The fatigue properties of metals are quite structure-sensitive. However, at the present time there are only a limited number of ways in which the fatigue properties can be improved by metallurgical means. By far the greatest improvements in fatigue performance result from design changes, which reduce stress concentration and from the intelligent use of beneficial compressive residual stress, rather than from a change in material. Nevertheless, there are certain metallurgical factors, which must be considered to ensure the best fatigue performance from a particular metal or alloy.
The fatigue properties of metals are
quite structure-sensitive. However, at the present time there are only a limited
number of ways in which the fatigue properties can be improved by metallurgical
means. By far the greatest improvements in fatigue performance result from
design changes, which reduce stress concentration and from the intelligent use
of beneficial compressive residual stress, rather than from a change in
material. Nevertheless, there are certain metallurgical factors, which must be
considered to ensure the best fatigue performance from a particular metal or
alloy.
Fatigue tests designed to measure the
effect of some metallurgical variable, such as special heat treatments, on
fatigue performance are usually made with smooth, polished specimens under
completely reversed stress conditions. It is usually assumed that any changes in
fatigue properties due to metallurgical factors will also occur to about the
same extent under more complex fatigue conditions, as with notched specimens
under combined stresses.
Fatigue properties are frequently
correlated with tensile properties. In general, the fatigue limit of cast and
wrought steels is approximately 50 percent of the ultimate tensile strength. The
ratio of the fatigue limit (or the fatigue strength at 106 cycles) to
the tensile strength is called the fatigue ratio.
Several nonferrous metals such as
nickel, copper, and magnesium have a fatigue ratio of about 0.35. While the use
of correlations of this type is convenient, it should be clearly understood that
these constant factors between fatigue limit and tensile strength are only
approximations and hold only for the restricted condition of smooth, polished
specimens which have been tested under zero mean stress at room temperature.
For notched fatigue specimens the
fatigue ratio for steel will be around 0,20 to 0,30. However, as yield strength
is increased by the various strengthening mechanisms, the fatigue limit usually
does not increase proportionately. Most high-strength materials are
fatigue-limited.
Several parallels can be drawn between
the effect of certain metallurgical variables on fatigue properties and the
effect of these same variables on tensile properties. The effect of
solid-solution alloying additions on the fatigue properties of iron and aluminum
parallels nearly exactly their effect on the tensile properties. Gensamer showed
that the fatigue limit of a eutectoid steel increased with decreasing
isothermal-reaction temperature in the same fashion as did the yield strength
and the tensile strength.
However, the greater structure
sensitivity of fatigue properties, compared with tensile properties, is shown in
tests comparing the fatigue limit of a plain carbon eutectoid steel heat-treated
to coarse pearlite and to spheroidite of the same tensile strength. Even though
the steel in the two structural conditions had the same tensile strength, the
pearlitic structure resulted in a significantly lower fatigue limit due to the
higher notch effects of the carbide lamellae in pearlite.
There is good evidence that high fatigue
resistance can be achieved by homogenizing slip deformation so that local
concentrations of plastic deformation are avoided. This is in agreement with the
observation that fatigue strength is directly proportional to the difficulty of
dislocation cross slip.
Materials with high stacking-fault
energy permit dislocations to cross slip easily around obstacles, which promotes
slip-band formation and large plastic zones at the tips of cracks. Both of these
phenomena promote the initiation and propagation of fatigue cracks. In materials
with low stacking-fault energy, cross slip is difficult and dislocations are
constrained to move in a more planar fashion. This limits local concentrations
of plastic deformation and suppresses fatigue damage.
While the concept has been useful in
understanding fatigue mechanisms, the ability to control fatigue strength by
altering stacking-fault energy has practical limitations. A more promising
approach to increasing fatigue strength appears to be the control of
microstructure through thermomechanical processing to promote homogeneous slip
with many small regions of plastic deformation as opposed to a smaller number of
regions of extensive slip.
The dependence of fatigue life on grain
size varies also depending on the deformation mode. Grain size has its greatest
effect on fatigue life in the low-stress, high-cycle regime in which stage 1
cracking predominates. In high stacking-fault-energy materials (such as aluminum
and copper) cell structures develop readily and these control the stage 1 crack
propagation. Thus, the dislocation cell structure masks the influence of grain
size, and fatigue life at constant stress is insensitive to grain size. However,
in a low slacking-fault-energy material (such as alpha brass) the absence of
cell structure because of planar slip causes the grain boundaries to control the
rate of cracking. In this case, fatigue life is proportional to grain diameter.
In general, quenched and tempered
microstructures result in the optimum fatigue properties in heat-treated
low-alloy steels. However, at a hardness level above about Rc 40, a bainitic
structure produced by austempering results in better fatigue properties than a
quenched and tempered structure with the same hardness. Electron micrographs
indicate that the poor performance of the quenched and tempered structure is the
result of the stress-concentration effects of the thin carbide films that are
formed during the tempering of martensite.
For quenched and tempered steels the
fatigue limit increases with decreasing tempering temperature up to a hardness
of Rc 45 to Rc 55, depend on the steel. The fatigue properties at high hardness
levels are extremely sensitive to surface preparation, residual stresses, and
inclusions. The presence of only a trace of decarburization on the surface may
drastically reduce the fatigue properties. Only a small amount of non-martensitic
transformation products can cause an appreciable reduction in the fatigue limit.
The influence of small amounts of retained austenite on the fatigue properties
of quenched and tempered steels has not been well established.
The results indicate that below a
tensile strength of about 200,000 psi (~1400 MPa) the fatigue limits of quenched
and tempered low-alloy steels of different chemical composition are about
equivalent when the steels are tempered to the same tensile strength. This
generalization holds for fatigue properties determined in the longitudinal
direction of wrought products. However, tests have shown that the fatigue limit
in the transverse direction of steel forcing may be only 60 to 70 percent of the
longitudinal fatigue limit. It has been established that practically all the
fatigue failures in transverse specimens start at nonmetalic inclusions.
Nearly complete elimination of
inclusions by vacuum melting produces a considerable increase in the transverse
fatigue limit. The low transverse fatigue limit in steels containing inclusions
is generally attributed to stress concentration at the inclusions, which can be
quite high if an elongated inclusion stringer is oriented transverse to the
principal tensile stress.
However, the fact that nearly complete
elimination of inclusions by vacuum melting still results in appreciable
anisotropy of the fatigue limit indicates that other factors may be important.
Further investigations of this subject have shown that appreciable changes in
the transverse fatigue limit which cannot be correlated with changes in the
type, number, or size of inclusions are produced by different deoxidation
practices. Transverse fatigue properties appear to be one of the most
structure-sensitive engineering properties.
The existence of a fatigue limit in
certain materials, especially iron and, titanium alloys, has been shown to
depend on the presence of interstitial elements. The S-N curve for a pure metal
will be a monotonic function with N increasing as stress decreases. The
introduction of a solute element raises the yield strength and since it is more
difficult to initiate a slip band, the S-N curve is shifted upward and to the
right. If the alloy has suitable interstitial content so it undergoes strain
aging, there is an additional strengthening mechanism.
Since strain aging will not be a strong
function of applied stress, there will be some limiting stress at which a
balance occurs between fatigue damage and localized strengthening due to strain
aging. With enhanced strain aging, brought about by higher interstitial content
or elevated temperature the fatigue limit is raised and the break in the curve
occurs at a lower number of cycles. In quenched and tempered steels, which do
not normally exhibit strain aging in the tension test, the existence of a
pronounced fatigue limit presumably is the result of localized strain aging at
the tip of the crack.