How to Conduct Failure Analysis 10/8/09 5:49 AM
How to Conduct
FAILURE ANAYLSIS
Proper failure analysis requires
following the proper sequence of tests,
collecting appropriate data,
and evaluating results.
This bolt failed because of ductile overload
Typical nondestructive examination methods
Typical chemical analysis methods
Typical mechanical testing methods
Failure analysis consists of investigations to find out how and why something failed. Understanding the actual reason for failures is
absolutely required to avoid recurrence and prevent failure in similar equipment. The analysis should also help with the
understanding and improvement of design, materials selection, fabrication techniques, and inspection methods.
This article reviews the steps to follow when conducting a full failure analysis investigation.
Acquiring background information
The failure analysis sequence generally follows an order of increasing “destructiveness” of the test and/or sample removal.
Significant deviation from this recommended hierarchy may prevent crucial evaluation techniques through damage caused by
previous tests. Naturally, methods may be added or deleted from this sequence provided consideration is given to the best order of
analyses.
• Part information: Detailed information about a failed component often facilitates selection of analytical methods and can provide
insight into some of the factors that may have contributed to the failure. Certain test methods may be suggested by knowledge of the
component manufacturing history, and this could lead to a quicker solution. This information should include as a minimum:
specifications, manufacturing information, part number and serial number, and drawings with a bill of materials.
• Service history: The history of a failed part is also of great importance to the analyst. All information concerning the actual record
of a part can serve to illuminate the causes of a failure. Even “typical” service, which may be ostensibly identical to similar
units in similar conditions, may initiate failure due to apparently innocuous or mundane differences that may not initially seem
worthy of mention.
• Investigation planning and sample selection: The planning portion of an investigation is crucial to determining the proximate cause
of failure. Proper planning can ensure that an investigation is efficient and cost effective. Particularly in the case of a high visibility
failure or if an assembly line shutdown is imminent, careful planning is necessary to hasten problem resolution. Haphazard or
unsystematic investigations are unprofessional and can be wasteful of time, effort, and manpower.
Equation for successful failure analysis
Proper background information
correct analyses and tests
Thorough knowledge of materials behavior and processing
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
How to Conduct Failure Analysis 10/8/09 5:49 AM
Pertinent experience
_________________________
Correct Analysis
Preliminary examination
All pertinent features should be examined visually, and thorough notes should be taken for each component associated with a failure
analysis. Many macroscopic features and characteristics can suggest certain failure modes and circumstances. Fracture surfaces can
exhibit an identifiable origin, progression marks, direction marks, and postfracture damage, all of which can be significant to the
failure.
All of the pertinent features noted in the visual examination should be well documented by photographs. The component condition as
well as any stamped information /part numbers should be illustrated. This should be done as soon as possible after failure, especially
if the fracture surfaces continue to corrode. If smaller sections are to be excised for analysis, this step can document their locations
and orientations. Photography is necessary because subsequent dissection of the sample may preclude further visual examination.
Failure analysis rules
The failure investigator has only one objective. to determine the failure
mechanism that caused the failure and to use that knowledge to prevent
another occurrence:
Start with and maintain an open mind. Emotion obscures objectivity and
must be edged from the investigation.
The theory, however elegant, must agree with the observed evidence,
however humble.
The simplest solution is the best solution.
Having the wrong solution is frequently much worse than having no
solution.
Major incidents are often triggered by very minor or apparently
innocuous details.
Only one thing is worse than knowing you have a crack growing in a
component, and that is not knowing you have a crack growing in a
component.
Cracks never get smaller, nor do they ever disappear. Either they stay the
same size (invariably in a minor, insignificant, or easily repaired
component), or they get bigger (usually in a critical component).
Material evaluation
Thoroughness and diligence mandate that prudent material evaluation be an integral part of any failure analysis. Many individuals
(with and without qualifying experience) feel sufficiently confident to render sophisticated failure analysis judgements without the
requisite testing. Absurd as this practice may seem, it is done every day. Naturally, a wise analyst will always select test methods that
are appropriate to the physical evidence of an investigation. A general rule in this regard is that failure analysis minus the analysis is
just a failure.
Equally deleterious is to disregard some of the necessary tests. Sometimes analysts take a “target of opportunity” approach, stopping
an investigation at a point where a discrepancy is found (improper chemical composition, low mechanical strength, etc.). The
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
How to Conduct Failure Analysis 10/8/09 5:49 AM
conclusions would then be composed around a hypothesis based on an incomplete investigation. Remember that most material
failures are complex and may contain several significant contributory causes. Although it may be economically attractive to conduct a
minimal evaluation, the proximate failure cause may be missed, and improper recommendations may be made as a result of this
ill-advised practice. The following tests should all be considered:
• Environmental testing: Environmental tests may simulate the corrosivity and temperature extremes experienced by a component in
service. Either the failed part or an exemplar part that has been processed similarly could be subjected to these procedures, which
may reveal behavior that could cause failure.
• Mechanical testing: Mechanical properties have been historically measured by material suppliers and are often reported to the end
user. Properties measured on a failed material can often be directly compared to pertinent specifications and prior property data.
Discrepant strength levels can be determined, and in many cases degradation of these properties during service can be identified.
• Chemical analysis: Chemical analysis is an integral part of an investigation, because it indicates whether a component is made of
the specified material. Subtle variations in composition can often dictate the strength and property values that processing can develop.
In addition, relatively low amounts of impurity elements can cause significant changes in these same properties.
• Metallography: Examination of cross sections of materials involved in a failure can provide important insights into the probable
cause of the incident. One of the basic expressions of materials science can be expressed as follows: Properties =
structure=processing.
Determination of material structure can indicate the likely mechanical and physical properties of a component. Similarly, this
examination can reveal crucial processing information, which may indicate incorrect or incomplete heat treatment or other required
structural alteration. Metallography can reveal macrostructure, such as depth of surface hardening or banding, and it can also reveal
microstructure, such as grain size and the phases present.
Carefully prepared cross excellent depth of field, which is a boon to morphological identification.
In addition to the fracture mode, electron microscopy can help identify types of mechanical damage, such as adhesive or abrasive
wear, corrosion, and other distinct features. To some extent, surface roughness or machining quality can also be evaluated. Corrosion
products and inclusions can be examined via electron microscopy, and these minute features can be analyzed and identified by X-ray
spectroscopy or other techniques.
In many cases, the skill of the metallographer is key to revealing the critical evidence that leads the analyst to the correct solution.
Poorly prepared Tension specimens can ruin the analysis by destroying critical inclusions, porosity, or other crucial evidence.
• Fractography: Fractography generally involves a stereoscope along with an electron microscope. It is an indispensable tool for the
analyst, because much fracture information cannot be revealed by any other technique. Magnifications as high as 10,000X can reveal
many features that cannot be seen by standard light microscopy. The limit of resolution of a microscope is the wavelength of the
imaging illumination, hence visible light would exhibit far less resolution than electron microscopy. Electron microscopes can also
offer excellent depth of field, which is a boon to morphological identification.
In addition to the fracture mode, electron microscopy can help identify types of mechanical damage, such as adhesive or abrasive
wear, corrosion, and other distinct features. To some extent, surface roughness or machining quality can also be evaluated. Corrosion
products and inclusions can be examined via electron microscopy, and these minute features can be analyzed and identified by X-ray
spectroscopy or other techniques.
Special testing
This group of tests includes all those that do not easily fit into the previous categories. The following are a few of the more important
techniques.
• Fracture mechanics: This tool can approximate the stresses surrounding a fracture to better explain its occurrence. By
making.certain assumptions and isolating the fracture location, simulations of crack
growth scenarios can be calculated. By altering the determined model parameters, the analyst can estimate the magnitude of the
mechanical stress applied to a component at the time of failure. This is often used in tandem with NDE-measured flaw sizes to
determine the reduced loads necessary to produce crack growth with a flaw of certain dimensions. Mechanical properties measured
during destructive physical analysis can be placed in this model to acquire more specific information.
• Finite element analysis: Finite element analysis, or FEA, is an advanced modeling technique that can help to predict the magnitude
of stresses on individual components within complex assemblies. This type of analysis also utilizes the computational capabilities of
current technology. Changing the 1ocanon and magnitude of simulated loads on the three-dimensional model may indicate the reason
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
How to Conduct Failure Analysis 10/8/09 5:49 AM
for a failure. In some circumstances, this type of engineering analysis is of utmost importance when physical evidence is not
available but mechanical loading is known.
• Simulation: In some instances, sophisticated test beds and test apparatus can be developed for approximating the service conditions
involved in a materials failure. Although this tool is more often used during product development, the ability to reproduce a failure is
a very clearcut proof of a failure hypothesis. By altering the physical simulation through inserting the specific parameters associated
with the failure, an exemplar part similar to the failed one can be tested. Naturally, the actual factors contributing to a failure may be
too numerous and varied to adequately simulate; however, failures in which mechanical stresses are the only significant causes are
not uncommon.
Typical nondestructive examination methods<
/FONT>
Acronym Technique Information provided
Detection of anomalies by differential
EC Eddy current electrical current response
Identification of features and structure
MRI Magnetic resonance imaging via introduction of
alternating magnetic fields
Detection of surface and near surface
MT Magnetic particle examination flaws in ferromagnetic materials by flux
leakage
Identification of flaws and cracks open to
PT Penetrant testing the surface in many materials via liquid
retention
Identification of material flaws and
RT Radiographic examination features via density differences measured
by penetrating radiation
Detection of anomalies by differential
UT Ultrasonic examination reflection of ultrasonic pulses
Typical chemical analysis methods
Acronym Technique Information provided
AA Atomic absorption Analysis of dissolved materials in a
gaseous flame.
AES Auger electron spectroscopy Composite analysis of surface layer
including depth profiling.
DSC Differential scanning calorimetry A measure of heat flow related to
temperature.
EDS Energy dispersive X-ray spectroscopy Chemical analysis of small features and
particles in the electron microstructure.
ESCA Electron spectrometry for chemical analysis
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
How to Conduct Failure Analysis 10/8/09 5:49 AM
FTIR Fourier transform infrared spectroscopy
GC Gas chromatography Quantitative analysis of organic
materials after volatilization
GDS Glow discharge spectroscopy Analysis of metallic materials vaporized
in an arc
GPC Gel permeation chromatography Form of liquid chromatography for
molecular weight distribution of
polymers
ICP Inductively coupled plasma spectroscopy
LC Liquid chromatography Quantitative analysis of dissolved
materials by separation techniques
MS Mass spectroscopy Detection of mass units of organic
materials in a magnetic field
OES Optical emission spectroscopy Analysis of metallic materials vaporized
in an arc
RS Raman laser spectroscopy Qualitative identification of organic
chemicals using Raman absorbance
Sims Secondary ion mass spectroscopy Surface analysis technique detects all
elements
TGA Therxnogravimetric analysis A measure of weight change related to
temperature
TMA Thennomechanical analysis A measure of physical properties related
to temperature.
XPS X-ray photoelectron spectroscopy Elemental analysis of surface layers.
XRD X-ray diffraction Analysis of materials and crystal
structures by X-ray impingement.
XRF X-ray fluorescence spectroscopy Bulls analysis of solids or liquids by
X-ray excitation.
Typical mechanical testing methods
Typical mechanical testing Test Information provided
methods
Bend Measures ductility of base materials and weldments
Brinell hardness Indentation hardness test has a high load and measures a relatively large area.
Compression Test measures inherent resistance to fracture under compressive loading. Compression modulus
can be calculated.
Creep / stress rupture High temperature tension test measures resistance to rupture by a creep mechanism.
Drawability Measures the formability of a material to determine its suitability for severe forming during
manufacturing.
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
How to Conduct Failure Analysis 10/8/09 5:49 AM
Durometer hardness Surface hardness testing method for plastics and elastomers.
Fatigue Cyclic loading test to determine fatigue resistance.
Flexure impact Three or four point bending test measures ductility. Rapid point loading test determines
impact resistance or toughness, often at simulated service temperature. (Charpy V-Notch,
Izod, etc.)
Knoop microhardness Microindentation hardness testing method for very small areas or features.
Rockwell hardness Indentation hardness tests measure an intermediate amount of surface area.
Scleroscope hardness Surface hardness measurement of softer materials.
Shear Measures resistance to fracture when shearing loads are applied. Most commonly performed
as single or double shear methods.
Superficial hardness Lower load varieties of the standard Rockwell tests for thin samples or thin surface layers.
Tension Mechanical test pulls a material apart and can determine the ultimate tensile strength, yield
strength, elongation, reduction in area, modulus of elasticity, and other properties.
Torsion Mechanical test loads a material in a twisting manner to measure its strength in torsion.
Vickers microhardness Microindentation hardness testing method for very small areas or features.
Analyze data and prepare a report
After all of the systematic examination and data collection has been completed, the information must be organized and interpreted for
its significance to the failure. The assembled results from the different tests must be considered collectively, because the final
hypothesis needs to be in substantial agreement with all physical evidence and test results. As the investigation reaches its fruition,
most of the possible failure causes can be conclusively discounted and a single, coherent explanation is indicated. At this stage of the
investigation, it pays to have kept an open mind throughout, because the results may not conform to preconceived notions or
preferred outcomes.
Presentation of the results in a technical report is a very essential portion of a failure analysis. The data should be arranged and
organized in such a manner that the information proceeds from general to specific, similar to any professional technical report. The
training and experience of the investigator are required to assemble the often large volume of information into a contiguous report
that logically leads to the conclusions.
In many failure analyses, the resolution of a single incident is subordinate to the goal of avoiding similar failures in the future. A
good investigation can provide specific recommendations concerning design, materials, and processing changes that will avert
identical failures in other products. One of the possible exceptions to this is in litigation, where determination of the specific failure
cause is paramount and may supersede other interests.
Naturally, changing the properties of any component in a complex assemblage of parts can alter the forces on adjacent components,
and these changes may not always be readily predicted. All recommended changes must be evaluated prior to implementation, as
other, potentially more severe failures may result.
http://www2.mtec.or.th/th/research/famd/lcenter/failure3.html
