CHAPTER

Failure Analysis and

8
Prevention

The repetitive loading of engineering materi-

als opens up additional opportunities for struc-
tural failure. Shown here is a mechanical test-
ing machine, introduced in Chapter 6, modified
to provide rapid cycling of a given level of me-
chanical stress. The resulting fatigue failure is a
major concern for design engineers. ( Courtesy
of Instron Corporation)
 Scale

Starting position
Pointer
Hammer

End of swing
h
Specimen

h′

Anvil

Figure 8-1 Charpy test of impact energy. (From H. W. Hayden, W. G. Moffatt,

and J. Wulff, The Structure and Properties of Materials, Vol. 3: Mechanical
Behavior, John Wiley & Sons, Inc., New York, 1965.)
 Testing temperature, ºF
–200 0 +200 400
100
70

80 60
Cu: 23000-061

Impact energy, ft•lb

50
Impact energy, J

60
40

40 30

20
20
Mg: AM100A 10

0 0
–200 –100 0 +100 200 300
Testing temperature, ºC
Figure 8-2 Impact energy for a ductile fcc alloy (copper C23000–
061, “red brass”) is generally high over a wide temperature
range. Conversely, the impact energy for a brittle hcp alloy
(magnesium AM100A) is generally low over the same range.
(From Metals Handbook, 9th Ed., Vol. 2, American Society
for Metals, Metals Park, Ohio, 1979.)
 Temperature, F

–100 0 100 200 300 400

250
175
0.11% C
200 150

Impact energy, ft•lb

125

Impact energy, J
150
0.20% C 100

100 75
0.31% C 0.60% C
0.41% C 0.49% C 50
0.69% C
50
25
0.80% C
0 0
–100 –50 0 50 100 150 200 250
Temperature, C
(a)

Temperature, F

–50 0 50 100 150 200 250 300

300
2% Mn 1% Mn 200
250
0.5% Mn
150

Impact energy, ft•lb

200
Impact energy, J

0% Mn
150
100

100
50
50

0 0
–50 –25 0 25 50 75 100 125 150
Temperature, C
(b)

Figure 8-3 Variation in ductile-to-brittle

transition temperature with alloy com-
position. (a) Charpy V-notch impact
energy with temperature for plain-carbon
steels with various carbon levels (in
weight percent). (b) Charpy V-notch
impact energy with temperature for Fe–
Mn–0.05C alloys with various man-
ganese levels (in weight percent). (From
Metals Handbook, 9th Ed., Vol. 1, Amer-
ican Society for Metals, Metals Park,
Ohio, 1978.)
 (a) (b)

Figure 8-4 (a) Typical “cup and cone” ductile fracture surface. Fracture
originates near the center and spreads outward with a dimpled tex-
ture. Near the surface, the stress state changes from tension to shear
with fracture continuing at approximately 45 ◦ . (From Metals Hand-
book, 9th Ed., Vol. 12, ASM International, Metals Park, Ohio, 1987.)
(b) Typical cleavage texture of brittle fracture surface. (From Metals
Handbook, 9th Ed., Vol. 11, American Society Metals, Metals Park,
Ohio, 1986.)
 a

Figure 8-5 Fracture

toughness test.
 1
! 1 2
!a
(flaw-induced fracture)

a critical
Log flaw size (log a)
Figure 8-6 A design plot of stress versus flaw size for a pressure vessel
material in which general yielding occurs for flaw sizes less than a
critical size, acritical , but catastrophic “fast fracture” occurs for flaws
larger than acritical .
Figure 8-7 Two mechanisms for improv-
ing fracture toughness of ceramics by
crack arrest. (a) Transformation tough- Tetragonal
ening of partially stabilized zirconia ZrO2 grain
involves the stress-induced transforma-
tion of tetragonal grains to the mono-
clinic structure which has a larger spe- Monoclinic
cific volume. The result is a local vol- ZrO2 grain
ume expansion at the crack tip, squeez-
ing the crack shut and producing a resid-
ual compressive stress. (b) Microc- Cubic ZrO2
racks produced during fabrication of matrix
the ceramic can blunt the advancing
crack tip. (a)

Microcrack

(b)
Stress
Tensile strength (T.S.)

Fracture

Time
Figure 8-8 Fatigue corresponds to the brittle fracture of an alloy after a total of N cycles to a stress
below the tensile strength.
 Flexible coupling Counter
Bearing housing
Specimen

Fulcrum High-speed motor

Fulcrum P/2
P/2
Bearing housing
Figure 8-9 Fatigue test. (From C. A. Keyser, Materials Science in Engineering,
4th Ed., Charles E. Merrill Publishing Company, Columbus, Ohio, 1986.)
 1000

T.S.
Applied stress, S (MPa)

500
Fatigue strength
or endurance limit

0
1 10 102 103 104 105 106 107 108
Number of cycles to failure, N
Figure 8-10 Typical fatigue curve. (Note that a log scale is required for the horizontal axis.)
 Specimen surface

(a)
Localized slip

(b)

(c)
Intrusion

(d)
Figure 8-11 An illustration of how repeated
stress applications can generate local-
ized plastic deformation at the alloy sur-
face leading eventually to sharp discon-
tinuities.
Crack length, a

da
1 2
dN da
a1 1 2
dN

a0

Cycles, N
Figure 8-12 Illustration of crack growth with number of stress cycles, N , at two dif-
ferent stress levels. Note that, at a given stress level, the crack growth rate, da/dN ,
increases with increasing crack length, and, for a given crack length such as a1 ,
the rate of crack growth is significantly increased with increasing magnitude of
stress.
 da
= A(DK)m
(log scale)

dN
dN
da
Crack growth rate,

I II III

Stress intensity factor range, DK (log scale)

Figure 8-13 Illustration of logarithmic relationship be-

tween crack growth rate, da/dN , and the stress in-
tensity factor range, 1K . Region I corresponds to
nonpropagating fatigue cracks. Region II corresponds
to a linear relationship between log da/dN and log
1K . Region III represents unstable crack growth
prior to catastrophic failure.
 (a)

(b) (c)

Figure 8-14 Characteristic fatigue fracture surface. (a) Photograph of an aircraft

throttle-control spring (1 21 × ) that broke in fatigue after 274 h of service. The al-
loy is 17–7PH stainless steel. (b) Optical micrograph (10 × ) of the fracture ori-
gin (arrow) and the adjacent smooth region containing a concentric line pattern
as a record of cyclic crack growth (an extension of the surface discontinuity shown
in Figure 8–11). The granular region identifies the rapid crack propagation at
the time of failure. (c) Scanning electron micrograph (60 × ), showing a closeup
of the fracture origin (arrow) and adjacent “clamshell” pattern. (From Metals
Handbook, 8th Ed., Vol. 9: Fractography and Atlas of Fractographs, American
Society for Metals, Metals Park, Ohio, 1974.)
 400
Pearlitic (80 – 55 – 06 as cast) unnotched
350
Fatigue strength, MPa

300

Fatigue strength, ksi

40
250

200

150
20
100

50
104 105 106 107 108
Fatigue life, cycles
(a)

200
26
21 C (70 F)
Fatigue strength, MPa

Fatigue strength, ksi

150 22

65 C (150 F) 18
100 C (212 F)
100 14

10
50
106 107 108
Stress cycles
(b)

Figure 8-15 Comparison of fatigue curves for (a) ferrous and

(b) nonferrous alloys. The ferrous alloy is a ductile iron.
The nonferrous alloy is C11000 copper wire. The nonfer-
rous data do not show a distinct endurance limit, but the
failure stress at N = 108 cycles is a comparable parameter.
(After Metals Handbook, 9th Ed., Vols. 1 and 2, American
Society for Metals, Metals Park, Ohio, 1978, 1979.)
 1
500 F.S. = T.S.
2
Fatigue strength, F.S. (MPa)

1
F.S. = T.S.
4

0
0 500 1000
Tensile strength, T.S. (MPa)

Figure 8-16 Plot of data from Table 8.4 showing how fatigue strength is gener-
ally one-fourth to one-half of the tensile strength.
 Increased cold work or surface
smoothness

Log N
Figure 8-17 Fatigue strength is increased by prior mechanical
deformation or reduction of structural discontinuities.
 25,000

Fused silica glass (wet)

20,000
Breaking stress, psi

15,000
Soda-lime
silica glass
(wet)
10,000

5,000

0
10–3 10–2 10–1 1 10 102 103 104 105
Stress duration, sec.
Figure 8-18 The drop in strength of glasses with duration of load (and without
cyclic load applications) is termed static fatigue. (From W. D. Kingery, In-
troduction to Ceramics, John Wiley & Sons, Inc., New York, 1960.)
 O
H H
+
Si O Si

Si OH

HO Si
Figure 8-19 The role of H 2 O in
static fatigue depends on its
reaction with the silicate net-
work. One H 2 O molecule
and one –Si–O–Si– segment
generate two Si–OH units.
This is equivalent to a break
in the network.
Crack growth by local shearing mechanism
(a)

H2O H2O

Crack growth by chemical breaking of

oxide network

(b)
Figure 8-20 Comparison of (a) cyclic fatigue in
metals and (b) static fatigue in ceramics.
 60
1800 cycles per minute
1 8
50
23˚C(73˚F) 7
2
40 6

Stress (103 psi)

Stress (MPa)

5
21˚C(73˚F)
30
4
3
20 4 3
66˚C(150˚F)
2
100˚C(212˚F)
10
1

0 0
103 104 105 106 107
Cycles to failure

1. Tensile stress only

2. Completely reversed tensile and compressive stress
3. Completely reversed tensile and compressive stress
4. Completely reversed tensile and compressive stress
Figure 8-21 Fatigue behavior for an acetal polymer at various temper-
atures. (From Design Handbook for Du Pont Engineering Plas-
tics, used by permission.)
 x-ray source

Test piece
(containing high-density
inclusion and low-density
pore)

Film
(showing corresponding
lesser and greater film
darkening)

Figure 8-22 A schematic of x-radiography.

 Initial pulse

Front
surface

Oscilloscope
display
Intensity

Back
Flaw surface

Time

Ultrasonic transducer

Water
bath

Flaw Sample

Figure 8-23 A schematic of a “pulse echo” ultrasonic test.

(Courtesy of Paramount Pictures and Twentieth Century Fox)