Upload
others
View
14
Download
1
Embed Size (px)
Citation preview
1
University of Waterloo
Department of Mechanical Engineering
ME 322 ME 322 -- Mechanical Design 1Mechanical Design 1
Partial notes – Part 4 (Fatigue)(G. Glinka)
Fall 2005
2
NNii = ?= ?
NNpp = ?= ?
NNTT = ?= ?
Metal FatigueMetal Fatigue is a process which causes premature irreversible damage or is a process which causes premature irreversible damage or failure of a component subjected to repeated loading.failure of a component subjected to repeated loading.
FATIGUE FATIGUE -- What is it?What is it?
3
Metallic FatigueMetallic Fatigue• A sequence of several, very complex phenomena encompassing several
disciplines:– motion of dislocations– surface phenomena– fracture mechanics– stress analysis– probability and statistics
• Begins as an consequence of reversed plastic deformation within a single crystallite but ultimately may cause the destruction of the entire component
• Influenced by a component’s environment• Takes many forms:
– fatigue at notches– rolling contact fatigue– fretting fatigue– corrosion fatigue– creep-fatigue
Fatigue is not cause of failure per se but leads to the final fracture event.
4
The Broad Field of Fracture MechanicsThe Broad Field of Fracture Mechanics
(from Ewalds & Wanhil, ref.3)
5
σσ
Intrusions and Extrusions:Intrusions and Extrusions:The Early Stages of Fatigue Crack Formation
6
Locally, the crack grows in shear;
macroscopically it grows in tension.
c
Schematic of Fatigue Crack Initiation Subsequent Gr owth Schematic of Fatigue Crack Initiation Subsequent Gr owth Corresponding and Transition From Mode II to Mode ICorresponding and Transition From Mode II to Mode I
7
The Process of FatigueThe Process of Fatigue
The Materials Science Perspective:The Materials Science Perspective:
•• Cyclic slip,Cyclic slip,•• Fatigue crack initiation,Fatigue crack initiation,•• Stage I fatigue crack growth,Stage I fatigue crack growth,•• Stage II fatigue crack growth,Stage II fatigue crack growth,•• Brittle fracture or ductile ruptureBrittle fracture or ductile rupture
8
Features of the Fatigue Fracture Surface of a Typic al Features of the Fatigue Fracture Surface of a Typic al Ductile Metal Subjected to Variable Amplitude Cycli c Ductile Metal Subjected to Variable Amplitude Cycli c
LoadingLoading
(Collins, ref. 22 )
A – fatigue crack area
B – area of the final static failure
9
Appearance of Failure Surfaces Caused by Appearance of Failure Surfaces Caused by Various Modes of LoadingVarious Modes of Loading (SAE Handbook)
10
Factors Influencing Fatigue LifeFactors Influencing Fatigue Life
Applied Stresses• Stress range – The basic cause of plastic deformation and consequently the
accumulation of damage• Mean stress – Tensile mean and residual stresses aid to the formation and
growth of fatigue cracks• Stress gradients – Bending is a more favorable loading mode than axial
loading because in bending fatigue cracks propagate into the region of lower stresses
Materials• Tensile and yield strength – Higher strength materials resist plastic
deformation and hence have a higher fatigue strength at long lives. Most ductile materials perform better at short lives
• Quality of material – Metallurgical defects such as inclusions, seams, internal tears, and segregated elements can initiate fatigue cracks
• Temperature – Temperature usually changes the yield and tensile strength resulting in the change of fatigue resistance (high temperature decreases fatigue resistance)
• Frequency (rate of straining) – At high frequencies, the metal component may be self-heated.
11
ComponentGeometry
LoadingHistory
Stress-StrainAnalysis
Damage Analysis
Allowable Load - Fatigue Life
MaterialProperties
Information path in strength and fatigue life prediction procedures
StrengthStrength --Fatigue Analysis ProcedureFatigue Analysis Procedure
12
Stress Parameters Used in Static Strength and Stress Parameters Used in Static Strength and Fatigue AnalysesFatigue Analyses
peak
tn
peak
K
S
σσ
σ
=
=
σσσσn
y
dn
0
T
ρ
σpeak
Str
ess
M
b)a)
σσσσn
y
dn
0
T
ρ
σpeak
Str
ess
S
S
13
Constant and Variable Amplitude Stress Histories;Constant and Variable Amplitude Stress Histories;Definition of a Stress Cycle & Stress Reversal
One cycle
����m
����max
����min
Str
ess
Time0
Constant amplitude stress historya)
Str
ess
Time
0
Variable amplitude stress history
One reversal
b)
max min
max mi
max min
mi
max
n
n
;
;
2 2
2
peak peak peak
peak peak
peak peakpeak
m
peak
pe
peakpe
ak
aka
R
σ σσ σσ σσ σσσσσ
σσσσ
σ σ σσ σ σσ σ σσ σ σ
σ σσ σσ σσ σσσσσσσσσ
σσσσ
∆ = −∆ = −∆ = −∆ = −
−−−−∆∆∆∆====
++++====
====
====
In the case of the peak stress history the important parameters are:
14
Str
ess
Time
Stress history Rainflow counted cycles
����i-1
����i-2
����i+1
����i
����i+2
Stress History and the “Rainflow” Counted Cycles Stress History and the “Rainflow” Counted Cycles
1 1i i i iABS ABSσ σ σ σ− +− < −
A rainflow counted cycle is identified when any two adjacent reversals in thee stress history satisfy the following relation:
15
A rainflow counted cycle is identified when any two adjacent reversals in thee stress history satisfy the following relation:
1 1i i i iABS ABSσ σ σ σ− +− < −The stress amplitude of such a cycle is:
1
2i i
a
ABS σ σσ − −
=
The stress range of such a cycle is:
1i iABSσ σ σ−∆ = −
The mean stress of such a cycle is:
1
2i i
m
σ σσ − +=
The Mathematics of the Cycle Rainflow Counting Meth od The Mathematics of the Cycle Rainflow Counting Meth od for Fatigue Analysis of Stress/Load Historiesfor Fatigue Analysis of Stress/Load Histories
16
17
Stress History
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14Str
ess
(MP
a)x1
02
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Reversal point No.
Str
ess
(MP
a)x1
02
18
Stress History
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14Str
ess
(MP
a)x1
02
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Reversal point No.
Str
ess
(MP
a)x1
02
19
Stress History
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14Str
ess
(MP
a)x1
02
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Reversal point No.
Str
ess
(MP
a)x1
02
20
Stress History
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14Str
ess
(MP
a)x1
02
-3
-2
-1
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Reversal point No.
Str
ess
(MP
a)x1
02
21
Number of Cycles Number of Cycles According to the According to the Rainflow Counting Rainflow Counting Procedure Procedure (N. Dowling, ref. 2)
22
The FatigueThe Fatigue SS--NN methodmethod(Nominal Stress Approach)
• The principles of the S-N approach (the nominal stress method)
• Fatigue damage accumulation
• Significance of geometry (notches) and stress analysis in fatigue evaluations of engineering structures
•
• Fatigue life prediction in the design process
23
AB
Note! In the case of smooth components such as the railway axle the nominal stress and the local peak stress are the same!
peakS σσσσ====Smin
Smax
WWööhler’shler’s Fatigue TestFatigue Test
24
Characteristic parameters of the S - N curve are:Se - fatigue limit corresponding to N = 1 or 2×106 cycles for
steels and N = 108 cycles for aluminum alloys,S10
3 - fully reversed stress amplitude corresponding to N = 103
cyclesm - slope of the high cycle regime curve (Part 2)
Fully reversed axial S-N curve for AISI 4130 steel. Note the break at the LCF/HCF transition and the endurance limit
Number of cycles, N
Str
ess
ampl
itude
, Sa
(ksi
)
Sy
Se
Part 1 Part 2 Part 3
Infinite life
Su
S103
10m A maS C N N= ⋅ = ⋅= ⋅ = ⋅= ⋅ = ⋅= ⋅ = ⋅
Fatigue SFatigue S --N curveN curve
25
Most of available S - N fatigue data has been obtained from fully reversed rotational bending tests. However, material behavior and the resultant S - N curves are different for different types of loading. It concerns in particular the fatigue limit Se.
The stress endurance limit, Se, of steels (at 106 cycles) and the fatigue strength, S103
corresponding to 103 cycles for three types of loading can be approximated as (ref. 1, 23, 24):
S103 = 0.90Su and Se = S106 = 0.5 Su - bending
S103 = 0.75Su and Se = S106 = 0.35 - 0.45Su - axial
S103 = 0.72Su and Se = S106 = 0.29 Su - torsion
103 104 105 106 1070.1
0.3
0.5
1.0
S103
Se
Number of cycles, Log(N)
Rel
ativ
e st
ress
am
plitu
de, S
a/S
u
Torsion
AxialBending
26
Approximate endurance limit for various materials:Approximate endurance limit for various materials:
Magnesium alloys (at 108 cycles) Se = 0.35Su
Copper alloys (at 108 cycles) 0.25Su< Se <0.50Su
Nickel alloys (at 108 cycles) 0.35Su <Se < 0.50Su
Titanium alloys (at 107 cycles) 0.45Su <Se< 0.65Su
Al alloys (at 5x108 cycles) Se = 0.45Su (if Su 48 ksi) or S e = 19 ksi (if S u> 48 ksi)
Steels (at 106 cycles) Se = 0.5Su (if Su 200 ksi) or S e = 100 ksi (if S u>200 ksi)
Irons (at 106 cycles) Se = 0.4Su (if Su 60 ksi) or S e = 24 ksi (if S u> 60 ksi)
S S –– N curveN curve
(((( )))) (((( ))))(((( ))))33
1 1 1
2
10101log lo
0
g
1
3
Am mm m
am m
e e
Aa aS C N or N C S C S
SSm and A
S S
N− −− −− −− −
= == == == =
= − == − == − == − =
= == == == =
⋅ ⋅⋅ ⋅⋅ ⋅⋅ ⋅
27
Fatigue Limit Fatigue Limit –– Modifying FactorsModifying FactorsFor many years the emphasis of most fatigue testing was to gain
an empirical understanding of the effects of variou s factors on the base-line S-N curves for ferrous alloys in the intermediate to long life ranges. The variables investigated inc lude:
- Rotational bending fatigue limit, S e’,
- Surface conditions, k a,
- Size, k b,
- Mode of loading, k c, Se = ka kb kc kd ke k f·Se’
- Temperature, k d
- Reliability factor, k e
- Miscellaneous effects (notch), k f
Fatigue limit of a machine part, S e
28
Ca
Effect of various surface finishes on the fatigue limit of steel. Shown are values of the ka, the ratio of the fatigue limit to that for polished specimens.
Below a generalized empirical graph is shown which can be used to estimate the effect of surface finish in comparison with mirror-polished specimens [Shigley (23), Juvinal(24), Bannantine (1) and other textbooks].
Surface Finish Effects on Fatigue Endurance LimitSurface Finish Effects on Fatigue Endurance LimitThe scratches, pits and machining marks on the surface of a material add stress concentrations to the ones already present due to component geometry. The correction factor for surface finish is sometimes presented on graphs that use a qualitative description of surface finish such as “polished” or “machined”.
(from J. Bannantine, ref.1)
k a
29
Size Effects on Endurance LimitSize Effects on Endurance LimitFatigue is controlled by the weakest link of the material, with the probability of existence (or density) of a weak link increasing with material volume. The size effect has been correlated with the thin layer of surface material subjected to 95% or more of the maximum surface stress.
There are many empirical fits to the size effect data. A fairly conservative one is:
• The size effect is seen mainly at very long lives.
• The effect is small in diameters up to 2.0 in (even in bending and torsion).
Stress effects in non-circular cross section members
In the case of non-circular members the approach is based on so called effective diameter, de.
The effective diameter, de, for non-circular cross sections is obtained by equating the volume of material stressed at and above 95% of the maximum stress to the same volume in the rotating-bending specimen.
' 0.097
' 0.097
1 0.3
0.869 0.3 10.0
1.0 8
1.189 8 250
b
b
e
e
e
e
if d inS
S d if in d in
or
if d mmS
S d if d mm
k
k
−−−−
−−−−
≤≤≤≤ = == == == =
≤ ≤≤ ≤≤ ≤≤ ≤
≤≤≤≤ = == == == =
≤ ≤≤ ≤≤ ≤≤ ≤
30
+
-
0.95σmax
σmax
0.05d/2
0.95dThe material volume subjected to stresses σ ≥ 0.95σmax is concentrated in the ring of 0.05d/2 thick.
The surface area of such a ring is:
( )max
22 20.95 0.95 0.0766
4A d d dσ
π = − =
The effective diameter, dThe effective diameter, d ee, for members , for members with nonwith non --circular cross sectionscircular cross sections
* rectangular cross section under bending
F
0.95
t
t
Equivalent diameter
max0.95 0.95 0.05
A Ft
A Ft t Ftσ
== − =
20.0766 0.05
0.808
e
e
d Ft
d Ft
=
=
31
Loading Effects on Endurance LimitLoading Effects on Endurance Limit
The ratio of endurance limits for a material found using axial and rotating bending tests ranges from 0.6 to 0.9.
The ratio of endurance limits found using torsion a nd rotating bending tests ranges from 0.5 to 0.6. A theoretical value obtaine d from von Mises-Huber-Hencky failure criterion is been used as the most po pular estimate.
( ) ( )(0.7 0.9)
0.7 0.9 ( 0.85)c
e axial e bendi
c
ngS S
suggested by Shigl y kek
≈ −≈ −≈ −≈ −
= − == − == − == − =
( ) ( )0.577
0.57( 0.59)e torsion e bendi
c
ng
c
S
suggested by Shigl kek y
ττττ ≈≈≈≈
= == == == =
32
From: Shigley and Mischke, Mechanical Engineering Design, 2001
, ,, ,
, ,, ;u T u T
e RT e RTu RT u R
dT
e T dSS S
S SS
kS
k= = == = == = == = =
Temperature EffectTemperature Effect
33
Reliability factor Reliability factor kkee
The reliability factor accounts for the scatter of reference data such as the rotational bending fatigue limit Se
’.
The estimation of the reliability factor is based on the assumption that the scatter can be approximated by the normal statistical probability density distribution.
1 0.08 ae zk = − ×= − ×= − ×= − ×The values of parameter za associated with various levels of reliability can be found in Table 7-7 in the textbook by Shigley et.al.
34
(source: S. Nishijima, ref. 39)
SS--N curves for assigned probability of failure; P N curves for assigned probability of failure; P -- S S -- N curvesN curves
35
Stress concentration factor, Stress concentration factor, KK tt, and the , and the notch factor effect, notch factor effect, kk ff
Fatigue notch factor effect k f depends on the stress concentration factor K t (geometry), scale and material properties and it is expressed in terms of the Fati gue Notch Factor K f.
1f
f
kK
====
36
1
σpeak
σn
σ22
σ11
σ3
3
3
2
A D B
C
F
F
σ22
σ22
σ22
σ33
σ33
σ11
C
A, B
D
Stresses in Stresses in axisymmetricaxisymmetricnotched bodynotched body
n
peakt n
FS
Aand
K
σ
σ σ
= =
=
37
Stresses in prismatic notched bodyStresses in prismatic notched body
n
peakt n
t
FS
Aand
K
K S
σ
σ σ
= =
==
A, B, C
σ22
σ22
σ22
σ33
D
E
σ11
σ peak
σn
σ22
σ11
σ3
33
2
A D B
C
F
F
1
E
38
Stress concentration factors used in fatigue Stress concentration factors used in fatigue analysisanalysis
peak
tn
peak
K
S
σσσσσσσσ
σσσσ
====
====
σn, S
x
dn
0
W
ρ
σ peak
Str
ess
S
σn
x
dn
0
W
ρ
Str
ess
M
σ peak
39
Stress concentration Stress concentration factors, Kfactors, K tt, in shafts , in shafts
Bending load
Axial load
S =
S =
40
S =
41
Similarities and differences between the stress fie ld near the nSimilarities and differences between the stress fie ld near the n otch otch and in a smooth specimenand in a smooth specimen
x
dn
0
W
ρ
σpeakS
tres
s
S
σσσσn=S σσσσpeak= σσσσn=S
P
42
The Notch Effect in Terms of the NThe Notch Effect in Terms of the N ominal Stressominal Stress
np
n1
∆∆∆∆S2
cyclesN0N2
Sesmooth
N (∆∆∆∆S)m = C
∆∆ ∆∆Sm
axS
tres
s ra
nge
S
n2
n3
n4 notchedsmoothe
ef
SS
K====
S
!!!f tK K≠≠≠≠fK −−−− Fatigue notch factor!
43
1
2
3
Nsm
ooth
22peak
smoothe
σ σσ σσ σσ σσσσσ
====
====
σpeak
Str
ess
1
2
Nnotched
notchedeσσσσ
M
notchedf notched
smoothsmoo he
e
tfo NNrKσσσσσσσσ
= == == == =
Definition of the fatigue notch factor Definition of the fatigue notch factor KK ff
1
σ22= σpeak
Str
ess
2
Nnotched
notchedeσσσσ
S
Se
44
( )11 1 1
1t
f t
KK q K
a r
−= + = + −+
1;
1q
a r=
+
PETERSON'sPETERSON's approachapproach
aaaa – constant,
r – notch tip radius;
11
1t
f
KK
rρ−= +
+
NEUBER’sNEUBER’s approachapproach
– constant,
r – notch tip radius
[[[[ ]]]]
[[[[ ]]]]
1.8
330010 .
.
u
u
a inS
for S in in
−−−− = ×= ×= ×= ×
45
The The NeuberNeuber constant constant ‘‘ ’’ for steels and aluminium alloysfor steels and aluminium alloys
46
Curves of notch sensitivity index ‘q’ versus notch radiusCurves of notch sensitivity index ‘q’ versus notch radius(McGraw Hill Book Co, from ref. 1)
47
Plate 1
W1 = 5.0 in
d1 = 0.5 in.
Su = 100 ksi
Kt = 2.7
q = 0.97
Kf1 = 2.65
Illustration of the notch/scale effectIllustration of the notch/scale effect
Plate 2
W2= 0.5 in
d2 = 0.5 i
Su = 100 ksi
Kt = 2.7
q = 0.78
Kf1 = 2.32
σσσσpeak
σσσσm1
ρρρρd1
W1
σσσσpeak
ρρρρ
d2
W2
σσσσm2
48
Procedures for construction of approximate fully re versedProcedures for construction of approximate fully re versedSS--NN curves for smooth and notched componentscurves for smooth and notched components
Nf (logartmic)
Juvinal/Shigley method
Nf (logartmic)
Collins method
Sar, σσσσar – nominal/local stress amplitude at zero mean stress σσσσm=0 (fully reversed cycle)!
σσσσf’
49
Procedures for construction of approximate fully re versedProcedures for construction of approximate fully re versedSS--NN curves for smooth and notched componentscurves for smooth and notched components
Sar, σσσσar – nominal/local stress amplitude at zero mean stress σσσσm=0 (fully reversed cycle)!
Sar
(loga
rtm
ic)
Nf (logartmic)
Manson method0.9Su
100 101 102 103 104 105 106 2 106
Se’kakckbkdke
Se’kakckbkdkek fSe
50
NOTE!NOTE!
• The empirical relationships concerning the S –N curve data are only estimates! Depending on the acceptable level of uncertaintyin the fatigue design, actual test data may be necessary.
• The most useful concept of the S - N method is the endurance limit , which is used in “infinite-life”, or “safe stress ” design philosophy.
• In general, the S – N approach should not be used to estimate lives below 1000 cycles (N < 1000).
51
Constant amplitude cyclic stress historiesConstant amplitude cyclic stress histories
Fully reversedFully reversed
σσσσσσσσmm = 0, R = = 0, R = --11
PulsatingPulsating
σσσσσσσσmm = = σσσσσσσσaa R = 0R = 0
CyclicCyclic
σσσσσσσσmm > 0 R > 0> 0 R > 0
52
Mean Stress EffectMean Stress Effect
sm< 0
sm= 0
sm> 0
No. of cycles, logN 2*106
Str
ess
ampl
itude
, log
Sa
time
sm< 0
sm= 0
sm> 0
Str
ess
ampl
itude
, Sa
0
53
The tensile mean stress is in general detrimental while the compressive mean stress is beneficial or has negligible effect on the fatigue durability.
Because most of the S – N data used in analyses was produced under zero mean stress (R = -1) therefore it is necessary to translate cycles with non- zero mean stress into equivalent cycles with zero mean stress producing the same fatigue life.
There are several empirical methods used in practice:
The Hiagh diagram was one of the first concepts where the mean stress effect could be accounted for. The procedure is based on a family of Sa – Sm curves obtained for various fatigue lives.
Steel AISI 4340, Steel AISI 4340,
SSyy = 147 = 147 ksiksi (Collins)
54
MeanMean Stress Correction for Endurance LimitStress Correction for Endurance Limit
Gereber (1874)S
S
S
Sa
e
m
u
+
=
2
1
Goodman (1899)S
S
S
Sa
e
m
u
+ =1
Soderberg (1930)S
S
S
Sa
e
m
y
+ = 1
Morrow (1960)S
S
Sa
e
m
f
+ =σ '
1
Sa – stress amplitude applied at the mean stress S m 0 and fatigue life N = 1-2x10 6cycles.
Sm- mean stress
Se- fatigue limit at S m=0
Su- ultimate strength
σσσσf’- true stress at fracture
55
MeanMean stress correction for arbitrary stress stress correction for arbitrary stress amplitude applied at nonamplitude applied at non --zero mean stresszero mean stress
Gereber (1874)
2
1a m
ar u
S S
S S+ =+ =+ =+ =
Goodman (1899) 1a m
ar u
S S
S S+ =+ =+ =+ =
Soderberg (1930) 1a m
ar y
S S
S S+ =+ =+ =+ =
Morrow (1960)'
1a m
ar f
S S
S σσσσ+ =+ =+ =+ =
Sa – stress amplitude applied at the mean stress S m 0 and resulting in fatigue life of N cycles.
Sm- mean stress
Sar- fully reversed stress amplitude applied at mean stress S m=0 and resulting in the same fatigue life of N cycles
Su- ultimate strength
σσσσf’- true stress at fracture
56
Comparison of various Comparison of various methods of accounting methods of accounting for the mean stress effectfor the mean stress effect
Most of the experimental data lies between the Good man and the yield line!
57
Approximate Goodman’s diagrams for ductile Approximate Goodman’s diagrams for ductile and brittle materialsand brittle materials
Kf
Kf
Kf
58
The following generalisations can be made when disc ussing mean stress effects:
1. The Söderberg method is very conservative and seldom used.
3. Actual test data tend to fall between the Goodman and Gerber curves.
3. For hard steels (i.e., brittle), where the ultimate strength approaches the true fracture stress, the Morrow and Goodman lines are essentially the same. For ductile steels (of > S,,) the Morrow line predicts less sensitivity to mean stress.
4. For most fatigue design situations, R < 1 (i.e., small mean stress in relation to alternating stress), there is little difference in the theories.
5. In the range where the theories show a large difference (i.e., R values approaching 1), there is little experimental data. In this region the yield criterion may set design limits.
6. The mean stress correction methods have been developed mainly for the cases of tensile mean stress.
For finiteFor finite --life calculations the endurance limit in any of the equations calife calculations the endurance limit in any of the equations ca n be n be replaced with a fully reversed alternating stress l evel corresporeplaced with a fully reversed alternating stress l evel correspo nding to that nding to that finitefinite --life value!life value!
59
Procedure for Fatigue Damage CalculationProcedure for Fatigue Damage Calculation
NT
n1
n2
n3
n4
cycles
σσσσe
NoN2
σσσσe'
Ni(�σi)m = a�σ2
N1 N3
�σ1
�σ3
Str
ess
rang
e, ��
1 1 2 2
1 1 2 2
1 21 2
11 2
.... ...
..... ...
i i
i i
ii i
n n nii i
n cycles applied atn cycles applied at n cycles applied atD
N cycles to failure at N cycles to failure at N cycles to failure at
n nn nD D D D
N N N N
σσσσσ σσ σσ σσ σσ σ σσ σ σσ σ σσ σ σ
∆∆∆∆∆ ∆∆ ∆∆ ∆∆ ∆= + += + += + += + +
∆ ∆ ∆∆ ∆ ∆∆ ∆ ∆∆ ∆ ∆
= + + + = + + + == + + + = + + + == + + + = + + + == + + + = + + + =∑∑∑∑
1 1 2 2
1 1....R
i i
LD n N n N n N
= == == == =+ ++ ++ ++ +
60
n1 - number of cycles of stress range ∆σ∆σ∆σ∆σ1 n2 - number of cycles of stress range ∆σ∆σ∆σ∆σ2n i - number of cycles of stress range ∆σ∆σ∆σ∆σi,
11
1D
N= - damage induced by one cycle of stress range ∆σ1,
11
1n
nD
N= - damage induced by n1 cycles of stress range ∆σ1,
22
1D
N= - damage induced by one cycle of stress range ∆σ2,
22
2n
nD
N= - damage induced by n2 cycles of stress range ∆σ2,
1i
i
DN
= - damage induced by one cycle of stress range ∆σi,
ini
i
nD
N= - damage induced by ni cycles of stress range ∆σi,
61
Total Damage Induced by the Stress HistoryTotal Damage Induced by the Stress History
1 1 2 2
1 1 2 2
1 21 2
11 2
.... ...
..... ...
i i
i i
ii i
n n nii i
n cyclesapplied atn cyclesapplied at n cyclesapplied atD
N cycles to failure at N cycles to failure at N cycles to failure at
n nn nD D D D
N N N N
σσσσσ σσ σσ σσ σσ σ σσ σ σσ σ σσ σ σ
∆∆∆∆∆ ∆∆ ∆∆ ∆∆ ∆= = == = == = == = =
∆ ∆ ∆∆ ∆ ∆∆ ∆ ∆∆ ∆ ∆
= + + + = + + + == + + + = + + + == + + + = + + + == + + + = + + + =∑∑∑∑
It is usually assumed that fatigue failure occurs w hen the cumulative damage exceeds some critical value such as D =1,
i.e. if D > 1 - fatigue failure occurs!
For D < 1 we can determine the remaining fatigue li fe:
1 1 2 2
1 1....R
i i
LD n N n N n N
= == == == =+ ++ ++ ++ +
LR - number of repetitions of the stress history to failure
(((( ))))1 2 3 .....R iN L n n n n= + + + += + + + += + + + += + + + + N - total number of cycles to failure
62
(((( ))))jN = am
jσσσσ−−−−
∆∆∆∆ if ∆∆∆∆σσσσj > σσσσe .
It is assumed that stress cycles lower than the fatigue limit, ∆σj < σe, produce no damage (Nj=∞) in the case of constant amplitude loading however in the case of variable amplitude loading the extension of the S-N curve with the slope ‘m+2” is recommended. The total damage produced by the entire stress spectrum is equal to:
1
j
jj
D D====
====∑∑∑∑
It is assumed that the component fails if the damage is equal to or exceeds unity, i.e. when D ≥ 1. This may happen after a certain number of repetitions, BL (blocks), of the stress spectrum, which can be calculated as:
BL = 1/D.
Hence, the fatigue life of a component in cycles can be calculated as:
N = BL××××NT, where, NT is the spectrum volume or the number of cycles extracted from given stress history.
NT = (NOP - 1)/2
If the record time of the stress history or the stress spectrum is equal to Tr, the fatigue life can be expressed in working hours as:
T = BL ××××Tr.
63
Main Steps in the Main Steps in the SS--NN Fatigue Life Estimation ProcedureFatigue Life Estimation Procedure
• Analysis of external forces acting on the structure and the component in question,
• Analysis of internal loads in chosen cross section of a component,• Selection of individual notched component in the structure,• Selection (from ready made family of S-N curves) or construction of S-
N curve adequate for given notched element (corrected for all effects),• Identification of the stress parameter used for the determination of the
S-N curve (nominal/reference stress),• Determination of analogous stress parameter for the actual element in
the structure, as described above,• Identification of appropriate stress history,• Extraction of stress cycles (rainflow counting) from the stress history,• Calculation of fatigue damage,• Fatigue damage summation (Miner- Palmgren hypothesis),• Determination of fatigue life in terms of number of stress history
repetitions, Nblck, (No. of blocks) or the number of cycles to failure, N.• The procedure has to be repeated several times if multiple stress
concentrations or critical locations are found in a component orstructure.
64
Example #2
An unnotched machine component undergoes a variable amplitudestress history Si given below. The component is made from a steelwith the ultimate strength Suts=150 ksi, the endurance limitSe=60 ksi and the fully reversed stress amplitude at N1000=1000cycles given as S1000=110 ksi.Determine the expected fatigue life of the component.
Data: Kt=1, SY=100 ksi Suts=150 ksi, Se=60 ksi, S1000=110 ksi
The stress history:
Si = 0, 20, -10, 50, 10, 60, 30, 100, -70, -20, -60, -40, -80, 70, -30,20, -10, 90, -40, 10, -30, -10, -70, -40, -90, 80, -20, 10, -20, 10, 0
65
Stress History
-100
-80
-60
-40
-20
0
20
40
60
80
100
0 5 10 15 20 25 30 35
Reversal point No.
Str
ess
(ksi
)
66
( )
4.1110886.1
log60log6
log110log3
log60log10log
log110log1000log
logloglog
26
6
=×=
=+=+
=+
=+
=+=
mC
Cm
Cm
Cm
Cm
CSmN
CSN
a
ma
67
S-N Curve
10
100
1000
1000 10000 100000 1000000 10000000
No. of cycles
Stre
ss a
mpl
itude
(ksi
)
68
Goodman Diagram
0
10
20
30
40
50
60
70
80
90
100
-100 -50 0 50 100 150
Mean stress (ksi)
Stre
ss a
mpl
itude
(ksi
)
SY
Suts
( )
( )
1
, 0
, 0,
1 1
11
m
m
a m maa r at S
e uts uts
a m aa r at S
ma r uts
uts
S S Sfor fatigue endurance S S
S S S
S S Sfor any stress amplitude S
SS SS
−
=
=
+ = = −
+ = =−
69
Calculations of Fatigue Damage a) Cycle No.11
Sa,r=87.93 ksi
N11= C×( Sa,N)-m= 1.866×1026×87.93-11.4=12805 cycles
D11=0.000078093 b) Cycle No. 14
Sa,r=75.0 ksi
N14= C×( Sa,N)-m= 1.866×1026×75.0-11.4=78561 cycles
D14=0.000012729 c) Cycle no. 15
Sa,r=98.28 ksi
N14= C×( Sa,N)-m= 1.866×1026×98.28-11.4=3606 cycles
D14=0.00027732
70
Results of "rainflow" counting D a m a g eNo. ∆∆∆∆S Sm Sa Sa,r (Sm=0) Di=1/Ni=1/C*Sa
-m
1 30 5 15 15.52 2.0155E-13 02 40 30 20 25.00 4.6303E-11 03 30 45 15 21.43 7.9875E-12 04 20 -50 10 10.00 1.3461E-15 05 50 -45 25 25.00 4.6303E-11 06 30 5 15 15.52 2.0155E-13 07 100 20 50 57.69 6.3949E-07 08 20 -20 10 10.00 1.3461E-15 09 30 -25 15 15.00 1.3694E-13 010 30 -55 15 15.00 1.3694E-13 011 170 5 85 87.93 7.8039E-05 7.80E-0512 30 -5 15 15.00 1.3694E-13 013 30 -5 15 15.00 1.3694E-13 014 140 10 70 75.00 1.2729E-05 1.27E-0515 190 5 95 98.28 0.00027732 0.000277
n0=15 D = 0.00036873 3.677E-04
D=0.000370 D=3.677E-04LR = 1/D =2712.03 LR = 1/D =2719.61
N=n0*LR=15*2712.03=40680 N=n0*LR=15*2719.61=40794