fatigue Interpretation
features can give a strong indication of the fracturing process. This is the science
of fractography. Since fracture surfaces may have exposed to various environments, generally contain surface debris, corrosion, or oxidation products, they
must be removed before the analysis.6
Figure 7.7:
Fructure surface.
Failure specialists usually inspect the fracture path to decide on the origin
of the crack (see Figure 7.7) as well as other important issues involved with the
failure. With the help of fractographic features, the specialists can also identify
the nature of the loading and decide how the failure occurred. As shown in
Figure 7.7, in general, fatigue failures start on the outside of a shaft where the
stress concentration is exist. in other words, fracture progression comes from
the edge. The instantaneous zone (IZ) carries the load in the instant before the
part breaks. Beach mark indicates the position of the crack tip.7
Figure 7.8: Comparison between a lightly and a heavily loaded shaft (reproduced from reference [8]).
Figure 7.8 shows a comparison between a lightly and a heavily loaded shaft
for both plain bending and rotational bending. By examining the size of the IZ,
one can tell the magnitude of the load on the part for each type of the loadings.
It is evident from Figure 7.8(b) that the rotational bending load causes the
crack to grow in a non-uniform manner. In general, when the divider of the
instantaneous zone does not point to the origin of the crack, it shows there was
a rotational bending involved in the failure cause.8
7.2
Structures are often found to have failed under the action of cyclic loading even
the applied stresses are below the yield strength of the materials. This way of
6 Fracture Surface Analysis. http://www.metallurgist.com/html/Definitions/Fracture
SurfaceAnalysis.htm, accessed: March 12, 2013.
7 Fracture Surface Analysis. http://www.asminternational.org/static/Static%20Files/IP
/Magazine/AMP/V165/I12/amp16512p021.pdf?authtoken=312dc07d8ee28c300b0e8863b90
e42cfaed4d8e8, accessed: March 12, 2013.
8 Neville Sachs, Root Cause Failure Analysis - Understanding Mechanical Failures.
http://www.plant-maintenance.com/articles/rcfa.shtml, accessed: march 14, 2013.
7.2.1
Low cycle fatigue region associated with high loads and short service life. Considerable plastic strain occurs during each cycle. In general, this model is considered for 103 cycles and below. Manson proposed a simplified formula known
as the mothod of universal slopes9
t = e + p = 3.5
Sut 0.12
0.6
N
+ 0.6
f N
E
(7.1)
Where, t is the total strain range which has two components (elastic strain
range, e and plastic strain range, p ), f is the true strain corresponding to
fracture in one reversal (fatigue ductility coefficient), E is Youngs modulus, and
N is the number of strain cycles to failure.
All the constants in Eq.7.1 for predicting the low-cycle fatigue resistance of
a variety of steels can be obtained from a monotonic test and an incremental
step test. Using Eq. 7.1, fatigue life, N can be calculated if the total strain
amplitude, t is known. However, calculation of the total strain would be
troublesome when the discontinuities exist.
7.2.2
In high-cycle fatigue situations, fatigue strength of machine components is analyzed by S-N diagram, obtained in constant amplitude fatigue test by using RR
Moore rotating beam fatigue test machine shown in Figure 7.9. Fatigue testing
machines apply repeated cyclic loads to test specimens.
As shown in Figure 7.10(b), the rotating beam testing machine applies cyclic
load through the use of dead weights as the specimen rotates. Standard test
specimens shown in Figure 7.10(a) are tested at different loads to obtain data
points for plotting an S-N curve. In general, the fatigue strength of materials
is documented by the S-N curve obtained from constant stress amplitude, a
imposed by a pure bending stress test. Typical S-N curve for a steel material
constructed using experimental data is shown in Figure 7.11. As seen from the
figure, experiment is repeated at each stress level to verify the results are consistent and not just an accident. In this figure, Sf0 and Se0 are called uncorrected
fracture strength and endurance strength of a material, respectively.
9 Manson, S. S., 1965. Fatigue: A Complex Subject Some Simple Approximations. Experimental mechanics, Vol. 5, No. 7, P.193.
As the specimen is under cyclic stresses, load acting on the specimen induces
fluctuating bending stress. Load on the specimen will be reduced gradually to
collect the failure data at each stress level until the failure will not occurs. At
this point, corresponding bending stress amplitude induces will be the endurance
strength of the materials. Endurance strength obtained in laboratory by the use
of RR Moore machine will be unrealistic for practical engineering use since the
specimen used for testing is highly polished and homogeneous with no surface
defect. Therefore, endurance strength is not material property such as yield
strength and ultimate strength. It depends on other parameters which should
be incorporated to obtain realistic endurance strength. These parameters are
represented as fatigue strength correction factors with value less than unity.
These correction factors will be discussed in section 7.3.
The endurance limit, Se0 shown in Figure 7.11 is a safe range of fluctuating
stress value, and below this value it is assumed that failure will not occur. The
endurance limit implies that structural members stressed under this limit will
have infinite life; N = 106 cycles. As shown in Figure 7.11, steel and titanium
alloys (ferrous alloys such as low strength carbon and alloy steel; some stainless
steels, irons, and titanium alloys; and also some polymers) have endurance limit,
the stress amplitude below which there appears to be no number of cycles that
will cause failure. On the other hand, structural metals such as aluminum,
copper, magnesium and, nickel do not show endurance limit and will eventually
S 0 = aN b
(7.2)
In Eq.7.2, b is the slope of the S-N curve. To determine the slope, b take log of
both sides of Eq. 7.2
log S 0 = log a + b log N
(7.3)
Coordinates of point A and C in Figure 7.12 must satisfy Eq.7.3. Using point
A, we have
log Sf0 = log a + b log Nf
(7.4)
(7.5)
log(Sf0 /Se0 )
log(Nf /Ne )
(7.6)
log(Sf0 /Se0 )
(log Ne )
log(Nf /Ne )
(7.7)
Sf0
0.9Sut
Nf
103
Se0
0.5Sut
Ne
106
Maximum shear
0.5Se0
Von-Mises
0.577Se0
0.9Sut
103
106
0.5Se0
0.577Se0
0.9Sut
0.9Sut
0.9Sut
0.9Sut
0.9Sut
103
103
103
103
103
100 kpsi
700 MPa
(0.25 0.5)Sut
(0.35 0.5)Sut
0.35Sut
0.35Sut
(0.45 0.65)Sut
106
106
106
106
106
0.5Se0
0.5Se0
0.5Se0
0.5Se0
0.5Se0
0.577Se0
0.577Se0
0.577Se0
0.577Se0
0.577Se0
a=
(Sf0 )2
Se0
(7.8)
a
a
a
(Sf0 )2
Se0
Sf0
= Se0 ( 0 )7/4
Se
Sf0
= Se0 ( 0 )8/5
Se
=
for
Nf = 103
and Ne = 106
(7.9)
for
Nf = 103
and Ne = 107
(7.10)
for
Nf = 103
and Ne = 108
(7.11)
Equation to calculate fatigue life, N for a given stress level, S can be derived
from Eq.7.2
S0
N = Nf ( 0 )1/b
(7.12)
Sf
S-N curve for a specific material can be constructed by using the parameters
given in Table 7.1.
EXAMPLE 7.1
The cantilever beam shown in Figure 7.13 has an ultimate strength of 600
MPa.
(a) Estimate the endurance limit, Se0 of the beam material for an infinite
life.
(b) Estimate the fructure strength, Sf0 of the beam material at 103 cycles.
(c) Determine the fatigue strength, S 0 corresponding to a life of 80 103
cycles.
Figure 7.13:
Cantilever beam.
SOLUTION
(a) From Table 7.1,
Se0 = 0.5Sut = 0.5 600 = 300 MPa
10
Sf0 7/4
)
Se0
From Table 7.1, Nf = 103 and Ne = 106 . Substituting these values into the
above equation, yields
a=
and
b=
5402
= 972 MPa
300
log(Sf0 /Se0 )
log(540/300)
=
= 0.085
log(Nf /Ne )
log(103 /106 )
7.3
Actual value of endurance strength, Se , is less than the ideal uncorrected endurance strength, Se0 obtained from RR Moore rotating beam fatigue test. Since
the specimen used in the laboratory to determine the fatigue strength is different
in size, surface finish, and operating conditions, the endurance limit obtained
from fatigue test should be modified.
Effect of surface finish, size, reliability and the stress concentration should
be considered to obtain the actual value of endurance strength, Se . Using appropriate modifying factors, the corrected endurance strength is defined by
Se = Se0 (ka , ks , kt , kr , kf , km )
Where
Se
Se0
ka
ks
kt
kr
kf
km
(7.13)
11
7.3.1
7.3.2
12
ks
ks
d 0.1133
)
0.3
d 0.1133
(
)
7.62
(
for 0.11 in d 2 in
for 2.79 mm d 51 mm
(7.14)
For larger size machine components, size correction factor, ks , varies from
0.60 to 0.75 for bending and torsion. Since there is no size effect for axial loading,
ks = 1 is assumed. The effect of size is mainly important for the rotating shafts
experiencing bending. When machine components do not rotate or do not have
a round cross section, the equivalent diameter, deq , is used in Eq.7.14. This
is obtained by equating 95% of the stress area of the rotating round shaft to
the same stress area of the component under consideration. The equivalent
diameter, deq , for a non-rotating round section under bending is given by2
deq = 0.37d
(7.15)
The equivalent diameter, deq , for a rectangular section under bending with width
(w) and thickness (t) is given by
d2eq = 0.65wt
(7.16)
EXAMPLE 7.2
A 2 inch steel shaft supported by rolling bearings and subjected a cyclic
loading. The shaft has ultimate strength of 100 kpsi with machined finished.
Estimate the surface factor and the size factor when:
(a) shaft is rotating, and
(b) shaft is not rotating
SOLUTION
(a) From Figure 7.14 for Sut =100 kpsi, surface factor, ka is 0.77. Size factor, ks
for rotating shaft is
ks
ks
d 0.113
)
0.3
2 0.113
(
)
= 0.807
0.3
(
(b) We have the same surface correction factor when the shaft is nonrotating.
Size factor for nonrotating shaft is
ks
deq
ks
deq 0.113
)
0.3
13
7.3.3
kr
1
0.897
0.868
0.814
0.753
0.702
0.659
The endurance limit, Se0 , is typically determined from the standard rotating
beam test machine at room temperature. Since higher temperature will cause
a reduction in strength, Eq.7.17 for temperature correction can be used for
reasonably high temperatures2 .
For T 450 C(840 F )
For T 450 C T 550 C
kt = 1
kt = 1 0.0058(T 450)
kt = 1 0.0032(T 840)
(7.17)
7.3.4
The S-N curve values are average values based on a number of tests resulting
in loads at failure stress, and the endurance strength values are average values
indicating a 50% survival rate. To calculate the endurance strength with a
higher survival rate than 50%, the endurance strength values must be reduced.
On the base of 8% standard deviation for both the stress and endurance strength,
reliability correction factors, kr shown in Table 7.2 are used.
7.3.5
Figure 7.15:
Stress concentration.
Usually, stress analysis calculations assume that the machine components have
uniform sections without any irregularities or discontinuities. however, many
machine components have discontinuities such as holes, grooves, fillets, shoulders, keyways, screws, threads, and notches that will affect the stress distributions in the neighborhood of these irregularities and will cause local stress to
increase. As shown in Figure 7.15, this stress raise is called stress concentration
and stress distributions become uniform away from the hole.
A dimensionless factor, known as stress concentration factor K, is used to
quantify how locally concentrated the stress is. The theoretical stress concentration factor Kt used for normal stress and Kts used for shear stress are defined
by
Kt =
max
o
(7.18)
Kts =
max
o
(7.19)
14
Figure 7.16: Notch sensitivity factors for steel and aluminum(reproduced from
Fundamentals of Machine Component Design, by Robert C. Juvinall and Kurt
M. Marshek, John Willey & Sons, Inc. New York, 2000).
where o and o are the nominal normal applied stress and nominal applied
shear stress, respectively. The value of stress concentration factor is a function
of the geometry, the type of discontinuity, and the type of loading applied on
the machine components. Stress concentration factors for different geometries
and loading types are shown in Figures D-1 through D-5 in Appendix D.
Once the theoretical stress concentration factor is known, the fatigue stress
concentration correction factor, kf can be calculated by
1
Kf =
(7.20)
kf r
Where the fatigue strength reduction factor, kf r , is given by
kf r = 1 + q(Kt 1)
(7.21)
Where q is the notch sensitivity factor given by Figure 7.16 for steel and aluminum.
EXAMPLE 7.3
A machined rotating groved shaft shown in Figure 7.17 is made of steel with
an ultimate strength of Sut =800 MPa. Assume that the shaft diameter,
D=40 mm and the radius of the groove root, r=3 mm. Determine the
fatigue stress concentration correction factor when:
Figure 7.17:
Rotating shaft.
15
SOLUTION
From Figure 7.17 we have D = 40 mm, d =40-2(3)=34 mm, and r=3 mm.
(a) From Figure D-3a (in Appendix D), geometric stress concentration factor, Kt
for bending (for the values D/d=40/34 = 1.18 and r/d = 3/34 = 0.09), is 1.81.
From Figure 7.16, for r=3 mm and Sut =600 MPa (87 Kpsi), notch sensitivity,
q is 0.83, then the fatigue reduction factor, kf r is
kf r
kf
1
1
=
0.6
kf r
1.67
(b) From Figure D-3b (in Appendix D), geometric stress concentration factor,
Kt for torsion (for the values D/d=40/34 = 1.18 and r/d = 3/34 = 0.09), is 1.41.
From Figure 7.16, for r=3 mm and Sut =600 MPa (87 Kpsi), notch sensitivity,
q is 0.85, then the fatigue stress concentration correction factor, kf is
kf r
1 + q(Kt 1)
1 + 0.85(1.41 1) 1.35
kf
=
=
1
kf r
1
0.74
1.35
Above calculations revels that notch sensitivity, q doesnt change much for either case. However, fatigue stress concentration correction factor, kf changes
relatively more.
7.3.6
The miscellaneous correction factor km is used for different factors that are not
easily quantifiable. These factors may include the effect of impact, corrosion,
environment, and metal spraying, among others. Hence, the endurance limit, Se0
determined in the laboratory must be modified by km .
EXAMPLE 7.4
In Example 7.3 if the expected reliability is 95% and the applied load is pure
bending, determine the endurance strength, Se . Assume that the machine
operating temperature is 30 C, shaft diameter, D is 46 mm, and fillet, r is
2mm.
SOLUTION
From Figure 7.14, surface correction factor, ka is found to be 0.74. From Eq.7.14,
the size factor is,
16
ks = (
d 0.1133
46 0.1133
)
=(
)
= 0.815
7.62
7.62
From Eq.7.17, temperature correction factor, kt is 1.0. From Table 7.2, reliability
factor for 95% is 0.868. From Figure D-3a (see Appendix D), stress concentration
factor, Kt for the below values is found to be 2.3.
D
d
r
d
=
=
Kt = 2.3
z }| {
46
= 1.12
42
2
= 0.048
42
Using values of Sut =800 MPa (116 ksi) and r=2 mm, notch sensetivity, q is 0.82
(see Figure 7.16). Fatigue reduction factor, kf r is,
kf r = 1 + q(Kt 1) = 1 + 0.82(2.3 1) = 2.066
and the fatigue stress concentration correction factor, kf is
Kf =
1
1
=
= 0.48
kf r
2.066
EXAMPLE 7.5
Rotating shaft in bending shown in Figure 7.18 is loaded by the stationary
force, F and supported by ball bearings at A and D. The shaft is machined,
has ultimate strength of 1800 MPa, and yield strength of 1200 MPa. If the
applied force is high enough to create a shaft bending, estimete the fatigue
life of the shaft for the desired reliability of 99%. All the fillets are 4 mm.
SOLUTION
P Investigate the critical section which will fail first. Summation of the forces,
Fy = 0, in the vertical direction gives
R1 + R2 = 8 kN
P
Summation of the bending moments,
M = 0 at point A is (consider counterclockwise is positive), yields
X
M = 8 300 + R2 520 = 0
From force and moment equations, reaction forces, R1 =3.4 kN and R2 = 4.6 kN
can be found. To investigate the critical section which will fail first, calculate
17
MC
MD
Since point B and D has the same fillet radius, section at point B is more critical
than the section at point D. Now, compare the stresses at points B and C sections
to find out the most critical point on the shaft.
MB
I/c
MC
C =
I/c
B =
=
=
32MB
32 0.68 103
=
= 108.3 MPa
d3
(0.04)3
32MC
32 1.020 103
=
= 93.6 MPa
d3
(0.04)3
Since the bending stress at the section B is larger than the bending stress at
section C, B has the critical section. To have a conservative result, we introduce stress concentration factor, Kt at the critical point B. From Figure D1b (in Appendix D), geometric stress concentration factor, Kt (for the values
D/d=48/40 = 1.2 and r/d = 4/40 = 0.1), is 1.3. Then the stress at point B will
be B =108.31.3=140.8 MPa.
Now, stress, B = 140.8 MPa should be compared with the endurance strength
for the life determination. Endurance strength, Se0 is estimated as
Se0 = 0.5Sut = 0.35 1800 = 630 MPa
and
Se = Se0 ka ks kt kr kf km
From Figure 7.14, for Sut =1800 MPa and for machined surface finish, we have
ka =0.51. Size factor, ks is
ks = (
40 0.1133
d 0.1133
)
=(
)
= 0.83
7.62
7.62
18
1
1
=
kf r
1 + q(Kt 1)
From Figure 7.16, for r=4 mm and Sut =1800 MPa (261 Ksi), notch sensitivity,
q is 0.95. Correction factor for reliability of 99% is kr =0.814 (see Table 7.2).
Then,
kf =
1
1
=
= 0.775
kf r
1 + 0.95(1.3 1)
Since other factors were not mentioned, kt = km =0. Therefore, fatigue endurance strength is
Se = 630 0.51 0.83 0.775 0.814 = 168.2 MPa
Stress at critical sectiont B, B =140.8 MPa is less than the fatigue strength,
Se = 168.2 MPa and less than the yield strength, Sy . This means that life is
infinite and no yielding will occur. Note that, although we have considerably
high value of safety factor aginst yielding (n= 1200/140.8 =8.5), safety factor
for infinite fatigure life is very low (n=168.2/140.8=1.2).
EXAMPLE 7.6
Overhung centrifugal compressor shown in Figure 7.19(a) is used in sulfuric
acid plants. Size a shaft to support maximum cyclic load shown in Figure
7.19(b). The shaft material is UNS steel with machined finished. Desired
reliability is 95% and no less than 3 mm fillet is allowable. Assume that
torsional stress is negligible. Shaft material properties are: Sy =400 MPa
and Sut =600 MPa.
SOLUTION
Initially, we size the shaft diameter on the base of static loading and then
perform the dynamic analysis to ensure that the size of the shaft is safe for the
fatigue loading. Assuming that the safety factor, n is equal to 2, permissible
stress, p is given by
Sy
n
400
= 200 MPa
2
Assume that the cyclic load, F =1,200 N is static and constant in direction, then
the bending stress is
Figure 7.19:
Overhung centrifugal
compressor.
M
32(F L)
=
I/c
d3
32(1200 350)
d3
19
b
32(1200 350)
d3
200
32(1200 350)
200
d 27.77 mm
3
d =
To be on the safe side, assume that the shaft diameter is 30 mm and the fillet
is 3 mm. Corrected endurance limit, Se is
Se
Se0 (ka ks kr kf )
where Se0 =0.5Sut =0.5 600 = 300 MPa. From Figure 7.14, surface correction
factor, ka =0.78. Size factor, ks for rotating shaft is
ks
d
0.76
0.1133
=
30
0.76
0.1133
= 0.66
kf
1
1
=
kf r
1 + q(Kt 1)
From Figure D-1b (in Appendix D), geometric stress concentration factor, Kt
(for the values D/d=36/30 = 1.2 and r/d = 3/30 = 0.1), is 1.64. From Figure
7.16, for r=3 mm and Sut =600 MPa (87.1 Ksi), notch sensitivity, q is 0.83.
Correction factor for reliability of 95% is 0.868 (see Table 7.2). Then,
kf
1
= 0.65
1 + 0.83(1.64 1)
and
Se
32(F L)
Kt
d3
32(1200 350)
1.64 = 260 MPa
303
Since b > Se , shaft will have finite life. For safe and infinite life, shaft
diameter should be increased. Iteration results for different shaft diameters are
shown in Table 7.3. As seen from Table 7.3, shaft diameter, D=56 mm gives
good safety factor.
20
d (mm)
30
40
50
56
ka
0.78
0.78
0.78
0.78
ks
0.66
0.64
0.62
0.61
kr
0.814
0.814
0.814
0.814
kf
0.65
0.64
0.60
0.59
Se (MPa)
87
76.8
70.9
68.6
b (MPa)
260
113.7
61.6
24.4
ns
0.33
0.68
1.15
2.8
Figure 7.20: (a) Fully-reversed cyclic load, (b) Repeated cyclic load, (c) Fluctuating cyclic load.
7.4
Fluctuating Stresses
7.4.1
Ductile Materials
Fluctuating normal stress combined with the steady state mean stress, m is
shown in Figure 7.21. Soderberg diagram shown in Figure 7.22, is used for
fatigue analysis of ductile materials. Any design on the line of the soderberg
diagram will be safe, with a safety factor of n=1. As seen from Figure 7.22, only
alternating stress component, a is affected by the fatigue strength reduction
factor, kf r , not the steady stress component, m . From similar trangles, OCD
and GED shown in Figure 7.22, we have the following relationship:
21
Figure 7.22: Soderberg diagram used for ductile material fatigue analysis.
CO
OD
EG
GD
(7.22)
where
Se
n
Sy
OD =
n
EG = kf r a
Sy
GD =
m
n
CO
(7.23)
Substituting Eq.7.23 into Eq.7.22 gives the safety factor for Soderberg criterion
in fluctuating normal stress:
nductile =
Sy
m + (Sy /Se )kf r a
(7.24)
If the loading is only fully reversed cyclic (m =0), the safety factor for ductile
material reduces to
nductile =
Se
kf r a
(7.25)
22
nbrittle =
Sse
kf rs a
(7.26)
where Sse is the endurance limit in shear, a is the shear stress amplitude, and
kf rs is the fatigue strength reduction factor in shear.
In Eq.7.24, when Kf r = 1, we obtain the Soderberg equation:
m
1
a
+
=
Se
Sy
nductile
(7.27)
If we assume that the design is on the Soderberg design line, further simplification can be made to determine the equivalent bending stress amplitude,
eq , which takes into account the combination of stress amplitude, a , and mean
stress, m .
eq = a (
Sy
)
Sy m
(7.28)
EG
GD
(7.29)
where
Se
n
Sut
OD =
n
EG = Kt a
Sut
GD =
Kt m
n
CO
(7.30)
23
Figure 7.24: Modified Goodman diagram used for brittle material fatigue
analysis.
Substituting Eq.7.30 into Eq.7.29 gives the safety factor for Goodman criterion
in fluctuating normal stress:
nbrittle =
Sut
Kt [m + (Sut /Se )e ]
(7.31)
If the loading is only fully-reversed cyclic, in other words, if the steady stress
component, m , in Eq.7.31 is equal to zero, then the safety factor in fluctuating
normal stress reduces to
nbrittle =
Se
Kt a
(7.32)
Sse
Kts a
(7.33)
where a is the shear stress amplitude and Kts is the stress concentration factor
in shear. In Eq.7.31, when Kt = 1, we obtain the modified Goodman equation:
a
m
1
+
=
Se
Sut
nbrittle
(7.34)
As in the case of ductile material, equation for Goodman design line can also
found by assuming safety factor is equal to 1.
eq = a (
Sut
)
Sut m
(7.35)
Eq.7.35 can only be used when the mean stress, m is due to tensile loading.
If min and max are known, a , m and r (stress range) can be determined
by
24
max min
2
max + min
2
2a
(7.36)
Gerber Criterion
Three design lines which take into account the effect of the tensile normal mean
stress in fatigue analysis are shown in Figure 7.25. Gerber diagram was the first
limit stress diagram proposed by Gerber in 1874 on the base of ultimate strength.
The use of Gerber failure rule in fatigue analysis is not common because of its
non-linear nature. As shown in Figure 7.25, Gerber design line is parabolic
whereas the other two design lines are linear.
In 1899, Goodman proposed a theoretical design line based on the ultimate
strength representing the available data to consider the effect of the tensile normal mean stress in fatigue analysis. Because of its simplicity, the modified
Goodman criterion is much more popular. As seen from Figure 7.25, Goodman
design lines is more conservative than the Gerber line. The Soderberg design line
is based on the yield strength and very conservative. Gerber parabolic relation
is given by
na
nm 2
+(
) =1
Se
Sut
(7.37)
When n=1 in Eq.7.37, we obtain equivalent bending stress for Gerber equation
eq =
a
m 2
1 ( Sut
)
(7.38)
25
EXAMPLE 7.7
A round rotating shaft in bending shown in Figure 7.26 is subjected to
a fluctuating load shown in Figure 7.27. The shaft material is steel fine
grounded and has the elongation more than 5%. If the shaft material has
ultimate strength of 600 MPa and yield strength of 400 MPa, what is the
safety factor against fatigue failure for the desired reliability of 99%.
Figure 7.27:
Fructure surface.
SOLUTION
Estimated fatugue strength is
Se0 = 0.5Sut = 0.5 600 = 300 MPa
From Figure 7.14, surface correction factor, ka =0.9. Size correction factor, ks is
ks = (
d 0.1133
48 0.1133
)
=(
)
= 0.81
7.62
7.62
Reliability correction factor, kr = 0.814 (see Table 7.2 for 99% reliability). Assuming other correction factors has no effect, corrected endurance strength is
Se = Se0 (ka ks kr ) = 300(0.9 0.81 0.814) = 178 MPa
Note that, fatigue stress concentration factor, kf is not included in estimating
endurance limit, Se . Instead, fatigue reduction factor, kf r will be considered in
safety factor calculation. Safety factor for ductile material is
nductile =
Sy
m + (Sy /Se )kf r a
From Figure D-3a (in Appendix D), geometric stress concentration factor, Kt
(for the values D/d=48/36 = 1.33 and r/d = 6/36 = 0.17), is 1.4. From Figure
7.16, for r=6 mm and Sut =600 MPa (87.1 Ksi), notch sensitivity, q is 0.84.
Then, fatigue reduction factor, kf r is
kf r = 1 + q(Kt 1) = 1 + 0.84(1.4 1) = 1.34
Stress amplitude, a and mean stress, m are calculated by
100 (20)
max min
=
= 60 MPa
2
2
max + min
100 + (20)
=
= 40 MPa
2
2
26
nductile =
7.4.2
Sy
400
=
= 1.8
m + (Sy /Se )kf r a
40 + (400/178) 1.34 60
Combined fluctuating normal and shear stresses may occur in shafts, crankshafts,
aircraft wings and in many other mechanical rotating components due to axial,
bending, and torsional forces acting together. Maximum stresses on the components will occur when the fluctuating normal and shear stresses are in phase.
This phenomenon should be considered in the design analysis so that there will
be a phase difference between two force components.
Combined Stresses Based on Maximum Shear Stress and Maximum Distortion Energy Theories for Ductile Materials
As shown in Figure 7.2811 , normal stress has two components. Steady state
normal stress, m and fluctuating stress component, kf r a . Shear stress also
has two components. Steady state shear stress, m and the fluctuating shear
component of kf rs a . Applied forces are shown in Figure 7.29 on element A
for a thickness of t. Let,
Figure 7.29:
Resulting forces
on an element A.
m + kf rs a
m + kf rs a
m + kf r a
(7.39)
In Eq.7.39, for the wost case, shear and normal forces are taken to be in phase.
A summation of forces in the tangential direction for a unit thicness yields
(7.40)
dy
dy
dx
cos y
sin + x
cos = 0
d
d
d
(7.41)
or
+ x
Replacing dy/d = sin and dx/d = cos and substituting Eq.7.39 into Eq.7.41,
gives
11 Adapted
from Arthur H. Burr, 1981. Mechanical Analysis and Design.Elsevier, New York.
27
(7.42)
sin cos =
sin 2
cos 2 + 1
1 cos 2
; cos2 =
; sin2 =
2
2
2
kf r a
m
=(
sin 2 + m cos 2) + (
sin 2 + kf rs a cos 2)
2
{z
} | 2
{z
}
|
(7.43)
As can be seen from Eq. 7.43, shear stress, has two components steady state
component, m
and alternating component, a . Figure 7.30 shows Soderberg
design line for the maximum shear stress creterion. From similar triangles BDC
and AOC (see Figure 7.30), we have
Figure 7.30:
Sodeberd design line for
shear stresses based on
maximum shear stress
theory.
DC
OC
BD
OA
(7.44)
where
DC = (Sy /2)/n m
; OC = (Sy /2)/n; BD = a ; OA = (Se /2)/n
a
(Sy /2)/n m
=
(Sy /2)/n
(Se /2)/n
(7.45)
= m + a
n
Sy /2 Se /2
(7.46)
or
Substituting m
and a from Eq.7.43 into Eq.7.46, gives
m
1
kf r a
m
kf rs a
=(
+
) sin 2 + 2(
+
) cos 2
n
Sy
Se
Sy
Se
|
{z
} |
{z
}
A
(7.47)
(7.48)
Differentiation of Eq.7.48 and equating to zero will provide the minimum value
of the safety factor n.
Figure 7.31:
Triangle for trigonometric
relationship.
d 1
( ) = 2A cos 2 4B sin 2 = 0
d n
Eq.7.49 yields the following trogonometric relationships (see Figure 7.31)
(7.49)
28
tan 2 =
A
2B
A
; cos 2 =
; sin 2 =
2
2
2
2B
A + 4B
A + 4B 2
(7.50)
Substituting Eq.7.50 into Eq.7.48 and rearranging the terms gives the minimum
safety factor, n on the base of maximum shear stress criterion.
n= q
(m +
Sy
Sy
2
Se kf r a )
+ 4(m +
(7.51)
Sy
2
Se kf rs a )
Simply by changing the coefficient of the shear stress term from 4 to 3 in Eq.
7.51 we obtain the equation for maximum distortion energy theory for the ductile
materials.
n= q
(m +
Sy
Sy
2
Se kf r a )
+ 3(m +
(7.52)
Sy
2
Se kf rs a )
n=
Sut
q
Sut
1
1
Kt2 (m +
2 Kt (m + Se a ) + 2
(7.53)
Sut
2
Se a )
2 ( +
+ 4Kts
m
Sut
2
Se a )
Changing the coefficient of the shear stress term from 4 to 3 in Eq. 7.53 we obtain
the equation for maximum distortion energy theory for the brittle materials
n=
Sy ut
q
Sut
1
1
2 ( +
Kt2 (m + SSut
a )2 + 3Kts
m
2 Kt (m + Se a ) + 2
e
(7.54)
Sut
2
Se a )
29
EXAMPLE 7.8
Figure 7.32 shows that torque transmission couplings for transmitting rotary power from a motor to a rotating machine. Since shaft is installed in
misaligned bearing, as the shaft rotates the shaft outer fibers go through a
tension-compression cycles (see Figure 7.33). Also, shaft misalignment (angular, parallel, or combination of two) through coupling creates an initial
system preload and consequently creating mean stress, as shown in Figure
7.33. Moreover, assume that shaft misalignment provides steady torsional
stress of m =1.5 Mpa from the driving shaft to driven shaft. Consider the
time dependent torque is negligible.
Figure 7.32:
Rotor system.
The shaft material is machined, has the ultimate strength of 1000 MPa, and
yield strength of 600 MPa with more than 5% elongation. Determine the
safety factor of the shaft for a 99% of reliability.
SOLUTION
Estimated fatigue strength is
Se0 = 0.5Sut = 0.5 1000 = 500 MPa
Figure 7.33:
Fluctuating load.
From Figure 7.14, surface correction factor, ka =0.71. Size correction factor, ks
is
ks = (
d 0.1133
40 0.1133
)
=(
)
= 0.83
7.62
7.62
Reliability correction factor, kr = 0.814 (see Table 7.2 for 99% reliability). Assuming other correction factors has no effect, corrected endurance strength is
Se = Se0 (ka ks kr ) = 500(0.71 0.83 0.814) = 240 MPa
Safety factor for ductile material is
n= q
Sy
(m +
Sy
2
S e kf r a )
+ 3(m +
Sy
2
Se kf rs a )
n= q
(m +
Sy
Sy
2
Se kf r a )
+ 3(m )2
30
max min
80 (40)
=
= 60 MPa
2
2
max + min
80 + (40)
=
= 20 MPa
2
2
Substituting all the calculated and known parameters into the safety equation,
gives
n= q
7.5
600
= 2.27
2
2
[20 + ( 600
240 )(1.63)(60)] + 3(1.5)
Miners rule first proposed by Palmgren in 1924 and was further developed by
Miner in 1945. It the simplest and most widely used cumulative damage models
for fatigue life estimation under variable amplitude loading. Miners rule states
that if there are k different stress level of amplitudes, a , then the damage
fraction, D, is:
k
X
ni
N
fi
i=1
(7.55)
n2
nk
n1
+
+ ... +
Nf 1
Nf 2
Nf k
(7.56)
D=
or
D=
EXAMPLE 7.9
Assume that the cantilever beam shown in Figure 7.34 has the material
properties of: Sut =80 kpsi and Sy =60 kpsi. Suppose the loads P1 and P2
(see Figure 7.34) on the beam produce fluctuating stress of 1 for a total
of n1 =5,000 cycles and 2 for a total of n2 =60,000 cycles. Determine the
damage and the remaining life of the beam.
SOLUTION
Bending stress due to bending moment, M = P L is
31
M
I/c
where
I
D3
(4)3
=
=
= 6.28 in3
c
32
32
1max =
9 40
= 57.33 kpsi
6.28
1min =
2 40
= 12.74 kpsi
6.28
1a
1max 1min
57.33 (12.74)
=
= 35.035 kpsi
2
2
1m
1max + 1min
57.33 + (12.74)
=
= 22.295 kpsi
2
2
1eq = 1a
Sy
60
= 35.035[
] = 55.75 kpsi
(Sy 1m )
(60 22.295)
Equivalent stress, 1eq = 55.75 kpsi includes the effect of the mean stress, 1m
for the damage calculation. Using similar calculations, following results are
obtained for the flactuating load, P2 :
2max =
4 40
= 25.48 kpsi
6.28
2min =
1 40
= 6.37 kpsi
6.28
and
2a
2m
Equivalent stress is
2max 2min
25.48 (6.37)
=
= 15.925 kpsi
2
2
2max + 2min
25.48 + (6.37)
=
= 9.555 kpsi
2
2
32
2eq
Sy
60
= 2a
= 15.925
= 18.94 kpsi
(Sy 2m )
(60 9.555)
Remaining life, RL is
RL = 100 D = 100 14.7 = 85.3%
7.5.1
As shown in Figure 7.36, riser is a long tensioned cylindrical hollow steel tube
that connects an offshore floating production structure or a drilling vessel to a
sub-sea system for drilling or production purposes. Riser is considered to be the
most critical component in an offshore pipeline development taking into account
the dynamic wave and current loads and hostile service conditions it has to
withstand. Riser ball joints are used in the upper and lower portion of the riser
to provide flexibility. When riser used in water depths greater than about 20
meters, it has to be tensioned to maintain stability. Riser not only protects the
drill pipe from wave and current forces but also provide a path for the return
of the drilling fluid back to the drilling vessel. It is evident that the drill pipe
undergoes a relatively sharp bend due to the deflection of the ball joints caused
by wave and current loads on the riser string. The bend at the ball joints can
cause fluctuating stresses and consequently results in fatigue damage to the drill
pipe that is inside the riser. In general, to be safe, in drilling from drilling vessels,
an allowable ball joint angle should not be more than approximately 5 degrees.
33
EXAMPLE 7.10 12
Develop a mathematical modelling to drive the bending moment equation
when a riser is bend and tensioned by a tensile force through riser tensioner
shown in Figure 7.36.
SOLUTION
An element of the deformed drill pipe experiences a bending moment and shear
forces while passing through the riser ball joint. To drive the bending moment
and shear force equations, consider the free-body diagram shown in Figure 7.37.
In this figure, is the ball joint angle which creates the bending moment in drill
pipe. For equilibrium of the element shown in Figure 7.37, summation of the
forces in the y direction should be equal to zero
X
Fy = 0
Q + (Q + dQ) = 0 dQ = 0 Q = constant
(7.57)
(7.58)
MA = 0
(CCW +)
(7.59)
12 Adapted from A. Ertas et al., The Effect of Tool Joint Stiffness on Drill Pipe Fatigue in
Riser Ball Joints, Transactions, ASME, Journal of Engineering for Industry, vol. 111, no. 4,
pp. 369374, 1989.
34
M + dM M + Qdx T dy = 0
(7.60)
dM
dy
T
= Q
dx
dx
(7.61)
Divide Eq.7.60 by dx
Taking the derivative of Eq.7.61 with respect to x yields (Note that from Eq.
the shear force Q is constant)
d2 M
d2 y
T
=0
dx2
dx2
(7.62)
M = EI
d2 y
dx2
(7.63)
EI
d4 y
d2 y
T 2 =0
4
dx
dx
(7.64)
or
y 0000
T 00
y =0
EI
(7.65)
k2 =
T
EI
(7.66)
(7.67)
y = C1 + C2 x + C3 cosh(kx) + C4 sinh(kx)
(7.68)
d y
dy
0
(b) x , M 0, dx
2 =0, y = dx
Applying boundary conditions for x=0, y=0, Eq.7.68 yields
C1 + C3 = 0 C3 = C1
Substituting Eq.7.69 into Eq.7.68 gives
(7.69)
35
y = C1 + C2 x + C1 cosh(kx) + C4 sinh(kx)
(7.70)
y0 =
dy
= C1 k sinh(kx) + C2 + C4 k cosh(kx)
dx
(7.71)
C2 + C4 k = 0 C4 =
C2
k
(7.72)
y = C1 [1 cosh(kx)] +
C2
[kx sinh(kx)]
k
(7.73)
y 00 =
d2 y
= k [C1 k cosh(kx) C2 sinh(kx)]
dx2
(7.74)
M = EI
d2 y
dx2
(7.75)
d2 y
dx2 =0,
yields
(7.76)
(7.77)
C1 k = C2 tanh(kx)
(7.78)
or
C1 =
Substituting Eq.7.79 into Eq.7.73
C2
k
(7.79)
36
y=
C2
C2
[1 cosh(kx)] +
[kx sinh(kx)]
k
k
(7.80)
dy
= C2 {[sinh(kx) cosh(kx)] + 1}
dx
(7.81)
1
dy
+
1
= C2
dx
1/ekx
(7.82)
or
y0 =
= C2
(7.83)
(7.84)
dy
= C2 {[sinh(kx) + (1 cosh(kx))]}
dx
(7.85)
d2 y
= k{[cosh(kx)] [sinh(kx)]}
dx2
(7.86)
M = EI
d2 y
dx2
(7.87)
M (x) =
T
{[cosh(kx)] [sinh(kx)]}
k
(7.88)
T
EI
(7.89)
37
EXAMPLE 7.11 13
By using S-N curve approach (see Figure 7.39), find the cumulative fatigue
damage inflicted on the drill pipe during one pass through riser ball joint.
One pass is defined as vertical movement of the drill pipe through 144 inches
while being rotated. Assume that the drill pipe is subjected to a constant
axial tension load of 100,000 lbf to prevent failure by buckling and has the
material with elongation less than 5%.
SOLUTION
During the drilling operation, the drill pipe is subjected to alternate bending
stresses (compression and tension) due to rotation. This bending stress is maximum at the ball joint and falloffs exponentially with distance from the ball
joint, both below and above. As a result, when the drill pipe makes one pass
through, the upper ball joint, as each joint of the drill pipe passes through the
ball joint, experiences cumulative fatigue damage from stress cycles of widely
varying amplitudes.
Calculation of Constants Used in Computation
Given Design Data
Modulus of elasticity of drill-pipe material, E
Yields strength of drill-pipe material, ys
Penetration speed of drill pipe, V
Outside diameter of drill pipe, Do
Inside diameter of drill pipe, Di
Rotational speed of drill pipe, n
Tension load, T
=
=
=
=
=
=
=
30 106 psi
90 kpsi
120 in./h
5.00 in.
4.276 in.
100 rpm
100 kips
(Do4 Di4 )
(54 4.2764 )
=
= 14.27in.4
64
64
13 Adapted from A. Ertas et al., The Effect of Tool Joint Stiffness on Drill Pipe Fatigue in
Riser Ball Joints, Transactions, ASME, Journal of Engineering for Industry, vol. 111, no. 4,
pp. 369374, 1989.
38
A=
T
100
=
= 18.96 kpsi
A
5.275
Since drill pipe material is brittle (material elongation less than 5%), we use
modified Goodman equation
Sut
eq = a
Sut m
{z
}
|
Fcr
The correction factor Fcr , which takes into account the mean stress, is
Fcr =
Sut
100
= 1.234
=
Sut m
100 18.96
The following steps are performed to determine the cumulative fatigue damage inflicted on the drill pipe during one pass through the upper ball joint:
Step 1 Divide a joint of the drill pipe into a number of finite elements.
As shown in Figure 7.40, smaller elements are taken close to the higher stress
region. As we will learn later, the fatigue damage calculation will show that the
bending moment beyond a distance of 72 in. of the ball joint has no significant
contribution to the life of a drill pipe.
Step 2 Calculate the bending moment using Eq.7.88. For = 1 (0.01744
rad) the bending moment M at x = 0 is
Figure 7.40:
Schematic of a
drill pipe in a
ball joint.
M (x = 0)
M (x = 0)
= (100,000) 1
180
sinh(0.01528)(0)]
1
0.01528
[cosh(0.01528)(0)
114.20 + 104.22
= 109.21 kips in.
2
Mave (EL.1) Do
109.21 5
=
= 19.13 kpsi
2I
2 14.27
Step 4 Determine the equivalent stress, eq , by the use of the equation representing the modified Goodman line to account for the mean stress due to nominal
tension.
39
x
(in.)
x =0
x =6
M (x)
(kips-in.)
114.20
104.22
Mave
(kips-in.)
109.22
av
(kpsi)
19.13
eq
(kpsi)
23.61
n
(cycle)
300
N
(cycle)
5 105
(n/N) 100
%
0.06
x=6
x =18
104.22
86.76
95.49
16.73
20.64
600
inf.
0.0
x =18
x =36
86.76
65.90
76.33
13.37
16.50
900
inf.
0.0
x =36
x =72
65.90
38.00
51.95
9.10
11.23
1800
inf.
0.0
rpm 60 L1
100 60 (6 in. 0 in.)
=
= 300 cycles
V
120
Step 6 Find the number of cycles N1 for failure at the stress level of 23.61
kpsi for the first element from the SN curve shown in Figure 7.39. Note that, if
the stress level is higher than 53 kpsi (see S-N curve given in Figure 7.39), total
life, N , should be assumed to be 104 cycles.
N1 = 500, 000 cycles
Step 7 Apply Miners rule to evaluate the cumulative fatigue damage for the
first element.
n1
300
100 =
100 = 0.06%
N1
500, 000
This procedure is repeated for the remaining elements, and the results are
summarized in Table7.4. Note that Figure 7.39 shows that the life is infinite at
stresses below 20 kpsi. Therefore, stresses below 20 kpsi, the damage is assumed
to be zero. The total damage to the drill pipe during its passage through the
location beginning 72 in. above and ending 72 in. below (one pass) the upper
joint can be calculated as follows:
n2
n3
n4
n1
+
+
+
100
2
N1
N2
N3
N4
2 (0.06 + 0 + 0 + 0) = 0.12%
Total damage =
=
40
7.6
41
42
Figure 7.43: Aloha Airlines Flight 243 extensive damage after an explosive
decompression in flight.3
and phosphorus are also factors affecting ductility negatively. Materials with
high carbon content may not be suitable for welding.
Corrosive environment High stresses combined with a corrosive environment
can cause critical components to crack and fail, sometimes with little warning.
Aloha Airlines Flight 243 While flying from Hilo to Honolulu suffered extensive
damage after an explosive decompression in flight (see Figure 7.43), but was able
to land safely at Kahului Airport. One flight attendant was blown out of the
airplane and another 65 passengers and crew were injured. This was a significant event in the history of aviation, with far-reaching effects on aviation safety
policies and procedures. Investigation showed that the accident was caused by
metal fatigue exacerbated by crevice corrosion. The plane was 19 years old and
operated in a coastal environment, with exposure to salt and humidity.16
To reduce the risk of catastrophic failure by fracture mentioned in this section, designers and analysts should become familiar with fracture control requirements. Fracture criticality of structural parts and components must be
documented and required analyses, inspections and other fracture control activities need to be implemented and monitored to ensure that the remaining life of
the components would not create failure risk.
7.6.1
43
at a distance r from the crack tip and at an angle with respect to the crack
plane will have the following stress field17 :
xy
3
I cos (1 sin sin )
2
2
2
2r
K
3
I cos (1 + sin sin )
2
2
2
2r
KI
(7.90)
(7.91)
(7.92)
44
Figure 7.46: Crack types: (a) Central through thickness crack, (b) single-edge
through thickness crack, (c) double-edge through thickness crack, (d) three point
bend specimen.
KI = a
(7.93)
KI = af (a/W )
(7.94)
Functions, f (a/W ) to correct the stress intensity factor for different crack configurations are given below 18 .
(a) Central through thickness crack (accuracy better than 0.5% for (a/W)
0.5):
f (a/W )
a
a 2
a 3
1 + 0.128
0.288
+ 1.525
W
W
W
(7.95)
(b) Single-edge through thickness crack (accuracy is 0.5% for (a/W) 0.6):
a
a 2
a 3
a 4
f (a/W ) = 1.122 0.231
+ 10.550
21.710
+ 30.382
W
W
W
W
(7.96)
18 H. Tada, P.C. Paris, and G. R. Irwin, 2000. The stress analysis of cracks handbook. The
Material Information Society.
45
(c) Double-edge through thickness crack (accuracy better than 0.5% for (a/W)
> 0.4):
a 3
a
a 2
+ 1.930
f (a/W ) = 1.12 + 0.203
1.197
(7.97)
W
W
W
(d) Single-edge through thickness crack in pure bending (accuracy is 0.2% for
(a/W) 0.6):
a
a 2
a 3
a 4
f (a/W ) = 1.122 1.40
+ 7.33
13.08
+ 14.0
W
W
W
W
(7.98)
(e) Three point bend specimen (accuracy is 0.5% for any (a/W):
=
6M
b2
where M =
PS
4
(7.99)
For S/b=4
1
1.99 a/W (1 a/W )[2.15 3.93a/W + 2.7(a/W 2 ]
f (a/W ) =
(7.100)
46
EXAMPLE 7.12
Assume that a=6 mm through thickness crack is subjected to a tensile stress
as shown in Figure
7.47. The steel plate has a critical stress intensity factor,
KIC = 25 MPa m. Determine the maximum tensile stress for failure if:
(a) the through thickness crack is at the center of the steel plate with
finite width (W=80 mm) and length (h=100 mm), and
(b) the through thickness crack is in a long and wide steel plate.
SOLUTION
(a) Since the crack is at the center of a plate with finite width and length,
functions, f (a/W ) to correct the stress intensity factor is
f (a/W )
=
=
=
a 2
a 3
a
0.288
+ 1.525
1 + 0.128
W
W
W
"
2
3 #
6
6
6
1 + 0.128
0.288
+ 1.525
80
80
80
1.009
Eq.7.94 can be modified to calculate the maximum tensile stress for failure as
Figure 7.47:
Through-thickness crack
in a plate.
max
KIC
af (a/W )
25 103
= 180.5 MPa
6 1.009
=
=
(b) Since the steel plate is is long and wide, we assume that a through-thickness
crack is at the center of the plate. Then, the infinite plate formula as in Eq.7.93
will be modified to calculate the maximum tensile stress for failure.
max
=
=
K
IC
a
25 103
= 182.13 MPa
6
47
Figure 7.48:
Elliptical crack in an
infinite solid.
a
a2
KI =
(sin2 + 2 cos2 )1/4
(7.101)
c
where is the elliptical integral and is defined as
Z
/2
q
( 1 k 2 sin2 )d
(7.102)
where
a
k 2 = 1 ( )2
c
The elliptic integral, has the following series expansions
=
1 c2 a2
3 c2 a2 2
1
...
2
2
4
c
64
c2
(7.103)
(7.104)
Figure 7.49:
Elliptical surface
crack.
KI = 1.1
a2
a (sin2 + 2 cos2 )1/4
(7.106)
KI2
4 2Sy2
(7.107)
where
a = a +
Including the effect of the crack shape around the crack front, Eq.7.106 can
be further modified for the maximum value of the stress intensity factor at the
minor axis ( = /2 ) as follows:
r
a
KI = 1.1
(7.108)
Q
Where Q is the crack-shape parameter given as
Q = 2 0.212
2
Sy
(7.109)
Finally, a magnification factor for deep cracks, such as those shown in Figure
7.50, can be used to find the maximum stress intensity factor for a semielliptical
surface flow as
20 G. R. Irwin, The Crack Extension Force for a Part Through Crack in a Plate, Transactions, ASME, Journal of Applied Mechanics, vol. 29, no. 4, pp. 651654, 1962.
48
r
KI = 1.1Mk
a
Q
(7.110)
EXAMPLE 7.13
Hydrogen is a renewable energy source alternative to petroleum-based fuels.
Because of the low energy density, hydrogen is compressed to a pressure of
20 MPa in a pressure vessel. Dimensions of the pressure vessel is is shown in
Figure 7.51. Assume that the hoop stress created by the internal pressure is
acting perpendicular to the semielliptical surface crack depth a (see Figure
7.52). Initial flow depth ai = 2 mm and an a/2c=0.25 ratio remains constant.
If the yield strength of the materail is 1000 MPa, calculate the stress intensity
factor corresponding to the hoop stress created by the internal pressure.
SOLUTION
From the thin-walled pressure vessel theory, stress created by the hoop stress
perpendicular to the semielliptical crack depth is
Figure 7.51:
Hoop stress created
by internal pressure.
hoop
=
=
pD
2t
20 0.3
= 500 MPa
2 6 103
a
Q
49
Q =
2
Sy
where
=
=
3 a 2
+
8
8 c
2
3
+
0.5 = 1.276
8
8
(1.276)2 0.212
500 2
= 1.575
1000
From Figure 7.50, since a/t is less than 0.5, Mk is equal to 1. Then the stress
intensity factor is
r
KI = 1.1 1 500
2 103
= 34.73 MPa
1.575
KIC
1.1Mk max
2
(7.111)
R. Irwin, Materials for Missiles and Spacecraft, McGraw-Hill, New York, 1963.
50
EXAMPLE 7.14
In Example 13, if the critical stress intensity factor of the pressure vessel
material is 45 MPa, will the pressure vessel leak?
SOLUTION
From Eq.7.110 critical crack size, acr is
acr
=
=
KIC
1.1Mk max
2
2
45
1.575
= 3.356 103 m
1.1 1 500
Since the thickness t=6 mm is larger than the critical crack length, acr =3.356
mm, pressure vessel will not leak.
(7.112)
where
a =
crack depth
number of cycles
KI
A and m are the constants for a particular material, environment, and loading
condition.
23 Rolfe, S. T. and Barsom, J. M., 1977. Fracture and fatigue control in structures: application of fracture mechanics. Prentice-Hall, Engle wood Cliffs, NJ.
51
2. Calculate the critical crack size acr , that would cause catastrophic failure.
3. Assume an increment of crack growth a.
4. Determine KI , using the proper equation for KI , with and aav ,
where
aav = ai +
a
2
(7.113)
5. Determine the number of cycles at a given stress level by using Eq. 7.111.
Direct integration should continue until a reaches acr . If acr is larger than
the thickness of the material, integration should stop when a is equal to
the material thickness.
EXAMPLE 7.1524
Using the theory of fracture mechanics, find the cumulative fatigue damage
inflicted on the drill pipe of Example 7-11 during one pass through the riser
upper ball joint.
=
=
=
=
=
=
=
=
=
=
30 106 psi
56 kpsi in.
90 kpsi
0.5
120 in./h
5.00 in.
4.276 in.
0.0312 in.
100 rpm
100 kips
52
SOLUTION
The drill pipe is assumed to have a semielliptical (part-through) surface crack,
as shown in Figure 7.54. This type of crack is typical of the starting-crack
shape customarily found. As shown, the drill pipe is assumed to be cut along a
longitudinal plane and opened into a plate. On the basis of quality inspection,
the estimated initial flaw size ai present in the drill pipe is assumed to
be 0.0312
in. The critical stressintensity factor KIC is assumed to be 56 kpsi in.
The sinusoidal fluctuating stress-time history for a drill pipe is shown in
Figure 7.56. From this figure, the stress range of the drill pipe is observed to be
r = = max min
where
max = m + ave
= r = 2ave
53
av
(kpsi)
19.13
16.73
13.37
9.10
(kpsi)
38.26
33.46
26.74
18.2
max
(kpsi)
38.09
35.69
32.33
28.06
n
(cycles)
300
600
900
1800
N
(cycles)
400,281
611,994
1.24 106
4.18 106
n/N
%
0.075
0.098
0
0
Table7.6 is obtained by using m = 18.96 kpsi and the values of av from Table
7.6 (see Example 7.10). The following steps are performed to determine the
cumulative fatigue damage.
Step 1 Divide a joint of the drill pipe into a number of finite elements as
shown in Figure 7.40.
Step 2 Calculate the average stress amplitude av for each of these finite
elements (see Example 7.11).
Step 3 Calculate the critical flaw size acr that would cause failure by brittle
fracture for the first element as follows:
As discussed previously, for catastrophic failure, the initial crack depth ai
must reach the critical crack depth calculated using
2
Q
KIC
acr =
1.1Mk max
3 a2
3
+
=
+ (0.5)2 = 1.276
2
8
8c
8
8
Q = (1.276) 0.212
38.09
90
2
= 1.59
Hence,
acr =
56
1.1 38.09
2
1.59
= 0.905 in.
Since the critical crack size is larger than the wall thickness, the drill pipe will
fail by leaking and not by catastrophic propagation to failure.
Step 4 Assume an increment of crack growth a and determine the number
of failure cycles for the crack to grow by this increment at the given stress level
for each finite element by using Eq. 7.111 with A = 0.614 1010 and m = 3.16
(specific to drill pipe material grade E).
da
3.16
= 0.614 1010 (KI )
dN
54
aav
Q
N =
1.976 109 a
q
3.16
ai +0.5a
Q
For the second iteration, initial crack size will be ai = 0.0312 + 0.01 = 0.0412
in. Hence, the number of cycles for the next a = 0.01 increment is
N =
Since the critical crack size was found to be larger than the wall thickness of the
drill pipe, iteration for incremental crack growth should stop when it reaches a
final crack size af , which is equal to the wall thickness of 0.362 in.
X
N (EL.1) =
N = 77, 600 + 52, 800 + = 400, 281 cycles
The damage for the first element is
300
n1
=
100 = 0.075%
N1
400, 281
The same procedure is repeated for the remaining elements, and the summary
of the results is given in Table ??. Note that, since the stress is changing, Q and
acr should be recalculated for each element.
The total damage to the drill pipe during one pass through the upper joint
is
Total damage
=
=
n1
n2
n3
n4
+
+
+
100
N1
N2
N3
N4
2 (0.075 + 0.098 + 0 + 0) = 0.346%
2
The results of Examples 7.10 and 7.14 show that the fracture mechanics approach
provides conservative results in determining the fatigue damage of drill pipe at
low-tension load such as T =100 kips and small riser angles such as =1 degree.
55
Machine components contain cracks may be subjected to more than one type of
loading. The principle of superposition can be applied for stress-intensity factors
for different types of loading of the same mode type. In other words, the stressintensity factors for different modes of crack growth con not be algebraically
added. Also, in summation process (superposition) the stress-intensity factors
must be associated with the same structural geometry, including crack geometry.
For example, stress-intensity factors associated with an edge crack problems
cannot be added to that of a semi elliptical crack problems.
EXAMPLE 7.16
Determine the stress-insensity factor of a plate subjected to a combined
tension and bending loadings shown in Figure 7.57.
SOLUTION
(a) Stress-intensity factor for tension load is
Figure 7.56:
Plate with through
thickness crack.
KIT = T af (a/W )
From Eq. 7.96, correction factor for a single edge through thickness crack is
f (a/W )
[1.122 0.231
+
30.382
4.8
40
4.8
40
+ 10.550
4.8
40
2
21.710
4.8
40
3
4
] = 1.215
KIT = 1.215 T a
where
P
P
=
A
BW
T =
KIM = M af (a/W )
From Eq.7.98, correction factor for a single edge through thickness crack for
pure bending is
f (a/W )
[1.122 1.40
14.0
4.8
40
4.8
40
+ 7.33
4
] = 1.04
4.8
40
2
13.08
4.8
40
3
56
KIM = 1.04 M a
where
M =
M
6M
=
I/c
B W2
The algebraic addition of the two above stress-intensity factors will give the
stress-intensity factor for combine loadings, KIC
KIC
P
6M
K T + K M = 1.215
a + 1.04
a
BW
B W2
a
6.24
1.215P +
=
BW
W
EXAMPLE 7.1725
Figure 7.58:
Blade configuration.
SOLUTION
From Eq.7.112, fatigue life is
a
A(KI )m
(a)
57
(b)
(c)
= 2a = range
(d)
Then
where
0.00072
A
acr =
2
2
1
KIC
1
32
=
= 0.28 m
max f (a/W )
30 1.13
Note that, in the above calculation for f (a/W ), (1.215 + 1.04)/2 = 1.13 is used.
Since acr is larger than the blade thickness, iteration will stop at the thickness
value of the fan blade (W=0.08 m). Average crack length, acr in Eq. (c) is
aav = ai +
a
2
58
aav = 0.001 +
0.079
= 0.0405 m
2
0.079
= 11, 190 cycles
9.5 1013 (25.25)4 .9