Predavanje Iz Zamora Materijala

Dynamic Loading
Sol id Mechanics Ander s Ekber g

1 (8)
Dynamic loading
More difficult to measure, analyze and estimate
dynamic loading compared to static loading
Loads can vary in time and space
Load range (or more accurately stress range) is the
most important parameter in fatigue analysis
This calls for a definition of a load cycle (or rather
stress cycle)
There can be large statistical scatter in the loading
High frequency content of the loading is difficult to
measure and/or analyze. This content may have an
effect on the fatigue behaviour
Standardized load spectra are often used as input to the
fatigue models.
Dynamic Loading
2 (8)
Varying amplitudes
Residual stresses
Overloads may introduce residual stresses due to
plastic deformations.
These stresses may supress the initiation of fatigue
cracks and/or lead to closure of existing cracks during
(parts of) the load cycle
This is normally beneficial
Add influence from different load cycles
This is far from obvious
I n continuum approaches (initiation), this is done by
damage accumulation
I n crack propagation analysis, it is done by the use of a
crack propagation law.
Dynamic Loading
3 (8)
Load cycles
A load cycle is a closed loop in load space
For harmonic loading, the load cycle
starts from a certain load magnitude
moves through a max-value and a min-value back to
the start magnitude (or the other way around)
The load cycle is then completely defined by the
amplitude and mid value
t
F
t
F
The problem in identifying a load cycle comes when we
are not dealing with harmonic loadpaths
Dynamic Loading
4 (8)
Rainflow counting
Depict the loading sequence as a function of time. For
conveni ence
start with largest maximum or smallest minimum
use straight lines between (local) minima and maxima
Start from the top and let a drop start from every maximum
and minimum. A drop stops if:
it starts from max and passes a larger or equal max
it starts from min and passes a larger or equal min
it reaches the run of another drop
I dentify closed loops by joining drops
Dynamic Loading
5 (8)
t
1
2
3
5
4
6
7
8
Rainflow counting example
1 passes an equally large maximum
2 passes a larger minimum
3 passes a larger maximum
4 reaches the run of drop 2
6 falls out
7 falls out
1 and 6, 2 and 5, 3 and 4; 7 and 8 are running
the same distances in the opposite directions.
These couples are forming closed loops in
load space and can thus be identified as
stress cycles with minimum and maximum
magnitudes and a mid value.
Dynamic Loading
6 (8)
Rainflow counting notes
Rainflow counting is used to identify the most damaging
stress cycles
When applying the rainflow count method, the
chronological order of applied loads diminishes. I n other
words, you no longer know at what instant of time the
overloads and (their pertinent residual stresses) occur
More than one load component rainflow counting
should be carried out on the stress
The rainflow count method is only applicable for
uniaxial loading. There are theories for how to extend the
method to multiaxial loading (however not generally
accepted)
The rainflow count method is fairly easy to implement in
a computer code. Several such codes exist and are in
practical use
Dynamic Loading
7 (8)
A Uniaxial Stress Cycle
Pulsating
compression
Alternating
tension/compression
time
Pulsating
tension
a
R =

min
max
m
=
1
2
(
max
+
min
)
a
=
1
2
(
max

min
)
Stress ratio Mid Stress Stress amplitude
R = 1
m
= 0
a
=
max
R = 0
m
=

max
2
a
=

max
2
R
m
a
=
=
=

min
min
2
2
(half the stress range)
Dynamic Loading
8 (8)
10 -20 -15 -10 -5 0 5
-5
-4
-3
-2
-1
0
1
2
3
4
5
R-ratio
R
=
m
i
n
/
m
a
x
min
The R-ratio
R
R
R
R
=
=
( )

min max
max min
max
max
a
a
a
1
2
2
1
2
1
The stress cycle is
defined by the
stress amplitude
and the R-ratio

max max
= +10
Uniaxial High Cycle Fatigue
1 (8)
High cycle fatigue initiation
Two primary causes
From stress concentrations, such as pores, inclusions,
initial cracks etc.
There will be a local increase of the stress levels
Due to a pile-up of dislocations, which will form slip
bands, which will grow to form cracks
Cracks form due to a local decrease in the fatigue
strength
Which of this two mechanisms that will dominate
depends on the purity of the material, the nature of the
loading, etc.
2 (8)
For some materials, there is a
stress amplitude below which
no fatigue failures will occur
This is called the fatigue limit
The fatigue limit can be
considered to be a material
parameter
Fatigue life at different stress levels
Designing for infinite life is to assure that no stress levels
exceeds the fatigue limit
But how do we translate the fatigue limit to other types of
l oadi ng?
10
1
10
2 3
10
4
10
10
5
10
6
Stress cycles to failure
N
Stress
amplitude
10
7
FL
Infinite fatigue life
Finite fatigue life
steel
aluminum
a
3 (8)
The Whler (S-N) curve
The Whler curve shows
fatigue life corresponding to a
certain stress amplitude
I t is also called a S-N-curve
The diagram is primarily valid
for uniaxial loading
The curve does not take into
account any effects of the mid
value of the stress during a
stress cycle
Consequently, the curve is only valid for loading with a certain
R-ratio
There are also Whler curves for entire components (e.g.
chains, wheel axles). Then, fatigue life is normally plotted
against applied load
10
1
10
2 3
10
4
10 10
5
10
6
Stress cycles to failure
N
Stress
amplitude
10
7
FL
Finite fatigue life
steel
aluminum
a
4 (8)
N
10
FL
steel
aluminum
1
10
2
10
3
10
4
10
5
10
6
10
7
a
Given stress amplitude
Gives
pertinent
fatigue life
Given
service life
Gives allowable stress amplitude
No fatigue damage is induced, the component
can sustain an infinite number of load cycles
The Whler
diagram can be
used to design for
finite (and infinite)
life
This can be done
either for a given
service loading or a
given service life
This slope on the Whler curve
can be described by the equation

a
m
N
f
= K
Using the Whler (S-N) curve
5 (8)
How to Construct a Whler Diagram
10
N
FL
1
10
2
10
3
10
4
10
5
10
6
10
7
a
FRA
For alternating loading,
FRA
=
UTS
For pulsating loading
FRA
=
UTS
2
Note that the Whler curve is
only valid for a certain R-value
( R =
min max
)
1000
0.9
UTS
FL
0.5
UTS
For low strength
wrought steel
For steel, the
fatigue limit
corresponds to
10
6
<N<10
7
6 (8)
FL
Y

UTS
FLP
Plasticity
m
The fatigue limits for two cases
fully reversed tension/compression (or
bendi ng)
pulsating tension (or bending) and the
yield limit, are needed to create the
di agram
The diagram is valid
for different R-ratios
The diagram is only valid
for uniaxial loading
The Haigh diagram
7 (8)
Reduction of the Haigh Diagram
Reduction is made on the amplitude axis
Reduction is normally made with respect to
Surface roughness (taking also the effect of corrosion into
account)
Size of the raw material
Loaded volume (no reduction in the case of a notch)
Fatigue notch factor K
f
The reduction factors are taken from diagrams (see Material Fatigue,
p.9-12)
The fatigue notch factor is determined from
K
f
=1+q(K
t
-1)
where K
t
is the stress concentration factor and q depends on the notch
radius (q<1, which gives K
f
< K
t
)
8 (8)
Create the Haigh diagram
a
Reduce the Haigh diagram
FL

FLP
UTS
FL
FLP
UTS
Y
I nsert your service stress, P, in
the reduced Haigh diagram
UTS
Y
P
Check if your in the safe
area. Calculate safety factors
m
P
Using the Haigh Diagram
( , ) K K
t m f a

Damage Accumulation
1 (2)
Assume that, during the service
life, we have 500 loadings of type 1
(defined by mid-value and
magnitude), 1000 loadings of type 2
and 10000 loadings of type 3
The Palmgren Miner rule states
that failure occurs when
n
N
i
i
i
I
=
=
1
1
where n
i
is the number of applied
load cycles of type i, and N
i
is the
pertinent fatigue life
Palmgren Miners Rule
1 2 3
1
1
1
2
1
3
1
4
1
5
1
6
N
7
1
a
3
n
i
N
i
=
500
10
3
i=1
I
+
10
3
10
5
+
10
4
= 0.51<1
Damage Accumulation
2 (2)
Damage
Define the damage induced in the i:th load cycle as
D
N
i
i
=
1
Then, Palmgren Miners rule states that fatigue failure
occurs when
D
i
i
I
=
=
1
1
Note that, in this case, the accumulation is made for all
load cycles and not all types of load cycles
The previous example can then be expressed as
D
i
i
I
= + + = <
=

1
3 3 5 4
500 10 10 10 10 10 0 51 1 .
Whether to accumulate damage or fatigue life is just a
matter of convenience
Fatigue Limit
1 (1)
Determining the Fatigue Limit
The Staircase Method (see Mat. fat. p.16-19)
Estimate mean value, S
m
, and standard deviation, d, of
the fatigue limit
Perform the first fatigue test at the stress level S
m
+d
I f thie specimen fails, decrease the stress level by d. I f
the specimen survives, increase the stress level by d
Continue until 15 to 30 specimens have been tested
From the test data, the real mean value and standard
deviation of the fatigue limit for the studied material
can be obtained
Finding the fatigue limit by experiments is very costly!
High Cycle Fatigue (HCF) part II
1 (20)
FL
FLP
UTS

a FL
=
m
= 0
time

a FLP
=
a

m FLP
=
Plastic
deformations
FLP
FLP
UTS
FL
FLP
Loaded
volume
Surface roughness
Size of raw
material
Haigh diagram I
Y
Haigh diagram
Reduced Haigh diagram
2 (20)
UTS
Y
P
m
P
( , ) K K
t m f a

Haigh diagram II
A
A
C
O
B
SF
a
AA'
AP
=
SF
m
OB'
OA
=
SF
am
OC'
OP
=
m
const =
a
const =
K
K
f a
t m
const
K q K
K K
f t
f t
= +
1 1 ( )
Service stress
Safety factors
3 (20)
Modern Fatigue Design
Background
Evolution in structural design due to
increased computational power
CAD/CAE - software
Need for new fatigue design methods that are
valid for a general type of loading
easy to implement in a computer code
Several options, but no method with general validity
HCF: equivalent stress is defined and compared to a
fatigue limit (expressed in the equivalent stress)
LCF: calculation of damage connected to the
constitutive model of the material. Fatigue damage
connected to the plastic deformation
LEFM: effective stress intensity range
4 (20)
Multiaxial high cycle fatigue initiation
Problem:
The Haigh diagram is valid for
Uniaxial loading
One stress component
Solution:
Assume that, in the general case,
fatigue behaviour is influenced by
Applied shear stress amplitude
Hydrostatic stress
Based on these assumptions, derive a fatigue initiation
criterion that defines a limiting stress magnitude for
which fatigue cracks will develop) for a general type of
l oadi ng.
Assumes undamaged material (continuum mechanics)
5 (20)
Hydrostatic stress
The hydrostatic stress is the mean value
of normal stresses acting on the material
point (positive in tension)
A tensile (positive) hydrostatic stress
opens up microscopic cracks (Stage II
crack growth)

h
= + +
( )
1
3
x y z
The hydrostatic stress is a stress invariant

ij
=
11 12 13
21 22 23
31 32 33

h
= = + +
( )
1
3
1
3
11 22 33 ii
regardless of coordinate system
6 (20)
Shear stress measures
The shear stress initiates slip bands which
leads to microscopic cracks (stage I crack
growth)
Since a static shear stress have no
influence on the fatigue damage, the
shear stress amplitude is employed
Two measures
Tresca shear stress

Tresca
=

1 3
2
von Mises stress
We need to define the amplitudes of these

vM
=
( )
+
( )
+
( )
1
2
1 2
2
2 3
2
3 1
2
7 (20)
Equivalent Stress Measures
Uniaxial Case
One stress component
Mid value and amplitude of this stress
component are taken to reflect the
fatigue properties
The stresses during a load cycles are
defined by a service stress
Multiaxial Case
Six stress components (general case)
Hydrostatic stress and shear stress
amplitude are taken to reflect the
fatigue properties
The stresses during a load cycles are
defined by a closed curve
FLP
m
Plastic
zone
UTS
FL
Service
stress
a
FATIGUE
NO FATIGUE
FATIGUE
FATIGUE
NO
FATIGUE
Shear stress
amplitude
NO
FATIGUE
Stresses
during one
load cycle
Plastic
zone
Plastic
zone
e3
c
3
e3
e3
8 (20)
Shear Stress Amplitude
General
I t has been found empirically that a superposed static
shear stress does not have any influence on the fatigue
initiation

FL FLP
= whereas
FL FLP
I n order to eliminate the influence of a superposed shear

stress, the shear stress amplitude is normally used in
multiaxial HCF-criteria
This amplitude is the difference between the current
shear stress magnitude and the mid value of the shear
stress for the current stress cycle
For the general case, this amplitude is rather
complicated to compute (see Fat igue a Sur vey,
Appendix I )
9 (20)
Shear stress Uniaxial case
Mohrs stress circle for loading in a uniaxial case
2
= 0
max
1
=
max
x
2
= 0
max
1
=
max
x
2
= 0
max
1
=
max
x
time
Max normal and shear stress correspond to the same
directions throughout the load cycle
45 45
45
time
max
time
P
time
max
mid
10 (20)
The deviatoric stress tensor
The stress tensor

ij
xx xy xz
yz yy yz
zx zy zz
=
Split into volumetric and a deviatoric part

ij
xx xy xz
yz yy yz
zx zy zz
xx xy xz
yz yy yz
zx zy zz
=
= +
h
h
h
h
h
d
1 0 0
0 1 0
0 0 1
I
The volumetric part contains the hydrostatic stress
The deviatoric part reflects influence of shear stresses
11 (20)
Mid value of the deviatoric stress tensor
In-phase
ij ij ij
a c f t = + ( )
a
ij
and c
ij
are constants
f t ( ) is a common time dependent function
Fixed principal directions
Every component of corresponds to a fixed direction
throughout the loading

ij
xx xy xz
yx yy yz
zx zy zz
t
t t t
t t t
t t t
t
t
t
a
a
d
d d d
d d d
d d d
d
d
d
d
( ) =
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
=
( )
( )
( )
=
1
2
3
11
22
0 0
0 0
0 0
0 0
0
dd
d
d
d
d
0
0 0
0 0
0 0
0 0
33
11
22
33
a
c
c
c
f t
( )
High Cycle Fatigue (HCF) part II Sol id Mechanics Ander s Ekber g
12 (20)
Movie 1 Click me!
13 (20)
Mid value of the deviatoric stress tensor III
Limitations In-phase loading
I n in-phase loading, the
stresscomponents have their
max- and min-magnitudes at
the same instant in time
I n out-of-phase loading, max-
and min magnitudes occur at
different instants of time for
different stress components
time
stress
time
stress
The case of out-of-phase loading is much more difficult to
analyse, for instance due to difficulties in
Defining a stress cycle
Defining a mid value of the shear stress
14 (20)
Mid value of the deviatoric stress tensor IV
Limitations Fixed principal directions
Rotating principal directions
" "
ij,p
d
d
d
d
=
1
1
3
0 0
0 0
0 0
corresponds to a rotating coordinate system
I nstead we have to look at the full deviatoric stress tensor and
find its mid value

ij
xx xy xz
yz yy yz
zx zy zz
,m
d
m
d
h
h
h
m
= =
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
15 (20)
Mid value of the deviatoric stress tensor V
Finding the mid value in a general case Click me!
16 (20)
Amplitude of the deviatoric stress tensor
The mid value of the deviatoric stress tensor is found as

ij,m
d
m
d
d
m
d
m
d
m
= =
1
1
3
0 0
0 0
0 0
(proportional loading)
( m denotes mid-value of component during stress cycle)
or as

ij
xx xy xz
yz yy yz
zx zy zz
,m
d
m
d
h
h
h
m
= =
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
(general)
the amplitude of the deviatoric stress tensor is defined as

ij ij ij
t t
,a
d d
,m
d
( ) = ( ) (or

a
d d
m
d
t t ( ) = ( ) )
17 (20)
Amplitude of the deviatoric stress tensor II
For in-phase loading with fixed principal directions (proportional
loading), we can express the amplitude of the Tresca and von Mises
stress using the amplitude of the deviatoric stress tensor

Tresca,a
1,a
d
3,a
d
2
( )
( ) ( )
t
t t
=

where (
1,a
d
1
d
1,m
d
( ) ( ) t t = etc)

vM,a
a a a a a a
( ) ( ) ( ) ( ) ( ) ( ) ( )
, , , , , ,
t t t t t t t =
( )
+
( )
+
( )
1
2
1 2
2
2 3
2
3 1
2
(it can be shown that using
a
or
a
d
gives the same results)
The max values are given as

Tresca,a
1,a
d
3,a
d
2
=

where (

1,a
d 1,max
d
1,min
d
2
=

)

vM,a
a a a a a a
=
( )
+
( )
+
( )
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
Amplitude of the deviatoric stress tensor II
For in-phase loading with fixed principal directions (proportional
loading), we can express the amplitude of the Tresca and von Mises
stress using the amplitude of the deviatoric stress tensor

Tresca,a
1,a
d
3,a
d
2
( )
( ) ( )
t
t t
=

where (
1,a
d
1
d
1,m
d
( ) ( ) t t = etc)

vM,a
a a a a a a
( ) ( ) ( ) ( ) ( ) ( ) ( )
, , , , , ,
t t t t t t t =
( )
+
( )
+
( )
1
2
1 2
2
2 3
2
3 1
2
(it can be shown that using
a
or
a
d
gives the same results)
The max values are given as

Tresca,a
1,a
d
3,a
d
2
=

where (

1,a
d 1,max
d
1,min
d
2
=

)

vM,a
a a a a a a
=
( )
+
( )
+
( )
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
18 (20)
Equivalent stress criteria
Sines criterion

EQS a
d
a
d
a
d
a
d
a
d
a
d
S h,mid eS
=
( )
+
( )
+
( )
+ >
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
c
Crossland criterion

EQC
a a a a a a
C h,max eC
=
( )
+
( )
+
( )
+ >
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
c
Dang van criterion

EQDV
1,a 3,a
DV h,max eDV
2
=

+ > c
19 (20)
Concluding remarks
Fatigue analysis
Calculate the state of stress
Apply the equivalent stress criterion, fatigue if

eq e
>
I n the case of no fatigue, calculate safety coefficient as
SF =

e
EQ
Pros Cons
Suitable for computer analysis Corrosion correction etc.
General state of stress Lack of empirical knowledge
I dentify critical parts of component Separates between fatigue / no fatigue
Have a physical basis
20 (20)
L
u
n
c
h
Low Cycle Fatigue (LCF)
1 (8)
Low Cycle Fatigue Introduction
Finite or Infinite Fatigue Life?
HCF
Both infinite or finite fatigue life is possible and
can be analyzed
LCF
Only finite fatigue life is possible and should be
analyzed using LCF-criteria
Stress or Strain?
HCF
Elastic material
Small strain increment -> large stress increment
LCF
Stresses close to (or at) the yield limit
Small stress increment -> large strain increment.
Best resolution if strains are employed in
fatigue model.
x
x
x
x

2 (8)
LCF Introduction contd
Damage Mechanisms
I nduced fatigue damage will be due to
global plasticity
local plasticity (same as in HCF)
For high load amplitudes (and/or high maximum
magnitudes of loading) global plasticity will be the
dominating cause
For low load magnitudes, the model should tend to
similar results as for HCF criteria (i.e. the Whler
curve)
Note that the static load carrying capacity must not be
exceeded. This must be checked separately
3 (8)
Morrow Design Rule
According to Morrow, the relationship between
strain amplitude,
a
, and pertinent fatigue life, N
f

can be written as

a a
el
a
pl
f
f f f
= + =

( )
+ ( )
E
N N
b c
2 2
N
f
is the number of load cycles to failure
log
a
log
f
log

f
E
a
el
a
pl
a
=
a
el
+
a
pl
0 2 4 6
log 2N
f
( )
el
pl
4 (8)
Mean Stress Effects
According to Morrow, the effect of a mean stress,
sm, can be taken into account by redefining the
criterion as

a a
el
a
pl
f m
f f f
= + =

( )
( )
+ ( )
E
N N
b c
2 2
log
a
log 2N
f
( )
0 2 4 6
m
< 0
(compressive)
m
> 0
(tensile)
5 (8)
Coffin Manson Design Rule
For the elastic part, the relationship between strain amplitude
and fatigue life can be approximated by

a
el
UTS
f
=

1 75
0 12
.
.
E
N
The fatigue life in the plastic part can be approximated by
a
pl
f
=

0 5
0 6
0 6
.
.
.
D N
where D is the ductility, defined as
D
A
A
ln
0
fra
fra
This yields the Coffin Manson relationship
a
=1. 75
UTS
E
N
f
0.12
+ 0.5D
0.6
N
f
0.6
6 (8)
LCF Design some notes
Comparison of the criteria
The Morrow criterion includes five material parameters
E b c , , , ,
f f
The Coffin Manson criterion includes three material
parameters
E D , ,
UTS
It is also possible to express the criteria using the strain
range, (and not the strain amplitude
a
) in the
Coffin-Manson criterion (see Eq. 5h
Approximations

a f
a f
= = =
= = =
1 2 10
0 5 1 10
3
4
% %
. % %
N
N
7 (8)
Stress and strain concentrations
Stress concentration
The stress concentration factor ahead of a
notch is defined as
K
max
In a similar manner, the strain
concentration factor ahead of a
notch is defined as
K
max
If we load above the yield limit, which is
the case for LCF conditions, we get

and K K
K
K
t
Y
K
t
K
> K

8 (8)
Neubers rule
K K K

=
f
2
where K q K
f t
= + 1 1 ( )
which yields

max max f
2
= =

K K K
Assuming elastic conditions far from the
notch, we get the Neuber hyperbola

max max
f
2
=

K
E
2
The equation of the Neuber hyperbola has
two unknown
But, the stress must also fulfil constitutive
relationship between stress and strain
Thus, two equations and two unknown
max
max
Constitutive
rel ati on
Neuber
hyperbol a
Fatigue crack propagation
1 (20)
Repetition Crack initiation and growth
Small cracks
Shear driven
I nteract with microstructure
Mostly analyzed by continuum
mechanics approaches
Large cracks
Tension driven
Fairly insensitive to
microstructure
Mostly analyzed by fracture
mechanics models
Stage II
Tension driven crack
(propagation)
Stage I
Shear driven crack
(initiation)
2 (20)
Stress intensity factors and fracture
I n static loading, the stress intensity factor for a small
crack in a large specimen can be expressed as
K f a
I
=
( )
, where f depends on geometry
I f the stress is kept constant, we will get fracture for
a certain crack length, a=a
C
, which will give
K
I
=K
IC
.
For a<a
C
(K
I
<K
IC
) the crack will not propagate (in
theory)
I n dynamic loading, we will still get fracture if the
stress intensity factor, for some instant of time, exceeds
K
I
=K
IC
However, for K
I
<K
IC
, the crack may still propagate.
Since this means that a (and K
I
) will increase, we
will eventually obtain fracture when a=a
C
.
3 (20)
Crack growth as a function of K
I n experiments, crack propagation has been measured as a
function of the stress intensity factor
I II III
log
da
dN
logK
K
th
K
C
There exists a threshold value of
K below which fatigue cracks
will not propagate
At the other extreme, K
max
will
approach the fracture toughness
K
C
, and the material will fail
A l inear relationship between
log
d
d
a
N
( )
and K in region II
Note that K depends on the
crack size. This is not shown in
the plot
4 (20)
Crack growth in region I
For small K (region I ), crack propagation is
difficult to predict since it depends on
microstructure and flow properties of the material
Here, the growth may even come to an arrest
Crack growth rate is sensitive to the size of the
grains. Finer grains gives
Closer spacing of grain boundaries,
which the crack has to break through
I ncreased yield stress (normally)
Decreased roughness of the crack
Crack growth predicted by
models of type d d
p
a N f =
( )
,
where
p
is plastic shear strain range
empirical adjustment of K- d d a Ncurve
I II III
log
da
dN
logK
K
th
K
C
5 (20)
Crack growth in region II and III
Region II
For larger magnitudes of K (region II), the crack growth rate
will be governed by a power law (such as Paris law)
The crack growth rate is fairly insensitive to the microstructure
(however, the constants m and C are, of course, different for
different materials)
I f region I I includes the dominating part of the fatigue life, the
fatigue life can be directly estimated by integrating Paris law
Region III
I f the stress intensity ratio is increased even further (region III),
the crack growth rate will accelerate and finally fracture will
occur
The behavior of this fracture is rather sensitive to the
microstructure and flow properties of the material.
6 (20)
Crack propagation laws introduction
I t has been found that, for dynamic loading of a crack, the
three most important factors determining the propagation
(growth) of the crack are
K K K
max min
the stress intensity range
R K K
min max
the stress intensity ratio

H the stress history
Thus, the crack growth rate (i.e. growth per stress cycle)
can be expressed as

d
d
a
N
f K R = ( , , ) H
where
d
d
a
N
is the crack growth per stress cycle
7 (20)
Paris law
Paris law can be written as
d
d
a
N
C K
m
=
where C and m are material parameters
One of the first (1962) and most widely used fatigue crack
propagation criteria
Algorithm
1. Find stress intensity factor for the current geometry
2. Find crack length corresponding to K K
max
=
C
3. Check if the requirements for linear elastic fracture mechanics are
fulfilled
4. I ntegrate Paris law
5. Solve for the number of stress cycles corresponding to failure
Important
I f the stress intensity factor includes a geometric function of a,
estimated (or analytic) values of this function has to be used
8 (20)
Paris law drawbacks
Compared to a general crack propagation criterion

d
d
a
N
f K R = ( , , ) H
Paris law does not account for
mean stress effects (described by the R-ratio)
history effects (introduced by H)
Further, Paris law is only valid in conditions with
uniaxial loading
long cracks
LEFM-condi ti ons
We will have a closer look at
short crack theory
retardation models due to overloads
crack closure effects
crack propagation in multiaxial loading
9 (20)
Short cracks
So far
d
d
a
N
f K = ( )
where K depends on the
amplitude of the normal stress
(and geometry)
But short cracks are shear
stress driven also LEFM is
not valid
Two types of short cracks
mechanically short cracks
propagate faster than large
cracks with same K
microstructurally short cracks
interact closely with the
microstructure and grow fast
Short Crack Theory
Microstructural
Threshold
C
r
a
c
k

S
p
e
e
d
Crack Length
1
>
2
a
1
< a
2
10 (20)
Variable amplitude loading (H)
crack
Plastic
zone
(tensi on)
Loading
Unloading
crack
Plastic zone
(compressi on)
Y

A (tensile) overload will introduce (compressive) residual
stresses
These residual stresses will influence K and thus the rate of
crack propagation
11 (20)
The Wheeler model
The Wheeler model is used to define the
reduction of the crack growth rate due to an
overl oad
The reduction factor is
defined as

R
c
0
d
=
+
a d

The reduced crack growth rate
is then calculated as
d
d
d
d
R
R
a
N
a
N
=
crack
d
0
a
d
c
Variable Amplitude Loading, contd
The Wheeler model is appropriate for single
overloads
The reduction of crack growth rate acts only as long as
the cracks current plastic zone is within the
plastic zone from the overload
Multiple overloads or stochastic loads
Cycle-by-cycle integration of
Appropriate crack growth law
that takes
Retardation/acceleration effects
into account
Normal crack propagation laws are usually
conservative
12 (20)
crack
13 (20)
Crack closure (R)
Elber, in 1970, discovered that crack closure exists in cyclic
loading, even for loads that are greater than zero
This crack closure will decrease the fatigue crack growth rate
by reducing the effective stress intensity range
The stress intensity rate
K K K
K K

=
[ ]
max min
min min
max , 0
Crack closure att K=K
op
gives
K K K
eff op

max
Paris law using effective stress intensity rate
d
d
eff
a
N
C K
m
=
Empirical relation
K R K
R R R R
op
=
= + +
( )
( ) . . .
max
0 25 0 5 0 25 1 1
2
14 (20)
Crack closure and arrestment
Crack closure
K
max
K
eff
K
op
K
min
K
I f the crack is closed throughout the stress cycle, the
crack will arrest
This is not the only mechanism of a crack to arrest!
15 (20)
Crack closure and arrestment II
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
-10
-8
-6
-4
-2
0
2
4
6
8
10
R
K
o
p

a
n
d

K
m
i
n
Kop
Kmin
Smallest magnitude of
K
min
in Paris law
K
Paris
K
Elber
K
Paris
Using Elber
correction in
Paris law is
conservative
(predicts a
longer
fatigue life)
The only
difference
when using
Elber
correction is
in a new,
higher K
min
16 (20)
Crack arrestment
d
d
eff
eff
max op
max
max min
op
max
max
min
max
op
a
N
C K C
K
K
K
C
K K
K
K
C
K
K K
K
K
K
C
K
K
K
K
K
K
K
C
m
m
m
m
m
= =

1
1
1 RR
K
K
K
m
op

For
1
1
0
=
R
K
K
op
we get
K K R K = ( ) =
op th
1
and
d
d
a
N
= 0
For
1
1
1
=
R
K
K
op
we get
K K
R
=
op
1
1
17 (20)
Crack growth treshold
By taking crack
closure into account
(using Elber
correction), we can
model a R-ratio
dependence
compressive mid
stress slower crack
propagati on
tensile mid stress
faster crack
propagati on
18 (20)
Crack arrest at different scales
A The load magnitude is
below the fatigue limit
we will not initiate any
(macroscopic cracks)
B The applied load gives a
stress intensity below the
fatigue threshold stress
intensity macroscopic
cracks will not continue to
grow
A
B
Fatigue
failure
loga
K
I,th
= U a
No fatigue failure
No propagation
log
log
e
19 (20)
Cracks in mixed mode loading
+
+
+
+
Cracks that are loaded in mixed mode, will normally tend to
propagate in pure mode I
One exception is when a crack propagates along a weak zone
(e.g. a weld). I n this case, an effective stress intensity factor can
be employed
K K K
eff I II
= +
( )
2
2
0 8 .
20 (20)
Crack propagation summary
Under one dimensional, elastic conditions and constant load range
Paris law, can predict fatigue life of large cracks
Under variable amplitude loading, plastic residual stress fields mostly
gives a decrease in crack growth rate.
Microstructurally small cracks interact closely with microstructure.
Mechanically small cracks propagate faster than long cracks.
Closure effects of large cracks can give a pronounced effect. I ts one
mechanism behind crack arrestment
I n multiaxial loading, most cracks tend to propagate in pure mode I
Less mature areas
Cases where LEFM is not applicable
The propagation of short, especially microstructurally short, cracks
Cases where crack closure and crack friction has a profound effect
Conditions of variable amplitude loading
Multiaxial loading conditions
Designing against fatigue
1 (4)
Design model
Design code
Fatigue life model
Design considerations
Predetermined fatigue life
Predetermined loads
Uncertanties
Applied load
Dynamical response
Material parameters
Decisions
Acceptable risc
Acceptable cost
2 (4)
Design using design models
Haigh diagram
Finite fatigue life
Whler curve
Morrows design rule
Coffin-Manson design rule
LEFM-analysis (Paris law, Elber correction etc.)
Fatigue testing
The uncertainties in the fatigue models are so high that
validating testing should always be performed (if
possi bl e)
Be aware that the fatigue life is dependent on the working
envi ronment!
3 (4)
Safe life design
Determine the fatigue life during which the component
shall be operational
Estimate the loading during this service life
Assure by fatigue analysis
Haigh diagram (infinite fatigue life)
Whler curve
Morrow / Coffin-Manson design rules
LEFM-anal ysi s
that the component can sustain this fatigue life
Assure that the material does not contain initial defects
larger than assumed (a that gives K K <
th
in continuum
design and a a =
0
in LEFM-design)
Validate the fatigue life by testing
4 (4)
Fail safe design
I dentify critical crack size, a
c
causing fracture
Apply safety factor, SF to the critical flaw size. This gives a
tolerable crack size, a
t
.
I dentify initial flaw size a
0
as the largest flaw size an inspection
may miss (much larger than the detectability limit)
The time to grow the crack from a
0
to a
t
is computed using
LEFM-analysis (Paris law)
Time in-between the inspections should be lesser than this time.
Preferably, one inspection should be able to miss the crack
without a failure of the component
I f a crack is detected
repair the component (or take it out of service).
make a new inspection at a time which is less than computed
time for the crack to grow from detected size to tolerable size
Also make analysis to ensure that the failure of a component
will not lead to catastrophic failure of the structure
Fatigue of non-metallic materials
1 (4)
Fatigue of concrete
Concrete is very brittle
Failure occur due to tensile (residual) stresses
Tensile residual stresses can occur even in compressive
loading due to microcracking
A suitable equivalent stress criterion should include the
amplitude of the largest principal stress and the mean
stress during a cycle
I n fracture mechanics analysis, the constitutive relations
should, preferably, account for the presence of micro-
cracks. This is difficult!
I n tensile loading, the fatigue (and static) resistance of
concrete is very low and the fatigue properties of a
structure will be determined by the reinforcement
2 (4)
Fatigue of polymers (e.g. plastics)
Plastics can be analyzed in a manner similar to that of
metal fatigue analysis
The material properties of plastics are very sensitive to
temperature
There exists a variety of microscopic deformation modes
in polymeric materials
More pronounced viscous behavior than metals. This can
lead to increased temperature under cyclic loading and
thermal softening
Polymeric materials exhibit only cyclic softening
behavior. Cyclic stability is rarely observed and only
under conditions of small strain amplitudes
Plastics are non-linear elastic
3 (4)
Fatigue of Wood
The fatigue properties of wood are, in general, good
The appearance of fatigue failure is similar to that of static
failure
No distinct fatigue limit is found for wood
Construction timber is sensitive to humidity
Rapid drying can cause the timber to crack
Repeated moisturing/drying can cause the cracks to grow
As for mechanical fatigue, the weak spots of a timber
construction lie in the joints and, to some extent, in the
cracks in the wood
These phenomena can be analyzed, for instance by a
fracture mechanics approach
Wood is a severely anisotropic material
4 (4)
Literature
Journals
Fatigue & Fracture of Engineering Materials and
Structures
I nternational Journal of Fatigue
Wear
Books
Bannantine, Comer and Handrock Fundamentals of
Metal Fatigue Analysis
Frost, Marsh and Pook Metal Fatigue
Suresh Fatigue of Materials
Andersson Kompendium i utmattning
Handbooks and Codes
SAE
ASM & ASTM
DI N and Eurocodes
Mekanresul tat
Fat igue a survival kit
1 (24)
FATIGUE
A Cr ash Cour se
Crack propagation
Retardation effects
Fatigue initiation
Limited High Cycle Fatigue life
Low Cycle Fatigue
Stress Concentrations in LCF
Stress cycle identification
Damage accumulation
Multiaxial fatigue initiation
Crack closure
Crack growth arrestment
Crack growth in mixed mode loading
2 (24)
Damage mechanisms
a
b
c
d
el
pl
FL
pl
FL

el
Plastic
shakedown limit
Elastic
shakedown limit
Yield limit
Fatigue limit
3 (24)
Fatigue design methods
Parameters
Magnitude of loading
Complexity of loading
Damage of material
Envi ronment
Possibility of
experi mental
val i dati ons
Experi ence
...
Low Stress
Magnitudes
No Cracks
Large Cracks
M
u
l
t
i
a
x
i
a
l
S
t
r
e
s
s
High Stress
Magnitudes
U
n
i
a
x
i
a
l

S
t
r
e
s
s
A D
B
C
E
F
G
H
4 (24)
Philosophies in Design
Safe life means that the component is designed to
last a previously determined life
Equivalent stress criterion (no cracks initiated)
Fracture mechanics analysis (crack does not
propagate enough to cause failure in operational
life)
Crcaks are not allowed to grow to failure.
Also, (partial) failure of a part of the component
through crack development must not endanger safety
Design steps
A flaw size, that will lead to fracture, is estimated.
A tolerable flaw size is defined
An initial flaw size is defined
The time to grow the crack from initial to
tolerable size is calculated
Based on this, inspection schemes are defined
Fail Safe
Safe Life
5 (24)
Fatigue
How?
Micro structural changes which
cause nucleation of permanent
damage. Persistent slip bands
(PSB)
Creation of microscopic cracks
Growth and coalescence of
microscopic flaws to form
"dominant" cracks
Stable propagation of dominant
macro crack(s)
Structural instability or complete
failure
Where?
I nitiation where
s
>1
Normal l y
weak spots
- inclusions
- small cracks
stress concentrations
- inclusions
- corrosion pits
Fatigue will then grow
(propagate) in the
direction of max
s
I nitiation propagation
6 (24)
Fatigue crack initiation and propagation
Small cracks
Shear dri ven
I nteract with microstructure
Mostly analyzed by continuum
mechanics approaches
Large cracks
Tension driven
Fairly insensitive to
microstructure
Mostly analyzed by fracture
mechanics models
Stage II
Tension driven crack
(propagation)
Stage I
Shear driven crack
(initiation)
7 (24)
Influencing factors
Stress concentrations
Give rise to increased stress levels
Volumetric effects
I ncreased volume subjected to high load magnitude gives
larger probability of a weak, highly stressed spot
Large raw material gives reduced fatigue resistance due to
manufacturing quality
Environmental effects
Corrosion
Corrosion pits which act as stress concentrators.
Cracks will always form due to corrosion -> no fatigue limit
Heat will often decrease the fatigue strength and the fatigue
limit may diminish at higher temperatures.
8 (24)
FL
FLP
UTS

a FL
m
0
time

a FLP
a

m FLP
Plastic
deformations
FLP
FLP
UTS
FL
FLP
Loaded
volume
Surface roughness
Size of raw
material
Haigh diagram I
Y
Haigh diagram
Reduced Haigh diagram
9 (24)
UTS
Y
P
m
P
( , ) K K
t m f a

Haigh diagram II
A
A
C
O
B
SF
a
AA'
AP
SF
m
OB'
OA
SF
am
OC'
OP
m
const
a
const
K
K
f a
t m
const
K q K
K K
f t
f t
+
1 1 ( )
Service stress
Safety factors
10 (24)
N
10
FL
steel
aluminum
1
10
2
10
3
10
4
10
5
10
6
10
7
a
Given stress amplitude
Gives
pertinent
fatigue life
Given
service life
Gives allowable stress amplitude
No fatigue damage is induced, the component
can sustain an infinite number of load cycles
The Whler
diagram can be
used to design for
finite (and infinite)
life
This can be done
either for a given
service loading or a
given service life
This slope on the Whler curve
can be described by the equation

a
m
N
f
K
Whler (S-N) curve
1
k
or be approximated
as a straight line
Val id onl y
f or a cer t ain
R-r at io
11 (24)
t
1
2
3
5
4
6
7
8
Stress cycle identification rainflow counting
1 passes an equally large maximum
2 passes a larger minimum
3 passes a larger maximum
6 falls out
7 falls out
1 and 6, 2 and 5, 3 and 4; 7 and 8 form
closed loops (i.e. stress cycles)
Depict the loading sequence as a function of time.
start with largest max or smallest min
use straight lines between min and max
Let drops start from every max and min nd stop if:
it starts from max and passes a larger or equal max
it starts from min and passes a larger or equal min
it reaches the run of another drop
I dentify closed loops by joining drops
12 (24)
Assume that, during the service
life, we have 500 loadings of type 1
(defined by mid-value and
magnitude), 1000 loadings of type 2
and 10000 loadings of type 3
The Palmgren Miner rule states
that failure occurs when
D
n
N
j
j
J
i
i
i
I

1 1
1
where D
j
is the damage of a load
cycle, n
i
is the number of applied
load cycles of type i, and N
i
is the
pertinent fatigue life
Damage accumulation Palmgren - Miners Rule
1 2 3
1
1
1
2
1
3
1
4
1
5
1
6
N
7
1
a
D
n
N
j
j
J
i
i
i
I
+ +
<

1
3
1
3
5
4
500
10
10
10
10
0 51 1 .
13 (24)
Equivalent stress criteria
Sines criterion

EQS
a a a a a a
S h,mid eS

( )
+
( )
+
( )
+ >
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
c

EQS a
d
a
d
S h,mid eS
+ >
3
2
ij ij
c
, ,
Crossland criterion

EQC
a a a a a a
C h,max eC

( )
+
( )
+
( )
+ >
1
2
1 2
2
2 3
2
3 1
2
, , , , , ,
c

EQS a
d
a
d
C h,max eC
+ >
3
2
ij ij
c
, ,
Dang Van criterion

EQDV
1,a 3,a
DV h,max eDV
2

+ > c
Val id onl y f or
pr opor t ional l oading
(in-phase and f ixed
pr incipal dir ect ions)
14 (24)
Equivalent stress criteria components
Shear stress measures
The shear stress initiates microscopic cracks
(stage I crack growth)
A static shear stress have no influence on
fatigue damage the shear stress amplitude
is employed
Hydrostatic stress
Mean value of normal stresses that opens
up cracks (Stage II crack growth)

h
+ +
( )
1
3
1
3
11 22 33 ii
regardless of coordinate system
(stress invariant)
15 (24)
The deviatoric stress tensor
The stress tensor can be split into deviatoric and volumetric part

ij
xx xy xz
yz yy yz
zx zy zz
xx xy xz
yz yy yz
zx zy zz
ij
ij kk
1
]
1
1
1
1
]
1
1
1
+
1
]
1
1
1
+ +
h
h
h
h
d
h
d
1 0 0
0 1 0
0 0 1
1
3
I
The volumetric part contains the hydrostatic stress
The deviatoric part reflects influence of shear stresses
Mi dval ue:

ij
xx xy xz
yx yy yz
zx zy zz
,m
d
m
d
d
m
d
m
d
m
d
m
d
m
d
m
d
m
d
m
d
m

1
]
1
1
1
1
(proportional loading)
Ampl i tude:
ij ij ij
t t
,a
d d
,m
d
( ) ( ) (or

a
d d
m
d
t t ( ) ( ) )
16 (24)
x
x
x
x
Low Cycle Fatigue

Stresses close to (or at) the yield limit
Small stress increment large strain
increment. Best resolution if strains are
employed in fatigue model
I nduced fatigue damage due to global
plasticity
Loading above yield limit, (LCF) gives

With stress concentration factor
K
max
and strain concentrationfactor
K
max
we get K K

K
K
t
K
> K
Y
t
K


17 (24)
Stress concentrations in LCF Neubers rule
At a stress concentration,Neubers rule
gives the ralation between stress and
strain as

max max
f
2
max,a max,a
f
2
,a

K
E
K
E
2
2
This equation has two unknown
Stress and strain must also fulfil constitutive
relationship (for cyclic loading)
2 equations and 2 unknown
max
max
Constitutive
rel ati on
Neuber
hyperbol a
18 (24)
LCF Design Rules
According to Morrow, the relationship between
strain amplitude,
a
, and pertinent number of load
cycles to failure, N
f
can be written as
a
f
f f f

( )
+ ( )
E
N N
b c
2 2
or, with a static mean stress 0
a
f m
f f f

( )
( )
+ ( )
E
N N
b c
2 2
According to Coffin Manson, the relationship
can be simplified as
a
1. 75
UTS
E
N
f
0.12
+ 0.5D
0.6
N
f
0.6
Mor r ow
Mor r ow wit h mean
st r ess cor r ect ion
Cof f in Manson
19 (24)
Fatigue crack growth
I n experiments, crack propagation has been measured as a
function of the stress intensity factor
I II III
log
da
dN
logK
K
th
K
C
There exists a threshold value of
K below which fatigue cracks
will not propagate
At the other extreme, K
max
will
approach the fracture toughness
K
C
, and the material will fail
Linear relationship between
log
d
d
a
N
( )
and K in region II
d d a N depends also on crack
size. This is not shown in the
plot
20 (24)
Paris law
Paris law can be written as
d
d
a
N
C K
m
(C and m are material parameters)
1. Find stress intensity factor for the current geometry
2. Find crack length corresponding to K K
max

C
3. (Check if LEFM is OK)
4. I ntegrate Paris law
5. Solve for the number of stress cycles corresponding to failure
Paris law does not account for
mean stress effects (described by the R-ratio)
history effects (introduced by H)
and is only valid for
uniaxial loading
long cracks
LEFM-condi ti ons
21 (24)
Variable amplitude loading
A (tensile) overload will introduce (compressive)
residual stresses
These residual stresses will influence K and thus the
rate of crack propagation
The Wheeler model is used to defines reduction of the
crack growth rate due to overload
The reduction factor is
defined as

R
c
0
d
_
,
a d

Reduced crack growth rate
is then calculated as
d
d
d
d
R
R
a
N
a
N
_
,

crack
d
0
a
d
c
22 (24)
Crack closure
Normally cracks only grow when they are open
The Elber accounts for crack closure, also for tensile loads by
defining an effective stress intensity range
Paris law
K K K
K K

[ ]
max min
min min
max , 0
Elber correction for crack closure att K=K
op
K K K
eff op

max
Modified Paris law
d
d
eff
a
N
C K
m

Empirical relation

K R K
R R R R
op

+ +
( )
( ) . . .
max
, where
0 25 0 5 0 25 1 1
2
23 (24)
Crack closure
K
max
K
eff
K
op
K
min
K
Crack closure
Using Elber correction in Paris
law is non-conservative (predicts
a longer fatigue life)compared to
standard Paris law
The only difference when
using Elber correction is
in a new, higher K
min
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
-10
-8
-6
-4
-2
0
2
4
6
8
10
R
K
o
p

a
n
d

K
m
i
n
Kop
Kmin
Smallest magnitude of
K
min
in Paris law
K
Paris
K
Elber
K
Paris
24 (24)
Crack arrest at different scales
A The load magnitude is
below the fatigue limit
we will not initiate any
(macroscopic cracks)
B The applied load gives a
stress intensity below the
fatigue threshold stress
intensity macroscopic
cracks will not continue to
grow
A
B
Fatigue
failure
loga
K
I,th
U a
No fatigue failure
No propagation
log
log
e

Predavanje Iz Zamora Materijala

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Predavanje Iz Zamora Materijala

Uploaded by

Copyright:

Available Formats

Dynamic Loading

Sol id Mechanics Ander s Ekber g

I n order to eliminate the influence of a superposed shear

Split into volumetric and a deviatoric part

Low Cycle Fatigue (LCF)

This yields the Coffin Manson relationship

Low Cycle Fatigue (LCF)

Low Cycle Fatigue

Fat igue a survival kit

You might also like