Mechanical Design Applications
Dr. M. O’Malley– Master Notes
Spring 2008
Dr. D. M. McStravick
Rice University
Reading
Chapter 6
Homework
HW 4 available, due 27
Tests
Fundamentals Exam will be in class on 221
Nature of fatigue failure
Starts with a crack
Usually at a stress concentration
Crack propagates until the material fractures
suddenly
Fatigue failure is typically sudden and
complete, and doesn‟t give warning
Fatigue Failure Examples
Various Fatigue Crack Surfaces [Text fig. 62]
Bolt Fatigue Failure [Text fig. 61]
Drive Shaft [Text fig. 63]
AISI 8640 Pin [Text fig. 64]
Steam Hammer Piston Rod [Text fig. 66]
Jacob Neu chair failure (in this classroom)
Fatigue Example 1
Fatigue Failure Example
Fatigue Failure Example
Fatigue Failure Example
Stamping Fatigue Failure Example
Schematic of Various Fatigue Failure
Jim Neu Chair Failure (Pedestal)
Fatigue Failure of Chair Shaft
Seat Fatigue Failure
Fatigue
Fatigue strength and endurance limit
Estimating FS and EL
Modifying factors
Thus far we‟ve studied static failure of machine elements
The second major class of component failure is due to dynamic loading
Variable stresses
Repeated stresses
Alternating stresses
Fluctuating stresses
The ultimate strength of a material (S
u
) is the maximum stress a
material can sustain before failure assuming the load is applied only
once and held
A material can also fail by being loaded repeatedly to a stress level that
is LESS than S
u
Fatigue failure
More Fatigue Failure Examples (ASM)
More Fatigue Failure Examples (ASM)
More Fatigue Failure Examples
Approach to fatigue failure in analysis and
design
Fatiguelife methods (63 to 66)
Stress Life Method (Used in this course)
Strain Life Method
Linear Elastic Fracture Mechanics Method
Stresslife method (rest of chapter 6)
Addresses high cycle Fatigue (>10
3
) Well
Not Accurate for Low Cycle Fatigue (<10
3
)
The 3 major methods
Stresslife
Based on stress levels only
Least accurate for lowcycle fatigue
Most traditional
Easiest to implement
Ample supporting data
Represents highcycle applications adequately
Strainlife
More detailed analysis of plastic deformation at localized regions
Good for lowcycle fatigue applications
Some uncertainties exist in the results
Linearelastic fracture mechanics
Assumes crack is already present and detected
Predicts crack growth with respect to stress intensity
Practical when applied to large structures in conjunction with computer
codes and periodic inspection
Fatigue analysis
2 primary classifications of
fatigue
Alternating – no DC component
Fluctuating – nonzero DC
component
Analysis of alternating stresses
As the number of cycles
increases, the fatigue strength
S
f
(the point of failure due to
fatigue loading) decreases
For steel and titanium, this
fatigue strength is never less
than the endurance limit, S
e
Our design criteria is:
As the number of cycles
approaches infinity (N ∞),
S
f
(N) = S
e
(for iron or Steel)
a
f
N S
o
q
'
=
) (
Method of calculating fatigue strength
Seems like we should be able to use graphs
like this to calculate our fatigue strength if
we know the material and the number of
cycles
We could use our factor of safety equation
as our design equation
But there are a couple of problems with this
approach
SN information is difficult to obtain and thus is
much more scarce than o÷c information
SN diagram is created for a lab specimen
Smooth
Circular
Ideal conditions
Therefore, we need analytical methods for
estimating S
f
(N) and S
e
a
f
N S
o
q
'
=
) (
Terminology and notation
Infinite life versus finite life
Infinite life
Implies N ∞
Use endurance limit (S
e
) of material
Lowest value for strength
Finite life
Implies we know a value of N (number of cycles)
Use fatigue strength (S
f
) of the material (higher than S
e
)
Prime („) versus no prime
Strength variable with a „ (S
e
‟)
Implies that the value of that strength (endurance limit) applies to a LAB SPECIMEN in
controlled conditions
Variables without a „ (S
e
, S
f
)
Implies that the value of that strength applies to an actual case
First we find the prime value for our situation (S
e
‟)
Then we will modify this value to account for differences between a lab specimen and our
actual situation
This will give us S
e
(depending on whether we are considering infinite life or finite life)
Note that our design equation uses S
f
, so we won‟t be able to account for safety factors until
we have calculated S
e
‟ and S
e
a
f
N S
o
q
'
=
) (
a
e
S
o
q
'
=
Estimating S
e
’ – Steel and Iron
For steels and irons, we can estimate the
endurance limit (S
e
‟) based on the ultimate
strength of the material (S
ut
)
Steel
S
e
‟ = 0.5 S
ut
for S
ut
< 200 ksi (1400 MPa)
= 100 ksi (700 MPa) for all other values of S
ut
Iron
S
e
‟ = 0.4(min S
ut
)f/ gray cast Iron S
ut
<60 ksi(400MPa)
= 24 ksi (160 MPa) for all other values of S
ut
Note: ASTM # for gray cast iron is the min S
ut
SN Plot with Endurance Limit
a
e
S
o
q
'
=
a
f
N S
o
q
'
=
) (
a
e
S
o
q
'
=
Estimating S
e
’ – Aluminum and Copper
Alloys
For aluminum and copper alloys, there is no endurance limit
Eventually, these materials will fail due to repeated loading
To come up with an “equivalent” endurance limit, designers
typically use the value of the fatigue strength (S
f
‟) at 10
8
cycles
Aluminum alloys
S
e
‟ (S
f
at 10
8
cycles) = 0.4 S
ut
for S
ut
< 48 ksi (330 MPa)
= 19 ksi (130 MPa) for all other values of S
ut
Copper alloys
S
e
‟ (S
f
at 10
8
cycles) = 0.4 S
ut
for S
ut
< 35 ksi (250 MPa)
= 14 ksi (100 MPa) for all other values of S
ut
Constructing an estimated SN diagram
Note that S
e
‟ is going to be our
material strength due to “infinite”
loading
We can estimate an SN diagram
and see the difference in fatigue
strength after repeated loading
For steel and iron, note that the
fatigue strength (S‟
f
) is never less
than the endurance limit (S
e
‟)
For aluminum and copper, note
that the fatigue strength (S‟
f
)
eventually goes to zero (failure!),
but we will use the value of S‟
f
at
10
8
cycles as our endurance limit
(S
e
‟) for these materials
Estimating the value of S
f
When we are studying a case of
fatigue with a known number of cycles
(N), we need to calculate the fatigue
strength (S‟
f
)
We have two SN diagrams
One for steel and iron
One for aluminum and copper
We will use these diagrams to come
up with equations for calculating S‟
f
for a known number of cycles
Note: Book indicates that 0.9 is not
actually a constant, and uses the
variable f to donate this multiplier.
We will in general use 0.9 [so f=0.9]
Estimating S
f
(N)
For steel and iron
For f=0.9
For aluminum and copper
( )
( ) ( ) b S a
S
S
b
aN N S
ut
e
ut
b
f
3 9 . 0 log log
9 . 0
log
3
1
'
÷ =


.

\

'
÷ =
=
( )
( ) ( ) b S a
S
S
b
aN N S
ut
e
ut
b
f
3 9 . 0 log log
9 . 0
log
3
1
'
÷ =


.

\

'
÷ =
=
For 10
3
< N < 10
6
For N < 10
8
Where S
e
‟ is the value of
S‟
f
at N = 10
8
5.7
Correction factors
Now we have S
e
‟ (infinite life)
We need to account for differences between the lab specimen and a real
specimen (material, manufacturing, environment, design)
We use correction factors
Strength reduction factors
Marin modification factors
These will account for differences between an ideal lab specimen and real life
S
e
= k
a
k
b
k
c
k
d
k
e
k
f
Se‟
k
a
– surface factor
k
b
– size factor
k
c
– load factor
k
d
– temperature factor
k
e
– reliability factor
K
f
– miscellaneouseffects factor
Modification factors have been found empirically and are described in section 69 of
ShigleyMischkeBudynas (see examples)
If calculating fatigue strength for finite life, (S
f
), use equations on previous slide
Endurance limit modifying factors
Surface (k
a
)
Accounts for different surface finishes
Ground, machined, colddrawn, hotrolled, asforged
Size (k
b
)
Different factors depending on loading
Bending and torsion (see pg. 280)
Axial (k
b
= 1)
Loading (k
c
)
Endurance limits differ with S
ut
based on fatigue loading (bending, axial, torsion)
Temperature (k
d
)
Accounts for effects of operating temperature (Not significant factor for T<250 C [482 F])
Reliability (k
e
)
Accounts for scatter of data from actual test results (note k
e
=1 gives only a 50% reliability)
Miscellaneouseffects (k
f
)
Accounts for reduction in endurance limit due to all other effects
Reminder that these must be accounted for
Residual stresses
Corrosion
etc
Surface Finish Effect on S
e
Temperature Effect on S
e
Reliability Factor, k
e
Steel Endurance Limit vs. Tensile Strength
Compressive Residual Stresses
Now what?
Now that we know the strength of our part under
nonlaboratory conditions…
… how do we use it?
Choose a failure criterion
Predict failure
Part will fail if:
o‟ > S
f
(N)
Factor of safety or Life of the part:
q = S
f
(N) / o‟
Where
b =  1/3 log (0.9 S
ut
/ S
e
) log (a) = log (0.9 S
ut
)  3b
b
a
N
1

.

\

'
=
o
Example Homework Problem 69
A solid rod cantilevered at one end. The rod is
0.8 m long and supports a completely
reversing transverse load at the other end of +/
1 kN. The material is AISI 1045 hotrolled
steel. If the rod must support this load for 10
4
cycles with a factor of safety of 1.5, what
dimension should the square cross section
have? Neglect any stress concentrations at the
support end and assume f= 0.9.
Solution:  See Board Work
Stress concentration (SC) and fatigue failure
Unlike with static loading, both ductile and
brittle materials are significantly affected by
stress concentrations for repeated loading
cases
We use stress concentration factors to modify
the nominal stress
SC factor is different for ductile and brittle
materials
SC factor – fatigue
o = k
f
o
nom+
=
k
f
o
o
t = k
fs
t
nom
= k
fs
t
o
k
f
is a reduced value of k
T
and o
o
is the nominal
stress.
k
f
called fatigue stress concentration factor
Why reduced? Some materials are not fully sensitive
to the presence of notches (SC‟s) therefore,
depending on the material, we reduce the effect of
the SC
Fatigue SC factor
k
f
= [1 + q(k
t
– 1)]
k
fs
= [1 + q
shear
(k
ts
– 1)]
k
t
or k
ts
and nominal stresses
Table A15 & 16 (pages 10061013 in Appendix)
q and q
shear
Notch sensitivity factor
Find using figures 620 and 621 in book (Shigley) for steels
and aluminums
Use q = 0.20 for cast iron
Brittle materials have low sensitivity to notches
As k
f
approaches k
t
, q increasing (sensitivity to notches, SC‟s)
If k
f
~ 1, insensitive (q = 0)
Property of the material
Example
AISI 1020 asrolled steel
Machined finish
Find F
max
for:
q = 1.8
Infinite life
Design Equation:
q = S
e
/ o‟
S
e
because infinite life
Example, cont.
q = S
e
/ o‟
What do we need?
S
e
o‟
Considerations?
Infinite life, steel
Modification factors
Stress concentration (hole)
Find o‟
nom
(without SC)
( ) ( )
F
F
h d b
P
A
P
nom
2083
10 12 60
=
÷
=
÷
= = ' o
Example, cont.
Now add SC factor:
From Fig. 620,
r = 6 mm
S
ut
= 448 MPa = 65.0 ksi
q ~ 0.8
( )  
nom t nom f
k q k o o o ' ÷ + = ' = ' 1 1
Example, cont.
From Fig. A151,
Unloaded hole
d/b = 12/60 = 0.2
k
t
~ 2.5
q = 0.8
k
t
= 2.5
o‟
nom
= 2083 F
( )  
( )   ( )
( ) F
F
k q
nom t
4583
2083 1 5 . 2 8 . 0 1
1 1
=
'
÷ + =
'
'
÷ + =
'
o
o
o o
Example, cont.
Now, estimate S
e
Steel:
S
e
‟ = 0.5 S
ut
for S
ut
< 1400 MPa (eqn. 68)
700 MPa else
AISI 1020 Asrolled
S
ut
= 448 MPa
S
e
‟ = 0.50(448) = 224 MPa
Constructing an estimated SN diagram
Note that S
e
‟ is going to be our
material strength due to “infinite”
loading
We can estimate an SN diagram
and see the difference in fatigue
strength after repeated loading
For steel and iron, note that the
fatigue strength (S‟
f
) is never less
than the endurance limit (S‟
e
)
For aluminum and copper, note
that the fatigue strength (S‟
f
)
eventually goes to zero (failure!),
but we will use the value of S‟
f
at
10
8
cycles as our endurance limit
(S‟
e
) for these materials
Correction factors
Now we have S
e
‟ (infinite life)
We need to account for differences between the lab specimen and a real
specimen (material, manufacturing, environment, design)
We use correction factors
Strength reduction factors
Marin modification factors
These will account for differences between an ideal lab specimen and real life
S
e
= k
a
k
b
k
c
k
d
k
e
k
f
Se‟
k
a
– surface factor
k
b
– size factor
k
c
– load factor
k
d
– temperature factor
k
e
– reliability factor
K
f
– miscellaneouseffects factor
Modification factors have been found empirically and are described in section 69 of
ShigleyMischkeBudynas (see examples)
If calculating fatigue strength for finite life, (S
f
), use equations on previous slide
Example, cont.
Modification factors
Surface: k
a
= aS
ut
b
(Eq. 619)
a and b from Table 62
Machined
k
a
= (4.45)(448)
0.265
= 0.88
Example, cont.
Size: k
b
Axial loading
k
b
= 1 (Eq. 621)
Load: k
c
Axial loading
k
c
= 0.85 (Eq. 626)
Example, cont.
Temperature:
k
d
= 1 (no info given)
Reliability:
k
e
= 1 (no info given)
Miscellaneous:
k
f
= 1
Endurance limit:
S
e
= k
a
k
b
k
c
k
d
k
e
k
f
Se‟ = (0.88)(0.85)(227) = 177 MPa
Design Equation:
( )
( )
kN 4 . 21
8 . 1 4583
10 x 177
8 . 1
4583
177
6
= =
= =
'
=
F
F
MPa S
e
o
q
Fluctuating Fatigue Failures
Alternating vs. fluctuating
Alternating
Fluctuating
I
Mr
A
P
a
m
=
=
o
o
Alternating Stresses
o
a
characterizes alternating stress
Fluctuating stresses
Mean Stress
Stress amplitude
Together, o
m
and o
a
characterize fluctuating
stress
2
'
min max
o o
o
+
=
m
2
min max
'
o o
o
÷
=
a
Alternating vs. Fluctuating
Modified Goodman Diagram
Fluctuating Stresses in Compression and
Tension
Failure criterion for fluctuating loading
Soderberg
Modified Goodman
Gerber
ASMEelliptic
Yielding
Points above the line: failure
Book uses Goodman primarily
Straight line, therefore easy algebra
Easily graphed, every time, for every problem
Reveals subtleties of insight into fatigue problems
Answers can be scaled from the diagrams as a check on the
algebra
Gerber Langer Plot for Fluctuating Stresses
Fluctuating stresses, cont.
As with alternating stresses, fluctuating stresses have been
investigated in an empirical manner
For o
m
< 0 (compressive mean stress)
o
a
> S
f
Failure
Same as with alternating stresses
Or,
Static Failure
For o
m
> 0 (tensile mean stress)
Modified Goodman criteria
q < 1 Failure
) S (or
max uc yc a m
S > ÷ = o o o
q
o o 1
= +
ut
m
f
a
S S
Modified Goodman Langer Equations
Fluctuating stresses, cont.
Relationship is easily
seen by plotting:
Goodman Line
(safe stress line)
Safe design region
(for arbitrary fluctuations
in o
m
and o
a
)
1 = +
ut
m
f
a
S S
o o
q
o o 1
= +
ut
m
f
a
S S
Note: o
m
+ o
a
= o
max
o
m
+ o
a
> S
yt
(static failure by yielding)
Important point: Part can fail because of fluctuations in either o
a
, o
m
, or both.
Design for prescribed variations in o
a
and o
m
to get a more exact solution.
Special cases of fluctuating stresses
Case 1: o
m
fixed
Case 2: o
a
fixed
a
a
S
o
q =
m
m
S
o
q =
Special cases of fluctuating stresses
Case 3: o
a
/ o
m
fixed
Case 4: both vary arbitrarily
m
m
a
a
S S
o o
q = =
ut
m
f
a
S S
o o
q
+ =
1
Example
Given:
S
ut
= 1400 MPa
S
yt
= 950 MPa
Heattreated (asforged)
F
mean
= 9.36 kN
F
max
= 10.67 kN
d/w = 0.133; d/h = 0.55
Find:
q for infinite life, assuming
F
mean
is constant
Example, cont.
Find o
m
and o
a
( ) ( )
( )( )
( )( )
MPa 28
MPa 228
MPa 200
Nm 800 3 . 0 10 0.67x 1
4
1
4
1
2 2
Nm 702 3 . 0 10 x 36 . 9
4
1
4
1
2 2
m 009 . 0
2
m x10 16 . 3 18 10 75
12
1
12
1
12
1
max
max max
max
max
3
max
max
max
3
max
4 8 3 3 3
= ÷ =
= =
= =
= = =

.

\


.

\

=
= = =

.

\


.

\

=
= =
= ÷ = ÷ = =
=
÷
m a
m
m
m
m
m
I
y M
I
y M
L F
L F
M
L F
L F
M
h
y
h d w bh I
I
My
o o o
o
o
o
Stress Concentration Factor
Example, cont.
Since this is uniaxial
loading,
o
m
= 200 MPa
o
a
= 28 MPa
We need to take care of
the SC factors
Su = 1400Mpa
k
t
~ 2.2 (Figure A152)
q ~ 0.95 (Figure 720)
k
f
= 2.14
( ) 1 1 ÷ + =
t f
k q k
nominal
( )( )
( )( ) MPa 428 200 14 . 2
MPa 60 28 14 . 2
= = = ' =
= = = ' =
nom
nom
m f m m
a f a a
k
k
o o o
o o o
Example, cont.
Find strength
Eqn. 78: S‟
e
= .504S
ut
Modification factors
( )
86 . 0
24 . 1
808 . 0
mm 51 d 2.8
: 19)  (7 Equation
: Size
107 . 0
2
1
eq
=
=
=
s s
÷
b
eq b
eq
k
d k
hb d
MPa 1400 S since MPa 700 ~
ut
= '
e
S
201 . 0
995 . 0
271
: Surface
=
÷ =
=
=
a
b
ut a
k
b
a
aS k
25)  7 (Eq. 1
Bending
: Load
=
c
k
( )( )( ) MPa 121 700 86 . 0 201 . 0 = =
e
S
Example, cont.
Design criteria
Goodman line:
For arbitrary variation in
o
a
and o
m
,
n
S S
ut
m
e
a
/ 1 = +
o o
121
1400
1
1400 121
= +
m a
o o
25 . 1
1400
428
121
60 1
1
1400 121
=
+ =
= +
q
q
q
o o
m a
Example, cont.
However, we know that
F
mean
= constant from
problem statement
o
m
= constant
4 . 1
60
84
MPa 84
1
1400
428
121
1
= = =
=
= +
= +
a
a
a
a
ut
m
e
a
S
S
S
S S
S
o
q
o
Less conservative!
Combined loading and fatigue
Size factor depends on loading
SC factors also depend on loading
Could be very complicated calculation to keep track of each load
case
Assuming all stress components are completely reversing and
are always in time phase with each other,
1. For the strength, use the fully corrected endurance limit for
bending, S
e
2. Apply the appropriate fatigue SC factors to the torsional stress,
the bending stress, and the axial stress components
3. Multiply any alternating axial stress components by the factor
1/k
c,ax
4. Enter the resultant stresses into a Mohr‟s circle analysis to find
the principal stresses
5. Using the results of step 4, find the von Mises alternating stress
o
a
‟
6. Compare o
a
‟ with S
a
to find the factor of safety
Additional details are in Section 614
More Fatigue Failure Examples