Canale L.C.F., Mesquita R.A., Totten G.E. Failure Analysis of Heat Treated Steel Components
Подождите немного. Документ загружается.
Microstructures that are formed upon cool-
ing and the proportions of each are dependent on
austenitization time, temperature, cooling his-
tory of the particular alloy, and the composition
of the alloy. The transformation products from
austenite decomposition change from a mixture
of ferrite and pearlite to bainite or martensite
with increasing cooling rate.
Bainite is a two-phase mixture of ferrite
and cementite consisting of fine lines of iron
carbide in acicular ferrite. Upper bainite has a
feathery appearance and forms just below the
temperature where fine pearlite is formed.
Lower bainite exhibits an acicular micro-
structure that is formed just above martensite,
which is produced at approximately 350
C
(660
F).
Martensite is a supersaturated solid solu-
tion of carbon in alpha iron (ferrite) that is
less densely packed than the c body-centered
tetragonal lattice and is a magnetic platelike
structure formed by a diffusionless shear type
of transformation of austenite below the mar-
tensite start (M
s
) temperature. The amount of
transformation depends on the martensitic
temperature range (M
s
to M
f
). (M
f
is the mar-
tensite finish temperature.) The three forms of
martensite are lath, plate, and tempered mar-
tensite.
Transformation from austenite to martensite
results in a volumetric expansion at the M
s
temperature. Dimensional changes are possible,
depending on the carbon content and micro-
structural transformation product formed. The
volume change (%) is [4.64–0.53 (%C)] for the
reaction from austenite to martensite.
The two most commonly used transforma-
tion diagrams are time-temperature transfor-
mation for isothermal transformation, and
continuous cooling transformation diagrams.
These diagrams can be used to predict steel
microstructures and hardness after heat treat-
ment, or they may be used to design a heat
treatment process. Heat treating processes
include hardening, austenitization, annealing
(full annealing, intercritical annealing and sub-
critical annealing, recrystallization annealing,
isothermal annealing, soft annealing, diffusion
annealing), normalizing, stress relieving,
quenching and tempering, and austempering,
and are summarized in Table 1 (Ref 1).
Hardening and tempering are common
heat treatment processes. If steel is cooled suf-
ficiently fast, without microstructural transfor-
mation, thermal stresses can develop. Under
these conditions, the surface of the part is initi-
ally cooled much more quickly than the core.
Therefore, the specific volume in the core is
greater than at the surface, and the reduction in
volume at the surface is resisted by the greater
volume in the core, resulting in the surface being
in tension and the core in compression. After the
cooling processes have been completed, the
residual-stress distribution between the surface
and core is obtained. If the surface stresses
exceed the hot yield strength of the material, it
plastically deforms, resulting in thermally
induced dimensional changes (Ref 3). When
steels that undergo transformational changes are
quenched, the possibility of the formation of
both thermal and transformational stresses must
be considered.
Steel parts are often tempered by reheating
after quench hardening to obtain specific
mechanical properties. The tempering process
involves heating hardened steel to some tem-
perature below the eutectoid temperature for the
purpose of decreasing hardness and increas-
ing ductility and toughness while relieving
quench stresses and ensuring dimensional stab-
ility. Tempering processes include temper-
ing of martensite, transformation of retained
austenite to martensite, tempering of the de-
composition products of martensite, and de-
composition of retained austenite to martensite.
In addition, tempering may also lead to dimen-
sional variation due to relaxation of residual
stress and plastic deformation, which is due
to the temperature dependence of yield
strength. Tempering may lead to an increase in
hardness if secondary hardening occurs, which
is due to precipitation of a compound or to
the formation of martensite or bainite from
retained austenite, decomposition during tem-
pering, or destabilization during this process
and then transformation during subsequent
cooling.
Quenchant selection and quenching con-
ditions are critically important parameters in
quench system design. For example, the dim-
ensional changes after austenitizing and then
quenching in water are greater than quenching in
oil (Ref 3).
Important Design Aspects
The importance of good design cannot be
overemphasized. Poor design can cause or
2 / Failure Analysis of Heat Treated Steel Components
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 2
Table 1 Heat treating process for carbon steels
Process Characteristics
Austenitization:
Complete transformation to austenite by
heating the steel above the critical
temperature for austenitic formation
The optimal austenitization temperature is 30–50
C (55–90
F) above Ac
3
for
hypoeutectoid steels and 30–50
C (55–90
F) above Ac
1
for hypereutectoid steels.
Ac
3
is the temperature at which the transformation of ferrite to austenite is completed
during heating. Ac
1
is the temperature at which austenite begins to form during heating.
The heating rate must be limited and uniform to avoid cracking or warpage and to control
thermal stresses in the range of 250–600
C (480–1110
F).
The carbon equivalent controls the propensity for steel to crack.
The holding time is dependent on geometrical factors related to the furnace (emissivities,
temperature, and atmosphere composition) and load (type of steel and thermophysical
properties).
Annealing:
Heat treatment consisting of heating and
soaking at suitable temperature followed by
cooling under conditions such that, after
return to ambient temperature, the metal will
be in a structural state closer to that of equi-
librium
The primary purpose of annealing is to soften
the steel to enhance its workability and
machinability. Also, it relieves internal
stresses, restores ductility and toughness,
refines grains, reduces gaseous content in the
steel, and improves homogenization of
alloying elements.
Full annealing: Heat 30–50
C (55–90
F) above Ac
3
for hypoeutectoid steels, then
furnace cool through the critical temperature range at a specified cooling rate. The aim
is to break the continuous carbide network of high-carbon steels. It improves
machinability.
Partial (intercritical) annealing: Heating within the critical temperature range (Ac
1
–Ac
3
),
followed by slow furnace cooling. It improves machinability.
Subcritical annealing: Heating 10–20
C (20–35
F) below Ac
1
followed by cooling in
still air. It can be used to temper bainitic or martensitic structures to produce softened
microstructures containing spheroidal carbides in ferrite. Improves the cold working
properties of low carbon steels (525% C) or softens high-carbon and alloy steel
Recrystallization annealing: Heat the steel for 30 min–1 h at temperature above the
recrystallization temperature (T
R
=0.4 T
m
), then the steel is cooled. The treatment
temperature depends on prior deformation, grain size, and holding time. The
recrystallization process produces strain-free grain nucleation, resulting in a ductile,
spheroidized microstructure.
Isothermal annealing: Heating the hypoeutectoid steel within the austenitic transformation
range above Ac
3
for a time sufficient to complete the solution process, yielding a
completely austenitic microstructure. At this time, the steel is cooled rapidly at a
specific rate within the pearlite transformation range until the complete transformation
to ferrite plus pearlite occurs, and then it is cooled rapidly.
Spheroidizing (soft annealing): Involves the prolonged heating of steel at a temperature
near the lower critical temperature (Ac
1
), then furnace cooling
Diffusion (Homogenizing annealed): Heat the steel rapidly to 1100–1200
C
(2010–2190
F) for 8–16 h, furnace cool to 800–850
C (1470–1560
F), and then
cool to room temperature in still air. It is performed on steel ingots and castings to
minimize chemical segregation.
Normalizing:
The aim is to provide a uniform
microstructure of ferrite plus pearlite (small
grains and finer lamellae than in annealing).
Heat the steel to 40–50
C (80–90
F) above Ac
3
for hypoeutectoid steels and 40–50
C
(80–90
F) above A
cm
for hypereutectoid steels. The holding time depends on the size,
and then the steel is cooled in still air. It produces grain refinement and improved
homogenization.
Stress relieving:
It is typically used to remove residual
stresses that have accumulated from prior
manufacturing processes. Stress relieving
results in a significant reduction of yield
strength in addition to reducing the residual
stresses to some “safe” value.
Heat to a temperature below Ac
1
for the required time to achieve the desired reduction in
residual stresses, and then the steel is cooled at a rate sufficiently slow to avoid the
formation of excessive thermal stresses. Below 300
C (570
F), faster cooling rates
can be used. No microstructural changes occur during stress-relief processing. The
recommended heating temperature range is 550–700
C (1020–1290
F), depending on
the type of steel. These temperatures are above the recrystallization temperature. Little
or no stress relief occurs at temperatures 5260
C (500
F), and approximately 90% of
the stress is relieved at 540
C (1005
F). The maximum temperature for stress relief
is limited to 30
C (55
F) below the tempering temperature used after quenching.
The results of the stress-relief process are dependent on the temperature and time.
Hardenability:
Ability to develop hardness to a given
depth after having been austenitized and
quenched
The hardenability depends on the concentration of dissolved carbon in the austenitic
phase, alloying elements, austenitizing temperature, austenitic grain size at the moment
of quenching, size and shape of the cross section, and quenching conditions.
Quenching:
Quench severity is the ability of a
quenching medium to extract heat from a
hot steel workpiece.
Specific recommendations for quench media selection for use with various steel alloys are
provided by standards such as SAE AMS 2759. Quench media include water, brine,
aqueous polymer, gas or air quenching, and caustic quenching.
Tempering:
Tempering is the thermal treatment of
hardened and normalized steels to obtain
the desired mechanical properties, which
include improved toughness and ductility,
lower hardness, and improved dimensional
stability.
The tempering process involves heating steel to any temperature below the Ac
1
temperature. During tempering, as-quenched martensite is transformed into tempered
martensite, which is composed of highly dispersed spheroids of cementite (carbides)
dispersed in a soft matrix of ferrite, resulting in reduced hardness and increased
toughness. The objective is to allow the hardness to decrease to the desired level and
then stop the carbide decomposition by cooling. The extent of the tempering effect
is determined by the temperature and time of the process.
Source: Ref 1
Component Design / 3
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 3
promote heat treatment failures before the
component is put into service, or it may reduce
service life, sometimes dramatically. The
objective of proper design for heat treatment is
to provide the minimum engineering require-
ments, the desired material properties at the
lowest total cost, and, in particular, to minimize
the expense of scrap due to rework of parts that
may have undergone excessive distortion or
cracked.
The Heat-Transfer Theory Applied to Heat
Treatments. The laws that govern heat trans-
mission are very important to the engineer in
heat treatment design. There are three different
types of heat transfer: conduction, convection,
and radiation (Ref 4). They have in common that
temperature difference (thermal gradient) must
exist and that the heat is always transferred in the
direction of decreasing temperature. When the
temperature profile does not change with time,
the fundamental relation for the unidirectional
steady flow of heat through a solid by conduc-
tion, Fourier’s first law, can be expressed by:
Q=7l
¶T
¶x
(Eq 1)
where Q is the quantity of heat flowing through
the unit area of a wall per unit time in the
direction of the x-axis and is directly propor-
tional to the thermal conductivity, l, and the
thermal gradient in x-direction. Thermal con-
ductivity has a nearly linear dependence on
temperature.
During heat treatments, the temperature var-
ies in time as well as in space; these processes
are called unsteady, nonstationary, or transient.
As the body heats, the temperature at each point
asymptotically approaches the temperature of
the medium. The temperature of points near the
surface of the body changes most rapidly. The
differential equation for one-dimensional tran-
sient heat conduction, Fourier’s second law, in
the absence of inner heat sources is:
¶T
¶t
=a
¶
2
T
¶x
2
(Eq 2)
where T is the temperature, t is the time, and a is
the thermal diffusivity of the metal and is:
a=
l
rC
p
(Eq 3)
where l is the thermal conductivity, r is the
density, and C
p
is the specific heat at constant
pressure of the material. The thermophysical
properties l, r, and C
p
vary with temperature.
The differential equation of heat conduction
establishes the relation between the time and
space variation of temperature at any point
of the body in which conduction takes place.
The factor of proportionality thermal diffusivity,
a, represents a physical property of the material,
is essential for transient processes of heat flow,
and defines the rate of change of temperature.
If the thermal conductivity, l, is the ability of
a solid to conduct heat, thermal diffusivity is
the measure of a material thermal inertia. The
quantity r C
p
is the volumetric specific heat;
this product is approximately constant for solid
metals. So, in the case of austenitic stainless
steels, low thermal conductivities correspond
to low thermal diffusivities. In other words,
equalization of temperature at all points of
space will proceed at a lower rate in austenitic
stainless steels, with respect to ferritic steel, due
to its lower thermal diffusivity, and there are
difficulties in homogenizing temperature during
heat treatments.
Heat transfer by convection occurs between
the surface of the body and surrounding fluids;
for this type of heat transmission, the following
Newton’s equation is in general use:
Q
h
=hDT (Eq 4)
It simply states that an invariable temperature
difference, DT, between a surface and a fluid in
contact with it causes a steady heat flow of Q
h
.
The factor of proportionality, h, is called the
coefficient of heat transfer.
The third type of heat-transfer mode is
radiation. The heat flow by radiation is com-
monly written:
Q=es T
4
1
7T
4
0
(Eq 5)
where e is the emissivity (1 for a black body),
T is the temperature, and s is the Stefan
constant (Ref 5). The three different types of
heat transfer, conduction, convection, and
radiation, are present during the heat treatment
processes.
Perhaps the most important physical property
of steel to be considered in design is its coeffi-
cient of thermal expansion. Most heat treating
problems could be solved if this coefficient
could be controlled. Because it cannot, it is
necessary to learn to design with it (Ref 6).
Almost all solids expand on heating. As the
temperature is raised, the thermal vibration
pushes the atoms apart, increasing their mean
4 / Failure Analysis of Heat Treated Steel Components
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 4
spacing. The effect is measured by the linear
expansion coefficient:
a=
1
L
dl
dT
(Eq 6)
where L is a linear dimension of the body
(Ref 5).
The relationship between thermal con-
ductivity and thermal expansion is important in
designing against thermal distortion. Thermal
gradients can cause a change of shape, which is a
distortion of the component. The strain is related
to temperature by:
e=a T
0
7TðÞ (Eq 7)
where T
0
is ambient temperature, and a is ther-
mal expansion (Ref 5).
The distortion is proportional to the gradient
of the strain, so it is proportional to the thermal
gradient. By Fourier’s first law, the heat flow is
proportional to the thermal gradient through
the thermal conductivity, l. For a given geo-
metry and heat flow, the distortion is mini-
mized by selecting materials with large values
of l/a (Ref 5). For example, austenitic stain-
less steels have low thermal conductivity and
high thermal expansion, related to ferritic steel,
so distortion during welding becomes a pro-
blem.
Thermal expansion has a strong influence on
the development of residual stress. Whenever
the thermal expansion or contraction of a body is
prevented, thermal stresses appear; if large
enough, they cause yielding, fracture, or elastic
collapse (buckling). For axial constraint, the
stress, Ds, produced by a temperature change of
1
C or the stress per
C caused by a sudden
change of surface temperature in one that is not
constrained is equal to aE, where a is the
expansion coefficient, and E is the elastic mod-
ulus of the material. For biaxial and triaxial
constraint, the stresses shall be multiplied by
(1n) and (12n) respectively, where n is
Poisson’s ratio. These stresses are large and can
cause a material to yield, crack, spall, or buckle
(Ref 5).
Linear thermal expansion (Table 2) in going
from room temperature to 700
C (1300
F) is
approximately:
10.3 mm/m (0.124 in./ft) for low-alloy steel
8.5 mm/m (0.102 in./ft) for martensitic
stainless steel (type 13Cr)
13.1 mm/m (0.157 in./ft) for austenitic
stainless steel (type 18Cr-8Ni).
Thermal expansion of austenite is larger than
that of ferrite. These thermal expansion values
are representative of solution-annealed material.
Subsequent precipitation hardening treatments
may affect thermal expansion (Ref 7).
In high-temperature components design,
incompatibility of thermal expansion becomes a
major problem. Choice of materials and designs
should take this into account. Bolts used to hold
high-temperature casings together must be
selected to have sufficient elevated-temperature
strength and make a good thermal expansion
match with the casing material. When rotor and
casing are made of ferritic steel, modified 12%
Cr bolts work well up to 565
C (1050
F), but
nickel-base superalloys are needed at 595
C
(1100
F) or higher.
The ability of a material to resist thermal
shock, due to a sudden immersion in a cold
medium, without cracking depends on its ther-
mal expansion coefficient, a; tensile strength,
s
t
, for metals; Young’s modulus, E; thermal
conductivity, l; and heat-transfer coefficient, h.
A temperature change of DT applied to a
constrained body or a sudden change D T of the
surface temperature of the unconstrained com-
ponent induces a stress:
s=
EaDT
C
(Eq 8)
where C is equal to 1 for axial constraint, (1n)
for biaxial constraint, (12n) for triaxial con-
straint, and n is Poisson’s ratio. If this thermally
Table 2 Linear thermal expansion for ferrous
materials
Steel
Linear thermal expansion
in temperature range
from room temperature
to 704 °C (1300 °F)
mm/m in./ft
Carbon and low-alloy steels:
C; C-Mn; C-Si; C-Mn-Si; C-
1
/
2
Mo to
1
1
/
4
Cr-
1
/
2
Mo; Mn-
1
/
2
Mo-
1
/
2
Ni
10.3 0.124
5Cr-1Mo and 29Cr-7Ni-2Mo-N steels 9.5 0.114
9Cr-1Mo steel 8.8 0.106
12Cr; 12Cr-1Al; 13Cr; and 13Cr-4Ni
steels
8.5 0.102
15Cr and 17Cr steels 8 0.096
27Cr steels 7.3 0.088
Austenitic stainless steels:
16Cr-12Ni-2Mo; 16Cr-12Ni-2Mo-N;
13.1 0.157
16Cr-12Ni-2Mo-Ti; 18Cr-8Ni
Austenitic stainless steels:
29Ni-20Cr-3Cu-2Mo; 20Cr-18Ni-6Mo;
12.25 0.147
22Cr-13Ni-5Mn
Source: Ref 7
Component Design / 5
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 5
induced stress exceeds the local tensile strength
of the material, yielding (permanent plastic
deformation) or cracking results (Ref 5). This
plastic flow causes permanent shape change
(distortion) and impacts the magnitude and dis-
tribution of residual stresses. Water quenching
gives a high h, and then the values of DT cal-
culated from the previous equation give an
approximate ranking of thermal shock resistance
(Ref 5).
However, when heat transfer at the surface is
poor and the thermal conductivity of the solid is
high, the thermal stress is less than that given by
the previous equations. A measure of the thermal
shock resistance that takes into account the finite
rate of heat transfer at the surface, a heat-transfer
coefficient that is never infinite, is given by:
BDT=
s
t
aE
(Eq 9)
where s
t
, a, and E were defined previously,
B=C/A, where C also was defined previously;
and A is:
A=
sh=l
1+sh=l
(Eq 10)
where s is a typical dimension of the sample in
the direction of heat flow, h is the heat-transfer
coefficient, and l is the thermal conductivity
(Ref 5). The quantity Bi=sh/l is usually called
the Biot modulus. If Bi41, heat flow is limited
by conduction. For fast water quench of metals,
the heat-transfer coefficient, h, is high (h=
10
4
W/m
2
K), and the thermal conductivity is
also high, so the factor A approaches 1. On the
other hand, for fast air flow (h=10
2
W/m
2
K),
the factor A results are equal to 3 · 10
2
(for
section s=10 mm), and the thermal shock
resistance DT is larger by the factor 1/A (Ref 5).
As an example of the use of the aforemen-
tioned equations, Fig. 1 shows schematically the
effect of the thermal expansion coefficient (a)
and the heat-transfer coefficient (h) in thermal
shock resistance (DT) for a hypothetical steel
with 800 MPa (~120 ksi) tensile strength, bi-
axial constraint, and thermophysical properties
constant with temperature. Three cases were
analyzed:
a) In the first case, a ferritic steel with low
thermal expansion and a very high-heat
transfer coefficient (fast water quench,
h=10
4
W/m
2
K) was considered. The max-
imum temperature change (thermal shock
resistance) that induces stresses below the
tensile strength, avoiding yielding or crack-
ing, is 240
C (470 F).
b) The second example similar to case (a) but
with high thermal expansion. The thermal
shock resistance results in temperatures
above 135
C (270
F) increasing the failure
risk.
c) The third example is similar to case (a) but
with a lower heat-transfer coefficient (air
flow, h=50 W/m
2
K). The thermal shock
resistance results in 280
C (535
F)
decreasing the risk of failure.
These factors (residual stresses and dimen-
sional changes) have the greatest influence on
Tensile strength
Thermal shock
resistance
0
100
200
(30)
300
400
(60)
500
600
(90)
700
800
(120)
900
0
50
(90)
100
(180)
150
(270)
200
(360)
250
(450)
300
(540)
Stress, MPa (ksi)
Induced thermal stress for very
high heat-transfer coefficient, h,
and low thermal expansion
Induced thermal stress for very
high heat-transfer coefficient, h,
and high thermal expansion
Induced stress for low heat-
transfer coefficient and low
thermal expansion
Temperature change, °C (°F)
Fig. 1
Schematic representation of thermal stresses resulting from a sudden change, D T , of the surface temperature and thermal shock
resistance
6 / Failure Analysis of Heat Treated Steel Components
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 6
the design process of a component. In addition to
thermal strains, many materials systems undergo
phase transformations as a function of tem-
perature. Often, the new phase(s) that forms has
a different volume and different coefficient of
expansion as well as different mechanical
behavior(s) than the parent phase(s). For exam-
ple, phase transformation from austenite to
martensite results in a volumetric expansion at
the martensite start (M
s
) temperature. These
differences increase the complexity of under-
standing the effect of thermal gradients on the
strains produced and the resulting plastic
deformation (Ref 8). Thermal and transforma-
tion-induced strains can result in substantial
plastic deformation and residual stresses. The
total induced strain is the result of the sum
of the strain produced by thermal expansion
(aE DT=e
th
) of a piece with initial length (E)
and the transformation strain (e
tr
). The total
induced strain must be accommodated through
either elastic (e
el
) or plastic (e
ep
) strain, which
sums to the total strain e
t
=e
th
+e
tr
=
(e
el
+e
ep
). In order to determine the accom-
modation strain values, Young’s modulus (E)
and the yield strength (s
ys
) are required as a
function of phase and temperature. Most of the
plastic deformation occurs during the heat-up
and cool-down stages of the process (Ref 8).
Primary Stresses, Secondary Stresses, Peak
Stresses, and Residual Stresses. Primary
stress is a normal or shear stress developed by
the imposed loading that is necessary to satisfy
the laws of equilibrium of external and internal
forces and moments. The basic characteristic
of a primary stress is that it is not self-limiting.
Primary stress that considerably exceeds the
yield strength will result in failure or at least
in great distortion. Secondary stress is a normal
or shear stress developed by the constraint of
adjacent parts or by self-constraint of a structure.
The basic characteristic of the secondary stress
is that it is self-limiting. An example of sec-
ondary stress is a general thermal stress.
The elastic stresses calculated previously are
nominal values, that do not take into account
local discontinuities such as holes, notches, or
section changes. Even on a structure where
stress intensity has been limited by yield criteria,
there may exist highly localized regions where
peak stresses are several times higher than yield.
Maximum local stresses on a structure can be
determined by considering nominal stresses
multiplied by a stress-concentration factor and
can be estimated through a detailed stress ana-
lysis or by using approximate formulas that
account for the most common cases. Design
should be verified to confirm whether there are
stress-concentration points that may activate
failure mechanisms due to brittle fracture, cor-
rosion, or fatigue.
Examples of peak stresses are thermal stresses
in the austenitic steel cladding of a carbon steel
vessel, thermal stresses in the wall of a vessel or
pipe caused by rapid change in temperature of
the contained fluid, and the stress at a local
structural discontinuity.
Residual stresses (Ref 9) can be defined as
those stresses that remain in a material or body
after being manufactured and processed in
the absence of external forces or thermal gra-
dients. Residual stresses can be defined as either
macro- or microstresses, and both may be pre-
sent in a component. Macroresidual stresses
vary within the body of the component over a
much larger range than the grain size. Micro-
residual stresses, which result from differences
within the microstructure of a material, operate
at the grain-size level or at the atomic level.
Microresidual stresses often result from the
presence of different phases or constituents in a
material.
Residual stresses develop during most
manufacturing processes involving material
deformation, heat treatment, machining, or
processing operations that transform the shape
or change the properties of a material. They arise
from a number of sources and can be present in
the unprocessed raw material, introduced during
manufacturing, or can arise from in-service
loading.
In heat treated parts, residual stresses may be
classified as those caused by a thermal gradient
alone or a thermal gradient in combination
with a microstructural change (phase transfor-
mation). When a steel part is quenched from
the austenitizing temperature to room tempera-
ture, a residual-stress pattern is established due
to a combination of a thermal gradient and a
local transformation-induced volume expan-
sion. Thermal contraction develops nonuniform
thermal (or quenching) stress due to different
rates of cooling experienced by the surface
and interior of the steel part. Transformational
volume expansion induces transformation stress
arising from the transformation of austenite into
martensite or other transformation product
(Ref 10).
Residual stresses may be sufficiently large to
cause local yielding and plastic deformation,
Component Design / 7
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 7
both on microscopic and macroscopic levels,
and can severely affect component perfor-
mance. Both the magnitude and distribution of
the residual stress can be critical to perfor-
mance and should be considered in the design
of a component. In any free-standing body,
stress equilibrium must be maintained, which
means that the presence of a tensile residual
stress in the component will be balanced by a
compressive stress elsewhere in the body.
Tensile residual stresses in the surface of a
component are generally undesirable, since
they can contribute to, and are often the major
cause of, fatigue failure, quench cracking, and
stress-corrosion cracking. Compressive resi-
dual stresses in the surface layers are usually
beneficial, since they increase both fatigue
strength and resistance to stress-corrosion
cracking and increase the bending strength of
brittle ceramics and glass. In general, residual
stresses are beneficial when they operate in
the plane of the applied load and are opposite
to it (for example, a compressive residual stress
in a component subjected to an applied tensile
load).
The origins of residual stresses in a compo-
nent may be classified as mechanical, thermal,
and chemical. Mechanically generated residual
stresses are often a result of manufacturing
processes that produce nonuniform plastic
deformation. They may develop naturally during
processing or treatment or may be introduced
deliberately to develop a particular stress profile
in a component. Examples of operations that
produce undesirable surface tensile stresses
or residual-stress gradients are rod or wire
drawing (deep deformation), welding, machin-
ing (turning, milling), and grinding (normal or
harsh conditions).
On a macroscopic level, thermally generated
residual stresses are often the consequence of
nonuniform heating or cooling operations.
These, together with the material constraints
in the bulk of a large component, can lead to
severe thermal gradients and the development
of large internal stresses. An example is the
quenching of steel (or aluminum alloys),
which leads to surface compressive stresses
balanced by tensile stresses in the bulk of the
component.
Chemically generated stresses can develop
due to volume changes associated with chemical
reactions, precipitation, or phase transformation.
Chemical surface treatments and coatings can
lead to the generation of substantial residual-
stress gradients in the surface layers of the
component.
The criterion applied to avoid plastic defor-
mation states that the calculated stress intensity
or effective stress must be lower than the yield
and design life creep-rupture stresses of the
material. When effective stress is exceeded
somewhere within the component, it does not
necessarily indicate plastic collapse of the entire
structure. Primary stresses may locally exceed
yield, within certain limits, provided that there is
enough ductility to allow the material to yield
without cracking. Plastic collapse occurs when
primary stresses are uniform on the entire
structure and exceed effective stress. To prevent
an incremental collapse or thermal stress ratchet
in each loading cycle, the total elastic stress-
intensity range, considering residual and applied
stresses, should be limited to twice the yield
stress.
Factors Leading to Size and Shape Changes
in Heat Treated Components. Within a typi-
cal component manufacturing process, there are
seven major factors that lead to size and shape
changes and the development of residual stres-
ses in heat treated components (Ref 8):
Variation in structure and material compo-
sition throughout the component, leading to
anisotropy in properties and transformation
behavior
Movement due to relief of residual stresses
from prior machining and forming opera-
tions
Creep of the part at elevated temperature
under its own weight or as a result of fix-
turing
Large differences in section size and asym-
metric distribution of material, causing dif-
ferential heating and cooling during
quenching
Volume changes caused by phase transfor-
mation
Nonuniform heat extraction from the part
during quenching
Thermal expansion
All of these factors, except relief of prior
residual stresses (second item) and creep at
elevated temperature (third item), can be
directly related to thermal and transformation-
induced strains in the component (Ref 8).
A simple example of how this thermophysical
property affects heat treating is given in Fig. 2
(Ref 11). As the shaft is quenched, the corner
cools first, and as it shrinks, it mechanically
8 / Failure Analysis of Heat Treated Steel Components
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 8
upsets the hot steel beneath it. As the quench
progresses, the entire shaft cools, but now,
because the end is hot upset, the diameter is too
small to accommodate the circumference. As a
result, the end is (usually) in a high state of
residual tensile stress, and if the steel is brittle,
quench cracks may develop.
The coefficient of expansion is a factor that
requires serious design consideration because it
affects a part during austenitization. With fur-
nace heating, a part is heated to the austenitiza-
tion temperature mainly by radiation (80 to
98%) and partly by convection (2 to 20%). By
radiantly heating particular portions of a part,
thin sections heat fastest, especially those that
expose a large surface area, such as a spline or
gear. A typical example is the gear on the left in
Fig. 3 (Ref 11). Because its thin sections are
relatively rapidly heated, this part could not be
made to the required tolerances. The redesign
shown on the right was an improvement, but the
necessary large access holes were still trouble-
some.
Several other factors at the design stage can
contribute to problems traceable to austeniti-
zation:
Combinations of components with widely
varying (nonuniform) section sizes
Designs requiring contact with furnace
hearths or placement near walls
Designs requiring processing that results in a
state of high residual stress before austen-
itization.
Parts that are very thin or long or parts that
are large in surface area, which are difficult
to heat treat because of distortion during
austenitization
Designs that are unsuitable for the type of
furnace equipment available
In designing a tool or die, various factors must
be considered. In practice, it is difficult to
separate the design stage from steel grade
selection because the two steps are inter-
dependent. The choice of a certain grade of steel,
such as one that must be brine or water quen-
ched, will affect all aspects of design and man-
ufacture. In general, any steel grade that requires
liquid quenching demands very conservative,
careful design. Air-hardening grades tolerate
some design and manufacturing aspects that
could never be tolerated with a liquid quench-
ing. The design must also be compatible
with the equipment available, for example,
heat treatment furnaces and surface-finishing
Fig. 3
Two gear designs showing the effect of coefficient of thermal expansion. At left is a widely used design, which is very
troublesome to heat-treat. A preferred design is shown at right. Source: Ref 11
Fig. 2
Effect of coefficient of thermal expansion in heat
treating a shaft. Source: Ref 6
Component Design / 9
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 9
devices. Designing tools and dies is more diffi-
cult than designing components made from
structural steels because of the difficulty in
predicting service stresses. Despite advances
made in design procedures, much of the design
work is still empirically based. Such experience
is primarily based on past failures; therefore, it is
important that the findings of the failure analyst
be incorporated into future work. Despite the
shortcomings of the empirical approach, there
is a vast body of common-sense engineering
knowledge available for guidance.
Analysis of many tool and die failures shows
that two relatively simple design problems
cause the most failures. These design short-
comings are the presence of sharp corners and
the presence of extreme changes in section mass
(Ref 12). A sharp corner concentrates and
magnifies applied stresses, stresses that arise in
tool and die manufacturing (such as during
quenching), or stresses that occur during service.
In addition to promoting cracking during liquid
quenching, sharp corners promote buildup of
residual stresses that may not be fully relieved
by tempering and can therefore reduce service
life. The largest possible fillet should be used at
all sharp corners. Air-quenching grades of steel
are more tolerant of sharp corners than liquid-
quenching grades and are preferred when only
minimal fillets can be used. Changes in section
size can be the locus of premature failures.
Figures 4 to 8 (Ref 13) show failures caused
by design errors and selection of unsuitable
material.
Figure 4(a) shows the fracture of a lathe tool
bit made of steel with approximately 1.45% C
and 1.4% Cr that was hardened in oil at 870
C
(1600
F), which was at least 20
C (35
F) too
high. The fracture propagated from a rectangular
cross-sectional transition that was not properly
filleted and moreover was rough machined, as
shown by the grooves in Fig. 4(b) (thereby the
notch effect was further aggravated). Many
failures in service, especially those caused by
shock or cyclic loads, can be caused by such
design errors.
Figure 5 shows a bolt from a self-service
elevator that failed as a result of reverse-bending
fatigue. In this case, the fracture also propagated
from a sharp-edged cross-sectional transition.
To avoid further damage and prevent potential
Fig. 4(a, b)
Lathe tool bit of 1.45% C and 1.4% Cr steel
with acute-angled and rough-machined cross-
sectional transition that fractured during hardening. (a) Fracture.
Original magnification: 1·. (b) View into angle. 2·. Source:
Ref 13
Fig. 5 Bolt of a self-service elevator that failed as a result of revers e-bending fatigue. Source: Ref 13
10 / Failure Analysis of Heat Treated Steel Components
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 10
accidents, 24 other bolts that had not yet failed
were examined metallographically or in bending
tests. Eight of these proved to have incipient
fatigue cracks in the cross-sectional transitions.
The bolts were partially normalized and partially
heat treated. Their strength was determined from
Brinell hardness to be between 440 and
700 MPa. Cracks had occurred in the annealed
as well as the heat treated bolts, that is, in soft as
well as hard bolts. The higher strength of the heat
treated bolts was made ineffective by the unfa-
vorable design.
Figures 6(a–e) show different parts of rock
drills of different durability and made of a steel
with composition 0.95% C, 1.2% Cr, and 0.25%
Mo. They had failed after a short period of ser-
vice in the hexagonal shaft, while others had
proved free of defects. At first glance, it could be
seen that the broken drills had sharp edges
(Fig. 6a), while those free of defects were
well rounded off (Fig. 6b). Fatigue fractures
propagated from the sharp edges. These, in turn,
led to catastrophic failures under shock loading
(Fig. 6c). The design differences could be
clearly seen in the cross section (Fig. 6d,e).
Etching showed that the failed drills also were
surface decarburized, which further reduced the
fatigue strength.
Figures 7(a,b) show a compressor transmis-
sion shaft with a fracture propagating from an
acute-angled keyway, and Fig. 8 shows a drive
shaft pinion with fatigue fractures propagating
from the acute-angular edge of the helical gear.
On the other hand, ignorance, carelessness,
and false economies in the selection of materials
cause many errors (Ref 12). Not every steel user
is in a position to select the most suitable
material for his purpose from the many varieties
available. When in doubt, consult the steel
manufacturer whose materials specialists pos-
sess the necessary knowledge of mechanical and
technological properties of the required mater-
ials. A close cooperation between the materials
specialist of the producer and the designer and
plant engineers of the user is the best formula for
success.
One of the overwhelming causes of steel
cracking and unacceptable distortion control
is part design (Fig. 9–11). Poor part design
promotes distortion, cracking, and nonsym-
metrical heat transfer during heating and cool-
ing.
Sometimes, designers make designs in which
combinations of parts intended to reduce costs
can actually increase cost due to problems
during austenitization. The classic example is
the gear-and-hub combination shown in Fig. 12
(Ref 11). As this part is heated, the thin extrem-
ities at the top of the hub heat faster than the
sections near the gear. Accordingly, this area has
a propensity to increase in size but is restrained
by the colder metal nearer the gear; therefore, it
upsets itself (yields in compression). Finally, the
entire part comes to the prescribed tempera-
ture. On cooling, however, and even without
quenching, the top end of the hub pinches in
because it has upset itself. This upsetting can
result in a serious taper condition in the bore. (If
the bore is broached before heat treatment, the
extremities of the hub will stretch and then close
Fig. 6
Different parts of rock drills of different durability and made of a steel with 0.95% C, 1.2% Cr, and 0.25% Mo. (a) Broken drills
had sharp edges in the hexagonal shaft. Original magnification: 2·. (b) Drills free of defects had well-rounded-off edges.
Original magnification: 2·. (c) Fatigue fractures propagated from the sharp edges. Original magnification: 3· . (d,e) Differences are
clearly seen in the cross section of the hexagonal shaft. Etching shows that the failed drills also were surface decarburized, which further
reduced the fatigue strength. Source: Ref 13
Component Design / 11
Name ///sr-nova/Dclabs_wip/Failure_Analysis/5113_1-42.pdf/Chap_01/ 18/8/2008 2:49PM Plate # 0 pg 11