Beruflich Dokumente
Kultur Dokumente
Fall 2005
1
FATIGUE - What is it?
Ni = ?
Np = ?
NT = ?
4
Intrusions and Extrusions:
The Early Stages of Fatigue Crack Formation
5
Schematic of Fatigue Crack Initiation Subsequent Growth
Corresponding and Transition From Mode II to Mode I
Locally, the crack grows in shear;
macroscopically it grows in tension.
6
The Process of Fatigue
7
Features of the Fatigue Fracture Surface of a Typical
Ductile Metal Subjected to Variable Amplitude Cyclic
Loading
(Collins, ref. 22 )
8
Appearance of Failure Surfaces Caused by
Various Modes of Loading (SAE Handbook)
9
Factors 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.
10
Strength-Fatigue Analysis Procedure
Stress-Strain
Analysis
Damage
Analysis
peak
Stress
peak
Kt =
peak
Stress
n
S
n
n
y peak
0 y 0
=
dn
T
dn
T
S
12
Constant and Variable Amplitude Stress Histories;
Definition of a Stress Cycle & Stress Reversal
a) Constant amplitude stress history In the case of the peak stress
max history the important parameters
are:
D peak = max
peak
- min
peak
;
Stress
One min
D peak max
peak
- mipeak
cycle
a
peak
= = n
2 2
0 Time
Variable amplitude stress history
b)
max
peak
+ min
peak
peak
m = ;
2
Stress
mipeak
R = penak
One
reversal Time
max
0
13
Stress History and the Rainflow Counted Cycles
i+1
Stress
i-1
i-2
i+2
Time
i
A rainflow counted cycle is identified when any two adjacent reversals in thee
stress history satisfy the following relation:
ABS i -1 - i ABS i - i +1
14
The Mathematics of the Cycle Rainflow Counting Method
for Fatigue Analysis of Stress/Load Histories
A rainflow counted cycle is identified when any two adjacent reversals in thee
stress history satisfy the following relation:
ABS i -1 - i ABS i - i +1
The stress amplitude of such a cycle is: The stress range of such a cycle is:
ABS i -1 - i D = ABS i -1 - i
a =
2
The mean stress of such a cycle is:
i -1 + i
m =
2
15
16
Stress History
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
Reversal point No.
17
Stress History
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
Reversal point No.
18
Stress History
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
Reversal point No.
19
Stress History
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
6
5
Stress (MPa)x10
4
2
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
Reversal point No.
20
Number of Cycles
According to the
Rainflow Counting
Procedure (N. Dowling, ref. 2)
21
The Fatigue S-N method
(Nominal Stress Approach)
22
Whlers Fatigue Test
S = peak
Smin
Smax
23
Infinite life
S103
Sy
Se
Number of cycles, N
Fully reversed axial S-N curve for AISI 4130 steel. Note the break at the LCF/HCF transition and
the endurance limit
Characteristic parameters of the S - N curve are:
Se - fatigue limit corresponding to N = 1 or 2106 cycles for
steels and N = 108 cycles for aluminum alloys,
S103 - fully reversed stress amplitude corresponding to N = 10 3S =C
N = 10
a Nm m A
cycles
m - slope of the high cycle regime curve (Part 2) 24
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.
S103
0.5
Bending
Axial
0.3
Torsion
0.1
103 104 105 106 107
Number of cycles,
Log(N)(at 106 cycles) and the fatigue strength, S 3
The stress endurance limit, Se, of steels 10
corresponding to 10 cycles for three types of loading can be approximated as (ref. 1, 23, 24):
3
1 1 A 1
S N curve - -
Sa = C
N m = 10 A
N m
or N = C m
( Sa ) m =C m
( Sa ) m
S
( )
2
1 S
103 103
m = - log and A = log
3 Se Se
26
Fatigue Limit Modifying Factors
For many years the emphasis of most fatigue testing was to gain
an empirical understanding of the effects of various factors on
the base-line S-N curves for ferrous alloys in the intermediate
to long life ranges. The variables investigated include:
28
Size Effects on Endurance Limit
Fatigue 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:
Se 1 if d 0.3 in
kb = =
Se' 0.869d -0.097 if 0.3 in d 10.0 in
or
Se 1.0 if d 8 mm
kb = ' =
Se 1.189d -0.097 if 8 d 250 mm
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, d e.
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.
29
max
The effective diameter, de, for members
0.95max with non-circular cross sections
+
A = Ft
A0.95 max = Ft - 0.95t = 0.05Ft
0.95t
t
Equivalent diameter
0.0766d e2 = 0.05 Ft
F d e = 0.808 Ft
30
Loading 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.
31
Temperature Effect
Su ,T Su ,T
Se ,T = Se , RT kd = Se , RT ; kd =
Su , RT S u , RT
32
Reliability factor ke
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.
ke = 1 - 0.08 za
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.
33
S-N curves for assigned probability of failure; P - S - N curves
(source: S. Nishijima, ref. 39)
34
Stress concentration factor, Kt, and the
notch factor effect, kf
1
kf =
Kf
35
F
22 Stresses in axisymmetric
notched body
C 2
peak C
22
F
22 n n = S =
33 A
A, B
and
11 1
A
peak = K t n
D B
33
22
D 3
11
33
36
Stresses in prismatic notched body
22 F
A,
B, C 2
F
n = S =
peak C
A
22
22 and
D
11 n peak = K t n
11
= Kt S
E
A
22 D B 1
E 33
33 F
37
Stress concentration factors used in fatigue
analysis
S M
peak peak
Stress
Stress
n, S
peak n
Kt =
n x
0 x peak 0
=
dn S dn
W W
38
Stress concentration
factors, Kt, in shafts
S=
Bending load
S=
Axial load
39
S=
40
Similarities and differences between the stress field near the notch
and in a smooth specimen
S P
Stress peak
n=S
peak= n=S
x
0
dn
W
41
The Notch Effect in Terms of the Nominal Stress
S
n1 N (DS)m = C
DSmax
S
n2 DS2
n3 Sesmooth
Stress range
Sesmooth
n4
Senotched =
Kf
np N2 N0 cycles
Stress
Stress
Nsmooth
notched
Se
2 enotched
e
1 1 1
3
Nnotched Nnotched
esmooth
K f = notched for N smooth = N notched
e
43
PETERSON's approach
Kt - 1
K f = 1+ = 1 + q ( K t - 1)
1+ a r
1.8
1 300
a = 10-3 [ in.]
q= ; a constant,
Su
1+ a r r notch tip radius;
for S u in [ in.]
NEUBERs approach
Kt - 1
K f = 1+ constant,
44
The Neuber constant for steels and aluminium alloys
45
Curves of notch sensitivity index q versus notch radius
(McGraw Hill Book Co, from ref. 1)
46
Illustration of the notch/scale effect
Plate 1
W1 = 5.0 in
d1 = 0.5 in.
peak peak
m1 Su = 100 ksi
m2
Kt = 2.7
q = 0.97
Kf1 = 2.65
Plate 2
d2
d1 W2= 0.5 in
d2 = 0.5 i
W1 W2 Su = 100 ksi
Kt = 2.7
q = 0.78
Kf1 = 2.32
47
Procedures for construction of approximate fully reversed
S-N curves for smooth and notched components
f
Collins method
Juvinal/Shigley method
Nf (logartmic)
Sar, ar nominal/local
stress amplitude at zero
mean stress m=0 (fully
reversed cycle)!
Nf (logartmic)
48
Procedures for construction of approximate fully reversed
S-N curves for smooth and notched components
Sekakckbkdke
Sekakckbkdkekf
Se
50
Constant amplitude cyclic stress histories
m = 0, R = -1 m = a R = 0
Cyclic
m > 0 R > 0
51
Stress amplitude, Sa Mean Stress Effect
sm> 0
0
sm= 0 time
sm< 0
Stress amplitude, logSa
sm< 0
sm= 0
sm> 0
53
Mean Stress Correction for Endurance Limit
2
Sa Sm
Gereber (1874) + =1 Sa stress amplitude
Se Su applied at the mean
stress Sm 0 and fatigue
life N = 1-2x106cycles.
Sa Sm
Goodman (1899) + =1 Sm- mean stress
Se Su Se- fatigue limit at Sm=0
Su- ultimate strength
Sa S m
Soderberg (1930) + =1 f- true stress at fracture
Se S y
Sa S m
Morrow (1960) + ' =1
Se f
54
Mean stress correction for arbitrary stress
amplitude applied at non-zero mean stress
2
Sa S m Sa stress amplitude
Gereber (1874) + = 1
S ar S u applied at the mean
stress Sm 0 and
resulting in fatigue life of
Sa Sm N cycles.
Goodman (1899) + =1
Sm- mean stress
S ar Su
Sar- fully reversed stress
Sa Sm amplitude applied at
Soderberg (1930) + =1 mean stress Sm=0 and
resulting in the same
S ar Sy
fatigue life of N cycles
Su- ultimate strength
Sa Sm
Morrow (1960) + =1 f- true stress at fracture
'
S ar f
55
Comparison of various
methods of accounting
for the mean stress effect
Most of the experimental data lies between the Goodman and the yield line!
56
Approximate Goodmans diagrams for ductile
and brittle materials
Kf
Kf
Kf
57
The following generalisations can be made when discussing
mean stress effects:
1. The Sderberg 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 finite-life calculations the endurance limit in any of the equations can be
replaced with a fully reversed alternating stress level corresponding to that
finite-life value!
58
Procedure for Fatigue Damage Calculation
n1 D
Ni(Di)m = a
n2 D
Stress range, D
n3 D
e
n4
'e
N1 N2 N3 No cycles
NT
n1 cycles applied at D 1 n2 cycles applied at D 2 ni cycles applied at D i
D= + .... + ...
N 1 cycles to failure at D 1 N 2 cycles to failure at D 2 N i cycles to failure at D i
i
n1 n2 ni n
D = Dn1 + Dn 2 + ..... + Dni = + + ... + =i
N1 N 2 Ni 1 Ni
1 1
LR = =
D n1 N1 + n2 N 2 + .... ni N i 59
n1 - number of cycles of stress range
D 1 n2 - number of cycles of stress range
D 2 ni - number of cycles of stress range
D i, 1
D1 = - damage induced by one cycle of stress range D1,
N1
n1
Dn1 = - damage induced by n1 cycles of stress range D1,
N1
1
D2 = - damage induced by one cycle of stress range D2,
N2
n2
Dn 2 = - damage induced by n2 cycles of stress range D2,
N2
1
Di = - damage induced by one cycle of stress range Di,
Ni
ni
Dni = - damage induced by ni cycles of stress range Di,
Ni
60
Total Damage Induced by the Stress History
n1 n2 ni i
n
D = Dn1 + Dn2 + ..... + Dni = + + ... + =i
N1 N 2 Ni 1 Ni
It is usually assumed that fatigue failure occurs when 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 life:
1 1
LR = = LR - number of repetitions of
D n1 N 1 + n2 N 2 + .... ni N i the stress history to failure
61
62
Main Steps in the S-N Fatigue 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 or
structure.
63
64
Stress History
100
80
60
40
20
Stress (ksi)
0
-20 0 5 10 15 20 25 30 35
-40
-60
-80
-100
65
66
67
68
69
Results of "rainflow" counting Damage
No. DS Sm Sa Sa,r (Sm=0) Di=1/Ni=1/C*Sa-m
1 30 5 15 15.52 2.0155E-13 0
2 40 30 20 25.00 4.6303E-11 0
3 30 45 15 21.43 7.9875E-12 0
4 20 -50 10 10.00 1.3461E-15 0
5 50 -45 25 25.00 4.6303E-11 0
6 30 5 15 15.52 2.0155E-13 0
7 100 20 50 57.69 6.3949E-07 0
8 20 -20 10 10.00 1.3461E-15 0
9 30 -25 15 15.00 1.3694E-13 0
10 30 -55 15 15.00 1.3694E-13 0
11 170 5 85 87.93 7.8039E-05 7.80E-05
12 30 -5 15 15.00 1.3694E-13 0
13 30 -5 15 15.00 1.3694E-13 0
14 140 10 70 75.00 1.2729E-05 1.27E-05
15 190 5 95 98.28 0.00027732 0.000277
n0=15 D= 0.00036873 3.677E-04
D=0.000370 D=3.677E-04
LR = 1/D =2712.03 LR = 1/D =2719.61
N=n0*LR=15*2712.03=40680 N=n0*LR=15*2719.61=40794
70