7/27/2019 Davidson.html
1/22
TMS Outstanding StudentPaper Contest Winner--1999 Undergraduate Division
An Introduction to Failure Analysis for Metallurgical Engineers
Thomas Davidson
FORWARD
The objective of this paper is to introduce the reader to therocedures generally followed when conducting a
metallurgical failure analysis. Due to the large number, of
ossible causes of failures, this report will not delve deeplyinto theory. Instead, six failure case reports are provided toallow the reader to learn by example. For this reason, thereader is expected to have some background knowledge ofailure mechanisms. However, the paper includes a detailed
bibliography containing several sources that were used duringmy summer employment to help carry out these cases. The sixcases presented are cases I worked on over the summer of 98or Noranda Technology Centre in the Materials Technologyor Failure Prevention group.
PROCEDURETo increase the odds of completing a conclusive failureanalysis while at the same time saving time and money,investigations should be carried out using a systemic approachsimilar to that outlined in Figure P.1. It is important to notehowever, that it is often impossible to foresee results that mightrequire the investigator to go back and repeat a test. A simpleway reduce the occurrence of this is to go into a case wellinformed on how similar systems have failed. An excellentsource of for this type of information is the ASM handbooks,particularly volume 10 on "Failure analysis and prevention".This book is an invaluable reference to the beginner and theexpert and should be consulted regularly. Another importantsource of information are the standards by which the part wasmanufactured. These standards give the investigator ameasuring stick by which to compare, as well as indicatingareas of importance. There are many organisations that producestandards for different applications and several organisationsstandards can overlap. It would be a good idea for theinvestigators to spend some time familiarising themselves with these organisations and how the
CONTENTS
FORWARD PROCEDURE CASE STUDIES
Introduction to CaseStudies Case Study 1: Crane Bolt
Failure Case Study 2: Rider
Roller Shaft Failure Case Study 3: Crane Pin
Failure Case Study 4: Shaft
Bearing Failure Case Study 5: Bronze
Bull Gear Failure
Case Study 6: Analysis of316L Reducer Failure
APPENDIX 1: EXAMPLEQUESTIONNAIRE
Bibliography
Page 1 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
2/22
standards are used. Table P.1 gives a brief list of the more common organisations that write standardsand their general area of coverage.
The first step in conducting any failure analysis is to gain a good understanding of the conditions underwhich the part was operating. The investigator must ask questions from those who work with, as well asthose who maintain the equipment and visit the site whenever possible. Contacting the manufacturermay also be necessary. A simple questionnaire, presented in Appendix 1, is a good place to start and will
lead the investigator to more detailed questions. Unfortunately, in many instances the investigator willreceive a failed part with little information about its history and operating conditions. In cases such asthese the physical evidence will have to be more heavily relied on.
The second step is to conduct a visual examination, cataloguing and recording the physical evidence atthe same time. This serves the functions of:
Familiarising the investigators with the evidence. Creating a permanent record that can be referred to in light of new information.
Samples should be examined, photographed and sketched taking particular care to identify and record
Figure P.1. Chart outlining the major steps that are usually taken when conducting a failureanalysis.
Table P.1--Common standard organisations and their general area of coverage.
Acronym Coverage
AISI Steel composition standards
ASTM Standards for materials and their manufacture
API Petroleum industry standards which are used by many other industries
ASME Responsible for Boiler Pressure vessel codes
NACE Codes for materials exposed to corrosive environments
SAE Automotive industry standards used by many other industries
UNS Classification for metals and metal alloys
Page 2 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
3/22
any area of particular importance, such as fracture surfaces and surface defects. Visual examination canbe aided by the use of a stereomicroscope with lights that can be easily directed. Shadows can givedepth to a surface making it easier to analysis and photograph. Pieces should always be examined andrecorded before any surface cleaning is undertaken. In some cases substances such as dirt, paint and Oilon the surface can themselves be important clues, indicating such things as how old the fracture surfaceis and in what kind of environment the piece was operating. A good general rule is to be conservativewhen destroying evidence of any kind. The visual examination is a good time for the investigator to
examine the fracture surfaces in detail and try to identify the mode of fracture (brittle , ductile, fatigue,etc.), points of initiation, and direction of propagation. Each mode of fracture has distinct characteristicsthat can be easily seen with the naked eye or the use of a stereomicroscope, however, sometimes ascanning electron microscope (SEM) will have to be used. There are several good books, some listed inthe bibliography, on fracture mechanism and compilations of fracture surface photographs that can beused by the investigator to identify the mechanism of fracture under investigation. As a reminder, somecommon fracture surface characteristics arc listed in Table P.2 with their corresponding mechanism.
The third step is to decide on a course of action. Based on the visual examinations and the backgroundinformation the investigator must outline a plan of action, which is the series of steps that will be neededto successfully complete the case. There are several resources that an investigator can draw on todetermine the cause of failure, which can classified into one of the following categories:
Macroscopic examination Non-destructive testing (NDT) Chemical analysis Metallographic examination Mechanical Testing
Many of these categories will require steps that use the same equipment and therefore much time can besaved with a little forethought. The macroscopic examination is best performed when cataloguing thesamples, however the investigator will often want to return to examine the part in more detail once otherevidence is gathered. Use of a scanning electron microscope (SEM) is often useful at this stage because
Table P.2--Fracture mechanisms and their fracture surface characteristics.
Mode of Fracture Typical fracture surface Characteristics
Ductile Cup and ConeDimples
Dull SurfaceInclusion at the bottom of the dimple
Brittle Intergranular ShinyGrain Boundary cracking
Brittle Transgranular ShinyCleavage fractures
Flat
Fatigue Beachmarks
Striations (SEM)Initiation sites
Propagation areaZone of final fracture
Page 3 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
4/22
of its large range of magnifications and its large depth of field. Since undamaged fracture surfaces arenot always available, it is often a good idea to open other cracks that may be present in the piece. Thisoften reveals good quality fracture surfaces similar to those that caused failure. Procedures for doing thiscan also be found in the ASM handbook volume 10.
Nondestructive tests (NDT) are a good way to examine parts without causing permanent damage. Oftentimes, results obtained from examining failed parts in the lab using NDT's can be used to examine parts
in the field and remove them from service before failure occurs. There are several NDT's that areavailable to the investigator and it would be a good idea to read up on each ones abilities. Table P.3gives an outline of NDT's available and what they are able to detect.
Chemical analysis is done on the bulk of the material to confirm the material composition. Dependingon the investigation, chemical analysis should also be done on any overlay materials or surface residues.There arc several techniques that can be used to check composition, and the choice of which to use oftendepends on accessibility and sample type. In many cases, the SEM can be a powerful tool for fastidentification of surface materials. Care should be taken not to contaminate samples taken for chemicalanalysis by surface residue or cutting instruments.
Metallographic examination involves the sectioning of samples to examine the microstructure. The
sections that are selected for examination are dependent on the type of piece and the mode of fracture.Sections from the sample should be taken in different planes so that any differences in themicrostructure can be seen. Sometimes it is useful to take a cross section through the fracture surface sothat the microstructure below the fracture and the surface profile can be examined. A section runningparallel to the fracture surface is also often taken for examination. Samples should be mounted, ground,and polished using metallographic techniques. They should be examined before etching for porosity,inclusions, and other defects. Microstructures should be identified and their properties researched. Thereare several referenced that the investigator can refer to for identification of uncertain structures.
Mechanical testing is done to verify that the mechanical properties of the material conform to the
Table P.3--Commonly used nondestructive tests and there capabilities in detecting defects.
NDT Method Capabilities
Radiography Measures differences in radiation absorption.
Inclusions, Porosity, Cracks
Ultrasonic Uses high frequency sonar to find surface and subsurface defects.
Inclusions, porosity, thickness of material, position of defects.
Dye Penetrate Uses a die to penetrate open defects.
Surface cracks and porosity
Magnetic Particle Uses a magnetic field and iron powder to locate surface and near
surface defects.
Surface cracks and defects
Eddy Current Based on magnetic induction.
Measures conductivity, magnetic permeability, physical
dimensions, cracks, porosity, and inclusions.
Page 4 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
5/22
standards. There are many types of mechanical testing that can be performed and their procedures can befound in the ASTM mechanical testing standards. The most common method used is hardness testingbecause of its relative simplicity, low cost, and the fact that for many materials tables exist to relatehardness with yield strength. A macrohardness is usually sufficient to determine material properties,however microhardnesss measurements are helpful in determining property variations within thematerial. Use the microhardness measurement to compare the surface hardness to that of the body or toverify the microstructure. Other mechanical testing such as tensile tests and impact tests can be used,
however their use is usually limited by insufficient material and high costs .
Once all the data is gathered, the investigator must come to a conclusion based on the evidence present.This requires that the investigator draw heavily on background experience and research performed. Thisstep can be difficult because when conducting the investigation clues will lead the investigator downpaths that seem to be the cause but which are merely consequences.
The final and most difficult step in any investigation is coming up with recommendations. Some caseswill be simple, however many cases are not obvious even though the cause and theory are known.Recommendations are not to be taken lightly. Serious failures can occur if recommendations are in error.The system may have to be redesigned or a new material put in place. Sometimes all you will be able to
recommend is that inspections be carried out more often.
CASE STUDIES
Introduction to Case Studies
These case studies are actual reports submitted in response to industrial failures. The purpose of thesereports is to demonstrate by example. Most of the cases mention the techniques that where used whenstating the results. They where written at a basic level due to the uncertainty of background of the readerand further reading is be recommended to better understand the failure mechanism. Most of the casesthat are presented here have comparable cases in the ASM failure analysis handbook.
Case Study 1: Crane Bolt Failure
Introduction:
One of two bolts supporting a load of 16 200 lbs failed while in service causing eight hours of downtimeon an essential machine to production. The bolts were in operation on a crane used to transfer anodesinto the machine. Figure 1.1 shows a drawing of the set-up and the location of fraction Just above thenut. The crane cycled 600 time a day 7 days a week.
The broken bolt (Figure 1.2) and a new unused bolt, recommended by the supplier for the application,
were supplied to conduct the investigation. The original designers of the crane specified a bolt thatconforms to SAE standards grade 5. The supplier of the new bolt confirmed that it was made to conformwith ASTM standard A 193 grade B7.
Page 5 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
6/22
Results:
Observations
Examination of the fracture surface revealed characteristics such as a beachmarks associated withfatigue (Figure 1.3). The zone of final fracture was located between two areas of fatigue propagationsuggesting the presence of bending forces. The surface area of final fracture was approximately 12% ofthe total fracture surface suggesting that the bolt was not overloaded. Cracks where also found betweenthreads near the fracture surface indicating that the bolt was highly susceptible to fatigue initiation.
Results from chemical analyses (Table 1.1) show that the original broken bolt had a carbon contentslightly below those required by the SAE standards for a grade 5 bolt. This lower carbon content wouldhave acted to decrease the material properties. The chemical composition of the new sample boltconformed to the ASTM standard A193/A grade B7 that requires an AISI-SAE 4140 composition.
Figure 1.1. Drawing of the bolt
and crane set-up.
Figure 1.2. Photograph of
broken bolt
Figure 1.3. Photograph of
fracture surface.
Table 1.1--Chemical analysis results on both bolts.
Element Original broken bolt (%)
SAE Standard
Grade 5 (%)
New Sample
Bolt (%)
ASTM Standard B7
AISI 4140 (%)
Carbon 0.20 0.28-0.55 0.42 0.37-0.49
Manganese 0.65 -- 0.85 0.65-1.10
Silicon 0.22 -- 0.22 0.15-0.35
Phosphor 0.013 0.048 max. 0.015 0.035
Sulphur 0.011 0.058 max. 0.030 0.040
Chrome 0.08 -- 0.79 0.75-1.20
Nickel 0.06 -- 0.07 --
Molybdenum 0.01 -- 0.15 0.15-0.25
Page 6 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
7/22
Microscopic examination of the bolts where done using longitudinal and latitudinal mounts for each.The sections taken from the fractured bolt were taken close to the fracture surface. Examination beforeetching of the two bolts showed no cracking or unusually large inclusions. The original broken bolt didshow some flaking at the base of the threads (Figure 1.4) but this is expected for a bolt that has been in
service. Etching the sections revealed a microstructure of coarse pearlite in a matrix of ferrite (Figure1.5). The SAE grade 5 standard requires that the bolt be quenched and tempered to conform andtherefore should have a tempered martensite structure. Martensite has higher material properties such asyield strength and hardness, which increases its resistance to fatigue initiation. The ferrite matrix of theoriginal bolt has low yield strength, which in turn reduces its resistance to fatigue initiation. The newbolt was found to be quenched and tempered as required by the ASTM standard (Figure 1.6). Howeverrolling seems where found at the tips of the treads (Figure 1.7). This is not a serious defect because ofthe defects location in a low stress area however, if the bolt was placed in a corrosive atmosphere theseseams would corrode and then act as fatigue initiation sites.
Tensile tests were done on the bolts to test their material properties in comparison with the standards.The results (Table 1.2) show that the yield strength and ultimate tensile strength of the original bolt areonly two thirds that required by the standards. This conforms to the microstructural observations. Theproperties of the new bolt conformed to the standard even though they were slightly elevated.
Figure 1.4. Micrograph offlaking found at the baseof a thread in thefractured bolt. 2% nital
100X
Figure 1.5. Micrograph offractured bolt. Ferritematrix with pearlite. 2%nital 200X
Figure 1.6. Micrograph ofnew bolt. Temperedmartensite. 2% nital500X
Figure 1.7. Micrograph ofthe new bolt threadshowing a rolling seam.2% nital 200X
Table 1.2--Results and standard requirements of tensile tests.
Original Broken Bol New Sample BolStandard Grade
5 SAE
Standard Grade
By AISI
Sample # 1 2 1 2
Ultimate Tensile Strength (KSI) 69.5 69.5 148 146 100 125
Yield Strength (KSI) 42.7 44.4 134 133 80 105
Elongation (%) 26 24 20 20 16 min. 16 min.
Surface Reduction (%) 67 67 59 59 50 min. 50 min.
Page 7 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
8/22
Conclusions and Recommendations:
Examination revealed that the bolt failed as a result of high cycle low load fatigue. Chemical analysisand tensile tests confirmed that the bolt did not meet the SAE grade 5 standards required by the originaldesign of the crane. The major cause for this lack of conformity is because the bolt was not quenchedand tempered. Since the resistance of steel to fatigue initiation in proportional to its yield strength, thelow properties of the steel in this case left it open to fatigue initiation.
Examination of the new bolt revealed that it conformed with the ASTM standards A 193 for a grade B7bolt, as the supplier specified. However, rolling seams were found in the thread tips. Due to therelatively low loads this area is subjected to this is not a major problem but if the bolt is subjected to acorrosive environments these seams could grow and become fatigue initiation sites.
The SAE grade 5 bolt specified by the original designers should continue to be used in future and theupgrade to the ASTM B7 is unnecessary.
Case Study 2: Rider Roller Shaft Failure
Introduction:
A section of a failed "rider roller" shaft was sent for failure analysis (Figure 2. 1). This shaft is designedto ride on top of cardboard as it is being rolled. It was first installed in December 97 replacing a shaft inwhich cracks were observed near the ends. In March 98 a crack was observed in the centre of the roll.Since no replacements were available at the time, welding was used to repair the crack. This caused theshaft to become out of round by 0. 140". To repair this a hydraulic Jack was used at the centre of the rollto bend it back leaving a 0.040" deflection that was corrected by machining. Nine days later, on April11th 98 at 21: 00, the shaft broke on the key-way side while the machine was being set up at low speed.The roll usually operates at 550 meters per minute, approximately 630 RPM.
The low carbon steel shaft was suppose to have a stainless steel weld overlay applied before installationto protect against corrosion in the mill environment. 17-4PH steel was used for this application beforeand failed to endure the high cycle low stress conditions.
Figure 2.1. Photograph of "riderroller" indicating approximatepoint of fracture.
Figure 2.2. Photograph offracture surface showinginitiation site, beachmarks fromfracture propagation, and smallarea of final fracture.
Figure 2.3. Photograph of shaftsurface indicating weld overlayflaw.
Page 8 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
9/22
Results:
Observations:
The fracture surface is characteristic of a high cycle fatigue failure caused by low torsion stresses(Figure 2.2). The area of final fracture is small, approximately 35% of total area, indicating that thematerial was adequate for the low applied stresses. The beachmarks (Figure 2.2), characteristics offatigue that radiate from the initiation site, and the location of final fracture, being off centre, indicatedthat initiation did not occur evenly around the circumference of the shaft. Around the circumference ofthe fracture surface, a layer was observed which fractured at a 45' angle to the plane of fracture. This ischaracteristic of the weld overlay. As well, there were many grooves running around the outside of theshaft that are weld overlay features (Figure 2.3).
Materials characterisation and evaluation:
Chemical analysis of the material revealed it to be low carbon steel. Compositions correspond to the
AISI 1019 specifications (Table 2.1). Using the alloy analyser, the weld overlay was found to be a lowalloy steel, probably type EFe, and not stainless steel as was thought.
Microscopic examination revealed the core to have a ferrite and a coarse pearlite structure characteristicsof low carbon steel (Figure 2.4). The weld overlay had pearlite matrix with some acicular ferrite (Figure2.5). A microhardness test revealed a hard surface that gets progressively softer towards the core (Table2.2). This concurs with the microstructure. The inclusions present in the core of the shaft whereacceptable (Figure 2.6).
Table 2.1--Result of shaft chemical analysis.
Element Analysed Composition of Shaft (%)AISI-SAE 1019
Standard Composition Ranges (%)
Carbon 0.19 0.15-0.20
Manganese 0.70 0.70-1.00
Silicon 0.26 --
Phosphorus 0.020 0.040 max.
Sulphur 0.020 0.040 max.
Chromium 0.10 --
Nickel 0.17 --
Molybdenum 0.02 --
Table 2.2--Results of microhardness measurements.
Distance from Surface (m) Hardness HVN-200g
Page 9 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
10/22
Examination of a longitudinal mount taken from near the point of major crack initiation sites showedlarge inclusions between weld passes (Figure 2.7). Examination of the fracture surface initiation sites(Figure 2.8), on the same sample, showed an initiation site on the fracture surface that is similar in shapeand size to the inclusions. This suggests that these inclusions acted as stress raisers and thus as fatigueinitiation sites. The bending of the shaft would have caused decohesion of the inclusions and increasedthe chances of fatigue initiation. Decohesion of the weld overlay between welding passes can also beseen around the circumference of the shaft (Figure 2.3). This indicates poor bonding between the weld
overlay and the base material.
Conclusions and Recommendations:
The failure was caused by high cycle low stress fatigue, which was initiated at inclusions in the weld
overlay. For this kind of failure, when there is an absence of other defects, the surface conditionsbecome an important factor in the prevention of crack initiation. Bending the shaft to correct itsalignment probably caused decohesion of the weld inclusions encouraging microcracks to form. Thiswould have increased local stress concentrations and the possibility of crack initiation. These inclusionsprobably originated from the weld being applied too quickly.
The use of a weld overlay to reconstruct existing rolls is an acceptable procedure provided the weld isapplied property. This would harden the surface and thereby make the shaft more resistant to fatigueinitiation at surface defects. A welding procedure should be developed that would involve the making ofblock samples in which the welding conditions, such as current and speed, are varied and optimised.
35 257
42 271
107 255
140 247
214 187
252 187
Core 156
Core 167
Figure 2.4.Micrograph of core
microstructurecomposed of ferriteand pearlite. 2%nital 100X
Figure 2.5.Micrograph of weld
overlaymicrostructurecomposed of apearlite matrix withthe presence ofacicular ferrite. 2%nital 500X
Figure 2.6.Micrograph
representing averageinclusion content ofthe low carbon steelcore. 2% nital 100X
Figure 2.7.Micrograph showing
two inclusions foundin the weld overlay2% nital 15X
Figure 2.8.Micrograph showing
the fracture surfaceinitiation site. 2%nital 15X
Page 10 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
11/22
Noranda Technology Centre can help in developing a procedure. A liquid penetrant inspection should beperformed to inspect the weld overlay for any cracks or porosity.
Future shafts should be made out of low alloy steel AISI-SAE 4340, heat-treated to a hardness of 35HRC. The properties of this material fall between those of 1019 and 174PH. It will resist crack initiationbetter than the former, due to its higher endurance limit, and will resist crack propagation better than thelatter, due to its higher fracture toughness (Table 2.3).
Other recommendations are:
Avoid bending of shafts that have been surface hardened or had weld overlay applied due to thehigh possibility of inducing surface cracks.
Avoid mechanical damage to the surface, such as scratches and dents, because they can act ascrack initiation sites.
Corrosion can be prevented in both cases by applying a coat of paint.
Case Study 3: Crane Pin Failure
Introduction:
After several failures, a pin connecting a chain to a load transfer bloc was sent for failure investigation(Figures 3.1a and 3.1b). The conditions of operation are similar to those under which the bolt in casestudy I was operating.
Table 2.3--Fatigue related properties of selected materials.
Material Endurance limit (MPa)
Fracture Toughness
1018 275 260
4340 450 110
17-4PH -- 53
Page 11 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
12/22
Observations :
The pin was broken in two locationsapproximately 2.4 and 5.2 centimetres from oneedge. These locations are shown in relation tothe mechanism in Figure 3.2. Examination of thesurface revealed that where the bolt came incontact with the chain, sever plastic deformationwas present. Examination of the 2.4 cm. fracturesurface (Figure 3.3), which was located in an
area of chain contact plastic deformation,revealed characteristics of fatigue. The fracturesurface had little to no zone of final fractureindicating that the loads perpendicular to thefracture plane where low. Fatigue characteristicsshowed that fracture initiated on the oppositeside to the deformation. This indicate thatbending forces were present in the pin. Bendingwould have caused one side of the pin to be intension and the other in compression. Thefatigue started on the tension side. Examination
of the 5.4 cm fracture surface located in the middle of the load transfer block revealed the same
characteristics of fatigue failure (Figure 3.4). However, a comparison of the two fracture surfaces on theadjoining Piece of the Pin revealed that the initiation sites were on opposite sides of the pin ( Figure 3.5).This indicates that bending forces at the two fractures were opposite.
Figure 3.1a. Pin industrial drawing. Figure 3.1b. Photograph ofbroken pin.
Figure 3.2. Industrial drawing of pin chain and block
mechanism.
Page 12 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
13/22
A chemical analysis performed on the body of the pin revealed it to conform to the SAE AISI standard1095. The original drawings for this application specify a SAE-AlSl 4140 (Table 3.1) Metallurgicalexamination of the mounted sample revealed plastic deformation at the edges as well as no significantinclusions. Examination of the microstructure revealed a ferrite matrix with spherodised carbides(Figure 3.6). The soft ferrite matrix increases the odds of fatigue initiation but will slow down fatiguepropagation.
Microhardness measurements show that the pin was slightly harder in the centre than on the surface(Table 3.2). The softer surface would have increased the possibility of fatigue initiation at the surface.
Figure 3.3. Photograph offracture surface of 2.4 cmfracture.
Figure 3.4. Photograph offracture surface of 5.4 cmfracture.
Figure 3.5. Photograph of pinindicating locations of fractureinitiation.
Table 3.1. Results of pin chemical analysis.
Element Pin SAE-AISI 1095
Carbon 1.06 0.90-1.03
Manganese 0.31 0.30-0.50
Silicon 0.25 --
Phosphor 0.011 0.040Sulphur 0.008 0.050
Chrome 0.03 --
Nickel 0.03 --
Molybdenum 0.01 --
Figure 3.6.Microphotograph of pinmicrostructure. Ferritematrix with spherodisedcarbides. 2% nital 1000X
Table 3.2. Microhardness results.
Hardness VHN (200g)
Location Longitudinal Section Transversal Section
Side 235 229
232 248
241 261
Page 13 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
14/22
Conclusions and Recommendations:
As the crane charges and unloads, the pin is subjected to bending forces. These forces create tensileforces on the surface at which point the probability of fatigue initiation is high.
Since the pin undergoes cyclic stresses, a steel for this application must have a high resistance tofatigue initiation. For these reasons, the original design material, SAE AISI 4140 hardened to arange of 45 - 50 HRC, was a good choice.
The block and chain should be examined for wear. If worn they would allow for larger bendingthen was originally allowed for in the design. If they are worn, they should be replaced.
If these measures do not correct the problem and the pin continues to break in future, the forces inthe original design should be revised.
Case Study 4: Shaft Bearing Failure
Introduction:
A bearing that had been in service for a year and a half was sent to undergo failure analysis (Figure 4.1).This bearing had been installed in the drive of a #P-40 centrifugal pump in the R-8 plant. It was locatedon a long shaft to separate the pump from the drive due to the presence of concentrated sulphuric acid.The shaft was belt driven at about 800 RPM. No special events were noticed in the pump operation.
Results:
Observations:
The inner raceway showed severe plastic deformation around its circumference in the form of a groove,which is located above the area designed to be the ball raceway (Figure 4.2). Spalling, a flaking andcracking of the surface, was observed in the groove but was not evenly distributed around its
275 268
Centre 294 294
Figure 4.1. Photograph ofbearing setup
Figure 4.2. Photograph ofinner ring showingspalling in groove.
Figure 4.3. SEMphotograph of spalling,flaking and cracking, inthe groove. 200X
Figure 4.4. SEMphotograph showingpresence of 45 sheerplanes. 500X
Page 14 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
15/22
circumference. Examination of the spalling using a scanning electron microscope (SEM) exposedflaking and the presence of surface cracks (Figure 4.3). Increased magnification of this area revealedfracture surfaces at forty-five degree angles indicating shear loads were present (Figure 4.4).
The inner raceway fracture surface is perpendicular to the groove and is located where the spalling ismost severe. Beachmarks and river lines, which are characteristic of fatigue failures, revealed severalinitiation sites situated in the base of the groove (Figure 4.5). Closer examination with the SEM
confirms that fatigue initiated from the spalling damage (Figure 4.6). Spalling was also seen to a lesserdegree on the balls surfaces. The outer raceway revealed no major defects.
Material characterisation and evaluation:
Both the compositions of the ball bearing and the inner raceway were found to fall within the norms for52100 steel, AISI-SAE standards (Table 4.1). The microhardness measurements of both pieces aretypical for this type of steel (Table 4.2). Surface hardness measurements for both ball and inner ring aresimilar, which is required by this type of application.
Figure 4.5. Photograph of the inner ring
fracture surface. Figure 4.6. SEM
photograph of the innerring fracture surfaceshowing fatigue initiating at
spall in the groove. 200X
Table 4.1--Result of chemical analysis.
Element
Analysed Composition of Ball
(%)
AnalysedComposition
of Inner Ring (%)
AISI-SAE 52100Standard
Composition
Ranges (%)
Carbon 0.97 1.02 0.98-1.10
Manganese 0.40 0.37 0.25-0.45
Silicon 0.24 0.23 0.15-0.30
Phosphorus 0.013 0.013 0.025
Sulphur 0.007 0.006 0.025
Chromium 1.21 1.36 0.025
Nickel 0.11 0.12 --
Page 15 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
16/22
Microscopic examination of a cross section of the inner raceway revealed surface cracks consistent withthe spalling observed (Figure 4-7). Etching the sample revealed a homogeneous macrostructure of atempered martensite matrix with undissolved carbides present (Figure 4.8). This microstructure agreeswith the chemical analysis and microhardness measurements.
Microscopic examination of a quartered ball bearing also revealed surface cracks (Figure 4.9). A largecrack extending towards the centre of the bearing was also found (Figure 4.10). The microstructure isheterogeneous, unevenly distributed; tempered martinsite with undissolved carbides. The large surfacecrack ties along a border of the heterogeneity (Figure 4.11). Some decarburization was observed on thesurface near spalling cracks.
Conclusions:
The failure was a result of vibrational fatigue initiated at spalling on the surface of the inner raceway.
Molybdenum 0.02 0.05 --
Table 4.2--Results of microhardness tests.
Ball Bearing Inner Ring
Hardness # Centre SurfaceDamagedSurface Centre
OutsideSurface
1 650 890 890 775 890
2 574 890 890 792 787
3 618 927 890 804 890
Figure 4.7.Micrograph ofcracks on the innerring surface. 200X
Figure 4.8.Microphotograph ofthe inner ringmicrostructurecomposed ofmartensite andundissolvedcarbides. 2% nital200X
Figure 4.9.Micrograph ofcracks on the ballsurface. 100X
Figure 4.10.Microphotographsof crack in a ball.15X
Figure 4.11.Microphotograph offigure 4.10 etchedwith 2% nitalshowingheterogeneousmartensite structurewith undissolvedcarbides. 15X
Page 16 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
17/22
The spalling, which is a characteristic of contact fatigue, originated from the bearing being InstalledIncorrectly or from it undergoing abnormal equiaxial radial loads in service, which caused adisplacement of the inner ring. This displacement increased the axial loads causing the plasticdeformation and spalling. Decarburization and uneven tempering of the balls as well as the extent ofplastic deformation indicate a temperature rise.
Case Study 5: Bronze Bull Gear Failure
Introduction:
A bronze bull gear was sent for failure investigation (Figure 5.1). It was used to rotate bleach washernumber 65B at a rate between 4 and 5 RPM. The contacting gear was a hardened steel worm gear, whichwas powered by a 50 horsepower 1800 RPM electric motor. The gear is a cast copper alloy with cutteeth and machined surface and was only in service for one month.
Observations:
Examination of the gear tooth revealed that there was a large amount of material loss. A measurementtaken near the base of the tooth where the material loss was most obvious revealed that tooth had gonefrom a thickness of 31 mm to 20 mm, a loss of I I mm. The contact surface had grooves running alongthe path the worm gear would have taken. Debris was also found along what was probably the exitingedge of the gear teeth (Figure 5.2). Along the front of the teeth, plastic deformation was seen near theedges where decreasing thickness could no longer support the load. Some cracking was observed inthese areas. When opened, they revealed that the mode of crack propagation was interdendritic.
Figure 5.1. Photograph showingthe bronze bull gear.
Figure 5.2. Photograph of the bull gear profileshowing debris and severe materials loss.
Table 5.1--Chemical composition of bull gear.
Composition %
Element Bull Gear Standard C90700
Copper 88.51 88.0-90.0
Page 17 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
18/22
Chemical analysis of the bronze gear revealed that it conformed most closely with the UNS standard forcopper alloy C90700 (Table 5.1). The lead and zinc content however were slightly above those allowedby the standard. Several samples where taken from the gear and examined microscopically. Theyrevealed large amounts of interdentritic shrinkage porosity (Figure 5.3) and interdentritic segregation
(Figure 5.4). The porosity reduces the amount of area supporting the load and therefore raises stresses inthe material. The heterogeneity of the structure is caused by rejection of tin into solution as the dendritesgrow while cooling. This segregation also reduces the mechanical properties of the material. Etching themicrostructure with 20 nil NH40H, 20 ml H20, 20 nil H202 (3%) revealed a coarse dendritemicrostructure (Figure 5.5). No plastic deformation of the working surface was observed whichindicates abrasive wear.
Hardnesses were taken on the cross section of a tooth which gave an
average Vickers hardness number of 76.6 VHN (5Kg) (Table 5.2). Thisis below the Brinell-500 Kg hardness number of 95 (100VHN) requiredby the ASTM standard, B427-93a "Standard Specification for GearBronze Alloy Castings". A lower hardness number also suggests that themechanical properties of the material would be below standards. Thisagrees with our metallographic examination.
Conclusions:
The bronze bull gear failed as a result of sever abrasive wear. The gear
Aluminium
7/27/2019 Davidson.html
19/22
did not meet ASTM materials specifications for this application and thisprobably had a great influence on the final failure. However, there areseveral possible causes of abrasive wear for which the system should be examined:
If the surface of the matching worm gear were damaged in any way, the difference in hardnesswould have led to severe wear.
If the lubricant was contaminated with an abrasive material wear will occur. If there was a misalignment between the two gears, the contact surface may be reduced increasing
contact loads above those that the material can withstand. If the system was overloaded, the rate of wear increases.
If one or a combination of these factors is present, it is then likely others failures would follow.
In this case, a large amount of porosity, a coarse dendrite structure, and interdentritic segregationcombined to reduce the properties of the bronze bull gear below those required by ASTM B427-93astandards. A possibility is that that when the gear is subjected to loads or overloading, these lowproperties would allowed the gear teeth to deflect. The gear surfaces would no longer meet as they weredesigned, decreasing the contact surface, which would have increased the loads and therefore wear.
Contamination of the lubricant would have followed, causing the wear to continue.
In future this bronze bull gear should be ordered specifying that it conform to ASTM standard B427-93afor the copper alloy UNS C 90700. As well the lubricant should be checked regularly for contaminationand both gear surfaces should be examined for damage.
Case Study 6: Analysis of 316L Reducer Failure
Introduction:
An 8" x 6", 316L stainless steel reducer was sent for failure analysis (Figure 6. 1). It had been in servicefor 13 months when a leak was noticed. The reducer was installed on #1 acid storage tank, equipmentnumber 50-200. The anodically protected carbon steel tank, contained off specification concentrated93% sulphuric acid. The flow rate through the reducer was 400 gal/min.
The tank was originally designed with a 4" diameter carbon steel nozzle, at floor level, that connected
Figure 6.1. Photograph ofreducer.
Figure 6.2. (a-left) Old tank installation. (b-right) Tank installation at the time ofreducer failure.
Page 19 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
20/22
directly to a valve (Figure 6.2a). This lasted seven to eight years without incident. The design waschanged to accommodate renovations so that an 8" carbon steel nozzle was installed 6" above the tankfloor. This nozzle lead into the failed reducer, which then connected to a valve composed of alloy 1-0steel (Figure 6.2b). This valve was said to be badly corroded. The valve then led to a 6" pipe made of316L stainless steel in which no problems were found. After the reducer failure, the piping arrangementswere changed so that the reducer is now after the valve.
Observations:
Visual examination of the reducer revealed an area at the top where little damage was observed (Figure6.3). This area, which was probably an air pocket, extended from the top of the 87' diameter flange intothe reducing pipe where is stopped just before the 6" diameter flange. Damage in this area consisted ofminor pitting (Figure 6.4). Damage, resembling a honeycomb structure in places, was most severe justbelow the air pocket in the reducing pipe near the 6" diameter end ( Figures 6.5a and 6.5b). This is wherethe leak was found (Figure 6.6). The damage becomes less severe in the pipe section towards thebottom. Only pitting was found in both the 8" and 6" flanges.
Chemical analysis of the flange and the pipe revealed that they both conform to AISI-SAE standards for316L stainless steel (Table 6.1)
Figure 6.3. Photograph ofthe top insider of thereducer showing the areaat the top where littledamage occurred.
Figure 6.4.Microphotograph ofpitting in air pocket. 15X
Figure 6.5. Photographs showing areas to the (a-left)right and (bright) left of the top relativelyundamaged surface. The red arrow in (a) indicatedwhere the leak occurred.
Table 6.1--Result of chemical analysis.
Element
Analysed Composition
of Flange (%)Analysed Composition
of Pipe (%)AISI-SAE 316L StandardComposition Ranges (%)
Carbon 0.031 0.034 0.03 max.
Manganese 1.85 1.28 2.00 max.
Silicon 0.57 0.35 1.00 max.
Phosphorus 0.014 0.011 0.045 max.
Sulphur 0.023 0.001 0.03 max.
Chromium 16.53 17.47 16.0-18.0
Nickel 10.85 11.46 10.0-14.0
Molybdenum 2.16 2.08 2.0-3.0
Page 20 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
21/22
Closer examination of the inside surface of the reducer with a SEM revealed dimples (Figure 6.7). Thesefeatures are typical of a ductile deformation, which indicates abrasion. The orientation of the featuresalso follows the direction of liquid flow. Pitting and uniform corrosion was also found in the region(Figure 6.8).
Conclusions and Recommendations:
A combination of two mechanisms caused the failure. Severe turbulence in the reducer caused adegradation of the passive layer that protects the stainless steel from corrosion. This would have left thesystem open to severe corrosion, which in turn would have lead to failure. The top of the reducer wasprobably protected by the presence of an air pocket.
The second mechanism was erosion, originating when air bubbles near the surface imploded causingmechanical damage, cavitation. Turbulence in the system may have formed bubbles from the air pocketat the top of the reducer. These bubbles would then have been carried into the reducer where increasingpressures would have caused them to implode. The highly corrosive environment would have increasedthe rate of degradation dramatically.
The new setup, placing the valve before the reducer, changed the dynamics of the system and may havesolved the problem, however existing reducers and valves should have their thickness monitored atregular intervals using an ultrasonic thickness gauge. If problems reoccur, the system should beevaluated for excessive turbulence and air pockets. A possible solution would be to use a PTFE liner inthe reducer. This would provide a barrier that protects against turbulence but not cavitation.
APPENDIX 1: EXAMPLE QUESTIONNAIRE
Bibliography
D.A. Ryder et al., "General Practice in Failure Analysis," inASM Metals HandbookVolume 11 "FailureAnalysis and Prevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)B.E. Wilde, "Stress-Corrosion Cracking," inASM Metals HandbookVolume 11 "Failure Analysis andPrevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)
Figure 6.6. Photograph takenon the outside of the reducer
showing the hole where thereducer leaked.
Figure 6.7. SEMphotograph of the inside
surface of the reducer inthe damaged area. 200X
Figure 6.8. SEMphotograph of the inside
surface of the reducer inthe damaged area. 500X
Page 21 of 22An Introduction to Failure Analysis for Metallurgical Engineers
23/09/2013http://www.tms.org/Students/Winners/Davidson/Davidson.html
7/27/2019 Davidson.html
22/22
K. H. Kamdar, "Liquid-Metal Embrittlement," inASM Metals HandbookVolume 11 "Failure Analysisand Prevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)Alan G. Glover et al., "Failures of Weldments," inASM Metals HandbookVolume 11 "Failure Analysisand Prevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)L. Windner, "Failures of Rolling-Element Bearings," inASM Metals HandbookVolume 11 "FailureAnalysis and Prevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)"Threaded Steel Fasteners," inASM Metals HandbookVolume 11 "Failure Analysis and Prevention",
Ed. Kathleen Mill (Ohio: ASM International, 1986)Walter J. Jensen, "Failures of Mechanical Fasteners," inASM Metals HandbookVolume 11 "FailureAnalysis and Prevention", Ed. Kathleen Mill (Ohio: ASM International, 1986)E. Alban, "Failures of Gears," inASM Metals HandbookVolume 11 "Failure Analysis and Prevention",Ed. Kathleen Mill (Ohio: ASM International, 1986)Michael Bauccio ed. Et al.,ASM Metals Reference Book, Third Edition, Ed. Kathleen Mill (Ohio: ASMInternational, 1993)Geaorge E. Dieter, Mechanical Metallurgy (Toronto: McGraw-Hill, Inc., 1986)Douglas A. Skoog and James J. Leary, Principles of Instrumental Analysis, Fourth Edition (Toronto:Sauders College Publishing, 1992)William D. Callister, Jr.,Materials Science and Engineering: An Introduction, Third Edition (Toronto:
John Wiley & Sons, Inc., 1994)Kathleen Mill ed. et al.ASM Metals Handbook: Metallography and Microstructures, (Ohio: ASMInternational, 1993)
Search Student Member Application Other Best Papers Student Members' Page TMs OnLine
Page 22 of 22An Introduction to Failure Analysis for Metallurgical Engineers