16

CHAPTER 6
FAILURE ANALYSIS – VARIABLE LOADING

We analyzed the machine elements for static/steady loading in Chapter 5. Often the
machine elements are subjected to variable or changing loads in time. The failure
analysis of mechanical systems under variable loading is known as fatigue analysis. In
this case, a mechanical element fails due to the repetition of the load. Micro cracks
would develop at places of stress concentration, which might get larger and propagate
over time eventually causing the failure of the element. There are basically 3 methods
for the fatigue analysis: 1) The Stress-Life method (Section 6-4 in the textbook), 2)
The Strain-Life method (Section 6-5) and 3) The Fracture Mechanics method (Section
6-6). We will follow the 1st method, i.e. The Stress-Life method.

6-4 The Stress-Life Method

The test specimens are subjected to repeated loads of known magnitudes and the
stress cycles are counted. The well-known testing is the R.R. Moore rotating beam
experiment, which subjects a special test specimen (Figure 6-9, pg. 282, in the
textbook) to pure (+) and () bending moments. Other test-devices are also available
for applying reversed axial, torsional or combined stresses. The results of the rotating
beam experiment is recorded in a semi-log S-N diagram, whose horizontal axis shows
the number of stress cycles (N) and the vertical axis is for the stress of failure (Fatigue
Strength = Sf). A stress cycle N =1 means that the load is applied once in one direction
and removed, and then applied in the opposite direction and removed again. A typical
S-N diagram for a ferrous metal or alloy looks like the one shown below:

Finite Life Infinite Life

Low Cycle High Cycle
Sut

Fatigue
Strength
(Sf)

Se

103 Ne=106
Number of Stress Cycles (N)

In the diagram above, Se is called the endurance limit of the material, which is very
important for the material because it indicates the boundary of the infinite life. If the
alternating stresses stay below this limit then the material can be used very long time
(infinite life) with no fatigue failure. The nonferrous metals and alloys do not have an
endurance limit.
 17

6-7 Endurance Limit

Although it is a very lengthy process, the fatigue testing of materials is now
frequently done, but a quick way of estimating the endurance limit would be very
useful. By comparing the results of simple tension and fatigue tests, we can try to find
a relation between the endurance limit of a material in a rotating-beam test ( Se' ) and
the tensile strength (Sut) of the same material. For steels, this relation is given as

0.5 S ut if S ut 1400 MPa

S e'  
700 MPa if S ut 1400 MPa

The endurance limits for different kinds of cast iron and fatigue strengths (not
endurance limits) for some aluminum alloys at N = 5108 stress cycles are given in
Table A-24, pgs. 1054 and 1055 in the appendix of the textbook.

6-8 Fatigue Strength

The curve for the high-cycle region shown above in the S-N diagram between N = 103
and 106 stress cycles is approximated by an equation given as
S f  aN b

where the coefficient a and the exponent b are found from the end conditions as

( f S ut ) 2
a
Se
and
1 f S ut
b= log ( ).
3 Se
In these equations, the fatigue strength coefficient f is found from Figure 6-18 in pg.
293 of the textbook.

Remember to replace Se with S e' in these equations for the rotating-beam specimen.

Note: Review Example 6-2, pg. 294 in the textbook.

6-9 Endurance Limit Modifying Factors

The endurance limit of an actual machine element (Se) will be different from the
endurance limit of the beam used for the rotating-beam test in the lab ( Se' ) . Because
in an actual working environment, conditions are different from those in the lab.
Hence, we define

Se = ka kb kc kd ke kf S e'

where the coefficients are detailed below.

 18

ka: Surface Factor

This factor is to account for the surface condition of the machine element in the actual
working place. A relation is defined as

b
k a  aS ut

where a and b are found from Table 6-2, pg. 296 in the textbook.

kb: Size Factor

Bending and Torsion
When the bar is rotating:
For a solid round cross-section of diameter d or hollow round cross-section of outside
diameter d, the size factor is given as

1.24 d 0.107 2.79  d  51 mm

kb    0.157
1.51 d 51  d  254 mm

When the bar is not rotating:

For a solid round cross-section of diameter d or hollow round cross-section of outside
diameter d, we define an effective diameter de = 0.37d and replace d with de in the
formulas above.

For a rectangular bar of cross-section h×b, the effective diameter is defined as

de=0.808 hb and again the same formulas above are used replacing d with de.

Axial or Tensile Loading

For all cases in axial loading, the size factor is taken as unity, i.e. kb = 1.

kc: Load Factor

The average load factor for different loadings are given as
1 bending

k c  0.85 axial
0.59 torsion


kd: Temperature Factor

When the machine element is operating at elevated temperatures, a temperature factor
kd must be accounted for in the endurance limit calculation. The temperature factor is
defined as

ST
kd 
S RT

where ST is the tensile strength of the material at the elevated temperature and SRT is
the tensile strength at the room temperature. The values of kd at different temperatures
are given in Table 6-4, pg. 299 in the textbook. An equation for kd is also given as
follows.
 19

kd = 0.9877 + 0.6507(103)Tc  0.3414(105)Tc2 + 0.5621(108)Tc3  6.246(1012)Tc4

where Tc is the elevated temperature in degree Celsius.

ke: Reliability Factor

This factor accounts for standard deviations of data from their mean values. Table 6-5
in pg. 301 of the textbook gives reliability factors for some standard specified
reliabilities.

kf: Miscellaneous-Effects Factor

This factor includes all the effects not covered in the above factors, such as residual
stresses, corrosion, metal spraying and so on. We will take this factor to be unity, i.e.
kf = 1.

6-10 Stress Concentration Factor for Fatigue: Kf (Do not mixed it up with kf !..)
The fatigue usually develops at the points of stress concentration. Hence, we define a
stress concentration factor for fatigue (Kf or Kfs), which is smaller than the theoretical
stress concentration factor defined before for the static loading (Kt or Kts). There are
two ways of finding Kf or Kfs:

1) Kf = 1 + q(Kt1) and Kfs = 1 + qshear(Kts1)

where Kt and Kts are found using Tables A-15 and A-16, pgs. 1034-1042 in the
appendix of the textbook, and the notch sensitivities q and qshear are determined from
Figures 6-20 and 6-21, pgs. 303 and 304 in the textbook. In these figures the notch
radius r of the horizontal axis is defined as shown below. For the cast iron only, it is
recommended that q = 0.2.

hole radius = r fillet radius = r fillet radius = r

Transverse Hole Shoulder Groove

K t 1
2) Neuber equation, K f  1 
1 a / r

where the Neuber constant a for bending or axial, and for torsion is given separately
in equations (6-35a) and (6-35b) on pg. 304 of the textbook.

Note: Review Examples 6-6 to 6-9, pgs. 304-308 in the textbook.

 20

6-11 General Fluctuating Stresses

Up until now, we assumed that the loads or stresses were completely reversed shown
as a sine curve below

max
Load or
Stress
a

m=0 Time

a

min

For a general fluctuating load or stress, the above curve looks like the following

max
Load or
Stress
a

m a

min
Time

The stress components shown in the above figures are defined as

 max   min
Midrange Stress Component = m =
2
 max   min
Alternating (Amplitude) Stress Component = a =
2

Remember that for the completely reversed stress min = max and hence
  ( max )
m= max = 0.
2

6-12 Fatigue Analysis for General Fluctuating Stresses

There are 4 methods for the fatigue analysis of a general fluctuating stress, namely:
1)Soderberg, 2)Modified Goodman, 3)Gerber and 4)ASME-Elliptic. These methods
 21

are defined as lines or curves in a diagram shown below, where the midrange
stress/strength components are shown on the horizontal axis and the alternating
stress/strength components on the vertical axis.

Sy

Yield Line Load Line

Slope=r=Sa/Sm=a/m
Gerber
Alternating Se
Stress or
Strength ASME-
Say Elliptic
C
Sa B Modified
Goodman
a A Soderberg

m Sm Smy Sy Sut

Midrange Stress or Strength

In the figure, point A is the stress point and point B is the intersection point between
the fatigue line (or curve) and the load line. Point B above is shown only for the
intersection of Modified Goodman line and load line. We define the fatigue factor of
safety as nf = Sa/a = Sm/m. Point C shows the intersection of the yield line with the
load line, from which we define the yield factor of safety as ny = Say/a = Smy/m. The
equations for the fatigue lines and yield line are given as

Fatigue Lines
Sa Sm
1) Soderberg:   1.
Se S y
n f σa n f σm
Since Sa= nf a and Sm= nf m , then the equation becomes   1 , from
Se Sy
1
which we find n f  .
σa σm

Se Sy
Sa Sm nf a nf m
2) Modified Goodman:   1 , or   1 , from which we find
S e S ut Se S ut
1
nf  . For this method refer to Table 6-6, pg. 315 in the textbook.
a m

Se S ut
 22

2
S S 
2
nf a nf m 
3) Gerber: a   m   1 , or     1 , which we solve it for nf.

S e  S ut  Se  S ut 
For this method refer to Table 6-7, pg. 315 in the textbook.

2 2
S  S   n f σa   n f σm 
2 2

4) ASME-Elliptic:  a    m   1 , or       1 , from which we

S S  S  S 
 e  y  e   y 
1
find n f  . For this method refer to Table 6-8, pg. 316 in the
2
 σa   σm 
2

    
S S 
 e   y 
textbook.

Yield Line
S ay S my
The yield line equation:   1.
Sy Sy
ny σa ny σm
Since Say= ny a and Smy= ny m , then the equation becomes   1 , from
Sy Sy
Sy
which we find n y  .
σa  σm

The question now is: which mode of failure is more dangerous or what is the mode of
failure? The fatigue or yield?. We can find an answer to this question in 2 ways:

1) If nf < ny then the fatigue is more dangerous than yield. Else, if ny < nf then the
yield is more dangerous than fatigue.

2) Alternatively, we can find a critical slope of rcrit that corresponds to the slope of the
load line going through the intersection point between the yield line and fatigue line.
These rcrit values are given in all the Tables of 6-6 (Modified Goodman), 6-7 (Gerber)
and 6-8 (ASME-Elliptic). Now, if r > rcrit then the fatigue is more dangerous than
yield. Else, if r < rcrit then the yield is more dangerous than fatigue.

Note: Review Examples 6-10 (pgs. 316-319) and 6-12 (pgs. 321, 322) in the
textbook.

6-14 Combinations of Loading Modes

So far we have assumed that there is only one kind of load acting on the machine
element, i.e. we have assumed that there is only axial or bending or torsion. How do
we proceed when the machine element is simultaneously subjected to axial, bending
and torsional stresses? The answer is to follow the following 5 steps:

1) Calculate the endurance limit Se with the assumption of only bending.

2) Compute the midrange and alternating components of axial, bending and torsional
stresses with corresponding fatigue stress concentration factors of Kf and Kfs.
 23

3) Multiply the alternating axial stress component by 1/kc,ax = 1/0.85 = 1.176.

4) Find the equivalent alternating and midrange stress components using the von
Mises approach as  'a   a2  3 a2 and  'm   m
2
 3 m
2
.

5) Calculate the factor of safety for fatigue and yield using  'a and  'm in the
formulas.

Note: Review Example 6-14, pgs. 326-329 in the textbook.