-
General Aspects of Failure AnalysisWaldek Wladimir Bose-Filho
and Jose Ricardo Tarpani,Universidade de Sao PauloMarcelo Tadeu
Milan, Instituto de Materiais Tecnologicos doBrasil Ltda.
FAILURE ANALYSIS is the process ofcollecting, examining, and
interpreting damageevidence. The objective is to understand
thepossible conditions leading to a failure andperhaps prevent
similar failures in the future.A failure analysis should provide a
well-documented chain of evidence that eitherexcludes or supports
possible interpretation ofthe damage evidence. Clear-cut
conclusionsdo not always occur, and the tendency ofdeveloping
preconceived interpretations shouldbe avoided.Various publications
(e.g., Ref 16) describe
the guidelines and methods of failure analysis,and this chapter
briefly outlines some of thebasic aspects of failure analysis. The
first sectiondescribes some of the basic steps and majorconcerns in
conducting a failure analysis. Thisis followed by a brief review of
failure typesfrom fracture, distortion, wear, and
corrosion.Fracture is a common damage feature, becausethe vast
majority of mechanical failures involvecrack propagationtypically
classified as duc-tile, brittle, and fatigue, as briefly
describedin more detail. Distortion, wear, and corrosionalso can be
important damage factors in failureanalysis.
General Guidelines of Failure Analysis
For a complete evaluation, the sequence ofstages in the
investigation and analysis of fail-ure, as detailed in Ref 5, is as
follows (Ref 2):
1. Collection of background data and selectionof samples
2. Preliminary examination of the failed part3. Nondestructive
and mechanical testing4. Selection, identification, preservation,
and/
or cleaning of specimens
5. Macroscopic examination and analysis andphotographic
documentation
6. Microscopic examination and analysis7. Selection,
preparation, examination, and
analysis of metallographic specimens8. Determination of failure
mechanism9. Chemical analysis10. Fracture mechanics analysis11.
Testing under simulated service conditions12. Analysis of all the
evidence, formulation of
conclusions, and writing the report
These stages or steps are briefly outlined asfollows.Collection
of Background Data and
Selection of Samples. There are basicallythree fundamental
principles to be carefullyfollowed when collecting damage
evidencefrom a fractured material (Ref 2):
Locate the origin(s) of the fracture. Thewhole fracture surface
should be visuallyinspected to identify the location of
thefracture-initiating site(s) and to isolate theareas in the
region of crack initiation thatwill be most fruitful for further
micro-analysis. Where the size of the failed partpermits, visual
examination should be con-ducted with a low-magnification
wide-fieldstereomicroscope having an oblique sourceof illumination
(Ref 3).
Do not put the mating pieces of a fractureback together, except
with considerable careand protection. Protection of the surfaces
isparticularly important if electron micro-scopic examination is to
be part of the pro-cedure (Ref 2). Appropriate packaging offailed
components for shipping is equallyimportant. Wrapping them directly
into aplastic bag, or placing pieces directly intoa plastic bottle
or container, can intro-duce unwanted hydrocarbon contaminants.
Failure Analysis of Heat Treated Steel Components
L.C.F. Canale, R.A. Mesquita, and G.E. Totten, editors, p
111-132
DOI: 10.1361/faht2008p111
Copyright 2008 ASM International
All rights reserved.
www.asminternational.org
-
Fingerprints on the failed surfaces can alsointroduce
contamination (Ref 4);
Do not conduct a destructive testing withoutconsiderable
thought. Alterations such ascutting, drilling, and grinding can
ruinan investigation if performed prematurely.Destructive testing
must be performed onlyafter all possible information has
beenextracted from the part in the original con-dition and after
all significant features havebeen carefully documented by
photography(Ref 2).
Preliminary Examination of the FailedPart. In addition to
locating the failure origin,visual analysis is necessary to reveal
stress con-centrations, material imperfections, presence ofsurface
coatings, case-hardened regions, welds,and other structural details
that contribute tocracking. A careful macroexamination is
neces-sary to characterize the condition of the fracturesurface so
that the subsequent microexaminationstrategy can be determined.
Corrodents oftendo not penetrate the crack tip, and this
regionremains relatively clean. The visual macro-analysis will
often reveal secondary cracks thathave propagated only partially
through a crac-ked member. These part-through cracks can beopened
in the laboratory and are often in muchbetter condition than the
main fracture (Ref 3).Nondestructive and Mechanical Testing.
A wide variety of nondestructive testing isavailable, including
dye penetrant, ultrasonics,x-ray, and eddy current, which can help
inthe failure analysis task in order to unveileven subtle and/or
internal defects in a part.Mechanical property tests are also ready
to use,ranging from a sample hardness test to elevated-temperature
tensile and impact testing. Thesetests are often used to determine
if degradation isrelated to fabrication or to the service
environ-ment. Sometimes, a standard test can be adaptedto simulate
manufacturing or in-service condi-tions more closely (Ref
4).Selection, Identification, Preservation,
and/or Cleaning of Specimens. Unless afracture is evaluated
immediately after it isproduced, it should be preserved as soon
aspossible to prevent attack from the environment.The best way to
preserve a fracture is to dry itwith a gentle stream of dry
compressed air, thenstore it in a desiccator, a vacuum storage
vessel,or a sealed plastic bag containing a desiccant.However, such
isolation of the fracture is oftennot practical. Therefore,
corrosion-preventive
surface coatings must be used to inhibit oxida-tion and
corrosion of the fracture surface. Theprimary disadvantage of using
these surfacecoatings is that fracture surface debris, whichoften
provides clues to the cause of fracture, maybe displaced during
removal of the coating.However, it is still possible to recover the
sur-face debris from the solvent used to removethese surface
coatings by filtering the spentsolvent and capturing the residue.
In regard tocleaning techniques, fracture surfaces exposedto
various environments generally contain un-wanted surface debris,
corrosion or oxidationproducts, and accumulated artifacts that must
beremoved before meaningful fractography canbe performed. Before
any cleaning proceduresbegin, the fracture surface should be
surveyedwith a low-power stereobinocular microscope,and the results
should be documented with ap-propriate sketches or photographs.
Low-powermicroscope viewing will also establish theseverity of the
cleaning problem and should alsobe used to monitor the
effectiveness of eachsubsequent cleaning step. It is important
toemphasize that the debris and deposits on thefracture surface can
contain information that isvital to understanding the cause of
fracture. Themost common techniques for cleaning fracturesurfaces,
in order of increasing aggressiveness,are (Ref 3):
Dry air blast or soft organic-fiber brushcleaning
Replica stripping Organic-solvent cleaning Water-based detergent
cleaning Cathodic cleaning Chemical-etch cleaning
Macroscopic Examination and Analysisand Photographic
Documentation. Moreoften than not, the investigation starts with
alow-magnification, if any, observation of thefailed part. This
visual examination can oftenquickly answer questions such as: What
was themode of failure? Did it crack, or was there auniform or
pitting corrosion failure? Did theprotective oxide film break down?
Were thewelds visibly contaminated? A variable magni-fication
stereoscope equipped with a ring lightand directional fiberoptic
lighting is a powerfultool for macroscopic visual
examination.Contemporary stereoscopes can operate over arange of
2.5 to 50 (Ref 4).Microscopic Examination and Analysis.
Once the area of interest is isolated, a smaller
112 / Failure Analysis of Heat Treated Steel Components
-
portion can be cut from the sample and mountedfor metallographic
polishing and microscopicexamination. The microstructure of
specimensmay be enhanced by a wide variety of metallo-graphic
techniques that include, for example,heat tinting, stain etching,
anodizing, and illu-mination by bright-field and polarizing
light.Optical microscopic examination generally be-gins at 50
magnification and continues through1000 or even 1500 . Higher
levels are bestsupplemented by differential interference con-trast
lighting, which allows theoretical resolu-tion of features as fine
as one-third of amicrometer. Features that are important
torecognize include the uniformity and size of thegrain structure,
the size distribution and shape ofintermetallic particles, and
inclusions. Scanningelectron microscopy (SEM) is most usefulwhere
extreme depth of focus and high magni-fications are needed.
Fractures generally arecomplex, undulating surfaces that are
difficult toimage, and an optical microscope can only focuson a
very narrow region because of the veryshallow depth of field.
However, the SEM excelsat imaging fracture surfaces, and it can
beoperated in many different modes. The mostcommon mode is
secondary electron imaging,which provides a detailed,
high-depth-focusimage that is easy to interpret. BackscatteredZ
contrast is used to identify regions ofimpurities within a matrix.
High-atomic-numberspecies produce a light appearance,
whereaslow-atomic-number species create a darkerappearance. The
topographic backscatteredmode enhances the surface topography of
thesample and accentuates height or elevation dif-ferences on a
fracture surface. The characteristicx-rays can be detected and
analyzed according totheir energy. This is called
energy-dispersivex-ray analysis. The x-ray wavelength corre-sponds
to the presence of a specific element, andits amplitude corresponds
to the quantity ofsuch element. This technique allows quantita-tive
characterization of elements within a givenphase. Bulk chemistry is
typically analyzedduring failure analysis to verify conformancewith
industry-accepted chemical limits. In thecase of reactive metals,
light elements canembrittle them due to improper processing
orservice conditions (Ref 4).Selection, Preparation, Examination,
and
Analysis of Metallographic Specimens. Oneof the worst things
that can happen to the sampleis inadequate handling, examination,
or pack-aging. It is imperative that the sample remains in
an undisturbed state prior to analysis, becausethe culprit is
often found in minute surfacefeatures or traces of impurities.
Fracture surfacesmust remain untouched so that high-magnifica-tion
images can accurately determine the failuremode. The sample must be
removed carefully.Important evidence can be destroyed by
over-heating or by allowing adjacent fracture surfacesto fret or
rub together during sectioning. Theideal method would be to unbolt
the componentor to provide adequate support so that a slow-speed
saw can be used to cut out the component.However, sawing lubricants
can mask ordestroy residual chemicals or elements on thefailed
surface, so precautions become extremelynecessary. If the component
has failed in themiddle of a large area, more aggressive
cutting/sectioning techniques may be warranted, butkeep a good
distance from the failed region(Ref 4).Determination of Failure
Mechanism
(with Adapted Text from Ref 7). A thoroughinvestigation should
ensure that all damage isfound and documented, because multiple
modesand mechanisms may be present in most real-world failure
analyses. It is also important torecognize that many unique
mechanisms may bedriven by more than one environmental factor,such
as stress, temperature, corrosion, wear,radiation, or electrical
factors.The term failure mechanism, or damage
mechanism, is meant to convey the specificseries of events that
describe both how thedamage was incurred and the resulting
con-sequences. Examples of damage mechanismsinclude
high-temperature creep, hydrogenembrittlement, stress-corrosion
cracking, andsulfidation. A failure or damage mechanismdescribes
how damage came to be present.This definition of failure mechanism
also
should not be confused with the description ofthe physical
characteristics of damage observed.For example, intergranular
fracture, buckling,transgranular beach marks, and pits can all
bethought of as damage modes. The term damagemode or failure mode
is best used to describewhat damage is present.Much confusion has
occurred because of
the tendency of engineers to use the termsmechanism and mode
interchangeably; in doingso, it is unclear that two distinct
characteristicsneed to be assessed. Sometimes this occursbecause,
within a given system, the samewording is used to describe both the
failuremode and mechanism. For example, pitting
General Aspects of Failure Analysis / 113
-
describes a damage mode because the surface ofa material is
pitted. In certain systems, pitting isalso a possible damage
mechanism. In boilertubing, for example, a pitting damage
mecha-nism describes a specific localized corrosionmechanism where
pits form through dissolutionof metal either from low-pH or
high-oxygenconditions. The metal under the pit surfaces
isunaffected. In this system, pitting is a specificdamage
mechanism, but many other damagemechanisms also result in a pitting
damagemode in boiler tubing, including hydrogendamage, phosphate
corrosion, and causticgouging.It is helpful to be as specific as
possible in
differentiating damage mechanisms in a system.For example,
fatigue is often identified as botha damage mode and a damage
mechanism. Afatigue damage mode is the observable damagethat occurs
under fatigue loading cycles (e.g.,the presence of beach marks).
Classifying fati-gue as a damage mechanism is not
necessarilycomplete because it does not point to the
specificenvironment that results in a fatigue damagemode. Instead,
specific mechanisms that canresult in a fatigue damage mode must
beexamined. Examples include corrosion fatigue,thermomechanical
fatigue, creep-fatigue inter-action, and mechanical
fatigue.Determination of damage mechanisms starts
by characterizing the component(s) being ex-amined. It is
impossible to know what is dif-ferent about a failure without first
understandingwhat is expected from unfailed components.In general,
the analyst should obtain as muchinformation as possible about a
part and itsbackground during the course of an investiga-tion. Some
key questions worth evaluatinginclude:
What was the part supposed to do? How wasit supposed to
work?
How was the part made? What processeswere involved in its
manufacture (e.g.,forming, joining, and heat treatment)?
Whatproperties were expected at the time ofmanufacture?
What were the specified dimensions andtolerances for the
as-manufactured part?
How was the part installed? To what service environment(s) was
the part
exposed? Typical environments to examineinclude operating
temperatures, stresses(steady state or slowly rising and
cyclic),oxidizing/corrosive environments, and wear
environments. What properties were re-quired during service? How
were propertiesexpected to change from service exposure?
How was the part inspected during serviceintervals? What
information was foundduring these inspections?
What material characteristics were specifiedfor the part (e.g.,
composition, strength,hardness, impact, and stress-rupture
proper-ties)? What specifications, industry stan-dards, and
contracts govern these properties?
What were the various ways the part couldfail?
The last item is a key question to repeatedlyask throughout a
failure investigation. The list ofvarious damage mechanisms by
which a part canfail can be narrowed down through two basicconcepts
(Ref 7). Limiting conditions that refinethe scope of explanations
for observed damagecan be defined by using the following two
rulesof thumb:
When the impossible is eliminated, whateverremains, however
improbable, must beconsidered (Sherlock Holmes rule).
When two or more explanations exist for asequence of events, the
simple explanation ismore likely to be the correct one
(Occamsrazor).
Chemical Analysis. In a failure investiga-tion, routine analysis
of the material is usuallyrecommended. There are two main
categories ofchemical analysis that are often used by
failureanalysts:
Bulk composition evaluation: often per-formed in order to
determine whether thecorrect alloy was used in the subject
com-ponent
Microchemical analysis: to find evidence ofcontamination, to
evaluate the compositionof microphases revealed on
ametallographicspecimen, or to evaluate corrosion products
Often, chemical analysis is done last, becausean analysis
usually involves destroying a certainamount of material. There are
instances wherethe wrong material was used, under which con-ditions
the material may be the major cause offailure. In many cases,
however, the difficultiesare caused by factors other than material
com-position.Extreme care must be used in interpretation of
chemical analysis work performed as part of afailure
investigation. Minor deviations from
114 / Failure Analysis of Heat Treated Steel Components
-
specified composition must not be interpreted asthe sole cause
of a failure, without much addi-tional supporting evidence. In most
instances,slight deviations from specified compositionsare not
likely to be ofmajor importance in failureanalysis. However, small
deviations in alumi-num content can lead to strain aging in steel,
andsmall quantities of impurities can lead to temperembrittlement.
In specific investigations, parti-cularly where corrosion and
stress corrosionare involved, chemical analysis of any
deposit,scale, or corrosion product, or a substance withwhich the
affected material has been in contact,is required to assist in
establishing the primarycause of failure.Where analysis shows that
the content of a
particular element is slightly greater than thatrequired in the
specifications, it should not beinferred that such deviation is
responsible for thefailure. Often, it is doubtful whether such
adeviation has played even a contributory part inthe failure. For
example, sulfur and phosphorusin structural steels are limited to
0.04% inmany specifications, but rarely can a failure inservice be
attributed to sulfur content slightly inexcess of 0.04%. Within
limits, the distributionof the microstructural constituents in a
materialis of more importance than their exact pro-portions. An
analysis (except a spectrographicanalysis restricted to a limited
region of thesurface) is usually made on drillings represent-ing a
considerable volume of material andtherefore provides no indication
of possiblelocal deviation due to segregation and
similareffects.Also, certain gaseous elements, or inter-
stitials, normally not reported in a chemicalanalysis, have
profound effects on the mechan-ical properties of metals. In steel,
for example,the effects of oxygen, nitrogen, and hydrogenare of
major importance. Oxygen and nitrogenmay give rise to strain aging
and quench aging.Hydrogen may induce brittleness, particularlywhen
absorbed during welding, cathodic clean-ing, electroplating, or
pickling. Hydrogen isalso responsible for the characteristic halos
orfisheyes on the fracture surfaces of welds insteels, in which
instance the presence of hydro-gen often is due to the use of damp
electrodes.These halos are indications of local rupturethat has
taken place under the bursting micro-stresses induced by the
molecular hydrogen,which diffuses through the metal in the
atomicstate and collects under pressure in pores andother
discontinuities. Various effects due to gas
absorption are found in other metals and alloys.For example,
excessive levels of nitrogen insuperalloys can lead to brittle
nitride phases thatcause failures of highly stressed parts.Various
analytical techniques can be used
to determine elemental concentrations and toidentify compounds
in alloys, bulky deposits,and samples of environmental fluids,
lubricants,and suspensions. Semiquantitative emissionspectrography,
spectrophotometry, and atomic-absorption spectroscopy can be used
to deter-mine dissolved metals (as in analysis of analloy), with
wet chemical methods used wheregreater accuracy is needed to
determine theconcentration of metals. Combustion methodsordinarily
are used for determining the con-centration of carbon, sulfur,
nitrogen, hydrogen,and oxygen.Wet chemical analysis methods may
be
employed for determining the presence andconcentration of anions
such as Cl, NO3
, andS. These methods are very sensitive.X-ray diffraction
identifies crystalline com-
pounds either on the metal surface or as a massof particles and
can be used to analyze corrosionproducts and other surface
deposits. Minor andtrace elements capable of being dissolved can
bedetermined by atomic-absorption spectroscopyof the solution.
X-ray fluorescence spectro-graphy can be used to analyze both
crystallineand amorphous solids, as well as liquids andgases.Stress
Analysis and Fracture Mechanics
Analysis. When confronted with a cracked,fractured, or deformed
component, the failureanalyst will usually seek to answer some
basicquestions:
Were the loads and stresses encountered bythe part at the level
anticipated duringdesign? Or did some unexpected condi-tion(s)
contribute to the failure?
Was the material in the area of the crackingor deformation
capable of meeting the con-ditions anticipated during design? Was
theresome deficiency or discontinuity that con-tributed to the
failure, or was there a localstress raiser at the critical
location? Was thistaken into account by the designer?
In general, there are two types of conditions thatmay lead to
structural failure:
Net-section instability, where the overallstructural cross
section can no longer sup-port the applied load
General Aspects of Failure Analysis / 115
-
The critical flaw size (ac) is exceededby some preexisting
discontinuity or whensubcritical cracking mechanisms (for exam-ple,
fatigue, stress-corrosion cracking, orcreep) reach the critical
crack size
Failures due to net-section instability typi-cally occur when a
damage process such ascorrosion or wear reduces the thickness of
astructural section. This type of failure can beevaluated by
traditional stress analysis or finiteelement analysis (FEA), which
are effectivemethods in evaluating the effects of loading
andgeometric conditions on the distribution of stressand strain in
a body or structural system.However, stress analyses by
traditional
methods or FEA do not easily account forcrack propagation from
preexisting cracks orsharp discontinuities in the material. When
apreexisting crack or discontinuity is present,the concentration of
stresses at the crack tipbecomes asymptotic (infinite) when using
theconventional theory of elasticity. In this
regard,fracturemechanics is a useful tool, because it is amethod
that quantifies stresses at a crack tipin terms of a
stress-intensity parameter (K).The fracture mechanics of cracking
from a dis-continuity or crack in a statically loaded com-ponent
has two possible situations:
The crack reaches a critical length with rapid(brittle)
separation.
The crack blunts, redistributing the stressstate, with continued
loading creating a tearzone (and sharpened crack-tip radius)
infront of the crack. In steels, this tear zonecan then cause the
critical crack length to beexceeded, such that unstable cleavage
frac-ture occurs or unstable microscale ductilefracture is
induced.
Which event occurs depends on the temperatureand the loading
rate, but in either event, crackpropagation is unstable (i.e., does
not require anincreasing load after creation of the tear
zone).Fracture mechanics is a tool to help evaluate theimplications
of preexisting discontinuities orcracks.Testing under Simulated
Service Con-
ditions. During the concluding stages of aninvestigation, it may
be necessary to conducttests that simulate the conditions under
whichfailure is believed to have occurred. Often,simulated-service
testing is not practical be-cause elaborate equipment is required,
and evenwhere practical it is possible that not all of the
service conditions are fully known or under-stood. Corrosion
failures, for example, aredifficult to reproduce in a laboratory,
and someattempts to reproduce them have given mis-leading results.
Serious errors can arise whenattempts aremade to reduce the time
required fora test by artificially increasing the severity of oneof
the factorssuch as the corrosive medium orthe operating
temperature. Similar problems areencountered in wear testing.On the
other hand, when its limitations are
clearly understood, the simulated testing andstatistical
experimental design analysis of theeffects of certain selected
variables encounteredin service may be helpful in planning
correctiveaction or, at least, may extend service life. Mostof the
metallurgical phenomena involved infailures can be satisfactorily
reproduced on alaboratory scale, and the information derivedfrom
such experiments can be helpful to theinvestigator, provided the
limitations of the testsare fully recognized.Analysis of All the
Evidence, Formulation
of Conclusions, and Writing the Report.Before starting this
final step, some questionsmust already be answered: Fracture
surface:a. What is the fracture mode?b. Is the origin of the
fracture visible?c. What is the relation between the fracture
direction and the normal or expected fra-cture directions?
d. How many fracture origins are there?e. Is there evidence of
corrosion, paint, or
some other foreign material on the fracturesurface?
f. Was the stress unidirectional or was itreversed in
direction?
The surface of a part:a. What is the contact pattern on the
surface
of the part?b. Has the surface of the part been deformed
by loading during service or by damageafter fracture?
c. Is there evidence of damage on the surfaceof the part by
manufacturing, assembling,repairing, or service?
Geometry and design:a. Are there any stress concentrations
related
to the fracture?b. Is the part intended to be relatively
rigid,
or is it intended to be flexible, like aspring?
c. Does the part have a basically flawlessdesign?
116 / Failure Analysis of Heat Treated Steel Components
-
d. How does the partand its assemblywork?
e. Is the part dimensionally correct? Manufacturing and
processing:a. Are there internal discontinuities or
stress concentrations that could cause aproblem?
b. If it is a wrought metal, does it containserious seams,
inclusions, or forging pro-blems, such as end grains, laps, or
otherdiscontinuities, that could have an effecton performance?
c. If it is a casting, does it contain shrinkagecavities, cold
shuts, gas porosity, orother discontinuities, particularly near
thesurface of the part?
d. If a weldment was involved, was thefracture through the weld
itself or throughthe heat-affected zone in the parent metaladjacent
to the weld? If through the weld,were these problems something like
gasporosity, undercutting, underbead crack-ing, or lack of
penetration? If through theheat-affected zone adjacent to the
weld,how were the parent metal propertiesaffected by the heat of
welding?
e. If the part was heat treated, was the treat-ment properly
performed?
Material properties:a. Are the mechanical properties of the
metal
within the specified range, if this can beascertained?
b. Are the properties of the metal suitable forthe
application?
c. Residual and applied stress relationship.The residual-stress
system that was withinthe part prior to fracture can have a
pow-erful effectgood or badon the perfor-mance of a part.
d. What was the influence of adjacent partson the failed
part?
e. Were fasteners tight? Assembly:a. Is there evidence of
misalignment of the
assembly that could have had an effect onthe fractured part?
b. Is there evidence of inaccurate machin-ing, forming, or
accumulation of toler-ances?
c. Did the assembly deflect excessively understress?
Service conditions: It is important to deter-mine if there were
any unusual occurrences,such as strange noises, smells, fumes,
orother happenings, that could help explain the
problem. The following questions shouldalso be considered:a. Is
there evidence that the mechanism was
overspeeded or overloaded?b. Is there evidence that the
mechanism was
abused during service or used under con-ditions for which it was
not intended?
c. Did the mechanism or structure receivenormal maintenance with
the recom-mended materials?
d. What is the general condition of themechanism?
Environmental reactions: The problemsrelated to the environment
can arise anywherein the history of the part:
manufacturing,shipping, storage, assembly, maintenance,and service.
None of these stages should beoverlooked in a thorough
investigation thatasks:a. What chemical reactions could have
taken
place with the part during its history?b. To what thermal
conditions has the part
been subjected during its existence? Report writing: Finally,
the report analyzing
the failure should be written in a clear,concise, logical
manner. It should be clearlystructured with sections covering the
fol-lowing (Ref 6):a. Description of the failed itemb. Conditions
at the time of failurec. Background history important to the
failured. Mechanical and metallurgical study of the
failuree. Evaluation of the material qualityf. Discussion of any
anomaliesg. Discussion of the mechanism or possible
mechanisms that caused the failureh. Recommendations for the
prevention of
future failures or for action to be takenwithsimilar pieces of
equipment
Irrelevant data should be omitted, and,depending on the nature
of the problem and thedata, not every report will need full
treatmentsfor every one of the sections listed previously.Many
times, the readership may include pur-chasing, operating, or
accounting personnelwho are not technically trained. If this is
thesituation, the report should be written so that itis
comprehensible to these persons. At least,those sections of the
report that bear on theirdecision-making or information needs
should bewritten in language that is accessible to them.Frequently,
a cover letter summarizing the mostimportant findings and the
suggested action is a
General Aspects of Failure Analysis / 117
-
good vehicle for reaching top executives whoare not as
interested in the technical specifics butneed key findings and
recommendations as abasis for decision making. Followup on
therecommendations is frequently a difficult taskbut should be
undertaken for the more criticalfailures. Cooperation between the
investigator,the designer, the manufacturer, and the user
iscritical in developing good, workable changes.
Fracture
The process of fracture, in general terms, canbe described in
terms of the mechanisms ofcrack initiation and/or crack extension
(growth).Different mechanisms may occur for crackinitiation and the
subsequent process of crackgrowth. For example, crack extensionmay
occurby the brittle mechanism of cleavage, eventhough extensive
elongation accompanied orpreceded crack initiation. The fracture
may beclassified as either ductile or brittle, dependingon whether
the mechanism is describing crackinitiation or crack growth,
respectively. Like-wise, the low-energy catastrophic fracture of
ahigh-strength aluminum alloy by microvoidcoalescence is also
difficult to classify because,although the fracture energy is low
and failureinitiates by fracture or decohesion of brittleparticles,
the growth and coalescence of themicrovoids occurs by plastic
deformation.Another difficulty is that cleavage fracture maybe
initiated by dislocation interactions that, bydefinition, involve
plasticity. This is why frac-tures are sometimes difficult to
logically classify(Ref 5). Therefore, it is helpful to be
clearwhether fracture mechanisms are describing theprocess of crack
initiation or extension. Crackextension also can be multimode over
time (e.g.,fatigue crack growth followed by overload).In terms of
fracture appearances (or fracture
modes, defined earlier in the section Deter-mination of Failure
Mechanism in this chap-ter), a general summary of the visual
andmicroscopic aspects of fracture surfaces formetallic materials
is provided in Table 1 (Ref 8).Several analytical procedures are
available fordistinguishing among the various types of frac-ture.
For example, the presence or absence ofplastic macrodeformation can
be determinedwith the unaided eye or by use of a steel scale,
amachinists micrometer, or a machinists ormeasuring microscope.
Differences in somedimensional attribute of parts (such as width
or
thickness) at andwell away from the fracture canserve to define
macrodeformation after assur-ance that both points of measurement
had thesame dimension before fracture.Fracture-surface matching is
also used to
determine the presence or absence of plasticdeformation. It is
very important, however, toresist the temptation to fit the
matching fracturesurfaces together, because this almost
alwaysdestroys (smears) microscopic features. Thefracture surfaces
should never actually touchduring fracture-surface matching.The
origin of a fracture may be indicated by a
discoloration or by the topography of the frac-ture surface. A
discolored area on a fracturesurface may be produced by a
preexisting crackwhose surfaces have been corroded or oxidized.For
example, the surfaces of a quench crack canbe oxidized during a
subsequent tempering heattreatment; the oxide film gives a
bluish-blackcolor to the surfaces of the crack.
Topographicalfeatures that often reveal the origin of a fractureare
either chevron or river patterns or a setof diverging ledges. If
the fracture surface isessentially featureless, the presence of a
shearlip can be used to locate, within limits, the originof a
fracture. For example, a shear lip is notformed at the origin of a
stress-corrosion crack,but when the crack begins to propagate
rapidly, ashear lip is formed wherever the crack frontexits from
the interior to the free surface. Beachmarks, which are associated
with fatigue-initiated fractures, also provide a definite
indi-cation of the crack origin; however, it should benoted that
fracture surfaces having an appear-ance similar to that of the
beach-mark patterncan be produced by stress corrosion.Generally,
cyclic loading produces only a
single crack, which is usually located at a site ofstress
concentration or of a metallurgical defect,whereas additional
cracks, formed indepen-dently of themain crack and at a distance
from it,may be observed on the surface of a structuralor machine
component subjected to corrosionfatigue or stress corrosion.On the
microscopic level, striations on the
fracture surface are unique to fatigue, and thecrack path,
although normally transgranular, canbe intergranular. For example,
intergranularfatigue cracking can occur in the case of a
car-burized steel or in a material that has a highdensity of
second-phase particles at the grainboundaries.Corrosion-fatigue and
stress-corrosion cracks
may propagate transgranularly, intergranularly,
118 / Failure Analysis of Heat Treated Steel Components
-
Table 1 Fracture mode identification chart
Method
Instantaneous failure mode(a) Progressive failure mode(b)
Ductile overload Brittle overload Fatigue Corrosion Wear
Creep
Visual, 1 to
50
(fracture
surface)
Necking ordistortion in
direction
consistent with
applied loads
Dull, fibrousfracture
Shear lips
Little or nodistortion
Flat fracture Bright or coarsetexture,
crystalline,
grainy
Rays orchevrons
point to origin
Flat progressivezone with beach
marks
Overload zoneconsistent with
applied loading
direction
Ratchet markswhere origins
join
Generalwastage, rough-
ening, pitting, or
trenching
Stress-corrosionand hydrogen
damage may
create multiple
cracks that
appear brittle
Gouging,abrasion,
polishing,
or erosion
Galling or storingin direction of
motion
Roughened areaswith compacted
powdered debris
(fretting)
Smooth gradualtransitions in
wastage
Multiple brittle-appearing fissures
External surfaceand internal
fissures contain
reaction scale
coatings
Fracture afterlimited
dimensional
change
Scanning
electron
microscopy,
20 to
10,000
(fracture
surface)
Microvoids(dimples)
elongated
in direction of
loading
Single crack withno branching
Surface slip bandemergence
Cleavage orintergranular
fracture
Origin area maycontain an
imperfection
or stress
concentrator
Progressivezone: worn
appearance,
flat, may show
striations at
magnification
above 500
Overload zone:may be either
ductile or brittle
Path of penetra-tion may be
irregular,
intergranular, or
a selective phase
attacked
EDS mayhelp identify
corrodent(c)
Wear debris and/orabrasive can be
characterized as to
morphology and
composition
Rolling-contactfatigue appears
like wear in early
stages
Multipleintergranular
fissures covered
with reaction scale
Grain faces mayshow porosity
Metallographic
inspection,
50 to 1000
(cross
section)
Grain distortionand flow near
fracture
Irregular,transgranular
fracture
Little distortionevident
Intergranular ortransgranular
May relate tonotches at
surface or brittle
phases internally
Progressivezone: usually
transgranular
with little
apparent
distortion
Overload zone:may be either
ductile or brittle
General orlocalized surface
attack (pitting,
cracking)
Selective phaseattack
Thickness andmorphology of
corrosion scales
May showlocalized
distortion at
surface consistent
with direction of
motion
Identify embeddedparticles
Microstructuralchange typical of
overheating
Multiple inter-granular cracks
Voids formed ongrain boundaries
or wedge-shaped
cracks at grain
triple points
Reaction scalesor internal
precipitation
Some cold flowin last stages of
failure
Contributing
factors
Load exceeded thestrength of the part
Check for properalloy and proces-
sing by hardness
check or destruc-
tive testing,
chemical analysis
Loading directionmay show failure
was secondary
Short-term,high-temperature,
high-stress rupture
has ductile
appearance
(see creep)
Load exceededthe dynamic
strength of the
part
Check for properalloy and
processing as
well as proper
toughness, grain
size
Loadingdirection may
show failure was
secondary or
impact induced
Lowtemperatures
Cyclic stressexceeded the
endurance limit
of the material
Check for properstrength, surface
finish, assembly,
and operation
Prior damage bymechanical or
corrosion modes
may have
initiated
cracking
Alignment,vibration,
balance
High cycle lowstress: large
fatigue zone;
low cycle high
stress: small
fatigue zone
Attack morphol-ogy and alloy
type must be
evaluated
Severity ofexposure
conditions may
be excessive;
check: pH,
temperature,
flow rate,
dissolved
oxidants, elec-
trical current,
metal coupling,
aggressive
agents
Check bulkcomposition and
contaminants
For gouging orabrasive wear:
check source of
abrasives
Evaluate effec-tiveness of lubri-
cants
Seals or filters mayhave failed
Fretting inducedby slight looseness
in clamped joints
subject to
vibration
Bearing or materi-als engineering
design may reduce
or eliminate
problem
Watercontamination
High velocitiesor uneven flow
distribution,
cavitation
Mild overheatingand/or mild
overstressing at
elevated
temperature
Unstable micro-structures and
small grain size
increase creep
rates
Ruptures occurafter long
exposure times
Verify properalloy
(a) Failure at the time of load applicationwithout prior
weakening. (b) Failure after a period of timewhere the strength has
degraded due to the formation of cracks, internal
defects, or wastage. (c) EDS, energy-dispersive spectroscopy.
Compiled by C.R. Morin, S.L. Meiley, and Z.B. Flanders, Packer
Engineering Associates, Inc.
General Aspects of Failure Analysis / 119
-
or by a combination of both modes. A distin-guishing feature of
stress corrosion is thebranching of the main crack. If corrosion
pitsor corrosion products are found only on theslow-growth region
of a fracture surface, theenvironment was in all probability
sufficientlycorrosive to affect the fracture mechanism.However, if
evidence of corrosion is found onboth the slow-growth and
fast-growth areas,some corrosion took place subsequent to
frac-ture, and the environment may or may not haveinfluenced
fracture.
Ductile Fracture
Ductile fracture takes place when a materialcapable of
undergoing plastic deformation issubjected to stresses that
culminate in its rup-ture. Macroscopically, the ductile fracture
pro-cess presents some peculiarities that allow it tobe identified
immediately. The first feature is thepresence of plastic
deformation that may beaccompanied by neck formation. In
tensiletestpieces of ductile materials, besides necking,the
fracture surface presents a fibrous aspect anda cup-cone geometry,
as seen in Fig. 1.
The fracture process begins in the center ofthe testpiece with
microvoid nucleation alonggrain boundaries or from interfaces such
as thosefound in base metal/inclusions boundaries. Asthe applied
stress increases, microvoids growand coalesce, forming a crack in
the center of thepart. This process, depicted in Fig. 2, ends up
inrapid crack propagation by shearing of theremaining ligament of
the neck region, at anangle of 45 in relation to the loading
direction.It is important to emphasize that a cup-conegeometry will
depend on the geometry anddimensions of the part and mechanical
proper-ties of the material. Thin sheets, for instance,present neck
formation and a fracture surfaceoriented at an angle of 45 in
relation to theapplied load, as observed in Fig. 3. Ductilefracture
takes place intergranularly, unless somesort of mechanism weakens
the grain bound-aries. The microscopic aspect of the
fracturesurface consists of several small ellipticalcavities, or
microvoids, as depicted in Fig. 4.
Brittle Fracture
Brittle fracture occurs with little or no plasticdeformation.
This type of fracture is often
Fig. 1 Ductile fracture showing the typical cup-cone
geo-metry
Microvoids
Fig. 2 Schematic representation of the cup-cone
geometryformation during the ductile fracture process
Fig. 3 Thin sheet testpiece of a low-carbon steel after
fracture
Fig. 4 Microvoids on the fracture surface of AA6061-T1tensile
testpiece
120 / Failure Analysis of Heat Treated Steel Components
-
associated with materials of high strength andlow ductility or
materials that were subjectedto an embrittlement process. The
crack, oncenucleated, propagates very quickly in a
directionperpendicular to the applied load. Figure 5 pre-sents an
example of a gray cast iron testpiece thatpresented brittle
fracture.Besides the mechanical properties, several
other factors may result in a brittle behavior,such as
temperature, loading rate, presence ofstress concentrators, and
dimensions. Low tem-peratures tend to reduce the ductility of
metals,especially those possessing a body-centeredcubic structure,
resulting in a typically brittlefracture. Figure 6 shows that as
the temperaturedrops, the brittle aspect on the fracture surface
ofimpact testpieces increases. The presence ofstress raisers or
larger dimensions introduces amore severe triaxial stress state
within thematerial, and thus, there is larger probability
thatbrittle fracture will occur. However, it is known
that the superposition of high hydrostatic stres-ses on the
material reduces the triaxiality levels,increasing ductility. High
applied loading ratesare likely to make plastic deformation
moredifficult because shearing processes are time-dependent,
resulting in brittle behavior.Crack propagation by brittle fracture
can
occur across the grains (transgranular) oralong the grain
boundaries (intergranular). Inthe transgranular mode, the fracture
processtakes place by cleavage along specific crystal-lographic
planes. Figure 7 presents cleavageregions in a microalloyed
low-carbon steel,which can be identified by flat regions on
thefracture surface. Additionally, it is worth men-tioning that
most parts of steels will presentalternate regions consisting of
cleavage areasand microvoids, evidencing a mixed mode ofcrack
propagation.In another situation, fracture can take place
intergranularly, because the grain boundary is a
Fig. 5 Tensile testpiece of gray cast iron presenting
brittlefracture
Fig. 6 Fracture surfaces of SAE 4140 impact testpieces.Tested at
room temperature, right, and at196 C, left
Fig. 7 (a) Cleavage region observed in low-carbon steel.(b)
Magnification of the region delimited by the rec-
tangle in (a) showing an inclusion in the center of the
cleavageregion
General Aspects of Failure Analysis / 121
-
weaker path for crack propagation. Normally,this fracture mode
will occur when someembrittlement process resulted in grain
bound-aries being more susceptible to crack propaga-tion than the
core of the grain, such as anunsuitable heat treating or by
environmentalfactors. Figure 8 presents an example of
inter-granular brittle fracture in an austenitic stainlesssteel SAE
316L, where grain boundaries canclearly be observed on the fracture
surface.
Fatigue Fracture
According to the definition given by ASTME1823, fatigue is the
process of progressivelocalized permanent structural change
occurringin a material subjected to conditions that pro-duce
fluctuating stresses and strains at somepoint or points and that
may culminate in cracksor complete fracture after a sufficient
number offluctuations. A material subjected to fatiguecan fracture
at applied stresses much lower thanthose necessary to fracture the
same materialunder monotonic conditions. The fluctuatingstresses
can be originated from mechanical,thermal, or vibration loading
conditions, and thephenomenon is responsible for more than 80%of
mechanical failures of components. For morethan 150 years, the
study of metals fatiguehas involved engineers, physicists,
chemists,and mathematicians, and everyday this studybecomes more
and more complex and impor-tant. The theory about fatigue is
extremely vast,and for each question answered, another one,more
instigating, appears, requiring a broadknowledge of materials
science. In the followingtopics, a brief overview is given about
the mainmechanisms and factors influencing the fatigue
life of a component during both the nucleationand crack
propagation phases.Fatigue Crack Initiation. Generally, fatigue
cracks are initiated at free surfaces, where thereis no
constraint to material deformation; how-ever, in some cases, cracks
may be initiated inthe interior of the material where interfaces
arepresent, such as the interface of a carburizedsurface layer and
the base metal or the interfaceof an inclusion and the base metal,
or from gasbubbles. In other cases, subsurface cracks werefound to
nucleate below the surface where highcompressive residual stresses
were introducedby shot peening or surface rolling.One of the
classic models of fatigue crack
nucleation considers that when a material isunder loading
(monotonic or cyclic), slips occurat the high-shear-stress planes,
creating steps onthe material surface. Under cyclic loading,the
formation of intrusions and extrusions isobserved, as schematically
represented in Fig. 9.Slip band intrusions are excellent stress
raisersthat can be sites of crack nucleation.Besides the applied
stress amplitude, DS/2,
several other factors are likely to affect thenucleation of a
fatigue crack, such as the meanstress, Sm, or load ratio, R;
geometry and surfacefinishing of the part; mechanical properties;
andenvironment. Here, the R ratio is defined as theratio between
the minimum and maximum loadsduring the fatigue cycle.A large
proportion of fatigue data found in the
literature refers to tests conducted at Sm= 0,that is, for a
load ratio R=1. However, inmany engineering situations, the
fluctuatingstresses are superimposed to a static stress.Larger mean
stresses reduce the nucleation timebecause they facilitate the
plastic deformationmechanism associated with this phenomenon. Inan
S-N graph, this can be represented by curvesshifted to the left and
down, as represented inFig. 10.
Fig. 8 Fractograph of SAE 316L showing intergranular
brittlefracture
Metal Surface
Intrusion Extrusion
Fig. 9 Schematic representation of an intrusion formation onthe
surface of a metallic material
122 / Failure Analysis of Heat Treated Steel Components
-
The mechanism proposed in Fig. 9 isadequate to explain the
initiation of crackson polished testpieces or components withoutthe
presence of geometric discontinuities.However, in engineering
components, there areseveral stress concentrators, such as
scratches,notches, machining marks, corrosion pits,
andmicroconstituents such as grain boundaries,triple points, and
inclusions, that individually orsynergistically can reduce the
initiation time.Since the initiation depends essentially on
plastic deformation mechanisms, high-strengthmaterials normally
present a higher resistance tofatigue crack nucleation. In this
sense, severalsurface-hardening treatments are employed
toselectively reinforce the material, aiming toretard crack
initiation and therefore to increasefatigue life.The chemical
composition and/or the micro-
structure of the surface can be modified bythermochemical
treatments, such as carburizingor nitriding, or by cold deformation
processes,such as shot peening or surface rolling.Mechanical parts
that necessarily present stressconcentrators, such as crankshafts,
gears, andbolts, can be subjected to these treatmentsto increase
the fatigue limit of the material.Figure 11 shows a micrograph of
the transversesection of a bolt, where the thread was coldformed by
surface rolling. As a consequence,surface grains are flattened due
to the mechan-ical deformation imposed. In this case,
besidesincreasing hardness and mechanical strength,the process
avoids the introduction of harmfulmachining marks.Surface
treatments may also increase fatigue
life by the introduction of compressive residualstresses on the
surface of the material. As long asthe material remains in linear
elastic conditions,the principle of stress superposition can be
employed to describe the actual stress state inmaterials
containing residual stresses. There-fore, the effective stress, S0,
is given by thesum of the applied stress, S, to the residual
stress,Sres:
S0=S+Sres (Eq 1)
Similarly, the effective minimum and maximumstresses are
defined, respectively, as:
S0max=Smax+Sres (Eq 2)
S0min=Smin+Sres (Eq 3)
Consequently, the effective stress amplitude,mean stress, and
load ratio are given, respec-tively, by:
DS0
2=
S0max7S0min
2=
(Smax+Sres)7(Smin+Sres)
2
=Smax7Smin
2=DS
2(Eq 4)
S0m=S0max+S
0min
2=
(Smax+Sres)+(Smin+Sres)
2
=Smax+Smin
2+Sres=Sm+Sres (Eq 5)
R0=S0minS0max
=Smin+Sres
Smax+Sres(Eq 6)
Therefore, the presence of a residual-stressfield does not
affect the stress amplitude butaffects the mean stress and the load
ratio.A compressive residual stress reduces the meanstress and the
load ratio, increasing the number
S/2
Nf
Increasing
Sm
Fig. 10 Mean stress effect on S-N fatigue curves
Fig. 11 Optical micrograph of the transverse section of athread
fillet machined by surface rolling. The ma-
terial consists of duplex stainless steel
General Aspects of Failure Analysis / 123
-
of cycles for crack nucleation and vice versa.In some
situations, where high surface com-pressive residual stresses are
found, such as inmaterials subjected to surface-hardening
treat-ments, a crack may initiate below the surface,where the
compressive residual-stress level islower. An example of subsurface
crack nuclea-tion is observed in Fig. 12 for a
surface-rolledductile cast iron subjected to
bending-rotatingfatigue.Fatigue Crack Propagation. Basically,
fati-
gue crack propagation can be divided into threestages: stage I
(short cracks), stage II (longcracks), and stage III (final
fracture).A fatigue crack, once initiated, propagates
along high shear-stress planes (45), as sche-matically
represented in Fig. 13. This is knownas stage I or the short crack
growth propagationstage. The crack propagates until it is
deceler-ated by a microstructural barrier, such as a grainboundary,
inclusions, or pearlitic zones, thatcannot accommodate the initial
crack growthdirection. Therefore, grain refinement is capableof
increasing fatigue strength of the material dueto the insertion of
a large quantity of micro-structural barriers, that is, grain
boundaries, thatmust be overcome in stage I of propagation.Surface
mechanical treatments, such as shot
peening and surface rolling, contribute to theincrease in the
number of microstructuralbarriers per unit of length due to the
flattening ofthe grains.When the stress-intensity factor, K,
increases
as a consequence of crack growth or higherapplied loads, slips
start to occur in differentplanes close to the crack tip,
initiating stage II ofpropagation. While stage I of propagation
isorientated 45 in relation to the applied load,propagation in
stage II is perpendicular to loaddirection, as depicted in Fig. 13.
An importantcharacteristic of stage II propagation is thepresence
of ripples on the fracture surface,known as striations, which are
only visible withthe aid of a scanning electron microscope. Notall
engineering materials exhibit striations. Theyare clearly seen in
pure metals and many ductilealloys, such as aluminum alloys. In
steels, theyare frequently observed in cold-worked alloys.Figure 14
shows examples of fatigue striationsin an interstitial-free steel
and in aluminumalloys. The most accepted mechanism for theformation
of striations on the fatigue fracturesurface of ductile metals (Ref
9) consists ofsuccessive blunting and resharpening of thecrack tip,
as represented in Fig. 15.Finally, stage III is related to the
unstable
crack growth as Kmax approaches KIc. At thisstage, crack growth
is controlled by static modesof failure and is very sensitive to
the micro-structure, load ratio, and stress state (plane-stress or
plane-strain loading).Macroscopically, the fatigue fracture
surface
can be divided into two distinct regions, asshown by Fig. 16.
The first region corresponds tothe stable fatigue crack growth and
presentsa smooth aspect due to the friction betweenthe crack-wake
faces. Sometimes, concentricmarks, known as beach marks, can be
seen onthe fatigue fracture surface as a result of suc-cessive
arrests or decrease in the fatigue crackgrowth rate due to a
temporary load drop or to anoverload that introduces a compressive
residual-stress field ahead of the crack tip.The other region
corresponds to the final
fracture and presents a fibrous and irregularaspect. In this
region, the fracture can be eitherbrittle or ductile, depending on
the mechanicalproperties of the material, dimensions of thepart,
and loading conditions. The exact fractionof area of each region
will depend on the appliedload level. High applied loads will
result in asmall stable fatigue crack propagation area,as depicted
in Fig. 16(a). On the other hand,
Fig. 12 Probable subsurface crack nucleation site in a
sur-face-rolled ductile cast iron testpiece tested under
bending-rotating conditions
Stage I Stage II
Su
rfa
ce
Fig. 13 Stages I and II of fatigue crack propagation
124 / Failure Analysis of Heat Treated Steel Components
-
if lower loads are applied, the fatigue crackwill have to grow
longer before the appliedstress-intensity factor, K, reaches the
fracturetoughness value of the material, resulting in asmaller area
of fast fracture (Fig. 16b).Ratcheting marks are another
macroscopic
feature that can be observed in fatigue fracturesurfaces. These
marks originate when multiple
Fig. 14 Fatigue striations in (a) interstitial-free steel and
(b)aluminum alloy AA2024-T42. (c) Fatigue fracture
surface of a cast aluminum alloy where a fatigue crack
wasnucleated from a casting defect, presenting solidification
den-drites on the surface. Arrow at top right indicates fatigue
striations.
(a)
(b)
(c)
(d)
(e)
Fig. 15 Proposed mechanisms of striation formation in stageII of
propagation. (a) No load. (b) Tensile load. (c)
Maximum tensile load. (d) Load reversion. (e) Compressive
load.Source: Ref 9
Fig. 16 Fatigue fracture surface. (a) High applied load.(b) Low
applied load
General Aspects of Failure Analysis / 125
-
cracks, nucleated at different points, join to-gether, creating
steps on the fracture surface.Therefore, counting the number of
ratchet marksis a good indicator of the number of nucleationsites.
Figure 17 presents in detail some ratchetmarks found on the
fracture surface of a largeSAE 1045 rotating shaft, fractured by
fatigue.Similar to the initiation phase, many factors
can affect long fatigue crack propagation rates.Among them,
special attention should be givento the effects of load ratio and
the presence ofresidual stresses.Increasing the load ratio has a
tendency to
increase the long crack growth rates in allregions of the
fatigue crack growth rate versusapplied stress-intensity factor
range curve, orsimply, da/dN versus applied DK curve. Gen-erally,
the effect of increasing load ratio is lesssignificant in the Paris
regime than in near-threshold and near-failure regions (Fig.
18).
Near the threshold stress-intensity factor,DKth, the effects of
R ratio are mainly attributedto crack closure effects, where crack
faces comein contact at an appliedKcl that is higher than
theminimum applied stress-intensity factor, Kmin.Several different
mechanisms may contribute
to premature crack closure. One of them consistsof
plasticity-induced closure, represented inFig. 19(a). As the crack
grows, the material thathas been previously permanently
deformedwithin the plastic zone now forms an envelope ofplastic
zones in the wake of the crack front. Thisleads to displacements
normal to the crack sur-faces as the restraint is relieved. This is
no pro-blem while the crack is open; however, as theload decreases,
the crack surfaces touch beforethe minimum load is reached,
shielding thecrack. This type of premature contact can alsooccur
due to the crack-wake roughness andirregularities (Fig. 19b) or by
the presence ofcorrosion subproducts, such as oxides (Fig. 19c).As
observed in Fig. 20, the effect of closure
produces a reduction in the effective DK rangebecause of the
increase in the effective Kmin,reducing the driving force for
fatigue crackgrowth. The effect is more significant near
thethreshold region because the crack tip openingdisplacements are
smaller and the crack facesare closer to each other. Additionally,
for thesame applied DK, higher R ratios increase theapplied values
of Kmax and Kmin, increasingDKeff.For most materials, the Paris
regime is con-
sidered closure-free and Kmax-independent, and
Fig. 17 Ratcheting marks, indicated by the arrows, in an SAE1045
shaft fractured by fatigue
K
Increasing
Rda/dN
Near threshold
Final failure
Paris regime
Fig. 18 Schematic representation of the R ratio effect onfatigue
crack growth curves. The near-threshold,
Paris regime, and final failure regions are also indicated on
thecurves.
Plastic deformation
envelope
(a)
(b)
(c)
Premature contact points Oxides
Plastic zone
Crack tip
Fig. 19 Crack closure mechanisms induced by (a) plasticity,(b)
roughness, and (c) oxide
126 / Failure Analysis of Heat Treated Steel Components
-
the crack growth rates are generally very similarfor tests
conducted under different R ratios. Nearthe final failure, the
effects of R ratio are relatedto the higher monotonic fracture
component asKmax approaches KIc. Therefore, for the sameapplied DK,
Kmax values are higher for testsconducted under higher applied R
ratios, andconsequently, da/dN values are higher.The effects of
residual stress on fatigue crack
growth are related to alterations in the R ratioand in the
applied DK. In other terms, the resi-dual stresses affect the two
parameters thatcontrol the crack driving force, that is, Kmax
andDKeff. When a crack is introduced in a platesubjected to a
residual-stress field, a residualstress-intensity factor, Kr,
arises that can eitherdecrease or increase the crack driving
forceparameters.The superposition principle can also be
applied in terms of the stress-intensity factor,provided that
the material remains linearlyelastic. In this sense,Kr can be added
toKmax andKmin:
K0max=Kmax+Kr (Eq 7)
K0min=Kmin+Kr (Eq 8)
As a result, R0 and DK0 are defined as follows. IfK0min40,
then:
R0=K0minK0max
=Kmin+Kr
Kmax+Kr(Eq 9)
DK0=K0max7K0min= Kmax+Kr 7 Kmin+Kr
=Kmax7Kmin=DK
(Eq 10)
If K0minj0, then:
R0=0 (Eq 11)
DK0=K0max=Kmax+Kr (Eq 12)
It is important to note that these equationsassume that the part
of the fatigue cycle duringwhich the crack is closed at its tip
(i.e., K050)makes no contribution to crack growth.
Distortion
Distortion is the least serious mode of failure,but it can lead
a part to failure or a structure tocollapse. It is easy to
recognize but very difficultto prevent. This is due to the fact
that distortiondoes not involve the part itself but its use
anddesign. There are four reasons for distortion:yielding,
buckling, creep, and residual stresses.Yielding. When a load is put
on a part, and it
causes the part to be permanently distorted, it isunable to
perform the intended function andtherefore must be considered
failed. In a well-designed part, the stresses never exceed the
yieldpoint, and the part deforms only elastically; thatis, when the
load is released, the part returns toits original dimensions.In a
good design, the part operates in the
elastic range, that is, below yielding point;beyond this, the
part will be permanentlydeformed, and greater loads will cause the
partto actually break. This point is considered to be avery basic
point to design and applies when theload on a part is applied in a
quasi-static way,such as the load on a building structure or
thestress in the legs of a desk. A ductile failure is
K
Time
Kmax
Kmin
KapKeff K
Time
Kmax
Kap=Keff
Kcl
(b)
Kcl
Kmin
(a)
Fig. 20 Load ratio effect on DKeff in a fatigue cycle. (a)
Kmin5Kcl. (b) Kmin4Kcl
General Aspects of Failure Analysis / 127
-
onewhere there is a great deal of distortion of thefailed part.
Commonly, a ductile part fails whenit distorts and can no longer
carry the neededload. However, some ductile parts break intotwo
pieces and can be identified because there isa great deal of
distortion around the fractureface, similar towhat would happen if
toomuch isplaced load on a low-carbon steel bolt.Buckling. The
failure of an engineering
component is not always caused by materialsfracture. In many
occasions, the componentdistortion may be sufficient to put it out
offunction. The distortion can be elastic or plastic.The elastic
distortions are temporary; however,they may be sufficient to cause
interference onthe mobile parts. The plastic distortion is
per-manent and can be a result of an overload orcreep deformation.
The overload causes per-manent plastic deformation when the
materialyield limit is overcome. This may happen in thepresence of
stress concentrators, high tempera-ture, inadequate heat treatment,
or incorrectmaterials selection for the component applica-tion.
Compressive overloads may lead thematerial to overcome the buckling
strengthlimit, such as the one shown in Fig. 21 for analuminum
part. The buckling strength is essen-tially a design problem (not
metallurgical), andthe load depends on the dimensions of the
partand the Youngs modulus of the material (theonly materials
factors involved).Creep is a time-dependent phenomenon that
causes a part failure if it is under both quasi-static load and
temperatures higher than 0.3 Tm(absolute melting temperature).
Creep strainmay produce sufficiently large deformation or
distortion that a part can no longer perform itsintended
function. The two general types ofcreep processes are
grain-boundary sliding andvoids at grain boundaries (cavitation
creep).The creep processes are easily identified by
the local ductility and large numbers of inter-granular cracks
that will depend on the tem-perature and strain rate imposed. In
general, ahigh strain rate combined with high temperatureresults in
ductile fracture, followed by a largeelongation and neck formation.
Additionally,the grains near the fracture surface tend to
beelongated. On the other side, the combination oflow strain rate
and high temperature results inintergranular brittle fracture, with
low elonga-tion or necking. Intergranular fracture in
suchconditions normally initiates by grain-boundarysliding from
triple points or at grain-boundaryintersections with second-phase
particles, caus-ing cavities on the material microstructure,
aspresented in Fig. 22.Once the crack nucleates, it propagates
by
grain boundaries, and given that some sig-nificant plastic
deformation may take place,the fracture surface tends to exhibit
grains ofequiaxial shape. Therefore, to increase creepstrength, the
material is normally heat treated toincrease the grain size,
reducing the ratiobetween the grain surface area and volume.
Inturbines that work at very high temperatures, thecreep mechanism
must be considered. In thiscase, the component may be produced
frommonocrystals that significantly increase thecreep
resistance.Most creep curves show three distinct stages
(Fig. 23). After the elastic strain, there is a regionof
increasing plastic flow at decreasing rate (firststage), followed
by a region of approximatelyconstant strain rate (secondary stage),
and finallya region of intense increase in the strain rate,which
rapidly extends to fracture (third stage).
Fig. 21 Aluminum part that suffered buckling
Cavities
(b)(a)
Fig. 22 Intergranular crack formation at high temperature
bygrain-boundary sliding at (a) triple points and
(b) inclusions
128 / Failure Analysis of Heat Treated Steel Components
-
Residual stresses can play a significant rolein explaining or
preventing failure of a compo-nent. One example of residual
stresses prevent-ing failure is the use of shot peening
processesthat increase the fatigue life of a component byinducing
surface compressive stresses.Unfortunately, there are also
processes or
processing errors that can induce excessivetensile residual
stresses in locations that maypromote failure of a component. The
internalstate of stress is caused by thermal and/ormechanical
processing of the parts. Commonexamples of these are bending,
rolling, orforging a part. Thermal residual stresses areprimarily
due to differential expansion when ametal is heated or cooled. Two
control factorsare thermal treatment (heating or cooling)and
restraint. Both the thermal treatment andrestraint of the component
must be present togenerate residual stresses. Residual stressescan
result in visible distortion of a component.However, in the case of
residual stresses, thedistortion can also be useful in estimating
themagnitude or direction of these stresses.
Wear-Assisted Failure
Wear may be defined as damage to a solidsurface caused by the
removal or displacementof material by the mechanical action of a
con-tacting solid, liquid, or gas. It may cause sig-nificant
surface damage, and the damage isusually thought of as gradual
deterioration.While the terminology of wear is unresolved,the
following categories are commonly used:adhesive wear, abrasive
wear, erosive wear,fretting, cavitation, rolling, contact fatigue,
andcorrosive wear.
Adhesive wear has been commonly identifiedby the terms galling
or seizing. It is caused bythe material transference from one
surface toanother during their relative movement due to
asolid-state welding process. Figure 24 shows aschematic
representation of this process. Highcontact pressure among the
surface roughnessresults in local plastic deformation and pointsof
microwelding. The movement between thesurfaces causes the rupture
of the junctions,resulting in a rough peak in one surface and
avalley on the other. Eventually, the tip of a peakmay break, and
an abrasive particle is formed.Abrasive wear, or abrasion, is
caused by the
displacement of material from a solid surfacedue to hard
particles or protuberances slidingalong the surface. The particles
may be foundfree between two surfaces or attached to one ofthem,
and the wear level depends on the relativehardness between the
particle and the surface(Fig. 25). The abrasion may also happen due
tothe protuberances or sharp asperities on one ofthe surfaces in
contact. The process of abrasiveerosion may be considered as
abrasive wear.Erosion, or erosive wear, is the loss of
material from a solid surface due to relativemotion in contact
with a fluid that contains solidparticles. In this case, the
particle is found to bedispersed in a fluid or gas means, and it
reachesthe surface under relatively high velocity(Fig. 25d). Figure
26 shows the microstructureof the transversal section of an H11
tool steelthat has been subject to abrasive erosion.Fatigue wear
can be characterized by the
formation of cracks superficially and/or sub-superficially and
the removal of posteriormaterial due to cyclic loading of solid
surfaces.The sliding contact and/or rolling between solids
Time
Stage I Stage II Stage III0
t
t = creep rate
Fracture X
Fig. 23 Schematic strain-time curve at constant load
andtemperature showing the three stages of creep
Adhesion
Particle
Fig. 24 Transference mechanism of a material from onesurface to
another and the formation of an abrasive
particle in the process of adhesive wear
General Aspects of Failure Analysis / 129
-
or the repetitive impact of solids and/or liquidsin a surface
are responsible for the superficialfatigue. When two surfaces of
this nature inter-act due to load application, the area
effectivelyin contact may be very small, resulting in
highcompressive and shear stresses that may lead tocrack
nucleation. If only rolling is present, themaximum shear stress
takes place just below thesurface, giving rise to cracks that
propagateparallel to the surface and emerge at the surface,causing
part of the material to separate from thecomponent, as shown in
Fig. 27.However, pure rolling is not found in in-
service conditions. Normally, there is somesliding between the
two surfaces, which altersthe stress field due to an increase in
the shearcomponent, displacing the resulting stress closerto the
surface. The cracks start to nucleate onthe component surface,
propagating at a veryshallow angle, as shown in Fig. 28.
Fretting fatigue is considered a phenomenonwhere the damage is
introduced by a conjunctionof events consisting of adhesion,
oscillatorymovement of very low amplitude, oxidation, andabrasion.
The small oscillatory movements maycause points of adhesion on the
surface thateventually break, forming oxidized particles that
(b)
(d)
(a)
(c)
Fig. 25 Abrasive wear. (a) Free particle between two sur-faces.
(b) Particle attached to one of the surfaces.
(c) Sharp asperity. (d) Erosion
Fig. 26 Fractography showing an H11 tool steel that hassuffered
abrasive erosion
Fig. 27 Schematic representation of contact fatigue underpure
rolling between two surfaces
Fig. 28 Damage by contact fatigue in rolling combined
withsliding conditions in gears produced from a quen-
ched and tempered AISI 8620 carburized steel. (a)
Transversalsection. (b) Frontal view from a formed cavity
130 / Failure Analysis of Heat Treated Steel Components
-
act as abrasives on the surface, since the small-amplitude
movements avoid their dispersionapart from the source point. Figure
29 presents amicrograph from a plasma nitrided Cr-Mo-Vsteel, where
a microcrack formed in the frettingregion.More than one mechanism
can be responsible
for the wear observed on a particular part. Themost critical
function provided by lubricants isto minimize friction and wear to
extend equip-ment service life. Gear failures can be traced
tomechanical problems or lubricant failure.Lubricant-related
failures are usually traced tocontamination, oil film collapse,
additivedepletion, and use of improper lubricant for
theapplication. The most common failures are dueto particle
contamination of the lubricant. Dustparticles are highly abrasive
and can penetratethrough the oil film, causing plowing wear
orridging on metal surfaces. Water contaminationcan cause rust on
working surfaces of gears andeventually destroy metal integrity. To
preventpremature failure, gear selection requires
carefulconsideration of the following: gear tooth geo-metry, tooth
action, tooth pressures, construc-tion materials and surface
characteristics,lubricant characteristics, and operating
environ-ment.
Environmentally Assisted Failure
Corrosion is chemically induced damageto a material that results
in deterioration ofthe material and its properties. Corrosioncan
seldom be totally prevented, but it can beminimized or controlled
by proper choice ofmaterial, design, coatings, and occasionallyby
changing the environment. Various types of
metallic and nonmetallic coatings are regularlyused to protect
metal parts from corrosion.Corrosion may result in failure of the
com-
ponent. Several factors should be consideredduring a failure
analysis to determine the effectcorrosion played in a failure, such
as type ofcorrosion, corrosion rate, the extent of the cor-rosion,
and the interaction between corrosionand other failure
mechanisms.Uniform, pitting crevice, galvanic, and stress-
corrosion cracking are the most common typesof corrosion.
Uniform corrosion is characterizedby corrosive attack proceeding
evenly over theentire surface area or a large fraction of the
totalarea. General thinning takes place until failure.On the basis
of tonnage wasted, this is the mostimportant form of
corrosion.Stress-corrosion cracking necessitates a
tensile stress, which may be caused by residualstresses, and a
specific environment to causeprogressive fracture of a metal.
Aluminumand stainless steel are well known for stress-corrosion
cracking problems. However, allmetals are susceptible to
stress-corrosion crac-king in the right environment.Pitting
corrosion is a localized form of cor-
rosion by which cavities or holes are producedin the material.
Pitting is considered to bemore dangerous than uniform corrosion
damagebecause it is more difficult to detect, predict, anddesign
against. Corrosion products often coverthe pits. A small, narrow
pit with minimaloverall metal loss can lead to the failure ofan
entire engineering system. Pitting corrosion,which, for example, is
almost a commondenominator of all types of localized
corrosionattack, may assume different shapes.Crevice corrosion is a
localized form of
corrosion usually associated with a stagnantsolution on the
microenvironmental level. Suchstagnant microenvironments tend to
occur increvices (shielded areas) such as those formedunder
gaskets, washers, insulation material,fastener heads, surface
deposits, disbondedcoatings, threads, lap joints, and clamps.
Crevicecorrosion is initiated by changes in local chem-istry within
the crevice.Galvanic corrosion (also called dissimilar-
metal corrosion or, wrongly, electrolysis) refersto corrosion
damage induced when two dis-similar materials are coupled in a
corrosiveelectrolyte. It occurs when two (or more) dis-similar
metals are brought into electrical contactunder water. When a
galvanic couple forms, oneof the metals in the couple becomes the
anodeFig. 29 Fretting fatigue at the surface of a Cr-Mo-V steel
General Aspects of Failure Analysis / 131
-
and corrodes faster than it would all by itself,while the other
becomes the cathode and cor-rodes slower than it would alone.
REFERENCES
1. D. Dennies, How to Organize a FailureInvestigation, ASM
International, 2005
2. D.J. Wulpi, Chapter 1: Techniques of FailureAnalysis,
Understanding How ComponentsFail, 2nd ed., ASM International,
2000,p 111
3. C.R. Brooks and A. Choudhury, Chapter
1:Introduction,Metallurgical Failure Analysis,McGraw-Hill, 1993, p
172
4. R. Graham, Strategies for Failure Analysis,Adv. Mater.
Process. Aug 2004, p 4550
5. D.A. Ryder, T.J. Davies, I. Brough, and F.R.Hutchings,
General Practice in Failure
Analysis, Failure Analysis and Prevention,Vol 11,Metals
Handbook, 9th ed., AmericanSociety for Metals, 1986, p 1546
6. G.F. Vander Voort, Conducting the FailureExamination, Prac.
Fail. Anal.,Vol 1 (No 2),April 2001, p 1446 andFailure Analysis
andPrevention, Vol 11, ASM Handbook, ASMInternational, 2002
7. A. Tanzer, Determination and Classificationof Damage, Failure
Analysis and Prevention,Vol 11, ASM Handbook, ASM
International,2002
8. G. Powell, Identification of Types of Failure,Failure
Analysis and Prevention, Vol 11,Metals Handbook, 9th ed., American
Societyfor Metals, 1986, p 7581
9. C. Laird, The Influence of MetallurgicalStructure on the
Mechanisms of FatigueCrack Propagation, Fatigue Crack Propa-gation,
STP 415, ASTM, p 131168
132 / Failure Analysis of Heat Treated Steel Components
-
Failure in Steel ForgingMd. Maniruzzaman and Richard D. Sisson,
Jr.,Worcester Polytechnic InstituteStephen R. Crosby, The Stanely
WorksCharlie Gure (deceased)
IN-PROCESS OR SERVICE FAILURES offorgings may occur for a
variety of reasons. Thestartingmaterial may be of insufficient
quality tobe adequately formed without cracking, or theforging
process may introduce various types ofdiscontinuities that cause
failure during services.For example, well-known forging-related
dis-continuities include:
Laps Bursts Flakes Segregation Cavity shrinkage Centerline pipe
Parting-line grain flow Inclusions
Forging discontinuities are discussed in moredetail in the texts
on forging (Ref 14).This article describes six case studies of
failures with steel forgings (summarized inTable 1). The case
studies illustrate difficultiesencountered in either cold forging
or hot forg-ing in terms of preforge factors and/or
dis-continuities generated by the forging process.Tables 2 and 3
summarize these factors forcold and hot forging, respectively.
Supporting
topics that are discussed in the case studiesinclude:
Validity checks for buster and blockerdesign
Lubrication and wear Mechanical surface phenomenon Forging
process design Forging tolerances
As case studies were being selected, each ofthe aforementioned
supporting topics wasreviewed for any impact that particular
studyhad on the case being examined. It is a well-known fact that
forging solutions have severalpossible avenues to follow. There is
no uniquetheory in plasticity that leads to the solution.Most of
the work reported here was performedusing the minimum amount of
energy to createthe particular product. Factors unrelated to
thedeformation process, such as chemistry, micro-structure, phase,
grain size, segregation, andprior strain history, are not addressed
here.Instead, factors directly related to the deforma-tion process
itself are presented in this abbre-viated discussion.Wear, plastic
deformation processes, and
laws of friction are introduced as a group of
Table 1 Failure analysis of steel forgings and components
Case study Defect Solution
Crankshaft underfill Unable to fill crankshaft flanges with
existing press
capacity
Introduce creep stages for last increment of
displacements
Tube bending Unable to control exterior wall thinning and
interior wall
thickening
Introduce induction heating and cooling to limit the
heated axial tube length prior to making the bend
Spade bit Unable to achieve center web thickness at
programmed
force and sufficient flow to wings
Adjust the die angle to create more shear stress, enabling
full flow to the wings
Trim tear Forge material tore at trimline when forging was
trimmed immediately following finish forging
Introduce a delay time after forge and prior to trim,
allowing the forge material to cool and gain strength
Upset forging Cracking at circumferential bulge after upset
Re-examine the strain and strain rate and process map
for stable flow
Flow-through laps
and avoidance
Material foldover at tops of rib and flange intersections
and cases of material flow under previously filled
flanges
Replace the input piece with a newly designed preform
piece, following the design procedures given in this
work
Failure Analysis of Heat Treated Steel Components
L.C.F. Canale, R.A. Mesquita, and G.E. Totten, editors, p
133-149
DOI: 10.1361/faht2008p133
Copyright 2008 ASM International
All rights reserved.
www.asminternational.org