2 Fatigue Chapter 2 v. 2016.03 III - Lecture Notes

Muzio M.
Gola
Costruzione di macchine - Machine Design
Fatigue - Chapter 2
a
ol
Chapters
.g
1 History and problem overview
2 Stress-life: material properties
m
3 Stress-life: component - infinite life
4 Stress-life: finite life io
5 Strain-life
6 Crack propagation and fracture
uz
m
 Pierluigi Lenoci
These slides are a support to the Machine Design, and related, classes held by prof. M.M. Gola at POLITO, Technical University of Torino – For educational purposes and classroom use only
Chapter 2 - Stress-life: basic material properties (1/2)
m.gola
2011
2014
Index of contents
Basic properties, data and
representations 8. The Haigh plot: data
a
1. Basic ideas on fatigue data approximation
ol
& Wöhler curve 9. The Haigh plot: linear
simplifications
.g
2. Rm vs. σD-1 for steel
3. Static strength values and 10. The Haigh plot: the
limit Goodman area
m
material tables
4. Fatigue limits from static 11. Haigh, Goodman-
properties Smith and Moore
io
5. Temperature factor 12. FKM refinement: the
uz
6. Influence of mean stress limit area
on fatigue 13. Use of the Haigh
m
7. The Haigh plot : diagram: the admissible

experimental data area
Version Muzio M. Gola - DIMEAS 2
March 2016 Politecnico di Torino - Italia
1. Basic ideas on fatigue data & Wöhler curve (1/10)
m.gola
2011
2008
The Stress-Life method is the classical method for fatigue analysis

of metals and has its origins in the work of Wöhler from about
1850.
a
A stress amplitude σ a is produced at a suitable location on
ol
specimen, and the number N of cycles to failure is found.
.g
Wöhler called a “safe stress level” the one below which failure does
not occur.
m
Above this safe stress level failure will occur within a certain “life”,
measured as number of cycles.
io
Crack growth is not explicitly accounted for in the Stress-Life
uz
method. Because of this, Stress-Life methods are often considered
crack initiation (or incubation) life estimates.
m

m.gola
2011
2012
Diameter of round specimens: steel 6÷ 8 mm

light alloys
Width, rectangular specimens: steel
8 ÷13
2÷ 5
mm
mm { frequently
used values
a
light alloys 4÷ 8 mm
ol
σ
.g
The basic definitions
m
of fatigue diagrams,
symbols and names io
are in Ch. 1 Sect. 5 t
The testing machines
uz
are described in
Ch. 1 Sect. 10
m
dauskardt.stanford.edu/.../bmenzel_research.html

m.gola
2011
2012
Experimental determination The basic procedures of

of Wöhler curves Wöhler test are outlined
in ch. 1 sect. 4
σa N/mm2
a
3 levels, 6 specimens each
ol
350 x xxxx x A group of
.g
specimens is
x xx xx x
tested at each of
300 (at least) three
m
x x x xx x
stress levels
x: broken
spanning the
io
specimens
expected range of
250
uz
finite life stress
amplitude
m
200
103 104 105 106 107 108 N
m.gola
2016
Data are used to obtain

life probability distribution the “life probability” curves
at each stress level, on
which the stress-life curves
are then fitted.
σa N/mm2
a
ϕ
% survival probability curves
ol
350
.g
x-µ
300
m
(x −µ )2
10% 1 −
ϕ= e 2σ 2
f
io ? f =σ 2
FWHM = 2 2 ln 2 σ
250 50%
uz
90%
m
200
103 104 105 106 107 108 N
October 2015 Politecnico di Torino - Italia
m.gola
2011
2015
The fatigue limit can be determined with the “staircase”

method. A small number of equally spaced stress levels is set
around the expected fatigue limit. The first specimen is tested at
σa N/mm2 the highest stress level. If failure occurs, a further test is carried
a
out at a one-step down stress.
ol
350 If failure does not occur (say, at 107 cycles) the test is stopped
and a further specimen is tested at a one-step up stress level.
.g
Up to 15÷20 specimens must be employed.
300
{ m
staicase method
x x x
20 specimens
x o x x x x x
io x o o
o
o o x o
4 levels
o o
250
uz
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
sequence of tested specimens

m
x: broken specimens
200 o: un-broken specimens

m.gola
2011
2008
Through the appropriate statistical

tools, the fatigue limit at 50% level of
probability is then estimated, with its
σa N/mm2
a
standard deviation...
ol
350
.g
number of un-broken specimens
300
{ m
3 0 x x x
x o x x x x x
6
2
2
5
io x o o
o
o o x o
250 0 2 o o
uz
number of broken specimens
m
σD at 50% x: broken specimens

200 o: un-broken specimens
probability

m.gola
2011
2008
life probability distribution
σa N/mm2
a
% survival probability curves
ol
350
.g
fatigue limit probability
distribution
300
m
10%
σD
50%
io σD= fatigue limit
at any σm
250
uz
90%
conventional limit life
m
200
103 104 105 106 107 108 N
m.gola
2011
2008
Wöhler curve can be divided in three parts
a
low cycle fatigue (N< 103) – strain controlled
ol
σa high cycle fatigue
.g
(~104<N<~106) – stress controlled
m
infinite life (N>~106~~107)
io
uz
σD
m
?
0 4 6 Log N
m.gola
2011
2012
Certain materials have a endurance limit which represents a

stress (amplitude) level below which the material does not fail
and can be cycled infinitely. If the applied stress (amplitude)
a
level is below the endurance limit of the material, the structure
ol
is said to have an infinite life. It is also called fatigue limit .
Curve A shows an
.g
σa σm ≠ 0
endurance limit, typical
of steel and titanium in
m
benign (non-corrosive) limit life
σD , fatigue &
environmental conditions
io endurance limit
If a “knee” on the curve
uz
is not easily detectable,
fatigue limit σD is defined
N
m
at N=106 or 2x106 or
5x106 or even 107 cycles.
diagram from

m.gola
2011
2012
Many non-ferrous metals and alloys, such as aluminum,

magnesium, and copper alloys, do not exhibit well-defined
endurance limits. These materials instead display a
a
continuously decreasing S-N response, similar to Curve B.
ol
In such cases a proper
.g
σa
fatigue limit σD for a
given number of cycles
m
must be specified.
An effective fatigue
io
limit for these materials
uz
is defined as the stress σD fatigue limit
that causes failure at
N
m
1x108 or 5x108 loading

cycles.
diagram from

Sect. 2 to 5 - FKM
m.gola
2011
2014
The main points to be remembered about the FKM-Guideline named

“Analytical strength assessment of components in mechanical
engineering” are:
a
Valid for components produced with or without machining or by welding of
ol
steel, of iron or of aluminum materials that are intended for use under
normal or elevated temperature conditions, and in detail:
.g
• for static loading • for milled or forged steel, also
stainless steel, cast iron materials
m
• for fatigue loading with more as well as aluminum alloys or
than about 104 cycles (HCF)
io cast aluminum alloys
constant or variable amplitude
• for components temperatures
uz
• for components with geometrical from -40°C to 500°C for steel
notches from -25°C to 500°C for cast iron
from -25°C to 200°C for alum.
m
• for components with welded

joints • for a non-corrosive environment

Sect. 2 to 5 - Basic material properties, data and factors
m.gola
2011
2012
The purpose of Sections 2 to 5 of this chapter is the investigation of

the main factors which influence fatigue as they are seen on the
specimens which are used to characterise the materials.
a
They are condensed in the following:
ol
• Re , KT,D and Rm from monotonic static tests corrected for deff ,
.g
“effective diameter” of the semi-finished product (sect. 3 sl. 1,5)
• σD-1,T= KT,D fW,s Rm = KT,D fW,s · Kd,m KA Rm,N
m
definition of Rm,N in sect. 3 sl. 1,2 ; KT,D in sect. 5 sl. 1, 2;
fW,s in sect. 4 sl. 1,2 ; Kd,m in sect. 3 sl. 3; KA in sect. 3 sl. 4
io
A selection of data is presented, together with formulas and their
uz
coefficients. It is stressed that data and formulas must be connected
in a coherent system.
m
Reference is made here to the FKM-Guideline 5th edition, representing the state-of-
the-art knowledge of German origin at the year 2003.
Warning: an updated 6th edition 2012 is now available; variations not of interest here.
2. Rm vs. σD-1 for steel (1/3)
m.gola
2011
2015
Experimental observation: the fatigue limit is in a certain relation

with static strength values.
a
A deep knowledge of the inner reasons for this relation is of great
ol
importance for material scientists when looking for evidence and
theories aiming at explaining the inner failure mechanisms in
.g
different failure modes. This will not be approached in these slides.
m
However, even an approximate empirical relation is of great practical
importance when exact fatigue data are not available, and the
io
fatigue limit (quite expensive to obtain) must be roughly estimated
from the (much cheaper) value of tensile strength.
uz
Next slides will show examples from the (ancient) literature.

m

m.gola
2011
2008
Empirical relation between static strength and fatigue limit
MPa σD-1
a
quenched and
ol
1000 tempered
steels
.g
800
m
600
mild carbon
steels
io these
400 data for
uz
bending
200 fatigue !
m
0 Rm
400 800 1200 1600 2000 MPa
600 1000 1400 1800
m.gola
2011
2008
Empirical relation between static strength and fatigue limit
MPa σD-1 austenitic stainless steels
a
martensitic stainless steels
ol
1000 ferritic stainless steels
.g
800
m
600
martensitic
io these
400 data for
uz
bending
200 fatigue !
m
0 austenitic Rm
400 800 1200 1600 2000 MPa
600 1000 1400 1800
3. Static strength values and material tables (1/11)
m.gola
2011
2008
Reliable material data must come from an accredited source !!!

Reference will here be made to those of Analytical Strength
Assessment, FKM – Forschungskuratorium Maschinenbau, 5th
a
ed., VDMA Verlag, Frankfurt am Main, 2003 (in what follows,
ol
quoted shortly as FKM) which are an ample and coherent standard
.g
Rm,N : static failure stress, i.e. , the maximum stress experienced in a tensile
test (minimum guaranteed value or else lower boundary of the tensile strength)
m
Re,N : minimum guaranteed value of yield strength
σD-1,N : fatigue limit for R=-1 (goes along with its statistical definition)
io
uz
The suffix ..N indicates that the values are given in the published tables
of the Standard (Norm).
m

m.gola
2011
2014
Rm,N : tensile strength, according to standard

- 97,5% probability
- tested at room temperature
a
- tested along the main direction of milling and forging
- in the case of steel it applies to the smallest dimension of a
ol
semi-finished product; in the case of cast iron and cast
.g
aluminium alloys to the test piece size defined in standards
Rp : yield stress, a generalization of ReH and Rp0,2
m
- the same conditions hold as above for Rm,N
σD-1,N : fatigue limit for R=-1, i.e. fully reversed stress
io
- however tested in tension-compression (FKM does not use
values from rotating bending for round specimens or
uz
from alternating bending for flat specimens); the reason for
this will be clear later
m
REMARK: the value given by FKM corresponds to 97,5% probability;

FKM safety factors are coherent with this.
m.gola
2011
2015
Material strength depends on the specimen/component size.

The material “Norm” parameters are experimentally determined
with specimens having the “effective diameter” deff,N of the semi-
a
finished product or the raw casting.
ol
Different materials have a different deff,N , specified in each FKM
.g
table.
m
FKM has formulas to modify Rm,N and Re,N to pass from deff,N , the
“Norm” (Standard) effective diameter, to the “component” effective
io
diameter deff through the technological size factor Kd,m :
uz
Rm= Kd,m KA Rm,N

m

m.gola
2011
2014
Definition of effective diameter deff Heat treatable steel

Non-alloyed struct.
steel (also forgings),
(also forgings), case
fine grained struct.
In the case of a solid round bar the Cross section hardening steel, heat
steel, norm. quench.
treatable cast steel,
and tempered steel,
diameter d is deff . GGG, GT or GG
Aluminium
a
In other cases, deff is calculated with the
ol
d d d
formulas defined in the table on the
right, which for cases in the first two
.g
cases on the right:
b b b
deff = 4 V/O, with:
m
V: specimen volume
s
O: lateral surface io 2s s
uz
In rod-shaped components made of
s 2s s
quenched and tempered steel, the
effective diameter deff is the diameter
m
existing while the treatment is 2bs

performed; if machining follows, deff is s s
b+s
the larger original diameter. b
m.gola
2011
2015
For all steels, except stainless steel, and for cast iron the technological size factor is:
-) deff ≤ deff,N ⇒ Kd,m =1 d0=7,5 mm
1−0,7686 ad,m log deff /7,5
a
-) deff,N,m < deff ≤ deff,max,m ⇒ Kd,m =
1−0,7686 ad,m log deff,N,m /7,5
ol
-) deff ≤ deff,max,m ⇒ Kd,m = Kd,m(deff,max,m) mm
(0,7686=1/log20 )
.g
In short
{ Kd,m=1 for deff ≤ deff,N
m
Kd,m<1 for deff > deff,N
io
For milled steel deff,max (m&p) =250 mm
uz
For other steels deff,max (m&p) = ∞
m
For yield strength here called Rp

replace the values Kd,m , deff,N,m , ad,m
with Kd,p , deff,N,p , ad,p
source: FKM
m.gola
2011
2014
Examples of (deff,N , ad,m ) from FKM 5th edition 2003

Non-alloy structural steels DIN EN 10025 (1994), deff,N,m=40 mm ad,m=0,15
grades S185, S235, S275, S355
a
deff,N,p=40 mm ad,p=0,3
ol
Weldable fine grain structural steels in normalised deff,N,m=70 mm ad,m=0,2
.g
condition DIN 17102 (1983) grades StE255 → StE500
Heat treatable steel in quenched and tempered deff,N,m=16 mm ad,m=0,3
m
condition DIN EN 10083-1 (1996)
grades C22 → C60, 28Mn6 → 51CrV4
io
Case hardening steels in the blank hardened cond. deff,N,m=16 mm ad,m=0,5
uz
DIN EN 10084 (1998) grades C10E →14NiCrMo13-4
m
Spheroidal graphite cast irons DIN EN 1563 (1997) deff,N,m=60 mm ad,m=0,15

grades GJS 350-22-LT → GJS 900-2

m.gola
2011
2014
In the case of wrought aluminium alloys FKM tables give

“component” values Rm and Re of the semi-finished product
indicated in the table, according to size. Example: strips & sheets,
a
extruded rods & bars, tubes & profiles of "DIN EN” norms.
ol
Norms other than FKM also for steel provide tables with size-
.g
dependent values (example, the italian standard UNI-EN 10083.1).
m
Whatever the method, the important fact is:
that these experimental material parameters are dependent on the
io
size in a way which cannot be neglected, specially on account of the
uz
fact that specimens are in the smaller size range, and strength values
decrease with increasing size.
m

m.gola
2011
2014
According to FKM, transition from values for static strength

coming from “normal” laboratory conditions, i.e. longitudinal test
specimen, to “real component” with material anisotropy is done
a
through the multiplying factor:
ol
KA = anisotropy factor , used in: Rm= Kd,m KA Rm,N
.g
which allows for the fact that the strength values of milled steel
and forgings are lower transversally to the direction of milling
m
and forging.
io
For strength in the direction of processing the factor is 1.
uz
For strength trasverse to the main processing direction:
(Rm in MPa) up to 600 from 600 to 900 from 900 to 1200 above 1200
m
Steel KA= 0,90 KA= 0,86 KA= 0,83 KA= 0,80

(Rm in MPa) up to 200 from 200 to 400 from 400 to 600
Alum. Alloys KA= 1,00 KA= 0,95 KA= 0,90 source: FKM

m.gola
2011
2014
The following tc b
Steel Type Rm,N Re,N σD−1,N σD−1,N
tables sample
some values S 235 JR 360 235 160 180
Weldable
a
relating static
fine grain S 275 JR 430 275 195 215
ol
tensile
structural S 355 JR 510 355 230 255
properties with
.g
steels * 160 180
the fatigue limit StE 255 360 255
DIN EN 10025/94
tc StE 500 610 500 275 300
m
σD− 1,N DIN 17102 (1983
tested Case 16MnCr5 1000 695 400 430

in tension-
io
hardening
compression. 20MnCr5 1200 850 480 510
uz
steels #
blank hardened 18CrMo4 1100 775 440 470
m
DIN EN 10084/98
Reduced and adapted from FKM

* deff,N = 40 mm #d
eff,N = 16 mm
m.gola
2011
2008
(continued)
tc b
Steel Type Rm,N Re,N σD− 1,N σD−1,N
Quenched and C35 630 430 285 310
a
tempered
ol
C45 700 490 315 345
steels
.g
in quenched C60 850 580 385 415
and tempered 34Cr4 900 700 405 435
m
condition
DIN EN 10083/96 34CrMo4 1000 800 450 480
io 50CrMo4 1100 900 495 525
uz
source: FKM 36NiCrMo16 1250 1050 560 595
deff,N = 16 mm
m

m.gola
2011
2012
A sample from the FKM material properties for an aluminium alloy
Alloy Type / condition Thickness tc b

from……to Rm Re σD− 1,N σD−1,N A10%
a
EN AW-2014 T3 0.4 1.5 395 245 120 140 14
Wrought
ol
AlCu4SiMg 1.5 6.0 400 245 120 140 14
aluminium
.g
T4 0.4 1.5 395 240 120 140 14
alloys
T451 1.5 6.0 395 240 120 140 14
strips
6.0 12.5 400 250 120 140 14
m
sheets
T451 12.5 40.0 400 250 120 140 10
plates io 40.0 100.0 395 250 120 140 7
………………………………………………………………………………………
DIN EN 485-2
uz
(1995) T6 0.4 1.5 440 390 130 150 6
source: FKM
T651 1.5 6.0 440 390 130 150 7
m
6.0 12.5 450 395 135 155 7

b Remark: Kd,m=1
Here we see also σD−1,N = fatigue limit in bending
A10% = elongation over a length 11,3 S0 (i.e. L=10d)
4. Fatigue limits from static properties (1/2)
m.gola
2011
2008
The fatigue limit σD-1, in fully reversed stress (R=-1), is either

presented in tables, as in the previous section, or expressed through
formulas, as it will be shown soon.
a
The second approach is particularly useful in all those cases where
ol
fatigue data are not provided by material tables, which is quite
.g
frequent (fatigue data are often proprietary data).
m
According to FKM the material fatigue strength values
presented in their own tables are intended for information only.
io
FKM requires to determine fatigue limits as described below
uz
(σD-1.. values here defined are tc, i.e. those in tension-compression):
m
tc tc tc
A) σD-1 = fW,σ Rm B) τD-1 = fW,τ σD-1
where coefficients fW,σ , fW,τ are found in the following table:
4. Fatigue limits from static properties (2/2)
m.gola
2011
2008
Type of material fW,σ ∗ fW,τ Type of material fW,σ fW,τ

GT- malleable cast
Case hardening steel 0,40 0,577# iron 0,30 0,75
a
0,577 0,85
ol
Stainless steel 0,40 GG - grey cast iron 0,30
Forging steel 0,40 0,577 Wrought alum. alloys ** 0,30 0,577
.g
Steels other than these 0,45 0,577 Cast alum. alloys **0,30 0,577
m
∗ fW,σ values for N=106 cycles
source: FKM
GS – steel castings 0,34 0,577
GGG – spheroidal io ** does not correspond to the
graphite cast iron 0,34 0,65 N=∝ limit, but to N=106
uz
Compare these values

Adjourned symbols today are: GGG ⇒ GJS
m
with Sect. 2 Sl. 2 GT ⇒ GJM

GG ⇒ GJL
# easy to check that 0,577= 1 / 3 , which is related to a Von Mises-like formula
5. Temperature factor (1/2)
m.gola
2011
2014
Fatigue limits are calculated in Sect. 4 for the material in the test
specimen form and at room (normal) temperature.
Fatigue limits are lower at increased temperature.
a
ol
The following are the ranges of “normal temperatures” where no
correction is needed (FKM):
.g
- fine grain structural steel: from: – 40°C to 60°C
- other kinds of steel: from: – 40°C to 100°C
m
- cast iron materials: from: – 25°C to 100°C
- age hardening alum. alloys: io from: – 25°C to 50°C
- non age-hard. alum. alloys: from: – 25°C to 100°C
uz
For temperatures outside the range:
m
σD-1,T= KT,D σD-1 τD-1,T = KT,D τD-1 ………

(for brevity the apex tc is here omitted)

5. Temperature factor (2/2)
m.gola
2011
2008
… … … where empirical interpolation formulas give: source: FKM
at “normal” temperature: KT,D =1
a
at low temperature: it is necessary to check specific guidelines
ol
at high temperature (T is in °C):
.g
-fine grain structural steel: KT,D= 1 - 10-3 T *
-other kinds of steel: KT,D= 1 - 1.4 10-3 (T-100)*
m
-GS: KT,D= 1 - 1.2 10-3 (T-100)*
-GGG, GT, GG: io KT,D= 1 - aT,D 10-3 (T-100)*
-aluminium alloys: KT,D= 1 – 1.2 10-3 (T-50)**
uz
m
material GGG GT GG * valid up to 500 °C

aT,D 1,6 1.3 1.0 ** valid up to 200 °C
Sections 6 to 9 - Mean stress and Haigh diagram
m.gola
2011
2012
Section 6 shows how the experimental Haigh diagram, for all lives,
can be derived from Wöhler curves.
Section 7 shows the “master diagram”, a popular diagram for design
a
purposes which accumulates fatigue data under different mean
ol
stresses and presents each line as the fatigue life under the net of
.g
maximum and minimum stresses in addition to mean stress and
alternating stress. Moreover, a “universal representation allows to
collect a family of material under one diagram.
m
Section 8 must be understood as an interpolation exercise, whith
io
under certain assumption permits to draw the Haigh diagram with
approximate curves that are convenient for later formula
uz
manipulations.
m
Section 9 the linear simplifications of the Haigh curve at infinite life, in

the way they are currently used by design standards.

6. Influence of mean stress on fatigue (1/4)
m.gola
2011
2008
When mean stress σ m is different from zero Wöhler curves will be

different from those obtained for σ m = 0 or stress ratio R =-1.
σ a σσmm,0,δ=>0σ m,γ > σ m,β > σ m,α
a
ol
σ max =,α0 = 0
σ m,ασ m
σa
.g
σm σ m,βσ m,β
∆σ
σ m,γσ m,γ
m
σ min σ m,δσ m,δ
io
0 t
uz
σ min
stress ratio R=
m
σ max 103 104 105 106

Log cycles to failure

m.gola
2011
2008
The diagrams below qualitatively The diagram on the right is

show the effect of mean stress called “Haigh” diagram.
σm on fatigue limits σD . The fatigue limit curve
a
σ a σσmm,0,δ=>0σ m,γ > σ m,β > σ m,α σ a
empirically connects
ol
fatigue limits with their
=,α0 = 0
σ m,ασ m
mean stresses
.g
σ m,βσ m,β (conventional infinite
life defined here at 106)
m
σ m,γσ m,γ
σ m,δσ m,δ io ♦ ♦σ D−1
♦ ♦σ D ,β σ
uz
♦ ♦ D ,γ
♦ ♦σ D ,δ
m
103 104 105 106 σ m,α σ m,β σ m,γ σ m,δ σm

m.gola
2011
2008
As shown below, a constant life curve is generated from a set of

Wöhler curves for different “mean stress” values.
a
σa σa
ol
σ m,α = 0
.g
σ m,β
m
σ m,γ
• •
σ m,δ •
io • N=104 curve
• •
uz
• •
m
103 104 105 106 σ m,α σ m,β σ m,γ σ m,δ σm

m.gola
2011
2008
Here the curves for three finite lives (103, 104, 105) plus the
fatigue limit (N≥106) are obtained from Wöhler curves and
represented in the Haigh diagram.
a
σa σa
ol
σ m,α = 0
• • 10
.g
3
σ m,β
• •
m
σ m,γ 104
σ m,δ • •
o
ioo 105
• o o •
uz
o o
o o
m
103 104 105 106 σ m,α σ m,β σ m,γ σ m,δ σm

7. The Haigh plot: experimental data (1/6)
m.gola
2011
2008
A Haigh diagram plots the mean stress, mostly in the tensile

range, along the x-axis and the oscillatory stress amplitude
along the y-axis. It sometimes called also “Goodman” diagram.
a
σa The next slide will
ol
show a “master
.g
diagram” i.e. a Haigh
diagram normalised
expressing them in
m
% of Rm. Why?
io Because it is expected
σD that such curves will
uz
be valid for a family of
σm materials:
m
in this case AISI 4340 ≡ 34CrNiMo6 AISI 2330 ≡ 60SiMn5

AISI 4130 ≡ 25CrMo4 AISI 8630 ≡ 30NiCrMo22
m.gola
2011
2008
a sample of steel alloy data units: non-dimensional
a
max. stress σmax in % of Rm
ol
Re
Rm
.g
m
io
uz
min. stress σmin in % of Rm

“master curve” for high
m
strength quenched alloy tension-compression fatigue

source: R.C. Juvinall, Engineering
steels, Rm in the range Considerations of Stress, Strain and Strength,
850÷1200 MPa Mc Graw-Hill, 1967

m.gola
2011
2012
a sample of aluminium alloy data units: non-dimensional
Re
Rm
a
ol
.g
σmax /Rm x100
m
io
uz
“master curve” for a 2000

m
σmin /Rm x100

series aluminium, alloy
source: R.C. Juvinall, K.M. Marshek
2023 T3; valid also for Fondamenti della progettazione dei componenti
2024 T4 and 2014 T6 delle macchine – Edizioni ETS, 1993, pag. 285

m.gola
2011
2012
Per garantire la privacy, il download automatico di questa immagine è stato bloccato da PowerPoint.
Re
a
ol
.g
σmax /Rm x100
m
io
uz
m
1
(σm − σ a ) 1 ⋅ 100 = σmin 1 ⋅ 100
Rm 2 Rm 2
changing scale on the horizontal axis read σmin/Rm
m.gola
2011
2015
This will be our preferred version

of the Haigh diagram
a
ol
stress ampl. σa in % of Rm
.g
m
io
uz
m
mean stress σm in % of Rm

m.gola
2011
2008
Some data for your reference
AISI 4340 or DIN 34NiCrMo6 2024 T3 or DIN AlCuMg2

C 0.38 - 0.43 % mechanical Al 90.7 - 94.7 % mechanical
a
Cr 0.4 - 0.6 % properties when Cr max 0.1 % properties:
ol
Mn 0.6 - 0.8 % normalized at Cu 3.8 - 4.9 %
Mo 0.2 - 0.3 % 870°C, measured Fe max 0.5 % E (GPa): 73
.g
Ni 1.65 – 2 % at T=25° C: Mg 1.2 - 1.8 % Rm (MPa): 480
Mn 0.3 - 0.9 % Re (MPa): 345
heat treatable, E (GPa): 190-210 Si max 0.5 % A (%): 18
m
low alloy steel Rm (MPa): 1275 Ti max 0.15 %
known for its Re (MPa):
io 860 Zn max 0.25 %
toughness and A (%): 22
capability of aircraft fittings,
uz
developing high gears and shafts,
strength in the bolts, clock parts,
heat treated computer parts,
m
condition while couplings, fuse

retaining good parts, hydraulic
fatigue strength valve bodies
8. The Haigh plot: data approximation (1/9)
m.gola
2011
2008
A very substantial amount of testing is required to generate a

Haigh diagram, and it is usually impractical to develop curves
for all combinations of mean and alternating stresses. Several
a
empirical relationships that relate alternating stress to mean
ol
stress have been developed to address this difficulty.
The two most widely accepted methods are those of Goodman
.g
and Gerber. Experience has shown that experimental points fall
between the Goodman and Gerber curves.
m
Goodman σD σm Goodman straight line is often used
+ = 1 due to mathematical simplicity and
England, 1899
io
σD −1 R m slightly more conservative values.
uz
2
Gerber σD σ 
Germany, 1874 +  m  = 1
σD −1  R m 
m
Soderberg σD σm
+ = 1 Soderberg is over-conservative.
USA, 1930 σD −1 R e
m.gola
2011
2012
A graphical comparison of these equations is shown below.

All methods should only be used for tensile mean stress values.
For cases where the mean stress is small relative to the
a
alternating stress (R << 1), there is little difference in the
ol
methods. The Soderberg method is conservative in excess.
.g
σa As R approaches 1,
(σa ⇒0) static limit
m
to yield predominates.
io This typically the
operation area of
uz
cyclically loaded high
strength bolts and of
m
vibrating turbine or
σm compressor blades.
Re Rm
m.gola
2011
2008
When all the necessary Wöhler curves are not available to deduce
the complete Haigh diagram as it was shown in Sect. 6 - sl. 2, 3, 4
the problem is to find a rule: 1) to interpolate data or 2) to
a
reproduce curves from a minimum set of data.
ol
In the next slides the second case is presented through a worked
out example.
.g
m
Please keep in mind that this reproduction is an assumption which is
justified on the basis of its ability to predict the behaviour of real
io
data to a certain degree of acceptable approximation. As such, the
method needs a confirmation based on experimental validation,
uz
then it must be used with caution.
m
Another thing to underline is that the particular choice of formulas

may be of advantage for data treatment (sl. 9 of this section and its
later consequences).
m.gola
2011
2012
It may be expedient to interpolate first the Haigh curve for the limit
life taken at N=106 cycles (however, some standards do it for 2*106
or 5*106). In the next slide we see a parabola through three points,
a
chosen at σm/Rm•100 = 0, 40, 100).
ol
The linear interpolation will be introduced later.
.g
For different lives the power law σ ka N = const is used to interpolate
the Wöhler curve at σm=0 , with the exponent k chosen so to pass
m
through the two points at lives 106 and 104
2
(in the case of this example we find: k =
io ≅10 )
Log 71,3 − Log 45
The through data are the five numbers in yellow boxes seen in the next two
uz
slides; the mathematics of interpolation is quite trivial and is omitted
m
Finally, the power law with the exponent determined as above is

used to generate Woehler curves for all other σm . This is going to
have useful consequences for data manipulation.
m.gola
2011
2012
100 100
90
σa/Rm σa/Rm
90
80 80
k=10
a
70 70
σm 60 60 N
ol
0 50 50 10^6
40 40
.g
30 30
20 20
10 10
σm/Rm
m
0 0
power law interpolation
1,E+03 1,E+04 1,E+05 1,E+06 1,E+07 0 10 20 30 40 50 60 70 80 90 100
102 …100 k 90 80 70 60 50 40 30 20 10 0
… N = const
σ a83,3
103 89,8 76,3
io 68,7 60,5 51,8 42,6 32,7 22,4 11,5 0,0
Life -cycles
uz
104 71,3 66,2 60,6 54,5 48,1 41,2 33,8 26,0 17,8 9,1 0,0
105 …
56,7 52,6 48,1 43,3 38,2 32,7 26,9 20,7 14,1 7,2 0,0
m
106 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
107 …
45,0 …
41,7 …
38,2 …
34,4 …
30,3 …
26,0 …
21,3 …
16,4 …
11,2 …
5,7 …
0,0
σm/Rm= 0 10 20 30 40 50 60 70 80 90 100
parabolic interpolation
m.gola
2011
2012
… 2 – Values for σa,i at all 1 - Values for σD= f (σm) at

Ni for σm=10% of Rm are n=106 are interpolated through
obtained through: a parabola …
a
1 / 10
σ a,i  10 6 
ol
= 41,7 
Rm  3 - These σa are taken
 Ni 
.g
etc. for all σm, one k for all equal to Rm- σm, i.e.
now 100-σm/Rm
m
power law interpolation
102 100 90 80
io 70 60 50 40 30 20 10 0
103 89,8 83,3 76,3 68,7 60,5 51,8 42,6 32,7 22,4 11,5 0,0
Life -cycles
uz
104 71,3 66,2 60,6 54,5 48,1 41,2 33,8 26,0 17,8 9,1 0,0
105 56,7 52,6 48,1 43,3 38,2 32,7 26,9 20,7 14,1 7,2 0,0
m
106 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
107 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
σm/Rm= 0 10 20 30 40 50 60 70 80 90 100
parabolic interpolation
m.gola
2011
2012
100 100
90
σa/Rm 90
σa/Rm
σm 80 80
0 k=10 N
a
70 70
10 10^3
60 60
20
ol
10^4
50 50
30 10^5
40 40
40 10^6
σm
.g
50 30 30
20 20
10 10
σm/Rm
m
0 0
1,E+03 1,E+04 1,E+05 1,E+06 1,E+07 0 10 20 30 40 50 60 70 80 90 100
102 100 90 80
io 70 60 50 40 30 20 10 0
103 89,8 83,3 76,3 68,7 60,5 51,8 42,6 32,7 22,4 11,5 0,0
Life -cycles
uz
104 71,3 66,2 60,6 54,5 48,1 41,2 33,8 26,0 17,8 9,1 0,0
105 56,7 52,6 48,1 43,3 38,2 32,7 26,9 20,7 14,1 7,2 0,0
m
106 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
107 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
σm/Rm= 0 10 20 30 40 50 60 70 80 90 100

m.gola
2011
2014
100 100
90
σa/Rm 90
σa/Rm
σm 80 80
0 k=10 N
a
70 70
10 10^3
60 60
20
ol
10^4
50 50
30 10^5
40 40
40 10^6
σm
.g
50 30 30
20 20
10 10
σm/Rm
m
0 0
1,E+03 1,E+04 1,E+05 1,E+06 1,E+07 0 10 20 30 40 50 60 70 80 90 100
102 100 90 80
io 70 60 50 40 30 20 10 0
103 89,8 83,3 76,3 68,7 60,5 51,8 42,6 32,7 22,4 11,5 0,0
exercise, find an explanation for the values above at life N=102
Life -cycles
uz
104 71,3 66,2 60,6 54,5 48,1 41,2 33,8 26,0 17,8 9,1 0,0
cycles , and for the values below at life N=107
105 56,7 52,6 48,1 43,3 38,2 32,7 26,9 20,7 14,1 7,2 0,0
check the curve for 103
cycles … any problem? all is right? sure?
m
106 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
107 45,0 41,7 38,2 34,4 30,3 26,0 21,3 16,4 11,2 5,7 0,0
σm/Rm= 0 10 20 30 40 50 60 70 80 90 100

m.gola
2011
2015
The Haigh curve for fatigue 100

σa/Rm
limit is: σD= σD(σm)
90 ~
σa / R m
80
Omitting reference to Rm, 70

σD −1 / R m
a
power law interpolation with
60
N
ol
σa / R m
50
the same value of k for all

10^6
40
σD / R m
values of σm produces:
.g
30
[σ (σσ N)] =N[σ= [(σσ (σ)]

20
k
a
k
a
m D Dm
k
m )
10] k6 6
10 10
σm/Rm
m
0
This means that independent 0 10 20

σm40
30 50 60 70 80 90 100
on the value σm:

The consequence is:
σ a (σ m )  10 6 
1/k
io
=  ~ σ a (σ m )
uz
σ D (σ m )  N  σa
= for any N
or, any Haigh curve (σa,σm) for σ D −1 σ D (σ m )
m
a given life N is proportional to

~
the curve (σD,σm) i.e. to the Later we shall s a = σ a / R m ⇒ sa = s a (s m )

also write: s m = σ m / R m s D −1 s D (s m )
one for life 106 .
9. The Haigh plot: linear simplifications (1/4)
m.gola
2011
2008
σa/Rm
100 100
90
σa/Rm
90
σm 80 80
0 k=10 N
a
70 70
10 10^3
60 60
20
ol
10^4
50 50
30 10^5
40 40
40 10^6
.g
50 30 30
20 20
10 10
σm/Rm
m
0 0
1,E+03 1,E+04 1,E+05 1,E+06 1,E+07 0 10 20 30 40 50 60 70 80 90 100
In many practical instances, the curves in the Haigh plot are

io
further simplified as straight lines, as indicated above.
uz
Only three values are needed: σD-1, Rm plus one point on the
106 Wöhler curve, or in alternative the value of exponent k.
m
 σm 
The equation for “Goodman” line σD is: σD = σD −1 1 − 
 Rm 
m.gola
2011
2008
A realistic plot of values for the empirical function σD=f (σm) is

shown below with the Goodman line; also negative σm have been
represented, where the Goodman line does not hold.
a
σa
ol
either
.g
σD σ
+ m =1
• • • •
σD −1 R m
m
σD-1
• • or
σD io • • σD +
σD −1
σ m = σD −1
σD • Rm
•
uz
m
Rm
σm

m.gola
2011
2008
FKM refinement The alternative, less restrictive,

simplified representation
according to FKM is based
a
σa (within the two limits R=-∞ and
ol
R=0) on the equation:
.g
σ D + Mσ σ m = σ D −1
• • where:
• • σ D −1
m
Mσ = (α > 1)
σD-1• • •σD αR m
io •
•
•
uz
m
Rm
• σm

m.gola
2011
2012
The values of Mσ for different materials are: Note: Rm is here the

component tensile
Mσ = aM 10 −3 R m + bM (R m in MPa) strength as defined in
Ch.2 Sect.3 sl.3
a
material steel* GS GGG GT GG wrought cast
ol
Al-alloys Al-alloys
.g
Source FKM
aM 0,35 0,35 0,35 0,35 0 1.0 1.0
bM -0.1 0.05 0.08 0.13 0.5 -0.04 0.2
m
*including stainless steel
io
Example for steel: Rm= 1000 MPa ⇒ Mσ=0,25
0.45 R m 0.45
uz
σ
Mσ = 0.35 − 0.1 = 0.25 = D −1 = ⇒α= = 1.8
α Rm α Rm 0.25
m
According to FKM, in the case of torsion: Mτ = fW ,τ Mσ , where

values fW,τ are in the table of section 4 slide 2 of this chapter.
Sections 10, 11, 12, 13 - Infinite life
m.gola
2011
2012
These Sections elaborate on the Haigh diagram in the case of one-

dimensional tensile (and compressive) stresses. Two-and-three-
dimensional stresses will be seen in Chapter 3.
a
Section 10 shows how the Haigh linearised plot for infinite life limit (for
ol
fatigue) and the yield limit (for maximum or peak load) can be plotted
.g
on the same diagram. Then how from limit curves a limit area can be
defined, and then an admissible area.
m
Section 11 illustrated the relation between the Haigh diagram and
other equivalent graphical representations which are frequently used
io
as well.
uz
Sections 12 and 13 show how the limit and admissible areas are
defined by the FKM standard, and may be used to assess the capability
m
to withstand fatigue stresses at infinite life and peak stresses at yield.

10. The Haigh plot: the limit Goodman area (1/3)
m.gola
2011
2008
As we have already seen before, the most conservative simplified

representation of σD values is given by the red lines (Goodman):
a
σa
ol
either
.g
σD σ
+ m =1
• • • •
500 σD −1 R m
m
σD-1•
or
• •
io • σD +
σD −1
σ m = σD −1
σD • Rm
•
uz
m
Rm
• σm
500

m.gola
2011
2008
The blue lines represents the yield condition: σ a + σ m = R e

σ a σm
+ =1
Re Re
a
σa
ol
Re σ
Re
σa
.g
σm
t
• • • •
500
m
σm
• • σa
io • • -Re
•
•
uz
m
Re Rm
• σm
500

m.gola
2011
2008
The simplified representation allows a conservative estimate for

design, without the great expense of a full set of experimentally
determined fatigue limits.
The combined “limit”
a
σa
conditions for strength:
ol
• eR
σ a ≤ σD
.g
σ a + σm ≤ R e
• • • 500 define the “limit” area
m
•
σD-1• •
in which fall points
io • • representing applied
σD • stresses.
σ a ≤ σD •
uz
σ max ≤ R e
m
Re Rm
• • σm
500

11. Haigh, Goodman-Smith and Moore (1/4)
m.gola
2011
2008
σa
• Re Haigh representation of material
fatigue is flanked by others,
E
•
σD-1
A
which are in use; it is necessary
a
X
C
to understand how they relate.
ol
σa
D B σ m σ max = σ a + σ m = XY + YZ
• •
.g
Y
Re Rm
σ max B Haigh diagram
m
σm on abscissas
C
Re• io •D σa on ordinates
X
σD-1•
uz σa
A
Goodman-Smith
Y 45°
E Diagram
m σm
σm σm on abscissas
Z
• σmax on ordinates
Re Rm
m.gola
2011
2014
Goodman-Smith diagram Moore

(positive σm) diagram
σ max σ max
σ min σ max
a
Re • Re •
ol
X
.g
σa
σD-1 • σD-1•
Y
σa
m
Z
io σm R
W
Re Rm -1 0 1
σ min σ max = σ m + σ a
uz
1) R = ;
σ min = σ m − σ a σ max σ min = R σ max = σ m − σ a
W Z = WY − Y Z 1+R 1−R
m
2) σ m = σ max ; σ a = σ max
2 2
σa σ σ 1 − R 1 + R 
3) + m = 1 ⇒ max  + =1
σ D −1 Rm 2  σ D −1 R m 
m.gola
2011
2008
The Goodman-Smith diagram lends itself to an easy

visualisation of stresses in time:
a
σ max
ol
σ min σ max
Re •
.g
X σ max
σa
σD-1 • t σ
m
Y m
σa
Z
io σ min
σm
uz
W
Re Rm
σ min = σ m − σ a
m
W Z = WY − Y Z

m.gola
2011
2008
The Moore diagram puts

σmax into evidence the
maximum stresses, which
Rm
a
are the main worry in
ol
design against static
Re •
failure, and takes fatigue
.g
into account by means of
σD−1 • the stress ratio R.
m
io
R
uz
-1 0 1
m
σa σ σ 1 − R 1 + R 
+ m = 1 ⇒ max  +  =1
σ D −1 R m 2  σ D −1 Rm 

12. FKM refinement: the limit area (1/2)
m.gola
2011
2008
If, in alternative to the Goodman line, the FKM line is used (orange
line, valid only within the two limits of the stress ratio R=-∞ and
R=0), and simultaneously we consider also the yield limits
a
σa (blue lines), then … … …
ol
• Re
σ
.g
R =0
• • • • t
m
R = −∞
0
σD-1• • •σD
io •
•
uz
•
m
Re Rm
• • σm
12. FKM refinement: the limit area (2/2)
m.gola
2011
2008
… then we get the FKM limit area. FKM description is slightly

more complex than here represented, but for the purpose of
these lessons we shall be satisfied. The limit area is hatched in
a
green: σa
ol
• Re
.g
• • • •
m
σD-1• • •
io •
•
uz
σ a ≤ σD •
σ max ≤ R e
m
Re Rm
• •
σm
13. Use of the Haigh diagram: the admissible area
m.gola
2011
2008
Once the “limit” Haigh thresholds and area have been decided
according to a standard, for a given material and for a certain
component, appropriate safety limits are applied (we shall
a
σa define them later),
ol
Re producing the admissible”
• area hatched in red.
.g
Applied stress ⊗satisfies
• • • •
500
m
design requirements when
σD-1• • • it falls inside that area.
io •
•
uz
•
⊗
m
Re Rm
• •
500 σm

2 Fatigue Chapter 2 v. 2016.03 III - Lecture Notes

Încărcat de

Informații document

Titlu original

Drepturi de autor

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

2 Fatigue Chapter 2 v. 2016.03 III - Lecture Notes

Încărcat de

Drepturi de autor:

Muzio M.

7. The Haigh plot : diagram: the admissible

The Stress-Life method is the classical method for fatigue analysis

Version Muzio M. Gola - DIMEAS 3

Diameter of round specimens: steel 6÷ 8 mm

Version Muzio M. Gola - DIMEAS 4

Experimental determination The basic procedures of

Data are used to obtain

% survival probability curves

The fatigue limit can be determined with the “staircase”

sequence of tested specimens

Version Muzio M. Gola - DIMEAS 7

Through the appropriate statistical

σD at 50% x: broken specimens

Version Muzio M. Gola - DIMEAS 8

life probability distribution

Wöhler curve can be divided in three parts

Certain materials have a endurance limit which represents a

Version Muzio M. Gola - DIMEAS 11

Many non-ferrous metals and alloys, such as aluminum,

1x108 or 5x108 loading

Version Muzio M. Gola - DIMEAS 12

The main points to be remembered about the FKM-Guideline named

• for components with welded

Version Muzio M. Gola - DIMEAS 13

The purpose of Sections 2 to 5 of this chapter is the investigation of

Experimental observation: the fatigue limit is in a certain relation

Next slides will show examples from the (ancient) literature.

Version Muzio M. Gola - DIMEAS 15

Empirical relation between static strength and fatigue limit

Empirical relation between static strength and fatigue limit

MPa σD-1 austenitic stainless steels

Reliable material data must come from an accredited source !!!

Version Muzio M. Gola - DIMEAS 18

Rm,N : tensile strength, according to standard

REMARK: the value given by FKM corresponds to 97,5% probability;

Material strength depends on the specimen/component size.

Rm= Kd,m KA Rm,N

Version Muzio M. Gola - DIMEAS 20

Definition of effective diameter deff Heat treatable steel

existing while the treatment is 2bs

1−0,7686 ad,m log deff /7,5

{ Kd,m=1 for deff ≤ deff,N

For yield strength here called Rp

Examples of (deff,N , ad,m ) from FKM 5th edition 2003

Heat treatable steel in quenched and tempered deff,N,m=16 mm ad,m=0,3

Spheroidal graphite cast irons DIN EN 1563 (1997) deff,N,m=60 mm ad,m=0,15

Version Muzio M. Gola - DIMEAS 23

In the case of wrought aluminium alloys FKM tables give

Version Muzio M. Gola - DIMEAS 24

According to FKM, transition from values for static strength

Steel KA= 0,90 KA= 0,86 KA= 0,83 KA= 0,80

Version Muzio M. Gola - DIMEAS 25

tested Case 16MnCr5 1000 695 400 430

Reduced and adapted from FKM

Quenched and C35 630 430 285 310

Version Muzio M. Gola - DIMEAS 27

A sample from the FKM material properties for an aluminium alloy

Alloy Type / condition Thickness tc b

6.0 12.5 450 395 135 155 7

The fatigue limit σD-1, in fully reversed stress (R=-1), is either

Type of material fW,σ ∗ fW,τ Type of material fW,σ fW,τ

Compare these values

with Sect. 2 Sl. 2 GT ⇒ GJM

σD-1,T= KT,D σD-1 τD-1,T = KT,D τD-1 ………

Version Muzio M. Gola - DIMEAS 31