Fatigue Failures 101

torsion probably ís an indication of the (a) initiation of multíple cracks ís more

Sheorcr~ck
presence of torsional vibration. likely; (b) striation spacing is increased;
Tensron crack --', Tenslon crock Axial Loading. In each of the preví- and (e) tbe region of final fast fracture
" '- ously discussed types of loading there is increased in size.
are stress gradients within the beam; With very high overstress, low-cycle-
(o) Longitudinal shear
stresses are greatest at tbe surface of the fatigue fractures are produced. The ar-
beam, wbich increases the normal Iikeli- bitrary but commonly accepted dividing
Sheor crack~ hood for fatigue cracks to initiate at the line between high-cycle and low-cycle
Tension crack~ <,:-TenSion crack surface. In pure axialloading of a simple, fatigue is cons.idered to be about 100,000
uniform beam, however, the stress is (105) eycles. In practice this distinction
(b) Transverse sheor constant across tbe cross section oí tbe is made by determining whether tbe
Fig, 10. Schematic representation 01 the beam, and a fatigue crack may initiate dominant component of the straín im-
way in which torsional fatigue cracks in at a discontinuity within the member posed during cyclic loading is elastic
a shait may begin. in longitudinal shear rather than at the surface. The appear- (higb cycle) or plastic (low cycle), which
01· transuerse shear. Dashed fines indi- ance of fatigue cracks caused by fluctuat- in turn depends on the properties of the
cate other crachs that can appear when ing tensile loads is often similar to that metal as well as the magnitude oí the
torsional stresses are reoersed. described for bending loads. The stress nominal stress. In extreme conditions,
had a hardness of Rockwell e 44.5 to 45.5: states witbin beams subjected to the two the divíding line between high-eycle and
the light bands (alloy-lean areas) , Rockwell types oí loading are similar, although low-cycle fatigue may be even less than
e 29 to 30. Surface hardness oC the shaCts purely axial loading is rarely found in 100 cycles.
from end to end ranged Irom Rockwell e 29 service. Stress Concentratlons. Notches,
to 33; surface hardness in the fracture area Loading of flat components differs from grooves, holes, fíllets, threads, keyways
was about Rockwell e 30. that oí long, somewhat cylindrical com- and splines are common design features.
Conclusions.The shafts Cailedin reversed- ponents. In flat components, biaxial ten- AH such sectional discontinuities in-
bending and torsional fatigue that resulted sion is more cornmon and torsion less crease the local stress level aboye that
from high stress conccntration at a low estimated on the basis oí minimum cross-
overstress. common. Under conditions of biaxial
Corrective Measure. The ahaft diameter tensión, fatigue life depends on the maxi- sectional area. In addition to the reduc-
in the failure area was increased Crom 1 in. mum shear stress rather than the prin- tion oí fatigue strength or fatigue life,
to 1.125 jn. Ior subsequent shafts, which cipal tensile stresses. Thus, fatigue increasing the severity of stress oonoen-
reduced the nominal stresses by about 25% Cailure may not occur when the loading tration has the following effects on fa-
and was expected to provide a safe operat- conditions result in low shear stresses but tigue-crack features: (a) in.itiation of
ing condition. high tensile stresses. multiple cracks is more likely; (b) beach
Torslonal Loadlng. Under torsional In sheet or plate materials, the fatígue- marks usually become oonvex toward tbe
loading oí a shaft, the maximum local crack front may extend under plane- point of crack origin; (e) under rotational
tensile stresses are at 45° 10 the axis of the strain conditions to give a wholly ñat- loading the beach marks may completely
shaft. Under a ftuctuating torsional load, Cace fracture (see Fig. 11a). Fatigue surround the final-fracture zone; and
fatigue cracks may develop normal to the fractures in very thin sheet subjected to (d) combined stress states may be íntro-
tensile stresses. Under altemating (re- high stress intensities may shift from duced, thereby influencing the direction
versed) torsional loading (Sm O,R = = flat face (plane-strain eonditions) to
shear face (plane-stress conditions), as
of crack growth.
Figure 12 schematicaJly shows how
-1), two sets oí fatigue cracks, perpen-
dicular to each other, may develop. shown in Fig. 11 (b). magn.itude of nominal stress, severity of
TorsionaI fatigue cracks can begin in stress concentration, and type of loading
longitudinal shear (Fig. lOa) or trans- Effect of Overstress and affect the appearance of fatigue-fracture
verse shear (Fig. 10b); the relatively Stress Concentration surfaces of components with round,
equal length of thc cracks in each pair The magnitude oí the nominal stress square and rectangular cross sections,
in F.ig. 10 indicates that the equal and on a cyclically loaded component is and those of thick plates. The chart in
opposite stresses have occurred during often measured by the amount oí over- Fig. 12 is in tended for use only as a
loading. At later stages oí fatigue-crack stress - that is, the amount by which guide, Deviations froro this chart will be
growth, one crack of a pair usually grows the nominal stress exceeds the fatigue found for various material, test and ser-
much fas ter than the other and even- limít or the long-life fatigue strength of vice conditions, Figure 12 is based on
tually causes rupture of the shaft. Beams the material used in the component. The the following principies:
that are subjected to ftuctuating torsión number of load cycles that a component 1 As local stress increases in regions of
(S", < So, R > O) will typically show fa- under low overstress can endure is high; potential crack initiation, so does the
tigue cracks in only one direction; the thus, the term high-cycle fatigue is often number of active crack nuclei or initia,
presence of perpendicular fatigue cracks applied. Increasing the magnitude of the tion sites. Thereiore, at high overstress
in a component subjected to ftuctuating nominal stress has the following effects: or in the presence oi a severe stress con-

(a) ( b)

Fig, 11. Fatigue-fracture zones in aluminum alloy 7075-T6 plates. (a) Fatigue crack that grew as a flat-face fracture with a shallow
conuex crack Iront. (b) Change in orientation of fatigue fracture from plane strain (at A) to plane stress (at B).
 102 Fatigue Failures
r----- High nominal stress ------, ,------- Low nominal stress -----~ centration, multiple crack origina :\ViII
No st,ess Mild stress Severe stress No stress Mlld stress Severe stress be seen. In most instanoes tbe cracks
"'--concentrotion---' 'Conc.nlraHon~ '-concentroflon-" r+eoncentrQ'ion~ '--concent'a,lon~ '-concentrotion-' from these origina will eventually unite
to form a single crack front. Before the
single crack front is formed, the in-
dividual microcracks will be separated
by small, vertical ledges, caUedratchet
marks. Alternatively, at just aboye the
fatigue limit, or minimum stress for
fracture, a single origin wiIl occur and
the entire fracture will emanate from
that point.
2 In the absence of stress concentrations
at the surface, cracks propagate more
rapidly near the center of a section
than at the surface. Tbis occursbecause
deformation constraints cause the
stresses to be triaxial and more severe
away from the surface. However,when
there is a stress-concentrating notch at
the surlace (such as a thread with a
sharp root), the stress near this notch
may be more severe than it is farther
below the surfaee, Under conditions of
severe notching, W-shape crack fronts
will sometimesbe observed.
3 For a given material, the size of the

® •
region of catastrophic fracture (or final,
fast fracture) relative to the size of the
region of subcritical crack propagation
,
,
,
,,- . . ..~... ,..
',(::
,.;.
,,~. will increase as nominal stress in-
~ creases. Under an overload that is

~m~
slightly more than adequate to cause
fracture, the region of final, fast frac-
ture wiIl be relatively smaIl; under a
. ... ········.··
' .... ' ...
much higher applied stress, this región
.
EJ· will be relatively large.

~~m
4 In a fracture causad by rotating bend-
ing, the final-fracture region will often
be rotated, or offset, toward the origin
in a direction opposite to the direction
of rotation. Also, all other conditions
being tbe same, tbe region of final frac-
ture will movetoward the center of tbe
section as the nominal stress increases.
5 Fracture initiation usually occurs at or
near the surlace, because in most engi-
neering situations such as in bending or
when stress concentrations are present,
the surface is subjected to the greatest
stress. Subsurface origins have been ob-
served in tension-tension or tension-
compression fatigue, or in Hertzian-
stress situations sucb as roIling-contact
fatigue of bearings and gear teeth, if
there is a large inclusion or ímperfec-
tion below the surface in tbe interior of
the specimen or part, but this is un-
usual. Crack initiation at comers or at
the ends of drilled holes may result from
the presence of burrs remaining at
these locations after machining.

ct)~.~
'- Rototional bendlng
•• -'
Effect of Frequency of Loading
Exclusive of environmental effects,
there are no distinguishing surface fea-
tures of a fatigue fracture produced at
high frequency that differentiate it from

~ ffi3* OO~ OO~

other types of fatigue fractures during
either visual or Iíght-microsoopeexam-
ination. At best, when examined with a
'----- --Torsion light microscope,a fracture surface ere- .
em Fost froeture zone
4
CJ Streas-ccncentrntionnotch ated at a high frequency will have a
Fig. 1'2. Sehernatio repreeerüation. o] marhe on eurjaces 01 fatigue fractures produced brittle appearanee, showing mostly a
in smooth and notched components with round, square and rectangular cross sections, platelike structure throughout the fa-
and in thick platee, under uarious loading conditions at high and: low nominal stress tigue zone. Beach marks may or may
 Fatigue Failures 103
not be present, depending on whether strength to the static yield strength, For
crack growth was steady or intennittent many purposes it is essential that the
or ií load variatíons occurred. The frac- stanc yíeld strengtb not be exceeded,
ture surface shown in Fig. 13(a) was pro- and this relation is intended to fuIñll
duced by subjecting a 0.020-in.-thick the conditions that neither fatigue failure
panel of aluminum alloy 7075- T6 to nor yielding occurs. The relations may be
high-frequency (200 Hz) vibration with written mathematically as:
superimposed loading at lower frequency Modi1ied Goodman law:
(1 Hz). Figure 13(b) shows the charac-
teristic platelike structure and well-de-
S« = S[1 - (SmIS ..)]
Gerber's law: S« = S[1 - (Sm/ S..) 2]
ñned striatíons that were created as a
Soderberg's law: S« = S[1 - (Sm/ SI/)]
result of the superimposed low-fre-
quency loads. Examination of the frac- where S« is the altemating stress associ-
ture surface with a scanning electrón ated with a mean stress S"" S is the alter-
microseope revealed fínely spaced (5 to nating fatigue strength, Su is the tensile
6 mícro-ín.) stríations attributable to strength; and 511 is the yield strength.
crack growtb resulting from the stresses An understanding of the Goodman, or
induced during high-frequency vibra- constan t-lí fe, diagram has resulted in
tion (Fig. 13c). The narrowly spaoed many varied and useful treatments for
striations are expected at high-frequency improving fatigue lile. According to a
loading because in each cycle the length constant-Iife diagram, an example oí
of time at peak loading is very short, and which is shown in Fig. 41 on page 120,
the increment of crack growth per eycle increased tension decreases the fatigue
is correspondingly small. lüe and increased compression increases
it. Because rnost cracks originate at the
Effed of Stress on surfaee of the part, placing the surfaee
Fatigue Strength under compressive stress should be
beneficial. Recognition of this has re-
Fatigue cracks generally form prefer- sulted in development of such surlace
entially at the surface, because tbe level treatments as nitriding, carburizing, shot
of stress generally is higher at the sur- peening, surface rolling and overstress-
face. Experimental results indicate that ing. When these treatments are prop-
fatigue can occur under high vacuum and erly applied, the surfaee is in a state oí
at low temperature, suggesting that the residual compression. However, if not
primary mecbanism of fatigue does not properly applied, they can have a detri-
involve corrosive attack or thermal acti- (a) and (b) Light Iractographs ot part o{ the
mental effect on fatigue )jfe.
vation, although both may contribute to Iracture suríaee at 20X and SOOX,respectively; Stress Amplltude. Because stress am-
final failure. (b) shows platelike strueture with striations plitude wiIl vary widely under actual
Mean Stress. A series of fatigue tests created by low-Irequency loading. (e) SEM Irac- loading conditions, it ís necessary to
tograph o{ part oí the fracture surface, at 5OOOX,
can be conducted at various mean predict fatigue life under various stress
showing fine striations between coarse ones.
stresscs and the results plotted as a series amplitudes. The most widely used
of S-N curves. For design purposes, it is Fig, 13. Faügue-jracture surlace 01 an method of estimating fatigue under com-
more useful to know how the mean stress aluminum alloy 7075-T6 panel, 0.020 in. plex loading is provided by the linear-
thich, produced by high-frequency uibra-
affects the permissible altemating stress tions with superimposed loadings at damage law, This is a hypothesis first
amplitude for a given life (number of louier [requency suggested by Palrngren and restated by
cycles). This usually is aceomplished by Miner, and is sometimes known as
plotting the allowable stress amplitude Miner's law (Ref 1). The assumption is
íor a specific number of cyc1es as a func- íAlternollng faligue slrenglh,S made that the application of 1li cycles at
tion oí the associated mean stress. At a stress amplitude SI, for which the aver-
zero mean stress, the allowable stress
r Gerber 's porobolo age number of cycles to failure is NI,
amplitude is the effective fatigue limit Mod,f,ed Goodmon Ilne causes an amount of fatigue damage that
for a specified number of cycles. As the is measured by the cumulative cycle
ratio 711/NI, and that failure will occur

=:
mean stress increases, the permissible
amplitudes steadily decrease until at a
Tensíle
when I(n;/ NI) =l.
mean stress equal to the ultimate tensile This method is not applicable in aIl
strength oí the material, the permissible cases, and numerous alternative theories
amplitude is zero, of cumulative linear damage have been
The two straight lines and the curve suggested (Ref 2). Some considerations
shown in Fig. 14 represent the three most Meon stress, S., of redistribution of stresses have been
widely used empirical relations. The Fig. 14. Effect 01 mean stress on the clarifíed, but there is as yet no approach
straight line joining the altemating fa- alterruuing stress amplitude, as shown that seems satisfactory in all situations.
tigue strength to the tensile strength is by the moditied Goodman line, Gerber'e The effect of varying the stress amplí-
the modified Goodman law. Goodman's parabola, and Soderberg lineo See texto tude (linear damage) may be evaluated
originallaw, which is no longer used, in- experimentally by means oí a test in
cluded the assumption that the altemat- that the early experiments of Wohler which a given number of stress cycles are
ing fatigue limit was equal to one-third of fitted closely to a parabolic relation, and applied to a test piece at one stress ampli-
the tensile strength; this has since been this is known as Gerber's parabola (curve tude and the test is continued to fracture
modified to the relation shown in Fig. 14, in Fig. 14). 'The third relation, known as at a different amplitude. Alternatively,
using tbe altemating fatigue strength de- Soderberg's law, is given in Fig. 14 by the the stress may be changed from one stress
termined experimentally. Gerber found straight line from the altemating fatigue amplitude to another at regular ínter-
104 Fatigue Failures
vals; such tests are known as block, or stresses where SI ~ S2 ~ 83, and if 8 is that OCCUl'S. (See the section on Concepts
interval, tests. These tests do not simu- the alternating uniaxial fatigue strength, Related to Fatigue, on page 95 in this
late service conditions, but may serve then the following criteria apply: article.) In sorne sítuatíons, values oí K«
a useful purpose for assessing the 1 Maximum principal-stress criterion can be calculated using the theory of
linear-damage law and indicating ita SI =S elasticity or can be measured using
limitations. Resulta of tests by several in- 2 Maximum shear-stress criterion SI- photoelastic plastic models. Many of
vestigators have shown that initial over- S3=S these values are reported in standard
stressing reduces both the fatigue limit 3 Shear-strain energy (SI - S2)2 + (S2- references (Ref 4 to 7).
and the subsequent fatigue life at stresses S3)2 + (S3 -SI)2 =S2 The mathematical theory of elasticity
aboye the fatigue limito The resulta also 4 Maximum principal-strain criterion is based on an ideal isotropic material
show that a slight overstress does nol =
SI -1l(S2 + S3) S, wherc Il is Pois- free of any internal díscontinuity, a
son's ratio.
markedly reduce the fatigue limit, even ir strictly accurate profile, and an increase
continued COI' a large proportion of the Because fatigue cracks usually propa- in stress concentration due solely to the
normal life of a material. However, this gate from the surface, where one oí the presence of the surface discontinuity. In
is not true Ior a high overstrcss. principal stresses is zero, only biaxial actual parta, the stress intensification is
Residual Stress. Fatigue fractures gen- stresses need be considered. When tbe affected not only by the surface disoon-
erally propagate from the surface. two principal stresses are of the same tinuity but also, to an undetermined ex-
Processing operations, such as grinding, algebraic sign, the criteria oí maximum tent, by the size oí the part, by local
polishing and machining, that work principal stress and maximum shear readjustments of stress because of plastíc
barden 01' increase residual stress on stress give the same ralationship. If tbe yielding, by surface roughness, and by
the surface can intluence the fatigue principal stresses are oí opposite alge- the heterogeneous structure of the ma-
strength, although there is no generalized braic sign, then aIl criteria give a dif- terial Itself, including anisotropy and in-
Iormulation that will predict the extent ferent relationship. Itis often convenient herent internal discontinuities. There-
of improved fatigue strength that can be to determine the suitabilíty of the vari- fore, the deleterious effect of a stress
derived from work hardening and resid- ous criteria by comparing the fatigue raiser on a part usualIy is determined ex-
ual stress. Compressive residual surtaco strength in torsion t and bending b. No perimentally and expressed in terms of a
stresses generally increase the fatigue single criterion adequately describes the fatigue notch factor, K¡. This is the ratio
strength, but tensile residual surface general behavior of the stresses; for duc- of the fatigue strength without stress
stresses do noto There may be a gradual tile materials, the closest correlation concentration to fatigue strength with
decrease in residual stress if the cyclic with the experimental results of (t/b) is stress concentration. In general, experi-
stresses cause sorne plastic deformation. provided by the shear-strain-energy cri- mentally determined values of K¡ are
Cornpressive residual surface stress pro- terion (Ref 3). somewhat less than the values of Kc cal-
vides greater improvement in the fatigue Frequency. The frequency range of culated for the same specímens.
strength of harder materials (like alloy 500 to 10,000 cycles per minute is gen- Under static loading conditions a
spring steel), and in softer materials (Iike erally employed in a fatigue test. In tbis stress raiser has little 01' no effect in the
low-carbon steel) work hardening effec- range the fatigue strength of most ma- majority of situations, provided that the
tively improves fatigue strength. This ís terials, based on a given number of cycles material is ductile. If the stress locally
because the harder material can sustain a to fracture, is little affected by frequency. exceeds the yield strength of the ma-
high leve) oí residual elastic surface In general, there ís a slight decrease in terial, plastic deformation occurs and
stress, and the tensile strength (and thus fatigue strength with decrease in fre- there is a redistribution of stress. Pro-
the fatigue limit) of the soíter material is quency, because the fatigue Iimit may vided the amount of plastic deformation
irnproved by work hardening. be related to the amount of plastic de- is not excessive, no adverse effect need be
In a notched high-strength steel , the formation that occurs during the stress anticipated. But if the part is subjected
beneficial effect of prestretching and the cycle. For instan ce, at high frequency, to fluctuating or alternating stresses,
detrimental effect of precompression are there is less relaxatíon time during each then, if a fatigue crack nucleates at a
much greater than in a plain carbon steel stress cycle for deformation to occur, stress below the yield strength of the
beca use oí the type of residual stress which resulta in less damagc, For steel, material, stress redistribution by plastíc
present at the notch. A compressive resid- the fatigue Iimit is not affected between yielding will not occur and the full effect
ual stress introduced during quenching 200 and 5000 cycles per minute. How- of the stress raiser will occur.
from a tempering tempera tu re will in- ever, at high frequencies up to 100,000 Stress concentrations affect the fatigue
crease the fatigue strength, particularly cycles per minute, steels that are 100% behavior oí different materials differ-
in notched spccimens. ferritic show marked increase in fatigue ently. For instance, relatively brittle
In general, residual stresses are intro- strength. In nonferrous metals, the fa- materíals such as quenched-and-tern-
duced by (a) misfit of structural parts; tigue strength increases continuously pered steels are more susceptible to the
(b) a chango in the specifíc volume of a with increase in frequency; in plain car- effects oí stress raisers than are ductile
metal accompanying phase changes; (e) bon steel, the fatigue strength reaches a materials su eh as normalized 01' annealed
a changa in shape following plastic defor- maximum value, then decreases with an stecls; also, cast irons, containing in-
mation; 01' (d) thermal stresses resulting incroase in frequency. numerable internal stress raisers, show
from rapid temperature changes such as )jttle furthct, adverse effects from exter-
occur in quenching. Effect of Stress Concentrations nally introduced stress raisers.
The inOuence of residual stress on fa- on Fatigue Strength Distrlbution of Stress. To visualize the
tigue strength is, in principie, similar to dislribution oC stress at a change in sec-
that of an extel'llally applied static stress. Fatigue cracks usually start at sorne tion size or shape, it is helpful to con-
A static compressive surface stress in- region of stress concentration resulting sider the part in terms oC electricity flow-
creases the fatigue strength and static from the presence of surface discontinui- ing through a conductor of similar croas
tensile surface stress reduces it. ties (stress raisers) , such as a step or section. In diagrammatic form, stress·
Complex Stresses. The criteria of shoulder, a screw thread, an oil hole or a flow can be represented as a series of
s~tic faill.1re have been applied to fa- boltho)e, or a surface f1aw. The stress- parallel lines, the stress being inverse)y
tigue failure. If SI, S2 and S3 are the concentration factor, Kt, oC the discon- proportional to the distance between the
amplitudes oí the principal alternating LinuiLyis a measure oC int.ensity of stress lines; that is, the lines bunch together
 Fatigue Failures 105
in regions of hígh stress. The flow of
stress associatedwith severa!of the stress
raisers typically found in parts in service
is shown in Fig. 15.
. Progressive increases in stress with
decreasing fillet radii are shown in Fig. (o) (b) (e)
15(a), (b) and (e), and tbe relativemagni-
tude and distribution of stress resulting
from uniform loading of tbese parts is
indicated in Fig. 15(d), (e) and (f).
Stress causedby the presence of an in-
tegral collar of considerable widtb is
shownin Fig. 15(g); Fig. 15(b) showsthe (d) (e) (I)
decrease in stress concentration that ac-
companies a decrease in collar width.
Stress conditions are very similar when
collars or similar parts are pressed or
shrunk into posítion. The stress flow at
the junction of a bolt head and a shank is
as represented in Fig. 15(j). (g) (h)
A single notch introduces a consider-
ably greater stress-eoncentration efiect
than does a continuous thread: the rea-
son for this is clear when the stress flow
is considered. The stress-concentration
efieet of a single sharp notch is as shown
in Fig. 15(k). The stress concentration
(k) (m) (n)
at the right of the arrow in Fig. 15(m) is
very similar to tbat in the narrow collar Fig, 15. Effect 01stress raisers on stress concerüration and distribution 01stress at several
in Fig. 15(h), becauseof the mutual relief changes 01 [orm in componente. See text [or discussion:
afforded by adjacent threads. To the left
of the arrow, however, tbe last thread is tangular slot in the base were abraded from sequently broken off and another crack was
relieved fromone side only and in conse- rubbing against the mating parto The sides observed in the casting; however, it was
quence there is a considerablestress con- oí the elongated hole in the slotted arm remote from and unrelated to the fracture
centration, similar to that of the single were polisbed by motion 01 the mating mem, under investigation. The lever had been
notch in Fig. 15(k). This is why bolts ber, There was negligible wear in tbe round magnetic-particle inspected during manu-
so frequentIy fracture through the last hole and in the elongated holes where frac- facture, and the crack was of a size that
full thread. ture occurred. should have been detected.
The crack initiated at the sharp córner The fracture surface under investigation
The effect of a groove or gouge on of the milled slot and propagated across to (see Fig. 16b) had beach marks initiating at
stress concentration (Fig. 15n) is less the outer wall, as shown in detail A in Fig. the sharp corner along tbe milled slot.
severe than that of a sharp notch, A 16. The surfaces of the part at the crack Changes in frequency or amplitude of vibra-
series oí grooves will have an effect were slightly offset, indicating some plastic tion caused different rates of propagation,
similar to that shown in Fig. 15(m). deformation, This deformation could have resulting in a change in pattem. Motion of
In the following example, a sharp occurred after tbe failure and during dis- the part after failure occurred caused sorne
comer in a milled slot was tbe point oí assembly of the lever from tbe fuel control. obliteration oí the fracture characteristícs.
stress concentration and the site of frac- To examine the fracture surface, tbe sec- Conclusions. The lever failed in fatigue as
ture initiation. tions were spread apart ahout 30·, result- a result of excessive vibration of the fuel
ing in a slight tearing in one corner of the control on the test engine. Cracking initiated
Example 2. Fatigue Fracture 01 a Cast Stain- section surrounding the elongated holeo in a sharp comer of a milled slot.
less Steel Lever That Was Initiated Bending of the part was easily done, indi- Correctíve Measure. Conditions on the test
by Stress Concentration at a cating tbat the general ductility of the part stand were more severe than those that tbe
Sharp Comer (Fig. 16) was satisfactory. The bent section was sub- engine was expected to survive in service

A main fuel control was returned to the

factory for examination after service on a Type 410 stolnless
steel costlng
test aircraft engine that bad experienced
high vibrations. There were no apparent
Rockwell e 26 to 31

problema with the fuel control, but it was Primor y

removed for examination to ensure that no crack
. problema had developed as a result of the
vibrations. When the fuel control was dis-
assembled, the lever shown in Fig. 16(a)
was found to be cracked.
The lever was cast from AMS 5350 (type
410) stainless steel, through-hardened to
Rockwell C 26 to 32, and passivated.
. Investigation. There were no corrosion
producís or stains on the lever; however, a
slight bluish cast, probably caused during
(o) Fuel-cantral leve r
chemical etching of the part number, was
found on the slotted ann. The general condi- Fig. 16. (a) Cast stainless steel juel-control leuer that [ailed when the aircraft engine in
tion of several holes and surfaces of the lever uihich.it was mounted experienced excessive uibration. (b) View 01 a fracture surjace;
'Vas good, although tbe sides of the rec- arrow indicates region.01fracture origino (Example 2)
106 Fatigue Failures
Two cadmium-plated flanged bushings, mum at the surface of the shaft). In re-
made of copper alloy 630 (aluminum gions near the keyway (zone B) the stress
bronze), were press-fitted into each bored distribution is distorted by the stresses
hole in the lug. A space between the bush- caúsed by contact between the key and
ings provided an annular groovefor a lubrí- keyway, resulting in maximum torsional
cant. A lubrication hole extended.from the
outer surface to the bore of the lug, as shown stress below the surface. When the end oí
in the "Original design" view in detail A in a keyway is formed with a sharp step, a
Fig.17. further concentration of stress occurs,
Investigation.Initial study of the fracture especially in the presence of bending '
surlaces indicated that the primary frac- stresses, and cracking very often begins
ture initiated from multiple origins on both at this point. This further stress concen-
sides of the lubrication hole, as shown at tration can be reduced considerably by
arrows in Fig. 17(b). Beach marks on the using a sled-runner type of keyway, in
fracture surlace indicate that cracking was which the transition at the end of the
initiated and propagated by fatigue until an
overloadstage was reached. keyway is gradual.
A fracture on the opposite side of the lug, Reduction 01 Stresses. The detri-
shown as "Secondary fracture" in detail A mental effeet of stress concentrations can
in Fig. 17,appeared to be typical for a static be reduced or eliminated by the use of
Secondory fracture overloadfailure. induced stresses. These stresses may be
A sectionwastaken normal to the fracture introduced by plastic deíormation of the
(o) Original design Improved design surface and in the region contacted by the surface, such as in cold working and shot
bushing flange, and prepared for metal- peening, by phase transformation, such
lographic examination. Results showed as in case hardening, or by proof Ioading
small fatigue-type cracks in the hole adja-
cent to the origin of primary fracture. There with a load high enough to cause local
were no indications of inclusions or dis- plastic flow in the notch but low enough
continuities in the plane of the section that to avoid general yielding.
could contribute to the initiation of lug Extreme or incorreet cold working
fracture. may have adverse effects under fatigue
Hardness of the material at the face of conditions, because it may give rise to
the lug adjacent to the fracture surface was minute cracks in the surface of the ma-
Rockwell B 85.5. Electrical conductivity of terial or at least make the surface prone
the material was 39% rAeS, typical for
aluminum alloy 7075. to such cracks. If the correet amount of
Conclusions.The arm faíled in fatigue cold working is applied, however, im-
cracking that initiated on each side of the provement in the fatigue strength is ap-
lubrication holeoNo material defects were preeiable. For instance, a screw thread
found at the failure origino produced by rolling is more resistant to
CorrectiveMeasures.The location of the fatigue failure than one that has been
lubrication hole was changed as shown in cut. Surface rolling and shot peeníng,
the "Improved design" view in detail A in especially oí springs, are other examples
Fig. 17. The faces of the lug were shot oí the successful application of cold
peened Ior added resistance to fatigue fail-
ure. Also, the forging material was changed working. Apart from the increase in
to aluminum alloy 7175-T736for its higher fatigue strength of the surface of the
mechanical properties. material arising from work hardening,
Fig. 17. (a) Landing-gear torque-arm as- the stress distribution is modified con-
sembly 01 aluminum alloy 7075-T73 that Keyways in components frequentlyact siderably. Figure 19(a) shows the stress
toas redesigned to elimina te fatigue frac- as stress raisers. When the stress is pre-
ture at a lubrication holeo (b) Fracture distribution in a specimen notched by
dominantly torsional, fatigue cracks machining and then subjeeted to a ten-
surjace, at abou,t 2x, shouiing fatigue
beach marks. (Example 3) usually start in the fillets of the keyway, sile load L = 1. This load just starts to
which is to be expected from the dis- produce yielding at the root of the notch,
turbance of the stress flow caused by the which has a stress-concentration factor of
without major damage and were solely re-
sponsible for failure of the lever. If any presence of the keyway. The transition of 3.0. Figure 19(b) shows the same speci-
corrective measure at all were necessary on torsional stress as it reaches a keyway is =
men loaded to L 1.5, which produces
the part, it wouldbe to specifv a large radius shown by comparison of shaded zones A yielding to a greater depth. The result of
in the corner where cracking originated. and B in Fig. 18. The distribution of tor- decreasing the load to L = 1 is shown in
This wouId reduce the stress-concentration sional stress in regions away from the Fig. 19(c). The beneficial effect of the
factor significantly, thus minimizing the keyway (zone A) is smooth and wedge- overload is indicated by the reduction of
susceptibility of the leverto fatigue. shaped (stress is zero at center and maxi- peak stress. The residual-stress distrí-
In the following example, failure oc- bution after removal of the externalload
curred at a lubrication hole located at a =
(L O) is shown in Fig. 19(d). The
region of high stress concentration. stress distribution produced in a speci-
men in which the notches have been shot
Example3.Fatigue Fracture of an Aluminum peened is shown in Fig. 19(e). The stress
AUoy7075-T73Landing-GearTorque Arm in the shot-peened speeimen after appli-
That Originatedat a LubricationBole
in the Lug (Fig. 17) cation of load is shown in Fig. 19(f).
If the amount of cold working is so
The torque-arm assembly for an aircraft adjusted that the residual compressive
nose landing gear, shown in Fig. 17(a),
failed after 22,779 simulated flights. The stress in the part effectively reduces the
part, made from an aluminum alloy 7075- Eig, 18. Comparison. 01 the distribution 01 applied peak tensile stress below the fa-
T73 forging, had an expected life oí 100,000 torsional stresses in a shaft at and away tigue limit of the material, fatigue failure
simulated t1ights. [rom. a heyuiay, See texto wiIl not occur. It is difficult in practice,