Chapter 6

Fatigue Failure Resulting from

Variable Loading

Introduction
Nature of failure due to static load

Under static failure, the stress on the member is constant.

Prior to a static failure: a very large deformation will occur on the structure or machine
members. Therefore the failure can be predicted by observation.

Nature of failure due to fatigue load

Alternating or fluctuating stresses on member will subject the member to fatigue

failure. Fatigue failure happens at a sudden without any hint (i.e. without any
deformation).

R. R Moore Rotating Beam

Testing specimen
Subject to pure bending (no traverse shear)
Perfectly machined and polished
No circumferential scratches

Fatigue Strength and Cycle Graph (S-N Graph)

Material: Steel
103
Low Cycle

Cycle

High Cycle

Se

Fatigue Strength Sf

Finite Life

Infinite Life

Life

Endurance
limit Se

NA

100

101

102

103

104
105
Cycle , N

106

107

108

Terminology

N : Cycle : 1 rotation of the specimen = 1 cycle of alternating stress

Sf : fatigue strength is the limit of strength where failure occurs when the alternating
stress is above the fatigue strength. However, when knee occurs on graph, fatigue
strength becomes constant. The value of fatigue strength is called Endurance Limit
(Se). For steel, the knee occurs when cycle is larger than 106.
For non-ferrous metal and alloys: no endurance limit. However, the value of
endurance limit in Table A-22 is equivalent to the fatigue strength @ 5(108).
Cycle: Cycle is related to the cycle of the specimen: boundary is 103
Low cycle N < 103, High Cycle N P 103
Life: Life is related to the stress of the specimen: boundary is Se.
Finite Life : N < 106

Infinite Life: N P103

Endurance limit
Steel

0 .5
'
e

S
= 0 .6
Su

980

840

0 .4

700

Endurance Limit

560

Se, MPa
420

280

140

0
0

280

560

840

Se 40% to 60% if Sut < 1425 MPa

Se = 724 MPa when Sut > 1425 MPa

1120

1400

Tensile strength Sut, MPa

1680

1960

2100

Endurance limit- cont

Therefore, after simplifying the observation, the estimation of the endurance limit is

0.5S ut
S 'e =
700Mpa

S ut 1400 MPa
S ut > 1400 MPa

Sut = 3.41 HB Mpa

Eq 6.8 pp 274

Eq 2.17 pp 37

Notes:
Various class of cast iron, polished and machined: Table A-22 pp 1023.
Aluminum alloy: does not have an endurance limit, fatigue strength from Table A-22
is set to 5(108) cycles

Fatigue Strength
To approximate relationship between S-N during high cycle, two stage approximation is
required which is
i) approximation of the fatigue strength fraction f @ 103
ii) approximation of the S N during high cycle
i)

Approximation of the fatigue strength f @103

Fatigue strength fraction:

f =

where

'F
Sut

f=

Sf
Sut

(2 10 3 ) B
B=

Eq 6-10 pp 276

log( 'F / Se )
log( 2 Ne )

Eq 6.12 pp 276

'F = Sut + 345 Mpa

Eq 6.11 pp 276
Using the above formula, a graph is plotted generates Figure 6-18

ii) Approximation of S N during high cycle

Calculate the fatigue strength when N is known

S f = aNb

Eq 6.13 pp 277

Calculate the cycle when Sf is known

N= a
a
where

1
b

Eq 6.14 pp 277

a: slope of the curve (log vs log)

b: Sf @ 103

fS
1
b = log( ut )
3
Se

Eq 6.14 and 6.15 pp 277

(f S ut ) 2
a=
Se
f: fatigue strength fraction

ENDURANCE LIMIT (Se)

Se is the value from experiment of laboratory controlled environment (Refer R-R Moore
test specimen). In reality, difference in material, manufacturing, environment and
design exist. Therefore, modifying factor is introduced, using Marin equation.

Se = k ak bk c k dk ek f S'e

where
Se : endurance limit based on R.R. Moore experiment
ka: surface factor
kb: size factor
kc: loading factor
kd: temperature factor
ke: reliability factor
kf: miscellaneous-effects factor

Surface Factor ka
Rotating specimen is highly polished without any circumferential scratches.
Therefore, surface factor for other finishes

k a = aS utb

Eq 6.19 pp 279
Factor a and b can be found in Table 6-2 pp 280

Size Factor kb
For rotating circular cross section subjected to bending or torsion

1.24 d0.107
kb =
0.157
1.51d

2.79 d 51mm
51 < d 254mm

kb = 1 (for axial load), please refer to kc

Eq 6.20 pp 280

For other condition such non-rotating circular shaft or non-circular cross section, effective
dimension de is used in Eq 6-20.
Effective dimension de non-rotating solid and hollow circular shaft
Rotating solid or hollow circular shaft: 95% stress area is

A 0.95 =

2
d (0.95d) 2 = 0.0766 d 2
4

Non-rotating solid or hollow circular shaft:

95% stress area is defined twice the area outside of two parallel chords having a
spacing of 0.95D, where D is the diameter. Therefore the exact computation is

A 0.95 = 0.0105D 2
Therefore

A0.95 = 0.0105D 2
0.0766 d e = 0.0105 D 2
2

d e = 0.37D

Effective dimension de of rectangular cross section with h x b dimensions due to nonrotating bending
95% stress area is

A 0.95 = 0.05(hb )

1
2

Therefore

de = 0.808(hb )

1
2

Similar principle is applied to other common non-rotating structural shapes and

effective dimension de is as in Table 6-3 pp 282

Loading factor kc
Bending

kc = 1

Axial

kc = 0.85

Eq 6.26

Temperature Factor kd

S
kd = T
SRT

ST: tensile strength at T temperature

SRT: tensile strength at room temperature

Refer to Table 6-4 pp 283: this table shows the endurance limit of the steel increases
kd > 1 for 200C < T < 2500C. This data obtained from Figure 2-9 pp 39 (result from
145 tests of 21 carbon and alloy steel).
The mathematics representation of data using 4th order polynomial

k d = 0.9877 + 0.6507(10 3 )TC 03414(10 5 )TC2 + 0.5621(10 8 )TC3

6.246(10 12 )Tc4

37 TC 540 0 C

Reliability Factor (ke)

This factor is intended to account the scattered data in Figure 6-17. Se = 0.5 Sut is
only mean of the data.
Refer to Table 6-5 pp 285

Miscellaneous-Effects Factor kf
This factor will take into account other factors that reduce the endurance limit such as
corrosion, electrolytic plating, metal spraying, cyclic frequency and/or frettage
corrosion.

Stress concentration and notch sensitivity

Stress increases at the vicinity of irregularities or discontinuities such as holes,
grooves or notches.
Theoretically, these discontinuities causes the stresses to increase
significantly by

max = K f 0

or

max = K fs 0

Eq 6.30 pp 287

Kf: stress-concentration factor

Notch sensitivity q is defined

K 1
q= f
Kt 1

or

qshear =

K fs 1
K ts 1

Eq 6.31 pp 287

Kt: stress concentration factor

Therefore, fatigue stress concentration factor (Kf)

K f = 1 + q(K t 1)

Eq 6.32 pp 287

q: notch sensitivity from Figure 6-20 pp 287

K fs = 1 + qshear (K ts 1)

Eq 6.32 pp 287

qshear : notch sensitivity from Figure 6-21 pp 288

Note: If the value of notch sensitivity can not be retrieved from the respective figures, the
most conservative assumption is that Kf = Kt or Kfs = Kts
Kt, Kts: can be obtained from Table A-13 to A-14

Stress concentration (Kt or Kts)

D= 30mm, d = 20mm and r = 2 mm and subjected to
Moment M
3.0

Figure A-13-9
2.6

Roundshaft withshoulder fillet

Mc
, where
I
d
d4
c = ,andI =
Kt
2
64

inbending, o =

2.2
D/d=1.5

1.8
1.54

1.4

geometrical property to be used

in calculating the stress.

1.0
0

0.05

0.10

0.15
r/d

0.20

0.25

0.30

Stress

Fluctuating Stresses
max
a:

amplitude stress

m:

midrange stress

min

Time

Stress Ratio:

Amplitude Ratio:

R=

min
max

A=

a
m

Midrange Stress

m =

Alternating Stress

a =

max + min
2

max min
2

Similarly for the fluctuating forces

Midrange Force

Alternating Force

Fm =
Fa =

Fmax + Fmin
2

Fmax Fmin
2

These forces will generate respective midrange and alternating stresses.

Assumption used in this book

Avoid localized plastic strain at a notch:

a = K f ao

and

m = K f m 0

When plastic strain cannot be avoided at the notch, conservatively

a = K f ao

and use Kfm = 1, that is

m = m 0

How to determine the midrange or alternating stress?

Stress

Moment on rotating shaft.

When the shaft rotates in ccw direction, element A will rotate from position 1, 2, 3 and 4
for one full cycle. At the same time, the stresses of A will fluctuate from 0, +s0, 0 and
-s0. Therefore, moment on rotating shaft will generate sa = Mc/I and sm = 0.

Stress

Torsion on rotating shaft

I A
4

ta = 0 and tm = Tr/J.

Time

Fatigue Failure Criteria:

graphical and empirical method

Sy

A: ( a , m )
B: (Sa1, S m1 )
C: (Sa2, Sm2 )
D: (Sa3, Sm3 )
Langer Line
Load line

D
C

Se

B
A

Gerber Line

Goodman Line

Sy

Sut

Graphical Method

Goodman line

n=

Sa1 Sm1
=
a
m

Gerber Failure Theory

n=

S a 2 Sm2
=
a
m

Langer Line

n=

S a 3 S m3
=
a
m

Empirical Method
Goodman Line
Basic line equation

Sm Sa
+
=1
Sut S e

For a known (sa, sm)

m a 1
+
=
Sut S e n

Eq 6.45 pp 298

Gerber Failure Locus (Parabolic Equation)

Basic parabolic equation

For a known (sa, sm)

Sm

S ut

S
+ a = 1
Se

n m
n a

+
=1
Se
S ut

Eq 6.46 pp 298

Langer Line (failure due to yielding)

Basic line equation

For a known (sa, sm)

Sm Sa
+
=1
Sy Sy
m a 1
+
=
Sy Sy n
Sy
a + m

Eq 6.48 pp 298

=n

For other failure theories, the same principles can be applied.

*In the following discussion, these three theories will be used.

Langer line vs. other theories

(i.e. Modified Goodman and Gerber Line )
a

Sy

Sy

Se

Se

Sy

Sut

Sy

Sut

If the stresses on the member is plotted on the shaded area, the load line of the
member will intersect with Langer line first. Therefore, failure due to yielding will occur
first. This is type of failure is called Langer-first-cycle failure. Empirically, it can be
determined when nlanger < nf(other).

Procedure in solving the fatigue problem

1.

Determine sa, sm,ta and tm.

2.

Combine the alternating stresses using DET

total( a ) = a2 + 3a2
3.

Combine the midrange stresses using DET

total(m ) = m2 + 3m2
4.

Use the theories to solve the problem